Lucy Allais - Manifest Reality - Kantâ S Idealism and His Realism (2015, Oxford)

OUP CORRECTED PROOF – FINAL, 8/7/2015, SPi
Manifest Reality
Manifest Reality
Kant’s Idealism and his Realism
Lucy Allais
1
3
Great Clarendon Street, Oxford, OX2 6DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
# Lucy Allais 2015
The moral rights of the author have been asserted
First Edition published in 2015
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2015931550
ISBN 978–0–19–874713–0
Printed and bound by
CPI Group (UK) Ltd, Croydon, CR0 4YY
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
To Geoffrey and Bernardine Barlow

Acknowledgements
My work on Kant has been improved by contributions and help from many people,
more than I can specifically acknowledge here, including the very many helpful
comments I have had in seminars and conferences where I have presented my
work. I owe a specific debt of gratitude to my teachers. I was introduced to Kant
by Michael Pendlebury’s brilliant lectures on the first Critique in the third year of my
undergraduate degree; it was love at first sight. John Kenyon, Quassim Cassam, Galen
Strawson, and Ralph Walker taught me in the B.Phil. and the last three also
supervised my D.Phil. I am very grateful for the different things I learned from
each of them, as well as for help and support given since the D.Phil. For reading
and giving comments on parts of the book (or early versions of the material in
papers), I’m grateful to Gary Banham, Graham Bird, John Callanan, Imogen Dickie,
Catharine Diehl, Christiane Diehl, Wolfgang Ertl, Hannah Ginsborg, Anil Gomes,
Stefanie Grüne, James Harris, David Martens, Colin Mclear, Thad Metz, Adrian
Moore, Beau Madison Mount, Michael Morris, Sasha Newton, James Pendlebury,
Oliver Pooley, Murali Ramachandran, Helen Robertson, Tobias Rosefeldt, Dennis
Schulting, Nick Stang, Andrew Stephenson, Clinton Tolley, James Van Cleve, Marcus
Willaschek, Daniel Warren, and Eric Watkins. I’ve been working on this project for
so long that it seems very likely that I have forgotten people; sincere apologies to
anyone I have left out. For reading and commenting on the entire manuscript,
particular thanks to Karl Ameriks, Colin Marshall, Helga Varden, Alnica Visser,
and anonymous readers for Oxford University Press. P. F. Strawson was my D. Phil.
examiner, and in addition to being utterly charming, he believed in the project, and
gave me encouragement at a crucial point; as did Galen Strawson, who advised me
not to quit at a time when I was thinking seriously about it, advice for which I am
grateful every day, since I can’t imagine a better job for me than the one I have.
It is very hard to write a book while teaching full time, which is at least one of the
reasons this has taken so long. I don’t think I would have been able to do it without
having had some time off teaching, and I am particularly grateful to time funded by
the Carnegie Corporation of New York and by the Alexander von Humboldt
Foundation.
Many of the thoughts and arguments presented in this book are based on thoughts
and arguments I have published previously in journal articles, although none
of the chapters simply reproduces these articles. Some of the material in Chapters
1, 2, and 6 was published in ‘Kant’s One World’, The British Journal for the History of
Philosophy, 12/4 (2004), 655–84. Some of the material in Chapters 5 and 6 was
published in ‘Kant’s Idealism and the Secondary Quality Analogy’, Journal of
the History of Philosophy, 45/3 (2007), 459–84. Some of the material in Chapter 7
viii ACKNOWLEDGEMENTS
was published in ‘Kant, Non-Conceptual Content and the Representation of Space’,

Journal of the History of Philosophy, 47/3 383–413. The central argument in
Chapter 8 was published in ‘Kant’s Argument for Transcendental Idealism in the
Transcendental Aesthetic’, Proceedings of the Aristotelian Society, Vol. CX, part 1
(2010), 47–75. Some of the material in Chapter 10 was published in ‘Intrinsic
Natures: A Critique of Langton on Kant’, Philosophy and Phenomenological Research,
73/1 (2006), 144–69. Some of the material in Chapters 3 and 4 was published in
‘Transcendental Idealism and Metaphysics’, Kantian Yearbook, 2 (2010), 1–31.
Permission from these journals is gratefully acknowledged.
Like Theseus’ ship, which existed continuously while every plank was replaced,
this book bears a historical relationship to my D.Phil. thesis. When I finished my
D.Phil. someone asked me what it would take to turn it into a book, and my father
suggested: ‘give it a happy ending’. As I see it, this book has a happy ending which, in
proper Hollywood style, also points to a sequel. While I am sure there are many
things in the book that I need to think about more, and many points which
could be better developed, I am convinced that I have shown that there is one
interpretation of transcendental idealism consistently presented in the first Critique.
Further, though I do not argue for this here, I think the interpretation presented here
makes sense of Kant’s strategy for dissolving the free will problem, and does so in a
way that will be helpful for the contemporary free will debate. There is clearly much
more work to be done.
Writing philosophy can be a lonely business. I’ve been kept sane and happy while
writing through the support and company of my family and friends, the overwhelm-
ing majority of whom have no interest in Kantian metaphysics and will never read
this book. I finished both the initial draft of the book as well as the final revised
manuscript (a year later) in the seclusion of Bernardine and Geoffrey’s beautiful
house in Kenton-on-Sea, and have dedicated it to my memory of them, in gratitude
for so much.
Contents
A Note on References to Historical Texts xi
Part One. Textual Evidence and an Interpretative

Pendulum
1. Navigating towards a Moderate Metaphysical Interpretation
of Transcendental Idealism 3
2. Why Kant is Not a Phenomenalist 37
3. Things in Themselves Without Noumena 59
4. Against Deflationary Interpretations 77
Part Two. Manifest Reality

5. Essentially Manifest Qualities 101
6. The Secondary Quality Analogy 125
7. Concepts and Intuitions 145
8. The Argument for Transcendental Idealism in the
Transcendental Aesthetic 176
Part Three. Kant’s Idealism and his Realism

9. Relational Appearances 207
10. Intrinsic Natures 231
11. The Transcendental Deduction: Relation to an Object 259
12. The Possibility of Metaphysics 290
Bibliography 309
Index 325
A Note on References to Historical Texts
Works of Immanuel Kant are referred to using the numbering in the Akademie Edition,
and the following abbreviations:
A/B Critique of Pure Reason

Anthropology Anthropology from a Pragmatic Point of View
C Philosophical Correspondence
CJ Critique of the Power of Judgment
CPR Critique of Practical Reason
Discovery On a Discovery, according to which any new Critique of
Pure Reason is made Superfluous through an Older One
Groundwork Groundwork of the Metaphysics of Morals
Inaugural Dissertation On the Form and Principles of the Sensible World
JL Jäsche Logic
LL Lectures on Logic
LM Lectures on Metaphysics
MFNS Metaphysical Foundations of Natural Science
MV Metaphysik Vigilantius
New Elucidation A New Elucidation of the First Principles of Metaphysical
Cognition
Progress What Real Progress has Metaphysics made in Germany
since the time of Leibniz and Wolff?
Proleg. A Prolegomena to any Future Metaphysics that will Present
itself as a Science
True Estimation Thoughts on the True Estimation of Living Forces
VL Vienna Logic
Historical texts by other authors are cited using the original date of publication, with
the date of the edition cited listed in the bibliography.
PART ONE
Textual Evidence and an
Interpretative Pendulum
1
Navigating towards a Moderate
Metaphysical Interpretation
of Transcendental Idealism
I Introduction to Part One

At the heart of Immanuel Kant’s critical philosophy is an epistemological and
metaphysical position he calls transcendental idealism; the aim of this book is to
understand this position. Kant sees transcendental idealism as a major philosophical
revolution which will enable us to solve problems that have troubled philosophers for
centuries. It is supposed to solve what he takes to be the unavoidable conflicts in
which reason becomes ensnared concerning such questions as freedom of the will,
and to avoid problems to which, he thinks, other philosophical systems lead, such as
scepticism about the external world. He thinks that it explains the possibility and
limits of metaphysics as well as the necessary foundations of empirical knowledge. As
the key to Kant’s resolution of the free will problem, transcendental idealism is at the
intersection between his metaphysics and ethics, and his central argument for the
validity of morality turns on an appeal to transcendental idealism.1 Making sense of
transcendental idealism is central to understanding Kant’s philosophy across a wide
range of areas.
Despite the centrality of transcendental idealism in Kant’s thinking, in over two
hundred years since the publication of the first Critique there is still no agreement on
how to interpret the position. As Karl Ameriks dryly notes, ‘Kant scholarship has yet
to have been overcome by consensus’ (1992: 329). Not only is there still dispute, there
is not even a tendency towards convergence, and recent publications continue to
represent such a wide spectrum of views that it sometimes scarcely seems possible
that they are all interpretations of a single position, put forward by a single philoso-
pher, primarily in a single book. Dominant interpretations pendulum between two
extremes, both of which seem to have textual and philosophical support. Here are
two examples. First, Arthur Collins:
1
Groundwork Part III.
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
Kant is not an idealist. (1999: 2)

[A]n interpretation that finds a kind of idealism in Kant, that ascribes to him a reduction of
objects to mental representations . . . fails to capture the originality, profundity, and merit of his
thought. (1999: 3)
Second, James Van Cleve:

As I interpret him, then, Kant’s transcendental idealism is idealism indeed, at least regarding
everything in space and time. (1999: 4)
[O]bjects in space and time are logical constructions out of perceivers and their states. That
makes Kant a phenomenalist. (1999: 11)
Not only is there disagreement about whether Kant is an idealist, there even is
disagreement about whether transcendental idealism is a metaphysical position at
all, as opposed to an epistemological project involving a rejection of metaphysics. It is
difficult to state the basic position in uncontested terms. As I understand it, tran-
scendental idealism has three central parts:
(1) Kant distinguishes between things in themselves (Dinge an sich), on the one
hand, and things as they appear to us, or appearances (Erscheinungen), on the
other. This is closely related to, but not exactly the same as, his distinction
between noumena and phenomena.
(2) Kant argues that the spatio-temporal objects of our experience (things as they
appear to us, appearances) are mere appearances or mere representations that
do not exist apart from a connection to possible perceptions.
(3) Kant claims that we do not and cannot have cognition (Erkenntnis) of things
as they are in themselves.
There is controversy surrounding the interpretation of all three claims. There is
disagreement as to how to understand Kant’s distinction between things as they are
in themselves and things as they appear to us, and about whether it is supposed to be
epistemological or metaphysical. There is dispute about whether Kant’s claim that
appearances are mere representations commits him to being an idealist and, if so, of
what sort. To make things more complicated, these distinctions cut across each other,
because the commentators who deny that Kant is an idealist include both some of
those who see transcendental idealism as a metaphysical position and some of those
who deny this.2 There is dispute about whether he is actually committed to the
existence of things in themselves or whether he merely thinks that the concept of
things in themselves is one to which our reason naturally leads us, without having a
commitment to there existing anything corresponding to the concept. Among those
2
For example, Paul Abela (2002) argues that transcendental idealism does not involve idealism because
it is an epistemological position involving the rejection of the so-called given, while Rae Langton (1998), in
contrast, does see the position as metaphysical, but also claims that it does not involve idealism, and instead
concerns a distinction between extrinsic and intrinsic properties.
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
who agree that he is committed to the existence of things in themselves, there is

disagreement about what this commitment amounts to, why he thinks we cannot
have knowledge of them, and whether we can say anything about them. For example,
there is dispute about whether he is committed to the existence of non-sensible
non-spatio-temporal objects in addition to the spatial objects of our knowledge, or
whether his position is that the objects of our knowledge have unknowable intrinsic
natures.
There are good textual reasons for seeing Kant as centrally concerned with
metaphysics—and good textual reasons for denying this. Kant introduces his pos-
ition as a revolution and compares it to the Copernican Revolution in astronomy
and, in a rough parallel with the way Copernicus explained the apparent movement
of the heavens as really being due to the movement of the earth, Kant says that he is
going to ascribe to the knowing subject some of what we normally think of as mind-
independent features of the spatio-temporal world. Famously, he suggests that, in
some respects, objects ‘conform’ to our knowledge rather than our knowledge
‘conforming’ to objects. This sounds like idealism. But some commentators have
argued that it is instead a move away from metaphysics in favour of epistemological
and meta-philosophical concerns. Kant clearly has such concerns. The Critique is
centrally addressed to the problem of how metaphysics is possible. Metaphysics, Kant
thinks, is neither an empirical science nor merely logic: its claims, he thinks, are
synthetic and a priori. Kant opens the Critique with the claim that synthetic a priori
judgments are mysterious and he addresses the book to the question: how are
synthetic a priori judgments possible? A large part of Kant’s answer is that traditional
(transcendent) metaphysics, which he sees as concerned with God, freedom, and
immortality (A3/B7), deals with topics with respect to which knowledge is not
possible for us. Rather than attempting, in vain, to answer these metaphysical
questions, he wants to diagnose the errors that lead us to think we can answer
them, and to argue that the only possible substantial a priori knowledge we can
have is knowledge of the a priori conditions of empirical cognition. This might seem
to support the view that Kant’s concern is to reject metaphysics in favour of
epistemological concerns, an impression that is strengthened by such statements as
the following:
[T]he understanding can never accomplish a priori anything more than to anticipate the form
of a possible experience in general, and, since that which is not appearance cannot be an object
of experience, it can never overstep the limits of sensibility, within which alone objects are
given to us. Its principles are merely principles of the exposition of appearances, and the proud
name of an ontology, which presumes to offer synthetic a priori cognition of things in general
in a systematic doctrine (e.g., the principle of causality), must give way to the modest one of a
mere analytic of the pure understanding. (A247/B303)
This passage has been taken to support an interpretation of Kant’s transcendental

idealism as an alternative to ontology, and a number of commentators think that to
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
interpret transcendental idealism as any kind of metaphysical position would be to

miss the point of Kant’s revolutionary programme.3
Though epistemological concerns are clearly an important part of it, I will argue
that Kant’s position also involves some metaphysics. At least some of the time when
Kant talks of ‘metaphysics’ or ‘ontology’ it is with the aim of criticising and rejecting
these subjects as conceived by his predecessors, but he also speaks of metaphysics in
other ways. On the one hand, he says that his ‘Copernican’ experiment seems to be
disadvantageous ‘to the whole purpose with which the second part of metaphysics
concerns itself ’, because it shows that ‘we can never get beyond the boundaries of
possible experience’ (Bxix). On the other hand, he also says that it ‘promises to
metaphysics the secure course of a science in its first part where it concerns itself with
concepts a priori to which the corresponding objects appropriate to them can be
given in experience’ (Bxviii, my italics; see also A845/B873).4 While he wants to show
that we cannot have knowledge of non-spatial, non-sensible objects like God and
Cartesian souls—the kind of knowledge, he thinks, to which metaphysics has trad-
itionally aspired—Kant also argues that we can have knowledge of a priori conditions
of the possibility of experience or empirical knowledge, which, he thinks, are
expressed in synthetic a priori claims. Kant wants to show, for example, that we
can know a priori that the spatial objects of our experience are in necessary causal
connections with each other and that they are made up of stuff which exists before
and after they exist (that substance is conserved). Despite his rejection of transcen-
dent metaphysics, it is not unreasonable to see this as giving us an account of
metaphysical claims that we can establish: a metaphysics of experience, as opposed
to a transcendent metaphysics. There are, therefore, some general reasons for
thinking both that Kant is rejecting one kind of metaphysics and also that he is
doing metaphysics, in another sense.
Even if Kant himself did use the term ‘metaphysics’ exclusively to refer to the
traditional metaphysics he is rejecting (the attempt to have a priori knowledge of
God, freedom of the will, and the soul), since this is clearly not what philosophers
generally mean by the term today, his rejection of metaphysics in this particular sense
would not be a reason to think that no aspect of his position is metaphysical in our
3
Allison (2004), Bird (2006).
4
In his Lectures on Metaphysics we find the same thing: sometimes Kant defines metaphysics and
ontology in terms of the kind of questions he is arguing that it is not possible for us to answer, but
sometimes he seems to see his project as part of metaphysics in another sense, for example when he speaks
of metaphysics as the science of a priori principles of cognition (LM 29: 749–54). He distinguishes between
different ways in which we can think of metaphysics, saying that ‘in cosmology and also in ontology there
are propositions which have objects in experience, and also those which do not—hence the critique of
reason must assume quite different basic propositions with respect to its immanent as opposed to its
transcendent use. We have classified metaphysics into the part which contains the immanent use of reason
and that which contains the transcendent’ (LM 29: 768; see also 29: 749–50, 29: 793, 29: 794, 29: 776, and 4:
274). And he also says that ‘All the despisers of metaphysics, who wanted to give themselves the
appearances of having clearer heads, also had their own metaphysics, even Voltaire. For everyone still
thinks something about his own soul’ (LM 29: 765).
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
current sense of the term. His account is partly concerned with such epistemological
questions as determining what kinds of things we can cognize and the a priori
conditions of empirical cognition, but it is also concerned with the nature of
reality—whether, for example, every event has a cause, and with the extent to
which spatio-temporal objects are independent of our minds or are dependent on
us. I will argue that while some of Kant’s central concerns are epistemological, his
transcendental idealism must be understood as containing substantial metaphysical
commitments: to the mind-dependence of things as they appear to us, and to the
existence of an aspect of reality that grounds the appearances of things, and which we
cannot cognize. And I will show how he takes this position to explain the possibility
of the kind of metaphysics that is possible for us: synthetic a priori claims about
spatio-temporal objects. However, I will also argue that transcendental idealism is
not the extreme idealist position it is sometimes taken to be. Further, I will argue that
understanding transcendental idealism as a (partly) metaphysical position is com-
patible with reading some of Kant’s key arguments, such as central parts of the
Transcendental Deduction of the categories, as epistemological.
Interpretations of Kant’s distinction between things in themselves and appearances
have tended to veer between two extremes. At one extreme are those metaphysical
interpretations which understand the distinction as one between non-sensible,
non-spatio-temporal things and appearances which exist merely as constructions
out of mental states. Understood literally, the term ‘noumena’ refers to objects which
are known by the intellect alone and which are not known through sense experience.
We can also call such objects intelligibilia. Kant thinks that objects which could be
known by an intellect alone would be non-spatio-temporal and non-sensible things,
such as Cartesian souls and Leibnizian monads, a fundamentally different kind of
thing than the spatio-temporal objects of our knowledge. The extreme metaphysical
reading of transcendental idealism takes Kant to be committed to the existence of
noumena in this sense (a position I call noumenalism), as well as to the claim that we
cannot know such objects, and also sees him as a phenomenalistic idealist with
respect to the objects of experience—things as they appear to us. At the other extreme
are proponents of deflationary views which deny that Kant’s transcendental distinc-
tion is an ontological one, seeing it instead as an epistemological or methodological
distinction between two ways of considering the same things.5 Although it precedes
them, the dispute between extreme metaphysical and deflationary or epistemological
interpretations is exemplified by the interpretations of P. F. Strawson (1966) and
Henry Allison (1983), respectively.6 The extreme idealist interpretation of Kantian
5
This kind of position is associated most prominently with Allison (1983; 2004) and Prauss (1971;
1974). There is some controversy with respect to both Allison and Prauss as to whether they really have
deflationary readings, but they are widely assumed to do so. See Westphal (2001) for the dispute with
respect to Allison, and Pippin (1974; 1976) and Ameriks (1982b) for discussion of Prauss.
6
See Ameriks (1982a) for a summary of the dispute and the allegiances of the disputants.
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
appearances has a long and distinguished history, dating to the very first review of the
Critique—the notorious Göttingen Review.7 At one point it was called standard,8
often by its opponents, and it may be dominant historically, but in the last fifty years
deflationary interpretations have proliferated9 and may be dominant among con-
temporary interpreters. In keeping with the historical pattern of oscillation between
the two extremes, there has recently been a resurgence of traditional phenomenalist
and noumenalist readings.10 I will present a position that avoids both extremes: a
moderate metaphysical interpretation.
The dispute between phenomenalist and noumenalist versus deflationary inter-
pretations is sometimes presented as a dispute between ‘two-world’ and ‘one-world’
interpretations.11 The idea is that interpretations of the first type are committed to
an ontological distinction between different kinds of entities (constituting different
worlds: the noumenal and the phenomenal worlds), whereas interpretations of the
latter type are concerned with a distinction between two ways of considering one set
of things (and therefore one world of entities). However, there are problems with this
terminology. Some of those who see Kant as an extreme, phenomenalistic idealist
deny that the term ‘two-world’ is appropriately applied to their views.12 On the other
hand, those who see transcendental idealism as an epistemological or methodological
thesis might argue that the characterisation of the debate as one between two-and
one-world interpretations illegitimately assumes an ontological starting point, which,
they maintain, Kant rejects. There is a great deal of variety among views that reject
the idea that Kant’s distinction is one between two kinds of objects: they include,
for example, Allison’s (1983) methodological reading, which sees Kant as con-
cerned with the conditions of cognition; Abela’s (2002) epistemological reading,
which sees Kant as concerned with the rejection of the ‘given’; and Langton’s
(1998) metaphysical distinction between intrinsic and extrinsic properties. Some
commentators deny that Kant is committed to there being a way things are in
themselves13 and read him as committed to only ‘one world’ in virtue of seeing
appearances as all that exists, rather than in virtue of seeing appearances as one
‘aspect’ of a single set of entities which also have a way they are in themselves. And
although talk of ‘two aspects’ of objects (as opposed to two objects or two worlds)
is associated with deflationary interpretations, it is also possible to have metaphy-
sical two-aspect views; such views can even be introduced by talking about two
7
Garve and Feder (1782).
8
See, for example, Hoke Robinson (1994: 415) and Allison (1983: Ch. 1).
9
For example, Bird (1962), and Prauss (1971; 1974), Allison (1973; 1983; 1996), Matthews (1982), and
Pippin (1982).
10
See for example Jauernig (forthcoming), Stang (forthcoming), and Hogan (2009a; 2009b).
11
I discuss it in these terms in Allais (2004). See Walker (2010) for criticism of this terminology.
12
Guyer provides an example of this. He says: ‘I have never held that Kant posits a second set of things
that are ontologically distinct from ordinary things or appearances’ (Guyer 2007: 12). He continues to
make explicit the fact that, on his view, Kant reduces ordinary empirical objects to mental representations.
13
See Hanna (2001), Senderowicz (2005), Bird (2006), and Hanna (2006: 15, 197–8).
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
different ways of considering objects.14 With this messy terminological back-

ground, I will avoid talk of ‘one-world’ and ‘two-world’ interpretations. My aim
is to reject two kinds of extreme views: on the one hand, deflationary interpret-
ations which don’t see Kant’s position as containing idealism and which don’t see
him as committed to there being a way things are in themselves and, on the other
hand, extreme metaphysical interpretations which see him as a noumenalist and a
phenomenalist. I argue for a moderate metaphysical interpretation which sees Kant
as holding that the things of which we have knowledge have a way they are in
themselves that is not cognizable by us, and that the appearances of these things
are genuinely mind-dependent, while not existing merely in the mind.
There is an abundance of apparent textual evidence as well as philosophical
considerations that can be appealed to in support of both extremes. At the same
time, both views have serious problems. Kant says that the spatio-temporal, physical
objects of our experience are appearances, that they are mere representations, that
they do not exist apart from their connection to a possible perception, and that they
do not give us insight into what things are like as they are in themselves. This sounds
like idealism. But Kant also vehemently asserts that his position is nothing like
Berkeley’s, distances himself from idealism, and even regrets having called his
position ‘transcendental idealism’.15 He insists that his transcendental idealism is
also an empirical realism. He foregrounds epistemological concerns by saying that by
‘transcendental’ he means an account of how a priori knowledge is possible: his
position is supposed to give us an explanation of the possibility of a priori knowledge
of the objects of experience, and to rule out the possibility of any other kind of a
priori knowledge. He characterises the idealism he rejects as one that sees the
immediate objects of perception as mental items on the basis of which external
objects are inferred; against this, he argues that external objects in space are the
immediate objects of perception. And he also says that his position enables him to
demonstrate the reality of the very spatial objects which Descartes doubts and
Berkeley denies, and that experience of these objects is immediate and primary, not
inferred from our awareness of our inner states.
14
For example, assuming that there is a metaphysical difference between intrinsic properties and
relational properties, or between primary and secondary qualities, we could introduce such a distinction
by talking about considering an object in two different ways. We could introduce the notion of a thing’s
intrinsic properties by considering the thing as it is apart from its possible relations with us and other
things. Similarly, someone who holds colour to be dependent on visual experience might introduce a
distinction between primary and secondary qualities by considering objects as they are in our perceptual
experience and objects as they are apart from their being perceived by us. Starting with these two
ways of considering objects may lead to a genuine metaphysical distinction. For discussion of the point
that two-aspect interpretations can be metaphysical, see Westphal (2001: 594–5; 1997a: 232) and
Rosefeldt (2007).
15
He suggests, in the Prolegomena, that he should have called it ‘formal idealism’ or ‘critical idealism’
(Proleg. 4:337: 375).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
The existence of strong considerations in favour of both sides as well as serious

problems with both sides seems to keep the literature in a state of oscillation.16
Problems with one extreme view are often taken as support for the other extreme.17
Many extreme idealist interpreters are rightly dissatisfied with deflationary readings
that cannot do justice to the parts of the text in which Kant expresses his idealism;
they frequently seem to assume that the only way to do justice to these texts is
through seeing Kant as a phenomenalist. On the other hand, many deflationary and
bare empirical realist interpreters are rightly dissatisfied with interpretations that see
Kant as a phenomenalist, and from this they conclude that he is not an idealist.
Similarly, noumenalist interpretations point out that Kant clearly expresses a com-
mitment to there being a way things are in themselves, independently of us. However,
they wrongly conclude that he is committed to the existence of non-sensible,
non-spatio-temporal entities which are distinct from the objects of our knowledge.
Deflationary and mere empirical realist interpretations rightly point out the prob-
lems with seeing Kant as committed to intelligibilia; they argue that it follows that the
idea of things in themselves does not commit him to an existing feature of reality at
all. To reach a stable interpretation we need an account of idealism that is not
phenomenalist and that does justice to Kant’s empirical realism, and we need an
account of what it means to say that things have a way they are in themselves which
does not involve a commitment to intelligibilia.
In my view, part of what makes Kant’s transcendental idealism so complex and
difficult to pin down, as well as so compelling, is that it is a position which aims to
accommodate competing philosophical concerns. This aspect of Kant’s thought is
captured by Lorne Falkenstein’s diagnosis of the impetus of Kant’s intellectual
development:
Kant was not the sort of person who had the intellectual courage to face up to a dilemma and
reject one alternative in favour of the other. Instead, when he felt himself pulled in opposite
directions by conflicting imperatives, his preference was to try to work out some way of
satisfying them both. This intellectual cowardice . . . is the characteristic that lead Kant to his
most brilliant discoveries (Falkenstein 1995: 19).
Setting aside the character judgment, it seems to me that this captures one of the
most interesting features of Kant’s thought: his attempt to incorporate and mediate
between competing philosophical pressures. Similarly, in my view, a compelling
interpretation of Kant’s position will be one which does justice to the undeniable
interpretative pressures in both directions. There are strong grounds for thinking
16
This can be seen in a single author: McDowell (1994) has a ‘two-world’ view which appears to switch
to a deflationary view without a commitment to the existence of things in themselves in his later
(1998a) work.
17
For example, Allison aims his arguments against those who see Kant as a noumenalist or a
phenomenalist, or both (2004: 4; also 5–9, 46, 51, 54; and 2006: 112), while Van Cleve (1999) seems to
take Allison’s two- aspect view as his only real target; it is also the main object of Guyer’s criticisms.
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
that Kant is an idealist who is also committed to the existence of an aspect of reality
that we cannot know. There are also strong grounds for thinking Kant is neither a
phenomenalist nor a noumenalist.18 My interpretation accommodates both these
sides of Kant’s thought.
The continual pendulum swing between the extreme readings might support the
idea that there is no single view unambiguously presented in the Critique—Kant is
simply inconsistent. Given the centrality of the position in Kant’s philosophy, an
agreed interpretation is clearly desirable. The aim of this book is to argue that there is
an interpretation of transcendental idealism that is unambiguously supported in the
first Critique, that accommodates the textual evidence that seems to support the
extreme views, and that is a coherent position. As I see it, the challenge is to give an
interpretation which does not undermine or trivialise Kant’s claims that we know
only mind-dependent appearances and that we cannot know things as they are in
themselves, without involving phenomenalistic idealism about appearances, or com-
mitting Kant to the noumenalism that he denies. Deflationary interpreters do not see
transcendental idealism as an ontological position at all, which, of course, the
strongly idealist interpreters do. On my reading, Kant’s distinction is based on
epistemological considerations, and has epistemological consequences, but it also
involves metaphysical claims about what exists and about the mind-dependence of
the aspect of reality of which we can have knowledge. Kant’s position is a careful
combination of realism and idealism, and of metaphysical and epistemological
claims. In my view, this is part of the broader philosophical interest of understanding
transcendental idealism: Kant attempts to accommodate as far as possible both
philosophical concerns that lead to idealism and those that lead to realism.
There are two central parts to my approach to transcendental idealism. One is
emphasising Kant’s concern with cognition rather than knowledge, and, as central to
this, paying detailed attention to the role of what he calls intuition in cognition. In my
view, the nature of Kant’s idealism, his argument for his idealism, and his reasons for
thinking that we cannot cognize things as they are in themselves all crucially turn on
his notion of intuition. I argue that Kantian intuitions are representations that give us
acquaintance with objects, and that since he thinks cognition requires intuition, he
thinks our cognition is limited to that with which we can have acquaintance: what
can be presented to us in a conscious experience. The other central part of my
strategy is to show that Kant’s position looks very different depending on the
18
A similar dispute between interpretative extremes is exemplified by those who, like P. F. Strawson,
think that everything that is of value in the Critique can be entirely separated off from Kant’s transcen-
dental idealism, and those who think that seeing any aspect of Kant’s arguments as not dependent on,
leading to, or otherwise intimately embedded in his idealism is to fail to take transcendental idealism
seriously and to miss the coherence of Kant’s work. I think we should take transcendental idealism
seriously, and it is clearly the key to many of Kant’s arguments and positions in the Critique, but it is
compatible with this that Kant may have some arguments and epistemological insights which are separable
from his idealism.
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
assumptions we bring to reading it, in particular assumptions about the nature of

perception. Like a number of other commentators, I argue that it is crucial to see that
Kant rejects what I will call a Cartesian picture of perception. This is a view according
to which perception is indirect in the following sense: perception centrally involves
having or being in a mental state which the subject could be in whether or not an
object were being perceived, and which counts as perceptual when it is caused by the
object in the right way. This makes perception ‘indirect’, because the object itself is
not a constituent of the mental state a subject is in when perceiving; rather, the object
is merely the cause of the mental state. According to this view, a subject could be in
the same mental state when perceiving an object as when hallucinating an object; the
fact that the former state is one of perception and the latter is not is a function of the
different causes of the states in the two cases, and not of their content. According to
an alternative to the Cartesian account of perception, a perceptual mental state is a
state which involves the presence to consciousness of the object perceived. The idea is
that the presence to consciousness of the object is part of what makes the mental state
the state that it is. Such views used to be called ‘direct realist’ accounts, but are now
often, following John Campbell, called relational accounts of perception: the term
‘relational’ marks the idea that ‘the object perceived is a constituent of the conscious
experience itself ’ (Campbell 2002a: 117). I argue that we should approach transcen-
dental idealism with a relational view as a starting point.
Indirect or representationalist theories of perception were dominant in the early
modern period (at least in the standard reading of the early modern period).19 Kant
claims that the spatio-temporal objects of our experience are mere appearances or
mere representations. Since the term ‘representations’ suggests mental intermediar-
ies, it is often taken to support a reading of Kant as a phenomenalist idealist.
However, as I will argue in detail in Chapter 2, it is extremely difficult to make
sense of Kant’s position as a phenomenalist idealism. A Berkeleyan idealist or a
phenomenalist starts with a conception of mental states (ideas, sense data, or mental
contents) as things of which we have direct awareness, and then argues that these are
all that exist and that physical objects are constructions out of them, or supervene on
them, or exist simply as a matter of certain truths about these mental states. But Kant
has no such starting point. He thinks that cognition of our own mental lives is not
primary, but rather requires immediate experience of things outside us, and that we
would not even be able to be aware of the temporal determination of our own mental
lives without this.20 He thinks that we do not know what our minds are as they are in
themselves and that we do not understand the essence of either mind or matter
sufficiently well to know whether they are really different kinds of things. One of my
19
See Yolton (1996; 2000) for an alternative view. Note also that ‘representational’ accounts of
perception, today, are not always understood as indirect; I ignore this, as it is typically taken to refer to
indirect accounts of perception in relation to the early modern period.
20
This is his argument in the Refutation of Idealism (B274–9).
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
aims is to show how fruitful it is for the interpretation of transcendental idealism if we

do not approach Kant with a representationalist or indirect account of perception as a
starting assumption.21 I argue that if we approach transcendental idealism with a
direct realist or relational, rather than an indirect or Cartesian, account of perception,
we can find a non-phenomenalist kind of mind-dependence which makes sense of the
way Kant expresses both his idealism and his claims about things in themselves, and
which is compatible with far more of his text than are phenomenalist readings. On this
view, restricting what is empirically real to what can feature in a possible perception is
not restricting it to what exists in the mind, but instead to what can be directly or
immediately presented to minds like ours. While ‘Vorstellungen’ is standardly trans-
lated as ‘representation’, it could just as well be translated as ‘presentation’.22 Rather
than saying that appearances are things which exist only in minds, or as constructions
out of what exists only in minds, Kant can be read as saying that what counts as part of
the empirically real world is only what can be presented to us in (relational) perceptual
experience. As I read him, Kant thinks that appearances are not just perceptible
things, things which manifest themselves to us (as a realist might think); they are
essentially perceptible or essentially manifestable. Here, we have a kind of idealism or
anti-realism which holds that spatio-temporal reality does not transcend what is
essentially manifestable to finite receptive creatures like us. Empirical reality is
restricted to what can be presented to consciousnesses like ours, but what can be
presented to consciousness is not something which exists merely in the mind.
Approaching transcendental idealism with a relational view of perception is closely
related to the other central strand of my approach—emphasising Kant’s notion of
intuition and the role of intuition in cognition. Increasing attention is being paid by
scholars to the fact that Kant’s central term Erkenntnis, now standardly translated as
cognition, is not the same as knowledge. Cognition, unlike knowledge, can be false
(B83), and what is relevant to whether or not something qualifies as cognition is not
whether it has some specified kind of justification or warrant, but rather the kind of
representation of objects with which it is able to provide us.23 Both at the empirical
and the a priori level, Kant’s primary concern is with what it takes for us to achieve a
certain kind of objective representation of the world (cognition), rather than with
what kind of warrant is required for knowledge. He thinks that cognition requires
general, conceptual thought, but he also thinks that concepts never uniquely indi-
viduate objects and never put us directly in touch with objects. This means, he thinks,
that it is never the case that merely having a concept enables us to know that there
exists something that corresponds to the concept, and that the use of concepts alone
is not enough to enable us to have a thought that succeeds in being about some
21
A number of philosophers have argued that one of Kant’s central achievements is the rejection of the
Cartesian conception of experience. Allison (1973), Guyer (1987), Willaschek (1997), McDowell (1998a),
Collins (1999), Abela (2002), Ameriks (2003: 5; 2006: Ch. 6), and Bird (2006).
22 23
As it is in the Pluhart (1996) translation. See Schafer (forthcoming).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
particular object. It follows that concepts on their own do not succeed in relating to
objects in the way that is necessary for cognition. Cognition requires, in addition to
conceptual thought, the possibility of acquaintance with the objects of cognition.
Acquaintance (unlike merely having a concept) is a relation to an object that
guarantees the existence of the object and which individuates a specific particular.
Kant thinks that we cannot have acquaintance with the objects of traditional meta-
physics (God, Cartesian souls, and Leibnizian monads), and that we cannot have
acquaintance with things as they are in themselves, and therefore that we cannot
cognize them. Not only do we not have knowledge of them, we do not even really
succeed in representing them. Our thoughts about them are merely coherent
thoughts, and not properly objective representations, or cognition.
My reading of intuition as giving us acquaintance with objects goes against an
interpretative trend of assimilating intuitions to sensations. This trend is associated
with the widespread reading that attributes to Kant the idea that it is the application
of concepts that organises the sensory input to give us presentations of individual
objects. This leads to seeing intuition as mere sensory input, and to paying insuffi-
cient attention to the role of intuition in Kant’s account of cognition: that of giving us
objects. Seeing intuition as presenting us with objects—giving us acquaintance with
objects—enables us to understand Kant’s idealism and his empirical realism. We are
directly presented (in intuition) with objects outside us in space; but spatio-temporal
reality does not transcend what can be given to us in intuition. I show that this
reading of intuition makes sense of Kant’s central argument for his idealism, the
argument in the Transcendental Aesthetic, which entirely turns on the notion of
intuition, rather than on an explanation of synthetic a priori cognition in general, as
it is sometimes read.
My account of intuition has implications for how we understand Kant’s answer to
his question about cognition of synthetic a priori metaphysical claims. Notably,
Kant’s question is not about how such judgments are justified, but how they are
possible. I will argue that Kant’s primary question is how it is possible for such
judgments to qualify as cognition. Their being cognition requires that they concern
objects that are given to us in intuition—objects with which we have acquaintance—
but Kant thinks that we can have acquaintance with objects that are independent of
us only if they affect our senses. This makes it hard to see how synthetic a priori
claims could concern objects with which we have acquaintance (because they are a
priori), and therefore hard to see how they could qualify as cognition. I will argue that
understanding Kant’s question about the possibility of cognition of synthetic a priori
claims in this way is key to understanding his argument for his idealism in the
Transcendental Aesthetic, as well as to his explanation of the possibility of
metaphysics.
Focusing on cognition helps with a traditional concern about whether Kant can be
entitled to assert both that we cannot know things as they are in themselves and to
assert that there is a way things are in themselves. Kant’s claim is not that we cannot
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
know that there is a way things are in themselves but rather that we cannot cognize
things as they are in themselves. It may also help with some meta-philosophical
concerns about the status of Kant’s theory. If transcendental idealism is both a
metaphysical theory and an explanation of metaphysical knowledge, this raises
questions about our knowledge of the theory itself. If, in contrast, we start with an
account of empirical cognition and its conditions, and argue that these conditions
include certain a priori representations, and also that these conditions have impli-
cations for the extent of reality that we can cognize, the meta-philosophical problems
become less severe.
It should be noted that my concern here is with the first Critique.24 While Kant
argues that we cannot have what he calls theoretical cognition of things as they are in
themselves, he does also think that we can achieve some kind of cognition of them
through practical reason. I do not discuss this, and when I talk about our being
unable to have cognition of things in themselves, this should be taken throughout to
refer to theoretical cognition.
This book is longer than I would have liked it to be, but many chapters can be read
independently. In Part One I look at the textual evidence for the various interpret-
ations of transcendental idealism and present arguments against the two traditional
extremes. The arguments in this section are negative arguments against other
people’s views, and readers whose interests do not include tracing the interpretative
debates in the literature about Kant can easily skip this part of the book. Despite the
length of the book, the enormous amount of writing there is on Kant means that
there is much that I leave out, and my use of the literature is necessarily selective:
my aim is simply to represent philosophical and textual motivation for the
competing positions, and not to give a comprehensive account of everything written
on the topic.
In Chapter 2, I argue against a phenomenalist interpretation of Kantian appearances.
In Chapter 3, I argue against two extreme interpretations of Kant’s notion of
things in themselves: on the one hand, a reading which sees things in themselves
as non-spatio-temporal, non-sensible things, objects of a distinct kind from those of
which we have experience (noumenalism), and, on the other hand, a reading which
holds that Kant’s only ontological commitment is to empirically real, spatio-temporal
objects (empirical realism alone). I argue that Kant is committed to there being an
aspect of reality of which we cannot have knowledge, but that he is not a noumenalist.
In Chapter 4 I argue against deflationary interpretations according to which Kant’s
distinction between things as they are in themselves and things as they appear to us is
concerned merely with conditions of knowledge. I argue that although Kant is
concerned with a priori conditions of empirical knowledge, this is not all there
is to his distinction between things in themselves and appearances. And I argue
24
I take seriously Kant’s claim that his revisions in the second edition of the Critique alter the
presentation but not the substance of his account (Bxxxvii), and therefore draw on both editions.
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
that the philosophical work for which he invokes transcendental idealism—the

resolution of those conflicts to which he thinks traditional metaphysics leads—
requires genuine metaphysical commitments.
The upshot of Part One is that there are serious textual reasons as well as
philosophical considerations that have led interpreters to extreme positions, both
of which are unstable and unable to accommodate the evidence for the alternative
position. To make sense of Kant’s position we need an account of mind-dependence
that does not involve existence in the mind, and which is compatible with thinking
that mind-dependent appearances are grounded in the way things are in themselves.
The rest of the book presents my positive account of this.
In Part Two, I present my account of the kind of mind-dependence had by Kantian
appearances and the way this relates to Kant’s account of things as they are in
themselves. I argue that Kant’s view is that spatio-temporal reality is essentially
manifestable: it does not transcend what can be presented in a conscious experience
to creatures like us.
In Chapter 5, I present a relational account of perception which allows that we can
directly perceive things without perceiving them entirely as they are in themselves,
apart from their perceptual appearing. My concern in this chapter is neither with
Kant’s account of perception nor with making the philosophical case for a particular
account of perception; rather, I simply want to present the possibility of a relational
view in sufficient detail to show how fruitful it is to approach Kant with this view in
mind, rather than starting with a representationalist or indirect view. Within this
account, I present the idea of manifest qualities—qualities which are presented to us
in perception—and then the idea of essentially manifest qualities—qualities which are
presented to us in perception and which do not present us with features that objects
have independently of their being presented to us in perception. I then present a
possible view of colour according to which colour is an essentially manifest quality.
The idea is that colour is a directly presented feature of external objects and also that
it is a property the existence of which does not transcend our possible perceptual
experience of it.
In Chapter 6, I use this account of essentially manifest qualities to explain Kant’s
idealism about appearances. In the Prolegomena, Kant explains his idealism about
appearances by comparing them with so-called secondary qualities such as colour.
How the secondary quality analogy enables us to interpret Kant’s idealism will of
course depend on what account of secondary qualities we draw on: I argue that the
account of colour as essentially manifest presented in the previous chapter enables us
to understand Kant’s idealism in a way that fits the texts very well. As I understand it,
Kant’s view is that we can cognize only essentially manifest features of reality.25
25
Other commentators have read Kant in a fashion similar to mine, most notably Paton (1936), Dryer
(1966), Collins (1999), and Rosefeldt (2007). I see my argument here as part of a common project with
theirs, but there are some differences between our positions. My biggest disagreement is with Dryer, as he
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
I argue that this gives us a robust and radical, but non-phenomenalist, form of
idealism which can do complete justice to all the ways Kant expresses his idealism,
to the philosophical work he wants it to do, to the way he talks about the relation
between things in themselves and appearances, and to the considerations which
count against viewing him as a phenomenalist—his empirical realism.
Chapters 7 and 8 concern the role of intuition in Kant’s idealism. This is crucial,
because he limits empirical reality to what could possibly be presented to us in an
intuition, and because his central argument for idealism is based on the role of a
priori intuition. Chapter 7 is a detailed argument for my reading of intuition: the
claim that intuitions are representations that give us acquaintance with the objects of
thought. Much of this chapter consists in engaging with debates in the literature.
Kant has often been taken, incorrectly, in my view, to hold that it is the application of
concepts that enables us to be presented with particulars; this view is widespread, and
therefore requires thorough refutation. In Chapter 8 I use my account of intuition to
explain Kant’s central argument for his idealism in the Transcendental Aesthetic, the
first part of the Critique, in which he presents his account of space and time. There is
much dispute about how this argument is supposed to work. I argue that under-
standing intuitions as giving us acquaintance with objects enables us to see why Kant
takes his conclusion to follow at the point at which he does, without the need for any
extra premisses. I then argue that the kind of idealism which follows from the
argument fits better with my essential manifestness form of anti-realism than it
does with a phenomenalist interpretation.
Part Three of the book completes my account of the tightrope Kant walks between
realism and idealism, of the metaphysical and epistemological components of his
position, and of the relation between his idealism and his explanation of the possi-
bility of metaphysics.
Chapter 9 is concerned with Kant’s account of empirical reality. I argue that,
according to Kant, the spatio-temporal objects that constitute empirical reality are
essentially sensory and are relational or non-categorical (though not merely in
minds, or properties of something merely mental, or merely mental results of mind-
independent dispositions). In explaining Kant’s position I compare it to three contem-
porary views. Like a certain form of contemporary anti-realism in philosophy of
sometimes implies that there is no idealism, or no mind-dependence, in Kant’s position at all. Both Paton
and Collins can be read as using the secondary quality analogy in a way similar to that in which I use it.
Understanding Paton in this way is controversial, however, and Collins attributes to him a phenomenalist
reading of appearances (Collins 1999: 162), as does Bird (1962: 1). It could be that Paton is sometimes
inconsistent; Barker claims this about Paton’s use of the language of appearing, since he sometimes calls
appearances ideas (Barker 1969: 282; see Paton 1936: 442). Collins’s position might seem different from
mine in that he denies that Kant is an idealist while I argue that there is a significant sense in which Kant is
an idealist, but it seems to me that this difference may be terminological. My position is closest to that of
Rosefeldt (2007), although the sense in which he sees secondary qualities as dispositional is not entirely
clear to me. Although Ameriks (2000; 2003; 2006) does not spend much time on the secondary quality
analogy in specific, my approach is also in line with his moderate metaphysical interpretation.
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
language, I argue that Kant’s idealism can be understood as rejecting experience-

transcendent reality, rather than asserting that objects exist as ideas in our minds.
Like anti-realism in philosophy of science, Kant can be understood as limiting scientific
knowledge to what is (in some sense) observable. On Kant’s account, as I understand it,
there is no conflict between the so-called manifest and scientific images: science studies
manifest reality, and scientific knowledge is limited to manifest reality. Finally, like
structural realism in contemporary philosophy of science, Kant argues that science
gives us knowledge only of structural or relational features of reality.
In Chapter 10, I show that Kant does not think that a coherent ontology could
contain only essentially manifest, essentially relational features. I argue that Kant’s
position includes the claim that we cannot have knowledge of non-relational features
of reality but that he also thinks there must be something non-relational which
grounds relational appearances.
Chapter 11 is concerned with the Transcendental Deduction of the Categories.
Kant’s idealism is sometimes associated with forms of conceptual idealism which
hold that our concepts shape our reality. Against this, I argue that the central
concerns of the Deduction are epistemological. The aim of the Deduction is to
show that a priori concepts (the categories) apply to spatio-temporal objects. As
I see it, Kant’s strategy in the Deduction is to argue that without applying a priori
concepts to objects we would not be able to apply empirical concepts to them: the
categories are the conditions of empirical concept application. This shows that the
categories can be used, as Kant puts it, to think empirical objects—because they are
conditions of such thoughts.
Finally, in Chapter 12, I evaluate the implications of the previous three chapters for
the interpretation of transcendental idealism and I examine the delicate relation
between transcendental idealism and Kant’s explanation of the possibility of meta-
physics. I examine the different roles a priori concepts and a priori intuitions play in
Kant’s account: I argue that the idealism is a function of the role and nature of a
priori intuition, and that the categories need not be seen as introducing further mind-
structuring to the world. Rather, their legitimate use (use for cognition) is limited to
the mind-dependence of what can be given to us in intuition. However, I still see
transcendental idealism as playing a central role in Kant’s explanation of the possi-
bility of metaphysics, and in establishing the Deduction’s final conclusion. As I read
Kant, we must distinguish between the argument that shows that the categories have
relation to an object (objective validity) and the argument which shows that all spatio-
temporal objects are subject to the categories. The former, as I argue in Chapter 11,
is an epistemological argument about the conditions of empirical concept application.
The latter is a result of combining the conclusion of the epistemological argument with
an idealism that limits spatio-temporal objects to the conditions of our cognizing them.
Similarly, Kant’s synthetic a priori principles are established as conditional claims
(claims about the conditions of empirical cognition); they are converted into uncon-
ditional claims about spatio-temporal objects once we grant that spatio-temporal
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
objects do not exist independent of the possibility of our cognizing them. On this
account, we can take seriously the role of the idealism in explaining the possibility of
metaphysics without taking the explanation to be that it is because our minds ‘make’
objects in certain ways that we can know a priori claims about objects. Finally, I briefly
sketch features of my view which I think may be helpful for understanding Kant’s
account of free will.
There are many questions about Kant’s transcendental idealism that this book
does not tackle, most obviously the role the position plays in Kant’s attempt to
resolve the free will problem. However, by showing that the text is not full of
contradictions and by providing a coherent interpretation that makes sense of the
texts that seem to pull in different directions, my interpretation lays the groundwork
for understanding these further questions.
The remainder of this chapter presents the basic textual evidence for my moderate
metaphysical reading. I start by showing that there is a strong textual basis for
attributing to Kant the idea that empirically real, spatio-temporal objects depend
on our minds in some sense. I argue, in Section II, that the text does not force us to
read Kant as a phenomenalist (or extreme idealist who takes physical objects to exist
literally in minds), but that there are overwhelming textual grounds for thinking that
he is an idealist in the broad sense in which this term covers a commitment to
physical objects being mind-dependent in some way. In Section III I present textual
evidence for three points concerning Kant’s notion of things in themselves. One,
Kant speaks of appearances and things in themselves as aspects of the same things.
Two, he is committed to there actually being an aspect of reality which we cannot
cognize. And three, he holds that this uncognizable reality grounds the mind-
dependent objects of our cognition. Many commentators have given reasons for
not taking the text at face value; I respond to these arguments in Chapters 2 to 4.
II Textual Evidence for Idealism

Kant refers to the spatio-temporal objects of our cognition as appearances, which is a
short hand for things as they appear to us. I argue that his use of the word
‘appearance’ is not, on its own, decisive with respect to the question of whether he
is an idealist. However, there are very many prominent passages in the Critique
which support a broadly idealist reading of Kantian spatio-temporal appearances.26
There are three striking common features of these passages: first, Kant claims that
spatio-temporal objects are appearances ‘in us’; second, he calls appearances repre-
sentations; and third, he says that the existence of appearances requires a connection
with actual perception. I will comment on each of these. In my view, they provide
such strong evidence of Kant’s being an idealist of some sort that in the absence of an
26
See A42/B59, B45, A46/B63, A104–5, A127, A376, A383, A490/B518, A492–3/B521, A494–6/
B522–4, A505–6/B533–4, and A514–15/B542–3.
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
alternative account of the sense in which appearances depend on our minds, phe-
nomenalist interpretations would have strong textual support. However, I contend
that none of these features of the text provides conclusive reason for seeing Kant as a
phenomenalist, and I argue in the next chapter that phenomenalism is inconsistent
with many of Kant’s core philosophical concerns. In Parts Two and Three I give an
alternative account of the mind-dependence of appearances. In my view, it may often
be due to alternative forms of idealism not being considered (rather, the alternatives
to phenomenalist interpretations are assumed to be realist interpretations) that the
strong textual evidence for Kant’s being an idealist is taken to show him to be a
phenomenalist, despite the serious textual cost of this reading. Once we have an
alternative account of the mind-dependence Kant invokes, we need not see the
textual evidence that shows him to be an idealist as requiring an interpretation of
him as a phenomenalist.
The fact that Kant calls spatio-temporal objects appearances27 (Erscheinungen)
need not, on its own, be taken to imply their mind-dependence. There are a number
of different ways the term ‘appearance’ can be used, including neutral realist uses. It
can be used to imply a mere seeming, as in ‘the straight stick appears bent in water’;
here, appearance is opposed to the way things really are. It is clear that this is not
Kant’s usage. He explicitly denies that his view makes objects ‘a mere illusion’
(Schein), or that bodies only ‘seem to exist outside of me’ (B69).28 ‘Appearance’
has also been used to refer to sense data which are understood as merely mental
representations of reality. For example, Russell says that ‘what we directly see and feel
is merely “appearance”, which we believe to be a sign of some “reality” behind’
(Russell 1912: 6).29 Understood in this way, Kant’s use of ‘appearance’ would
correspond to the extreme idealist interpretation of transcendental idealism accord-
ing to which the existence of the empirical world consists in the existence of actual
and possible sense data or mental states. However, there is also a neutral use of
‘appearance’, in which appears means ‘is manifest’, or ‘becomes manifest’, and has no
implication of mind-dependence, as in ‘when you come over the hill, the church
appears’.30 If we understand Kant’s use of appearance in this way, calling spatio-
temporal objects appearances need not have any idealist implications. It simply
implies that they are things of which we have sense experience—things which are
presented to us. Kant has been read in this neutral way. Dryer, for example, says that
‘when Kant is translated as asserting that empirical knowledge can be got only of
27
For example A20/B34, A42/B59, and A46/B53.
28
Similarly, he says that ‘when appearances are in question, and this term is taken to have the same
meaning as semblance, one is always poorly understood’ (MFNS 4: 555).
29
He says that ‘what the senses tell us immediately is not the truth about the object as it is apart from
us, but only the truth about certain sense-data’, ‘the various sensations due to various pressures of
various parts of the body cannot be supposed to reveal directly any definite property of the table, but at
most to be signs of some property which perhaps causes all the sensations’ (Russell 1912: 6).
30
Kant’s initial explanation of appearance is that it is the undetermined (unconceptualised) object of an
empirical intuition; this has no implication of mind-dependence (A20/B34).
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
“appearances”, what he is saying is that it can be got only of what presents itself to
empirical intuition’ (Dryer 1966: 50631). Similarly, Bird says that Kant ‘is a transcen-
dental idealist because [he thinks] it is quite unwarranted to make claims about
objects which are not open to any sort of perceptual inspection. Kant’s empirically
neutral term “appearance” was thus designed to limit the range of our possible
experience to the objects that can be presented to our senses’ (Bird 1962: 50). On
this reading, the significance of saying that the objects of our knowledge are appear-
ances lies in restricting our cognition to things that we can, in principle, perceive and
thereby rejecting the claim that we have knowledge of any non-sensible objects, such
as Leibnizian monads, God, or immortal souls.32 I do not rest my interpretation on
any reading of ‘appearances.’ In my view, even though the neutral realist use of
appearance is defensible (at least some of the time), there are sufficient other reasons
to see Kant as an idealist. It is certainly a central part of his position that we can
cognize only things which we can, in principle, perceive (which can be presented to
us in sensible intuition, which can appear to us). However, Kant makes further,
specific claims about the status of appearances: throughout the Critique he says that
spatio-temporal appearances exist ‘in us’, that they are representations (Vorstel-
lungen), or mere representations (bloße Vorstellungen), and that their existence
requires a connection to possible perception.33 These claims strongly support seeing
Kant as an idealist about appearances; I discuss each of them in turn, and look at
some of the texts in which they feature.
The first feature of the text that supports seeing Kant as an idealist is his repeated
claim that appearances exist ‘in us’. For example, he says:
We have therefore wanted to say that . . . if we remove our own subject or even only the
subjective constitution of the senses in general, then all the constitution, all relations of objects
in space and time, indeed space and time themselves would disappear, and as appearances they
cannot exist in themselves, but only in us. (A42/B59, my italics)
The transcendental idealist . . . allows this matter and even its inner possibility to be valid only
for appearance—which, separated from our sensibility, is nothing—matter for him is only a
species of representations (intuition), which are called external, not as if they are related to
objects that are external in themselves but because they relate perceptions to space, where all
things are external to one another, but that space itself is in us. (A370)
We have sufficiently proved in the Transcendental Aesthetic that everything intuited in space
or time, hence all objects of an experience possible for us, are nothing but appearances, i.e.,
mere representations, which, as they are represented, as extended beings or series of alter-
ations, have outside our thoughts no existence grounded in itself. (A490–1/B518–19)
31
See also Bird (1962: 46, 50, 148).
32
See also Bxxvi, A20/B34, A26/B42, and A239/B298. Against the neutral reading of ‘appearance’, see
A490–1/B518–19.
33
B45; A98; A101; A104; A109; A113; B164; A190/B235; A197/B242; A369; A370; A372; A383; A385;
A386; A490–1/B518–19; A493/B521; A494/B522; A499/B527; A507/B535; A563/B591; A793/B821; Proleg.
5: 288, 289, 319, 341, 342.
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
Space itself, however, together with time, and, with both, all appearances, are not things, but
rather nothing but representations, and they cannot exist at all outside our mind. (A492/B520)
It is not hard to see why such passages have led many commentators to see Kant as a
phenomenalistic or Berkelyean idealist who thinks that appearances exist literally in
our minds or as constructions out of ideas which exist in our minds. Berkeley, as he is
standardly read, thinks that physical objects are collections of ideas which exist in our
minds; in these passages Kant might be thought to be expressing exactly this view.
However, Kant clearly distinguishes between what he calls empirical and tran-
scendental senses in which objects can be ‘in us’ (in uns) and ‘outside us’ (ausser uns)
(A373); it is crucial to read the above claims with this disambiguation in mind.34 In
the passage in which Kant describes his disambiguation he says that:
[T]he expression outside us carries with it an unavoidable ambiguity, since it sometimes
signifies something that, as a thing in itself, exists distinct from us and sometimes merely
something that belongs to outer appearance, then in order to escape uncertainty and use this
concept in the latter significance—in which it is taken in the proper psychological question
about the reality of our outer intuition—we will distinguish empirically external objects from
those that might be called “external” in the transcendental sense, by directly calling them
“things that are to be encountered in space”. (A373)
Kant says that what is outside us in the empirical sense are things which exist in
space whereas what is transcendentally external is ‘something that, as a thing in
itself, exists distinct from us’ (von uns unterschieden existiert, my italics). The
transcendental sense of ‘outside us’ (to which appearances being ‘in us’ is opposed)
is the idea of things which exist distinct from us. A straightforward way of under-
standing Kant’s point here is that he is distinguishing the thought of a thing’s being
outside us in space from the thought of a thing’s existing independently of us: what is
transcendentally ideal can be empirically outside us (can exist in space), but is not
independent of us. Although Kant’s disambiguation does not give a particular
account of what this latter dependence amounts to, it sets some constraints on
understanding it. A non-phenomenalist kind of idealism will make more sense of
his contrast between what is transcendentally in us (dependent on us) and what is
empirically in us (what is inside our minds, rather than outside us in space) than will
phenomenalism. Phenomenalism, broadly, holds that the existence of objects in
34
This is forcefully argued by Henry Allison (2004: 24) and Karl Ameriks (2000: 111–12). Allison’s
interpretation of this distinction is that, understood empirically, the terms ‘in us’ and ‘outside us’ mark a
distinction between objects of inner and outer sense, respectively, but understood transcendentally, they
mark a distinction between ‘two manners in which objects can be considered in relation to the conditions
of human sensibility’ (Allison 2004: 24). Allison is clearly right to point out the significance of the
disambiguation between the empirical and transcendental senses of ‘in us’, but this does not settle the
interpretative question, or establish any particular account of the latter. Notably, Kant’s disambiguation of
the empirical and transcendental senses of ‘in us’ and ‘outside us’ says nothing about two ways in which
objects can be considered in relation to the conditions of human sensibility, as Allison’s interpretation
requires. Allison’s interpretation is discussed in Ch. 4.
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
space does not amount to anything more than the existence of certain actual and
possible mental states (and what can be constructed out of these).35 Kant says here, in
contrast, that objects do not exist in our minds, yet their existence is not independent
of our minds. Against merely empirical realist and deflationary interpretations,
however, it is clear that his account of what it means to be transcendentally ‘in us’
involves some kind of mind-dependence: he says that spatial objects do not exist
distinct from us. This means that to make sense of Kant’s saying that appearances are
‘in us’, our aim should be to look for an account of mind-dependence that does not
involve existence in the mind.
Another way of responding to the apparent support provided for phenomenalist
interpretations by Kant’s assertion that objects are ‘in us’ is to say that the expression
‘in us’ is metaphorical. It has been argued that some early modern writers talk of
objects being in the mind as a way of expressing the idea of their being understood,
known, or apprehended.36 This is an important point. In light of Kant’s disambigu-
ation of the transcendental and empirical senses of ‘in us’, it is reasonable to think
that the expression is not meant to refer to what is literally or merely in our minds.
The point of the disambiguation, after all, is to say that objects which are not literally
in our minds (since they are outside us in space) might still not exist distinct from us.
However, the fact that ‘in us’ can be read metaphorically does not show that Kant’s
position is not idealist. The expression is clearly meant to express some kind of
relation to mind, and Kant says that this relation to mind is a condition of the
existence of appearances: he says not just that empirical objects are in us, but that
they are merely in us and that they exist only in us. In other words, even if ‘in us’ can
mean apprehended by us or cognized by us, Kant is saying not just that objects are
apprehended or cognized by us, but that their existence depends on the possibility of
their being apprehended or cognized by us (A490–1/B518–19). This is an idealist
claim. However, it need not be understood in a phenomenalistic sense if we can give
some other account of the way in which they are dependent on being cognizable
by us.
The second and most prevalent feature of the passages that seem to support a
phenomenalist interpretation is the fact that in them Kant calls spatio-temporal,
empirically real objects, ‘representations’ (Vorstellungen) or ‘mere representations’
(bloße Vorstellungen). For example, he says that:
[W]hat we call outer objects are nothing other than mere representations of our sensibility.
(B45, my italics)
35
It is worth noting that despite the many features of the text which indicate the mind-dependence of
appearances, Kant in fact never says that appearances are constructions out of mental states: phenomenalist
interpretations explain Kantian appearances as constructions, but Kant himself never does.
36
See Yolton (1984). Similarly, Aquila argues that for Brentano, ‘To say that actual objects of sense are
“in” one’s sensory awareness is just to say that there is some awareness of them, not that those objects are,
in any literal sense, in the mind at all’ (Aquila 1974a).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
[A]ppearances are not things in themselves, but rather the mere play of our representations,
which in the end come down to determinations of the inner sense. (A101)
[A]ppearances themselves are nothing but sensible representations, which must not be
regarded in themselves, in the same way, as objects (outside the power of representation).
(A104, my italics)
[A]ll objects of an experience possible for us, are nothing but appearances, i.e., mere represen-
tations, which, as they are represented, as extended beings or series of alterations, have outside
our thoughts no existence grounded in itself. This doctrine I call transcendental idealism.
(A490–1/B518–19, my italics)
I understand by the transcendental idealism of all appearances the doctrine that they are all
together to be regarded as mere representations and not as things in themselves, and accordingly
that space and time are only sensible forms of our intuition, but not determinations given for
themselves or conditions of objects as things in themselves. (A369, see also A370, quoted earlier)
[I]f I were to take away the thinking subject, the whole corporeal world would have to
disappear, as this is nothing but the appearance in the sensibility of our subject and one
mode of its representations. (A383)
rather nothing but representations, and they cannot exist at all outside our mind. (A492/B520,
my italics.)37
Kant’s use of the term ‘representations’ is seen by many commentators as pro-

viding strong support for a phenomenalist interpretation.38 It is repeated many
times throughout the Critique, and not only does Kant call appearances represen-
tations, he says that they are mere representations, and he frequently links calling
appearances ‘representations’ to asserting their mind-dependence. For example, at
37
See also A98, A109, A113, B164, A190/B235, A197/B242, A369, A372, A385, A386, A493/B521,
A494/B522, A499/B527, A507/B535, A563/B591, A793/B821, Proleg. 288, 289, 319, 341, and 342. It should
be noted that some of the most strongly phenomenalist sounding passages are in the fourth Paralogism in
the A edition, a notoriously controversial section, which Kant dropped from the second edition. Also,
although Kant consistently upheld the distinction between appearances and things in themselves in the
theoretical philosophy published after the Critique, he seldom again used the phenomenalist-sounding
language that features there. Here are some representative descriptions of transcendental idealism from
later works: ‘the world as appearance is merely the object of possible experience’ (Progress: 20: 290); ‘nature
is . . . the sum total of all things, insofar as they can be objects of our senses, and thus also of experience.
Nature, in this meaning, is therefore understood as the whole of all appearances, that is, the sensible world,
excluding all non-sensible objects’ (MFNS 4: 467; also 4: 477); ‘The teaching of the Critique therefore
stands firm: that no category can contain or bring forth the least cognition, if it cannot be given a
corresponding intuition, which for us human beings is always sensory, so that the use of it in regard to
the theoretical cognition of things can never extend beyond the limits of all possible experience’ (Discovery:
8: 198); ‘the Critique . . . shows that in the corporeal world, as the totality of all objects of outer sense, there
are, indeed, everywhere composite things, but that the simple is not to be found in it at all. At the same
time, however, it demonstrates that if reason thinks a composite of substances as thing-in-itself (without
relating it to the special character of our senses), it must absolutely conceive of it as composed of simple
substances’ (Discovery: 8: 209); and ‘This ideality of space and time is nevertheless, at the same time, a
doctrine of their perfect reality in regard to objects of the senses (outer and inner) qua appearances, i.e., as
intuitions so far as their form depends on the subjective constitution of the senses’ (Progress 20: 268).
38
See, for example, Van Cleve (1999: 123).
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
A492/B520, he says that space, time, and appearances are nothing but represen-
tations and cannot exist outside our minds.
The term ‘representations’ has a strong association with representational theories
of perception and is easily read as referring to something which exists only in the
mind. However, like the term ‘appearance’, it could also be read in a more meta-
physically neutral way. ‘Appearance’ can be taken to refer merely to what appears to
us: what manifests itself to us. Relatedly, ‘representation’ can be used to refer to that
which is represented or presented to the mind. As already mentioned, an alternative
translation of Vorstellungen is presentation.39 To say that the objects of our know-
ledge are Vorstellungen is to say that they are things that are presented to us; it need
not follow that they are mental intermediaries. In fact, as I will argue in the next
chapter, one of Kant’s aims in the Critique is to reject a view of perception on which
we are directly or immediately in touch only with mental intermediaries. This
provides further reason for thinking that the connotations of the term ‘representa-
tion’—the suggestion of representational theories of perception—are unfortunate
and misleading in this context. Since the term ‘Vorstellung’, despite its associations,
need not be read as indicating mental intermediaries, Kant’s saying that appearances
are Vorstellungen does not provide conclusive support for a phenomenalist reading.
However, it does support thinking that he is an idealist of some sort. This is because
he does not say simply that empirically real, spatio-temporal objects manifest
themselves to us or are presented to us. He says that they are mere presentations
which do not exist independently of the possibility of their being presented to our
minds. While this may not mandate a phenomenalist reading, it strongly supports
seeing Kant as asserting some kind of mind-dependence.
The third feature of the text which seems to support a phenomenalist reading is the
fact that Kant repeatedly says that outer objects exist or are real in perception only
(wirklich nur in der Wahrnehmung), and that the actuality of a thing requires ‘its
connection with some actual perception in accordance with the analogies of experi-
ence’ (A225/B272, my italics). Some of these passages are strongly reminiscent of
Berkeley’s claim that the being (esse) of empirical objects is perception (percipi).40
Kant says that
The real in outer appearance is thus actual only in perception, and cannot be actual in any other
way. (A376, my italics)
[T]he objects of experience are never given in themselves, but only in experience, and they do
not exist at all outside it. That there could be inhabitants of the moon, even though no human
being has ever perceived them, must of course be admitted; but this means only that in the
possible progress of experience we could encounter them; for everything is actual that stands in
one context with a perception in accordance with the laws of the empirical progression. Thus
39
It is translated this way in the Pluhar (1996) translation. See McLear (2013) for a defence of this
reading.
40
Berkeley (1710: 78).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
they are real when they stand in an empirical connection with my real consciousness, although
they are not therefore real in themselves, i.e., outside this progress of experience. (A493/B521,
my italics)
[A]ppearances, as mere representations, are real only in perception, which in fact is nothing
but the reality of an empirical representation, i.e. appearance. To call an appearance a real
thing prior to perception means either that in the continuation of experience we must
encounter such a perception, or it has no meaning at all. For that it should exist in itself
without relation to our senses and possible experience, could of course be said if we were
talking about a thing in itself. But what we are talking about is merely an appearance in space
and time, neither of which is a determination of things in themselves, but only of our
sensibility; hence what is in them (appearances) are not something in itself, but mere
representations, which if they are not given in us (in perception) are encountered nowhere
at all. (A493–4/B521–2)
Some merely empirical realist interpreters see Kant’s claim about the link between
the existence of appearances and our possible perception of them as simply express-
ing the thought that our cognition is limited to objects of which we can have sense
experience. For example, Bird says that
Transcendental Idealism holds, then, that all our knowledge is based upon experiences, and
that we have no knowledge of anything that cannot be experienced . . . the force of Transcen-
dental Idealism is only to suppose that we have experiences, and that these play an essential
part in our knowledge. (Bird 1962: 148. See also Dryer 1966: 84–5, 500, 506)41
These views are acceptable to a certain sort of realist: to say that there are conditions
under which objects can become objects of knowledge is not to make the objects of
knowledge mind-dependent, and many straightforward empirical realists would
agree that we can have knowledge only of objects which affect our senses. Restricting
our cognition to spatio-temporal things which causally interact with our senses is
clearly a crucial part of Kant’s position and while this may not seem controversial
from the point of view of contemporary empiricism and naturalism,42 in the context
of the Leibnizian rationalism he is responding to, it is an important point. So these
readings are right as far as they go: Kant does restrict our cognition to things which
affect our senses. But he also argues that what is given to us in intuition, what appears
to us, are only representations, which cannot exist ‘outside us’, and whose existence is
dependent on a connection with actual perceptions. The way Kant often puts this is
in terms of the idea of possible experience: he says that everything that is empirically
real is part of the extent of possible experience. Kant says not just that empirically real
objects are known only in experience (or that we can cognize only objects of which
we can have sense experience), but that they exist only in the extent of possible
41
Similarly, Prauss says that to consider things as appearances is to consider them as objects of the
intuition of our sensibility (Prauss 1974: 37).
42
Although, see Hanna (2006), who argues that much contemporary scientific realism is committed to
noumenal realism, because it is committed to unobservable objects.
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
experience or in our representations. This claim goes beyond the limitation of our
knowledge to objects which can be presented to our senses. In saying that appear-
ances exist merely ‘in us’ and that they are merely representations, Kant expresses
some kind of mind-dependence; he then explains this idealism through the claim that
the existence of appearances requires a connection to actual perception (wirkliche
Wahrnehmung). In order to establish a non-phenomenalist interpretation of tran-
scendental idealism, we need an account of existence being linked to actual percep-
tion which does not involve existence in the mind. I present such an account in
Chapters 6 and 9.
The fact that Kant says that empirically real objects are mere representations,
which exist in us and which require a connection to possible perception, provides, as
we have seen, strong reason for seeing him as an idealist in some sense. Much of the
appeal of phenomenalist interpretations, despite what is, as we will see, a substantial
list of serious objections to them, is due to the fact that many of the available
alternatives do not give a satisfactory account of the mind-dependence of Kantian
appearances.43
III Things in Themselves

So far my concern has been with the textual evidence for transcendental idealism’s
being a genuine form of idealism. In the remainder of this chapter I document textual
evidence for the claim that Kant is genuinely committed to thinking that there is an
aspect of reality that we cannot cognize: the way things are in themselves. I present
three groups of texts, which exhibit three striking features of the way in which Kant
talks about things as they are in themselves. First, Kant repeatedly speaks of things as
they appear to us and these same things as they are in themselves. Second, he
frequently and clearly talks as if he thinks that there actually is a way things are in
themselves—that there exists an aspect of reality that is independent of us. Third, he
frequently makes such claims as that things in themselves are the ground of appear-
ances, and he also speaks of things in themselves as the cause of appearances and of
their affecting us.
The most straightforward way of doing justice to all these texts is to see Kant as
committed to thinking that the empirically real things that we experience are
43
Many commentators argue for phenomenalist interpretations simply by arguing that Kant is an
idealist, and without considering other possible accounts of the mind dependence in question. For
example, Van Cleve argues that only phenomenalism is compatible with the mind-dependence of
appearances, because the only way it is possible for objects to owe any of their traits to our manner of
cognizing them is if ‘the objects in question owe their very existence to being cognized by us’ (Van Cleve
1999: 5). He says that only a view which has it that the esse of appearances is percipi can make sense of their
mind-dependence. Van Cleve’s argument is not compelling, because the claim that the only explanation of
objects conforming to our cognition of them is phenomenalism is disputable: positions such as the anti-
realism of Dummett and Wright hold that objects must conform to our knowledge of them, without
thereby thinking that they exist in the mind. See Wright (1992; 1996) and Dummett (1993).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
grounded in, or are an aspect of (or something given under an aspect of 44),
something that exists independent of us, which we cannot cognize, and which is
ontologically fundamental in the sense that it is somehow responsible for what we
experience.45 However, each part of this claim has been denied by some Kant
scholars. Some interpreters have denied that Kant’s position is consistent with
thinking that we can even think coherently about things in themselves.46 Many
interpreters deny that Kant is committed to a metaphysical position according to
which there actually exists an aspect of reality that we cannot know.47 This may be
part of a generally deflationary approach, but it is also advanced by some who see
transcendental idealism as an ontological position committed to empirically real
things only.48 Some argue that the notion of things in themselves is merely a limiting
concept—a coherent concept which we cannot help using but which is such that we
cannot have knowledge that there is actually something to which it applies. Both the
notion of an actually existing aspect of reality which we cannot know and the idea of
things in themselves as the ground or cause of things as they appear to us are claimed
by some to be inconsistent with Kant’s own account of the conditions of knowledge
and the legitimate use of the categories.49 At the other extreme (and against the first
set of texts), some see Kant as committed to the existence of non-sensible, non-
spatio-temporal things which are distinct from the things of which we have experi-
ence.50 I argue in detail against each of these interpretations in Chapters 3 and 4; for
the moment, my aim is simply to record a number of texts that challenge them.
In the first group of texts, we see that Kant frequently presents his distinction as
being between things as they are in themselves, and those same things as they appear.
He says:
[T]he unconditioned must not be present in things insofar as we are acquainted with them
(insofar as they are given to us), but rather in things insofar as we are not acquainted with them,
as things in themselves. (Bxx, my italics)
44
Talk of Kant as distinguishing between two ‘aspects’ of things has been strongly associated with
Allison-style deflationary interpretations. However, there is nothing intrinsically deflationary about such
talk, and it is possible to have metaphysical two-aspect views. See Westphal (2001), Allais (2004; 2006;
2007; 2010), and Rosefeldt (2007).
45
See Willaschek (2001) for helpful discussion of this point.
46
An example is Melnick’s (1973) interpretation, according to which the notion of a thing in itself is the
notion of an object quite literally incomprehensible to us and is a purely limiting concept (Melnick 1973:
152). Another example of this kind of position is Matthews’s suggestion that ‘we might contrast the world
as we thus describe it, using our conceptual framework, with the world that we thus describe, the world to
which our concepts are applied. The latter world would be ex hypothesi indescribable and, in a sense,
unthinkable. Nothing could be said in detail about it’ (Matthews 1982: 137). However, it is clear that Kant
thinks that we can have coherent thoughts about things as they are in themselves. Keller argues that this is
also a problem for Allison’s view (Keller 1998: 226–8).
47
For example, Bird (1962; 2006), Grier (2001), Allison (2004), Senderowicz (2005), and Hanna (2006).
48
For example, Hanna (2006).
49
See for example, Prauss (1974) and Senderowicz (2005).
50
See P. F. Strawson (1966: 236), Jauernig (forthcoming), and Stang (forthcoming).
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
Now if we were to assume that the distinction between things as objects of experience and the
very same things as things in themselves, [eben denselben, als Dingen an sich selbst] which our
critique has made necessary, were not made at all, then the principle of causality and hence the
mechanism of nature in determining causality, would be valid of all things in general as
efficient causes. I would not be able to say of one and the same thing, e.g., the human soul, that
its will is free and yet that it is simultaneously subject to natural necessity, i.e., that it is not free,
without falling into contradiction; because in both propositions I would have taken the soul in
just the same meaning, namely as a thing in general (as a thing in itself), and without prior
critique I could not have taken it otherwise. But if the critique has not erred in teaching that the
object should be taken in a twofold meaning, namely as appearance or as thing in itself; if its
deduction of the pure concepts of the understanding is correct, and hence the principle of
causality applies only to things taken in the first sense, namely insofar as they are object of
experience, while things in the second meaning are not subject to it; then just the same will is
thought of in the appearance (in visible actions) as necessarily subject to the laws of nature and
to this extent not free, while yet on the other hand it is thought of as belonging to a thing in
itself as not subject to that law, and hence free, without any contradiction hereby occurring.
(Bxxvii–xxviii, my italics)51
[T]he same objects can be considered from two different sides, on the one side as objects of the
senses and the understanding for experience, and on the other side as objects that are merely
thought at most for isolated reason striving beyond the bounds of experience. If we now find that
there is agreement with the principle of pure reason when things are considered from this twofold
standpoint, but that an unavoidable conflict of reason with itself arises with a single standpoint,
then the experiment decides for the correctness of that distinction. (Bxvii–xixn, my italics)
[A]ppearance . . . always has two sides, one where the object is considered in itself (without
regard to the way in which it is to be intuited, the constitution of which however must for that
very reason always remain problematic), the other where the form of the intuition of this object
is considered. (A38/B55)
[T]hese a priori sources of cognition determine their own boundaries by that very fact (that
they are merely conditions of sensibility), namely that they apply to objects only so far as they
are considered as appearances, but do not present things in themselves. (A39/B56, my italics)
We have . . . wanted to say that all our intuition is nothing but the representation of appearance;
that the things that we intuit are not in themselves what we intuit them to be, nor are their
relations so constituted in themselves as they appear to us. (A42/B59, my italics)
[I]n the appearance the objects, indeed even the properties that we attribute to them, are always
regarded as something really given, only insofar as this property depends only on the kind of
intuition of the subject in the relation of the given object to it then this object as appearance
(dieser Gegenstand als Erscheinung) is to be distinguished from itself (von ihm selber) as object
in itself. (B69)
51
A possible view, suggested by Adams (1997), is that while Kant need not be seen as committed to
thinking that empirical objects have a way they are in themselves, he does think that subjects have a way
they are in themselves as well as a way they appear to us. See also Ameriks (1982a: 6). See Marshall (2013a)
for a detailed argument that Kant’s commitment to noumenal and phenomenal subjects being one thing
provides a general reason for thinking that appearances and things in themselves are not distinct things
(though this allows that there could also be distinct noumena which do not have a phenomenal nature).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
[W]e call certain objects, as appearances, beings of sense (phaenomena), because we distin-
guish the way in which we intuit them from their constitution in itself. (B306, my italics)
[W]e do not understand through pure reason what the things that appear to us might be in
themselves. (A277/B333, my italics)
[T]he human being . . . obviously is in one part phenomenon, but in another part, namely in
regard to certain faculties, he is a merely intelligible object. (A546/B574)
What matter is, as a thing in itself (transcendental object) is of course, entirely unknown to us.
(A366)
[T]he doctrine of sensibility is at the same time the doctrine of the noumenon in the negative
sense, i.e., of things that the understanding must think without this relation to our kind of
intuition, thus not merely as appearances but as things in themselves. (B307)
As deflationary interpreters have argued, these texts show that Kant’s distinction is
between two ways of considering the same objects, or two aspects of objects, rather
than between supersensible noumena and ontologically distinct mental items.
A phenomenalistic idealism can regard appearances that exist only in our minds as
representations of things which have a way they are in themselves, but it is hard to see
how such phenomenalistic appearances could be aspects of the very same things
which have a way they are in themselves, so these passages strongly count against
phenomenalist readings of Kant’s idealism. However, as will be discussed in more
detail in subsequent chapters, the idea that Kant distinguishes between two aspects of
things does not commit us to seeing his distinction as merely epistemological or
methodological: there are metaphysical two-aspect readings. Further, as we will see,
deflationary interpretations are threatened by the next two groups of texts.
Simply distinguishing between considering things as they appear to us and as they
are in themselves might leave it open as to whether there actually is anything more to
things than what we can know of them. One could draw this distinction and then
argue that experience presents us with things as they are in themselves, or that the notion
of things in themselves is coherent but there does not turn out to be in reality anything
that falls under it. But in the second group of texts, Kant clearly indicates that he thinks
that there actually is an aspect of reality which we cannot cognize.52 He says that
[O]ur . . . cognition reaches appearances only, leaving the thing in itself as something actual for
itself, but uncognized by us. (Bxx)
Space represents no property at all of any things in themselves nor any relation of them to each
other, i.e., no determinations of them that attaches to objects themselves and that would
remain even if one were to abstract from all subjective conditions of intuition. (A26/B42)
52
Erich Adickes documents this point in detail. He says: ‘Was Kant an zahlreichen Stellen als notwendig
fordert und als selbstverständlich annimmt, is nicht der Begriff des Dinges an sich, sondern die extramentale
Existenz einer Vielheit uns affizierende Dinge an sich’ (What Kant assumes as self-evident and puts forward
as necessary in many places, is not the concept of things in themselves, but the existence outside our minds
of a multiplicity of things in themselves that affect us. Adickes 1924: 3, my translation). See also Willaschek
(2001: 225).
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
[S]pace is not a form that is proper to anything in itself, but rather . . . objects in themselves are
not known to us at all, and . . . what we call outer objects are nothing other than mere represen-
tations of our sensibility, whose form is space, but whose true correlate, i.e., the thing in itself, is
not and cannot be cognized through them, but is also never asked after in experience. (A30/B45)
What may be the case with objects in themselves and abstracted from all this receptivity of our
sensibility remains entirely unknown to us. We are acquainted with nothing except our way of
perceiving them, which is peculiar to us, and which therefore does not necessarily pertain to
every being, though to be sure it pertains to every human being. (A42/B59, my italics)
Even if we could bring this intuition of ours to the highest degree of distinctness we would not
thereby come any closer to the constitution of objects in themselves; . . . what objects may be in
themselves would still never be known through the most enlightened cognition of their
appearances, which is alone given to us. (A43/B60, my italics)
[E]verything in our cognition that belongs to intuition . . . contains nothing but mere
relations . . . But what is present in the place, or what it produces in the things themselves
besides the alteration of place, is not given through these relations . . . outer sense can also
contain in its representation only the relation of an object to the subject, and not that which
is internal to the object in itself. (B66–7, my italics)
[the mind] intuits itself . . . as it appears to itself, not as it is. (B68–9)
[A]ppearances are only representations of things that exist without cognition of what they
might be in themselves. (B164, my italics)
[W]e have to do only with our representations; how things in themselves may be (without
regard to representations through which they affect us) is entirely beyond our cognitive sphere.
(A190/ B235)
[An appearance is] a representation, the transcendental object of which is unknown. (A191/B236)
What matter is, as a thing in itself (transcendental object) is of course, entirely unknown to us.
(A366)
I grant by all means that there are bodies without us, that is, things which, though quite
unknown as to what they are in themselves, we yet know through the representations which
their influence on our sensibility procures us, and which we call bodies. (Proleg. 5: 289)
In fact, if we view the objects of the senses as mere appearances, as is fitting, then we thereby
admit at the very same time that a thing in itself underlies them, although we are not
acquainted with this thing as it is in itself, but only with its appearance, i.e., with the way
in which our senses are affected by this unknown something. Therefore the understanding,
just by the fact that it accepts appearances, also admits to the existence of things in
themselves, and to that extent we can say that the representation of such beings as underlie
the appearances, hence also of mere intelligible beings, is not merely permitted but also
inevitable. (Proleg. 5: 314–15)53
53
See also Discovery, where Kant says ‘no recourse remains but to admit that bodies are not things-in-
themselves at all, and that their sensory representation, which we denominate corporeal things, is nothing
but the appearance of something, which as thing-in-itself can alone contain the simple, but which for us
remains entirely unknowable’ (Discovery 8: 209). See also Groundwork 4: 451.
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
Opponents of the idea that Kant thinks there is some existing reality of which we
cannot have knowledge say that his view is merely that we cannot do without the
concept of a thing as is it is in itself: the concept of things in themselves is an
unavoidable posit of reason, but not a commitment to something actually existing.
As the above quotations show, these commentators will have to discount some of
Kant’s texts, to explain them away, or to regard Kant as inconsistent.
There is one passage in which Kant might be thought to assert that we do not know
whether there are things in themselves. Towards the end of the ‘Amphiboly’ section, he
says that ‘The understanding . . . thinks of an object in itself, but only as a transcendental
object, which is the cause of appearance’ and that it ‘remains completely unknown whether
such an object is to be encountered within or without us’ (A288/B344). This passage occurs
in the middle of a discussion in which Kant is arguing against supersensible objects. In
contrast, as we have seen, there are very many passages in which Kant says that things in
themselves ground appearances, and that we cannot know things as they are in themselves.
In the third group of texts, Kant indicates that the aspect of reality which we
cannot know is in some sense metaphysically ultimate or more fundamental than the
appearances we know, since it grounds the appearances we know, causes appearances,
and, notoriously, ‘affects’ us. Kant says:
The representation of a body in intuition . . . contains nothing at all that could pertain to an
object in itself, but merely the appearance of something and the way in which we are affected by
it. (A44/B61, my italics)
[W]e have to do only with our representations; how things in themselves may be (without
regard to representations through which they affect us) is entirely beyond our cognitive sphere.
(A190/B235, my italics)
[I]t . . . follows naturally from the concept of an appearance in general that something must corres-
pond to it which is not in itself appearance, for appearance can be nothing for itself and outside of our
kind of representation; thus, if there is not to be a constant circle, the word “appearance” must already
indicate a relation to something the immediate representation of which is, to be sure, sensible, but
which in itself, without this constitution of our sensibility (on which the form of our intuition is
grounded), must be something, i.e., an object independent of sensibility. (A251–2)
[T]he understanding . . . thinks of an object in itself, but only as a transcendental object, which
is the cause of appearance (thus not itself appearance). (A288/B344, my italics)
[T]he transcendental object that grounds both outer appearances and inner intuitions is neither
matter nor a thinking being in itself, but rather an unknown ground of those appearances that
supply us with our empirical concepts of the former as well as the latter. (A379–80, my italics)
How is outer intuition . . . possible at all in a thinking subject? . . . it is not possible for any
human being to find an answer to this question, and no one will ever fill this gap in our
knowledge, but rather only indicate it, by ascribing outer appearances to a transcendental
object that is the cause of this species of representations, with which cause, however, we have
no acquaintance at all, nor will we ever get a concept of it. (A393, my italics)
The sensible faculty of intuition is really only a receptivity for being affected in a certain
way with representations . . . which, insofar as they are connected and determinable in these
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
relations (in space and time) according to laws of the unity of experience, are called objects.
The non-sensible cause of these representations is entirely unknown to us. (A494/B522)
The cause of the empirical conditions of this progress, the cause, therefore, of which members
of it I might encounter, and also the extent to which I may encounter them in the regress, is
transcendental, and hence necessarily unknown to me. (A496/B524)54
As we will see, some commentators argue that Kant’s talking as if there is a way
things are in themselves and his talking of things in themselves as grounding or
causing appearances are inconsistent with his own restrictions on what we can know,
in particular, his restriction of the use of the categories, including that of causation, to
empirical knowledge, or to appearances. I respond to this worry in Chapters 3 and 10.
For now, we can simply note that Kant does in fact, frequently, say that things in
themselves ground appearances. An interpretation which makes the texts maximally
consistent will make sense of Kant’s saying that the things that we cognize have a way
that they are in themselves which grounds the way we appear to us, and which we
cannot cognize.
In closing this chapter, I note a few points about Kant’s claim that we cannot
cognize things as they are in themselves. Kant seems to assert that there is a way
things are in themselves and to deny that we can cognize things as they in themselves.
This may be thought problematic. There is concern that asserting both these claims is
inconsistent, or that the former is unmotivated if the latter is true: if we cannot know
things in themselves, why think that any such things exist? In my experience, this
worry quickly occurs even to undergraduate students when you introduce them to
transcendental idealism.
In response to this worry, first, it is important to see that Kant does not both claim
that there is a way things are in themselves and that we cannot know that there is a
way things are in themselves. As I read him, he starts with the claim that there is a
way things are in themselves: things have properties or natures which are independ-
ent of their relations to us and other things. He then argues that we are not able to
have any specific cognition of what the intrinsic natures of things are like. This
involves no contradiction.
Second, we should not exaggerate Kant’s account of the lack of our knowledge.
Kant does not deny that we can make any justified claims at all about things in
themselves; he denies that we can cognize things as they are in themselves. He
frequently asserts general or formal claims about things in themselves, such as the
claim that they ground appearances, and his view allows analytic propositions to be
54
In the Metaphysics Mrongovius Kant says ‘When we look upon the appearances, they all fit together
according to the laws of nature. But still all appearances also have a transcendental cause which we do not
know’, and that ‘There must . . . be a transcendental cause that contains the ground from which this
appearance arises. This cause is unknown to us; but because it . . . does not belong to the sensible world,
it also cannot be determined by other causes in it, consequently it does not stand under the laws of nature
or of the sensible world’ (LM 29: 861).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
true of things as they are in themselves.55 Rather than insisting on attributing to him
the view that we cannot make any knowledge claims at all with respect to things in
themselves, in the face of the fact that he does make some general claims, we should
take what Kant says about things in themselves to guide the way we understand his
claim that we cannot cognize them. Cognition, for Kant, has specific conditions,
including acquaintance, and cognition gives us a certain kind of specific, determinate
representation of things. He thinks that we have no acquaintance with things as they
are in themselves and no way of representing their specific natures. We do know
some general claims about things as they are in themselves, but we cannot cognize the
specific natures things have, as they are in themselves. As Ameriks explains, Kant
means to exclude ‘only positive determinate theoretical knowledge of things in
themselves’ (2003: 17 n25).
Third, Kant’s claim that there is a way things are in themselves appears unmoti-
vated only if read a certain way: if we start with the empirically real objects of our
knowledge and then postulate that there are, in addition to these, unknowable
noumena. But this is not how Kant presents his position. A striking feature of the
way in which Kant first starts discussing things in themselves is that he says virtually
nothing to explain or introduce the notion or to indicate that it requires explanation,
and he does not give any arguments for thinking that there are things in themselves.
A simple explanation of this is that Kant does not take it to be a technical or unusual
notion which requires introduction and definition: he simply starts with the idea that
there are things and that things have a way that they are independent of their
relations to other things, including us.56 Similarly, Locke says of primary qualities
‘We have by these an idea of the thing as it is in itself ’ (Essay, II, VIII, 23).57 Locke
opposes primary qualities to two kinds of relational qualities: qualities which are
understood in relation to us and qualities of things which are understood as powers
to affect other things. He assumes that things must have a way they are independent
of their relations to us and their powers to affect other things and speaks of this as the
way things are in themselves. It is possible to argue that at the most fundamental level
reality contains nothing non-relational: contemporary ontological structural realists
hold this. However, this is a position which requires argument, which some think to
be incoherent, and which is not, in my view, an intuitive starting point. As I read
them, both Kant and Locke do not take it to be an option—they simply assume that if
a thing exists it has a non-relational nature, a way that it is independent of its
55
There is reason to think that analytic propositions do not qualify as cognition, for Kant.
56
Ameriks (2006: 74–5) and Adickes (1924: 9) argue that the most straightforward account of why Kant
does not argue for things in themselves is precisely that he uses the notion in this way. As Ameriks says,
Kant starts with, rather than argues towards, the reality of things in themselves, and that ‘the very lack of an
argument by Kant shows his insight into the oddity of insisting that one must be had’ (2006: 74–5; cf.
127–8); Ameriks (2003: 23, 33). See also Willaschek (2001: 221).
57
As we will see in Chs. 9 and 10, the matter is considerably more complicated for Kant, since his
notion of things in themselves fills some of the traditional roles of primary qualities but not others.
TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 
relations to other things, including relations to perceiving subjects. They both use the
expression ‘things as they are in themselves’ without introduction, to refer to this
independent nature.58 Kant does not give a complete account of empirical knowledge
and then argue that, in addition, we need to posit non-spatio-temporal intelligibilia;
rather, he starts with the thought that there are things and that these things have a
way that they are, independently of their relations to other things, including their
relations to us. To understand Kant in this way does not require us to take him as
asserting that reality consists of unknowable, non-spatio-temporal noumena—a
thesis which, as we will see, he maintains could not be known by us and which he
explicitly says is not his view. At the same time, it does enable us to do justice to his
claim that there is a way things are as they are in themselves and that this grounds the
appearances of which we have knowledge.
Bird objects to the idea that a commitment to things in themselves could be a
starting point, saying:
To deny knowledge of things in themselves is certainly compatible with a belief in their existence,
but to deny such knowledge and assume that existence as a premise in a metaphysical system
must be unsatisfactory. How can a system seriously both accept an essential premise and at the
same time deny that we can have any knowledge of its truth? (Bird 2006: 553)
We can avoid the unsatisfactory position Bird describes if we see Kant’s starting point
not as a commitment to things in themselves understood as mysterious supersensible
entities but rather a commitment to things, understood neutrally,59 as well as to the
idea that things must have a nature that is independent of us. As I understand him,
Kant’s central concern in the Critique is not to oppose the Cartesian sceptic;60 he starts
by assuming that there are things. He then argues that our cognition of these things is
limited to mind-dependent appearances of them (to aspects of them which exist only in
relation to us) and that we cannot know them as they are apart from their mind-
dependent appearances—as they are in themselves. On this reading, what is radical in
Kant’s position is not the claim that there is a way things are in themselves, but the
claim that we cannot cognize this nature. Kant starts with the idea that there are things
and that things have a nature, a way that they are independently of their relations to
other things, including conscious subjects, and he never denies or questions this
starting assumption. He goes on to argue that we cannot cognize their natures as
they are independently of their appearing to us.
As we have seen, Kant repeatedly, throughout the Critique, presents his position in
terms of a distinction between things as they are independently of us and these same
58
This is compatible with there being relations between things as they are in themselves.
59
I take it that Bird would not disagree with this; he simply disagrees that there is more to such things
than what is presentable in experience. As I make clear in Chs. 2 and 3, I agree with his detailed rejection of
interpretations which see Kant as a noumenalist and a phenomenalist as well as those which see kant as
centrally concerned with rejecting Cartesian scepticism.
60
Bird (2006) gives detailed argument for this.
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
things as they appear to us. He says that the way things are in themselves grounds the
way they appear and that we cannot cognize things as they are in themselves. We
have also seen that Kant holds that what appear to us, spatio-temporal objects, are
transcendentally ideal, which means that although they are not literally in our minds
(empirically in us), their existence is not distinct from the possibility of our cognizing
them. My aim in the rest of this book is to present an interpretation which does
justice to all these features of the text. We need an account of a form of mind-
dependence which allows that the mind-dependent appearances given to us in
intuition are things which also have a way they are in themselves which we cannot
cognize. Before presenting my positive account of Kant’s position, I first, in the next
three chapters, discuss problems with the main competing interpretations.
2
Why Kant is Not a Phenomenalist
I Introduction
We have seen that there is clear textual evidence that Kant thinks that empirically real
objects depend on our minds in some sense: he is an idealist. Probably the historically
dominant reading of Kant’s idealism is that it is a kind of phenomenalism: that
Kantian empirically real objects are mental entities or constructions out of mental
entities, mental states, or mental activities. A classic example of an extreme idealist
interpretation is that of P. F. Strawson. He explains Kant’s transcendental idealism as
the doctrine that the physical world ‘only appears to exist, [that it] is really nothing
apart from our perceptions’ (P. F. Strawson 1966: 238; see also 240, 242–6).1 In this
chapter I argue that there are very strong textual and philosophical reasons for
thinking that Kant’s transcendental idealism is not a form of phenomenalism. I start,
in this section, by saying something about the range of interpretations I regard as
falling under the label of phenomenalism; in particular, I explain why I regard
many adverbial and intentional object interpretations as subject to the objections to
phenomenalist interpretations. In Section II, I argue that there are overwhelming
reasons to think that Kant does not have a phenomenalist or mentalist account of
appearances. However, as we saw in Chapter 1, there are strong textual reasons for
thinking that transcendental idealism involves a commitment to some form of
idealism. In the absence of an alternative account of the sense in which appearances
depend on our minds, it might be thought that the Critique is simply inconsistent.
Thus, my case against the phenomenalist reading must rest not only on the negative
arguments in this chapter alone, but also on the positive proposal presented in the
rest of the book, especially in Part Two.
I am using the term ‘phenomenalism’ loosely, to include the view that empirically
real, spatial objects are things that exist only in the mind, like Berkeleyan ideas, or
that they exist as constructions out of mental items or mental states or mental
1
For a sample of other phenomenalist interpretations, see Turbayne (1955), Bennett (1966: 23, 126),
Wilkerson (1976: 180–4), Guyer (1987), and Van Cleve (1999). There is also a distinguished line of
commentators who have argued in detail against any kind of phenomenalist reading of Kantian appear-
ances, including Bird (1962; 2006), Melnick (1973), Prauss (1974; 1971), Matthews (1982), Pippin (1982),
Allison (1983; 1973), Langton (1998), Collins (1999), and Ameriks (2006: Ch. 3).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
activities, or that they supervene on properties of mental states.2 As such, I use the
term interchangeably with ‘mentalism’ and ‘extreme idealism’. It is important to note
that phenomenalists can appeal to possible as well as actual mental items and also to
constructions out of mental items. Excluding phenomenalist interpretations of
Kantian appearances requires arguments against sophisticated phenomenalist views
that see Kantian appearances in terms of constructions out of facts about possible
mental entities, mental states or activities, or ways of mentally being. A construction
out of mental items is not a mental item and it can have properties which the mental
items lack; for example, a construction out of fleeting items need not be fleeting. This
kind of sophisticated phenomenalism or mentalism represents a harder case for those
who want to reject phenomenalist interpretations.
While phenomenalist interpretations of things as they appear to us have been
associated with what has been called a ‘two-world’ interpretation of Kant’s distinc-
tion between things in themselves and appearances, not all the interpretations I count
as phenomenalist are straight-forwardly ‘two-worlds’ interpretations. Some phenom-
enalist interpreters deny that they understand Kant as being committed to the
existence of more than one kind of object and therefore deny that their interpret-
ations deserve the ‘two-world’ or ‘two-object’ label. Van Cleve, for example, argues
that there are not two realms of objects, precisely because appearances are virtual
objects whose only existence is phenomenalistic: ‘appearances do not, strictly speak-
ing, form a second class of existents alongside things in themselves’ (Van Cleve 1999:
150). Similarly, Guyer denies that he has a ‘two-world’ interpretation in one sense: he
argues that Kant ‘does not postulate a second set of ghostlike nonspatial and non-
temporal objects in addition to the ordinary referents of empirical judgments’ (Guyer
1987: 334). However, he says that Kant ‘does something just as unpleasant—namely,
degrade ordinary objects to mere representations of themselves, or identify objects
possessing spatial and temporal properties with mere mental entities’ (Guyer 1987:
335; see also 2007: 12). While both deny that there are two sets of entities, Guyer sees
Kantian appearances as ‘mere mental entities’, and Van Cleve says that they are
nothing over and above the existence of subjects with certain mental states.3 As I am
using the term, phenomenalist interpreters need not think of appearances as distinct
mental objects; I argue that we must reject all readings which mentalise Kantian
appearances, whether or not this involves positing distinct mental objects.
2
So I am not using the term in the strict sense in which it was used in early twentieth-century
philosophy, to refer to a theory of meaning according to which object statements are translatable into
statements describing mental states.
3
Despite the fact that phenomenalist readings of things as they appear to us may not regard them as a
second set of entities, we can still see why commentators have wanted to group such interpretations under
the ‘two-world’ heading: mental states, mental activities, or ways of mentally being are not aspects or
features of the things which have a way they are independently of our being able to experience them. This
means that, on this view, there is no sense in which appearances are an aspect of the very things which have
a way they are in themselves, so the view is committed to appearances having a clearly distinct ontological
status from the things of which they are appearances.
WHY KANT IS NOT A PHENOMENALIST 
Some interpretations try to avoid understanding Kantian phenomena as distinct

mental objects by comparing Kant’s position with adverbial and intentionalist
analyses of perception. The rest of this section explains why I regard standard
versions of this move as versions of phenomenalist interpretations. It is surely no
coincidence that interpretations of Kantian appearances have gone down many of the
routes explored by philosophical accounts of perception, as one of the key problems
in spelling out Kant’s idealism is making sense of the (extremely tricky) language of
perceptual appearing. The motivation for introducing intentionally inexistent objects
and adverbial analyses of appearing in interpretations of transcendental idealism is
often related to the motivation for this move in philosophy of perception: in both
areas the desire is to avoid postulating mental objects as the direct objects of
perception.4 However, at least with respect to some versions of these theories,
appealing to them to interpret Kant avoids mental objects but still results in a
phenomenalist-type interpretation, because the perceptual states with which Kant-
ian appearances are compared are still characterised entirely subjectively, mentally,
or internally. Intentional and adverbial theories of perception reject the classic form
of sense-data representative realism which has us perceiving mental entities of some
sort but, at least on some versions of these theories, they still try to analyse
perception in terms of a combination of something mental—something charac-
terised entirely subjectively—and an external cause which is extrinsic to the indi-
viduation of the relevant subjective state.5 This means, as with Guyer’s and Van
Cleve’s views, that the effect of (at least some) intentional object and adverbial
interpretations of Kantian appearances is to keep appearances in the mind without
making them into mental entities. I say something briefly about each of the adverbial
and intentional interpretations.
Adverbial theories of perception characterise sensory qualities in terms of modi-
fications of ways of mentally being or mental activities, together with external causal
correlates of these. They differ from traditional sense-data indirect realism in that the
subjective (mental) element is a kind of mental act rather than a mental object, but,
despite not postulating a set of mental objects which represent the objects perceived,
they reject the direct realist idea that there are psychological states which essentially
4
Adverbial theories of perception hold that the role of the grammatical object of experience is not to
indicate an actual mental object but simply to characterise the kind of experience being attributed to the
subject. See Howard Robinson (1994) and Putnam (1994: 453–4). For discussion of intentional object
readings of transcendental idealism, see Aquila (1983), Cummins (1968), and Pereboom (1988). For an
adverbial reading, see Baldner (1990; 1988). Here I am considering those accounts of Kantian appearances
which try to explain them through an analogy with adverbial accounts of perception. An alternative
‘adverbial’ approach to transcendental idealism is that of Prauss (1974), who argues that Kant’s phrase ‘in
itself ’ is meant to be used adverbially, to characterise a way of considering things, rather than adjectivally,
to characterise things, and argues that this leads to deontologising Kant’s distinction. I discuss deflationary
views in Ch. 4.
5
See Crane (2005: 14) for helpful discussion.
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
involve the objects perceived.6 Emphasising mental states or activities rather than
mental objects is entirely consistent with still understanding perception in terms of
inner states which do not intrinsically involve the objects perceived as one compo-
nent and an appropriate causal relation between the mental state/activity and the
objects as another.7 This does not do justice to the direct realist idea that the object
itself is present in the perceptual state.8 Interpretations of Kantian appearing that
appeal to this kind of adverbial theory—that explain the status of Kantian appear-
ances as reducing to the existence of certain mental states or activities—will keep
things as they appear to us firmly in the mind, despite not making them a second set
of entities. I do not argue here that this analysis necessarily applies to all adverbial
accounts of perception—more complex theories are not my topic here9—and cor-
respondingly I cannot say that it is true of all adverbial interpretations of Kant.10 My
point is simply that to the extent that adverbial theories of perception understand
perception in terms of merely mental states with appropriate external correlates, the
interpretations of Kantian appearing which draw on these accounts of perception will
keep appearances in the mind: they mentalise appearances without making them
mental objects.
The situation is more complicated when it comes to those who appeal to inten-
tionalist accounts of perception. At least some of those who want to explain Kantian
appearances by appealing to the notion of intentional objects, or to the intentional
content of perception, do this in order to deny that appearances are things which
have a merely mental existence.11 There is no simple or quick way to assess this move
because intentional theories of perception differ considerably and are controversial,
as is the status of intentional objects.12 This also means that rather than explaining
the status of Kantian appearances, appealing to intentional objects leaves the
explanatory work still to be done, in explaining the latter. I will say something briefly
6
See Howard Robinson (1994: Ch. 7).
7
To this extent, they may accept that there is a subjective (mental) element common to perception and
hallucination: the mental states or activities.
8
As Hilary Putnam argues, it is a mistake to think that ‘all one has to do to be a direct realist . . . about
visual experience, for example, is to say “[w]e don’t perceive visual experiences, we have them” ’ (1994: 453).
9
See Howard Robinson (1994) for an overview of the problems of both views, and Jackson (1975) for
problems with the adverbial account.
10
For example, I am not sure about Baldner’s interpretation in this respect. He suggests an adverbial
analysis of Kantian appearances, saying that adjectives describing appearances ‘need to be recast as adverbs
or adverbial phrases describing our experiences of objects’ (Baldner 1990: 10). Here, everything depends on
how ‘our experiences’ is to be understood: if this is something that can be characterised purely subjectively,
as it usually is in the adverbial analysis of perception, then we still have a view which keeps appearances in
the mind. On the other hand, if this is not the idea, then the parallel with the adverbial theory of perception
may be misleading, and we need to hear much more detail about how the adverbial analysis is supposed
to work.
11
See Adams (1994: 219).
12
Howard Robinson argues that ‘intentional inexistence’ ‘is more of a name for a problem than it is a
solution to one’ (1994: 164).
WHY KANT IS NOT A PHENOMENALIST 
about two of the ways in which the notion of intentional objects of perception may be
appealed to in order to explain the status of Kantian appearances.
A key feature of intentional objects as they are standardly understood, as objects of
thought, is that they may not exist: I can think about things which do not exist.13 This
has been thought by some to enable an intentional account of perception to respond
to the traditional arguments against direct realism based on hallucination and non-
veridical perception: they point out that while something which does not exist can be
the object of my thought, this does not imply that in the case in which I think about
something which does exist, the object of my thought is anything other than the
object itself. Similarly, it is suggested that the fact that I can, for example when
hallucinating, represent things which do not exist, does not show that, in the case of
veridical perception, the direct object of perception is anything other than the
physical object perceived. When you are thinking about something which does not
exist, the object of thought is merely represented; similarly, it may be argued that we
can account for the occasional illusory character of perceptual consciousness in terms
of representational mental states or intentional objects. The idea is that hallucination
involves objects which are merely intentional,14 whereas in cases of actual perception
the intentional objects of perception are the external objects themselves. Whether or
not this is a way in which a direct realist about perception can deal with the argument
from hallucination,15 it is not clear that it can help us characterise Kantian appear-
ances in a non-phenomenalist way. This is because there are two ways in which the
analogy between Kantian appearances and intentional objects or intentional content
can be understood, one of which will lead to seeing Kant as a phenomenalist, and the
other to seeing him as a straightforward realist. On the one hand, Kantian appear-
ances could be understood in terms of an analogy with the merely intentional objects
in terms of which intentional theories of perception characterise hallucination. On
the other hand, the point of the analogy with intentional theories of perception could
be to compare Kantian appearances with the mind-independent objects which the
theory holds to be the intentional (direct) objects of perception in cases of veridical
perception. I will discuss each option briefly.
The first possibility is for Kantian appearances to be compared with merely
intentional objects, or representational mental states which can be characterised
entirely subjectively. As we have seen, in the theory of perception at least part of
13
Intentional objects have also been appealed to by accounts of perception which analyse perceptual
states in terms of beliefs. Since there are many objections to assimilating perception to belief (see Howard
Robinson 1994: 66 and Smith 2002: 42–7, 94–121), I assume here that the representational mental states
appealed to by intentional theories could be understood in terms other than belief.
14
This view need not require that merely intentional objects be understood as some kind of entity—
Smith says that ‘some objects do not exist’ (Smith 2002: 234). As Smith explains the view, talk of ‘merely
intentional objects is, however, an invocation of “nothing” only in an ontological sense. We do, indeed,
need an ontologically reductive account of intentionality. Non-existent intentional objects supervene on
intentional experiences’ (Smith 2002: 244).
15
See Smith (2002: Ch. 9).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
the point of introducing intentional objects is to enable us to characterise hallucin-

ations by appealing to representational mental states rather than mental entities,
while asserting that in the case of non-hallucinatory perceptual experience the
intentional objects of the experience are the external objects themselves. This
means that to compare Kantian appearances with merely intentional objects is to
keep appearances in the mind: we simply have a more sophisticated version of a
phenomenalist interpretation. An example of this kind of view is provided by
Aquila’s interpretation. He says: ‘[t]o exist as an appearance is to exist in what
I shall call a “phenomenalistic” sense; it is to exist, in a certain sense, merely
“intentionally” ’, adding that it is ‘not strictly speaking possible to suppose that the
very same thing might exist in both of the senses in question’ (Aquila 1983: 89–90).
Thus, this kind of intentionalist interpretation can be grouped with the phenomen-
alist interpretations I argue against in the next section.
However, some of those who appeal to intentional objects to explain Kantian
appearances have in mind something different from intentionally inexistent objects
of thought. On our second way of reading the analogy, the intentional objects in
terms of which Kantian appearances are explained are supposed to be those to which
the mind is directed in experience—the ordinary physical objects of experience.
Adams appeals to this sense of intentional objects in explaining both Leibniz and
Kant.16 Adams denies that phenomena (his primary concern is with Leibnizian
phenomena but he thinks it applies to Kant as well) are mental images and says
they should instead be understood in terms of intentional objects, where these are
things that are objects for minds or objects of perception, or the ‘representational
contents of our perceptions—what appears to us’.17 In the sense in which intentional
objects are not intentionally inexistent objects of thought, but rather ‘what appears to
us’, they can, and do, feature in straightforwardly realist accounts of perception.
A realist thinks that external objects can be and often are objects for minds, or objects
of perception, and that the ‘representational contents of our perceptions’ are features
of the way things mind-independently are. But if Kantian appearances are compared
to the independent things which are the objects of veridical perception, then the
appeal to intentional objects does not explain any sense in which appearances are
mind-dependent.
In summary, on the one hand, appearances could be compared with the merely
intentional objects, or representational mental states, involved in hallucinatory
perceptual events. In this case, appearances would once again be equated with
something which is characterised entirely mentally, which will result in a (sophisti-
cated) phenomenalist reading of transcendental idealism. On the other hand, appear-
ances could be compared with the actual objects which are the intentional objects of
perception in normal veridical perception, but now the comparison with intentional
16 17
Adams (1994: 219). Adams (1994: 220).
WHY KANT IS NOT A PHENOMENALIST 
objects fails to capture any sense in which Kant’s position is idealist or any sense in
which empirically real objects have a necessary connection with possible perception.
This argument is not intended to exclude the possibility of alternative intentional
object interpretations.18 It simply shows that appealing to the notion of intentionality
is the beginning, rather than the end, of an explanation of Kantian appearances, and
that at least some intentionalist and adverbialist interpretations can be grouped
together with phenomenalist readings and will be vulnerable to objections to the
latter, which I discuss later. The objections I develop in the next section are not
merely to views that take Kantian appearances to be mental objects but to all
mentalised readings of appearances. As in philosophical accounts of indirect percep-
tion, some interpreters of Kantian appearances have gone to great lengths to avoid
postulating mental objects, while continuing to understand Kant’s empirically real
objects in terms of merely mental, inner states; this, I argue, is inconsistent with the
Critique. Putnam says that the false belief that perception must be analysed in terms
of a mere affection of a person’s subjectivity by external things ‘is at the root of all the
problems with the view of perception that, in one form or another, has dominated
Western philosophy since the seventeenth century’ (1994: 454). It seems to me that it
is also a cause of problems with interpreting Kant.
II Seven Reasons Why Kant is Not a Phenomenalist

Since the case against phenomenalist interpretations has been made by a number of
commentators, my discussion of some objections will be relatively brief. At the same
time, the historical dominance of phenomenalist interpretations, their recent revival,
and the textual evidence which apparently supports them make it necessary to bring
these arguments together to show the strength of the case.
(1) Kant claims that his notion of appearances implies things which appear. While
phenomenalist interpreters appeal to Kant’s use of the word ‘representations’, their
opponents can appeal to the fact that Kant says repeatedly that his notion of
appearance implies the existence of the thing which appears and that it would be
absurd to suppose otherwise. He says:
[E]ven if we cannot cognize these same objects as things in themselves, we must at least be able
to think them as things in themselves. For otherwise there would follow the absurd proposition
that there is an appearance without anything that appears. (Bxxvi–xxvii, my italics)
[I]t . . . follows naturally from the concept of an appearance in general that something must
correspond to it which is not in itself appearance, for appearance can be nothing for itself and
outside of our kind of representation; thus, if there is not to be a constant circle, the word
“appearance” must already indicate a relation to something the immediate representation
18
For example Pereboom’s (1988) account.
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
of which is, to be sure, sensible, but which in itself, without this constitution of our
sensibility . . . must be something, i.e., an object independent of sensibility. (A251–2, my italics)
The understanding therefore, by assuming appearances, grants also the existence of things in
themselves, and thus far we may say that the representation of such things as are the basis of
appearances . . . is not only admissible but unavoidable. (Proleg. 5: 315, my italics)
The reason that these passages count against a phenomenalist interpretation is that
virtual objects, phenomenalist objects, or Berkeleyan objects simply do not entail the
existence of things of which they are appearances. As Langton puts it: ‘what could be
easier than to suppose that there could be an appearance without there being
something that exists in itself, apart from our state of mind, on the phenomenalist
interpretation?’ (1998: 22). Berkeleyan objects, as collections of ideas, require some-
thing other than the ideas—the subject who has the ideas—but they do not imply the
existence of things of which the collections-of-ideas are appearances. Similarly,
someone who believes that objects as we know them are collections of sense data
or ideas may plausibly think that there is something external to subjects which is in
some way responsible for the order and existence of the ideas or sense data (for
example Berkeley’s God), and may thus be committed to something like the ‘two-
world’ view of transcendental idealism.19 However this is an abductive inference
rather than an analytic entailment of there being appearances, and there would be no
need to call the sense data appearances of this cause. It would be odd to say that
Berkeley’s empirically real objects are appearances of God. This is closely related to a
point we saw in looking at the textual evidence concerning things in themselves: Kant
very frequently speaks of the things as they appear to us and these same things as they
are in themselves. He talks about things in so far as we are acquainted with them as
opposed to in so far as we are not acquainted with them (Bxx), about things as objects
of experience and the very same things as things in themselves (Bxxvii), and about
considering the same objects in two ways (Bxviin; A38/B55; A39/B56). It is of course
possible for a phenomenalist to regard things that exist in our minds (or as con-
structions out of what is in our minds) as representations of some other things, but
they are not aspects of these things, and there is no sense in talking about these
representations as the same things as the things which also have a way they are
in themselves.
(2) Unlike Berkeley, Kant appears to distinguish between primary and secondary
qualities within the empirically real world. In the Aesthetic, Kant distinguishes space
from qualities like colour, saying that colour does not belong to the empirically real
object and is merely an alteration of the subject (B45). In contrast, one of Berkeley’s
first moves is to deny that secondary qualities have a different kind of reality to that
which pertains to primary qualities. This counts against phenomenalist interpret-
ations because it can be argued that Kant could not hold that objects exist as mental
19
See Foster (2000) for a defence of this kind of view.
WHY KANT IS NOT A PHENOMENALIST 
items but do not have the properties they have in sensation.20 However, this
argument may not count against sophisticated phenomenalism, which, unlike the
cruder view which identifies appearances with actual mental items, need not
attribute to objects-as-constructions-out-of-mental-items all the properties which
are had by the mental items. A further problem is that Kant’s view of primary and
secondary qualities is unclear and controversial, and the distinction is one to which
he devotes very little time, so it cannot provide a conclusive basis for any inter-
pretation. In a note a few pages after the claim just considered (B45), he says that
‘the predicates of appearance can be attributed to the object in itself, in relation to
our sense, e.g., the red colour or fragrance to the rose’ (B69–70n), suggesting a
different view of colour from that at B45. Finally, while Kant’s primary/secondary
quality distinction can be used to argue against phenomenalist interpretations,
different conclusions can also be drawn from it. Wilson, for example, can be
read as attributing to Kant a kind of phenomenalism according to which what is
empirically real is a scientific construction which accords no reality to the manifest
image (i.e. objects of ordinary sense experience as sensed) (Wilson 1984: 160–70).21
I discuss this issue in Part Three.
(3) Empirically real objects are public and are in one public space. Further, Kant
has a clear distinction between what is in the mind and what is out of it in terms of
what is accessible only in inner sense as opposed to what is presented in both inner
and outer sense. This suggests that empirically real objects cannot be literally in the
mind. Collins presents this kind of objection to phenomenalist readings, arguing that,
for Kant, ‘the very object that one perceiver detects can also be detected by another’
(Collins 1999: 1, 18, 107–2022). This is a central part of Kant’s position, and it rules
out understanding Kantian appearances as private mental entities.
One way a Berkeleyan phenomenalist can respond to this is by claiming that
empirical objects do exist entirely independently of us (of individual perceivers)
because they exist permanently in the mind of God. In this case, the point of
Berkeleyan idealism is not to say that objects depend on our minds—on our perceiv-
ing them—but rather that their nature is somehow essentially mental (perhaps they
20
See, for example, Allison (1973), Wilson (1984: 161, 165), and Langton (1998: 144, 142–7).
21
Wilson says that Kant has a kind of phenomenalism because his empirically real world is a reductivist
scientific construction, elaborated within the framework of the merely ideal ‘forms of intuition’ (1984: 170).
In other words, on her view, Kant denies that what is given in sensory experience is part of what is
empirically real, but still sees the physical world as a mental construction in some sense. The problem with
this interpretation is Kant’s repeated claim that there is nothing empirically real without a connection to
possible perception (A218/B266; A225/B272; A376; A493/B521). I will argue, in Part Three, against a
reading of Kantian appearances which denies the reality of the manifest image. On the contrary, for Kant,
only what is possibly manifest is empirically real. This issue is complicated by the details of what Kant
means by possible experience, which I discuss in the final section of Ch. 6 as well as in Ch. 9.
22
See also Allison (1973), Melnick (1973: 137), Ameriks (1982a: 267), Matthews (1982: 134), and
Rosefeldt (2001: 265–6; 2007: 168).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
are made up of mental stuff ) and is dependent on a mind, though not ours. This kind
of Berkeleyan position is clearly not Kant’s view, because Kant explicitly denies that
we could know that physical objects are ultimately made up of mental or spiritual
stuff, and whatever mind-dependence his position involves, it is dependence on our
minds that is at issue. In the Paralogisms, Kant argues that our ignorance of the
nature of things in themselves rules out our asserting that the ultimate nature of
empirical objects is essentially mental, and rules out our having grounds for saying
that mental and material stuff are, ultimately, different kinds (A359; also A360). At
the level of empirical reality there is a clear difference between what is mental (in the
mind) and what is in space. (Thoughts cannot be intuited among outer appearances
(A357).) At the level of things in themselves, we do not know enough to know
whether the mental and the material are fundamentally different. Thus, material
objects are not regarded by Kant as consisting of mental stuff or of ideas at either the
empirical or the transcendental level. This means that his idealism cannot be
captured by the kind of Berkeleyan view which holds that physical objects exist
outside of individual finite minds but are ideal in the sense that they are made up of
mental stuff.
An alternative way for the phenomenalist to account for the public nature of
objects is to appeal to possible perceptions; these can be public in the sense that we
could all have them. Mill, for example, says that ‘the permanent possibilities [of
sensation] are common to us and to our fellow-creatures; the actual sensations are
not’ (1865: 182). Similarly, the phenomenalist can account for the sense in which
objects are in space in terms of ways in which some of our ideas are ordered. It is part
of the phenomenalist’s position that everything we want to say about physical objects
in space can be captured by claims about mental states, so the phenomenalist will be
able to find a way to capture what Kant says about objects being in space. However,
the possibility of a phenomenalist’s carrying out such a translation cannot be enough
to establish that this is how Kant intends his claims to be understood (any more that
this possibility shows Locke to be a phenomenalist) and it is far from a straightfor-
ward reading of the text. Kant never says that what it is for an object to be spatial is
for there to be ideas which have certain sorts of coherence relations and that what it is
for an object to be public is for there to be the right sorts of relations between my
ideas and yours. He simply says that the public objects that we experience are in one
space and are immediately given to us as such.
A further difference between Kant’s account of the spatiality of objects and
Berkeley’s is that Berkeley thinks that what is presented to us in perception is not
itself three-dimensional and that three-dimensional spatial properties are something
our minds construct, as a kind of interpretation, on the basis of the ideas we are
presented with. Kant, in contrast, does not think that we are experientially presented
with things which lack spatiality (although he does think that our minds are
responsible for the spatiality of perception). He thinks that the particulars we
perceive are extended, shaped, and spatially located. And he also thinks that space
WHY KANT IS NOT A PHENOMENALIST 
is not constructed, as it is for Berkeley, but is directly given (intuited).23 Finally, he

thinks that the way we represent space and time gives us the structure within which
we can experience objects and sets limits on what objects we are able to construct.
Since space and time constitute the framework that enables us to have empirical
intuitions, they cannot be a construction out of what is given in intuition.
(4) For Kant, very small things, very distant things, and entities posited by science
can all be part of what is empirically real, despite the fact that we cannot actually
perceive them.24 He talks about stars that exist even if no one will ever perceive them
(A496/B524), and magnetic force, for Kant, is part of appearances, but it is not
something we perceive—we lack the requisite sense apparatus to perceive it. He says
that the ‘crudeness’ of our senses ‘does not affect the form of possible experience in
general’ (A226/B237). This suggests that Kant is not a Berkeleyan idealist who
reduces physical objects to ideas or perceptions. However, this point is tricky. On
the one hand, Kant seems to admit as part of what is empirically real things which are
too small or too far away for us to perceive, or for which we do not have the requisite
sensory apparatus. On the other hand, he thinks that the empirical reality of objects
is not separable from their having a connection to a possible perception and that
what is not part of the extent of possible experience is not empirically real. Exactly
what this connection amounts to is a delicate point, which I discuss in more detail
in Chapters 6 and 9. The dependence of the empirically real on a connection to
possible experience gives us, it seems to me, clear reason to see Kant as an idealist,
but the fact that Kant allows things which we cannot actually perceive to be part of
what is empirically real counts against seeing Kant’s idealism as a phenomenalistic
idealism. Berkeley denies that quantities less than the minimum sensibile are
possible (1710: no. 132, p. 119), but Kant says that ‘perception . . . too weak in
degree to become an experience for our consciousness . . . still belong[s] to possible
experience’ (A522/B550). This is a stronger argument than that based on the
secondary quality distinction, since while Kant’s view of secondary qualities may
be unclear, there is no doubt that he attributes existence to empirical objects which
we cannot actually sense.
A possible phenomenalist response to this is to point out that while Berkeley
denies the reality of anything less than the minimum sensible not all phenomenalists
have to do so. Here the distinction between those who see empirical objects as
actual mental items and sophisticated phenomenalists who see them as construc-
tions out of mental items is crucial. The latter can characterise things which are too
far away or too small for us to see in terms of possible rather than actual percep-
tions and in terms of constructions out of perceptions; such a reading might be
23
See Justin (1974) and Falkenstein (1995) for helpful discussion. Kant thinks conceptually governed
synthesis is needed for us to cognize space, but not for it to be immediately presented to us. This is
discussed in Chs. 8 and 11.
24
See Langton (1998: 144–5) for a development of this objection.
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
thought to be supported by Kant’s claim that the actual existence of an empirically

real object requires some connection with possible perceptions (A218/B266; A225/
B266; A376; A493/B521). The phenomenalist can argue that, for example, the
existence of magnetic force can be explained in terms of constructions out of
perceptions of other things. However, this is not how Kant accounts for it. As
Allison points out, Kant does not explain the reality of magnetic force through
appeal to counterfactuals about what sensations we would have had under certain
conditions or in terms of constructions out of what we do in fact sense (2004: 41).
On the other hand, Kant does clearly say that what is empirically real requires a
connection to an actual perception, so to dismiss the phenomenalist reading, we
need to give an alternative account of this.
(5) A central part of Kant’s account of empirical reality is established by his
arguments in the Analogies for the claims that empirically real objects are made
up of stuff that exists at all times, they exist unperceived, and they stand in causal
relations. Reconciling this with a phenomenalist interpretation is far from
straightforward. In the first and third Analogies, it is clear that Kant thinks
that (empirically real) substance exists unperceived. In the first Analogy, he
says that ‘the real in appearance [substance] . . . as the substratum of all change
always remains the same . . . its quantum in nature can also be neither increased
nor diminished’ (A181/B225), that ‘[o]nly in that which persists are temporal
relations possible’, and that ‘we can grant an appearance the name of substance
only if we presuppose its existence at all time’ (A185/B228). He contrasts our
changing perceptions with the permanence of substance, saying that ‘our appre-
hension of the manifold of appearance is always changing’ (A182), and that this
is why it must be grounded by something which always exists. In particular, it
must be grounded in
something lasting and persisting, of which all change and simultaneity are nothing but so
many ways (modi of time) in which that which persists exists. (A182/B225–6)
Kant argues for something which endures through time and he contends that
perceptions are not like this. In the third Analogy, Kant says that ‘things are
simultaneous insofar as they exist at one and the same time’ (A211/B258), and that
when one thing is coexistent with another, the former exists while we are looking at
the latter—i.e. the former exists unperceived. Kant contrasts the continued existence
of simultaneously existing objects with our successive perceptions of simultaneously
existing objects; at any given time, only one of these perceptions is present in the
subject, and the others are not (B257). Unlike our perceptions of things, things exist
when we are not looking at them.
In the second Analogy Kant maintains that there really are necessary connections
between things as they appear. As Hume argues, the necessary connections we think
of as existing between objects could not be the regularity relations between ideas.
Kant argues that there must be genuine causal relations (necessary connections)
WHY KANT IS NOT A PHENOMENALIST 
between appearances, and this suggests that things as they appear to us are not
mental items or properties of mental items.25
In response to these points, it might be said that the sophisticated phenomenalist
account holds that everything we want to say about the persistence, existence
unperceived, and causal relations between objects can be captured by relations
between possible sensations or constructions out of mental states.26 This means
that insisting that Kant thinks of empirically real objects as external, public, persist-
ing, existing unperceived, and as standing in necessary causal relations does not show
that he could not be a phenomenalist. But the fact that, rightly or wrongly, phenom-
enalists think their view can accommodate the central claims Kant makes about
external objects cannot be taken to show that Kant is a phenomenalist—just as the
fact that the phenomenalist thinks we can restate everything Locke wants to say
about physical objects in terms of constructions out of sensations does not make
Locke a phenomenalist. Kant explicitly says that he is presenting a position which
contains some kind of idealism, but he also explicitly says that substance is perman-
ent and exists unperceived, and he does not say that the permanence of substance
must be understood as a construction out of our possible perceptions. He does not
say that we must regard our perceptions as if they represented things which persist,
exist unperceived, and are in necessary connections. Nor does he say that these
principles are rules for constructing objects. Rather, he argues that the empirically
real, spatial objects of our knowledge exist unperceived (which he says is unlike our
perceptions of objects which are there only when we are perceiving them), persist
through time, and are made up of stuff which persists through all time. He does not
explain the existence of unperceived objects by talking about the sensations we would
have had if we had been perceiving them. In contrast, Mill, for example, makes it
explicit that he accounts for unperceived existence in terms of possible sensations. He
says that what it means to say that external objects exist unperceived is that there are
certain sensations we would have under certain circumstances. We appeal to ‘sensa-
tions that are not in our present consciousness, and individually never were in our
consciousness at all, but which in virtue of the laws to which we have learnt by
experience that our sensations are subject, we know that we should have felt under
supposable circumstances, and under these same circumstances might still feel’ (Mill
1865: 178–9). We find nothing like this in the Critique.
Phenomenalists think that they can account for everything we want to say about
objects as public, spatial, and existing unperceived. If we approach Kant’s text with
the assumption that phenomenalism is the only way to take seriously Kant’s idealism
(the claim that appearances are representations which do not exist apart from a
connection to a possible perception), we can then use phenomenalist analyses to
account for Kant’s view that appearances are spatial, public objects which exist
25 26
See Allison (1973: 53). See, for example, Mill (1979: 178–9).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
unperceived. But this analysis will be something which is read into the text, not
something Kant himself ever says. Since Kant does not present his principles as
principles that must be used for constructing an objective world out of sense data, but
rather as principles that are true of empirically real objects, we should start by looking
for an interpretation of his idealism that fits with this.
Ironically, two commentators who argue that no phenomenalist reading of
appearances is compatible with the Analogies, and who are also broadly sympathetic
to the arguments of the Analogies, hold extreme idealist positions on appearances:
Van Cleve (1999) and Guyer (1987). Both consider a reading which mentalises
appearances to be incompatible with Kant’s arguments in the Analogies, but they
take this merely to show that Kant is inconsistent. Van Cleve says that the first
Analogy is a reasonable argument, but not for an idealist:
logical constructions are precisely modes and not substances—they are adjectival on the
entities out of which they are constructions. So, it appears that for Kant, nothing in the
world of space and time qualifies as a substance, and there can be no hope of establishing
the First Analogy in its intended sphere. (Van Cleve 1999: 120)
Similarly, he says that the most plausible interpretation of the Second Analogy is at
the cost of inconsistency with a phenomenalist interpretation of Kant (Van Cleve
1999: 128). The inconsistency between the Analogies and phenomenalism provides
very strong evidence that Kant is not a phenomenalist. These are two of the core
projects of the Critique: Kant’s distinction between things in themselves and things as
they appear to us and his project of describing (what he sees as) the necessary
structure of the empirical world. An interpretation of Kant’s idealism which makes
them inconsistent must be a last resort.
(6) We do not cognize mental states as they are in themselves. This central
Kantian claim creates severe difficulties for a phenomenalist reading because such
a reading must combine an analysis of appearances in terms of constructions out of
mental states with the claim that we do not know anything about subjects and mental
states as they are in themselves. Compare this to the standard readings of Locke and
Berkeley. As the standard story has it, Locke thinks we have direct and immediate
knowledge of the contents of our minds, our ideas, which represent the world to us.
Berkeley agrees that we know the contents of our minds, our ideas, and then argues
that we do not know anything other than what exists in our minds. Both start from
the view that we know our mental states. Similarly, a phenomenalist like Mill starts
with something he takes us to have direct access to and knowledge of: mental states.
The phenomenalist then argues that all it is for there to be objects is for us to have
certain mental states, or for certain constructions out of mental states to be possible.
Kant, in contrast, thinks that we do not know either objects or our minds as they are
in themselves. This means that on Kant’s account we have no knowledge of the
material on which the phenomenalist construction would be based. And in addition
to thinking that we do not cognize mental states as they are in themselves, Kant
WHY KANT IS NOT A PHENOMENALIST 
thinks that our cognition of our own mental states requires immediate awareness of
objects outside us. He thinks that cognition of outer objects is prior to cognizing our
mental states and we could not have the latter without the former.
If Kantian appearances were constructions out of mental states they would have to
be either constructions out of empirical mental states or constructions out of
noumenal mental states—mental states as they are in themselves. The former is
straightforwardly false: for Kant, empirical objects are not dependent on empirical
mental states. The latter is problematic because we do not know anything about
subjects as they are in themselves or mental states as they are in themselves. The
phenomenalist interpretation, as Rosefeldt nicely sums up the situation, has to say
that objects are constructions out of the noumenal correlation of mental states of which we are
not conscious and this leaves one with a strange kind of phenomenalism that makes empir-
ically accessible objects logical constructions out of completely unknown data. Moreover, it is
unclear why one should call such data “representations” or “mental states”, because we seem to
have no more right to characterize things in themselves as “mental” than to call them
“physical”. (Rosefeldt 2001: 266)27
Since we do not know what noumenal ‘mental states’ are, we do not know what is
meant by the claim that the existence and nature of empirically real objects consist in
subjects as they are in themselves having certain noumenal ‘mental’ states. In
contrast, in Berkeley’s account, the point of the reduction of physical objects to
ideas is to avoid positing something unknowable and indescribable as the real object
of experience and to appeal instead to something we all directly experience. To
translate the phenomenalist view into Kantian terms, we need a picture according
to which the empirical world is constructed out of mental ‘data’ which we do not
know or understand, rather than out of sensations, which are empirical causal
products of empirical causal processes. In addition to the fact that the phenomenalist
view is harder to state in Kantian terms than is sometimes realised, it is hard to see
how Kant could regard phenomenalism as knowable. If ‘transcendentally ideal’
meant ‘in the noumenal mind’, ‘comprising noumenal sense data’, or ‘comprising
noumenal mental items’, we would not be able to have knowledge that anything is
transcendentally ideal. According to Kant, we can and do have knowledge of the
doctrine of transcendental idealism, but we cannot have knowledge of such things as
‘noumenal mental states’. Another way to put this point is to say that, unlike Berkeley
and contemporary idealists such as Foster (1982) and Howard Robinson (1982),
Kant thinks that the mental cannot be known to be ontologically basic.
27
Making a similar point, Walker says that ‘Phenomenalists hold that physical objects and states of
affairs are logical constructions out of sense-impressions, or something analogous to sense impressions.
However, for one thing, sense-impressions themselves must, for Kant, belong to the world of appearances
for we are aware of them and they belong to the temporal order. If it is appropriate to use the metaphor of
“construction” at all in this context—as Kant himself never does—temporal sense-impressions must
themselves be constructed, and so cannot be the basis for the construction’ (2010: 831–2).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
(7) Kant himself, famously, vehemently denies that his idealism is like Berkeley’s.
In the Refutation of Idealism he repudiates what he calls the dogmatic idealism of
Berkeley, saying that he disproved it in the Aesthetic (B274) and he accuses Berkeley
of degrading bodies to mere illusion (B71). In the Prolegomena Kant rejects the
‘mystical and visionary idealism of Berkeley (against which and other similar phan-
tasms, our Critique contains the proper antidote)’ (Proleg. 4: 293; see also 4: 374).
While many questions have been raised about Kant’s rejection of Berkeley, I argue
that it gives us extremely strong grounds to reject any mentalised reading of appear-
ances. However, it has not always been taken in this way.28 There are at least three
possible reasons for this. One, it could be argued that Kant misunderstood Berkeley’s
position, so his rejecting it does not imply that he is not an idealist of the Berkeleyan
type. Two, it could be argued that Kant’s objection to Berkeley is that Berkeley has
insufficient means to characterise the coherence of experience within his empirical
reality. On this account, what Kant thinks is wrong with Berkeley’s view is not his
position on the mental status of appearances, but his neglect of synthetic a priori
principles, as well as space and time, as ways of organising appearances. The idea is
that rather than seeing Kant as objecting to Berkeley’s mentalising empirical objects,
Kant’s objection is that Berkeley has an insufficiently sophisticated phenomenalism.
A third phenomenalist response would be to argue that Kant’s rejection of Berkeley
concerns the fact that Berkeley is a mere phenomenalist and that Kant’s position
differs in that he postulates, in addition to phenomena, unknowable things which are
somehow their ground. I will discuss these points in turn, and will then argue that, on
the contrary, the way Kant rejects a Berkeleyan reading of his position provides the
strongest reason for thinking that he is not a phenomenalist.
It has been argued that Kant’s criticisms of Berkeley show basic misunderstandings
of the good bishops position, since Kant accuses Berkeley of making objects mere
illusion (B71), whereas Berkeley has the means to distinguish illusion from empirical
reality within his system.29 Turbayne (1955) famously argues that Kant would have
known that he and Berkeley had the same idealism, the same refutation of scepticism,
and the same empirical realism, and that Kant’s seemingly crude misinterpretations
of Berkeley and repudiation of Berkeley’s views are in reality an attempt to distance
himself from the vulgar view of Berkeley as a mystic and visionary. Even less
flatteringly, he speculates that Kant may have wanted to hide his debt to Berkeley.
28
See, for example, esp. Turbayne (1955) and P. F. Strawson (1966: 35).
29
See, for example, Falkenstein (1995: 318–19). Against this, see Justin (1974), who discusses both
Berkeley’s and Kant’s views on space, and the means Berkeley has for making use of coherence as a way of
distinguishing illusory perceptions. He argues that, for Kant, Berkeley’s failure to have something like the
synthetic a priori principles of experience means that he does not have an adequate account of objectivity,
and that ‘failing some independently justifiable principles which determine a precise sense of what counts
as experiential cohering, Berkeley’s appeal to coherence is vacuous. And in this sense Kant’s charge that
Berkeley is unable to distinguish reality from illusion is appropriate’ (Justin 1974: 30).
WHY KANT IS NOT A PHENOMENALIST 
Whether or not Kant misunderstood Berkeley is a complex question,30 as is the

question of how Berkeley’s position should be understood. My concern here is not
with whether Kant gets Berkeley entirely right; rather, my aim is to establish that,
whether or not he got Berkeley right, his idealism is different from (standard readings
of) Berkeley’s—that Kantian appearances cannot be understood as mental entities or
constructions out of mental entities or states, as Berkeley’s empirically real objects are
normally taken to be.31
The main reason to think that Kant’s rejection of Berkeley is based on a
misunderstanding is that Kant accuses Berkeley of making bodies illusory. Berkeley
clearly does not think that the sensible things we perceive are illusions or imaginary
entities and has the means to distinguish illusions from objects within his system.
Berkeley has his own form of empirical realism. However, we can make sense of
Kant’s accusation of making bodies illusory in terms of the mentalised status that
objects are given by Berkeley. Perceptual experience subjectively presents as if it
puts us in touch with mind-independent objects; this means that there is a clear
sense in which the view that physical objects are ideas in our minds renders
ordinary perception illusory in an important manner. This holds even though
Berkeley can distinguish illusion from reality within his system. Thus, as a way of
distinguishing their positions, Kant’s claim that Berkeley makes the spatiality of
objects an illusion has a point.
The second, and I think strongest, phenomenalist response to Kant’s apparent
distancing of himself from Berkeley is to say that the important difference between
Kant and Berkeley concerns the principles that Kant uses to determine what counts
as experiential coherence and not the mentalised status of appearances.32 Evidence
for this is provided by one of Kant’s responses in the Prolegomena to the Berkeleyan
reading of his position in the 1782 Göttingen review. Kant says that the difference
between him and Berkeley concerns the fact that Berkeley regards space as empirical,
whereas he regards it as a priori (Proleg. 4: 375). This suggests that the difference
between their positions does not concern the mentalising of empirical objects, but
rather the fact that Kant thinks that these objects require an a priori structure that
Berkeley has not accounted for. This shows that not all Kant’s objections to Berkeley
can be used to reject a phenomenalist reading of appearances. However, it does not
mandate a phenomenalist interpretation: one could think both that Berkeley has
insufficient means for determining what counts as coherence within his empirical
reality and that Kantian appearances cannot be mentalised.
30
See Turbayne (1955), Allison (1973), Miller (1973), Justin (1974), Ayers (1982), Matthews (1982:
134), Wilson (1984), Walker (1985), Ameriks (2006: Ch. 3), as well as Kant’s early critics, in Sassen (2000).
Allison, Ayers, Justin, Miller, Walker, and Wilson all argue that Kant’s idealism differs significantly from
Berkeley’s.
31
Whether this is the correct understanding of Berkeley is also disputed; see, for example, Yolton
(2000).
32
Justin (1974), Walker (1985).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
The third phenomenalist argument is that Kant rejects Berkeley’s position because
Berkeley is a mere idealist who leaves out things in themselves. According to this
view, the crucial difference between Kant and Berkeley does not lie in their accounts
of the nature of empirical reality, but in the fact that Kant posits, in addition to
empirical objects, things in themselves of which appearances are the appearances. My
objection to this phenomenalist argument is that Kant’s rejection of Berkeleyan
idealism in the Refutation of Idealism does not mention Berkeley’s failure to allow
for things in themselves, but rather focuses on Berkeley’s characterisation of appear-
ances. Given Kant’s explicit distancing of himself from Berkeley’s view, if the key
difference between their positions were that Berkeley leaves out Kant’s things in
themselves, we might expect him to have said this. He never gives this as his reason
for rejecting Berkeley and instead always attacks Berkeley’s view of empirical reality.
This brings us to the part of Kant’s rejection of Berkeley that seems to me to be the
strongest argument against phenomenalist readings of Kantian appearances: the way
in which Kant rejects the mentalisation of empirically real objects in Berkeley’s
account and likewise rejects the Cartesian view of mental priority.33
In the Refutation of Idealism, Kant argues that we are immediately presented
with external objects in space and that our knowledge of the temporal order of our
inner mental states depends on this. However, it might be objected that since Kant
is an idealist who thinks that external objects in space are really in our minds,
when he says he is going to prove that the immediate objects of perception are
external objects in space, what he really means is that he is going to show
something about the way we represent some of the ideas in our minds. Perhaps,
for example, as with the Prolegomena passage discussed above, he is going to argue
that Berkeley has insufficient materials to explain how some ideas come to
represent external objects, and how we come to have knowledge of the temporal
order of our mental states. However, the problem with this objection is that
although it invokes a move that an idealist might make, it is simply not what
Kant says. It only seems that we have to find a way to attribute something like this
thought to Kant if we approach the Refutation with the assumption that Kant has
to be a phenomenalist, so that we must find a way of making what he argues there
compatible with phenomenalism. If, however, we pay attention to what it is about
33
A number of commentators argue that one of Kant’s purposes in the Critique was to reject the
Cartesian assumption that we are primarily acquainted with the contents of our own minds and that our
knowledge of physical objects is based on an inference from this. For example, Abela argues that one of the
Critique’s ‘principle aims is to shatter the image that frames the original epistemic situation in terms that
privilege inner content and relegate considerations of empirical reference and truth to secondary status’
(Abela 2002: 4). This position, which Collins and Abela defend at book-length (Collins 1999; Abela 2002),
is also argued by Allison (1973: 45), and Guyer (1987), and Ameriks (2006: Chs. 3 and 7). Guyer reads the
Refutation of Idealism in this way, which means that, given his mentalised reading of appearances, he
thinks it is inconsistent with transcendental idealism. Abela and Collins attribute phenomenalist readings
to the immersion of interpreters within the Cartesian outlook.
WHY KANT IS NOT A PHENOMENALIST 
Berkeley’s position that Kant rejects, we can see that it tells very strongly against
seeing Kant as a phenomenalist.
In both the Refutation of Idealism and the Fourth Paralogism in the A edition,
Kant characterises the idealism he rejects as a view in which the immediate objects of
perception are not external objects in space, and what is immediately perceived are
mental items—perceptions—on the basis of which external objects are inferred
(A367; B274–9). He divides his idealist opponents into those who, dogmatically,
deny the existence of matter or external objects (such as Berkeley) and those who,
sceptically, doubt the existence of external matter (such as Descartes). He identifies
the external objects Descartes doubts with the material objects that Berkeley denies.
He identifies the direct access to ideas that Descartes considers indubitable with the
access Berkeley allows to his empirically real objects. He then rejects both Descartes’
sceptical worries concerning the inference from these mental entities to the external
things that are their cause and Berkeley’s denial that the objects of our experience are
anything other than directly perceived mental items. Kant says that he wants to
prove, against both these views, that the immediate objects of perception are external
objects in space. Clearly, Berkeleyan empirical objects are immediately perceived, but
Kant takes it that they are not external objects in space, but rather something
mental. Kant wants to show, on the contrary, that the immediate objects of percep-
tion are external, material objects in space. Crucially, Kant’s contrast between
our access to our own mental states and our access to external objects in space
does not concern an account of how we need to construct experience, but concerns
what is immediately given.
While the details of the argument of the Refutation of Idealism are somewhat
obscure and remain controversial, Kant’s presentation of his aims and what he is
arguing against is clear. He says that he wants to reject the view that our experience of
our own minds is more immediate than our experience of outer objects, and he wants
to establish the reality of those very outer objects whose existence (he thinks) Berkeley
denies and Descartes renders doubtful. It is much harder to argue that Kant misun-
derstood Descartes on this point than that he misunderstood Berkeley. The external
objects whose existence Descartes renders doubtful are neither mental items nor
constructions from mental items, and Kant does not suppose that they are. The fact
that Kant argues for the existence of the very external objects which Berkeley denies
and Descartes doubts is not compatible with an interpretation on which external
objects are merely constructions out of the mental entities or states which Berkeley
and Descartes admit. A phenomenalist reading of appearances makes Kant’s state-
ment of his aim in the Refutation of Idealism more than just confused; it makes it
actually inconsistent with his transcendental idealism.
Kant describes the Cartesian position he rejects as one in which it is ‘assumed that
the only immediate experience is inner experience, and that from that outer experi-
ence could only be inferred, but, as in any case in which one infers from given effects
to determinate causes, only unreliably, since the cause of the representations that we
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
perhaps falsely ascribe to outer things can also lie in us’ (B276).34 The way Kant
describes his strategy for refuting this kind of idealism is that he is going to show that
‘even our inner experience, undoubted by Descartes, is possible only under the
presupposition of outer experience’ (B275), in other words, experience of objects
outside us, in space. The argument turns on a contrast between the kind of experience
we have of our (inner) mental states (which Descartes does not regard as subject to
doubt) and experience of objects outside us. Kant argues that ‘the determination of
my existence in time is possible only by means of the existence of actual things that
I perceive outside myself ’ (B276). Irrespective of the details of how the argument is
supposed to work, it is clear that Kant is trying to establish the reality of the spatial
objects that Descartes doubts and Berkeley denies, and that he is trying to do this by
arguing that immediate experience of outer objects is necessary for knowledge of the
temporal order of my inner states. Rather than constructing spatial representations
by ordering inner states in certain ways, Kant argues that we can have knowledge of
the temporal order of our inner states only because we have direct and immediate
experience of objects outside us in space.
Neither Berkeley’s position nor sophisticated phenomenalism can accommodate
Kant’s rejection of the view he sees as common to Berkeley and Descartes—that the
immediate objects of experience are mental. And these positions cannot accommo-
date the thesis he proposes as an alternative to this view: that we have immediate
experience of objects in space, from which our knowledge of our own mental states is
built up. A contrast between inner, mental states and objects outside us in space is
Kant’s starting point, and spatial objects are not something he tries to construct out
of mental states.
III Two Sides of Kant’s Idealism

In the Prolegomena, in responding to what he saw as reviewers’ misreading of his
work, Kant says:
Idealism consists in the assertion that there are none but thinking beings; all other things which
we believe are perceived in intuition are nothing but representations in the thinking beings, to
which no object external to them in fact corresponds. On the contrary, I say that things as
objects of our sense existing outside us are given, but we know nothing of what they may be in
themselves, knowing only their appearances . . . Consequently, I grant by all means that there
are bodies without us, that is, things which, though quite unknown to us as to what they are in
themselves, we yet know through the representations which their influence on our sensibility
procures us, and which we call bodies. Can this be termed idealism? It is the very contrary.
(Proleg. 4: 289)
34
Making a similar point in the Fourth Paralogism in the A edition, he says ‘the inference from a given
effect to its determinate cause is always uncertain, since the effect can have arisen from more than one
cause’ (A368).
WHY KANT IS NOT A PHENOMENALIST 
Here we find him denying that the things we believe we perceive in intuition are
merely representations in thinking beings. He is clear that his transcendental ideal-
ism is quite different from other positions usually thought of as idealist and,
specifically, from any view that would make the immediate objects of knowledge
into things with a purely mental existence or nature. On the other hand, he still
indicates the idealist nature of his view in saying that appearances are representa-
tions. And although he denies that his view can be called idealism in this passage, he
does continue to call his view transcendental idealism and to assert the dependence of
appearances on us, the claim that they are mere representations, and the claim that
they would not exist without us.
Kant again distances his view from Berkeley’s when he presents transcendental
idealism as the solution to the Antinomies. He says:
One would do us an injustice if one tried to ascribe to us that long decried empirical idealism
that, while assuming the proper reality of space, denies the existence of extended beings in it, or
at least finds this existence doubtful, and so in this respect admits no satisfactorily provable
distinction between dream and truth. As to the appearances of inner sense in time, it finds no
difficulty in them as real things; indeed, it even asserts that this inner experience and it alone
gives sufficient proof of the real existence of their object (in itself) along with all this time-
determination.
Our transcendental idealism, on the contrary, allows that the objects of outer intuition are
real too, just as they are intuited in space, along with all alterations in time, just as inner sense
represents them. (A491/B519–0)
However, he immediately goes on to say:

rather nothing but representations, and they cannot exist at all outside our mind; and even the
inner and sensible intuition of our mind (as an object of consciousness), the determination of
which through the succession of different states is represented in time, is not the real self as it
exists in itself, or the transcendental subject, but only an appearance of this to us unknown
being, which was given to sensibility. (A492/B520)
Here we have two important sides of his position. On the one hand, Kant clearly and
explicitly distances his account from a purely mentalised, phenomenalist reading of
appearances. On the other hand, he clearly and explicitly states that appearances are
mind-dependent. To make these claims consistent, we need an account of mind-
dependence that is not phenomenalist. What is needed is an account which allows
that empirical objects can exist outside our minds, in space, and not as constructions
out of something merely mental, while at the same time depending on (the possibility
of) our representing them, or their being presented to us. If we can give an alternative
account of the way in which appearances depend on our minds, which explains
Kant’s saying that they have an essential connection to perception and his calling
them representations, then there will be no reason to accept a phenomenalist reading
of Kantian appearances. Those who show that Kant is committed to some kind of
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
mind-dependence often take this to demonstrate that he is a phenomenalist. Those

who show that there are serious problems with seeing Kant as a phenomenalist take
this to mean that he is not an idealist. Because there are clear textual and philosoph-
ical reasons to see Kant as an idealist, as well as to think that he is not a phenom-
enalist, each extreme is able, relatively easily, to disprove the alternative view, leading
to an ongoing pendulum swing between opposing extreme views. To reach a stable
interpretation we need to accommodate both: we need a non-phenomenalist account
of Kant’s idealism.
3
Things in Themselves
Without Noumena
I Introduction
A central part of transcendental idealism is the claim that we cannot cognize things as
they are in themselves. In my view, Kant is committed to the claim that the things
(objects) of which we have empirical cognition and knowledge have natures or
aspects that are entirely independent of us, which ground their appearances. There
are three prominent alternative readings. One, the traditional ‘two-world’ interpret-
ation, adds a specific and stronger commitment: the idea that Kant thinks there exist
non-spatio-temporal, non-sensible objects—objects of a completely different kind
from the spatio-temporal objects of which we have knowledge.1 This position goes
beyond thinking that Kant is committed to there being an aspect of reality we cannot
know which grounds spatio-temporal appearances (a claim with which I agree); it
gives a specific characterisation of the nature of this aspect of reality. I call this
position noumenalism. At the other extreme, there are interpretations which see
Kant’s position as including only his empirical realism and which deny that he is
committed to the existence of anything other than the spatio-temporal appearances
which are empirically real objects. A third possibility is that held by those who deny
that Kant has a commitment to things in themselves because they deny that tran-
scendental idealism is a metaphysical position. They see the notion of things in
themselves as simply a way of thinking about the spatio-temporal objects of our
knowledge. I call these deflationary interpretations. In this chapter I argue against
noumenalist and merely empirical realist interpretations and in the next chapter
against deflationary readings. There is clearly a degree of overlap between some
versions of the second and the third alternatives, so some of the arguments in this
chapter also apply to deflationary interpretations and some of the arguments in the
next chapter apply to those who see Kant as an empirical realist only.
In Section II of this chapter, I argue, against noumenalist readings, that Kant’s
commitment to there being a way things are in themselves should not be understood
as a commitment to there being non-spatio-temporal intelligibilia in addition to the
1
See, for example, P. F. Strawson (1966: 236).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
objects of which we have experience. In Section III I argue, against merely empirical
realist interpretations, that a commitment to there being a way things are in them-
selves is consistent with Kant’s project in the Critique and that he clearly expresses
such a commitment. A worry about my reading comes from objections to thinking
that there can be an identity relation between things in themselves and appearances.
I discuss this in Section IV. I argue that we do not need to see Kant as committed to
problematic object-identity statements in order to see him as committed to the claim
that the spatio-temporal objects we experience are appearances of a more funda-
mental or ultimate aspect of reality which we cannot cognize and which is somehow
responsible for the aspect we experience.
Kant frequently speaks about things as they are in themselves as if he is referring to
an existing aspect of reality and there are both textual and philosophical reasons for
thinking that he is committed to there being a way things are in themselves. However,
there are also a number of grounds for denying that Kant is committed to there being
an additional kind of object (a kind of object different to the spatio-temporal objects
of our experience) of which we cannot have knowledge. Perhaps the four most
obvious reasons are the following. One, as we have seen, Kant frequently talks of
things as they appear to us and these same things as they are in themselves. Two,
Kant explicitly denies that we could have knowledge of the existence of non-
sensible, non-spatial objects. Three, he says that the notion of noumena is prob-
lematic and can be used only in a negative sense. Four—closely related to the
previous two points—unlike saying that things in themselves exist but that we
cannot cognize them (which may involve some tension but does not involve
contradiction), the noumenalist claim does seem to involve contradiction: it gives
a positive characterisation of things in themselves that does not seem consistent
with Kant’s denial that we can cognize them. These considerations have been used
to motivate both merely empirical realist interpretations and deflationary inter-
pretations. There are interpretative pressures in competing directions here, similar
to those we saw in the previous two chapters with respect to Kant’s idealism. There,
we saw that there is clear textual evidence for seeing Kant as an idealist; at the same
time, there are problems with seeing him as a phenomenalist. Similarly, there are
strong textual and philosophical reasons for thinking that Kant is not committed to
the existence of intelligibilia, but equally good textual and philosophical reasons for
thinking that he is committed to the existence of an aspect of reality of which we
cannot have knowledge.
II Positive and Negative Conceptions of Noumena

By ‘noumenalism’ I mean the view that Kant’s account of things in themselves
involves a commitment to the existence of objects which are distinct from and
entirely different from the things of which we have knowledge: objects which are
THINGS IN THEMSELVES WITHOUT NOUMENA 
not spatial, and which cannot be sensed. Kant thinks that such objects could be
known only by a non-sensible (intellectual) intuition, which would be an intu-
ition which could have acquaintance with objects without being affected by them.
Such things would be intelligibilia, the literal meaning of noumena: things
which can be known by the intellect as opposed to the senses. From the point
of view of theoretical cognition, Kant thinks that we can form a thin conception
(though we cannot fully understand it) of a kind of object that could be presented
to an intellectual intuition: objects which are not spatio-temporal and do not
interact with the senses.2 To get a grasp on what might be meant by such non-
spatial, non-sensible objects, we can think of such things as Cartesian souls,
Leibnizian monads, God, perhaps Platonic forms, and Platonically understood
numbers. In the Critique, Kant clearly argues that we cannot cognize such
objects, we do not have knowledge such objects exist, and we do not even really
understand what they would be. He states explicitly that his notion of things in
themselves should not be understood as a commitment to such objects. Kant is
not a noumenalist.
The section ‘On the Ground of the Distinction between Noumena and Phenom-
ena’ (A235–60/B294–315) is the place where Kant most explicitly repudiates the idea
that things in themselves should be understood as intelligibilia. Here, Kant distin-
guishes between a positive and a negative conception of noumena; this distinction is
crucial to understanding things in themselves. The distinction concerns how much
content and what kind of content we can give to the notion of noumena—to the
notion of things which are grasped by the intellect rather than by the senses.3 As we
will see, in terms of this distinction, the noumenalist sees Kant as committed to
noumena in the positive sense, but Kant makes it clear that his notion of things as
they are in themselves should be understood in terms of a negative conception of
noumena (B307; B311; A251–2).
The distinction between the positive and negative conceptions of noumena is
drawn far more clearly and explicitly in the second edition. Understood in a positive
sense, the concept of noumena is the concept of a kind of object which could be
cognized by the understanding alone, which means, for Kant, something that could
be ‘an object of a non-sensible intuition’ (B307). He says that this notion assumes a
special kind of intuition, intellectual intuition, ‘which, however, is not our own, and
2
Ideas of such objects play a role in Kant’s moral philosophy, where he also allows some kind of
cognition of them.
3
It is not a distinction between different kinds of noumena. There may be different kinds of noumena,
for example, there may be noumena which are not aspects of the things which appear to us (like God), but
the distinction between a positive and a negative conception of noumena is not an attempt to sort noumena
into kinds. Rather, there is a positive notion of non-sensible objects (intelligibilia) and a purely negative
notion which simply abstracts from what we know of objects through sensibility, and makes no claims at all
about what this leaves.
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
the possibility of which we cannot understand’ (B307). He is clear that this is not how
his notion of things in themselves should be understood:
the doctrine of sensibility is . . . the doctrine of the noumenon in the negative sense, that is, of
things which our understanding must think without reference to our mode of intuition,
therefore not merely as appearances but as things in themselves. (B307)
The division of objects into phenomena and noumena, and the world into a world of the senses
and a world of the understanding, is therefore quite inadmissible in the positive sense. (B311)
[The critique of pure understanding] thus does not allow us to create a new field of objects
beyond those that can come before it as appearances, and to indulge in intelligible worlds, or
even in the concept of them. (A289/B345)4
Not only does Kant say that we do not have knowledge that there are intelligibilia;
even more strongly, he says that we do not know if they are even really possible. Kant
distinguishes between logical possibility and what he calls real possibility. He holds
that the mere lack of contradiction in the concept of an object does not tell us that it is
the concept of a really possible object, and he says that we can know that an object is
really possible only if it could be presented to us in intuition—it could be given to us
(B308). Giving a positive conception of a noumenon involves attempting to charac-
terise objects of a sort which could be known by an intellect alone without being
given in sensible intuition; Kant thinks we cannot do this. Where we have concepts
only, we have mere logical possibility; we do not even know whether, in Kant’s terms,
we have the concepts of really possible things. He thinks that we mistakenly think we
are entitled to say that there are such objects because we are committed to there being
a way things are in themselves that grounds the way they appear to us. We imagine
that we can understand what it would be for there to be intelligibilia because we
mistakenly think that using concepts alone enables us to reach positive conclusions
about such objects—such as the thesis that they are simple, thinking things, like
Leibnizian monads (this is his argument in the Amphiboly section).
Kant says that the notion of a noumenon in the negative sense is ‘a thing insofar as
it is not an object of our sensible intuition, because we abstract from the manner of
our intuition of it’ (B307). Understood in the negative sense, the concept simply
involves thinking about the spatio-temporal objects of our experience and abstract-
ing what we know about them through the senses. Because the negative conception of
noumena involves abstracting from our sense experience of objects, some merely
empirical realist and deflationary commentators argue that this conception is merely
an abstract thought. They argue that in saying that we must understand things in
themselves as noumena in a negative sense, Kant is denying that we should think
there actually is an aspect of things that we cannot cognize. However, as we will see,
4
Similarly, he says ‘[w]e can never know objects of sense (of outer sense and of inner sense) except as
they appear to us, not as they are in themselves. Similarly, supersensible objects are not objects of
theoretical knowledge for us’ (C 12: 224).
THINGS IN THEMSELVES WITHOUT NOUMENA 
Kant makes it clear that he is committed to something actually existing. He thinks we

cannot characterise this something: when we abstract from what we know of objects
through the senses we are left with no positive content so although there is a way
things are in themselves we have a merely negative understanding of this. We have no
determinate representations of what things are like as they are in themselves.
Although Kant draws the distinction less clearly in the first edition, we can also
find there a distinction between different ways of understanding noumena, and this
corresponds to the more explicitly drawn distinction in the B edition. In the first
edition, Kant starts off by saying that ‘If . . . I suppose there to be things that are
merely objects of the understanding and that, nevertheless, can be given to an
intuition, although not to sensible intuition . . . then such things would be called
noumena’ (A249). He argues that a natural movement of thought leads us to the
idea of such noumena: the interrelated notions of appearances, beings of sense, and
phenomena inevitably lead us to the notion of things which are objects purely for an
intellect, independent of sensibility. In thinking that there are things that appear to
the senses, we inevitably think that there is something independent of us that
grounds the way things appear. However, he says that for ‘a noumenon to signify a
true object, to be distinguished from all phenomena, it is not enough that I liberate
my thoughts from all conditions of sensible intuition, but I must in addition have
ground to assume another kind of intuition than the sensible one’ (A252). On this
account of noumena, we attempt to characterise them not just by abstracting from
what we know of objects through sensibility, but by thinking of a different kind of
intuition, one which could have acquaintance with objects without being affected by
such objects. Kant then argues that it is a mistake to consider this (as he sees it)
inevitable movement of thought as giving us insight into the existence of intelligibilia.
He says that we could not prove that such a kind of intuition is possible and we
therefore do not even know whether such objects are possible. He says that ‘one might
have thought that the concept of appearances . . . already yields by itself the objective
reality of noumena’ (A249, my italics), but argues that this is not the case. He makes it
clear that his notion of appearances does not require noumena in this sense: he says of
‘the object to which I relate appearances in general’ that it cannot be called the
noumenon, ‘for I do not know anything about what it is in itself, and have no concept
of it except merely that of the object of a sensible intuition in general’ (A253).
In the same section, immediately after saying that his notion of appearance does
not require intelligibilia, Kant says that from the concept of appearances it follows
naturally that something must correspond to the appearance—that something
appears—and from this arises a concept of a noumenon ‘which is not at all positive’
(A252). He says that the word ‘appearance’ must indicate a relation to something ‘the
immediate representation of which is, to be sure, sensible, but which in itself . . . must
be something, i.e. an object independent of sensibility’ (A252). Here it is clear that
he is committed to the existence of something other than what we know through
the senses, something which grounds appearances. Kant says that there must be
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
something other than appearances that appears, denies that this ground of appear-
ance is noumena in the sense of distinct non-sensible objects, and says that it is
noumena in another sense.
Certainly, the A version is not a model of clarity: Kant sometimes seems to say that
the things in themselves to which appearances are referred are noumena and
sometimes to deny this. He sometimes suggests that we must refer appearances to
noumena and sometimes that the notion of noumena is a merely coherent limiting
concept which we cannot use for any positive work and of which we cannot even say
that there are such things. But a clear line of thought can be disentangled from these
passages, by picking up on Kant’s hint that the notion of noumena to which
appearances are referred ‘is not at all positive’, while the ‘true’ notion of a noumenon
is that of a purely intelligible object, concerning the very existence of which we can
have no knowledge. This reading of the A edition is supported by the fact that this is
exactly the distinction Kant draws in the B edition.
The distinction between the positive and negative conceptions of noumena enables
us to navigate through what might otherwise appear contradictory features of Kant’s
position. He explicitly presents a conception of noumena as intelligibilia, and expli-
citly denies that his conception of things in themselves should be understood in this
way. We find ourselves led to the thought of positive noumena (and the idea of
intelligibilia plays a role in our thinking) but we do not even know so much as that
they are really possible. On the other hand, he thinks that we are committed to there
being something that exists independently of what we know of objects through
sensibility; but he thinks that we can give no positive characterisation of this
something. We are committed to thinking that the objects of our cognition have a
way they are independently of our sensory cognition of them, but we have no
characterisation at all of these natures.
In the section on Phenomena and Noumena Kant introduces the idea of a
problematic concept, which is one which contains no contradiction (contains no
contradictory concepts), but the objective reality of which ‘can in no way be cognized’
(A254). He says that the concept of positive noumena (intelligibilia) is problematic in
this sense. He says that ‘the concept of a noumenon, i.e. of a thing that is not to be
thought as an object of the senses, but rather as a thing in itself (solely through a pure
understanding), is not at all contradictory, for one cannot assert of sensibility that it is
the only possible kind of intuition’ (A255/B310). However, he also says that ‘we have
no insight into the possibility of such noumena, and the domain outside of the sphere
of appearances is empty (for us)’ (A255/B310). We do not even have insight into
their real possibility. Kant thinks that problematic concepts can play essential
limiting or regulative roles in our thinking despite the fact that they cannot be
cognized. He says that the concept of a noumenon is admissible as a boundary
concept, which has a negative (and unavoidable) use, but that in that case ‘it is not a
special intelligible object for our understanding; rather an understanding to which it
would belong is itself a problem’ (A256/B311). Since Kant thinks that the notion of
THINGS IN THEMSELVES WITHOUT NOUMENA 
intelligibilia is problematic, it cannot be that his transcendental idealism centrally

involves a commitment to there being such things. This has been taken by some
commentators to support thinking that the notion of things in themselves is prob-
lematic: that it is a concept we unavoidably posit but which should not be understood
as involving commitment to an existing aspect of reality that we cannot cognize.
I argue against this in the next section.
III Against Bare Empirical Realism

In Chapter 1 we saw textual evidence that Kant very frequently both asserts and
implies that there is a way things are in themselves. However, some commentators
argue that these texts need to be explained away, because they take them to be
inconsistent with other things Kant says. I will reject these accusations of inconsis-
tency, and conclude from this that the straightforward reading of the text—which
says that there is a way things are in themselves—stands.
From the fact that Kant denies a commitment to intelligibilia, says that we can use
the notion of noumena only in a negative sense, and says that the notion is
problematic (in his technical sense in which this means that we cannot know whether
there exist any objects to which it applies), some commentators conclude that he does
not think there actually is an aspect of reality which we cannot know. In addition,
some interpreters argue for the stronger position that both a commitment to the
existence of things in themselves and the idea that they cause or ground appearances
would be inconsistent with Kant’s overall position in the Critique, including his
restriction of the categories to a use only within the realm of experience.5 They argue
that Kant’s talk of the unavoidability of the notion of things in themselves and his
saying that the notion of appearance implies that there is something that appears
(Bxxvi; A251–2; Proleg. 4: 315) can be taken to express a commitment merely to the
unavoidability of the concept of things in themselves, and not a commitment to
something that exists corresponding to this concept. I call these positions merely
empirically realist interpretations, because they see Kantian empirically real objects
as all his ontology contains. I argue against merely empirical realist interpretations in
this section. Deflationary interpreters also reject a commitment to things in them-
selves; in the next chapter, I argue that deflationary characterisations of things in
themselves cannot do the philosophical work Kant needs them to do. These argu-
ments are also relevant to the case against mere empirical realist interpreters.
Examples of merely empirical realist interpretations include those of Bird (2006),
Hanna (2006), and Senderowicz (2005). Bird, for example, says of the division
5
See, for example, Schrader, who says that when Kant talks about things in themselves as causes of
appearances, and as accessible through practical reason, he is using the notion of things in themselves
dogmatically (Schrader 1968: 185). See also Prauss (1974: 197), Senderowicz (2005), and Hanna (2006:
422–6).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
between phenomena and noumena in the negative sense that ‘such a division is not
like an empirical distinction between two genuine kinds of object, but only between
phenomena, which are the things we ordinarily perceive and know about, and the
empty (but not inconsistent) concept of a non-phenomenon’ (Bird 1962: 74). He
argues that while a two-aspects interpretation ‘usefully avoids the mistake of sup-
posing that the two concepts have different objects, it may mislead by suggesting that
appearances and noumena are simply two sides of the same coin, or equal partners in
the same joint enterprise. The truth seems, however, that these are two ways of
looking at the same thing only because, on Kant’s view, there is only one thing at
which to look, namely appearances’ (Bird 1962: 29).
If we were to think that the only way to understand things in themselves is as
noumena in the positive sense, we would have strong grounds for denying that Kant
thinks there actually is a way things are in themselves, because Kant denies that we
have knowledge that there are intelligibilia. However, we have seen that this is not the
only way to understand things in themselves, and Kant’s denial of our knowledge of
the existence of intelligibilia is not the same as a denial that there is an aspect of
reality that is independent of us. If it were, he would be seriously inconsistent, since
he repeatedly states that there is an aspect of reality that is independent of us which
we cannot cognize. Even in the less clear A edition, Kant says that the ‘object to which
I relate appearance in general . . . cannot be called the noumenon; for I do not know
anything about what it is in itself ’ (A253). Here he both denies a commitment to
intelligibilia and asserts that appearances are appearances of something we do not
know. So in the very section in which he says that he is not committed to intelligibilia,
Kant repeatedly affirms the existence of a way things are in themselves. While he
explicitly denies that we can even know whether intelligibilia are possible, he
nowhere says that we cannot know whether it is possible that there is a way things
are in themselves; on the contrary, he says that in recognising that there are
appearances, we are committed to there being a way things are in themselves.
It is a mistake to think that Kant’s saying we can use the notion of noumena only in
a negative sense implies that he has no commitment to there being a way things are in
themselves. Kant says that by ‘noumenon’ in the negative sense ‘we understand a
thing insofar as it is not an object of our sensible intuition’ (B307). In the negative
sense, the idea of noumena involves taking the objects of which we have sense
experience and subtracting, in thought, what we know of them through sense
experience. Kant argues that since we cannot have cognition without intuition and
our intuition is sensible, it turns out that we are unable to give any positive
characterisation of objects when we abstract from them what we know of them
through the senses. Saying that the use of the notion of noumenon here is negative is
not a denial that there is anything existing corresponding to the notion; the claim is
simply we give the notion no positive content. When we try to characterise things as
they are independently of what we know about them through the senses, the only way
we can do so is negatively: we simply have the idea that after we take away what we
THINGS IN THEMSELVES WITHOUT NOUMENA 
know about objects through sense experience, something remains which is respon-
sible for (grounds) what we experience, although we can have no substantial know-
ledge of this something (and cannot even understand what it might be). This means
that Kant’s assertion that things in themselves are to be understood only as noumena
in the negative sense is not a denial that there is a way things are in themselves; it is a
denial that we can characterise things as they are in themselves (including character-
ising them as intelligibilia, or noumena in the positive sense).
The merely empirical realist interpretations read Kant’s notion of things in
themselves as functioning like what he calls transcendent metaphysical ideas,6 but
he does not say that the notion should be understood in this way. One of Kant’s
central concerns in the Critique is with transcendent metaphysical ideas like the ideas
of God, the soul, and the universe as a systematically unified totality. A whole section
of the Critique—the Transcendental Dialectic—is devoted to giving a detailed, clear,
explicit account of these ideas, their role in our thinking, how reason leads us to
them, and to arguing that we can never know that any of them are instantiated. Kant
argues that our thinking about the world naturally leads us to think about the self as
simple, about the world as a whole, and to the idea of God, but that we cannot have
knowledge that there are soul-substances or that there is a God and that we can know
that the idea of the spatio-temporal world as a whole cannot be instantiated. He also
diagnoses what he thinks are the mistaken thoughts that lead us to think that we can
have knowledge of these things. In contrast, the claim that the notion of things in
themselves is a merely limiting notion which does not involve a commitment to
anything actually existing is not something Kant says anywhere.7 He does not say
about the notion of things as they are in themselves that it is an unavoidable postulate
of reason which does not involve a commitment to anything existing. He does not say
that the notion of things in themselves functions in a way that is anything like his
account of the ideas of reason (the concepts of transcendent metaphysics). If Kant’s
view were that the role for which he introduces the notion of things in themselves was
as empty of metaphysical commitment as the notion of a unified totality of appear-
ances or a Cartesian soul he surely would have said so, since he is so clear about this
with respect to the latter notions. Transcendental idealism runs right through the
Critique and features both in his positive project of characterising a metaphysics of
experience and in his negative project of rejecting transcendent metaphysics. Given
the absolute centrality of things in themselves in Kant’s transcendental idealism and
the centrality of his analysis of problematic metaphysical concepts which play
6
Hanna argues that the notion of things in themselves is an idea our minds cannot help positing, but
that we have no more rational entitlement to say that this notion is instantiated than we have with respect
to the concept of a being containing all perfections (Hanna 2006: 197). Along similar lines, Senderowicz
argues that the notion of a noumenon is relevantly like an idea of reason (Senderowicz 2005: 14); for Kant,
such ideas have a necessary regulative role in empirical knowledge, but cannot be known to be instantiated.
7
Unlike the notion of noumena in the positive sense, which he does sometimes treat as similar to
transcendent ideas.
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
regulative, limiting roles, but which cannot be known to have instantiations, the
claim that things in themselves are supposed to have the status of transcendent
metaphysical ideas is not credible.
Kant’s claim that the notion of noumena is problematic is argued by some to show
that he cannot be committed to the existence of a way things are in themselves.
However, as we have seen, it is the positive conception of noumena, intelligibilia,
which Kant says is a problematic concept. In the ‘Remark on the Amphiboly of
Concepts of Reflection’, Kant confirms that the problematic concept of a noumenon is
the representation of a thing of which we can say neither that it is possible nor that it is
impossible, since we are acquainted with no sort of intuition other than our own sensible one
and no other sort of concepts than the categories, neither of which, however, is suited to an
extrasensible object. (A287/B343)
Kant thinks that, as a problematic concept, the notion of intelligibilia plays an

important role in limiting the pretensions of sensible experience to cognize every-
thing that there is (since we have no justification for asserting that there are not
intelligibilia) and he thinks that the notions of God and the soul play an important
role in our thinking about morality. But he simply never says that the notion of things
as they are in themselves is the representation of something of which we can say
neither that it is possible nor that it is impossible. Rather, he says that if there are
appearances there is a way things are in themselves and that things in themselves
ground appearances.
Some commentators have argued that other core philosophical commitments in
the Critique, in particular Kant’s account of the conditions of knowledge and
cognition, rule out Kant’s being committed to the existence of an aspect of reality
which we cannot cognize. One of Kant’s aims in the Critique is to argue that the a
priori concepts he calls the categories (concepts such as causation and substance) can
be used in substantive knowledge claims only when applied to spatio-temporal
objects which could be objects of experience (for example, A96, A139/B178, and
B147). He argues that outside of the intuitive (spatio-temporal) interpretation we
give such concepts they are merely logical forms which cannot be used to cognize any
objects. For example, the concept of causation is merely the notion of the dependence
of a consequence on a ground, and the concept of substance is merely the notion of
something which can be thought only as a subject. Independently of applying it to
persisting causal unities in space, we have no idea what corresponds to the notion of
something which can be thought only as subject, and we have no way to apply it, or to
use it to cognize any objects. This is thought to show that Kant cannot consistently
say that things in themselves cause appearances.8
A first response to this objection is simply to note that, as we have seen, Kant does
speak of things in themselves as the cause of appearances. Further, as a number of
8
See, for example, Prauss (1974: 197).
THINGS IN THEMSELVES WITHOUT NOUMENA 
commentators have pointed out,9 outside of the spatio-temporal framework of our

knowledge, we can still use the unschematised categories (the categories understood
independently of the spatio-temporal conditions which enable us to apply them to
objects of experience). When we speak of things in themselves as grounds of
appearances, we are using the category of causation outside of the spatio-temporal
conditions within which we are able to have knowledge of causes. This means that
not only are we unable to cognize the way things are in themselves, but we also do not
understand how they are responsible for appearances and do not even understand
what kind of dependency the relevant grounding might be (apart, perhaps, from
some formal features of the relation). We cannot use the category to give us any
determinate knowledge of what kind of relation this is. This result is entirely
consistent with what Kant says about things as they are in themselves: we have no
knowledge of things as they are in themselves or of how they are responsible for
appearances. But to say that we have no specific or substantial (as opposed to merely
formal) understanding of the grounding relation is not to say that we are not entitled
to assert that there is one, and Kant speaks throughout the Critique as if he takes
himself as entitled to assert that there is one. I discuss this grounding relation further
in Chapter 10.
Some commentators argue that a commitment to there being things in themselves
involves an existential commitment which is not made within experience and
therefore is not, in Kantian terms, legitimate. This is argued by Senderowicz
(2005), who says that knowledge of noumena is incompatible both with Kant’s
‘principle of significance’10 and with his account of synthetic a priori knowledge,11
and by Wood, who says that the category of existence applies only to phenomena
(2007: 6).12 In response to this, we can first note that, following the above argument,
we can use the unschematised category of existence. Further, we are not making an
unjustified existential commitment, because Kant seems to think that it is a concep-
tual truth that where there is something that appears in a certain way to us there is
something that has a way it is in itself (Bxxvi, A251–2). Similarly, he thinks that it is
a conceptual truth that what is relational requires something non-relational
(A284/B340) and he thinks that appearances are entirely relational. (I discuss this
in detail in Chapter 11.) Anything we can know to be true as a matter of logic we can
know to be true of everything that exists, so if something exists and logic commits us
9
See, for example, Ameriks (2003: 32).
10
The ‘principle of significance,’ which P. F. Strawson attributes to Kant, says that ‘we can make no
significant use of concepts in propositions claiming to express knowledge unless we have empirical criteria
for the application of those concepts’ (Strawson 1966: 241).
11
Senderowicz agrees that there is textual evidence for the opposing view: he thinks that Kant ‘was
divided regarding the question of whether the actual existence of things in themselves should be part of his
transcendental theory of experience’ (Senderowicz 2005: 10). Similarly, see Hanna (2006: 422).
12
See also Bird (2006: 553–80).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
to thinking that there is a way things are in themselves, then we can know that there
is a way things are in themselves.
Hanna argues that these analytic entailments cannot be used to show that there are
things in themselves, because it would be inconsistent with Kant’s critique of the
ontological argument to think that we can make existence claims on the basis of
analytic statements. Kant argues that all existential claims are synthetic, that we
cannot infer the existence of anything from mere concepts, and that, although reason
leads us to the concept of an absolutely necessary being, we cannot conclude from
this that the concept is instantiated (A592–603/B620–31).13 This would have force as
an argument against intelligibilia, but the objection will not hold if we understand
things in themselves as I do, where the idea is simply that the things which we
cognize have a way that they are independently of our cognizing them. In saying that
there is a way things are in themselves we are not positing distinct supersensible
things but only talking about an aspect of the things we know, so we are not making
new, unjustified existential commitments. There is no commitment to a new object
but rather a claim about the objects of our knowledge: they have a nature which
grounds the way they appear. It is not inconsistent with Kant’s critique of the
ontological argument to think that we can make assertions about the things we
already know to exist on the basis of analytic judgments that apply to them. If it is a
conceptual truth that appearances are appearances of something, then it does not
even make sense to think that there are appearances without there being a way things
are in themselves. Similarly, if it is a conceptual truth that relations require something
non-relational, and if there are appearances that are entirely relational, then we are
entitled to assert that there is something which is not relational. This does not involve
making a new existential commitment merely on the basis of an analytic judgment;
rather, it involves drawing a conclusion about the objects of our knowledge on the
basis of (as Kant sees it) a conceptual truth. As we will see, Kant thinks that some-
thing’s being logically possible is not enough to tell us that it is really possible,
however, something’s being logically impossible is enough to tell us that it is really
impossible. And Kant thinks that logic tells us that relations require something non-
relational and appearances require something that appears. I discuss this issue in
detail in Chapter 11.
13
Hanna argues that Kant’s critique of the ontological argument shows that ‘analytic entailments of the
concept of existence guarantee at most the logical possibility of the thing to which the concept of existence
necessarily applies’ (Hanna 2006: 197). However, it might be argued that the point of Kant’s critique is not
to argue that analytic entailment of the concept of existence does not imply that a concept is instantiated,
but rather that existence is not the kind of concept which can be analytically entailed by a conceptual
account of a thing’s nature (A597/B625). Further, Kant’s critique of the ontological argument is concerned
specifically with thinking of existence as what he calls a ‘determining’ predicate.
THINGS IN THEMSELVES WITHOUT NOUMENA 
IV The Same Things?

One way in which the question of the relation between things in themselves and
appearances is sometimes formulated is in terms of the question of whether there is
one class of objects or two distinct classes of objects.14 As we have seen, there are
many texts in which Kant talks about things as they appear to us and these same
things as they are in themselves. It is sometimes thought that ‘one-object’ views must
be committed to thinking that things in themselves and appearances must be
understood as numerically identical objects. However, both noumenalist interpreters
and some deflationary interpreters have argued that there are problems with this.
One possible problem is that Kant seems to assert that things in themselves and
appearances have contradictory (incompatible) properties (such as being spatial and
not being spatial), but two things with incompatible properties cannot be the same
object.15 Another objection is that regarding things in themselves and appearances as
numerically identical objects might be thought to give us more information about
things in themselves than we should, in Kant’s terms, properly have. It seems to allow
us to make some claims, for example, about how things in themselves are individu-
ated (on the basis of claims about how appearances are individuated).16 It is also
argued that since, in Kant’s view, identity conditions for things in themselves and
appearances are different, there is no one thing that can be considered from two
perspectives.17 Noumenalists may take these considerations to show that things in
themselves and appearances cannot be numerically identical and thus must be
distinct entities,18 while deflationary interpreters may take them to show that Kant
is not ontologically committed to anything other than what appears to us.19 I will
argue that both sides are right to point out that we are never in an epistemic position
in which we can pick out things in themselves and things as they appear to us and
14
Van Cleve (1999), Walker (2010), and Stang (forthcoming).
15
A number of commentators have argued that this creates difficulties for deflationary interpretations
that see Kant’s distinction as merely epistemic or methodological. As Hoke Robinson points out, ‘We
would not, for instance, accept the possibility of a round square on the suggestion that we distinguish
methodologically the consideration of it as round from the consideration of it as square’ (Hoke Robinson
1994: 422). Guyer argues that a conception of objects that abstracts from certain of their properties does
not in itself give us reason to change our view about what properties objects have (Guyer 2007: 13–14).
Similarly Van Cleve says that ‘the double-aspect view has always seemed to me unfathomably mysterious.
How is it possible for the properties of a thing to vary according to how it is considered?’ (Van Cleve 1999:
8). Hanna argues that the two-aspect theory is incoherent if interpreted ontologically, since it entails the
existence of a class of otherwise unspecified objects each of which instantiates two contradictory sets of
properties, and unexplanatory if interpreted epistemically, since it does not tell us why we perversely persist
in ascribing contradictory properties to the same objects nor does it justify our beliefs in the objective
correctness of those ascriptions (Hanna 2001: 109). See Marshall (2013a) for discussion of why a one-
object view need not be committed to asserting that appearances and things in themselves are numerically
identical objects.
16
Walker (2010) and Stang (forthcoming).
17
Walker (2010). This is also argued by Senderowicz (2005: 13, 161).
18
Walker (2010), Jauernig (forthcoming), and Stang (forthcoming).
19
Senderowicz (2005) and Hanna (2006).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
then identify them, and therefore that talking about them being numerically identical
objects is problematic. However, I will argue that this does not count against the idea
that the things which appear to us also have a way they are in themselves, which we
cannot know.
Normally, when we ask with respect to two putatively distinct things whether they
are in fact numerically identical we have some way of individuating and referring to
each. We can ask the question of things picked out at different times (a caterpillar
and a butterfly), through different modalities (lightning and thunder), or under
different sortal concepts (a statue and a lump of clay). On Kant’s account of
cognition, the only way we have of individuating or picking out things is at the
level of spatio-temporal appearances. This means that we cannot pick out the objects
that are appearances and pick out objects that are things in themselves and ask the
question whether or not they are the same objects or distinct objects. This is taken by
some ‘two-world’ or ‘two-object’ interpreters to show that Kant cannot think that
appearances and things in themselves are numerically identical things, thus disprov-
ing two-aspect interpretations (in both metaphysical and epistemological versions).20
However, I suggest that it should rather be taken to show that the question of whether
things in themselves and appearances are numerically distinct or numerically iden-
tical objects is not the right question: the question is well posed only where we have
objects individuated in two different ways and we want to know whether we can
identify them. This is what we do not have: we have no way of individuating things in
themselves. We cannot pose a question about the numerical identity or distinctness
of things in themselves and appearances without having some way of identifying the
objects that are things in themselves. We do not have any way of doing this.
‘One-world’, ‘one-object’, or ‘two-aspect’ views are sometimes seen as asserting the
identity of things in themselves and things as appearances. In other words, they are
seen as posing the same question as to whether or not things in themselves and
appearances are numerically identical and giving a positive answer. However, the
denial of intelligibilia, together with the claim that everything that appears to us is an
appearance of something which has a way that it is independently of appearing to us,
need not be understood as saying that every appearance is a thing that is numerically
identical with a thing that exists in itself. I have argued that Kant starts with the
assumption that there are things. In my view (defended in detail in Chapter 10), Kant
thinks that things have a nature that is independent of their actual and possible
relations to other things: a way they are in themselves. Kant argues that we cannot
know this nature. He does not start with a domain of things identified in terms of
their intrinsic natures and a separate domain of things identified in terms of their
relational appearances and then ask whether we are dealing with one class of objects
or two. Rather, he starts with things, things which have intrinsic natures and
20
Walker (2010: 824).
THINGS IN THEMSELVES WITHOUT NOUMENA 
relational properties, and from there he argues that we cannot know these things’
intrinsic natures and can know only their relational appearances.
Van Cleve (1999: 147) argues that Kant asserts that the identity of indiscernibles
applies to things in themselves but not to appearances, and that things in themselves
are simple (indivisible) whereas there is no simplicity in things as they appear to us.
He takes this to count against all ‘two-aspects’ interpretations, because it involves
attributing contradictory properties to things in themselves and appearances. While,
as I will argue later, we can make sense of some of Kant’s contradictory property
attributions, there is a prior objection to Van Cleve’s argument. This is that in the
Amphiboly, the section where Kant talks about the identity of indiscernibles and
simplicity in things in themselves, Kant is talking about what we seem to be able to
say of noumena in the positive sense—objects which could be known purely through
the understanding, understood as Leibnizian monads.21 These claims are not ones
Kant actually asserts: the Amphiboly argues that we think we have insight into
noumena through mere concepts, but that this is a mistake. We are not entitled to
assert these claims. He argues that if we think about the objects of our knowledge
without paying attention to the spatio-temporal conditions of our knowledge, and if
we think that relations between concepts alone give insight into reality, we will be led
to think that reality consists of Leibnizian monads (and that the principle of the
identity of indiscernibles applies to them, that they are simple, etc.). But it does not
follow from this that Kant thereby attributes to things in themselves the properties he
diagnoses us as wanting to attribute to Leibnizian monads. And this would be clearly
incompatible with his restrictions on our cognition. So it is incorrect to ascribe to him
the claim that things in themselves are simple. Kant’s contrast in the Amphiboly is
not between the properties of appearances and the properties of things in themselves,
but between the properties of appearances and the properties we (incorrectly) think
we are entitled to attribute to intelligibilia, or positive noumena.
However, the worry about contradictory properties remains, because Kant does
say that things as they appear to us are spatio-temporal and that things as they are
in themselves are not. Here we still have the problem of contradictory properties
being attributed to the same things. The response to this is that there is no
contradiction in saying that things have properties in certain relations which they
do not have independently of these relations. For example, if we accept a broadly
Lockean account of the mind-dependence of colour, we can allow that things are
coloured in our experience of them, but are not coloured as they are in themselves.
This does not involve asserting that there are coloured objects that are numerically
21
See Discovery, in which Kant says: ‘The representation of an object as simple is a merely negative
concept, which reason cannot avoid . . . The concept does not, therefore, serve to extend our cognition, but
merely designates a something, so far as it needs to be distinguished from objects of the senses . . . nobody
can have the least knowledge of whether the super-sensible which underlies appearance as substrate is, as
thing-in-itself, either composite or simple’ (Discovery 8: 209n).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
identical with non-coloured objects, but neither does it involve asserting two
distinct objects; rather, the thought is that colour is a property which objects have
only in certain relations they stand in to subjects. Of course, we will still need to see
why Kant thinks he is entitled to assert that things in themselves are not spatial
despite saying that we cannot have knowledge of them; I return to this question in
Chapters 8 and 10.
A final problem about the idea that things in themselves and appearances are in
some sense aspects of the same things is that it might be thought to give us more
knowledge of things in themselves than we should have (in Kant’s terms) and to give
us a way of individuating things in themselves. Kant thinks that the only way we can
individuate objects is at the level of experience. However, to say that the things we
experience have a way they are independently of their appearing to us is not to say
that picking out objects at the level of experience tells us how they are individuated in
themselves. By way of analogy, grant, for the sake of argument, that there is a
distinction between the visual colour we experience and the properties picked out
by science that ground visual colour, such as spectral reflectance properties. Empir-
ical science tells us that the physical properties which ground red, for example, are
enormously varied. This means that picking out redness, at the level of experience,
does not individuate properties at the level of the non-experiential properties physics
studies, even if the former are grounded in the latter.
This point is related to a concern that my view might be thought to commit Kant
to saying that appearances represent things in themselves, whereas he denies this. Not
only does Kant deny that we have any knowledge of things in themselves, further, in
opposing Leibniz, he denies that phenomena should be thought of as unclear or
confused representations of things in themselves, saying, on the contrary, that they
do not represent things in themselves at all. Kant says that it would be wrong to think
that we have confused representations of things in themselves, rather ‘[w]e do not
apprehend them in any fashion whatsoever’ (A44/B62). He argues that, for Leibniz,
‘appearance was the representation of the thing in itself ’ (A270/B327) and he rejects
this view, saying that appearances ‘do not represent things in themselves’ (A276/
B332). The worry is that if we say that things have a way they are in themselves which
grounds the way they appear to us this might be thought to imply, against Kant’s
position, that our experience is some kind of apprehension of things in themselves
and therefore that things in themselves do appear to us.22
In response to this, it is important to see that we can deny that we have represen-
tations of things as they are in themselves without denying that there is a sense in
which we have representations of things themselves. In fact, this is something we find
Kant doing. While he denies that we have representations of things as they are in
22
This objection is developed by Van Cleve (1999: 138).
THINGS IN THEMSELVES WITHOUT NOUMENA 
themselves, including confused ones, he also sometimes says that appearances

represent things, things themselves:
[A]ppearances are only representations of things that exist without any cognition of what they
might be in themselves. (B164)
[T]he word “appearance” must already indicate a relation to something the immediate represen-
tation of which is, to be sure, sensible, but which in itself, without this constitution of our
sensibility . . . must be something, i.e., an object independent of sensibility. (A252, my italics)
Here he says that appearances are the immediate and sensible representation of
something which is an object independent of our sensibility. The passages in which
Kant denies that we represent things in themselves in any way occur in places where
his aim is to reject Leibniz’s account of confused representations of things as they are
in themselves (A44/B61–2; A271/B327; A276/B327). In other words, he rejects the
idea that we have confused representations of the natures things have independently
of their relations to our sensibility; rather, we have no representation, not even
confused ones, of these natures.
Against Leibniz, Kant rejects the claim that sensibility gives us confused represen-
tations of things as they are in themselves—confused representations of things’ non-
relational, intrinsic natures, the way they are independent of their relations to us. He
rejects Leibniz’s view that appearance is the representation of the thing in itself, not
because he is rejecting the idea that appearances represent things—things
themselves—which, as we have seen, he sometimes asserts, but rather because he
rejects the idea that appearances represent things as they are in themselves, the
intrinsic natures of things. Immediately after criticising Leibniz’s view, Kant says that
‘we do not understand through pure reason what the things that appear to us might be
in themselves’ (A276/B332, my italics), and here it is clearly the very things which
appear to us that have a nature in themselves which we cannot know. Appearances
are appearances of things, so there is a sense in which they represent things
themselves and a sense in which they do not—they do not represent things as they
are in themselves.
Rather than saying that things in themselves and appearances are two different
kinds of objects (which, as we have seen, Kant denies) or that they are numerically
identical objects (which is problematic), we can simply say that the things that appear
to us have a way they are in themselves. Like the traditional ‘two-object view’, this
allows that there exists something entirely independent of us which grounds the
spatio-temporal objects of our experience, and which we cannot know. However, like
deflationary interpretations, merely empirical realist interpretations, and traditional
‘one-object’ interpretations, it does not involve asserting that there are intelligibilia,
and it is compatible with thinking that appearances are aspects of things which also
have a way they are in themselves.
In this chapter I have argued that Kant is not a noumenalist: his claim that there is
a way things are in themselves is not a claim that there exist objects of a distinct
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
kind—non-sensible, non-spatio-temporal objects which would be objects for the

understanding alone. I have also argued against those who think that Kant is
ontologically committed only to the spatio-temporal reality that we experience.
Rather, he is committed to the claim that there exists an aspect of reality that we
cannot cognize, that is more metaphysically fundamental than the spatio-temporal
aspect that we experience, and that somehow grounds what we experience. We can
know nothing about the nature of this reality—including that which we would be
required to know in order to understand it in terms of intelligibilia.
Like the view that phenomenalism is the only way of understanding Kant’s
idealism, the idea that a commitment to intelligibilia is the only way of making
sense of his commitment to things in themselves leads to unstable interpretations
and regular swings between extreme views. Deflationary and merely empirical realist
interpreters show that Kant is not committed to an additional kind of object and
conclude that he is not committed to an existing aspect of reality that we cannot
cognize. Noumenalist interpreters show that he is committed to an existing aspect of
reality that we cannot know and take this to show that he is committed to objects of a
different kind from the spatio-temporal objects of our knowledge. Each extreme is
able, relatively easily, to show that the opposite extreme view is untenable; each takes
this to support their own position. To reach a stable interpretation, we need a way of
understanding Kant’s commitment to an aspect of reality of which we cannot have
knowledge which does not involve a commitment to there being intelligibilia and
which makes sense of his talking about things as they appear to us and these same
things as they are in themselves.
4
Against Deflationary
Interpretations
I Introduction
I have argued that Kant is not an extreme phenomenalistic idealist and I have
suggested that his distinction between things as they are in themselves and things
as they appear to us is, in some sense, a distinction between two aspects of things.
Many philosophers writing on Kant agree with this, and many of them would defend
this reading by saying that Kant’s point is to introduce an epistemological or
methodological distinction between two ways of considering things. I refer to such
interpretations as ‘deflationary’ since they aim to take the metaphysics out of Kant’s
distinction between things in themselves and appearances. (It is, of course, compat-
ible with this that such views see Kant as having some metaphysical commitments.)
Deflationary interpreters frequently deny that Kant is committed to the existence of
anything other than the objects of which we have knowledge. They may deny that
Kant is an idealist, although some, such as Allison, aim for positions that account for
Kant’s claim that objects conform to our cognition; therefore they may see their
position as idealist in some sense.1 There are a number of different deflationary
readings. One kind of interpretation sees Kant’s transcendental idealism as a state-
ment of the underdetermination, non-uniqueness, or incompleteness of our theory of
the world.2 Another sees Kant’s point as the limitation of our knowledge to objects
which can be given in experience (Dryer 1966). Alternatively, Kant’s distinction has
been interpreted as being between two perspectives on things (Hoke Robinson 1994)
and between the world and perspectives on it (Matthews 1982). A different kind of
epistemological interpretation, although also one which finds no idealism in Kant, is
that of Abela, on whose epistemic interpretation transcendental idealism is a rejec-
tion of the so-called given (Abela 2002).
A prominent deflationary interpretation, associated most famously with Henry
Allison, is that transcendental idealism concerns ‘epistemic conditions’ (conditions
of representing objects or an objective world) and the correlative idea of considering
1
There is some debate about whether Allison’s position is really deflationary. See Westphal (2001).
2
Walker (1978) and Gordon Nagel (1983) are read in this way by Guyer (1989: 140–1) and Van Cleve
(1999: 4).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
objects in abstraction from these conditions, combined (in Allison’s more recent
work) with an emphasis on the discursivity of our knowledge (the need for both
concepts and intuitions). Allison says that ‘the transcendental distinction, which
constitutes the heart of transcendental idealism, is a bit of metaphilosophical therapy
rather than a first-order metaphysical doctrine’ (Allison 2004: 398). On this view, the
point of saying that we cannot know things in themselves is not to assert that there is
an actual aspect of things which we cannot know, but merely to draw attention to the
idea of there being conditions of cognition and to insist on the possibility of the
abstract thought of objects apart from such conditions.
The motivations for having a deflationary interpretation include Kant’s speaking
of considering things in two ways, his rejection of transcendent metaphysics in
favour of epistemological and metaphilosophical concerns (such as his concern
with a priori conditions of cognition), and the inadequacy of the traditional extreme
metaphysical interpretation. Frequently, deflationary interpretations target their
objections on opponents who see Kant as a noumenalist or a phenomenalist, or
both.3 For example, Allison says that
According to many of its critics, transcendental idealism is a metaphysical theory that affirms
the uncognizability of the “real” (things in themselves) and relegates cognition to the purely
subjective realm of representation (appearances). It thus combines a phenomenalistic, essen-
tially Berkelean, account of what is actually experienced by the mind . . . with the postulation of
an additional set of entities which, in terms of the very theory, are uncognizable. (Allison 2004: 4;
see also 5–9, 46, 51, 54; 2006: 112)
Allison devotes much of his critical discussion to this kind of position. If noumen-
alism and phenomenalism are seen as the alternatives, it is understandable why
deflationary interpretations are appealing, since, as we have seen, both of the former
positions are incompatible with parts of Kant’s text.4 However, we have also seen that
there is considerable textual evidence for the view that Kant holds that spatio-
temporal objects depend on our minds in some sense and that he thinks that there
is a way reality is in itself, which grounds the way things appear to us. In this chapter
I argue against deflationary interpretations and complete laying the ground for a
moderate metaphysical interpretation. In Section II, I examine the deflationary
attempt to understand transcendental idealism through the idea of considering
objects in terms of certain conditions of knowledge and what it means to consider
objects apart from this. I argue that this does not give a satisfactory account of Kant’s
3
Rather than, for example, moderate metaphysical interpretations, such as Ameriks’s (2006; 2003;
2000) position.
4
On the other side, defences of extreme idealist readings are often based on the argument that
deflationary readings fail to accommodate any sense in which Kant is an idealist. For example, Van
Cleve says that reading much contemporary commentary, ‘one can begin to wonder whether Kant’s
transcendental idealism has anything much to do with idealism at all’ (Van Cleve 1999: 4). Similarly,
Guyer argues that attempts to make Kant’s doctrine into a form of epistemic modesty are of no avail in the
face of ‘Kant’s firm announcements that things in themselves are not spatial and temporal’ (Guyer 1987: 334).
AGAINST DEFLATIONARY INTERPRETATIONS 
views of things in themselves or of his view of appearances. I focus on a broadly

Allisonian deflationary interpretation, as his is so important and influential, and the
idea of epistemic conditions comes from him, but I cannot do full justice to his
complex and subtle account in such a short space (and I focus only on his account of
transcendental idealism, not the many other parts of his views on Kant). In
Section III, I argue that Kant needs his notion of things as they are in themselves
to do metaphysical work, and that this requires a commitment to there actually being
a way things are as they are in themselves.
II Epistemic Conditions and Idealism

Allison (1983) argues that transcendental idealism is a ‘theory about the nature and
scope of the conditions under which objects can be experienced or known by the
human mind’ (Allison 1983: 25) and that Kant’s most revolutionary insight is his
realisation that human knowledge has a priori epistemic conditions, such as the
principle of causation, and the fact that we must represent an objective world as
spatial. He says that Kant’s claim that objects must conform to our knowledge must
be understood as saying that objects must conform to the conditions under which we
can represent them as objects. There are three central features of Allison’s interpret-
ation that I look at. First, Allison says that Kant’s distinction between things in
themselves and appearances is methodological—or, more precisely, transcendental
idealism is a methodological ‘standpoint’ (Allison 1983: 25; 2004: 4) that concerns
ways of considering objects, not ways of being. In particular, it involves considering
objects in terms of various epistemic conditions and considering them apart from
such conditions (Allison 1983: 7). Second, as is developed in more detail in his later
accounts, Allison adds to this analysis an emphasis on the discursivity of human
knowledge (Allison 1996; 2004; 2006; 2007), which, roughly, is Kant’s claim that we
need both intuitions and concepts for cognition. Third, Allison explains his inter-
pretation of Kant’s transcendental idealism by contrast with the opposing doctrine of
transcendental realism, as he understands it. He argues that transcendental realism
tries to consider objects from a ‘God’s eye’ view and that part of the point of Kant’s
revolution is that we should stop trying to consider objects in this way. As Allison
sees Kant’s position, the point is that once we realise that for creatures like us there
are conditions of representing things as objective and realise that we can know things
only in terms of such conditions, we will stop asking about what things are like in
themselves, apart from such conditions. I will discuss each of these aspects of his
position in turn, with most emphasis on the influential idea of epistemic conditions.
I argue that Allison’s interpretation is appealing because noumenalism and phenom-
enalism are wrong, because Kant does start with two ways of considering things,
because Kant is concerned with a priori conditions of empirical knowledge, and
because Kant is concerned with understanding the nature of cognition for finite,
receptive creatures like us. However, I argue that the notion of epistemic conditions
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
explains neither Kant’s idealism nor his claim that we cannot cognize things as they
are in themselves and that it appears to do so only if we conflate different under-
standings of such conditions.
The first crucial part of Allison’s interpretation, then, is the notion of an epistemic
condition. This is a notion Allison himself introduces and defines, rather than one
explicitly presented and defined by Kant: Allison acknowledges that, despite the
importance he ascribes to the notion of an epistemic condition for transcendental
idealism, it is never more than ‘implicit’ in the Critique (Allison 1983: 3, 10). He says
that epistemic conditions are conditions which are necessary for our representing
objects and which define ‘object’ or ‘objective’ (Allison 1983: 10; 2004: 11). He further
explains the notion of epistemic conditions negatively, by distinguishing them
from what he calls psychological and ontological conditions. Psychological condi-
tions of knowledge are conditions which provide a genetic account of a belief or an
empirical explanation of the way we perceive things. Ontological conditions are
conditions of the possibility of the being of things (Allison 1983: 10–11; 2006:
115–16). An example of a Kantian epistemic condition is our representation of
space: Kant says that space is necessary for us to represent objects, or to represent
an objective world. Allison also calls such conditions ‘objectivating conditions’.
They are conditions which have a subjective aspect (they are necessary for our
representation or experience) and an objective aspect (they are necessary for us to
represent the world as objective). This distinguishes epistemic conditions from, for
example, the condition that light needs to strike objects in order for us to see them;
this is not part of the content of our representation of objects. So the idea is that
epistemic conditions give us ways we need to represent objects in order to represent
them as objective.
Allison argues that the notion of an epistemic condition enables him to explain
what Kant means by the claim that objects conform to our cognition and also to
explain why we cannot have knowledge of things as they are in themselves. He gives
two explanations of the thought that epistemic conditions involve some kind of
idealism. In his 1983 account, Allison explains this by saying that epistemic condi-
tions reflect the cognitive structure of the (human) mind rather than the object as it is
in itself (Allison 1983: 27; see also 2006: 116). On this account, reflecting the
cognitive structure of mind is understood by contrast with reflecting a structure
that things have in themselves. His more recent work says that the notion of an
epistemic condition ‘brings with it an idealistic commitment of at least an indeter-
minate sort, because it involves the relativisation of the concept of an object to
human cognition and the conditions of its representation of objects’ (Allison 2006:
116; also 2004: 12). I argue that both of these moves are problematic and that neither
succeeds in capturing a kind of idealism. My argument, in a nutshell, is that Allison
appears to get a kind of idealism out of the notion of an epistemic condition only
as a result of conflating two different accounts of such conditions: epistemic
conditions are understood, on one account, as conditions which are necessary for
AGAINST DEFLATIONARY INTERPRETATIONS 
the representation of objects or objectivity and, on the other, as conditions which

reflect the cognitive structure of the mind.
In his earlier work, Allison (1983) says that it is because epistemic conditions
reflect the cognitive structure of the (human) mind rather than the object as it is in
itself that appealing to epistemic conditions gives us a sense in which the objects of
our knowledge must conform to our cognition, and therefore captures some kind of
idealism. His argument goes as follows: he says that since epistemic conditions are, by
definition, conditions which are necessary for the representation or experience of
something as objective, to deny that they reflect the cognitive structure of the mind
involves asserting, contradictorily, that knowledge is possible apart from the condi-
tions under which it is possible. He says that ‘[t]o claim otherwise is to assume that
the mind can somehow have access to an object (through sensible or intellectual
intuition) independently of the very elements that have been stipulated to be the
conditions of the possibility of doing this in the first place’ (Allison 1983: 27). The
problem with this argument is that it simply conflates the two notions of epistemic
conditions. The argument asserts that to deny that epistemic conditions reflect the
cognitive structure of the mind is to assert that knowledge is possible apart from
epistemic conditions. But the notion of an epistemic condition as one which is
necessary for the representation of something as an object or objective state of affairs
is not the same as the notion of an epistemic condition as a condition that reflects
only the cognitive structure of the mind and not the nature of objects in themselves.
The latter is what the argument needs to establish, and it cannot establish it simply by
asserting that we cannot have knowledge of objects apart from the conditions of
representing them as objects.
In his 2004 account, Allison explains the idea that objects must conform to our
knowledge as saying that objects must conform to the conditions under which alone
we can represent them as objects (2004: 37). Either he is claiming that the objects that
we can know conform to the conditions under which we can know them, or he is
saying that all objects must conform to the conditions under which we can know
them. The former has a straightforward realist reading; the latter is not argued for,
and does not follow from the idea of epistemic conditions. Clearly, if there are
conditions objects must meet to be cognized by us, all the objects that can be
cognized by us will meet these conditions. Allison defends the idea that his position
involves a kind of idealism by saying that the idea of epistemic conditions involves
the relativisation of the concept of an object to human cognition and its epistemic
conditions. This could be read as the idealist claim that what counts as an object,
per se, is dependent on epistemic conditions of human cognition. This would capture
idealism, but it would not follow simply from the idea of conditions which are
necessary for us to represent things as objects. It seems that Allison does not make
this idealist claim since he says that ‘the claim is not that things transcending the
conditions of human cognition cannot exist (this would make these conditions
ontological rather than epistemic) but merely that such things cannot count as objects
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
for us’ (Allison 2004: 12). A weaker reading says that the objects we cognize meet the
conditions of our cognizing them, but this does not give us any kind of idealism. For
illustration, suppose we agree that space is a condition of our cognizing objective
particulars and that, as such a condition, it gives content to what we mean by an
objective particular: an objective particular is a spatial object.5 It does not follow from
this that the spatiality of the objects we cognize is relativised to our cognition; it
follows simply that we can cognize only spatial objects. If space is a condition of our
representing objective particulars, all the objective particulars we are able to represent
will be spatial. But here the idea that objects conform to our knowledge is a merely
conditional claim, which is in no tension with realism. It does not imply that objects
depend on our being able to represent them; rather, our being able to represent them
depends on their meeting the condition. To make it plausible that the position
captures some kind of idealism, Allison needs the second reading—the claim that
all objects must conform to space as an epistemic condition. Yet when arguing that
his position follows from the notion of an epistemic condition, he appeals to the first
reading.
Similar points apply to Allison’s account of things in themselves. From his notion
of epistemic conditions Allison gets the result that we cannot know things as they are
in themselves only in the trivial sense that we cannot know things apart from the
conditions of knowing them. This does not give us any reason to think that we
cannot have knowledge of things in themselves if the notion is understood as
referring to mind-independent things. It seems to me that Allison gives the impres-
sion of a more substantial result only if we conflate two different ways in which
‘things considered independently of our being able to cognize them’ can be under-
stood. One way of understanding this is as ‘things as they are to the extent that they
do not meet the conditions of our cognizing them’. Understood in this way, it follows
trivially that we cannot cognize things as they are independently of our being able to
cognize them. It is not even clear that we can give any content to the idea of things as
they are to the extent that they do not meet the conditions of our cognizing them. But
an alternative reading of ‘things considered independently of our being able to
cognize them’ is ‘things considered as they mind-independently are, whether we
can cognize them or not’. Suppose representing space is a condition of our being able
to cognize things. On the first reading, considering a thing as it is apart from our
epistemic conditions is considering a thing to the extent that it is not spatial.
On the second reading, it is considering a thing as it is independently of whether
spatiality is an epistemic condition for us. Spatiality being an epistemic condition
for us is compatible with the thing being spatial, so considering it as it is inde-
pendently of our epistemic conditions does not tell us whether or not it is spatial
in itself. To get a conclusion about the way things mind-independently are
5
This is something that realists have argued (see, for example, P. F. Strawson 1963 and Evans 1985a).
AGAINST DEFLATIONARY INTERPRETATIONS 
from the thought that spatial representation is an epistemic conditions, we would

need a reason to think that whether or not objects meet the conditions of our
representing them somehow depends on these conditions being conditions for us.
This certainly does not follow from the notion of there being conditions that we
need to represent things.
Consider an object which I experience—for example, a table. According to Kant,
one condition of my cognizing a table is that I represent it spatially. Now consider the
table as it is independently of the conditions under which I can cognize it. On the first
reading, what this means is: consider it as it is apart from its being spatial, or to the
extent that it is not spatial (whatever that is supposed to amount to). On the second
reading, it means: consider the table as it is whether or not creatures like us are able to
cognize it, whether or not we have the conditions of cognition that we have. With
respect to the first reading, it is not quite clear how to give content to the idea of
considering a table as it is apart from its being spatial, but it seems fair to say that it is
not spatial, considered as is it is apart from its being spatial. If this is how we
understand the notion of things in themselves (as things considered as they are to
the extent that they do not meet the conditions of our cognizing them), the claim that
objects are not spatial as they are in themselves is true but trivial. However, this tells
us nothing at all about the table as it is independently of us. The table could be spatial
whether or not we are able to cognize it, or independently of whether we are able to
cognize it, since its being spatial does not depend on spatiality’s being an epistemic
condition for us. If we read it in the second way, the claim that spatiality is a
condition of representing objects for us gives us no basis for claiming that objects
are not spatial, considered as they are in themselves, independently of their meeting
the conditions of our cognizing them. In order to get this further claim, we would
need a reason to think that whether or not objects meet the conditions of our
representing them as objects depends on whether or not these are conditions for
us. This is not a trivial claim and does not fall out of the notion of there being
epistemic conditions. A realist could hold that space is an epistemic condition of our
representing objects, and that objects are spatial independently of our being able to
cognize them, because their meeting the conditions of our cognizing them does not
depend on these conditions being conditions for us.
It seems that Allison must intend for his account to be read in the first way (as
things considered as they are to the extent that they do not meet the conditions of
our cognizing them), since he says that considering objects as they are in them-
selves leaves us with a notion of them that is cognitively vacuous (Allison 2004: 56).
He says that
in considering things as they appear, we are considering them in the way they are presented to
discursive knowers with our form of sensibility. Conversely, to consider them as they are in
themselves is to consider them apart from their epistemic relation to these forms or epistemic
conditions, which, if it is to have any content, must be equivalent to considering them qua
objects for some pure intelligence, or “mere understanding”. (Allison 2004: 16–17)
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
In order to say that the notion of considering things as they are apart from their
epistemic relation to us could have no content other than the idea of considering
them as objects for a pure intelligence, Allison must be assuming the first of the two
readings I have given of the idea of considering things apart from our cognizing
them. On this reading, thinking of things as they are in themselves is thinking of
them as they are to the extent that they are not spatio-temporal and sensible. It is very
difficult to give any content to this way of considering things, and perhaps Allison is
right that the best way of doing so is by thinking of things as objects for a pure
intelligence (God). But it does not follow from this that we can give no content to the
idea of things as they are independently of our cognizing them.6 The problem for
Allison is that the realist notion of things which have a way they are independently of
us is not ruled out by the idea of epistemic conditions or by the correlative claim that
we cannot know objects apart from their meeting epistemic conditions. Allison says
that epistemic conditions ‘condition the objectivity of our representations of things
rather than the very existence of the things themselves’ (Allison 2004: 11). But this
means that things could meet these conditions, independently of us and independ-
ently of what the conditions of our representing objects are. If it is true that space is
necessary for us to represent objects, this tells us something about all the objects that
we can represent. Thus, this epistemic condition would limit the notion of an object-
of-human-knowledge to spatial objects, but it does not follow from this that these
objects are not spatial independently of our experiencing them; nor does it follow
that spatiality is thereby relativised to human knowledge. To get a substantial
conclusion from the idea of epistemic conditions, we need some reason to think
that whether or not things meet the conditions of our cognizing them somehow
depends on the fact that these conditions are conditions for us. But Allison’s idea of
epistemic conditions is just the idea of conditions we need to represent things as
objects, and he has no argument for thinking that the further claim follows merely
from there being conditions of representing objects.
It seems to me that these conflations makes the Allisonian position sometimes
difficult to pin down. Claims which look innocuous when they trade on the first
6
Similarly, Allison says that there is an ambiguity in the notion of considering something as it is in
itself: he says that on an ontological reading, to take a thing as something that exists in itself is to take it as a
substantia noumena, whereas in an epistemological sense, it is ‘considering it as it is independently of its
epistemic relation to human sensibility and its conditions’ (Allison 2004: 52). The ambiguity I have been
discussing concerns how to understand the latter half of this. In terms of the former, Allison is not entitled
to equate the straightforward realist notion of a thing which exists independently of us with that of a
substantia noumena which is an object for a pure intelligence, and this way of considering things in
themselves does not follow from his account of epistemic conditions, and the idea of thinking of things
apart from such conditions. When Locke regards knowledge of primary qualities as knowledge of objects as
they are in themselves, he is not talking about knowing them as supersensible objects in the way that they
would be known by an intuitive intellect; rather, he is talking about knowing them as they are independ-
ently of us. Allison might respond to this that Locke does not have the notion of epistemic conditions or of
Kant’s Copernican Revolution. But Locke is concerned with understanding us as finite knowers, and, as
Allison admits, Kant does not use the term ‘epistemic conditions’ either.
AGAINST DEFLATIONARY INTERPRETATIONS 
reading are taken to establish conclusions that depend on the second. This is brought
out in the very clear account of the deflationary position presented by Angela
Breitenbach (2004). She explains that deflationary readings such as those of Prauss
(1974), Allison (1983), and Buchdahl (1992) circumvent the traditional worry that
there is supposed to be some tension in claiming both that there is a way things are in
themselves and that we cannot know things as they are in themselves. She argues that
the claim
that there exist things in themselves, thus has to be understood as no more than the claim that
there are things which can be thought of in abstraction from the way they appear to us.
(Breitenbach 2004: 143)
And she says that
On the deflationary interpretation, the problem of the thing in itself vanishes: all that can be
said to exist is the object as we find it in the world. The thing in itself understood as the
transcendental object cannot be said to be a real existing entity. (Breitenbach 2004: 142)
She says that the problem of the thing in itself vanishes because things in themselves
‘are objects considered in abstraction from the conditions of knowing them, and this
is why we cannot know anything about them’ (Breitenbach 2004: 143–4). It is of
course trivially true that we cannot know anything about objects as they are con-
sidered in abstraction from their meeting the conditions of our knowing them.
However, this claim does not entitle us to assert that all that can be said to exist is
the object as it appears: it does not give a basis to reject the claim that there is
something that exists independently of us. To get to this further claim, we need to
equate the thought of the way things are independently of our cognizing them with
the thought of the way they are apart from the way we cognize them as being. Once
this conflation has been made, the Allisonian position looks not just appealing, but
unavoidable: clearly we cannot cognize anything about objects in abstraction from
their meeting the conditions of our cognizing them, and it would be just senseless to
assert that things as they are when we subtract from them what we know about them
in our cognition are as they are in our cognition of them. But without this conflation,
it is an open question whether things as they are independently of our experiencing
them are as we experience them to be. Deflationary interpreters want to avoid
asserting that our experience gives us only a partial understanding of something
which has other properties that we cannot cognize, but this is exactly what Kant
asserts: ‘what objects may be in themselves would still never be known through the
most enlightened cognition of their appearances, which is alone given to us’ (A43/
B60); ‘our cognition reaches appearances only, leaving the thing in itself as something
actual for itself, but uncognized by us’ (Bxx).
I have argued that the sense in which Allison is able to establish that objects are not
spatial independently of us is trivial: they are not spatial considered as they are
independent of their spatiality. The triviality objection has been made by a number of
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
commentators.7 Kant thinks that transcendental idealism as a whole is a revolution-

ary doctrine, which will enable him to solve previously irresolvable philosophical
disputes; thus trivialising is a serious objection to any interpretation of the position.
Perhaps ironically, the Allisonian position is defended against this accusation by one
of its opponents, Strawson, who argues that, on one reading of transcendental
idealism:
we are merely offered the cautious and surely legitimate reminder that human knowledge
cannot exceed the bounds of human cognitive capacities. It would be unfair to say that the
doctrine then reduces to the tautology that we can know of things only what we can know of
them, for this would be to ignore the brilliant and largely persuasive demonstration of the
necessary structural features of human knowledge and experience which makes the first
Critique a work of unique philosophical importance. (Strawson 1994)8
Graham Bird also defends Allison’s view against the charge of triviality. He argues
that the point is not that ‘we can have no knowledge of things in abstraction from the
conditions of knowledge’ but that ‘we can have no knowledge of things in (complete)
abstraction from the senses’, which is neither trivial nor analytic (Bird 2000: 107).
Two further responses to the objection of trivialising are given by Allison, who argues
that the charge misses the point that Kant’s limitation of knowledge to appearances is
supposed to be therapeutic, and that the fact ‘that a conclusion follows analytically,
once a distinction is in place, does not render it trivial’ (Allison 2004: 19).
As these defences of Allison’s position correctly point out, there is much in his
position which is not anodyne, such as the account of the epistemic conditions which
structure our knowledge. However, his account does trivialise Kant’s reasons for
thinking that we cannot know things as they are in themselves. What is right about
the deflationary position is that Kant thinks that our idea of things in themselves is, in
a sense, empty: we can give it no positive content. But this is because we cannot
cognize things as they are in themselves. In contrast, the deflationary reading
introduces the notion of things in themselves as an analytically empty notion, and
then posits this, trivially, as an explanation of why we cannot have knowledge of
So far I have focused on Allison’s idea of epistemic conditions, and the way this is
supposed to explain Kant’s idealism and his claim that we cannot know things in
themselves. There are two further parts to his account: the discursive nature of our
cognition and Kant’s rejection of transcendental realism. In his 2004 account, Allison
argues that the notion of epistemic conditions gets us only a ‘generalised kind of
idealism’, and that we need to bring in Kant’s commitment to the discursivity of our
knowledge to get the full position. What Allison calls the discursivity thesis says that
7
Van Cleve argues that Allison makes Kant’s key claims turn out to be tautologies (Van Cleve 1999: 4),
and asks: ‘if transcendental idealism is a tautology, why did Kant write such a long book defending it?’
(1999: 8; see also Guyer 1987, Ameriks 1992: 334, and Langton 1998: 9–10).
8
See also Wood (2007: 8).
AGAINST DEFLATIONARY INTERPRETATIONS 
we have both a receptive sensibility and a spontaneous understanding—in other

words, that human cognition requires both concepts and sensible intuitions (Allison
2004: 13; 2007: 34). Allison is clearly right that the claims that we require both
concepts and intuitions for cognition, and that we can have (empirical) intuitions
only as a result of being affected by objects, are crucial parts of Kant’s system and, as
I will argue in Part Two, central to his idealism. But if the notion of an epistemic
condition, on its own, does not give us any kind of idealism, then it is hard to see how
the idea that we cannot have knowledge without both concepts and intuitions could
do so. There is no obvious reason why realists and empiricists could not agree that
sensibility requires ordering of the given data and that we do not know an object until
we have conceptualised it.9 Allison says that Kant’s views about discursivity involve
three epistemological assumptions: one, cognition requires that objects are given;
two, for us this requires affection; three, sensible intuition is not enough for cognition
so we need to bring objects under concepts. All three of these claims could be
accepted by realists. Kant’s idealism is based on the further claims that being given
objects in sensible intuition requires an a priori form of intuition, that empirical
concept application requires a priori concepts, and, further, that the a priori forms of
intuition do not present us with things as they are in themselves. It is these further
steps that get us to his transcendental idealism, not the idea that cognition requires
both concepts and intuitions.
The third part of Allison’s position is his appeal to the notion of transcendental
realism, understood as a God’s eye view, in order to characterise, by contrast,
transcendental idealism. I have argued that transcendental idealism is a metaphysical
position, which includes a commitment to an existing aspect of reality which we
cannot cognize and which claims that the things that we experience are mind-
dependent in some sense. Allison would, I think, object that seeing Kant’s position
in this way involves illegitimately approaching it from the point of view of tran-
scendental realism. Similarly, where I have argued that the idea that there are
conditions which are necessary for the representation of objects does not entail
that objects could not meet such conditions as they are independently of our
experiencing them, Allison might object that to ask how things are independently
of our being able to cognize them is to ask an illegitimate, transcendentally realist
question. Allison defines transcendental realism as a view which assumes that
‘human knowledge is to be measured and evaluated in terms of its conformity (or
lack thereof) to the norm of a putatively perfect divine knowledge’ (Allison 2006:
114). Further, he sees it as defined by a failure to recognise the discursivity of our
knowledge: the fact that we need both intuitions and concepts (Allison 2006: 118).
I argue that the few texts in which Kant talks about transcendental realism do not
support Allison’s case.
9
See Westphal (2001: 596), who argues that the discursivity thesis is available to a realist. Like Allison,
Prauss also seems to assume that the discursivity thesis is not available to a realist (Prauss 1974: 63, 67).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
Kant has very little direct discussion of transcendental realism; he nowhere talks
about it in terms of putatively divine knowledge. In the two main texts in the Critique
in which he talks about it, he writes:
I understand by the transcendental idealism of all appearances the doctrine that they are all
together to be regarded as mere representations and not as things in themselves, and accord-
ingly that space and time are only sensible forms of our intuition, but not determinations given
for things in themselves. To this idealism is opposed transcendental realism, which regards
space and time as something given in themselves (independent of our sensibility). The
transcendental realist therefore represents outer appearances (if their reality is conceded) as
things in themselves, which would exist independently of us and our sensibility and thus would
also be outside us according to pure concepts of the understanding. It is really this transcen-
dental realist who afterwards plays the empirical idealist; and after he has falsely presupposed
about objects of the senses that if they are to exist they must have their existence in themselves
even apart from sense, he finds that from this point of view all our representations of sense are
insufficient to make their reality certain. (A369)
or in time, hence all objects of an experience possible for us, are nothing but appearances, i.e.,
ations, have outside our thoughts no existence grounded in itself. This doctrine I call tran-
scendental idealism. The realist, in the transcendental signification, makes these modifications
of our sensibility into things subsisting in themselves, and hence makes mere representations
into things in themselves. (A490–1/B518–19)
While these passages may not dictate a particular interpretation, they are straight-
forwardly compatible with a metaphysical interpretation of transcendental realism,
and there is nothing in them that forces us to see Kant’s opposing position as ‘an
alternative to ontology’ (Allison 2006: 123) rather than as a position which itself has
metaphysical commitments. First, notice that the texts seem to indicate the mind-
dependence of appearances pretty clearly: Kant says that appearances are mere
representations, modifications of our sensibility, which have no existence apart
from our thought. Second, the key error that Kant attributes to transcendental
realism is that of confusing appearances with things themselves; a straightforward
way of reading this is to say that the transcendental realist thinks that things as they
appear to us—what we experience—are things as they exist independently of our
experiencing them. In this sense, both empiricists and rationalists could be transcen-
dental realists, not because they need to be committed to thinking that we cognize
objects with an intuitive intellect or from a God’s eye point of view, but rather to the
extent that they think that the objects we are presented with in experience are the way
they are independently of our experiencing them.
Allison argues that things as they are in themselves are things as thought of in
abstraction from the conditions of our having knowledge of them. I have argued that
there is a sense in which this is right and in which there is clear textual support for it:
Kant says that the notion of a noumenon in the negative sense is ‘a thing insofar as it
AGAINST DEFLATIONARY INTERPRETATIONS 
is not an object of our sensible intuition, because we abstract from the manner of
our intuition of it’ (B307). However, while Kant does think we have a merely abstract
conception of things as they are in themselves (we have no positive, substantial
characterisation of them), he does not take this to imply that there is nothing actually
corresponding to this notion. He does not argue that we are unable to cognize things
as they are in themselves merely because the notion of things in themselves is the idea
of things considered as they are apart from our cognition. Rather, he starts with the
idea of things as they are independently of us and argues that we cannot cognize their
mind-independent natures and can cognize them only as they are in relation to our
minds. Since we can cognize only aspects of things that they have in relation to us, all
we are left with in terms of the idea of things as they are in themselves is the idea of
things as they are apart from the way we cognize them. We are committed to there
being a way things are as they are in themselves, but we have no determinate
representation of this way things are, so we have a merely negative conception of
it. Starting with the idea of two ways of considering things, and the idea of epistemic
conditions as conditions which are necessary for cognition of an objective world,
we reach an account according to which that of which we have experience
comprises only mind-dependent appearances, which are grounded in something
mind-independent, of which we cannot have knowledge.
III The Role of Metaphysics in the Dialectic

Transcendental idealism plays a central role in Kant’s attack on traditional, tran-
scendent metaphysics in the Dialectic. The fact that attacking a certain kind of
metaphysics is such an important part of Kant’s project in the Critique might be
taken to support the view that transcendental idealism should not be understood as
making any metaphysical claims. However, I argue that the way Kant invokes
transcendental idealism in the Dialectic in fact requires it to include some metaphys-
ical commitments, and that merely distinguishing between considering objects in
terms of the conditions of knowledge and apart from these conditions simply cannot
do the work he wants transcendental idealism to do, and so cannot be how Kant
intends his transcendental idealism to be understood. His diagnosis of what goes
wrong in transcendent metaphysics in fact requires a commitment to an actual,
ontologically fundamental, ground of appearances and requires that appearances
(spatio-temporal, empirically real objects) are not ontologically fundamental. This is
crucial to make sense of the ways in which he thinks appearances can be incomplete
or indeterminate.
In the Dialectic, Kant wants to show that the attempts of rationalist metaphys-
icians to attain a priori knowledge of things which are not given in experience, such
as Cartesian souls, the world as a whole, and God, can never succeed. As well as
arguing that such knowledge is not possible, Kant aims to diagnose the features of
our thought which, he argues, mistakenly but unavoidably lead us to think that we
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
can have knowledge of such things. He thinks that the nature of our reason gives rise
to an illusion which leads us into erroneous metaphysical conclusions, and that
although we cannot avoid the illusion, we can avoid making the erroneous conclu-
sions. He also thinks that the very features of reason which lead to metaphysical
illusion play a necessary role in empirical knowledge.10
Kant thinks that an essential and inescapable feature of reason is that reason seeks
(what he calls) conditions for every conditioned. For everything caused, contingent,
or not entirely self-explanatory, reason seeks for an explanation, ground, or cause of
its existence and its nature; without this we would not have a coherent use of reason.
He says that transcendental illusion arises from the fact that we mistake the
(unavoidable) principle which tells us to seek for the conditions for every conditioned
for an objective claim about the nature of the world: the claim that there is
a condition for every conditioned. Further, Kant considers the claim that there is a
condition for every conditioned to be equivalent to saying that there is something
unconditioned, something with respect to which no further explanation or ground
could be asked for (A307–8/B364). This is why (as Allison argues) the demand for
conditions for every conditioned (the demand for the unconditioned) can be under-
stood as a requirement for ultimate, complete explanation (Allison 2004: 381; see also
Grier 2001).11 Therefore, Kant argues, our natural and unavoidable dissatisfaction
with explanatory incompleteness misleads us into thinking that we have insight into
the existence of the unconditioned (which would be required for explanatory
completeness).
Kant thinks that reason’s search for a condition for every conditioned plays a role
in empirical knowledge because, roughly, the demand for an explanation for every-
thing conditioned drives science. The particular form this takes in science is the
demand for systematic unity: a theory that explains everything, in a unified, con-
nected way. However, not only do we lack an entitlement to assert that this is possible
(to take reason’s demand as insight into reality), further, Kant thinks that asserting
that there is something unconditioned in the world of our experience (that ultimate
explanatory completeness is possible with respect to the spatio-temporal world, or
that the spatio-temporal world exists as a complete whole) generates paradoxes or
antinomies. He takes these paradoxes to show that the completeness reason seeks
cannot be taken to be true of the spatio-temporal world of our experience. Kant
10
My discussion here is much indebted to the excellent discussion in Grier (2001). While she has a
deflationary reading of transcendental idealism, it seems to me that much of her analysis of transcendental
illusion, the errors involved in rationalist metaphysics, and the role of the ideas of reason in empirical
knowledge is compatible with a more metaphysical interpretation of transcendental idealism (although
perhaps not with a noumenalist and phenomenalist interpretation).
11
Grier argues that Kant identifies transcendental illusion ‘with the propensity to take the subjective or
logical requirement that there be a complete unity of thought to be a requirement to which “objects”
considered independently of the conditions of experience (things in themselves) must conform’ (Grier
2001: 8; A297/B353).
AGAINST DEFLATIONARY INTERPRETATIONS 
thinks that once we realise that we have knowledge only of appearances, and not
things in themselves, we can explain both how it could be that systematic unity is not
actual in the world of our experience and why we seek it. Crucially, Kant’s explan-
ation turns on the claim that the spatio-temporal world is not metaphysically
fundamental: this is how we can make sense of the claim that it does not exist as a
complete, systematically unified totality. Kant says that the idea that there is a
condition for every conditioned would be justified if the objects of our knowledge
were things as they are in themselves, but we can avoid the claim if the objects of our
cognition are only appearances.
The part of the Dialectic which relates most clearly to transcendental idealism is
the section on the Antinomies and, in particular, the first two Antinomies. The
Antinomies are four pairs of arguments for contradictory conclusions, in which we
seem to be committed to each side of the pair, and thus to be committed to
contradictions. The Antinomies arise, Kant thinks, when we try to think about the
spatio-temporal (empirically real) world as a whole. In the first Antinomy we have an
argument which shows that the world must have boundaries in space and time and
an argument which shows that it cannot have such boundaries. In the second
Antinomy, we have an argument which shows that matter must be composed of
simple parts and an argument which shows that it cannot be composed of simple
parts. Kant says that the Antinomies provide indirect support for transcendental
idealism, since positing transcendental idealism enables us to dissolve these conflicts.
My concern here is not to examine Kant’s arguments for the Antinomies, or to assess
whether he demonstrates that the Antinomies are unavoidable paradoxes, but rather,
to see how he thinks transcendental idealism is supposed to resolve these conflicts.
This gives us considerable insight into how Kant understands transcendental ideal-
ism. I argue that if we pay attention to the way Kant introduces transcendental
idealism in his solution to the Antinomies, as well as to the philosophical work which
he requires the position to do in this section, we will see that both deflationary and
merely empirically realist accounts cannot do justice to it.
The question at issue here is what interpretation of transcendental idealism makes
sense of the claim that these conflicts of reason can be resolved by seeing that the
objects of experience are not things in themselves. The first thing to pay attention to
is the way Kant describes transcendental idealism in his introduction to the reso-
lution of the Antinomies. The section entitled ‘Transcendental Idealism as the Key to
solving the Cosmological Dialectic’ begins as follows:
or in time, hence all objects of an experience possible for us, are nothing but appearances, i.e.,
ations, have outside our thoughts no existence grounded in itself. This doctrine I call tran-
scendental idealism. The realist, in the transcendental signification, makes these modifications
of our sensibility into things subsisting in themselves, and hence makes mere representations
into things in themselves. (A490–1/B518–19)
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
Kant immediately goes on to distinguish this view from empirical idealism, which
denies the existence of objects in space. He insists that he allows for the reality of
spatial objects, but says that
rather nothing but representations, and they cannot exist at all outside our minds; and even the
inner and sensible intuition of our mind (as an object of consciousness), the determination of
which through the succession of different states is represented in time, is not the real object as
it exists in itself, or the transcendental subject, but only an appearance of this to us unknown
being, which was given to sensibility. (A492/B520)
He also says:
[T]he objects of experience are never given in themselves, but only in experience, and they do
not exist at all outside it. (A492–3/B521)
The non-sensible cause of these representations is entirely unknown to us, and therefore we
cannot intuit it as an object. (A494/B522)
The cause of the empirical conditions of this progress, the cause, therefore, of which members
of it I might encounter and also the extent to which I may encounter them in the regress, is
transcendental, and hence necessarily unknown to us. (A496/B524)
At the crucial point where Kant is about to appeal to transcendental idealism to resolve
the Antinomies, and is describing it for this purpose, he clearly both distances himself
from positions which allow secure knowledge only of inner experiences (as do phe-
nomenalist interpretations), but also asserts that the spatio-temporal objects of our
experience are mind-dependent, and he also clearly implies that there is something
which exists apart from mind-dependent appearances as their cause or ground.
The next thing to look at is how the resolution is supposed to work: how this
transcendental idealist position is supposed to enable us to resolve the Antinomies.
Crucially, Kant says that the idea that for every conditioned there must be a condition
is something which would be true of the objects of our knowledge if they were things
in themselves. He says that it is a conceptual truth that ‘if the conditioned is given,
then through it a regress in the series of all conditions is given to us as a problem’
(A498/B526). He also says that ‘[i]f the conditioned as well as its conditions are
things in themselves, then when the first is given not only is the regress to the second
given as a problem, but the latter is thereby already given along with it’ (A498/B526;
see also A506/B534 and A535/B563). In terms of the ways of understanding things in
themselves we have canvassed so far, there are at least three different things that
could be meant by this. One, following the deflationary interpretation, is that if the
conditioned and its conditions are thought of in abstraction from the conditions of
knowledge, completeness of the series of conditions would be given. The second,
which is my reading, is that if the conditioned objects of our knowledge were entirely
independent real things, they would exist as a complete or determinate totality. The
third, noumenalist, reading says that Kant’s point here is to say that if the conditions
AGAINST DEFLATIONARY INTERPRETATIONS 
of the conditioned are intelligibilia, the completeness of the series of conditions must
be given. I submit that the first and the third are not plausible readings of what Kant
has said, nor do they fit with the way he goes on to explain his solution. This counts
strongly against both deflationary and noumenalist interpretations.
The third reading seems implausible since Kant says that intelligibilia are a kind of
object of which we do not even really understand what they are. It would therefore be
strange to find him making positive claims about intelligibilia such as that complete-
ness in the series of conditions of such things must be given. The first reading also has
Kant making an unmotivated claim: why should abstracting from the conditions of
knowledge force us to say anything about whether or not the unconditioned must be
thought of as given? Further, since, according to most deflationary readings, abstract-
ing from the conditions of our cognition leaves us with an entirely cognitively empty
thought, this would give us no basis for making such claims as that things in
themselves exist as a totality, contain something unconditioned, and are completely
determinate (A571–6/B599–604).12
On the second reading, Kant starts with the thought of completely mind-independent
reality and makes a very general claim about it: whatever it is like, it is complete
and determinate. He does not appear to offer argument for this claim, but he does
assert it. He says that ‘if the world is a whole existing in itself, then it is either finite
or infinite’ (A506/B534; see also A498–9/B526–7 and A504/B532). That this is what
he is saying is supported by what he immediately goes on to say, in contrast, about
the status of appearances. He says that
on the contrary, if I am dealing with appearances, which, as mere representations are not given
at all if I do not achieve acquaintance with them . . . then I cannot say with the same meaning
that if the conditioned is given, then all the conditions (as appearances) for it are also given.
(A498–9/B526–527)
Here, Kant says that what explains how it can be that appearances are not complete is
their mind-dependence: appearances, as ‘mere representations are not given at all if
I do not achieve acquaintance with them’. The thought is that mind-independent
reality must be complete; this explains why, when we assume that the objects of our
cognition are mind-independent, we assume that they exist as a complete totality.
However, this (Kant argues) leads to contradictions when we think about the extent
of the world in space and time. We can avoid the contradiction if the spatio-temporal
objects of our cognition do not exist as a complete totality, and what enables us to
make sense of how this could be is seeing that they are not mind-independent and
metaphysically fundamental. Kant does not give much argument for the claim that
what exists entirely independently must be determinate in magnitude, but it is clearly
12
See Watkins (2005: 313). As Colin Marshall (personal correspondence) points out, abstracting gives
you less, not more. It should not enable us to go from something that is not a complete totality, and does
not contain the unconditioned, to something that does.
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
his view (A571–6/B599–604). He thinks that it is only because we are dealing with
mind-dependent appearances of things, which exist only in the possibility of their
appearing to us, that we can make sense of the claim that they do not exist as a
complete totality. Crucially, what enables us to make sense of the claim that the world
of sense has no absolute magnitude (A521/B549) is precisely the claim that ‘appear-
ances are nothing outside our representations’ (A506/B534). It is the dependence of
appearances on our finite minds and the specific way in which they are mind-
dependent which are necessary for Kant’s solution.
This account of Kant’s general position about the unconditioned nature of mind-
independent reality matches the specific explanations Kant gives of the resolution of
the first and second Antinomies. Kant claims that when we assume that the spatio-
temporal objects of our knowledge exist independently of us we come up with
conclusive arguments for the claim that ‘the world is finite in magnitude’ and the
claim that ‘the world is infinite in magnitude’ (A504/B532; A506/B534). This contra-
diction is the problem that needs to be resolved. Kant thinks that if the spatio-
temporal world existed entirely independently of us, it would have a determinate
extent in space and time, because it would exist as a determinate, complete totality.
However, once we realise that the world of which we have experience does not exist
independently of our being able to cognize it, we can explain how it could be that it
does not exist as a determinate totality. We cannot cognize it as a determinate totality
(either a finite or an infinite totality) and therefore it does not exist as one. Both in
respect of its boundaries in space and time, and in respect of its divisibility, Kant says
that ‘the world of sense has no absolute magnitude’ (A521/B549). He says that
because ‘the world does not exist at all (independently of the regressive series of
my representations), it exists neither as an in itself infinite whole nor as an in itself
finite whole’ (A505/B533).13 Kant thinks that we can explain how it can be that the
world of sense has no determinate magnitude once we realise that the empirically real
world is not (entirely) independent of our experience of it. It is because our cognition
is of appearances only that we are not forced to say that the world is ‘determined in
itself regarding its magnitude’ (A504/B532).
It is hard to see how Kant’s rejection of the claim that the unconditioned can be
said to exist in appearances, as well as his assertion that it would be found in things in
themselves, can be explained merely in terms of the idea of considering things as
objects of knowledge and considering things in abstraction from what we know about
them.14 None of Kant’s discussion of the resolution of the Antinomies speaks about
13
Kant does claim that space and time are infinite wholes and that space is a ‘totum’ rather than a
composite (A438/B466); however, Kant defines a totum as that where the parts are possible only in the
whole, and this is also what Kant emphasises in the Metaphysical Expositions in the Transcendental
Aesthetic. This sense of ‘totum’ does not require full determinacy, as an indeterminate whole could be prior
to its parts. Thanks to Colin Marshall (personal correspondence) for this point.
14
As Aquila argues, ‘the incompatibility of the two ways of considering things must itself stem from
some fact about things themselves’ as the transcendental realist’s mistake is not ‘considering things
AGAINST DEFLATIONARY INTERPRETATIONS 
epistemic conditions and Allison himself, in his discussion of the Antinomies, does
not appeal directly to the notion of epistemic conditions and is primarily concerned
with Kant’s rejection of the idea that the unconditioned can be said to exist in
appearances. However, Allison’s account of the empty thought of things considered
as they are independently of their meeting the conditions of our cognizing them
offers us no explanation of what Kant says about the necessary postulation of the
unconditioned in things in themselves. Since the thought of things in themselves is
supposed to be empty, it is hard to see how it could commit us to the idea that things
considered in this way must be a determinate totality and must be fully determinate
in their properties. And the idea that we can know things only in terms of certain
conditions of knowledge does not explain Kant’s claims about the indeterminate
extent of the empirically real world.
The idea that Kant has a metaphysical commitment to an unknown reality
which grounds the world we experience, and to the objects of our experience
being (partly) mind-dependent, also provides the most straightforward reading of
his (admittedly not terribly clear) solution to the third Antinomy. Since this
notoriously complex and controversial section raises many problems that cannot
be dealt with here, I will simply make a few comments. Kant’s aim in the third
Antinomy is to demonstrate that a strong kind of (libertarian) freedom of the will
is coherent, and that nothing we know about the world can be taken to rule out the
possibility that we have it. The metaphysical nature of his account of freedom of
the will is suggested by the fact that his conception of freedom involves a kind of
causality that is different in certain crucial respects from causality in accordance
with empirical laws (A533/B561). Further, his solution, which is supposed to
enable us to show that this other kind of causality is possible (although we cannot
know it to be actual) depends on the possibility of an intelligible cause which is not
a part of the empirically real world. This possibility, in turn, can be explained
because the empirically real objects of our knowledge are appearances (as opposed
to completely mind-independent things) that depend on some mind-independent
ground. He says that
since these appearances, because they are not things in themselves, must be grounded in a
transcendental object determining them as mere representations, nothing hinders us from
ascribing to this transcendental object, apart from the property through which it appears, also
another causality that is not appearance, even though its effect is encountered in appearance.
(A539/B56715)
considered in one way not in that way’, but ‘supposing that objects in space and time . . . are things that exist
in themselves’ (Aquila 1983: 90). Similarly, Ameriks argues that the claim that human knowledge is
governed by certain conditions will not account for ‘Kant’s own stronger conclusion, which is that there
are objects which in themselves have genuine ultimate properties that do not conform to those conditions’
(Ameriks 1992: 334).
15
See also Groundwork 4: 453–60, CPR 5: 95–8, and Ameriks (2006: 110).
 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM
Rather than merely invoking the idea of two ways of considering objects, or the idea
of conditions of knowledge, Kant says that since appearances are mere representa-
tions, they must have grounds that are not appearances. It is because appearances are
not metaphysically complete or ultimate things, but must rather be grounded on
something which exists in itself, that we can allow for the possibility that there is a
kind of causality in the world in itself which is distinct from the causality of empirical
law. Allison sees Kant’s solution to the third Antinomy as the claim that there is no
fact of the matter as to whether we are free or not free; rather, claims about
determinism and claims about freedom are both relativised to a point of view
(Allison 2006: 121).16 But the place where Kant denies that there is a fact of the
matter is with respect to the extent of the world in time and space. With respect to
freedom, he wants to show that it is (at least logically) possible that there is another
kind of causality than the causality that falls under empirical laws. He says that
if the effects are appearances, is it also necessary that the causality of their cause, which (namely
the cause) is also appearance must be solely empirical? Is it not rather possible that although
for every effect in appearance there is required a connection with its cause in accordance with
laws of empirical causality, this empirical causality itself, without the least interruption of its
connection with natural causes, could nevertheless be an effect of a causality that is not
empirical, but rather intelligible? (A544/B572; A536–7/B564–5; see also A538/B566)
Kant’s proposed solution here is extremely difficult to understand17 but it certainly

seems to invoke an aspect of reality which is independent of our experience of it.
As I will argue in Chapter 10, a similar argument to that I have made with respect
to the Antinomies can be made with respect to the section called the ‘Amphiboly to
the Concepts of Reflection’ (A260–92/B316–49) and, in particular, with respect to
Kant’s discussion of the claim that relations require something non-relational,
although the point may be more controversial here. The Amphiboly, which follows
straight after the section on the distinction between noumena and phenomena, is not
straightforward to read because in it Kant is discussing Leibniz’s views in a way which
may sometimes fail to make it entirely clear what he is asserting and what is merely
an account of a Leibnizian argument he rejects. However, throughout the section he
clearly states that it is a conceptual truth that what is relational requires something
non-relational and he also repeats the view (introduced in the Aesthetic) that
appearances contain only relations. He says that ‘through mere concepts, of course,
I cannot think of something external without anything inner, for the very reason that
16
As Watkins points out, a problem with this solution is that it leads one to ask what the nature of the
difference between the standpoints that precludes the possibility that they could be held at the same time is.
(Watkins 2005: 321)
17
Watkins explains one way of making sense of the idea that things could be determined as appearances
but not determined in themselves while appearances are dependent on things in themselves: he says that it
could be that how things are in themselves (including the choices of noumenal subjects) is responsible for
which deterministic laws are true in the world of appearances (2005: Ch. 5).
AGAINST DEFLATIONARY INTERPRETATIONS 
relational concepts absolutely presuppose given things and are not possible without
them’ (A284/B340). He denies that we can cognize anything other than relations
(something absolutely inner) in the objects of our experience and he says that the
necessity of positing something non-relational, or something absolutely inner, does
not apply to the objects of our experience. But he thinks that this is not something it
would make sense to say of entirely independent reality. He says that it is ‘startling’ to
be told that objects consist of nothing but relations, but that this can be explained
once we realise that the things we are talking about ‘consist in the mere relation of
something in general to the senses’ (A284–5/B341). As in the Antinomies, Kant
makes claims about the objects of our experience which he does not think it makes
sense to make about entirely independent things. He also holds that we can make
sense of this precisely by seeing that the objects of our experience are not entirely
independent and that they are grounded on something that is.
IV Conclusion to Part One

We have seen that there is substantial textual evidence that Kant thinks spatio-
temporal objects (appearances) are mind-dependent, and that he thinks there exists
an aspect of reality that we cannot cognize. Transcendental idealism is not just an
epistemological or methodological position. Kant argues that appearances exist in us,
that they are mere representations, that their existence requires a connection to
possible perception, and that the way things are in themselves grounds appearances.
He frequently asserts that there is a way things are in themselves and his critique of
transcendent metaphysics requires such an aspect of reality, as well as requiring that
appearances are not metaphysically fundamental. However, I have argued that Kant
is not committed to there being intelligibilia. I have also argued that there are a
number of problems with an extreme idealist, phenomenalist interpretation of
appearances including the following. It is not consistent with Kant’s rejection of
what he takes to be common to Descartes and Berkeley: that the objects of our
immediate experience are in our minds. It does not fit comfortably with his distinc-
tion between empirical and transcendental senses in which objects can be in us and
his regarding appearances as the same things as the things which have a way they are
in themselves. To make sense of this, we need an account of mind-dependence that
allows appearances to be dependent on the possibility of our cognizing them while
not existing merely in minds, and which sees appearances as grounded in a way
things are in themselves. The rest of the book presents my account of this.
PART TWO
Manifest Reality
5
Essentially Manifest Qualities
I Introduction to Part Two

In Part One, I argued that there is overwhelming textual evidence for the following
claims:
(1) Kant’s transcendental idealism includes idealism about spatio-temporal

objects, i.e. that there is a substantial sense in which such objects are mind-
dependent;
(2) Kant is committed to a robust empirical realism which is incompatible with a
phenomenalist idealism; and
(3) Kant thinks that the things which appear to us have a way they are in
themselves, independent of their appearing to us.
I argued that it is partly because there is textual evidence for all of these claims that
disputes between phenomenalist idealist interpreters and deflationary non-idealist
interpreters seem irresolvable: each side can always find textual evidence which
challenges the other’s view. The goal of Part Two is to present an account of mind-
dependence which, unlike phenomenalism, does not involve the existence of empir-
ical objects merely in the mind (or as a construction out of what is merely in the
mind), which fits with the claim that things have a way they are in themselves that
grounds their appearances, and which fits with the way Kant presents his argument
for transcendental idealism, including his specific concern with intuition.
While my concern in this part of the book is to present an account of Kant’s
idealism, the present chapter is not concerned with Kant: it presents an account of
mind-dependence that does not involve existence in the mind. I describe this in
terms of the idea of properties that are essentially manifest: their existence is not
independent of the possibility of their being presented to us in a conscious experi-
ence. I then draw on this account in Chapter 6, to show how well it fits with the texts
concerning Kant’s idealism. Chapter 7 is concerned with a crucial piece of background
for understanding Kant’s idealism: his account of the role of intuition in cognition.
Kant holds that intuition is one of the two essential and distinct ingredients in
cognition. His limits on what is empirically real centrally concern intuition: empirical
reality is limited to what can possibly be presented to us in intuition. Getting the notion
of intuition clear is therefore central for understanding transcendental idealism. I argue
 MANIFEST REALITY
that Kantian outer intuitions are representations that give us acquaintance with
particulars outside us. I then use this account of intuition, in Chapter 8, to make
sense of Kant’s argument for transcendental idealism in the Transcendental Aesthetic.
This argument centrally turns on the notion of intuition and, when we read intuition in
the way I propose, Kant’s argument goes through (given his assumptions) at exactly the
place he thinks it does. He argues that our representations of space and time are a priori
intuitions. He takes this to show that they are mind-dependent and do not present
things as they are in themselves. I then show that the kind of idealism that follows from
this argument fits well with the essentially manifest account presented in Chapters 5
and 6. Transcendental idealism limits empirical reality to things with which we can
have acquaintance: what can be manifest to us.
The first step, in this chapter, is to present an account of mind-dependence that
allows that we could be presented with things that are independent of us—but be
presented with them in a mind-dependent way that does not reveal how they are
independently of minds like ours. This aim of this chapter is to present an account of
perception and a way of understanding mind-dependent perceptual qualities that
enable us to make sense of this, independently of Kant.
As we saw in Chapter 2, in both the Refutation of Idealism and the Fourth
Paralogism in the first edition, Kant discusses a kind of idealism he rejects, which
he characterises as the view which says that the immediate objects of perception are
mental items and not external objects in space (A367–80; B274–9). Kant takes the
proponents of this position to include Descartes and Berkeley, who, he says, both
understand the immediate objects of perception to be merely mental items. Berkeley
says that this is all empirically real objects are, while Descartes adds that there are
objects which are inferred to be the causes of those mental items that we immediately
perceive. The ideas in us that Berkeley thinks constitute empirical objects, and that
Descartes thinks are the immediate objects of perception, are not, Kant says, external
objects in space. Kant says that he rejects both Descartes’ and Berkeley’s views; in
opposition to them, he wants to prove that the immediate objects of perception are
external objects in space. He argues that our mental representations of external
objects put us in immediate or direct contact with these objects, and that it is only
because we have direct and immediate experience of outer objects that we are able to
have knowledge of the temporal order of our inner life. This kind of position is
associated most obviously with direct realism about perception. But Kant is an
idealist. We thus need to find a way to put together the thought that we have direct
experience of things outside us in space with the claim that experience does not
present things as they are in themselves and that what is presented to us are mind-
dependent appearances. Making sense of the required kind of mind-dependence is
the aim of this chapter.
In a famous but tricky passage, Kant compares the idealism he is introducing with
what he takes to be the mind-dependence of secondary qualities like colour (Proleg. 4:
289, quoted in full in the next chapter). The central difficulty with making use of this
ESSENTIALLY MANIFEST QUALITIES 
comparison to explain Kant’s idealism is that there is so much disagreement about

how to understand the metaphysical status of colour. In addition, there is disagree-
ment about how to interpret Locke’s account of colour (which Kant alludes to), about
whether Kant is committed to Locke’s account (however he interprets it), and about
how Kant himself understands the metaphysical status of colour. Explaining Kant’s
idealism through a comparison with the mind-dependence of secondary qualities like
colour could result in any number of different positions, including straightforward
realism and extreme (phenomenalist) idealism: it depends on what account of the
metaphysical status of properties like colour is slotted into the analogy. In this
chapter, I present an account of metaphysical status of colour that enables us to
use the secondary quality analogy to explain Kant’s idealism in a way that fits with
the interpretative constraints I have argued for in the previous chapters. This chapter
is not concerned with Kant, and my aim is not to defend the view of colour I present.
Rather, my aim is to spell out the account of colour and the understanding
of perception within which it needs to be understood. In the next chapter I look
at how the account of colour presented in this chapter can be used to explain
Kant’s idealism.
We are looking for an account of colour according to which it is a directly
presented, external, but mind-dependent feature of mind-independent things.
I argue that this account of colour is one which must be understood in the context
of a realist, relational, or direct account of perception. Such an account of perception
holds that external objects are directly presented to us in perception and allows that
colour is a property of the objects of perception, and not a property of merely mental
states which represent objects. But Kant is not a realist: he is an idealist, and the point
of his comparison with secondary qualities is to invoke properties which are mind-
dependent. We need an account of colour which sees it as a property of objects (not
of mental states, internal modifications of subjects), but which sees it as a mind-
dependent property of objects. We need to explain the possibility of directly per-
ceiving features of objects that are mind-dependent but not in the mind, and to
explain how we could directly perceive outer things while not perceiving them as they
are in themselves. Both colour and perception are enormously complicated areas of
philosophy. I do not argue for or defend an account of either here; my aim is simply
to present a possible account of colour. Once the possibility of a relational or direct
realist account which allows that perception may not present us with things as they
are in themselves is taken seriously, the support for a phenomenalist reading
disappears, because we are able to present an alternative which fully accommodates
and explains Kant’s idealism, while doing much better justice to his empirical
realism, and his claims about the relations between things in themselves and appear-
ances, than does the phenomenalist reading.
I start by presenting a realist, relational, or direct account of perception. I then
present a possible version of such a view which allows that we can directly perceive
non-veridical qualities of things, and therefore which allows that perception
 MANIFEST REALITY
directly presents us with things, but can present us with them in ways other than
the way they are in themselves, independently of their perceptually appearing to us.
Drawing on this, I present a possible account of the metaphysical status of colour:
the idea that colour is a directly perceived quality of things that does not present us
with things as they are in themselves. I call this view the essential manifestness
account of colour.
It is, of course, not necessary to demonstrate that an account of perception or
colour is true in order to show that it is useful for interpreting Kant. Phenomenalist
interpreters do not take it that they need to demonstrate even that phenomenalism
is a fully consistent position, let alone true, in order to attribute phenomenalism
to Kant. However, it is necessary to spell out the account of perception we need
in sufficient detail to show that it is an alternative to indirect views of perception,
and looking at some of the arguments for and against the position is helpful in
explaining it.
One of the many difficulties in writing about perception is that it is hard even to
state the opposing views using clear and non-question-begging terminology. For a
long time debate about the immediate objects of perception was framed in terms of a
contrast between so-called direct and so-called representationalist or indirect
accounts. These terms are metaphorical and are potentially confusing in a number
of ways. For example, direct realists may agree that perceptual states represent the
world to us and therefore that there is a sense in which perception is representational.
And representationalists may argue that the use of the term ‘direct’ amounts to no
more than table-thumping; they may argue that what it is to directly perceive objects
just is to be in mental states (with certain contents) which are caused by objects (in
the right way). Describing perception clearly and without prejudice to the philo-
sophical questions is tricky, and there may be no problem-free terminology. This
problem with terminology is present with Kant: as we have already seen, Kant’s use of
the term Vorstellung, especially when this is translated (as it usually is) as ‘represen-
tation’, is often simply assumed to associate his view with traditional indirect or
representationalist views of perception. Kant’s saying that appearances are Vorstel-
lungen or representations is one of the chief pieces of textual evidence that phenom-
enalist interpreters appeal to. However, given the difficulties, for everyone, that are
involved in talking about perception, and the fact that, for many of Kant’s purposes,
‘Vorstellung’ could as well be translated as ‘presentation’, we should not take this
word to settle all interpretative issues. As I will show, a direct account of perception
can do full justice to the way Kant talks about appearances as representations.
Because of the extra distractions that may arise from Kant’s using the term ‘repre-
sentation’, my discussion of perception below avoids the traditional language of
‘direct’ and ‘indirect’, or ‘representationalist’, views of perception. Instead of ‘direct
realism’, I follow John Campbell’s (2002a) terminology by using the phrase ‘the
relational view of perception’ to describe the position within which, I argue, we need
to situate the account of colour that will enable us to make sense of Kant’s
ESSENTIALLY MANIFEST QUALITIES 
comparison with secondary qualities.1 I use the term ‘Cartesian’ to describe the
alternative view. I sketch, in broad outlines, the two main contrasting positions
I need to present my account.
My starting point is a relational view of perception according to which the external
object perceived is a constituent of perceptual mental states. A relational view has
important similarities to the contemporary theory called disjunctivism.2 Discussing
some of these similarities will be useful for presenting the view. However, I am not
arguing that a particular theory of perception should be attributed to Kant; in
particular, I am not claiming that he is a disjunctivist. ‘Disjunctivism’ and ‘the
relational account’ of perception are contemporary names for contemporary theor-
ies. In my view, there is no systematically presented ‘theory of perception’ in our
contemporary sense in the Critique, which suggests that there is a clear sense in
which it is problematic to argue either that Kant did or did not hold one of these
theories. Still, this does not rule out finding in Kant some of the ideas which these
theories of perception are trying to capture.
An objection to my strategy might be to say that representationalist or indirect
views of perception were dominant in the early modern period, and, for this reason,
appealing to contemporary relational and disjunctivist accounts of perception in
order to explain Kant’s idealism is anachronistic, and therefore implausible. Whether
indirect perception was the standard view of the period may be contested,3 but even if
it were, and even if no other philosopher Kant ever read had an alternative view, this
would not show that Kant could not have had something like the relational view.
Kant’s work is, of course, connected with and informed by his context and his
predecessors in many ways; it is also full of groundbreaking new conceptions. Direct
or relational views are not anachronistic in the sense of requiring scientific know-
ledge not available in Kant’s time. Many of the philosophers today who hold this kind
of view argue for it on the basis of considerations available to Kant, rather than
subsequent developments in empirical science. There is no reason to think that the
possibility of such a view could not have been seen by one of the most brilliant
thinkers in the history of philosophy, whose work is full of deep, revolutionary
insights.
II A Relational Account of Perception

The first step towards presenting the account of colour we need is to present a
relational (non-Cartesian) account of what is involved in things perceptually appear-
ing to us. At this point, I am concerned with the notion of perceptual appearing in a
1
This does not mean that Campbell would agree with the position I present; for example, he argues for
realism with respect to colour.
2
See Hinton (1973), Child (1994: Ch. 5), McDowell (1998a; 1998b), and Snowdon (2005).
3
See Yolton (1996; 2000).
 MANIFEST REALITY
neutral, realist sense; it simply captures the idea that we perceive things. Things
appear to us in perception. The aim is to characterise what is involved in objects’
perceptually appearing to us in a way that allows that what directly appears to us are
external objects in space (and instances of their properties), and not merely mental
representations of those objects.
The relational view is opposed to what I am calling the Cartesian view. One way to
explain the Cartesian view is that it holds that that the subject could be in exactly the
same mental state whether or not they were perceiving an object. Another way to put
this point is to say that perception involves a mental state (modifications of an
internal state of a subject) which is the final point in a causal chain involving the
object perceived, but where that very mental state could also have had completely
different causal origins. On this view, mental states count as instances of perception
when they have the right causal origin (the object), but the same mental state that is
present in an instance of perception could have been produced in a different way.
The idea is that the mental state represents the object whether or not the object is
present, but it counts as a perception of the object only when the object is present and
causes the mental state in the right way.4 In contrast, on the non-Cartesian or
relational view, perceptual mental states essentially involve the presence to con-
sciousness of the object perceived. The idea is that the presence to consciousness of
the object is part of what makes the mental state the state that it is. Campbell says that
term ‘relational’ marks the idea that ‘the object perceived is a constituent of the
conscious experience itself ’ (Campbell 2002a: 117). According to the Cartesian view,
perception involves mental states which a subject could be in irrespective of the
presence of the object perceived; the mental states can therefore be understood as
purely internal—as, in Kant’s terms, modifications of a state of the subject. According
to the relational view, in contrast, a perceptual mental state would not be the very
state that it is if it did not involve the presence to consciousness of the objects
perceived; this presence is essential to its being a state of perceiving these objects.
A perceptual mental state is not merely a modification of an inner state of a subject
but a relational state essentially involving the object and a conscious subject.
Another way philosophers have explained direct views of perception is as follows.
Phenomenologically, perception seems presentational, or transparent, in the follow-
ing sense. First, perception seems to make external objects immediately available for
demonstrative identification and cognitive scrutiny in a non-mediated way.5 Second,
perception subjectively presents as if (at least most of ) the qualitative aspects of
perceptual experience are aspects of the objects perceived, and not intrinsic features
of subjects’ merely mental states.6 The relational view holds that perception is as it
subjectively presents in these respects: it makes objects available for demonstrative
4 5
Excluding deviant causal chains. See Foster (2000: 50) and Snowdon (2005).
6
This needs to be qualified: the account can allow that there are some aspects of perceptual experience
which do not subjectively present as if they are aspects of the objects perceived, for example, blurriness.
ESSENTIALLY MANIFEST QUALITIES 
identification and cognitive scrutiny in a non-mediated way, and the qualitative

features of perceptual experience are features of the objects perceived (particular
instances of properties) rather than properties of entirely inner, merely mental states.
Perceptual mental states are intrinsically object-involving.
A common way of contrasting relational and Cartesian views is in terms of how
they characterise the difference between perception and a hallucination that is
subjectively indiscernible from the perception. According to the Cartesian view, a
subject could be in the same perceptual mental state when hallucinating as that they
are in when actually perceiving an object: what distinguishes perception from
hallucination is not the content or nature of the mental state itself, but how it is
caused. A relational view, in contrast, allows that perception can involve states that
are subjectively indistinguishable from hallucinations, without granting that these
must, for that reason, be type-identical mental states. The view holds that perceptual
mental states are mental states that essentially involve the presence to consciousness
of the objects perceived, and which therefore are metaphysically different from the
purely internal, or merely mental, states involved in hallucination. Perceptual states
are not just internal modifications of a subject, together with the right causal history;
they are conscious states that contain objects (and particular instances of properties)
as their constituents.
The relational view, here, is similar to disjunctivism, since a denial of a common
factor between perceptual states and hallucinatory states is what characterises dis-
junctivism.7 Disjunctivists reject the idea that there are putative perceptual states
(states which appear to be perceptual) which serve as a common factor between
actual perceptual mental states and hallucinatory mental states. Rather, they propose
that we should think in terms of a disjunction of two essentially distinct kinds of
states: perceptual states, which essentially involve their objects, and states which
merely seem to be perceptual (hallucinations). Some versions of disjunctivism have
commitments which go beyond a general relational account of perception,8 such as
the idea that perceptual experience must be understood as involving something like
the grasping of demonstrative thoughts about objects.9 This further claim is not what
I am drawing on. What is relevant for present purposes is what is common to
relational views and disjunctivism: a way of thinking about perception which allows
that when you see an object, the object itself is a constituent of your perceptual
mental state.
7
See Hinton (1973), Child (1994: Ch. 5), McDowell (1998a; 1998b), and Snowdon (2005).
8
There are also internal disputes among disjunctivists which need not concern us here. For example,
whether hallucinatory states can be characterised only in a derivative way—as states which are subjectively
indistinguishable from perceptual states we could actually be in. See, for example, Child (1994: Ch. 5).
A possible problem with this position is that we can hallucinate things which we could not veridically
perceive. See Johnston (2004) for a positive characterisation of hallucination.
9
Child (1994) and McDowell (1998b). Campbell argues that this robs experience of its explanatory role,
which should be prior to, and not require, demonstrative thoughts (Campbell 2002b: 136).
 MANIFEST REALITY
The different ways in which relational and Cartesian views characterise hallucin-
ation can be explained by considering the traditional argument from hallucination
against direct realism.10 The argument starts from the possibility of experiencing a
hallucination that is subjectively indistinguishable from perceiving something and
takes this to show that what is immediately before the mind in cases of perception is
something merely mental. The thought is that since it seems possible to be in
hallucinatory states that are indistinguishable from states of perceiving, in such
cases the mental states are in fact metaphysically identical kinds of states: the mental
state a subject is in when perceiving an object is a mental state the subject could also
be in while merely hallucinating (though in the latter case the mental state would
have a different causal history). This argument assumes that subjectively indiscern-
ible mental states are not distinct types of state. In cases of hallucinating an external
object, the external object is not a component of the mental state. If the perceptual
mental states involved in actually perceiving objects are not type-distinct from those
involved in hallucination, then it follows that external objects are also not compo-
nents of actual perceptual states. This is taken by proponents of the argument to
show that perception cannot be direct. However, relational views reject the assump-
tion that subjective-indiscernibility implies metaphysical identity. According to
relational views, perceptual mental states involve the presence to consciousness of
objects, while hallucinatory mental states do not. From this starting point, the
possibility of the subjective indistinguishability of hallucination and perception
does not establish the Cartesian, indirect view of perception, since it does not show
that perception involves being in inner mental states which could occur without the
object being present. It merely shows that subjects cannot always tell, from the inside,
whether they are perceiving an object. What the argument from hallucination brings
out is a standoff: what it establishes depends on the starting assumptions. The
possibility of hallucinations that are subjectively indistinguishable from perceptual
mental states establishes the Cartesian view of perception only with the starting
assumption that metaphysical differences between perceptual and hallucinatory
mental states would have to be discernible to the subject, from the inside.
Other considerations which have been taken to support a Cartesian conception of
experience are scientific accounts of optics11 and, more recently, detailed knowledge
of brain processing. Those who argue for a Cartesian view by appealing to brain
processing often stress differences between the sensory input and the nature of our
perceptual experience of the world. For example, Colin Frith (a cognitive neurosci-
entist who presents what, from the point of view of philosophy, is a Cartesian-like
position12) claims, among other things, that we have an illusion of perceiving the
world as stable, when in fact the image on the retina is in constant motion (Frith
2007: 102). The thought seems to be that what we are directly given is a causal input
10
See, for example, Valberg (1992), Howard Robinson (1994), and Foster (2000).
11 12
See Yolton (1984) for a summary. In a book written for non-specialist readers.
ESSENTIALLY MANIFEST QUALITIES 
delivered to sense-receptors (such as the image on the retina), and that if processing
this input is required in order for us to have a conscious representation of a stable
world, then the product of this processing would have to be an inner or mental image
constructed by the brain from the retinal image, other causal inputs, and perhaps also
its own rules of inference or association.13 The relational view, in contrast, holds that
mental processing enables us to be in direct contact with the world.
Arguments against direct realism based on cognitive processing are of particular
interest for understanding Kant’s account because of the many ways in which Kant
thinks the mind must be actively involved in structuring, combining, and synthesis-
ing the sensory input to produce experience of a world. This is a feature of Kant’s
thought which has interested philosophers interested in cognitive processing in
contemporary philosophy14 and also one which might be taken to support his
being read as a phenomenalist who sees the empirically real world as a construction
out of sense data. Those who take it to follow from the existence of brain processing
that our experience is an entirely inner, merely mental, representation may think that
Kant must draw this conclusion, given his extensive interest in various kinds of
mental processing. However, just as this view is not a compulsory conclusion about
cognitive processing, it is not a compulsory interpretation of Kant.
It is important to notice that Kant argues that something like cognitive processing
is needed at every level of experience,15 but he does not argue for idealism on this
basis, because he does not take his idealism to follow from the need for empirical
synthesising. He takes idealism to follow, specifically, from (as he sees it) the need
for a priori ways of organising the sensory input. So he does not take the mere fact
that mental input is needed in arranging and organising the data of the senses to
lead to idealism. Further, Kant does not think that the direct or immediate objects
of perception are sensations. For Kant, the mental representations that directly
present us with particular things are intuitions, and sensations are not the same as
intuitions: intuitions are the product of sensory input being ordered using our
forms of intuition, space and time (this is discussed in detail in Chapters 7 and 8).
While he speaks of sensations as a mere causal input, or a mere awareness by a
subject of their internal states, he says that (outer) intuitions immediately present
us with external objects. For Kant, therefore, a spatial object’s being immediately
13
Frith says that ‘by hiding from us all the unconscious inferences it makes, our brain creates the
illusion that we have direct contact with objects in the physical world’ (2007: 17). He argues that we seem to
perceive the world instantly and without effort, but that this is an illusion, because there is a temporal delay
during which the brain engages, in his terms, in unconscious inferences. He further argues that while we
seem to perceive the whole visual scene in vivid, full-coloured detail, this is illusory because colour
information only reaches us from the centre of the visual scene, and because although we have a blind
spot from which we get no input from the world, we experience a world without a hole in it (2007: 41).
14
Kitcher (1990) and Brook (1997).
15
This includes the combining processes he calls synthesis (discussed in detail in Ch. 11) as well as the
binding involved in representing spatio-temporal locations, and the kinds of associative processes we share
with animals.
 MANIFEST REALITY
present to consciousness is not the same as its being the last link in a causal process
involving the object, the optical nerve, and the brain.
According to a relational view of perception, what is directly present to conscious-
ness is the world of objects outside us—not something in our brains or minds, such as
the causal effects of objects on the optical nerve, the upside-down image on the
retina, or a mental state we could be in whether or not an object was present.
Campbell suggests a metaphor to help us get a grip on the idea that mental
processing is needed to enable us to be in direct contact with objects. Suppose, he
says, that the relational view understands perception as like viewing things through
a pane of glass. In the metaphor, a representationalist or Cartesian view involves
seeing an image on the glass, which represents things behind the glass. In contrast,
for the relational view, we see through the glass to the objects themselves. Our
worry was that knowledge of cognitive processing seems to threaten the idea that
perception can be like this, because it tells us that, unlike a passive pane of glass, the
brain is actively involved in constructing a representation of the world. In response,
Campbell suggests the following analogy:
Suppose we have a medium which, like glass, can be transparent. But suppose that, unlike glass,
it is highly volatile, and needs constant adjustment and recalibration if it is to remain
transparent in different contexts. Suppose, in fact, that the adjustment required is always
sensitive to the finest details of the scene being viewed. The upshot of the adjustment, in each
case, is still not the construction of a representation on the medium of the scene being viewed;
the upshot of the adjustment is simply that the medium becomes transparent. You might think
of visual processing as a bit like that. It is not that the brain is constructing a conscious inner
representation whose intrinsic character is independent of the environment. It is, rather, that
there is a kind of complex adjustment that the brain has to undergo, in each context, in order
that you can be visually related to the things you see around you; so that you can see them, in
other words. (Campbell 2002a: 119)
Another illustration of the idea is to think about the role played by corrective glasses
in bringing the world into focus. When a short-sighted person puts on their glasses,
the scene is transformed from fuzziness to focus. The glasses are an addition to the
total visual system which alters the focusing of light, but the fact that they make a
contribution which affects visual processing does not entail that they make us see
images on the lenses. Rather, the addition to the visual system enables the system to
bring the scene into focus.
In contrast to the Cartesian view, relational views of perception do not take the
existence of cognitive processing to show that we perceive the world in virtue of being
in merely mental, inner states that represent it. And they do not take the possibility
that perceptual states and hallucinatory states can be indistinguishable to demon-
strate that perceiving is a matter of being in representational, inner, merely mental
states with the right causes (as opposed to involving the object as a constituent).
I have not argued for this view; so far I have merely suggested that two traditional
concerns taken to rule it out (hallucination and cognitive processing) need not be
ESSENTIALLY MANIFEST QUALITIES 
taken in this way. My aim is simply to present the view as a possible starting point in
thinking about perception; I will argue that this kind of relational view is also a
starting point for making sense of Kant’s idealism.
III A Relational Account of

Non-transparent Perception
I have mentioned objections to direct realism on the basis of hallucination and on the
basis of cognitive processing. Another traditional argument against direct realism is
based on non-veridical perception. Roughly, whereas hallucination involves seeming
to perceive something which is not actually present, non-veridical perception
involves perceiving an object but perceiving it as being, in some way, different
from the way that it mind-independently is.16 Thinking about non-veridical percep-
tion within a relational account of perception is the next step in the characterisation
of the account of colour we are working towards.
As in the argument from hallucination, non-veridical perception is sometimes
taken to show that the content of perceptual states, or what is immediately presented
in perceptual states, cannot be an external physical object. The thought is that
perceptual experience cannot directly present objects themselves if it sometimes
presents them in ways that are different from the way they are. This is why non-
veridical perception is supposed to provide an objection to direct realism. As Howard
Robinson presents the argument, the claim is that the phenomenon of non-veridical
perception shows that ‘in some cases of perception there is something of which the
subject is aware which possesses sensible qualities which the physical object does not
possess’,17 and thus that, in these cases, it is not the physical object that is directly
perceived. For example, consider the case of a straight stick that is perceived as being
bent in water. The argument against direct realism says that, in this case, what the
subject is aware of possesses sensible qualities the stick does not possess (being bent).
This, it is argued, shows that what the subject is directly or immediately aware of is
not the external object (the stick). The idea is that since bentness is perceived and is
not a property of the stick, the sensible experience of bentness must involve the
subject being in direct mental contact with something other than the external,
physical object. The argument then generalises this to all cases of perception, leading
to the conclusion that what is directly or immediately perceived is never a physical
object. In support of the generalising move, the proponent of the argument can
appeal to the apparent continuity between veridical and non-veridical perception—
the way in which veridical perception can gradually become non-veridical. For
example, a scene can gradually move out of focus as objects recede, and a stick in
16
There may not always be a sharp distinction between these, and there are many different kinds of
cases of both.
17
Howard Robinson (1994: 57).
 MANIFEST REALITY
water can gradually appear more bent, as the viewer’s position changes. The general-
ising move rides on the thought that it does not seem plausible that the direct objects
of awareness are mental when the short-sighted subject takes her glasses off, physical
when she puts her glasses on, and that there is some point at which they switch from
one to the other.
Direct realists have a number of strategies for dealing with non-veridical percep-
tion. Two of the possibilities include a disjunctivist move of the sort that can be made
for hallucination and the denial that non-veridical perception occurs. The disjuncti-
vist move with respect to non-veridical perception would be to say that such
awareness is not awareness of something outside of the subject; it merely seems to
be, and in this respect is entirely different from, veridical perception. On this view,
states which misrepresent the object (or some property of the object) do not involve
the presence to consciousness of something outside the subject, while states which
represent the object as it mind-independently is do. Alternatively, an extreme realist
response consists in maintaining that all apparently illusory features that can be
presented in non-veridical perception are in fact mind-independent features of
reality that are presented to us just as they mind-independently are. On this view,
our perception of the apparent bentness of the stick involves our being presented
with something that really is, mind-independently, outside us, just in the way it is
perceived as being. The illusion is to be accounted for in terms of the mistaken
judgment that, for example, the refracted light we are seeing is in fact the stick.
In common with the indirect realist, the disjunctivist about non-veridical percep-
tion assumes that where our perceptual presentation of an object presents it as being,
in some respect, a way it is not in itself, what we are directly presented with cannot be
the external object itself, or a property of the object itself, or, more generally,
something outside the mind. In other words, both these views assume that to the
extent that we are directly presented with something external, we must be presented
with it as it mind-independently is: either we are directly presented with mind-
independent things as they mind-independently are, or we have merely mental,
internal states, which we may mistakenly take for presentations of external objects,
or features of external objects. The indirect realist argues that in the case of an illusion
such as the stick that appears bent in water what we are sensibly aware of has
properties the external physical object lacks, and therefore that in perceiving these
properties what we are directly aware of cannot be the stick. The disjunctivist about
non-veridical perception agrees that the apparent bentness of the stick cannot
involve our being directly presented with the stick, but rejects the idea that we can
generalise from this to the conclusion that everything presented to us in perception is
merely in the mind. The apparent continuity between veridical and non-veridical
perception is taken merely to show that a subject cannot always tell, from the inside,
whether they are directly presented with an object (or a property of an object).
An alternative possibility is a version of a relational or direct view that stresses the
fact that perception involves a relation not just between an object and a perceiving
ESSENTIALLY MANIFEST QUALITIES 
subject, but an object, a perceiver, and a particular environment or context of

perception. For example, with respect to the bent stick in water, it can be argued
that what we directly perceive is something outside us (not something merely
mental) but what is perceived is not simply the stick, but the stick presented in a
particular environmental context, viewed from a particular angle. Campbell uses the
term ‘the relational view’ to mark the idea that ‘the object perceived is a constituent of
the conscious experience itself ’ (Campbell 2002a: 117). In responding to indirect
realism, the object side of this relation is stressed: the idea that external objects are
components of the mental states involved in perception. But the relationality goes the
other way as well: it is equally important that perceptual states contain not just
objects on their own, but objects presented to consciousness, and presented in a
context. This means that versions of a relational view can allow that the way an object
is presented in consciousness may differ, to some extent, from the way it is inde-
pendently of its being presented to consciousness. The way the object is presented in
consciousness is something more than the object alone, as it is outside of this relation,
so we are not forced to say that the relationality or directness of perception implies
that the object can be perceived only as it is apart from its appearing to consciousness
(as it is in itself ). This version of the relational view agrees that non-veridical
perception involves being presented with a feature the object does not have as it is
independently of its being perceived but denies the indirect realist’s conclusion.
Rather, it attributes to the stick (the external object) the sensible, relational property
of appearing in a certain way: the sensible property of appearing bent. The way it
appears differs, to some extent, from the way it is apart from its appearing.
The idea that perception can present things as being, to some extent, different
from the way they are apart from their being perceived can be illustrated by
introducing a variation on Campbell’s analogy of the glass-like transparent medium
of perception. Imagine that a malfunction in the adjustment mechanism causes the
medium to fail to be completely transparent and, instead, distorts the shapes of things
to some extent, as, for example, a fish-eye lens might. According to the version of the
relational view I am going to draw on, just as when the medium is transparent, it is
not the case that the distorted things are seen by means of seeing an image on the
glass—it is the things themselves, beyond the glass, which we directly see. If the fully
transparent glass allows us to be directly in touch with the objects themselves, we can
also be directly presented with the objects themselves when they are seen through a
distorting glass. On the account I am presenting, the curved appearance of the
distorted-appearing objects is not a sensation objects produce in us, nor is it a feature
of the objects as they are in themselves; it is a feature of the objects which are seen,
but which are seen in a way which differs, to some extent, from the way they are apart
from their being seen.
We can see here that this relational account allows a distinction between things as
they are in themselves and things’ perceptual appearances. A straightforward way of
using the expression ‘things as they are in themselves’ is to refer to things as they are
 MANIFEST REALITY
independently of their relations to other things. Since perceptual appearing is

relational (it involves objects being presented to subjects’ consciousnesses), it can,
in principle, be distinguished from the way things are in themselves outside of this
relation. In the case of the distorting glass, we can distinguish between the shapes of
things as they are apart from their being seen (or as they are apart from their visually
appearing) and the way the shapes of things appear; they do not appear as they are in
themselves, apart from the context and the relation of perceptual appearing. The
distorted-appearing objects are the objects—the very objects themselves—directly
present to consciousness, but they do not appear to consciousness entirely as they are
in themselves.
Relational accounts take perceptual appearing to involve objects being present in
subjects’ consciousnesses as opposed to objects causing merely inner effects in
subjects (inner effects which represent the objects). If we take the expression ‘objects
as they are in themselves’ to refer to the way objects are independently of their
relations to other things, this will include the way they are independently of their
perceptual appearing to subjects. Of course, relational views allow that objects could
appear to subjects just as they are independently of their being perceived: they could
appear as they are in themselves. However, it is open to a relational view to allow that
objects can, at least sometimes, and to some extent, perceptually appear in ways
which are different from the way they are in themselves. It is important to move
slowly here: this is not meant to be an account of the status of Kantian appearances (it
is an account of empirical illusion). The point is to present a way of thinking about
‘non-veridical’ perception that allows that, although perception is direct in the sense
that the object is essentially present to consciousness, it does not follow from this that
the perceptual presentation of the object is exactly the way the object is apart from its
being seen. Crucially, this move does not involve mentalising the sensible appearance
of the object, or understanding it as something merely in the mind, a modification of
a subject’s internal state. There are three further features of non-veridical perception,
on this account, to which I want to draw attention here. First, on this account, non-
veridical appearances can be public. Second, we can make sense of the use of the
term ‘representation’ in this context, without mentalising appearances. Third, non-
veridical appearings can be mind-dependent.
The first point is that despite the fact that non-veridical appearance is illusory in
the sense that it presents things as being different to the way they are in themselves,18
it can be public (if it is relativised to a specific, but publicly available perceptual
context). We can all see the apparent difference in length of the Müller-Lyer lines and
we can all see the bent appearance of the stick in water.19 For example, by pointing at
it, we can bring someone else’s attention to the bent appearance of the stick. We can
18
It need not be illusory in the sense that the illusion is believed by the subject.
19
These non-veridical appearings arise in very different ways, the former, arguably, from the common
structure of our minds, the latter from the laws of light refraction. How we characterise non-veridical
ESSENTIALLY MANIFEST QUALITIES 
draw a picture of it, and disagree about whether someone has drawn it accurately.
This contrasts with private mental presentations, such as my itches.
Second, we can see a use for the term ‘representation’ in this context without
taking it to refer to merely mental intermediaries. Since, in the above example, the
distorted shapes things appear to have differ from the shapes those things have apart
from their appearing, it seems natural to say that visual experience presents or
represents the shapes of things as being more curved than they are.20 Rather than
introducing merely mental intermediaries, the term ‘representations’ could be used
simply to emphasise the fact that perception involves things appearing a certain way
for subjects: it presents things to subjects. When an object is presented to a subject’s
consciousness, there is a way perceptual experience presents the object. In this sense,
we could say that perceptual appearances are representations. As we saw in Chapters
1 and 2, Kant’s calling appearances ‘representations’ is one of the main features of the
text which is taken to support the extreme idealist view, but we can now see that the
term could be used to mark the presentational character of perception.21
The third point about non-veridical perception is closely related: despite being
public and not mere mental items, it can be argued that non-veridical perceptual
presentations are essentially mind-dependent. In the extension of Campbell’s pane of
glass case in which the shapes of things are seen in a distorted way, the objects, as they
are in themselves, do not have the curved edges they appear to have. It might be
argued that the curved appearances are not mind-dependent, even though they do
not belong to the objects as they are in themselves; rather, they belong to the object in
the specific context of perception, which includes the distorting pane. They could be
relational properties without being mind-dependent. However, they will depend on
minds if minds (conscious perceptual subjects) are essential to the relation in which
the presented property exists. If the curved-appearing edges do not exist apart from
the way the objects appear for the relevant subjects in the particular context, they can
be thought of as mind-dependent. This is not a kind of mind-dependence that
involves existing merely in the subject’s mind. On the account presented, it is the
objects that have this mind-dependent feature; the perceptual appearing is not
merely a modification of a subject’s internal state. It is a feature the objects have in
appearings may depend on the details of case, and relational views may not treat them all alike; these
complexities go beyond my concern here.
20
I am not arguing that this is the best, or most fundamental, analysis of what is going on in all cases. It
might be argued, for example, that visual experience presents objects as they are independently of us, in the
particular perceptual context, and that it is our judgments about the objects that introduce error. On this
view, perceptual experience does not represent objects as having properties that are different from the ones
they have in themselves, rather, it simply presents particular instances of properties in particular contexts
which we then may, mistakenly, take for presentations of properties in other contexts of perception. For a
defence of something like this position, see Genone (2014). My point is simply to note that we can see why
someone might use the term ‘representation’ in this context, without taking this to imply mental
intermediaries.
21
For detailed and helpful discussion of this point, see Collins (1999, esp. Ch. 5).
 MANIFEST REALITY
relation to the possibility of their being experienced by minds of a certain sort. How
exactly to characterise and formalise such mind-dependence is a complex philosoph-
ical question.22 The point for here is simply to note the possibility of holding
perceptual appearances to be mind-dependent to the extent that they do not present
objects as they are independently of their appearing to a subject in a particular
context of perception.
We can combine the points about mind-dependence and understanding non-
veridical perceptions as representations. I have suggested that a relational view could
make sense of the use of the word ‘representations’ to mark the presentational
character of perception. Further, I have argued that there is a possible view which
holds that in the case in which perceptual experience presents objects in a non-
veridical way, what is presented does not exist in the object as it is apart from its
perceptual appearing. In such cases, we could say that the perceptual appearance
exists merely in the representation of the object. The point to note here is that in this
case not only calling perceptual appearances representations but even calling them
mere representations does not require seeing them merely as inner modifications of a
subject.
Partly through looking at the arguments given for the opposing positions, I have
presented a possible view of perception which has the following features. As a direct
realist or relational view, it understands perceptual mental states as states which
present objects outside us to consciousness. It denies that perceptual mental states are
merely mental states which we could be in whether or not any object were present to
us. However, the fact that perception presents objects to consciousness is not taken to
mean that it must always and only present them as they are independently of their
perceptual appearing. The very relationality of the view—the fact that it involves
objects, consciousness, and a context—means that perceptual states depend not just
on objects. The view allows that it is possible for the objects themselves to be
presented to consciousness in ways which differ from the way they are independently
of their being presented to consciousness.
IV Essentially Manifest Qualities

The next step in my argument is to take the account of perception just given, and use
it to present a candidate account of the metaphysical status of colour. The account of
22
Resolving it will include questions about how to understand the dependence of relata on the relations
in which they exist. It might objected that we can say that objects, in themselves, have the property of being
such that they would appear in certain ways to certain kinds of subjects. For example, independently of
minds, the objects have the shape properties which are responsible for their having the curved appearances
they have, in a particular context of perception, to particular subjects. However, that objects have mind-
independent properties which ground their perceptual appearances does not show that the perceptual
appearances are mind-independent. The form of mind-dependence invoked here is relatively weak; it
might be thought only to require the possible existence of perceiving subjects.
ESSENTIALLY MANIFEST QUALITIES 
perception allows for the possibility of a view that understands colours as features of
external objects which are directly presented to us in perception but which are not
qualities objects have independently of the possibility of their perceptually appearing
to us. Like perception, colour is a deeply difficult, complex, and controversial topic in
philosophy; my task here cannot be to argue for a particular account. Rather, as I did
with the relational account of perception, I will simply present a possible account on
which I will draw. For our purposes, it will be enough if this account is coherent.
We started with a relational view of perception which insists that the object
perceived is a constituent of a perceptual experience (and not simply a cause of
merely mental representational states). We added to this the possibility of drawing a
distinction between the way things appear in perceptual experience and the way they
are apart from this, and the possibility of allowing that perceptual experience may not
always present objects as they are independently of their perceptual appearing. The
next step is to introduce the idea of qualities which belong only to the perceptual
appearances of objects: qualities which objects have as they are presented to subjects
but which they do not have as they are apart from their appearing, as they are in
themselves. I will call these essentially manifest qualities.
I start with the idea of manifest qualities, using a relational account of perception.
To say that something appears to us can simply express the thought that it becomes
manifest to us; it becomes present to consciousness. Appearing is used in this sense in
such claims as, ‘when you get over the hill the church appears’. In this sense, to say
that something could appear to us simply marks the fact that it is the kind of thing (or
the kind of quality of a thing) that can manifest itself to us in perception—that can be
presented to us in conscious experience. I will call qualities of objects which can be
presented to us in perception manifest qualities. It is essential to manifest qualities
that they are not merely theorised or postulated: they are presented to our conscious
experience. Like the neutral notion of perceptual appearing, the idea of a manifest
quality does not imply mind-dependence. It could be that mind-independent fea-
tures of objects manifest themselves to us in perception.
The next step is to introduce the idea of essentially manifest qualities, drawing on
the relational account of non-veridical perception presented above. Essentially mani-
fest qualities will be qualities of objects that appear to us—they are manifest qualities.
However, in addition (and relevantly similar to non-veridical appearings), they are
not qualities objects have independently of their perceptually appearing to us. They
belong only to the perceptual appearing of objects. They are features of the way
objects appear to us and nothing but such features. This is a possible and, in my view,
appealing, account of the metaphysical status of colour. Part of what makes it
appealing is that, on this view, we can take seriously many of the reasons philo-
sophers have given for thinking that colour is mind-dependent, as well as, on the
other hand, reasons for thinking that colour is a property of objects and not of merely
mental representations or sensations. However, I cannot argue for the position here.
My aim is simply to present the view as a possibility.
 MANIFEST REALITY
To get a grip on the idea of colour as essentially manifest, it is helpful to contrast it

with two alternative positions which pull in opposite directions: on the one hand, a
version of realism about manifest colours, which I will call primitivist objectivism;23
and, on the other hand, the idea that colour should be understood in terms of the
production by objects of merely mental ideas or sensations in us. Primitivist object-
ivism is the view that objects have colour in the way in which naive perceptual
experience takes them to: mind-independent colours are manifest to us in visual
experience. In taking colours to be mind-independent manifest properties, primitivist
objectivism differs from those objectivist views which see colour as a micro-physical
surface property, as a causal power, or, more generally, as any kind of property that is
specified in the language of science, as these are not the visual properties which
transparently reveal themselves to us in colour perception; they are not manifest
qualities. We can call the latter scientific objectivism.24 According to the primitivist
objectivist, colours as they manifest themselves in visual experience are mind-
independent properties, presented to consciousness, whose natures are revealed to us
in perceptual experience.25 Primitivist objectivism is a robust form of realism about
colours: it takes colours to exist independently of us, and it takes the colour properties
which exist independently to us to be the very properties which are presented to us in
perceptual experience, the colour properties which are manifest to us.
The other view with which I want to contrast essentially manifest qualities holds
that colours should be understood in terms of objects having dispositions or powers
to produce ideas or sensations in us. There are many possible versions of the idea that
colours are dispositions or powers: dispositional views can regard colour as the
mental effect of mind-independent causes, as the mind-independent cause of the
mental effect (this would be a form of scientific objectivism), or both as the mind-
independent cause and the mental effect it produces. The view with which I am
contrasting primitivist objectivism is the first of these.26 I will refer to it as the idealist
dispositional account. The idea is that (visual) colours are ideas or sensations or
representational mental states or properties of ideas or sensations (modifications of a
state of a subject), which typically have mind-independent causes. This view is
situated within a Cartesian account of perception. It holds that colour is a mental
effect produced by a mind-independent property of the object and that colour
experience does not reveal the nature of the mind-independent property that causes
it. Colour experience reveals only the merely mental effect produced by the cause.
23
This term is used by Byrne and Hilbert (1997a).
24
See, for example, Armstrong (1997), Byrne and Hilbert (1997a), and Jackson and Pargetter (1997).
25
See Hacker (1989), Campbell (1997), and Stroud (1999).
26
The dispositional view I am drawing on is situated within a representative view of perception; it is
controversial whether dispositional views of colour have to be so situated. McGinn (1983: 132) argues that
they do not. My aim is simply to draw on a dispositional account which does understand colour in this
way—as a power to produce sensations or ideas in us—and therefore which sees colour as belonging only
to merely mental bearers of representational content.
ESSENTIALLY MANIFEST QUALITIES 
The idea of an object’s having powers to produce sensations in us can be illustrated

with less controversial examples than colour. For example, consider a pill which has
the power to produce sensations of dizziness. Here, the pill has certain causal powers
(or dispositional properties) that produce inner effects in me. In this case it is clear
that the sensation of dizziness is in me; it is, as Kant puts it, a modification of an
internal state of the subject. The sensations of dizziness can be characterised in a way
which is entirely internal to the subject, does not require reference to its mind-
independent cause, and could have had a different cause. And experiencing the
sensation of dizziness does not reveal or make manifest the causal powers of the
pill which enable it to produce the sensation. The idealist dispositional view of colour
sees colour in something like this way. Like the dizziness that the pill causes, colour is
something that exists only in the subject. Just as the pill has causal powers that
produce the dizziness in subjects, objects have causal powers that produce colour
sensations in subjects. It is not the causal powers that are manifest in colour
experience (any more than the causal powers of the pill are manifest in experience
of dizziness), but rather the colour sensations. Experience of the colour sensations,
like experience of the dizziness, does not reveal or make manifest the causal powers.
As mentioned earlier, there are different versions of dispositionalism. Another
possible view is one which identifies colour with the mind-independent causal power
that produces the mental effects. It would be possible to pick out the power pills have
to cause dizziness in terms of its effects, and calls this causal ground ‘dizziness’
because of its link to the relevant effects. Analogously, there is a version of the
dispositional view of colour which sees colour as a causal property of objects that
is picked out in terms of the mental effects it produces. This view is a version of
scientific objectivism: it identifies colour with a property specified in the language of
science, something other than the property that is manifest to us in visual experience,
leaving the visual, manifest property in the mind.27 Both scientific dispositionalism
and idealist dispositionalism agree in thinking that visual colour, the colour
we experience, is a merely mental effect of a causal power and that the nature
of the causal power that produces the mental effect is not revealed in the sensations
of colours.
Phenomenologically, colour experience is very unlike experience of sensations
such as dizziness. Colour is, while dizziness is not, experienced as presenting an
external feature of objects. The sensation of dizziness is not a property of the pill and
is not experienced by us as being one: it is not part of how the pill perceptually
appears. In contrast, colour experience presents colour as a property of the object
perceived. It is part of how experience presents objects as being. Both the scientific
27
As Stroud argues, the dispositional view either makes colours sensations or it says that colours are not
the properties we see when we think we experience colour (Stroud 1999:116–17). Van Cleve has a similar
view of Locke’s account of secondary qualities: he says that if colour is identified with the cause of the
colour-ideas then direct perceptual experience gives us no knowledge what colour is like (Van Cleve 1995).
 MANIFEST REALITY
dispositional account and the idealist dispositional account are therefore, in a sense,
error theories, since they deny that colour is the manifest property of objects that it
appears to be. The scientific dispositionalist denies this by identifying colour with
something other than the visual quality which manifests itself to us. The idealist
dispositionalist, in contrast, holds that colour is the visual quality that is manifest to
us in perception, but denies that this quality is a quality of the object. Both say that
the quality we seem to see objects as having when we see them as coloured in fact
belongs to something merely mental. Primitivist objectivism takes the colours pre-
sented in perceptual experience as mind-independent, manifest features of objects.
Both idealist and scientific dispositional views take the visual colour presented in
perceptual experience to belong to something merely mental, and not to reveal
the mind-independent properties of objects that cause colour experience. Like
the primitivist objectivist, the essentially manifest view takes visual colours to be
properties of objects that manifest themselves to us in perception. But like the idealist
dispositional view, it holds that the perceptual presentation of these properties does not
reveal the mind-independent natures of the objects that ground colour experience.
Philosophical accounts of colour are subject to pressures in opposite directions,
produced by two commonly accepted features of colour. One pressure is to preserve
the idea that colour is, as it phenomenologically appears to be, a property of objects.28
The other pressure is a function of the idea that colour has an essential relation to
visual experience: it is a feature of the way things look.29 This idea is central to a
28
A common objection to the powers view of colour is that it wrongly assimilates perception to
sensation (Hacker 1989; Harman 1997; Stroud 1999: 98–9). Stroud argues that if colours are sensations
there is not even a coherent thought to the effect that a physical object could be coloured, and could be
ascribed to objects, any more than pain could (Stroud 1999: 106). Another common objection to the
dispositional view is that colours do not appear to be dispositions. Johnston argues that ‘a property cannot
appear as a disposition unless it appears as being a relation of the bearer of the disposition to the
manifestation of the disposition and the circumstances of the manifestation’ (Johnston 1997: 142). He
argues that colour could at best be the manifestation of a disposition. In a related argument, Jackson and
Pargetter argue that if redness were a disposition ‘objects looking red could not be the apprehension of that
redness, because dispositions do not cause their manifestations. Their categorical bases do that’ (Jackson
and Pargetter 1997: 69, see also Boghossian and Velleman 1997). See McGinn (1983: 135–7) for responses,
although McGinn (1996) agrees that colours don’t look like dispositions.
29
See Boghossian and Velleman who argue that ‘the experience of seeing red is unmistakeably an
experience of a quality that could not be experienced other than visually’ (Boghossian and Velleman 1997:
96). In defence of this idea, we may appeal to facts about colour epistemology which, it seems, can be
grasped only from the point of view of colour experience, such as that purple is darker than yellow, that red
is closer to purple than it is to green, and that orange is binary whereas red is unitary. See Broakes (1997),
Hardin (1997), Shoemaker (1997), and Maund (2002). Maund points out that ‘there are crucial features of
colours that are not reproduced at the microstructural level of the physical objects, nor are they explained
at that level’ (Maund 2002: 16), and Shoemaker says that ‘There is good reason to think, for example, that
the phenomenological distinction between “unique” and “binary” hues (e.g., ones like “pure” red, on the
one hand, and ones like orange, on the other) is grounded in a feature of our visual system, and has no basis
in the intrinsic physical properties of the objects we see as coloured’ (Shoemaker 1997: 230). Of course, the
mere fact that colour distinctions do not map onto distinctions drawn at the level of micro-physics does
not show that colour is mind-dependent; the argument requires the plausibility of the further claim that
these claims about colours are essentially linked to visual experience and cannot be grasped apart from this.
ESSENTIALLY MANIFEST QUALITIES 
traditional argument for the view that colour is mind-dependent: the claim that
colour can be accessed only by sight (unlike shape and size, which can be accessed by
more than one sensory modality). This argument for the mind-dependence of colour
is an ancient one, and it does not depend on knowledge about the relation between
colour and wave-lengths of light. A. D. Smith argues that it is the strongest argument
for the secondary quality status of colour (although he characterises this status a bit
differently), saying that
it is not a contingent peculiarity of color qualia that they are not enjoyed through the exercise
of smell or touch: for they define the sense modality of sight. When an object is perceived
through a sense modality, the intrinsic nature of that sense determines how that object
sensorily appears to the subject. (Smith 1990: 239)
The idea is that we cannot make sense of accessing colour through any modality
other than vision, because colour has an essential relation to visual experience; it is
essentially a feature of the way objects look.
The primitivist objectivist view accommodates the first pressure by rejecting the
second; it says that colours are qualities of objects and denies that they are essentially
visual. The idealist dispositional view does the reverse; it sees colours as essentially
visual, but as not really features of objects. The idea that colours are essentially
manifest properties accommodates both. According to this view, both the disposi-
tional view and the objectivist view have something right and something wrong.
What is right about the dispositional view is the idea that there is an essential
connection between colour and visual experience. But, unlike the dispositional
view, the essentially manifest view does not think that this should be captured by
seeing colour as belonging to merely mental states. What is right about the objectivist
view is that colours are qualities of external objects that can be accessed in perception
and not something that belongs to merely mental states. However, unlike the
objectivist position, the essentially manifest view does not hold that colour experi-
ence reveals a property which objects have independently of their perceptually
appearing to us. Like the primitivist objectivist, the proponent of essential manifest-
ness holds that colours are qualities of objects which are manifest to us in experience.
But the account goes beyond this in holding that colours are not just manifest
qualities, but are qualities objects have only in their perceptual appearing to subjects
like us—essentially manifest qualities. They are mind-dependent relational qualities
of objects.
This account of colour requires the view of perception outlined in the previous
section: it requires an account of perception which allows that we can be directly
presented with features of external objects which they do not have independently of
(the possibility of) their perceptual appearing to us. As with our relational view of
non-veridical perception, the essentially manifest view does not understand colour
experience in terms of merely mental effects. Our relational account of perception
allows that perceptual mental states may present objects as being in some respects
 MANIFEST REALITY
different from the way they are independently of their appearing. This allows for the
possibility that colour, like non-veridical appearances, could be part of the way
objects are in our perceptual experience of them while not being a feature objects
have as they are in themselves. According to the account we are considering,
essentially manifest properties and non-veridical appearings have in common the
fact that they belong to perceptual appearances, and not to things as they are in
themselves apart from their perceptual appearing. But they are also importantly
different. In the case of non-veridical perception that we considered using the
metaphor of the distorting glass, the shapes of things are presented as being (to
some extent) other than they really are. Essentially manifest qualities, in contrast, are
qualities objects have only in their perceptual appearing to subjects. This means that
colour experience is not (on this view) a kind of non-veridical perception, because it
does not misrepresent mind-independent features; rather, it does not present mind-
independent features of things at all. It presents qualities things have only in their
perceptually appearing in the visual experience of subjects like us. Colour experience
does not misrepresent the mind-independent qualities of reality which appear red.
Colour experience presents colours, and these are the qualities which are manifest in
experience, not the mind-independent features of objects which are responsible for
their appearing coloured as they do. It is only within visual perception that it makes
sense to talk about non-veridical perception of colour.30
The essentially manifest view of colour conceives of colours as relational qualities
which are partly dependent on how coloured objects are in themselves and partly
dependent on subjects (and also environmental context). This means that there are
features of objects which they have independently of their appearing to us that are
responsible for the colours we experience them as having or that are the mind-
independent correlates or grounds of the colours that appear to us. According to the
essentially manifest view, the nature of colour is revealed in colour experience, and
colour experience does not reveal the nature of the mind-independent features of
objects which appear coloured, or which ground colour. In this sense, colour
experience does not present any feature of things as they are in themselves: colour
experience presents us with qualities of things (things themselves) but it does not
present them as they are in themselves. It presents qualities which belong only to
things’ perceptual appearing.
The idea that colours are essentially manifest goes beyond the claim that they are
manifest; the thought is that they do not exist apart from the possibility of visual
experience like ours, so being red, for example, is essentially a feature of the way an
object looks. The intuitive thought is that it follows from this that there is an essential
30
When we distinguish colour-blindness from normal vision, or talk about things seeming to have
different colours from the colours they actually have, this is done with reference to the visual experience of
normal observers in normal conditions and not with respect to the underlying physics, or causal
explanations involving properties other than visual colour.
ESSENTIALLY MANIFEST QUALITIES 
connection between redness and conscious experience; how exactly to characterise

and precisely formalise this kind of mind-dependence is a question requiring much
further philosophical work. According to the view of colour as essentially manifest, it
depends both on objects and on conscious subjects; appearing is relational. How to
understand relations in general (even before we get to the specific question of these
relational properties) and what their ontological requirements are is an area of
complicated philosophical debate. With respect to dispositional properties, for
example, there is dispute about whether and in what sense an object’s having such
a property depends on the existence, or the possible existence, of the things it is a
disposition to affect. For example, intuitively, we might wish to say that salt would
continue to have the property of being soluble in water even if all the water in the
world dried up. On the other hand, the property requires reference to water to be
understood: we cannot describe or understand what the property is without describ-
ing how the salt would relate to things other than itself—in particular, water.
Attributing this property to salt seems to require at least the possible existence of
water, and it might be thought to require the actual laws of nature as well. Arguably,
unless we index salt to the actual world, with the actual laws of nature, we will have to
attribute to things many dispositional properties we think they don’t actually have
(wood might have the property of being soluble in water in worlds with different
laws, but we don’t take this to imply that wood is soluble). This suggests that we
might want to say that salt’s having the property of being soluble in water depends on
the possible existence of water as well as the laws that govern water and salt in the
actual world. The point here is not to resolve how to understand the relationality of
dispositions like solubility; it is simply to note the complexity of understanding such
relational properties. They depend on their relata in some sense, even if this does not
mean that they depend on the actual existence of both relata. The relational view is
not a traditional dispositional view of perception: perceptual appearing is not
analysed in terms of objects having dispositions to produce ideas in us (merely
mental modifications in subjects). However, like dispositional properties, perceptual
appearing is essentially relational. It attributes to objects ways they appear in the
conscious experiences of subjects, and this involves reference to minds. Properties
that objects have only as they appear to subjects are, at the very least, something that
cannot be understood without the possibility of (specific kinds of) conscious subjects:
objects do not have essentially manifest properties independently of the possible
relations those objects have to (specific kinds of) conscious subjects. Essentially
manifest qualities are features of how objects would perceptually appear to suitable
subjects, and they are nothing but such features.
We could express the idea that colour is a manifest quality (which is compatible
with realism about colour) in terms of a conditional: if something is coloured, then
subjects in appropriate conditions, who are appropriately receptive, will perceive its
colour. To capture the idea that colour is essentially manifest we need something
stronger: an object is coloured only if there is a way it would appear to subjects who are
 MANIFEST REALITY
suitably situated and suitably receptive. Here, the object’s being coloured is dependent
on its possible appearance to conscious subjects. Note that this conditional claim
expresses a relation to mind but need not be taken as telling us any more about the
nature of colours than that they have an essential relation to conscious experience.
Saying that a property is mind-dependent is not the same as explaining the nature or
essence of the property, so we need not take the claim to tell us what colour is. Similarly,
for the primitivist objectivist to say that if something is coloured then a suitably situated
and suitably receptive observer will perceive its colour does not tell us anything about
the nature of colour other than that it is a manifest quality. The conditional formula-
tion captures the idea that there is an essential relation between colours and visual
experience, but does not tell us more than this about what kinds of properties colours
are. According to this account of colour, colour is a property we see (it is part of how
objects appear to us, or manifest themselves in our visual experience) and colour is
essentially such a property; it is essentially linked to visual experience.
Another (perhaps not equivalent) way we might try to capture the idea of there
being an essential relation between colour and visual experience is the denial of
experience-transcendence: the thought is that as essentially visual qualities colours do
not have natures which transcend what can be presented in visual experience. Both
the conditional and the experience-transcendence formulation capture the idea that
colour has an essential relation to visual experience. Neither the conditional nor the
experience-transcendence formulations say that colours exist only in particular
events of their actually being perceived; both allow that we can say that objects
have colour when they are not currently being perceived, that objects had colours
before sentient creatures evolved, and would continue to have colours if sentient
creatures ceased to exist. Both allow that there are ways of being mind-dependent
other than existing merely in minds, as modifications of inner states of subjects. On
these positions, colour is a quality of objects which they have in relation to the
possible visual experiences of subjects like us.
To understand Kant’s suggestion that his idealism can be understood as making
spatial properties like shape and size mind-dependent like colour, we need an
account of the mind-dependence of colour. I suggest that the account we need is
one that holds that colour is an essentially manifest quality. Essentially manifest
qualities are relational, mind-dependent qualities of things which can be present in
perceptual experience, and which do not present us with qualities things have as they
are in themselves, independent of their perceptually appearing to us. Kant’s idealism
about appearances can be understood as the claim that our experience and empirical
knowledge is limited to essentially manifest features of reality. This claim will be my
subject in Chapter 6. Kant’s idealism, however, is not based on general features of
perceptual experience, or our being affected by objects, but on specific claims about
space and time; this will be the subject of Chapters 7 and 8.
6
The Secondary Quality Analogy
I Introduction
In Chapter 5, I presented an account of essentially manifest qualities using colour as
an example. My suggestion here is that we can explain Kant’s idealism by taking this
account of colour and generalising it to all the qualities of which we can have
experience. On my account, Kant’s idealism about things as they appear to us is
the view that what appear to us are essentially manifest features of reality and we can
cognize only essentially manifest features of reality. In this chapter I argue that this
reading fits with the textual evidence concerning the mind-dependence of appear-
ances, as well as the relation between appearances and things in themselves.
Kant himself famously suggests that his idealism can be understood in terms of an
analogy with secondary qualities like colour.1 He suggests the analogy in the following
passage:
That one could, without detracting from the actual existence of outer things, say of a great many
of their predicates: they belong not to these things in themselves, but only to their appearances,
and have no existence of their own outside our representation, is something that was generally
accepted and acknowledged long before Locke’s time, though more commonly thereafter. To
these predicates belong warmth, color, taste, etc. That I, however, even beyond these, include (for
weighty reasons) also among mere appearances the remaining qualities of bodies, which are
called primarias: extension, place, and more generally space along with everything that depends
on it (impenetrability or materiality, shape, etc.), is something against which not the least ground
of uncertainty can be raised; and as little as someone can be called an idealist because he wants to
admit colors as properties that attach not to the object in itself, but only to the sense of vision as
modifications, just as little can my system be called idealist simply because I find that even more
of, nay, all of the properties that make up the intuition of a body, belong merely to its appearance:
for the existence of the thing that appears is not thereby nullified, as with real idealism, but it is
only shown that through the senses we cannot cognize it at all as it is in itself. (Proleg. 4: 289)
1
His suggestion is taken up by Putnam (1981: 59) and Collins (1999: 11–12). Van Cleve (1995) argues
that no version of the secondary quality analogy makes sense of Kant’s idealism. See also Bird (1962: 46–9).
I develop an account of the secondary quality analogy in Allais (2007); see Rosefeldt (2007) for a similar
view. Collins (1999) denies that he reads Kant in terms of a secondary quality analogy, since he thinks that
a secondary quality account of colour is committed to mentalising colour, something he accuses Putnam of
doing. However, I think that my account of the mind-dependent status of colour is close to what Collins
calls subjectivism about colour, so that the difference here may be more verbal than substantial.
 MANIFEST REALITY
The central problem in making use of this analogy to explain Kant’s idealism
is deciding on the account of colour to read into the analogy. Clearly, using
different accounts of colour will lead to very different interpretations of tran-
scendental idealism. If we take the primitivist objectivist view mentioned in
Chapter 5 (which sees manifest colour as a mind-independent property of
objects), the analogy will give us a version of transcendental idealism which
says that we can have knowledge only of manifest qualities. This view limits
our knowledge to things we can experience but it is not idealist, since it sees
experience as presenting us with features of things as they are independently of
us and of our being able to experience them. On the other hand, the subjectivist
dispositional view of colour considered in Chapter 5 would give us an extreme
idealist (phenomenalist) interpretation of Kant’s idealism. According to this
position, colour exists only as a property of merely mental inner states, modifi-
cations of a state of the subject: saying that the appearances of which we can have
knowledge are mind-dependent in the way that colour is would be saying that
appearances exist merely in minds. Neither of these ways of understanding
colour enables us to avoid the traditional interpretative extremes in understand-
ing Kant’s idealism.2
Kant does not spell out the account of secondary qualities he wants to appeal to in
the analogy. It might be thought that the most straightforward reading would be
to use a Lockean subjectivist dispositionalist view of secondary qualities. This view
was influential and well known, the passage in which Kant presents the analogy
mentions Locke, and Kant, at least sometimes, seems to ascribe colour solely to
merely mental states or sensations (B45). These factors might support thinking that
Kant’s secondary quality analogy should be read in terms of a subjectivist disposi-
tional account of colour, and therefore understood to support an extreme idealist,
2
Van Cleve argues, plausibly, that reading Kant’s suggested analogy using Lockean and Berkeleyan
accounts of properties like colour, as Van Cleve understands Locke and Berkeley, has the same problem: it
does not enable us to avoid the traditional interpretative extremes in understanding Kant’s idealism. As he
sees it, for Locke, ‘a secondary quality is something quite definitely in the object, namely a power’, whereas
for Berkeley, ‘a secondary quality is a quality that exists only in the mind’ (Van Cleve 1995: 84). (Of course,
for Berkeley, this does not make such properties secondary, as all properties of physical objects are like
this.) To read the secondary quality analogy in terms of the Berkeleyan account of qualities like colour
would clearly lead to the extreme idealist account of Kantian appearances. On the other hand, according to
Locke (as he is interpreted by Van Cleve), secondary qualities are dispositions of objects, and Locke’s
‘subjectivism about colors, tastes, and the like is expressed not by saying that secondary qualities exist only
in the mind, but by saying that our ideas of secondary qualities (which do exist only in the mind) do not
resemble anything in the object that causes them’ (Van Cleve 1995: 84). (For an alternative view of Locke
see Yolton 2000). If we follow this reading of Locke, we will have a choice about how to understand the
secondary quality analogy: we could explain the status of appearances by comparing them either to the
mind-independent powers that cause our ideas of them (this will be similar, in some respects, to Rae
Langton’s 1998 reading), or to the merely mental ideas in the subject which fail to resemble their causes.
The second option once again reduces to Berkeleyan idealism. The first option fails to capture any sense in
which Kantian appearances are mind-dependent, and, implausibly, identifies Kantian appearances with
qualities whose natures are not presented to us in perceptual experience.
THE SECONDARY QUALITY ANALOGY 
phenomenalist interpretation of transcendental idealism. I will argue against

this way of understanding the analogy, in favour of using the essentially manifest
account of colour.
In addition to the problem that Kant does not tell us what account of secondary
qualities to use in his analogy, it can also be argued that we should not make use of
the secondary quality analogy at all, since it is presented only in the Prolegomena,
while in the second edition of the Critique Kant explicitly rejects explaining tran-
scendental idealism through a comparison with qualities like colour. He says that he
wants to prevent us
from thinking of illustrating the asserted ideality of space with completely inadequate
examples, since things like colors, tastes, etc., are correctly considered not as qualities of things
but as mere alterations of our subject, which can even be different in different people. (B45)
Arguably, the Critique must take precedence over the Prolegomena in interpretative
disagreements, since the latter was written to popularise the former, and the second
edition of the Critique in which the denial of the analogy occurs was written after the
Prolegomena. This might be thought to show the analogy to be a non-starter.
However, an alternative possibility is that we can take the explicit denial in the
Critique to constrain the account of secondary qualities to which we can appeal in
reading the Prolegomena passage; we should not interpret the analogy in the Prole-
gomena in terms of the account of secondary qualities with respect to which Kant
denies the analogy in the Critique.
The two texts can be made compatible if we allow that Kant has a different account
of secondary qualities in mind in each. Kant clearly denies that his idealism can be
illustrated by comparison with secondary qualities on one understanding of those
qualities—an understanding which sees them as merely states of the subject, in no
way belonging to the object (B45). This means that we must not interpret the
secondary quality analogy using an account of secondary qualities on which colours
exist merely as subjective effects in individuals’ minds, or mere alterations of the
subject (the idealist dispositionalist account of colour). The account of secondary
qualities we need for the analogy need not be Kant’s settled view of such qualities (if
he had one); in the Prolegomena passage he could simply be saying that there is a
common view that colour is mind-dependent in some way and that his position can
be seen as saying that shape is mind-dependent in the same way. I suggest that
we read the analogy using the essentially manifest account of secondary qualities
like colour.
The interpretative desiderata developed in Chapters 1–4 suggested that in order to
explain Kant’s idealism we need an account of perceptual appearing which does not
see appearances as mental entities but which still captures a significant sense in which
appearances are mind-dependent. In the passage in which he presents the analogy we
find Kant trying to express exactly this idea: he says that there are properties of things
which belong to things as they perceptually appear and not as they are in themselves.
 MANIFEST REALITY
He says that there is a sense in which these properties are mind-dependent and have
no existence ‘outside our representation’, but he also says that saying that all
properties that are given in intuition are like this is not idealist. And immediately
before the passage in which he presents the analogy, Kant defines idealism in terms of
the idea that the immediate objects of perception have a merely mental existence and
he denies that this is his position. He says:
Idealism consists in the claim that there are none other than thinking beings; the other things
that we believe we perceive in intuition are only representations in thinking beings, to which in
fact no object existing outside these corresponds. I say in opposition: There are things given to
us as objects of our senses existing outside us, yet we know nothing of what they may be in
themselves, but are acquainted only with their appearance, that is, the representations that they
produce in us because they affect our senses. (Proleg. 4: 289)
Kant seems both to assert and to deny that the things we perceive in intuition are
representations in us. He says that appearances have no existence outside our
representations (hätten außer unsere Vorstellung keine eigene Existenz), but also
that they are not mere representations in thinking beings (Vorstellungen in den
denkenden Wesen). In the passage discussing the analogy with secondary qualities,
Kant says that everyone agrees that there are properties which belong only to the
appearances of bodies, such as Lockean secondaries. He says that the move he is
introducing is to make all the properties of bodies which we experience into mere
appearances. And he says that this move does not amount to idealism, which he has
just defined as a view which holds that the ‘things that we believe we perceive in
intuition are only representations in thinking beings’. He asserts that there are
entirely mind-independent things, that these things are given to us as objects of the
senses existing outside us, but that we know nothing about them as they are in
themselves.
So long as we remain with a Cartesian account of perception according to which
mental representations are only in us, it is extremely difficult to make sense of this.
However, the idea of essentially manifest qualities gives us exactly what we need. On
this account, appearances do not exist merely in mental states, but neither do they do
exist in things as they are apart from their perceptually appearing to us. They have no
existence outside our representations, but they are not representations in thinking
beings. This enables us both to give a genuine sense in which appearances are mind-
dependent, and at the same time to do justice to their empirical reality. The claim that
empirical reality is essentially manifest is a form of idealism, though it is not
phenomenalist. Realist views of Kantian appearances like those of Dryer (1966)
and Bird (2006) see Kant as limiting our knowledge to what can appear to our
senses. In contrast, the essentially manifest view is not realist: it does not say merely
that what appear to us are perceptible things (things which can be manifest to us);
they are essentially perceptible, or essentially manifest. It is not just that we can have
knowledge only of things which affect our senses, but rather that the features of
THE SECONDARY QUALITY ANALOGY 
reality which we can know have an essential relation to our senses and do not present
us with things as they are independently of their perceptually appearing to us.
II Kant’s Distinction between Things

in Themselves and Appearances
As we have seen, Kant seems to assume, without argument, that there is a way things
are as they are in themselves, and he frequently talks about things as they are in
themselves and these same things as they appear,3 and about things in themselves as
grounding appearances, or as the causes of appearances.4 All of these features of the
text are awkward for a phenomenalist reading of appearances, but are straightfor-
ward for the essential manifestness view.
On the essential manifestness view, the things that perceptually appear to us are
things which have a way they are independently of us and the notion of perceptual
appearing requires this. Understanding appearances as essentially manifest does not
cast doubt on the starting assumption that there are things in themselves; rather,
it reinforces this assumption. This is exactly how Kant seems to see his position.
He says:
It . . . follows naturally from the concept of an appearance in general that something must
correspond to it which is not in itself appearance, for appearance can be nothing for itself and
outside of our kind of representation; thus, if there is not to be a constant circle, the word
“appearance” must already indicate a relation to something the immediate representation of
which is, to be sure, sensible, but which in itself, without this constitution of our sensibility (on
which the form of our intuition is grounded), must be something, i.e., an object independent of
sensibility. (A251–2)
In contrast, if we were to understand Kantian appearances in terms of collections of

ideas in minds (with the right principles of organisation), and add that we can have
knowledge only within experience, his continued commitment to mind-independent
things in themselves is in need of much more argument than he gives it. As we saw in
Chapter 2, an argument frequently given against phenomenalist interpretations of
appearances is the fact that Kant claims that his notion of appearance implies the
existence of something that appears. This is not straightforwardly a feature of a
phenomenalist analysis of perceptual appearances. Berkeley would deny it of his
empirically real objects. Of course, Berkeleyan idealists may, like Berkeley, think that
something exists other than our ideas that is responsible for their existence or for
their order and coherence, but Berkeley would not say that objects are appearances of
this something: that empirically real objects are appearances of God. In contrast,
Kant’s claim makes immediate sense on the view that appearances are essentially
3
For example, Bxx, Bxxvii, A38/B55, A39/B56, A42/B59, B69, B306, A276/B332, A546/B574, and A360.
4
A44/B61; A190/B235; A251–2; A288/B344; A379–80; A393; A494/B522; A496/B524.
 MANIFEST REALITY
manifest: essentially manifest appearances are appearances of something independ-

ent of us and, on the relational view, the idea that things have a way they appear to us
does indeed imply that there are things which are appearing.
Throughout the Critique, Kant talks about his distinction as being one between
things as objects of experience and these same things as they are in themselves. There
is no sense in which merely mental ideas, modifications of subjects’ states, can be said
to be the aspects of the very same things that exist outside us. On the other hand, this
is entirely straightforward for the essentially manifest view. Red perceptual appear-
ances are appearances of those very things which have a way in which they are
independently of us, even though these appearances do not present things as they are
in themselves.
As we have seen, Kant says both that appearances represent things and that they
do not represent things as they are in themselves.5 The essential manifestness view
explains this. The idea that red is essentially manifest means that perceptual appear-
ings of red do not present mind-independent things as they are in themselves;
redness is a quality that belongs only to appearances. Experience of red things does
not reveal or manifest the mind-independent qualities that are responsible for, or
appear as, red. This means we cannot identify redness with its mind-independent
grounds, and it also means that redness does not represent this ground as it is in
itself. There is thus a sense in which colour presents things, things themselves, but it
does not present them as they are in themselves.
In addition to talking about things in themselves and those same things as they
appear to us, Kant sometimes talks about two ways of considering objects. This gives
some support to deflationary interpretations, but is difficult for the phenomenalist.
On the other hand, deflationary or non-idealist readings cannot do justice to Kant’s
idealism. The essential manifestness view can account for both of these sides of
Kant’s position, and can show both the uses and the limitations of explaining Kant’s
position by appealing to the thought of two distinct ways of considering objects.
One can use the idea of two different ways of considering objects to introduce a
metaphysical distinction. Consider the essentially manifest view of colour. Someone
wanting to introduce the idea of the mind-dependence of colour might start by
saying that we can distinguish between considering objects (or perhaps surface
properties) as they are in perceptual experience and as they are apart from this.
The idea of considering things in these two different ways is a helpful way of
introducing the idea of there being a difference between the surfaces of things as
they appear and as they are in themselves. However, appealing to these two ways of
considering objects does not get us very far: the idea of considering things in these
two ways does not explain the difference between mind-independent surface prop-
erties and the mind-dependent appearances. And the fact that considering things in
5
B164/A252; A44/B61–2; A271/B327; A276/B327.
THE SECONDARY QUALITY ANALOGY 
these two ways is a useful way of introducing a distinction between colours and their
mind-independent grounds does not imply that this distinction is simply one
between two ways of considering things: the difference between the surface of the
object as it is in itself and its visual appearance is not just a difference in the way it is
considered. Although the deflationary view rightly points out that Kant talks about
two ways of considering things, this is simply a starting point: Kant goes on to make
substantial claims about the things considered in these two ways. Unlike deflationary
and empirically realist views, the essentially manifest view accommodates Kant’s
idealism; unlike phenomenalist readings, it accommodates his talking about a dis-
tinction between two ways of considering the same objects.
Kant frequently speaks about things in themselves as grounding appearances and
even as causing appearances (see Chapter 1 for references). This has been thought to
create problems for his position on any interpretation, and it is an issue I return to in
Chapter 10. For now, I simply comment that the claim that colour is essentially
manifest is not just compatible with but in fact is naturally taken to require the
existence of some mind-independent features of things that are, in some sense,
responsible for the colours things appear to have in perceptual experience or that
ground colours. How this ‘responsibility’ should be understood is a tricky question,
but the point for now is simply that it fits neatly with the way Kant talks about the
relation between things as they are in themselves and appearances.
In explaining the relation between things in themselves and appearances there is a
further useful feature of the analogy with qualities like colour, as the essentially
manifest view understands such qualities. The essentially manifest view holds that
colours are relational appearances and thus that there is a way things are in
themselves that is responsible for or grounds colour experience. Suppose that an
object’s micro-physical surface properties, or wavelengths of light, or spectral reflect-
ance properties, or some combination of these, is the mind-independent ground of
visual colour. It turns out that there is no simple correlation between particular
colours and any of these features.6 Instead, the physical bases of colour are multiply
disjunctive, and it may be that the only thing that unites the various features of mind-
independent reality that we experience as red is their red appearance. There are
features of mind-independent reality that are, in a sense, what appear to us as red, but
there is no simple correlation between these features and the redness that features in
experience, and the visual presentation of red does not present or reveal the mind-
independent features of things that appear red as they are in themselves.
Of course the analogy with Kant’s position here is not perfect: the fact that we are
able to talk about the complex relation between colours and spectral reflectance
properties, or wavelengths of light, is a function of the fact that we (take ourselves to)
have knowledge of the mind-independent bases of colour appearances. Kant thinks
6
See Byrne and Hilbert (1997b).
 MANIFEST REALITY
that all the properties we experience are mind-dependent appearances and that we
have knowledge only of mind-dependent appearances, and the explanations of
things’ properties we give in science also involve essentially manifest qualities. This
means that, unlike with visual colour and, for example, spectral reflectance proper-
ties, we are not in a position to have any insight at all into what kinds of correlations
there are between mind-dependent appearances and the features of things in them-
selves that ground them. Still, the knowledge we have of colour in this respect is
helpful in thinking about the relation between things in themselves and appearances.
It gives us an actual case in which there is no neat, one-to-one mapping between
manifest qualities, qualities which appear in visual experience, and the features of
things as they are in themselves that ground or explain these qualities. It enables us
to see why one might want to say that it is things in themselves that we experience
as coloured, since these natures ground colour experience, but it also enables us
to see why one might want to say that colour cannot be identified with the mind-
independent grounds of colour appearances: visual colour exists only in the
mind-dependent appearances. The nature of red is transparently revealed in visual
experience of colour, but the mind-independent grounds of red are not revealed at all
in colour experience. Experience of red does not even give us any insight into how
mind-independent reality is divided up, since it is only at the level of visual experi-
ence that the world is divided into red and non-red things. On the one hand, there are
mind-independent qualities of objects which ground red perceptual appearances,
and which are, in a sense, what appear to us as red, but, on the other hand, these
qualities do not appear to us as they are in themselves. These features of colour
experience (on this account of colour) fit well with significant parts of Kant’s view of
the relation between appearances and things in themselves. He thinks that experience
of appearances gives us no insight at all into what things are like in themselves; it does
not tell us what joints there are in reality in itself and it gives us no insight into the
mind-independent natures of things which are responsible for the appearances
things have. He also thinks, however, that the nature of things in themselves grounds
appearances, and that it is things—things themselves—that appear to us. I return to
this issue in Chapter 10.
III Kant’s Idealism

The central textual grounds for idealist readings are Kant’s saying that appearances
are representations that exist in us, and his saying that their existence requires a
connection to possible perception.7 The idea that appearances are essentially mani-
fest is able to account fully for these features of the text.
7
A42/B59; B45; A46/B63; A98; A101; A104–5; A109; A113; A127; B164; A190/B235; A370; A197/B242;
A369; A372; A376; A383; A385; A386; A490/B518; A492–3/B521; A494–6/B522–4; A499/B527; A505–6/
B533–4; A514–15/B542–3; A563/B591; A793/B821; Proleg. 4: 288, 289, 319, 341, 342.
THE SECONDARY QUALITY ANALOGY 
The passages in which Kant says that appearances are mere representations are
among the most extreme-idealist sounding in the Critique. For example, Kant says:
or time, hence all objects of an experience possible for us, are nothing but appearances, i.e.,
ations, have outside our thoughts no existence grounded in itself. (A490–1/B518–19)
We saw in Chapter 5 that there is a way of making sense of talking of representations

within the context of a direct realist or relational account of perception: such an
account can say that perceptual experience represents objects to us. Further, we
considered a version of a direct realist or relational account which allows that
perception may sometimes present things as being other than the way they are
apart from their perceptual appearing. In this case, not only does perception repre-
sent things in certain ways, but there are aspects of what is presented that belong only
to the perceptual representation of the object and are not features of the object as it is
apart from its perceptual appearing. They are mere representations. According to the
relational view of perception presented in the previous chapter, both non-veridical
perceptual appearances and essentially manifest qualities (like colour) represent
objects to us as being other than the way these objects are independently of their
perceptually appearing. Although they are not merely modifications of subjects’
inner states, these representations are mind-dependent and do not present objects
as they are in themselves. However, this does not entail that they are merely
alterations of internal states of subjects. They are the mind-dependent way some-
thing mind-independent is presented to us. As Kant puts it, ‘What is not to be
encountered in the object in itself at all, but is always to be encountered in its relation
to the subject and is inseparable from the representation of the object, is appearance’
(B70n). Thus, the essential manifestness view can fully take account of and explain
Kant’s calling appearances representations, and does not need to explain it away or
see Kant as inconsistent. In keeping with what such passages suggest, the essential
manifestness view is a robust and radical form of idealism.
The next feature of the text that seems to support phenomenalism is Kant’s saying
that appearances exist ‘in us’ and this might be thought to be less well captured by the
essentially manifest account. One response to this is to say that the claim that
appearances are ‘in us’ is simply a metaphorical way of expressing idealism or mind-
dependence, leaving it open to interpretation exactly how this mind-dependence
should be understood. While I think this is right as far as it goes, the essentially
manifest view is able to do better than this; it can explain both the sense in which Kant
says appearances are in us and the sense in which he denies this.
As we saw in Chapter 1, Kant distinguishes between a transcendental and an
empirical sense in which objects can be outside us (A373), which implies corres-
ponding transcendental and empirical senses in which objects can be in us. In the
transcendental sense, to say that an object is in us is not to deny that it is outside us in
 MANIFEST REALITY
space but to say that it does not exist distinct from us or independently of us. Kant
describes a kind of idealism that he explicitly rejects as holding that ‘the other things
that we believe we perceive in intuition are only representations in thinking beings’
and says that on his view, in contrast, mind-independent things are given to us in
sensibility, although we know only their appearances and are not given them as they
are in themselves (Proleg. 4: 289). And he rejects a version of the secondary quality
analogy that understands colours as ‘mere alterations of our subject’ (B45). On the
other hand, he says that spatio-temporal objects are appearances which exist tran-
scendentally in us. Ideally, an interpretation should be able to explain both of these
senses of ‘in us’, and to show why appearances are in us in the one sense but not the
other; the essentially manifest view enables us to do this. It is crucial to see that
relational views of perception give a different account than does the Cartesian view of
what it is for something to be ‘in’ a mental state. Unlike the Cartesian view, the
relational views I presented in Chapter 5 hold that perceptual mental states have
external objects as their constituents: external objects can be in mental states.
Perceptual mental states differ from (possibly subjectively indistinguishable) hallu-
cinations in that the object perceived is actually present in the perceptual mental
states. So relational views can allow a sense in which an object can be in a mental state
which is not a matter of being merely a modification of a subject’s internal state. For
the Cartesian view, something which is in a mental state is thereby only in a subject’s
mind. For relational views, in contrast, external objects can be in mental states;
objects are present in conscious mental states when we perceive them. The version
of the relational view I have appealed to allows the possibility of features of objects
which exist only in their perceptual appearing—essentially manifest qualities. Since,
as we have just seen, there is a sense in which, on the relational view, objects can be in
mental states (in us), similarly, features of objects which they have only as they
perceptually appear can be said to exist only in us. They do not exist as objects
distinct from us, and therefore are not outside us in the transcendental sense. At the
same time, such appearances are not in us in the Cartesian sense of things which exist
merely as modifications of inner states. The essential manifestness view is therefore
able to make sense of Kant’s both asserting and denying that appearances are in us: of
how appearances can be transcendentally in us (they don’t exist distinct from us)
while not being empirically in us (they are outside us, in space).
The essentially manifest view is therefore able to accommodate the two sides of
Kant’s idealism that we saw at the end of Chapter 1. On the one hand, Kant
characterises a form of idealism he rejects as one which says that what we immedi-
ately or directly perceive are ideas in the mind. He associates Berkeley with the idea
that there are only such things (and minds), and Descartes with the idea that what we
immediately perceive are things which exist merely in the mind, but that in addition
there are mind-independent things which are the causes of the mental ideas with
which we are in immediate contact. Kant rejects both views, and says that, in contrast
to them, he will show that we have immediate experience of objects outside us in
THE SECONDARY QUALITY ANALOGY 
space, and that even our being able to cognize the temporal order of our own inner,
mental states depends on this. The phenomenalist account cannot give a straight-
forward interpretation of this. In contrast, an explanation of this is easy for the
essential manifestness account, since this account does not make empirically real
objects into ideas in minds. In a sense, we have not abandoned direct realism in
moving to transcendental idealism. We started with a relational or direct realist view
that allows us to directly perceive essentially manifest qualities, qualities which
belong to objects only in relation to their possible perceptual appearing to us. The
transcendental idealist move is to say that all of the qualities that are directly
perceived are mind-dependent essentially manifest qualities which do not reveal
the mind-independent natures of the things that are appearing, and that our know-
ledge is limited to features of the world which can be manifest to us. While this
position makes the radical claim that we can experience and have knowledge only of
features of reality that are mind-dependent (that exist only in the perceptual appear-
ing of objects to us), it straightforwardly allows that the direct objects of perception
are objects in space, not constructions out of merely mental states. We have both a
radical form of idealism and a robust empirical realism.
A possible worry about illustrating the status of Kantian appearances by compar-
ing them with essentially manifest qualities like colour is that, for Kant, appearances
are objects, not properties, and it might be objected that my account makes them into
properties.8 This impression is misleading. The position makes a claim about the
status of the properties that constitute physical objects. We characterise objects by
characterising their properties; Locke, for example, characterises objects as having
primary, secondary, and tertiary qualities (which have different metaphysical sta-
tuses, as secondary qualities are mind-dependent). This does not mean that Locke
turns objects into properties, and we could express Lockean realism by describing the
mind-independent nature of primary qualities, in his account. In giving an account
of the metaphysical status of the properties of spatio-temporal objects, Kant gives an
account of the metaphysical status of such objects. Kant thinks, as I read him, that the
essentially manifest qualities of things we are able to cognize exhaust the character-
istics of spatio-temporal empirically real objects. He thinks that empirical knowledge
and science gives us knowledge only of essentially manifest objects and does not need
anything more. This does not make spatio-temporal, empirically real objects into
properties; rather, it gives us a description of the status of the properties of the objects
we know, and thereby gives us an account of the metaphysical status of the objects.
As we have seen, Kant sometimes says that our experience presents us with things
themselves but also denies that our experience represents things as they are in
themselves. Similarly, he often uses ‘object’ for entities that exist within appearances
8
Barker, for example, argues that Kant is committed to a sense-data language because he speaks of
appearances as the objects of perception; Barker argues that if he were using the language of appearing, he
ought to speak of things in themselves as the objects of perception (Barker 1969: 207).
 MANIFEST REALITY
(e.g. A34/B50, B306), but he also often uses it in the context of things in themselves.9
According to my interpretation, this is what we should expect. Everything we know
about objects is at the level of appearances, and appearances do not represent any
things as they are in themselves. Spatio-temporal objects are mere representations.
On the other hand, there is a sense in which appearances are appearances of things
themselves: appearances present us with things, but do not present us with these
things as they are in themselves. Spatio-temporal objects are mind-appearances that
are grounded in something we do not cognize, and which we do not need to
characterise in order to cognize spatio-temporal objects.
At the beginning of this chapter, I noted that there is debate about how Kant
himself understood secondary qualities like colour; this might be thought to create
difficulties for my use of the secondary quality analogy. I have suggested that the
account of colour we need to use to understand the Prolegomena analogy need not be
Kant’s settled view of the status of colour (assuming he had one). Most importantly, it
should not be the view of colour Kant mentions in the B45 passage where he denies
that his idealism can be illustrated by comparison with qualities like colour. However,
an objector might say that no one who thinks of colour as Kant seems to do in that
passage—as something entirely in the mind and not really a property of objects—
could have the direct realist account of appearances to which I have appealed. Our
relational account of appearances needs to be able to allow that some aspects of what
perceptually appears to us are not objective or empirically real, since this is how Kant
seems to think of colour in the Critique, and since he allows for empirical illusion.
There are a number of ways a direct realist of the sort I have discussed can
accommodate this; I will sketch two of these.
One option is to say that, although we directly perceive objects, some of the
qualities which perceptual experience presents objects as having are not qualities of
objects at all, but are only ‘modifications of the sense of the subject’ (B45). There are
some features of perceptual experience which are not experienced as presenting
features of the world at all—for example, the blurriness objects appear to have
when a short-sighted person takes their glasses off. A possible view could hold that
there are, in addition, some features of perceptual experience that are experienced as
presenting features of the world but do not really do so. A direct realist could think
that most of the qualities objects are presented as having in perception are qualities of
those objects, even where they are mind-dependent, but that some are only features of
something in subjects’ minds (and therefore not features of empirically real outer
objects), even though they are experienced as features of objects. For example,
consider my looking at an uplifting scene: it might be said that while I directly
perceive the size and colour of the large purple mountain, the ‘upliftingness’ aspect of
9
See A26/B42, A30, A34/B51, A36/B53, A39/B55, A42/B59, A43/B60, A44/B61, A46/B63, B69, A104,
A105, and A109.
THE SECONDARY QUALITY ANALOGY 
my perception is a feature of my mental state projected on the scene, not a quality—

not even a perceiver-dependent quality—of the scene itself.
A second option is to say that as well as our being directly presented with relational
appearance qualities that are relativised to us as a group—that are part of the way we
all represent things as being—and therefore are empirically real, there can also be
directly presented qualities which are relativised to us as subgroups, or as individuals,
as well as qualities which can be relativised to individuals and to specific contexts of
perception. Such qualities would still be perceivable qualities of objects, rather than
aspects of mental states, but would not be part of the appearance of objects to all
humans, and therefore not part of an objective (public) empirically real world. This is
a possible view of colour which allows for genuine variation in colour experience
without making colour a feature of merely mental states.
It seems, on the basis of some brief comments, that in the Critique Kant adopted
the first alternative with respect to colour, though he seems to adopt the second with
respect to some other kinds of non-veridical appearings. Kant clearly thinks he can
accommodate a contrast between illusion and reality within the empirically real
world. He says that ‘[w]e ordinarily distinguish quite well between that which is
essentially attached to the intuition of appearances and is valid for every human
sense in general, and that which pertains to them only contingently, because it is
not valid for the relation of sensibility in general, but only for a particular situation
or organization of this or that sense’, and says that this gives us an account of
empirical appearance (A45/B62).10 Since either view is a possibility for my direct
realist or relational account of colour, my account can allow Kant to say that
qualities like colour are not empirically real. In contrast, a phenomenalist view
has difficulties with what Kant says about primary and secondary qualities at B45:
if colour and taste are contrasted with space by being only properties of something
merely mental, it is harder to see how space could also be a property of something
merely mental.
IV Possible Experience
Kant says that the existence of appearances is essentially linked to our possible
perceptual experience of them. Understanding what Kant means by possible experi-
ence, and the sense in which objects are dependent on it, is therefore key to
understanding his idealism. This issue is so crucial that I give it a separate section
here, and return to it in Chapter 9. The first thing to see is that the way Kant links
empirical reality to possible experience is actually captured better by the essential
10
Similarly, he says that ‘appearance, taken in the transcendental sense, where we say of things that they
are appearances (phaenomena), is a concept quite different in meaning from that whereby I say that this
thing appears to me in this way or that, which is meant to indicate the physical appearance, and can be
called apparency, or seeming’ (Ak. 20: 269).
 MANIFEST REALITY
manifestness account than it is by phenomenalism. We then need to look at what

Kant means by ‘possible’ and by ‘experience’.
Kant says:
the objects of experience are never given in themselves, but only in experience, and they do not
exist at all outside it . . . Thus they are real when they stand in an empirical connection with my
real consciousness, although they are not therefore real in themselves, i.e., outside this progress
of experience. (A493/B521)
[A]ppearances, as mere representations, are real only in perception, which in fact is nothing
but the reality of an empirical representation, i.e. appearance. To call an appearance a real
thing prior to perception means either that in the continuation of experience we must
encounter such a perception, or it has no meaning at all. (A493–4/B521–2; see also A225/
B272 and A375–6)
To an ear tuned only to the opposition between realism and phenomenalism, these
passages sound phenomenalist. Berkeleyan idealism sees the existence of physical
objects as a matter of subjects being in certain mental states. Phenomenalism sees the
existence of objects as a construction out of what exists in merely mental states. On
both accounts, for a spatio-temporal object to exist is a matter of our being in certain
mental states. If we approach Kant with the assumption that perception must be
analysed in terms of external causes and entirely inner modifications of subjects,
Kant’s saying that appearances exist only in possible perception might be taken to
suggest one or the other of these two forms of idealism. However, as we have seen,
this is not the only way of understanding perception. And the passages in which Kant
expresses a connection between appearances and possible experience do not say that
appearances exist merely as ideas in subjects’ minds, or as constructions out of ideas
which exist in subjects. He does not say that the existence of appearances is a matter
of subjects being in certain mental states, but rather that the existence of empirically
real objects does not go beyond the extent of the possible experience of subjects like
us. According to the view that colour is essentially manifest, something is coloured
only if it would perceptually appear coloured to suitably situated and suitably
receptive subjects. Colour does not exist as something merely mental, a modification
of the subject’s inner state, but an object’s being coloured ‘means that in the
continuation of experience we must encounter such a perception’ (A494–B522).
Kant never says and in fact denies that appearances exist only as modifications
of an internal state of a subject, but also says that they are not real apart from
the possibility of our perceiving them. This is exactly the status of essentially
manifest qualities.
An objection commentators have made to phenomenalist interpretations is that
Kant distinguishes objects in space from things in our minds, regards objects in space
as public, says that they exist unperceived, and contrasts them in this regard to our
changing perceptions. The view of colour as essentially manifest can easily accom-
modate these aspects of Kant’s account of empirical reality. The account allows that
THE SECONDARY QUALITY ANALOGY 
objects can be coloured when we are not looking at them and that their colours are
public; to say that an object is coloured is not to say that it produces an idea in our
minds, but rather that if we were to see it (in the right conditions) we would see its
colour. This is something that can be true of objects we are not currently perceiving,
as well as of objects whose existence pre- or post-dates the existence of sentient
perceivers. Of course, phenomenalists claim that they can give an analysis of all these
features of objects in terms of claims about constructions out of possible perceptions.
However, as I argued in Chapter 2, the fact that phenomenalists, rightly or wrongly,
think that they can capture everything that Kant says about external objects in terms
of claims about possible sensations does not show that Kant is a phenomenalist, any
more than the fact that phenomenalists think that their analysis can capture all the
features Locke thinks physical objects have makes Locke a phenomenalist. Kant
himself does not explain the public nature of objects, their existing unperceived,
and their being unlike perceptions in this respect, in the way that a phenomenalist
does. He does not, for example, explain the fact that the same object can be perceived
by both you and me in terms of our having certain similarities in our sensations. He
does not say that what it means for objects to exist unperceived is for us to construct
sensations in certain ways. He denies that they are merely internal states of subjects.
Yet he says that their existence requires a connection to possible experience, and that
we can say that they exist only if it is true that there are some possible conditions in
which we would perceive them. Just as Kant says of appearances, it is true of
essentially manifest qualities that they are given only in experience and do not
exist outside it, and if something is an essentially manifest quality, to say that it is
real is to say that we could experience it.
Kant’s analysis of the relation between what is empirically real and possible
experience is not phenomenalist. Some commentators think that it is realist. Langton,
for example, argues that Kant’s point is not that ‘it is because we would have certain
experiences that a thing counts as existing, and being a part of our world: rather, it is
because the thing exists, and is already a part of our world, that we would have certain
experiences of it’ (Langton 1998: 143–5). There are certainly some passages in which
Kant describes his position in a way which is compatible with realism. A realist could
agree that everything actual is connected to a perception of which we are conscious,
in accordance with the analogies of experience (roughly, in accordance with the idea
of mutual causal interactions between persisting spatio-temporal objects made up of
persisting stuff). This does not require thinking that everything is essentially percep-
tible, but rather that everything that exists is part of one causal system: in this case,
everything actual will be causally connected to something we perceive. When Kant
describes actuality in the Postulates of Empirical Thought, the section concerned
with analysing actuality, possibility, and necessity, he says that ‘cognizing the actual-
ity of things requires perception, thus sensation of which one is conscious—not
immediate perception of the object itself the existence of which is to be cognized, but
still its connection with some actual perception in accordance with the analogies of
 MANIFEST REALITY
experience, which exhibit all real connection in an experience in general’ (A225/

B272). Here, although he introduces an essential link with possible perception, he
describes this as necessary for cognizing reality rather than for something’s being real.
This could be understood simply as expressing empiricism, not idealism. The realist
empiricist can read him as saying that we can cognize things only through a causal
chain that affects something we directly perceive. The same could be said about his
claim that ‘wherever perception and whatever is appended to it in accordance with
empirical laws reaches, there too reaches our cognition of the existence of things’
(B273). Thus, it is clear that there are indeed some texts which are compatible with
the realist empiricist view that we can have knowledge only of what affects our senses,
and that everything real is causally connected to something that affects our senses.
This is part of Kant’s view. However, he goes beyond this. Kant does not just say that,
if something is empirically real, it will be causally connected to something we
perceive. He says that what is empirically real is essentially something presentable
in a perceptual experience (in an empirical intuition) and he says that what is
presented to us are mere representations that exist in us (A493–4/B521–2). This is
neither realist nor phenomenalist; it is closer to a form of realism which denies the
existence of anything experience transcendent (a topic to which I return in
Chapter 9). Transcendental idealism says that our cognition is limited to what is
essentially manifest.
The final point for this chapter is to see the link between possible experience and
intuition. As I will argue in detail in Chapter 7, Kant’s notion of intuition is centrally
connected to what can be given to us in consciousness. The connection between this
and possible experience can be seen by looking at what Kant means by ‘possible’ and
what he means by ‘experience’. Kant gives an analysis of actuality, necessity, and
possibility in the Postulates of Empirical Thought, but it is his analysis of actuality,
rather than that of possibility, that is relevant to understanding what it means to say
that something is possibly perceivable. What is empirically real is what actually exists,
whereas the Postulate of Possibility gives conditions of thinking about things which
do not actually exist but could exist.11 However, the discussion of possibility in the
postulates is a starting point in understanding the idea of possible experience.
Kant distinguishes between logical possibility and what he calls real possibility,
and the Postulates are concerned with the latter.12 He thinks that something is
logically possible if the concept of it does not contain a contradiction. However, he
also thinks that there being no contradiction in the concept of an object is not enough
to tell us that it is the concept of (what he calls) a really possible object. He says that ‘I
can think whatever I like, as long as I do not contradict myself, i.e., as long as my
concept is a possible thought, even if I cannot give any assurance whether or not
there is a corresponding object somewhere within the sum total of all possibilities’
11 12
See Milmed (1969). See Adams (1997: 819–21).
THE SECONDARY QUALITY ANALOGY 
(Bxxvin), and ‘that the not-being of a thing does not contradict itself is a lame appeal
to a logical condition, which is certainly necessary for the concept, but far from
sufficient for real possibility’ (A244/B302). One of Kant’s examples of something that
is logically possible but not really possible is a two-sided figure. He says:
in the concept of a figure that is enclosed between two straight lines there is no contradiction,
for the concepts of two straight lines and their intersection contains no negation of a figure;
rather the impossibility rests not on the concept itself, but on its construction in space, i.e., on
the conditions of space and its determinations; but these in turn have their objective reality, i.e.,
they pertain to possible things, because they contain in themselves a priori the form of
experience in general. (A220–1/B267–8)
Other examples Kant gives of things whose logical possibility does not suffice to show
that they are really possible are
a substance that was persistently present in space yet without filling it (like that intermediate
thing between matter and thinking beings, which some would introduce), or a special funda-
mental power of our mind to intuit the future (not merely, say, to deduce it), or finally, a faculty
of our mind to stand in a community of thoughts with other men (no matter how distant they
may be)—these are concepts the possibility of which is entirely groundless, because it cannot
be grounded in experience and its known laws, and without this it is an arbitrary combination
of thoughts that, although it contains no contradiction, still can make no claim to objective
reality, thus to the possibility of the sort of object that one would here think. (A222–3–B270)
For Kant, we can know that something is really possible only if it is consistent with
the formal conditions of experience: the categories, space, and time (A218/B265). As
he understands real possibility, anything which could exist in space and time, made
up of matter and in causal relations with everything else that exists, is a really possible
object of experience. An object which can be known to be really possible is one which,
if it existed, could be presented to us in intuition and could be cognized by us.
A really possible object of course need not be an actual object, so this is not the
notion of possibility we need to understand what Kant means by possible experience
when he claims that something is empirically real only if it is part of the extent of
possible experience. When Kant links empirical reality to possible experience, he is
not talking about empirical objects which do not exist but could; rather, he is giving
an account of what it means for something actually to exist. As Bella Milmed
explains, in the sense in which the first Postulate defines possibility, ‘it is possible
to perceive a mermaid, for example (because a mermaid would be sense perceptible if
there were one), though not other people’s thoughts’ (Milmed 1969: 306). This
notion of possibility is not what explains the idea of the extent of possible experience
in the sense in which this idea features in Kant’s account of what actually exists.
Mermaids are really possible because, if they did exist, we would be able to perceive
them: they could be given to us in spatio-temporal intuition, as objects that persist
through time and are in causal relations with other spatio-temporal objects. How-
ever, mermaids are not part of the extent of possible experience that constitutes
 MANIFEST REALITY
empirical reality: mermaids are not empirically real, and we will not encounter one in
what Kant calls the regress of possible experience. In addition to logical possibility
and real possibility, we need the idea of what I will call actual possibility. As Milmed
says, in the sense of possibility that we need to understand the Kantian notion of
possible experience, it is not possible to perceive a mermaid, while it is possible to
perceive the house next door (Milmed 1969: 306). The house next door, and the far
side of the moon, actually exist and it is actually possible for us to perceive them, even
if we never will.
Kant thinks that something is part of empirical reality only if it is actually possible
for us to perceive it: what is empirically real must ‘stand in an empirical connection
with my real consciousness’, and if something is real it must be encountered in
perception in the continuation of experience (A493–4/B521–2). To explain this, we
need to know who ‘us’ refers to (any human being? a human being with especially
good senses? an ideal human being? any finite rational creature that has space and
time as the forms of its intuition?) and we need to know what counts as possible
perception. Kant allows us to say of things that are too small for us to perceive and of
things that are too far away for us to perceive that it is actually possible for us to
perceive them. Perhaps more unexpectedly, he also allows as part of the extent
of possible experience things which we lack the appropriate sense organs to perceive.
He says that
one can also cognize the existence of the thing prior to perception of it . . . if only it is connected
with some perceptions in accordance with the principles of their empirical connection (the
analogies). For in that case the existence of the thing is still connected with our perceptions
in a possible experience, and with the guidance of the analogies we can get from our actual
perceptions to the thing in a series of possible perceptions. Thus we cognize the existence
of a magnetic material penetrating all bodies from the perception of attracted iron filings,
although an immediate perception of this matter is impossible for us given the constitution
of our organs. For in accordance with the laws of sensibility and the context of our
perceptions we could also happen upon the immediate empirical intuition of it in an
experience if our senses, the crudeness of which does not affect the form of possible
experience in general, were finer. Thus wherever perception and whatever is appended to
it in accordance with empirical laws reaches, there too reaches our cognition of the existence
of things. (A225–6/B273)
Kant here says of objects or properties with respect to which we do not in fact have
the appropriate sense apparatus to perceive them that such properties can be part
of possible experience. He says that such things can be ‘connected with our percep-
tions in a possible experience’. He allows as part of the extent of (actually)
possible experience (and therefore as part of what is empirically real) things which are
so small or so far away that actual human beings, where we are in space, cannot perceive
them, as well as qualities which we do not have the right sense organs to perceive.
Phenomenalist interpreters will argue that they can explain the existence of things
which are too small or too far away for us to perceive in terms of constructions out of
THE SECONDARY QUALITY ANALOGY 
our possible perceptions. However, crucially, Kant himself does not explain them in
this way. With respect to things which are too small or too far away for us to see
them, Kant seems to think that they are in fact present in consciousness; they are just
present so weakly that we are not aware of their presence.13 He says that ‘perception
too weak in degree to become an experience for our consciousness still belongs to
possible experience’ (A522/B550). With respect to ‘magnetic material’, crucially,
Kant says that we could have an immediate empirical intuition of ‘magnetic material’
in an experience if our senses were finer. Intuitions are representations which present
us with objects; empirical intuitions do this as a result of objects affecting our senses.
Something of which we could have an empirical intuition is something which could
be present to the consciousness of a finite receptive creature like us, with spatio-
temporal intuition, by affecting its senses. The passage implies that for something to
be possibly perceivable (empirically real) it must be the case that it could be presented
in an empirical intuition to a possible sense, and that it is causally related to things
which are actually present to our consciousness. What determines the extent of
possible experience, then, is not simply what we, with our actual senses, can perceive,
but rather whether there is a way a thing could be presented in or manifest to
consciousness, through some possible extension of our senses. The key idea is not
what can be constructed on the basis of what is actually present in our mental states;
rather, something is empirically real if it is possible for it to be given to or present in a
consciousness like ours, through an at least possible sense or a possible extension of
our senses.
Interestingly, this is both stronger and weaker than a phenomenalist analysis.
A phenomenalist could account for the existence of things which are too small, too
far away, or for which we lack the appropriate senses in terms of there being
appropriate relations, or ways of constructing relations, between the contents of
mental states. Kant’s analysis does not consist of a construction out of what is actually
in minds, and in this respect its connection with what is in consciousness is less
demanding. On the other hand, for the phenomenalist it might suffice that the
appropriate relations exist among things that are actually present to consciousness.
The phenomenalist need not require of everything that counts as empirically real that
it is something that could actually be present in a consciousness experience; Kant
does require this. For Kant, in order for something to be empirically real it is not
enough that appropriate relations exist among other things that are present to
consciousness; something that is empirically real is something that could be manifest
in or present to consciousness.
13
This applies to distant stars, which we are not consciously aware of seeing, but which, he thinks, are
still present to us. He also allows the reality of things which we are not and never will be in the right
position to perceive, such as inhabitants of the moon (if there are any); the empirical reality of these
objects is more straightforward, since we can say of them that if we were present where they are we would
perceive them.
 MANIFEST REALITY
Suppose there is a shade of blue which no human eyes are actually able to detect,
but which, if our eyes were more sensitive, we would detect. According to the
essentially manifest view, colours exist only in the perceptual appearing of objects
to subjects like us. On this view, to say that the postulated missing shade of blue exists
is to say something about a quality of objects we would perceive, if our senses were
finer or more receptive. The claim is not about what merely inner states subjects
could be in; neither is it about a property things have independently of the possibility
of their appearing to sighted perceivers; it is about the way in which objects could
perceptually appear to finite receptive creatures. On the essential manifestness view,
it will be (partly) in virtue of a way things are in themselves (the way they are
independently of their perceptual appearing) that the missing shade of blue exists,
but the actual visual blueness exists only in relation to the possibility of its percep-
tually appearing. On this view, to say that there is a missing shade of blue is not to
say something about actual mental states (as a Berkeleyan claim about empirical
reality would), nor is it to say something about relations that could be constructed
between mental states. Rather, it is to make a claim about a feature of reality that
could be present to consciousness—a feature of reality that does not exist inde-
pendently of the possibility of its being present to consciousness. This is the way
Kant expresses the connection between something’s being empirically real and its
being part of possible experience. For something to be empirically real it is
necessary that it could be present in a possible empirical intuition; in addition,
Kant thinks that it is necessary that it is causally connected to something that is
actually present in an empirical intuition. On the one hand, for something to count
as part of possible experience it is not necessary that we actually perceive it—or
even that we could (with our present sense organs) perceive it. On the other hand,
anything which counts as part of possible experience must be causally linked to
something we do actually perceive, and, further, its counting as part of possible
experience requires that it be something that could be given in an empirical
intuition: something that could be present or manifest in a perceptual experience
by affecting a creature’s senses.
The key idea here is what can be given to us. On Kant’s account, the representa-
tions that give us objects are intuitions. Our next step, therefore, is to pay attention to
Kant’s notion of intuition: the subject of Chapter 7. Chapter 8 then looks at how Kant
takes his idealism to follow from the role of a priori intuition, and shows that the
idealism which the argument supports is best understood in terms of my essential
manifestness account.
7
Concepts and Intuitions
I Introduction
The subject of this chapter is Kant’s notion of intuition. Since he understands the two
ingredients in cognition, intuitions and concepts, by contrast with each other, the
chapter is also indirectly concerned with his account of concepts. Getting clear about
both is crucial to understanding his account of cognition, his transcendental ideal-
ism, and to the relation between cognition and the idealism. I have presented an
account of mind-dependence which does not involve existence in the mind, and
which fits the textual evidence concerning Kant’s distinction between things in
themselves and appearances. In my view, Kant’s idealism can be expressed in
terms of the claim that spatio-temporal objects are essentially manifest: their exist-
ence is not separable from the possibility of their being presented to us in a perceptual
experience. It is central to both manifest qualities and essentially manifest qualities
that they can be present in a conscious experience. Kant has a specific account of the
kinds of representations that present things to consciousness: intuitions. The notion of
intuition is central to Kant’s idealism for two reasons. First, as we saw in Chapter 6, the
way Kant understands possible experience centrally includes the idea of what can
possibly be presented to us in intuition: what can be given to us. Something counts as
experienceable only if it could be presented in an intuition. Second, Kant’s central
argument for transcendental idealism is based on the notion of a priori intuition: he
takes transcendental idealism to follow from the claim that space and time are a priori
forms of our intuition. This chapter and the next complete our account of essential
manifestness by exploring these two parts of Kant’s position. This chapter concerns
Kant’s notion of intuition in general; in the next chapter I argue that Kant’s central
argument for his transcendental idealism turns on a specific feature of intuition when
this is understood in combination with apriority. This will add to our understanding of
essential manifestness as a form of idealism, as well as to the arguments against both
phenomenalist and realist interpretations of Kantian empirical reality.
One of Kant’s most famous and fundamental claims in the Critique is that all
cognition requires two ingredients, intuitions and concepts (A51–2/B75–6; A320/
B377). Kant repeatedly insists that the two ingredients in cognition are distinct, that
each makes an essential contribution to cognition, and that neither can replace the
other or play the other’s role. Without both in play, we do not have cognition,
 MANIFEST REALITY
properly so called (A50–1/B74–5). In my view, much work on Kant’s transcendental

idealism and on his account of cognition has failed to take sufficiently seriously the
role of intuition in cognition: it is frequently assimilated to the role of sensation, so
Kant is seen simply as saying that cognition requires some kind of causal-sensory
input from objects. But this leaves out an interesting and central part of Kant’s view:
the idea that acquaintance with objects (not merely being affected by them) is an
essential ingredient in cognition and the idea that empirical reality is limited to things
with which we can, in principle, have acquaintance. This part of Kant’s view is central
to his epistemology and, as I will argue, to his transcendental idealism. However,
although it has many implications for his idealism, the fundamental contrast between
intuitions and concepts is epistemological (it concerns different kinds of mental
representations and the roles they play in cognition) and, as Kant presents it, could
be accepted by a realist. I examine the implications for his idealism in subsequent
chapters, but the concern in this chapter is with the epistemological contrast between
intuitions and concepts and the role each plays in our cognition.
Kant introduces his notions of intuitions and concepts partly by contrast with each
other. The crucial contrasts are that intuitions are singular and immediate represen-
tations, through which we are given objects, and concepts are general and mediate
representations, which enable us to think objects. Kant says that intuition is ‘imme-
diately related to an object and is singular’ whereas the way concepts relate to
particulars is ‘mediate, by means of a mark, which can be common to several things’
(A320/B377; see also A19 and A50/B74). He says: ‘Concepts differ from intuition by
virtue of the fact that all intuition is singular. He who sees his first tree does not know
what it is that he sees’ (LL: 905; see also A713/B741, LM 29: 800, 888, LM 29: 970–3,
and LL 905). For Kant, concepts are (in some sense) rules (A141/B180), essentially
general, and essentially constituents of judgments (A68/B93; A69/B94) and therefore
of inferential thinking.
Kant says repeatedly that the role intuitions play in cognition is that of ‘giving’ us
objects. He opens the Transcendental Aesthetic, the first section of the Critique, by
saying that intuition gives us objects, that only intuition gives us objects, that it is
only through intuition that cognition can relate to an object immediately, and
that thought requires given objects in order to be cognition. He says that ‘all
thought . . . must . . . ultimately be related to intuitions, thus, in our case, to sensibility,
since there is no other way in which objects can be given to us’ (A19/B33). Again,
after the Deduction and the Principles, Kant says: ‘For every concept there is
requisite . . . the possibility of giving it an object to which it is to be related. . . . Now
the object cannot be given to a concept otherwise than in intuition’ (A239/B298). He
repeats exactly the same idea near the end, saying that ‘all of our cognition is in the end
related to possible intuitions, for through these alone is an object given’ (A719/B747).1
1
See also A31–2/B47, A68/B93, A140–1/B180, A320/B377, A713/B741, Proleg. 4: 282, LM 29: 800, 888,
MV 29: 970–3, and VL 905.
CONCEPTS AND INTUITIONS 
My central concern in this chapter is with understanding the claim that intuitions
are singular and immediate representations that give us objects. I argue that the
singularity of intuitions should be understood as their presenting perceptual particu-
lars,2 that the immediacy of intuitions should be understood as their giving us
acquaintance with or presenting the particulars they represent, and that this is what
it means to say that intuitions give us objects. Intuitions give us acquaintance with
objects. I discuss this in detail later, in Section II, so for now I simply comment briefly
on each of these three points. First, singularity: to say that intuitions present us with
particulars is to say that a subject who has an outer intuition has awareness of some
thing outside of and other than themself, and that the subject is in a position to
perceptually discriminate this thing: to perceive it as distinct from other things and
from the background. Kant thinks that this involves (at least) representing its spatio-
temporal unity, being able to discriminate its spatio-temporal boundaries, and
representing it as spatio-temporally located and spatio-temporally related to other
things.3 Second, immediacy: on my view, intuitions present us with particulars
immediately because they involve the presence to consciousness of the things they
represent.4 They do not merely represent objects, but present them. Third, the claim
that intuitions give us objects: intuitions give us acquaintance with the objects of
cognition. They are not merely, like sensations, the causal outcome of an object
affecting us: they are presentations of the objects about which we think.5 Kant thinks
that it is never the case that merely having a concept puts you directly in touch with
the object the concept represents and he thinks that concepts do not pick out unique
particulars. He takes this to mean that merely having a concept (or a number of
concepts connected in thoughts) is always compatible with the possibility that there
exists no object to which the concept corresponds, and merely having a concept
cannot, on its own, connect you to a particular thing. This is why conceptual thought,
on its own, lacks what Kant calls objective validity; it lacks connection to a world. The
objective validity of cognition requires concepts and intuitions together, with intu-
itions putting us directly in touch with the objects that concepts enable us to think
2
At this point, ‘particular’ should understood minimally, as a thing which a subject singles out as a
perceptual unit—a distinct, bounded thing to which the subject can pay perceptual attention. This could be
a causally unitary object, but could also be less than, or more than, an object. A subject may intuit, for
example, a desk and the lamp attached to it, or may attend just to the light bulb in the lamp. A spot of light
moving on a wall could be a perceptual particular—it is something outside of and other than the subject,
that the subject can pick out as a unit.
3
This is compatible with there also being empirical, biological principles of association which deter-
mine what a creature perceives as particular.
4
See Buroker (2006: 37), Ewing (1934: 29), and Setiya (2004: 66). Parsons questions whether phenom-
enological immediacy requires actual presence of the object, saying that imagination can immediately
represent a non-present object (1992: 83). On my reading, imagination presents an image immediately, but
not an object immediately, whereas intuition immediately presents its object.
5
See A19/B33, A31–2/B47, A68/B93, A140–1/B180, A239/B298, A320/B377, A719/B747, A713/B741,
Proleg. 4: 282, LM 29: 800, 888, LM 29: 970–3, and LL 905.
 MANIFEST REALITY
about.6 In my view, Kant thinks that cognition requires (possible) acquaintance with
the objects of thought, that intuition is what provides this, and that only intuition can
provide this.
The most controversial part of my view is that I argue that intuitions do not
depend on concepts to play their role of presenting us with particulars. For a long
time, the dominant view in Kant scholarship has been that Kant holds that we could
not have any kind of awareness of an object (a distinct individual or particular thing)
without the application of concepts (at least of the a priori concepts he calls the
categories).7 The idea this view attributes to Kant is that without concepts we would
merely have a mass of sensory input, an unorganised sensory mush, rather than any
awareness of distinct, discrete individuals. This has strongly affected the way Kant’s
notion of intuition has been understood. In my view, it has led to commentators not
paying enough attention to Kant’s idea that intuitions give us objects and that
concepts do not, as this claim is quite hard to make sense of on the dominant
account. Kant has an interesting and powerful account of the contribution minds
must make to organising the sensory input to enable us to cognize a world. Focusing
on the contribution of concepts leaves out half of this account. The dominant view
sees concepts as carving out boundaries or organising the mass of sensory input to
enable us to have conscious representations of particulars. On this account concep-
tual organisation is needed for us to represent distinct particulars, and the independ-
ent contribution made by sensibility (intuition) is that of causal, sensory input.
Cognition, on this view, requires conceptual (rule-governed) organisation of some
causal, sensory input. This leaves out a central part of Kant’s account: the need for
acquaintance with the objects of cognition.
In my view, the dominant reading of intuition distorts our understanding of Kant’s
idealism in two ways. First, it is likely to lead to understanding transcendental
idealism as centrally a kind of conceptual idealism—a position which is about seeing
empirical reality as carved up by or dependent on our concepts. However, Kant takes
both the mind-dependence of appearances and the claim that we cannot cognize
things in themselves to be established in the Transcendental Aesthetic, and to follow
from considerations concerning intuition. Second, the conceptualist reading is not
able to take sufficiently seriously Kant’s claim that intuitions give us objects, and this
is likely to lead to paying insufficient attention to the centrality of givenness in Kant’s
6
This does not entail that we have to have an intuition of every object we cognize: in Kant’s account, we
can have cognition of objects with which we are not immediately acquainted if they are causally connected
to objects with which we are acquainted, and they are the kinds of things with which it is possible for us to
be acquainted. We cannot have knowledge by description alone of objects with which it is not possible for
us to be acquainted, such as Leibnizian monads.
7
Falkenstein, for example, says, ‘[f]or Kant, our senses are insufficient for the perception of particular
objects. Perception occurs only when the information acquired by the senses is recognised by us as an
instance of an object of a certain kind’ (2006: 141). See also Longuenesse (1993), McDowell (1994; 1998a),
Ginsborg (2006), and Griffith (2010).
CONCEPTS AND INTUITIONS 
idealism. Thus, defending my view of intuition is central to my understanding of

I will call the view that I am rejecting conceptualism. There are different versions
of conceptualism. One variant, which I will call strong conceptualism, involves
denying that intuitions present us with particulars and giving some other account
of what is meant by their being singular and immediate representations which give us
objects. A strong conceptualist reading of intuition sees intuition as giving us
something less than presentations of individuals.8 A second possibility, which I will
call moderate conceptualism, is to agree that intuitions present us with particulars,
and perhaps even to agree that this is what it means to say that they give us objects,
but to argue that they depend on concepts to play this role.9 On the moderate
conceptualist reading of intuition, the application of concepts is necessary for intuitions
to be intuitions: to be singular and immediate representations whose contribution to
cognition is to give us objects. I reject both these readings. I argue that intuitions do not
depend on concepts to play their role of presenting us with objects.
On the conceptualist reading of Kant, a creature which does not have concepts
could not be perceptually presented with a distinct thing outside it.10 It has, pre-
sumably, a blooming buzzing mass of sensations—a sensory mush that does not
enable it to perceive anything individual, unified, or distinct. In contrast, as I read
him, Kant’s account allows that a creature that has a spatio-temporal intuition but
lacks concepts (in his demanding sense of what concepts are) could perceive things
outside it—could be presented with distinct particulars. In what follows, when I talk
8
Lorne Falkenstein argues that we cannot make everything Kant says about intuition cohere, because
either intuitions are not singular representations or they are not distinct from concepts (1995: 63, 59). He
thinks that Kant is inconsistent, sometimes using ‘intuition’ to mean raw data and sometimes using it to
mean experience of individual objects, and that in the latter case it involves concepts. Wilfred Sellars,
similarly, sees Kant’s notion of intuition as confused between singular representations (which are in some
sense conceptual), and a sheer manifold of receptivity (which is not conceptual) (1968: 3–7). This
confusion arises only if, unlike Kant, we fail to distinguish between the roles of sensation and intuition.
9
See, for example Griffith (2010).
10
Explaining the point in terms of thinking about what a (finite, receptive) creature without concepts
could intuit is, I think, a helpful heuristic device for thinking about the logical independence of intuition
from concepts for the purposes of presenting us with perceptual particulars. While my argument here is
not attempting to give an account of Kant’s views on animal cognition, it is worth noting that there is clear
evidence both that Kant thinks that animals have consciousness and that he thinks that they perceive the
world. In a letter to Marcus Herz, Kant says that without the conditions necessary to represent an object
‘I would not even be able to know that I have sense data; consequently for me, as a knowing being, they
would be absolutely nothing. They could still (if I imagine myself to be an animal) carry on their play in an
orderly fashion, as representations connected according to empirical laws of association’ (C 11: 52). In
arguing that we could not derive necessity from experience, as this would be to ‘substitute subjective
necessity, that is, custom, for objective necessity’, Kant explains subjective necessity in terms of the idea
that ‘we may expect similar cases (just as animals do)’ (CPR: 13, my italics). While Kant controversially
thinks that we have only indirect duties to avoid giving animals pain, he clearly thinks they experience pain
(LM 6: 443) and his attribution of the (pathological) power of choice to animals implies that they have
consciousness (A534/B562; A802/B830). See Naragon (1990), Ameriks (2000: 242–5), and McLear (2011).
Ameriks argues that Kant’s claim that animals can physically distinguish (but not logically distinguish)
between two items involves attributing to them not only behaviour, but also conscious representations.
 MANIFEST REALITY
about conceptualism, both of the moderate and the strong variety, I am talking
specifically about conceptualist readings of Kant’s notion of intuition (as opposed to
conceptualism about perception more broadly).
The conceptualist reading has many roots, of which, in my view, the most
important lie in the Transcendental Deduction of the Categories—in particular, in
two features of Kant’s argument in the Deduction. One is the conclusion Kant argues
for: that the application of the categories is necessary for us to have ‘relation to an
object’ (A109) or for anything to ‘become an object for me’ (B138). He says that the
Deduction aims to show that without the categories ‘nothing is possible as object of
experience’ (A93/B126) and that all objects that can come before the senses fall under
the categories (B160). In isolation, these claims certainly seem to suggest the view
that there is no awareness of objects without concepts. However, our starting point in
approaching the Deduction should be to look for an account of what Kant means by
‘relation to an object’ that takes seriously and is informed by his claim that intuitions
are singular and immediate representations that give us objects, and that this is
something concepts could never do. All accounts, not just non-conceptualists, need
to give an account of the difference between being ‘given an object’ (as Kant uses this
expression in relation to intuition) and having ‘relation to an object’ (as Kant uses
this expression in the Deduction); I argue that non-conceptualists are better situated to
do this. As I will argue in this chapter (Section IV), and in Chapter 11, Kant’s concern
with ‘relation to an object’ in the Deduction is something very specific: he is concerned
with the possibility of concepts being used in a way that succeeds in referring to objects.
The second feature of the Deduction that may be taken to support conceptualism is
Kant’s account of the role of what he calls ‘synthesis’ in organising and unifying the
manifold of intuition; Kant argues that this synthesis is governed by the categories.
The way this is typically understood by the conceptualist view is that Kant thinks that the
conceptually governed syntheses that are necessary to cognize anything as an object are
what produce a unified particular object of perception out of an unorganised sensory
mass. Against this reading, I argue that the syntheses the Deduction is concerned with
are primarily something that is done to intuition, not something that produces intuition.
It is a manifold of intuition that is synthesised and this implies that we already have
intuition. Kant argues that synthesising intuition introduces into it a kind of unity
without which we could not cognize a world; he does not argue that synthesising
intuition is necessary for intuition to be intuition, and to play its role in giving us objects.
In what follows, I will mostly be taking about outer intuitions. Kant distinguishes
between inner sense and outer sense, and he sometimes speaks of the mind intuiting
itself (A22/B37), but when he talks about the role of intuition in giving us objects, he
is concerned with outer intuition giving us outer objects: ordinary macroscopic
spatio-temporal objects. I argue that outer intuitions are mental representations
that directly present us with spatio-temporal particulars.
The idea that, for Kant, intuitions present us with particulars has recently been
discussed in relation to the contemporary debate about the role of concepts in
CONCEPTS AND INTUITIONS 
perception.11 In these terms, the conceptualist reading could be presented as the

claim that Kant’s view is that without concepts we cannot perceive objects, whereas
my alternative could be put by saying that intuition is enough for perception of
objects. However, I do not put the point in this way here, because the question of
what Kant means by perception is not straightforward. He sometimes uses the term
‘Perzeption’ and sometimes ‘Wahrnehmung’, and it is not clear that his usage of
either is entirely consistent.12 ‘Wahrnehmung’ sometimes seems to refer to the
state of having a certain kind of awareness of what one is perceiving (e.g. A120):
in this sense, when I see a house but am not aware of the individual bricks as parts
of the house, I do not perceive the bricks, even though I am seeing them.13 If this is
what Kant means by perception, it may be something different from the kind of
perception animals have, and it is something more than simply being presented with
an individual thing outside you. Kant does not, in my view, intend to present a
systematic theory of perception in the contemporary sense: a theory designed to
answer questions such as the argument from hallucination against direct realism or to
give an account of how we think about animal perception.14 His main concern in the
Critique is with cognition, and we do not find in the text clear statements concerning
all the questions we might have with respect to a theory of perception. It is clear that
Kant does think that our perception—the perception of ordinary adult human
beings—involves both sensibility and concepts and that this makes an important
difference to the nature of our perceptual experience. This is, however, compatible
with thinking that being presented with perceptual particulars in itself does not
require concepts, so it is not enough to establish conceptualism about intuition. In
my view, this dispute cannot be resolved simply by appealing to texts in which Kant
mentions perception. A better way forward is to focus on understanding what kinds
of representations intuitions are, the role for which Kant introduces them—that of
giving us objects—and the extent to which they depend on concepts to play this role.
Before going on to explain my reading of intuition, I will mention one argument
for the conceptualist view which, though it does not seem to me very strong, is so
common that it requires some response. This argument turns on the famous passage
where Kant says that
intuition [Anschauung] and concepts [Begriffe] therefore constitute the elements of all our
cognition [Erkenntnis], so that neither concepts without intuition corresponding to them in
11
Hanna (2005), Ginsborg (2006; 2008), and Allais (2009).
12
See Tolley (forthcoming) for a detailed account of why Kant’s use of Wahrnehmung is technical, and
his thinking that concepts play a role in Wahrnehmung does not support seeing him as a conceptualist
about perception in the contemporary sense.
13
See Golob (2011: 7).
14
It is compatible with this that we can find, in Kant’s texts, ways of situating parts of his position in
relation to these debates (See McLear 2013, for a systematic account of Kant’s theory of perception, and
McLear 2011 for Kant on animal perception). My claim is simply that his concern is not to present a
systematic theory of perception.
 MANIFEST REALITY
some way, nor intuition without concepts can yield a cognition. . . . Without sensibility no
object would be given to us, and without understanding none would be thought. Thoughts
without content [Inhalt] are empty, intuitions without concepts are blind. . . . The understand-
ing is not capable of intuiting anything, and the senses are not capable of thinking anything.
Only from their unification can cognition arise. (A50–1/B74–5, my italics)
This passage is sometimes taken to rule out intuitions having the kind of represen-
tational content that is involved in presenting us with particulars. However, it is
crucial that in this passage Kant is talking about the interdependence of concepts and
intuitions for cognition (Erkenntnis).15 Kant’s analysis of cognition is not an analysis
of mere perceptual experience; nor is it an analysis of possible mental representations
in general. Since the passage states that we need both ingredients for cognition, it is
compatible with the claim that we can have conscious mental representations of
some sort with only one ingredient present. This is something Kant explicitly allows,
because he allows that we can have coherent conceptual thought without intuition
but claims that such thought does not amount to cognition. Transcendent meta-
physics (which makes claims about things which are not possible objects of
experience, thus which Kant thinks it is impossible for us to cognize) consists of
making judgments using concepts for which we cannot have corresponding
objects given in intuition (this is what is wrong with it), but it certainly has some
kind of content, since some of its ideas play an important role in Kant’s ethics.
Thus, his account of the need for both intuitions and concepts for cognition does
not rule out the possibility of conscious mental representational states without
concepts, nor does it imply that mere perceptual presentation of particulars
requires concepts.
What is often taken as decisive in this passage is Kant’s use of the term ‘blind’ in
the claim that intuitions without concepts are ‘blind’ (A51/B75). ‘Blind’ is metaphor-
ical; it needs to be interpreted, just as does ‘empty’ in the claim that concepts without
intuitions are empty. Without the full context (including Kant’s account of the role of
transcendent metaphysical concepts in moral thought), the claim that concepts
without intuitions are ‘empty’ might be thought to suggest that they have no content.
This would clearly be wrong: concepts without intuitions are not, for Kant, literally
empty or contentless.16 Similarly, on its own, ‘blind’ might suggest something which
15
Kant in fact sometimes uses ‘cognition’ in a way which allows that the ingredients of cognition can
count as cognition, and this is appealed to by non-conceptualist interpreters of Kant’s account of intuition
(see Hanna 2006: 98). However, the conceptualist could argue that this is a marginal, non-standard, or
inconsistent use of Kant’s notion of cognition. My argument allows that even if we do not count intuitions,
on their own, as cognitions, they can present us with particulars.
16
Taking the claim that concepts without intuitions are ‘empty’ literally is just as problematic as taking
the blindness claim literally. McDowell seems to take this line, since he explains Kant’s claim that thoughts
without concepts are empty by saying that ‘for a thought to be empty would be for there to be nothing that
one thinks when one thinks it; that is, for it to lack what I am calling “representational content”. That
would be for it not really to be a thought at all’ (McDowell 1994: 3–4). This cannot be Kant’s view. While
CONCEPTS AND INTUITIONS 
is in no sense representational, but in the context, and in the light of everything else
Kant says about the role of intuitions and concepts, it is preferable to read the passage
as claiming that without concepts intuitions are blind in the sense that they do not
present objects as classified or as subject to the normative requirements of inferential
judgments. Immediately after the famous togetherness passage, Kant goes on to say:
‘It is thus just as necessary to make the mind’s concepts sensible (i.e., to add an object
to them) as it is to make its intuitions understandable (i.e., to bring them under
concepts)’ (A51/B75). This fits with the idea that intuitions give us objects (they ‘add
an object’, not unorganised sensory mush, to the mind’s concepts) and that concepts
are necessary not in order for us to be given objects, but in order for us to understand
(think about, classify, or judge) the objects that are given to us.
In Section II, I defend in detail my view that seeing intuitions as giving us
acquaintance with perceptual particulars is the most straightforward reading of
Kant’s claim that they are singular and immediate representations that give us
objects, and argue that both strong and moderate conceptualists have a much harder
time making sense of these features of intuition, as well as of Kant’s claim that
intuitions and concepts are essentially different kinds of representations which play
essentially distinct roles in cognition. In Section III, I show that central arguments in
the Transcendental Aesthetic are inconsistent with conceptualism. Section IV exam-
ines the considerations from the Deduction that are taken to support conceptualism
and shows that they do not.
II Distinct Perceptual Particulars

In this section I present my understanding of intuition. To do this I need to explain
(1) the idea that intuitions present us with particulars (their singularity); (2) the idea
that intuitions involve the presence to consciousness of their objects (their immedi-
acy); and (3) the idea that intuitions do not depend on concepts to play their role of
presenting us with particulars (the distinct nature of each of the two ingredients of
cognition). The first two points are closely related. However, one could grant the first
two and reject the third: the moderate conceptualist could agree that the role of
intuitions is to present us with particulars but deny that they could do this without
being subsumed under concepts. And one could reject conceptualism (grant the third
claim) while denying the object-dependence of intuition (claim 2).17 As I see it, all
three points are related: it is because outer intuitions immediately present us with
outer particulars that they can play their distinct role in cognition: that of giving us
transcendent metaphysics, for Kant, is less than cognition, he clearly thinks that it is, to some degree,
understandable, and that it contains thoughts which play a crucial role in ethics. Empty thoughts are ones
for which there is no possibility of our securing referents and which therefore are not, in Kant’s terms,
objectively valid, and do not qualify as cognition proper.
17
Stephenson (2015).
 MANIFEST REALITY
objects. Against strong conceptualists I argue that we cannot make sense of intuitions
as singular and immediate representations that give us objects if intuitions are merely
a mass of unorganised sensory data (or individual items of sensory input). Against
moderate conceptualists I argue that intuitions are not dependent on the application
of concepts to play their role as singular, immediate representations that give us
objects. And against representationalists, who do not see intuitions as object-
dependent, I argue that if intuitions did not involve the presence to consciousness
of the objects they represent, they would not be singular and immediate and could
not play their role in giving us objects: uniquely picking out objects, guaranteeing the
existence of objects, and putting us in immediate contact with objects.
To understand Kant’s position, as I read it, we need to have some grip on what it is
to have a presentation of a perceptual particular without concepts. Being perceptually
presented with a particular involves a subject’s being aware of something outside of
and other than themself, as opposed to being aware of a sensation, such as an itch,
which is simply awareness of a state of the self. Further, it requires that the subject is
able to discriminate the thing perceptually presented from other things and from its
surroundings: the subject is aware of a distinct, bounded, discrete, and individual
thing.18 Typically, the subject will be in a position to attend to the thing, to track it,
and perhaps to do things to it (such as reaching out and grabbing it), and move
around it. Pace the conceptualist, the subject does not simply have a mass of
unstructured sensation. The idea that intuitions are singular means that there is a
particular thing the intuition presents; pace the representationalist, this would not be
guaranteed by images or mental intermediaries, which could represent more than
one (qualitatively identical) thing.
Being presented with a perceptual particular can be distinguished from a different
way of representing an object that involves using a concept to pick out or introduce
something as the subject of a judgment and then attributing some property to this
thing: having a thought about an object.19 The latter way of representing an object
18
Consider a dog that is perceptually presented with a desk lamp that is attached to a desk. Suppose its
behaviour indicates that it can discriminate the boundaries of the desk and the lamp, and thus that it can
see the desk and the lamp; this provides no reason to think that it perceives the lamp and the desk as
separate objects. Similarly, the dog may not be able to represent the light bulb as a part of the light; it may
not be aware of it as a part of the light. As noted earlier, one of the ways in which Kant uses the term
‘perception’ in the Deduction is to refer to being aware of what is being perceived (B160). In this sense, the
dog does not perceive the bulb as a part of the lamp, since it is not aware of the bulb as a part of the lamp,
although it sees the lamp and it sees the light bulb. However, from the fact that the dog does not perceive
the bulb as a part of the lamp it does not follow that the dog is not perceptually presented with the lamp and
its bulb. It perceives something outside it that it can discriminate from the background, discriminate from
other things, and move around. Importantly (and contra Land 2011), the non-conceptualist can assert that
it is the bulb and the lamp, the very things that can be brought under concepts and understood as objects,
that are presented to the dog in intuition; the objects given in intuition are the objects we think about
using concepts.
19
See Dickie (2011) for a clear statement of the difference between being perceptually presented with a
particular and having a referential thought about it, as well as a survey of empirical evidence for the idea
that the former does not require concepts. See Campbell (2002a; 2002b) for an account of the idea that
CONCEPTS AND INTUITIONS 
involves grasping that it is an object, or thinking of it as an object, and this clearly

involves concepts. Some philosophers think that being presented with perceptual
particulars is dependent on this capacity for thought; the conceptualist attributes this
view to Kant. For now, I simply note that the possibility of distinguishing between
these two different ways of representing an object means that it does not automat-
ically follow from the fact that Kant argues that the categories are necessary for
judgmental thought in which we think of an object that he thinks that the categories
are necessary in order for us to be presented with distinct perceptual particulars.20
The fact that this is a possible philosophical view does not of course mean that it is
Kant’s view; further argument is needed to show this. However, this point cuts both
ways: argument is also needed for the conceptualist view that Kant’s concern with
relation to an object in the Deduction is about the conditions of being presented with
perceptual particulars, since this does not automatically follow from the idea that the
categories are necessary to represent something as an object. All interpretations need
to give an account of the difference between what Kant means by being given an
object and having relation to an object, and to explain why intuitions play the former
role and concepts the latter. I argue that distinguishing between being in a position to
discriminate a perceptual particular (having acquaintance with an object) and being
in a position to think about it as an object gives us a way of doing justice to these
different features of the text in a way which fits Kant’s arguments in both the
Aesthetic and the Deduction.
Commentators have paid great attention to Kant’s account of the way conceptu-
alisation transforms perceptual experience; it is important to see that we can take this
seriously without thinking that it establishes conceptualism about intuition. The
thought that intuition can play its role of presenting us with perceptual particulars
independent of the application of concepts is compatible with thinking that concep-
tually governed syntheses affect how things are given in sensibility, that this could
make a difference to what we can perceive, and that Kant thinks it is necessary for us
to be able to cognize what is given in sensibility. The important point is that it does
not follow from any of this that these syntheses are conditions of objects’ being given
to us in intuition. It is plausible that there are discriminations which a non-concept-
having creature is not able to make, but this does not show that having concepts is
necessary to discriminate perceptual particulars at all. On the contrary, thinking
about the more extensive discriminations subjects are able to make when they are in a
position to represent objects as objects using concepts enables us to explain further
the difference between the two ways of representing objects. Without concepts, a
perception involves a presentation to consciousness of objects which is prior to conceptual thought. See
also Smith (2002: 84).
20
The idea is not that Kant has an ambiguous or equivocal account of what is meant by a particular (as
objected by Bauer 2012: 221), but simply that there is a difference between having an object of acquaintance
(a perceptual particular) and being in a position to think about this object using concepts.
 MANIFEST REALITY
subject cannot think of a particular with which it has acquaintance as an object. This
means, in Kant’s view, that it cannot think about an object as having properties, so it
cannot think about how an object’s various properties will interact with each other or
about the implications this has for how the object will interact with other things. It
cannot think about and investigate what kinds of changes the object can survive. This
entails that there will be limits to the extent to which it is able to track the thing over
time, and limits to the extent to which it is able to distinguish the thing from other
things. For example, it may not be able to represent a butterfly and a caterpillar as
stages of the same creature. It may not be able to represent the ivy growing up the tree
as a different thing from the tree. A subject may have acquaintance with something
outside of and other than itself (it may be able to discriminate the thing’s boundaries
and may be able to attend to and follow it) which is less than or other than an object,
such as a spot of light, something consisting of two objects, or a shadow. This is
compatible with its being aware of distinct perceptual particulars outside it.
In addition to saying that intuition is singular, Kant says that it is immediate. As
I read Kant’s immediacy claim, he thinks that intuitions are object-dependent in the
sense that we have an intuition of an object only when that object is in fact present to
us: a dream or a hallucination of an object does not count as an outer intuition. The
idea is that intuitions do not merely represent objects, or give us mere images of
things, but in fact present them.21 Kant says that ‘Imagination is the faculty for
representing an object even without its presence in intuition’ (B151),22 and he
21
Falkenstein takes immediacy as bound up with receptivity/passivity, and therefore as understood in
contrast with mental representations that involve processing. He therefore thinks that the immediacy of
intuition means that it should be understood as raw sensory input. This does not follow: it is indisputably
Kant’s view that our minds make a contribution to the representation of empirical intuitions through
organising them in the framework of a priori intuition, but Kant does not take this to undermine either the
immediacy or the receptivity of sensibility. The receptivity claim says that sensible representations are the
way they are because of the way objects are and because objects affect us. Consider a tomato affecting our
senses (putting aside possible distinctions between primary and secondary qualities). We perceptually
represent the tomato as round and red because it is round and red, and its roundness and redness affect our
senses. However, colour and shape are processed in different processing streams, and the visual system
needs to ‘reassemble’ or bind information from the different streams in order to represent a single round
red thing. Processing that is not conceptualising is involved in this, but this does not undermine the idea
that we represent the tomato as round and red because it is round and red and its roundness and redness
affect our senses. Processing might be necessary for us to be in immediate contact with objects; in contrast,
raw sensory data need not be immediate, in the sense that it not be given to consciousness, and it might
play a sub-personal role.
22
A passage in the Critique which might cast doubt on the idea that intuitions involve the actual
presence to consciousness of the objects they represent is found in the Refutation of Idealism, where Kant
says ‘[f]rom the fact that the existence of outer objects is required for the possibility of a determinate
consciousness of our self it does not follow that every intuitive representation of outer things includes at the
same time their existence, for that may well be the mere effect of the imagination (in dreams as well as in
delusions)’ (B278). I take it as significant that this passage does not actually talk about intuitions, but
intuitive representation (anschauliche Vorstellung). Further, as McLear (forthcoming) argues, Kant can
allow that we have intuitions without objects’ being present to us, so long as these are not outer intuitions.
Arguing against my view, Andrew Stephenson (2015) documents a number of passages from the Anthro-
pology which might be taken to show that intuition does not involve presence, and points out two passages,
CONCEPTS AND INTUITIONS 
explains the contrast between the mediacy of concepts and the immediacy of
intuitions in terms of whether or not we are representing an object by means of a
further representation. With respect to concepts, he says ‘since no representation
pertains to the object immediately except intuition alone, a concept is thus never
immediately related to an object, but is always related to some other representation of
it (whether that be an intuition or itself already a concept). Judgment is therefore the
mediate cognition of an object, hence the representation of a representation of it’
(A68/B93; A19/B33; A239/B298).23 The object-dependence of intuition makes sense
of Kant’s contrast here between saying that a concept is always related to an object
through a representation of it while an intuition relates to an object immediately.
My reading of both the singularity and the immediacy of intuition is supported
by the fact that these claims should be understood in contrast with Kant’s account
of the generality and mediacy of concepts. The generality of concepts (for Kant)
means that concepts always apply, in principle, to more than one object; this entails
that concepts do not uniquely pick out their objects—descriptive criteria do not
uniquely individuate.24 This contrast loses its point if intuitions are not presenting
us with individuals.25 Closely related, the mediacy of concepts means that it is never
the case that merely in virtue of having a concept you are thereby in touch with its
object (A239/B298). Kant says of concepts on their own (without being related to
something given in intuition) that ‘I would not even know whether there could be
anything that corresponded to this determination of thought if empirical intuition
did not give me the case for its application’ (B149; see also B146 and A239/B298).
When we do not have intuitions, we do not know if there are any objects (or even if
there could be any objects) to which our concepts apply. On my account, the way
intuitions guarantee that there are objects corresponding to our concepts is by
presenting us with them. If intuitions did not present us with objects, but were
merely mental intermediaries which represent objects, it is hard to see how their
from the Prolegomena (4: 281–2) and the Critique (B151), that support my reading less clearly than I have
previously taken them to (2015: 8–20). However, my reading is not simply based on Kant’s expression in a
few passages, but rather on making philosophical sense of Kant’s saying that intuitions are singular and
immediate, his contrast between this and the generality and mediacy of concepts, and the way in which this
enables intuitions to play their role in cognition. Stephenson suggests that the immediacy and singularity of
intuition should be taken to mean that intuitions present their objects in a peculiarly phenomenological
way—the particular objects we perceptually experience at least seem to be really there’ (2015: 22). This does
not do justice to the crucial idea that an intuition guarantees the existence of its object, and that it presents a
unique object. In relation to the Anthropology passages, it seems significant to me that many of these
passages speak of ‘the faculty of intuition’, rather than ‘intuitions’. However, it may be that Kant did not
always use the word ‘intuition’, in the same sense, or strictly precisely.
23
As Houston Smit puts it, ‘the immediacy of an intuition can be characterized—albeit negatively—as
its not relating to an object by means of some other representation of that object’ (2000: 263).
24
Kant rejects Leibniz’s idea of complete individual concepts, and thinks that no concept picks out a
single particular thing (he says that we can make a singular use of concepts, but this is always a
derivative use).
25
Note how much Falkenstein’s view undermines the point of the notion of intuition: he says that
intuitions are not really singular and do not really present objects, Falkenstein (1995: 67–8).
 MANIFEST REALITY
role could be contrasted to that of concepts in this respect—how they could play
their role in guaranteeing the existence of objects corresponding to our concepts.
Further, mental intermediaries, such as mere images, are not singular—they do not
present their particular objects. Merely mental intermediaries do not guarantee the
existence of their objects nor do they uniquely individuate their objects; Kant thinks
that intuition does both. The thesis that intuition involves the presence of the
objects it represents makes sense of this.
Kant takes the immediacy and the singularity of intuition to go together.26 He says
‘intuition is namely an immediate representation of an object. This latter can thus be
only singular’ (MV 29: 970). The idea that intuitions present us with particulars
makes sense of this: a representation which immediately presents us with a particular
thing presents that thing and no other. This is not true of a mere image, which could
represent more than one thing. An image is not singular in the way in which a
presentation of a particular is, yet Kant here says that singularity follows from
immediacy. This supports thinking that what he means by immediacy is a mental
representation’s directly presenting us with a particular. Further, note that what
intuition is said to give us an immediate representation of is an object, not a mere
manifold of sensation.
Understanding the singularity and immediacy of intuition as saying that intuitions
present us with particulars makes sense of the role of intuition, which is to give us the
objects about which we think. This can be understood in terms of the notion of
acquaintance: Kant thinks that intuitions give us acquaintance with objects. As
Russell explains the notion of acquaintance:
I say that I am acquainted with an object when I have a direct cognitive relation to that object,
i.e., when I am directly aware of the object itself.27 When I speak of a cognitive relation here,
I do not mean the sort of relation which constitutes judgment, but the sort which constitutes
presentation. In fact, I think the relation of subject and object which I call acquaintance is
simply the converse of the relation of object and subject which constitutes presentation. That
is, to say that S has acquaintance with O is essentially the same thing as to say that O is
presented to S. (Russell 1910–11: 108)
26
Jaakko Hintikka suggests that immediacy and singularity are not distinct criteria (1969: 42).
Hintikka also suggests that Kant’s notion of intuition is close to what we would call a singular term,
as does Charles Parsons, although Parsons argues against Hintikka’s view that the singularity and
immediacy criteria are distinct, and understands immediacy in terms of presence to the mind (1992). As
Manley Thompson, Robert Howell, and Kirk Wilson point out, seeing intuitions as like singular terms
makes Kant’s notion of intuition inappropriately conceptual; intuitions are not linguistic expressions
(Thompson 1973, Howell 1973, Wilson 1975). Smit argues that Hintikka, Thompson, and Howell are
wrong to understand immediacy as a matter of not relating through marks, because, he argues,
Kant allows both intuitions and concepts to have marks. On his view, the marks of intuitions are
relevantly like tropes, in that the property is represented as a single instance, rather than as general
(2000: 238, 255).
27
Translated into Kantian terms, the ‘object itself ’ here would be the empirically real object, which does
not present the object to us as it is in itself.
CONCEPTS AND INTUITIONS 
There are of course enormous differences between Russell’s and Kant’s positions: for
one thing, for Russell, the only objects of acquaintance are sense data, whereas for
Kant, we are directly acquainted with spatio-temporal objects. For Russell, we have
transparent epistemic access to all features of that with which we have acquaintance,
which is not Kant’s view. However, the common idea is that of things which are
present to consciousness rather than being represented by some intermediary. On my
reading, intuitions present objects to consciousness. If an object is present to con-
sciousness, it is not merely something that causally affects a subject’s mental states.
Rather, the object is in the subject’s consciousness; it is directly and immediately
available in consciousness for the subject to attend to. It is making objects available to
us in this way that is, as I see it, the role of intuition in Kant’s account of cognition:
this is precisely what concepts cannot do.
It is important to see that something which is present to or before a subject’s
consciousness is not therefore merely in the subject’s consciousness: it is not a merely
mental representation (as traditional representationalist or indirect views of percep-
tion might assume). As I have argued in Chapters 1, 2, 5, and 6, we should not
approach the Critique with the assumption of an account of perception which
involves mental intermediaries. One of Kant’s aims in the Critique is, arguably, to
reject such an account,28 and, in my view, central parts of his position make sense
only in the context of this. This can be seen most clearly in Kant’s argument against
Cartesian scepticism, the Refutation of Idealism, where he argues, against Descartes
and Berkeley, that external objects in space are the immediate objects of perception
(B274–9; Bxxxix–Bxli). It can also, I argue, be seen in his notion of intuition.
The idea of a mental representation which involves its object’s being present to
consciousness is not simply that of a mental representation whose content essentially
depends on its cause.29 This is because there are possible versions of the externalist
thesis—the idea that what a mental state represents depends on what causes it—that
are compatible with intermediary representations, so long as these intermediaries
have the right causal links to the world. In contrast, acquaintance is a direct relation,
which does not involve intermediaries. Causal processes might be necessary to enable
us to be directly presented with objects,30 but the idea of causal input is not enough to
capture the idea of acquaintance.
A strong conceptualist could say that intuitions involve a causal input (sensation),
that this is what is meant by ‘giving’ us objects, and that this is how intuitions
28
See Willaschek (1997), Collins (1999), Abela (2002), and Bird (2006).
29
See Willaschek (1997: 545–6), who argues that intuitions are essentially caused by the objects they
represent. In my view, this fits with less of the text than the idea that intuitions involve the presence to
consciousness of their objects, since the latter (unlike the mere causal connection) explains both singularity
and immediacy, and can be common to a priori and empirical intuition. Kant thinks that empirical
intuitions involve receptivity (objects affecting us), but, unlike the immediacy and singularity criteria, this
is not part of the definition of intuition, given that Kant has a notion of intellectual intuition.
30
As I argue in Ch. 5, the idea that perception involves causal processing does not rule out a direct or
relational account of perception, and it is not taken by Kant to do so.
 MANIFEST REALITY
guarantee the existence of objects. The idea would be that we are passively ‘given’
some input from objects when they affect our senses and that cognition requires
intuitions because it requires this input. In addition to the fact that this would not
explain the singularity of intuition,31 a central problem with it is that it assimilates the
role of intuitions to that of sensations. It is not implausible that some philosopher
might talk about being given objects in order to refer to the idea of being given some
sensory input from objects through sensation. And it is clear that Kant thinks we
need input from the objects of cognition. He thinks we need sensation. However, if
being given objects were simply having some causal input from objects, then the
contribution intuitions make to cognition would be the same as the contribution
sensations make. But there is simply no doubt that, on Kant’s account, intuitions are
not sensations. On Kant’s account, empirical intuitions are the outcome of sensation
being ordered in the a priori forms of intuition, and it is intuitions, not sensations,
that give us objects. Kant says that ‘[t]he effect of an object on the capacity for
representation, insofar as we are affected by it, is sensation’ (A20/B34) and that
sensation provides the matter of empirical intuitions, a matter which must be
arranged in a form before we have empirical intuitions, implying that sensation
cannot be the same as intuitions. He says that ‘an object can appear to us only by
means of such pure forms of sensibility, i.e., be an object of empirical intuition’ (A89/
B121), and that ‘with every manner in which we are affected there are two parts:
matter, i.e. the impression of sensation, and form, i.e., [the] manner in which
impressions are unified in my mind. Otherwise I would have millions of impressions
but not intuition of a whole object’ (LM 29: 800, my italics). Notice that it is the form
which gets us from impressions (sensation) to ‘intuition of a whole object’, and the
form of intuition is of course a priori intuition. In the Inaugural Dissertation, Kant
says that ‘all our intuition is based on a certain principle of form, and it is only under
this form that anything can be apprehended by the mind immediately or as singular,
and not merely conceived discursively by means of general concepts. But this formal
principle of our intuition (space and time) is the condition under which something
can be the object of our senses’ (2: 396). Notice, here again he sees immediacy and
singularity as bound up with each other.
Kant does of course think we need a causal, sensory input (sensation), but he
thinks that sensations and intuitions are fundamentally different, and play different
roles in cognition. Sensation is a mere causal result, a function of objects affecting us,
and it does not present us with objects or with a world. A mere causal, sensory input
from objects need not be thought of as presenting us with anything, and while it
31
The strong conceptualist could perhaps argue that the sense in which intuitions are singular is that
they are individual points of sensory data. However, Kant does not see individual points of sensory data as
representations of which we are usually conscious (which are given to us) and they certainly do not give us
objects. Further, this reading would not explain the singularity of intuition in contrast with the generality of
concepts, which must surely be central to an account of Kant.
CONCEPTS AND INTUITIONS 
indicates the existence of something other than itself, it does not guarantee the
existence of the particular objects of cognition. In the first Critique, Kant’s dominant
use of ‘sensations’ indicates that they are nonintentional or nonreferential: they do
not, themselves, present objects to the mind, but ‘refer to the subject as a modification
of its state’ (A320/B376). In contrast, he says that intuitions are immediate, singular
representations that give us objects.32 Kant does not say that intuitions are the
representations through which objects have an impact on how we represent them,
and that together with conceptualisation this sensory input gives us objects, but that
intuitions give us objects.
As I discuss in Section III, Kant thinks that our representations of space and time
are a priori intuitions. The strong conceptualist idea that intuitions give us objects in
the sense of giving us causal input from objects makes it hard to see how we can give a
unified account of how empirical and a priori intuitions are both intuitions, and both
give us their objects. The idea that intuitions are immediate presentations of par-
ticulars to consciousness does, in contrast, enable us to do this: both empirical and a
priori intuitions present us with their objects, the things they represent. Kant says
that intuitions exhibit objects and that they display objects (A156/B195), and he
makes these claims in talking about providing a priori objects for mathematical
concepts. He says that we need to ‘display the object that corresponds to . . . [the
concept] in intuition, since without this the concept would remain as one without
sense, i.e., without significance’ (A240/B299, my italics).
Things might look more promising for the moderate conceptualist who agrees that
intuitions present us with particulars, but argues that they depend on the application
of concepts to play this role. A problem with this reading is that it is in serious tension
with Kant’s insistence on the essential heterogeneity of intuitions and concepts, and
the fact that they cannot play each other’s roles. Kant says that intuitions and
concepts depend on each other for cognition, and that only from the unification of
concepts and intuitions can cognition arise, ‘[b]ut on this account one must not mix
up their roles, rather, one has great cause to separate them carefully from each other
and distinguish them’ (A52/B76). Kant says that concepts cannot give us objects and
that they depend on intuitions for this (A19/B33; A51/B75; A239/B298; Proleg. 4:
282); he does not say that intuitions depend on the contribution of concepts to be
intuitions and to play their role in presenting us with objects. If sensations need to be
organised by concepts to generate intuitions, which then give us objects, then
32
See George (1981) and Westphal (2004: 44). As George explains, sensations are nonintentional in the
sense that, although they may be objects (of thought or attention), they don’t have objects: ‘Condillac, Reid,
and also Kant held that if one is aware of a sensation, one is aware of something, but that one is not, in such
a case, aware of anything other than one’s act or state’ (1981: 235). I agree that this is Kant’s dominant
account of sensation in the first Critique, but, unlike George and Westphal, I think this is entirely consistent
with thinking that intuitions present us with particulars, since intuitions are not sensations. George reads
Kant as thinking that all reference to objects requires judgment; similarly, Westphal seems to think
that the synthesis that brings about the referential and representational role of sensations must be a
function of judgments.
 MANIFEST REALITY
intuitions and concepts are not fundamental and distinct ingredients of cognition
that play distinct roles. Rather, sensations and concepts are the fundamental ingre-
dients, which collectively enable us to have intuitions that give us objects. The textual
cost of this reading is significant. In contrast, the idea that intuitions present us with
particulars makes straightforward sense of the idea that intuitions are singular and
immediate representations whose distinct contribution to cognition is to give us
objects. That intuitions are singular and concepts are general, that intuitions give us
objects and concepts cannot do this, and that these two ingredients make essential
and distinct contributions to cognition are not features of Kant’s account that are
provisional or open to revision. These claims are fundamental to Kant’s position, and
asserted many times throughout the critical works.
A further severe textual objection to the moderate conceptualist reading is that
Kant explicitly denies it, in the transition to the Transcendental Deduction, the very
section that is thought to give it most support. He says:
since an object can appear to us only by means of such pure forms of sensibility, i.e., can be an
object of empirical intuition, space and time are thus pure intuitions that contain a priori the
conditions of the possibility of objects as appearances, and the synthesis in them has objective
validity. The categories of the understanding, on the contrary, do not represent the conditions
under which objects are given in intuition at all, hence objects can indeed appear to us without
necessarily having to be related to functions of the understanding, and therefore without the
understanding containing their a priori conditions. . . . For appearances could after all be so
constituted that the understanding would not find them in accord with the conditions of its
unity, and everything would then lie in such confusion that, e.g., in the succession of
appearances nothing would offer itself that would furnish a rule of synthesis and thus
correspond to the concept of cause and effect, so that this concept would therefore be entirely
empty, nugatory, and without significance. Appearances would nonetheless offer objects to
our intuition, for intuition by no means requires the functions of thinking. (A89–90/B122–3,
my italics)
In contrast to both strong and moderate conceptualism, Kant explicitly says that
intuition does not require the functions of thinking and that objects can be given to
us independently of thought. I have quoted the passage at length, because a common
response to my position is that in this passage Kant is considering a possibility he is
going to go on to rule out, and which does not represent his considered view.33
Indeed, he is going to rule out a possibility that he mentions here: the possibility that
appearances are ‘so constituted that the understanding does not find them in accord
with the conditions of its unity’. This, he thinks, is something which seems prima
facie possible, and which we need to show is not the case. He wants to show that the
objects which appear to us are necessarily related to the functions of the understand-
ing and that the understanding does provide their a priori conditions. But he
33
For an example of this response, see Griffith (2010: 7). This objection is often made to me in
discussions.
CONCEPTS AND INTUITIONS 
straightforwardly asserts that the categories are not conditions under which objects
are given in intuition at all, and he says that even in the case of the possibility which is
he going to rule out (appearances not being in accord with the conditions of the
understanding), appearances would still offer objects to our intuition.34 He does not
express this as provisional, or as something he is going to go on to deny; he simply
straightforwardly asserts that objects can appear to us without being related to the
understanding, and immediately following this he states, as the explanation of this
possibility, that ‘intuition by no means requires the functions of thinking’ (A91/
B123). This makes it absolutely clear that what we should be looking for is an
argument which shows that the objects which are given to us in intuition are ‘related
to the functions of the understanding’ but which does not do this by attempting to
show that being so related is a condition of being given to us in intuition. Showing
that objects of intuition must accord with the conditions the understanding requires
for the synthetic unity of thinking is simply not the same as showing that objects
must so accord in order to be presented in intuition. In the A89–90/B122–3 passage
Kant explicitly denies the latter. A strategy which tries to vindicate the categories by
showing them to be conditions of our being given objects in intuition should be an
interpretation of last resort, since Kant opens the Deduction by denying this.
As I read him, Kant thinks that cognition requires the possibility of acquaintance
with its objects, and he thinks that intuitions give us this. Intuitions present us with
the particulars about which we think: they are singular and immediate. It is a relation
to an object that is logically independent of conceptual thought about an object,
which puts us in touch with the object, and this is necessary for us to be in a position
to have thought about the object.
III The Transcendental Aesthetic: A Priori Intuition

In the Transcendental Aesthetic, Kant introduces the idea of a priori intuition, and he
argues that our primary representations of space and time are a priori intuitions. It is
extremely difficult to make sense of Kant’s conclusions, his premisses, and his
argumentative strategy on the conceptualist reading of intuition. He thinks that
having singular presentations of particulars (empirical intuitions) requires that the
sensory input is organised in a spatio-temporal form, and that these forms (our
representations of space and time) are themselves singular and immediate represen-
tations.35 His arguments for space and time being intuitions include: (1) the claim
34
As Paton points out, in the A89/B122 passage Kant says that objects can (können) appear to us
without necessarily having to be related to the functions of the understanding, whereas he considers (and
rejects) the possibility that appearances could (könnten) be so constituted that the understanding would
not find them in accord with the conditions of its unity (1936: 324n).
35
And, in the Prolegomena argument from incongruent counterparts, he argues that this form has
content that cannot be captured conceptually. The argument from incongruent counterparts explicitly
depends on the idea that there can be a difference in the way two things are presented to us even where
 MANIFEST REALITY
that they are presented to us as immediately given individuals; (2) the claim that they
are presented to us with a different kind of unity than the unity we can achieve
through conceptual representation; and (3) the claim that these a priori forms are the
conditions of our representing distinct particulars.
Kant argues that because our primary representations of space and time are
singular and immediate presentations of space and time, these representations are
intuitions and not concepts. The moderate conceptualist account allows that intu-
itions are presentations of particulars or individuals but insists that they depend on
conceptualisation to be this. But Kant argues that it is precisely because space and
time present us with individuals that our primary representations of them are not
conceptual. Kant’s claim that space and time are presented to us as individuals is of
course difficult to understand. They are clearly not particulars in the sense in which
physical objects are. A rough take on the thought with respect to space is to say that
we have an immediate presentation of one structure or arena in which all outer
empirical intuitions (objects of perception) are located and related to each other, and
this structure is immediately presented to us, not constructed on the basis of
something immediately given. (Though this is already a bit misleading, in that it
might be taken to imply that the structure is something like a whole object, whereas
Kant’s view is that it is a form, a structural way in which outer particulars are related
to each other.) The idea that space is immediately presented to us as one and as
infinite should not be understood to mean that the whole of space is given to us at
once (which would be impossible) but that space is immediately presented to us as a
whole that is prior to its parts, rather than something which is constructed from its
parts, and as something the form of which is that it goes on. It is immediately
presented to us as that in which all spatial objects are related to each other. While
the idea is difficult, what is entirely clear is that Kant’s arguments for the claim that
our representations of space and time are intuitions (arguments 3 and 4 in the
B edition) turn on claiming that we represent space and time as immediately given
and as one or as singular (A24–25/B39–40). He argues that ‘we can represent only a
single space’ (A25/B39) and that ‘[d]ifferent times are only part of one and the same
time. That representation, however, which can only be given through a single object,
is an intuition’ (A31–2/B47). An argument for space and time being intuitions and
not concepts could not turn on the claim that space and time present us with
individuals if our representations of space and time needed to be organised concep-
tually to present us with individuals.
there is no conceptual difference. He says: ‘[w]e can therefore make the difference between similar and
equal but nonetheless incongruent things (e.g., oppositely spiralled snails) intelligible through no concepts
alone, but only through the relation to the right-hand and left-hand, which refers immediately to intuition’
(Proleg. 4: 286). He says of two incongruent counterparts that ‘nothing will be found in either, when it is
fully described by itself, that is not also in the description of the other’ (Proleg. 4: 286, my italics); ‘there are
no inner differences here that any understanding could merely think; and yet the differences are inner as
far as the senses teach’ (Proleg. 4: 286).
CONCEPTS AND INTUITIONS 
A second significant feature of the argument in the Aesthetic is that Kant appeals
to the claim that space and time are presented with a different kind of unity from the
unity that is achieved through conceptual synthesis. One argument for conceptual-
ism is based on the claim that, for Kant, the way we represent both unity and
singularity involves conceptualising; it is argued that this means that there is no
way of accounting for how unity could be introduced into the sensory manifold
without bringing in concepts, and, relatedly, that there is no way to determine what a
creature will single out and represent as one—how a creature could represent
singularity—without concepts.36 It is true that Kant thinks there are conceptually
governed ways of representing unity and singularity; the non-conceptualist need
not deny this. The non-conceptualist position requires that there are also ways
of representing unity and singularity that are not conceptual; not only can we
find precisely this view in Kant, it features centrally in one of the arguments in
the Aesthetic. Kant distinguishes between aesthetic (intuitive) and conceptual
unity, between aesthetic and conceptual ways in which representations can be
distinct,37 and between aesthetic and conceptual ways in which representations
can be ordered, and he clearly argues that intuitional unity gives us singular
representations: individual things. Kant thinks that conceptual synthesis is
needed to represent a singular thing as a complex of parts, a unified complex
object: to represent the manifold in it as manifold. This does not show that a
subject could not be presented with a unified perceptual particular without con-
cepts. On the contrary, Kant thinks that without concepts a subject can only
represent the things it is presented with in intuition as singular (as one whole)
and cannot represent them as complexes of parts. Aesthetic unities, Kant thinks,
are presented as one primitively, all at once (they are presented as singular),
whereas conceptually grasped unities require apprehending a combination of
the parts of a thing.
This point about intuitional unity is crucial to Kant’s third argument in the
Aesthetic. Kant’s third argument for the claim that our representation of space is
an a priori intuition turns on the idea that there is a way of representing unity or
singularity that is different from conceptual unity and that we are presented with
space and time as unified in this way.38 Kant says that the way we represent the
oneness of space is as a single given whole, rather than through first representing
parts of space and putting them together to represent the whole of space,
and he says that representing space in this way is prior to being able to represent
36
These claims are frequently made to me in discussion.
37
Intuitions are distinct when we can spatially discriminate their parts, whereas concepts are distinct
when we know what sub-concepts fall under them. Kant says that my representation of the milky way is
indistinct when I represent it as a whitish streak, but when I can (with the help of a telescope) distinguish
the individual stars it becomes distinct (JL, 35).
38
My argument here is in agreement with, and draws on, the excellent discussion in McLear
(forthcoming).
 MANIFEST REALITY
its parts (A25/B39).39 He thinks that this is the opposite of the way in which we
grasp a whole/unity/oneness conceptually, which requires running through parts
and putting them together. In the fourth argument, Kant says that the way the
infinity of space is given to us would not be possible if our primary representa-
tion of space were conceptual. On Kant’s view, the conceptual representation
of unity requires a synthesis of the parts of the unified thing; representing an
infinite totality in this way would require the completion of a synthesis of an
infinite number of parts of space. This is not possible for a finite mind. Yet we do,
according to Kant, represent space as an infinite given whole (A25/B39).40
So Kant’s arguments turn on the claim that representing unity, wholeness, or
oneness aesthetically (in intuition) is entirely different from the way we represent
unity conceptually, because the former involves being presented with a thing all at
once, as a whole (singular, individual), as opposed to representing it as a unified
complex of parts or properties, which, Kant thinks, requires conceptually governed
synthesis (A438/B466). The former is precisely the unity we need in order to be
presented with perceptual particulars, and it undermines the idea that the unity
introduced by conceptualising is a requirement of being presented with unified,
distinct perceptual particulars.
A further feature of the argument in the Aesthetic that is difficult for conceptualists
to make sense of is that one of Kant’s arguments for the thesis that our representa-
tions of space and time are a priori turns on the claim that space and time are the
conditions of our representing distinct things outside us. In Kant’s first argument for
the claim that space is a priori he says:
Space is not an empirical concept that has been drawn from outer experiences. For in order for
certain sensations to be related to something outside me (i.e., to something in another place
from that in which I find myself), thus in order for me to represent them as outside and next to
one another, thus not merely different but as in different places, the representation of space
must already be their ground. (B38)
There is much debate about how to understand this argument—in particular, about
whether Kant is arguing for the claim that space is a condition of the possibility of
39
As McLear puts it, we can perceive spaces (parts of space, a space) only by perceiving them as limits of
a whole which is presented to us: ‘for any representation of a bounded region, that experience is made
possible in virtue of a representation of a later region which encompasses the bounded region, and so on’
McLear (forthcoming: 20).
40
As McLear argues, the idea is not that we perceive the whole of an infinitely large object, but rather
that our immediate representations of any part of space presents it as a part of a larger whole, of which it is
a limitation. Space is immediately given to us as having the form of always going on, not having
boundaries. See Golob (2011: 15) and McLear (forthcoming: 15). See also Moore, who explains Kant’s
point by saying that ‘any experience, and any accumulation of experiences, has written into its very form
the possibility of further experience, of how things are elsewhere and elsewhen’ and ‘what I see presupposes
the possibility of seeing further’ (1992: 481).
CONCEPTS AND INTUITIONS 
representing distinct particulars, or arguing from it.41 My view is the latter. What is
clear, however, is that the argument centrally involves the idea that the way we get
from sensations to representations of something distinct from and outside ourselves
is by ordering sensations according to the a priori form of intuition, our a priori
representation of space (A20/B34). Kant says that representing a distinct outer
particular involves representing something as outside of and other than me and as
distinct from other things. He then says that because the representation of space plays
the role of enabling us to represent things as outside me (as in different places, as
spatially related and as distinct from each other) the primary representation of space
is a priori. The conceptualist thinks that there could be no distinct particulars in
intuition alone, but what Kant argues is that it is because representing objects
spatially is the way we represent them as distinct from ourselves and other things
that the representation of space is a priori. The conceptualist makes the role of a
priori intuition in organising the sensory manifold redundant: if the role of both a
priori intuition and a priori conceptually governed synthesis is to enable us to be
presented with distinct, unified perceptual particulars, then a priori intuition and
a priori concepts do not make essentially distinct and complementary contributions
to cognition.
The non-conceptualist position requires that subjects have ways of discriminating
individual things from other things, of perceiving their boundaries, and of repre-
senting them as unified, discrete, and distinct from each other and from the subject
that are independent of conceptualising. Kant thinks that (an essential part of) the
way we do this is by representing things as spatio-temporally located and spatio-
temporally related to other things, and he thinks that this is a way of representing
things that is fundamentally different from conceptualising.42 He thinks that the
representation of spatial location and of a directly presented spatio-temporal frame-
work is logically prior to (cannot be derived from) representation of particulars, is
necessary for presentation of particulars, and has content that is not simply descrip-
tive and is not empirical. He thinks this a priori and non-descriptive form is the
condition of anything being immediately presented to us as singular, rather than
grasped discursively (using concepts), and that this is required to individuate objects
uniquely, to guarantee that there are objects corresponding to our concepts, and to be
in immediate cognitive contact with objects. The a priori forms of intuition are
41
Warren (1998) argues in detail against readings of this passage which sees Kant as trying to show that
space is a necessary condition of individuating or distinguishing objects.
42
In addition to spatio-temporal form, there are other ways in which representations could be unified
that need not involve concepts, such as associative principles which are a function of a creature’s biology
(what it eats and is eaten by), and of its context (how close, for example, it is to something). These can
determine what a creature focuses on or picks out as singular, and therefore how it unifies the sensory
input. Though Kant’s account is concerned only with the a priori conditions of representing particulars,
nothing in his account excludes these kinds of factors, and he clearly allows that non-concept-having
animals have ways of associating. See Golob (2011).
 MANIFEST REALITY
required for our having singular and immediate presentations of the objects about
which we think.
IV The Transcendental Analytic

Conceptualists frequently argue that the Aesthetic must be reread in the light of the
Transcendental Deduction and that this enables us to discount the apparent support
the Aesthetic provides for non-conceptualism. The latter section certainly provides
the strongest grounds for conceptualism, and in it Kant does say that we need to
revisit some of his claims in the Aesthetic in the light of the later arguments. As I have
noted, there are two main features of the argument in the Deduction which seem to
support conceptualism: Kant’s claim that the categories are necessary for us to have
‘relation to an object’, and his account of the role of what he calls synthesis in
unifying and organising intuition. In this section I argue that both these parts of
the text can straightforwardly be read in a way that is compatible with a non-
conceptualist reading of intuition.
There are a number of ways in which representations could have relation to an
object, and the phrase ‘relation to an object’ could be used to mean a number of
different things—including being caused by an object, presenting an object, or
depicting an object. I will argue that Kant’s concern in the Deduction is with
something specific: what it takes for thought (concepts) to refer to an object.
I argue that Kant aims to show that the categories are conditions of having referential
thoughts in which we apply empirical concepts to objects. This is a radical position.
Ginsborg says that the non-conceptualist reduces Kant’s exciting idea that concepts
are needed for perception to the unexciting idea that concepts are needed for thought
about objects (2008). In my view, Kant’s argument in the Deduction is not for the
trivial claim that concepts are needed for thought about objects, but for the radical
claim that a priori concepts (concepts containing necessity and universality) are
needed for referential empirical thought about objects. For Kant, referential thought
is not the same as acquaintance with objects, and it is not required to have acquaint-
ance with objects, but rather, is dependent on our having acquaintance, but is also
dependent on the application of the categories. While I give a detailed account of
Kant’s arguments concerning the role of the categories in enabling thought to have
‘relation to an object’ in Chapter 11, the point already stands here that his claiming
that the categories are needed for ‘relation to an object’ cannot be assumed to entail
conceptualism about intuition. We should not start with a general account of what
‘relation to an object’ might mean, independent of the context, and then insist that
Kant’s concern in the Deduction is with showing the categories are necessary for this.
Rather, we need an interpretation of this phrase in the Deduction that is informed
and constrained by the fact that Kant says that the role of intuition is to give us
objects, and that this is something concepts can never do, as well as by his arguments
in the Transcendental Aesthetic. There is absolutely no reason to approach the
CONCEPTS AND INTUITIONS 
Deduction with the default assumption that to have relation to an object is to have a
perceptual particular presented to consciousness.
The conceptualist position might seem to be supported by a central feature of the
argument in the Deduction: Kant’s account of the role of what he calls synthesis in
unifying the manifold of intuition—and, in particular, the threefold synthesis of
apprehension, reproduction, and recognition.43 Synthesis is a way of combining
representations. Kant says ‘By synthesis in the most general sense . . . I understand
the action of putting different representations together with each other and compre-
hending their manifoldness in one cognition’ (A77/B103). While, as we will see, the
relation between synthesis and concepts is delicate, it is clear that Kant thinks that the
categories are rules which govern synthesis—ways in which we combine the manifold
of intuition—and Kant says this synthesis introduces unity to the manifold of
intuition. The conceptualist reading sees Kant as saying that categorially governed
synthesis is necessary to produce intuitions—presentations of unified, single par-
ticulars. However, the syntheses Kant argues for in the Deduction are something
that is done to intuitions, and therefore they need not be understood as what
produces intuition, or as something that is necessary to have intuitions. This point
is crucial. The idea that unifying a manifold of intuition is producing intuitions is
by no means a compulsory, or even the most natural, reading of the text. Intuitions,
not sensations, are what we are synthesising; intuitions are not produced via this
synthesis.
It is important that we should not read ‘synthesis’ (and, in particular, the syntheses
with which the Deduction is concerned) as a general term covering any possible
organisation of the sensory input by the mind.44 It is indisputable that Kant thinks
that organising or structuring sensation spatio-temporally involves a contribution by
the mind that is, in the first instance, not conceptual, since it involves the a priori
43
See Griffith (2010: 10).
44
In addition to talking about synthesis Kant refers to ‘combination’, and because ‘combination’ sounds
less like a term of art than ‘synthesis’, it might seem natural to read Kant’s talk of combination as including
all mental binding or organising of sensory input. However, this is simply not how he introduces the notion
of combination. He says that all combination, whether we are conscious of it or not, whether it is a
combination of the manifold of intuition or of several concepts, and in the first case either of sensible or
non-sensible intuition, is an act of the understanding, which we would designate with the general title
synthesis in order at the same time to draw attention to the fact that we can represent nothing as combined
in the object without having previously combined it ourselves (B130). Kant says that combination is an act
of the understanding, and that it is something we can do both to the manifold of intuition (the particulars
presented to us) and to concepts. The idea that organising sensation in the a priori forms of intuition is an
act of the understanding would make nonsense of the Transcendental Aesthetic. And the idea that
‘combination’ is a semi-technical term is supported by Kant’s saying that ‘Combination is the represen-
tation of the synthetic unity of the manifold. The representation of this unity cannot, therefore, arise from
the combination; rather, by being added to the representation of the manifold, it first makes the concept of
combination possible’ (B130–1). He thinks that there are ways of associating or binding which do not lead
to a representation of the synthetic unity of the manifold (for example, the subjective association animals
have), which means that combination refers to something more specific than associating or binding
in general.
 MANIFEST REALITY
forms of intuition (A20/B34); he does not refer to this organisation as synthesising.45

In the terms of contemporary cognitive psychology, we could put the point by saying
that the conceptually governed synthesis Kant is concerned with in the Deduction
need not be read as his solution to what has come to be called the binding problem.
The binding problem can be described as follows:
Sensory information arrives in parallel as a variety of heterogeneous hints, (shapes, colors,
motions, smells and sounds) encoded in partly modular systems. Typically many objects are
present at once. The result is an urgent case of what has been labeled the binding problem. We
must collect the hints, bind them into the right spatial and temporal bundles, and then
interpret them to specify their real world origins. (Treisman 2003: 97)
An example often given to illustrate the binding problem is the fact that the visual
system processes colour and shape in different streams, and needs some way of
organising or binding together, for example, redness and roundness as both belong-
ing to a round red tomato, as opposed to just informing the subject that redness is
present and roundness is present. Since synthesis, in Kant’s account, involves orga-
nising, ordering, and unifying sensory representations, it might be thought of as
Kant’s account of perceptual binding. This seems to me to be a mistake. Before he
introduces the notion of synthesis, in the Deduction, Kant has already got an account
of something like binding, in the Transcendental Aesthetic. Kant thinks the sensory
input needs to be arranged in the a priori forms of intuition (space and time) in order
for us to have empirical intuitions, or representations of whole, distinct particulars.
Kant does not talk about synthesising a mass of sensation to produce intuition (as the
conceptualist reading requires) but instead talks about synthesising a manifold of
intuition to make possible cognition. While there may be reasons for re-reading the
Aesthetic in the light of the Deduction, we surely should read the Deduction in a way
that is informed by what Kant has told us in the Aesthetic. He has told us that
intuitions are singular and immediate representations which give us objects, and that
this involves matter which comes from sensation being organised in an a priori
intuitive form. This means that a manifold of outer intuition is an immediately
presented array of distinct, spatio-temporally located and related particulars. The
manifold of intuition that needs to be synthesised and brought to the transcendental
unity of apperception is a mass of unclassified outer individuals (perceptual particu-
lars that have been produced by binding), not a mass of unorganised sensory data.
In addition to talking about synthesising the manifold of intuition, Kant talks about
synthesising the manifold in an intuition. In his introduction to the synthesis of
apprehension in the A edition, Kant says that an intuition gives us ‘one representation’
45
In my view, it is best not to use the term ‘synthesising’ to refer to the structuring or ordering
introduced by a priori intuition: Kant never talks of synthesis in this regard. However, if we were to accept
using the word ‘synthesis’ broadly, to include any contribution made by the mind in structuring the input
of the senses, this would not help the conceptualist about intuition. It would have the result that synthesis
would play a role in constituting intuition, but would not show that concepts are involved in this.
CONCEPTS AND INTUITIONS 
and does not enable us to represent what it presents as ‘containing a manifold in itself ’,
or as having a manifold within it (A99). Conceptualists take this to mean that we do not
have a presentation of a unified individual before we have synthesised intuition. But an
alternative, and at least equally natural, reading is that intuition presents us with things
only as singular (aesthetic unities), and until we have synthesised these singular
representations, we are not able to represent them as complex objects (as having a
manifold within them). Synthesising the manifold in an intuition is needed to grasp a
particular we are presented with as a complex of parts: to represent the manifold in it as
manifold and therefore to represent a complex object as a unified complex. This
synthesis does not produce singular, unified representations in the first place; rather,
it is something that is done to singular representations, so that we can represent the
complexity in them.
The thought that there is a way of being presented with something as one (i.e.
singular/unified) that involves representing it as a whole prior to grasping its parts,
and prior to being in a position to represent it as having parts, is in fact necessary to
explain how conceptual synthesis can get off the ground. If we grasped an object as a
whole by synthesising together all its parts, we would have to start with its smallest
parts (or representations of smallest parts), but Kant thinks that objects do not have
smallest parts (and we do not represent them). This means that it cannot be the case
that we first represent the smallest parts, combine or synthesise them, and then
conceptually grasp the resulting unity. Rather, we first need to be presented with
something presented as singular (a unit); this puts us in a position to represent the
thing as a complex of parts through a conceptually governed synthesis of the
manifold in it. Presentation of basic perceptual unities is also necessary, Kant thinks,
to get the process of measurement off the ground and to represent things as
aggregates of parts that occupy determinate extents of space. Kant says that cognizing
the extent of an object in space requires the category of magnitude, which involves
representing ‘the successive addition of one (homogeneous) unit to another’ (A142/
B182).46 But the representations of homogenous basic units could not be produced in
this way: for basic units to be a product of synthesis, they would have to be produced
46
As Golob’s helpful discussion of this point explains, Kant allows a notion of a basic measure, where
this corresponds to a basic perceptual unity. Kant says that every intuition is represented as absolute unity
and also that any intuition can always be divided into a manifold of parts. As Golob argues, ‘[t]his naturally
raises the following question: why do my perceptions represent one level of decomposition as opposed to
another? The core of Kant’s answer lies in his conception of a ‘‘basic measure’’ [Grundmaß] which we ‘‘take
in directly in one intuition’’’ (Golob 2011: 7). The basic measure ‘refers to the explanatorily primitive
capacity for the representation of a determinate spatial extent’, or a unit of spatial representation (2011: 7).
What a creature tends to attend to as a basic perceptual unit can vary, depending on context (I could see the
hill and the trees on it as a unity, from a distance, and a leaf as a singular thing up close) and can be
determined by biologically driven principles of association (what size, for example, its prey typically is). He
argues that the basic measure is determined by imagination (2011: 9), but that in our perception it is
affected by being brought under the synthesis of apprehension. As Golob points out, iteration of the basic
measure occurs successively, but the basic unity is presented in one moment (A99).
 MANIFEST REALITY
by successive addition and synthesis of more basic units.47 To avoid a vicious regress,
the process must start with something which is presented as a unity in a way which
does not require conceptual synthesis: this is exactly what intuition gives us. Far from
being in tension with the idea that there is a way of representing unity or particularity
that is prior to conceptualisation, the Deduction, and its arguments concerning
synthesis, as well as those in the Axioms of Intuition, require this. This argument
undermines any reason for thinking that this section mandates conceptualism about
intuition. While they are sometimes read as being about what it takes to constitute
intuition, the principles are conditions of cognition, and in the Axioms Kant talks
about what it takes to apprehend appearances and about cognizing determinate
spaces and times. He argues that appearances can be cognized only through succes-
sive synthesis in apprehension which corresponds to representing them as extensive
magnitudes (A163/B204). This successive synthesis cannot be what is responsible for
there being perceptual units since it requires perceptual units, and what this synthesis
is a condition of is cognizing and apprehending appearances, not being given them.
I have argued that there are strong reasons for understanding synthesis (and at
least the syntheses Kant is concerned with in the Deduction) as primarily something
done to intuitions. In brief summary: synthesis is not binding; Kant opens the
Deduction denying conceptualism about intuition; intuitions are not sensations,
and Kant speaks about synthesising intuitions, not about synthesising sensation;
the syntheses he is concerned with in the Deduction and Principles require the
presentation of singular/whole perceptual units, which can be aggregated (to repre-
sent magnitudes) and which can be represented as complexes of parts or complexes
of properties as a result of conceptually governed synthesis. Once we see the
possibility of reading Kant in this way, we are able to see how many key passages
thought to support conceptualism can straightforwardly be read in alternative ways,
which enables us to avoid the heavy textual costs associated with conceptualism
without attributing inconsistency to Kant. For example, Kant famously says that ‘the
same function that gives unity to the different representations in a judgment also
gives unity to the mere synthesis of different representations in an intuition’ (A79/
B104–5). This passage is taken to support conceptualism when it is taken as saying
that the same function that operates in judgment is involved in producing intuitions.
But it need not be read in this way. Rather, an alternative is to see the function as
introducing the kind of unity to the manifold of intuition (the manifold of singular,
discrete things) that is required to cognize it.
I have argued that we should not take Kant’s term ‘synthesis’ to refer to absolutely
any contribution the mind makes to ordering and organising the sensory input, and
in particular that it does not refer to the ordering of sensation in the forms of
intuition which enables us to be presented with appearances (A20/B34). However,
47
See Golob (2011) for detailed and extremely helpful discussion of this point.
CONCEPTS AND INTUITIONS 
there is a further question about whether synthesis always involves concepts, and
how synthesis relates to the understanding (I return to this in Chapter 11). It is clear
that Kant thinks that there are close relations between synthesis and concept appli-
cation: in the Deduction, he says that the understanding brings synthesis to concepts
(A78/B103), that concepts are rules which govern synthesis, and that applying a
concept involves consciousness of the unity of synthesis (A103). However, Kant also
frequently attributes synthesis to the imagination, saying that synthesis in general is
‘the mere effect of the imagination, of a blind though indispensable function of the
soul’ (A78/B103).48 Nuanced versions of moderate conceptualism may appeal to
ways of understanding synthesis that are neither merely the ordering that is contrib-
uted by the a priori forms of intuition, nor the explicitly conceptual ordering that
involves judging, but something between the two. It can then be argued that this is
what is necessary in order for us to be presented with perceptual particulars in
intuition, that it does not require explicit judgments or the application of reflected
concepts, but that it is essentially related to concepts or to the understanding in some
sense. Interesting possibilities here are offered by Longuenesse, Grüne, and Land.
Longuenesse distinguishes between the categories understood as ways of grasping a
unity of synthesis, and as ‘reflected concepts’ or discursive rules (Longuenesse 1998:
46–7). Land suggests that we should distinguish between the understanding as a
capacity for thought and understanding as a capacity for apperception, he argues that
it is the latter that is necessary for intuitions to give objects to us.49 And Grüne argues
that we need to distinguish obscure from clear concepts and that obscure categorial
synthesis is needed to produce objects of perception, but that obscure categories are
concepts neither in the contemporary sense, nor in the full-blown Kantian sense,
since they do not involve the capacity to classify objects or knowledge of inferential
relations between concepts (which would involve having clear and distinct con-
cepts).50 According to these strategies, synthesis could be needed to constitute
intuitions even if the full-blown application of reflected concepts in judgments is not.
In my view, all three of these strategies make an important contribution to
understanding Kant’s strategy in the Deduction. There is significant textual evidence
as well as philosophical reasons for thinking that synthesis should not be equated
with applying reflected concepts, and that pre-conceptual synthesis carried out by the
imagination is important to his account. If Kant’s strategy required full-blown
reflected concepts, it would be difficult to see how anyone other than transcendental
philosophers could have experience of objects. However, the non-conceptualist about
48
See also A120, A118, A119, A123, A124, and B151. The B130 ‘combination’ passage is sometimes
taken to show that Kant thinks that synthesis necessarily involves the understanding, but this is not
completely clear from the passage: it could be taken to be introducing combination as a form of synthesis
which involves the understanding, and this would be more in keeping with the fact that Kant frequently
attributes synthesis to the imagination, and not the understanding (see A78/B103, A120, A118, A119,
A123, A124, and B151).
49 50
Land (forthcoming). Grüne (forthcoming).
 MANIFEST REALITY
intuition can take on board these contributions, because saying that there is a role for
a form of synthesis that is prior to judgment or reflected concepts (which seems to me
to be right) still does not imply that this synthesis produces intuitions. These pre-
conceptual syntheses can still be understood as something done to intuitions.
I make two final comments about the Deduction. One is that it might be argued
that Kant’s aim in the Deduction is inconsistent with my non-conceptualist reading
of intuition, since he wants to show that everything given in intuition must fall under
the categories.51 However, demonstrating the latter is straightforwardly not the same
as showing that everything given in intuition must fall under the categories in order
to be presented to us in intuition. Similarly, the idea that we need to justify our
entitlement to apply the categories to objects does not imply that this can be justified
only by showing that the categories are conditions of being given objects in intuition.
As we have seen, Kant opens the Deduction by explicitly stating that he wants to
prove that everything given to us in intuition falls under the categories and also
by saying that being brought under the categories is not a condition of things
being given to us in intuition. This would make it entirely perverse to insist that
his strategy is to show that it is because falling under the categories is a condition of
being presented in intuition that everything presented to us in intuition falls under
the categories.
Finally, a comment about what is, in my view, one of the strongest bits of textual
evidence for the moderate conceptualist view. In a famous but controversial footnote
at the end of the Deduction in the second edition Kant says that
Space, represented as object (as is really required in geometry), contains more than the mere form
of intuition, namely the comprehension of the manifold given in accordance with the form of
sensibility in an intuitive representation, so that the form of intuition merely gives the manifold,
but the formal intuition gives the unity of the representation. In the Aesthetic I ascribed this unity
merely to sensibility, only in order to note that it precedes all concepts, though to be sure it
presupposes a synthesis, which does not belong to the senses but through which all concepts of
space and time first become possible. For since through it (as the understanding determines the
sensibility) space or time are first given as intuitions, the unity of this a priori intuition belongs to
space and time, and not to the concept of the understanding. (B160)
A conceptualist argument can use this to support the claim that the Aesthetic needs
to be re-read in the light of the Deduction, and that Kant thinks the understanding
affects sensibility. However, the idea that synthesis affects the way we represent
space, and that this is necessary to enable us to represent space as the unified object
that geometry studies, is fully compatible with my account. The conceptually
governed syntheses the Deduction is concerned with affect the way we represent
space in ways that are necessary for cognition of space as an object (‘as is really
required in geometry’). As Kant says, this is something more than simply
51
I am frequently given this objection in discussion.
CONCEPTS AND INTUITIONS 
having the form of space as the form of intuition (a non-concept having animal that
has a spatial form of intuition does not represent space as the unified object that is the
object of study of geometry). Kant thinks that we need conceptually governed
synthesis to cognize determinate spaces (determinate extents in space) and to cognize
space as the unified complex object studied by geometry: to grasp the manifoldness in
space as manifold. This is compatible with thinking that we also have a representa-
tion of space as an intuition (the form of intuition) that is independent of concep-
tually governed synthesis. However, the conceptualist can respond that Kant seems
to be saying in this passage that the unity of space and time, and their being given as
intuitions, depends on a synthesis. I make a few comments about this. One, although
Kant does introduce the claim that some kind of synthesis produces our unified
representations of space and time, and refers to the understanding, he also explicitly
denies that this synthesis involves concepts (‘it precedes all concepts’). So it does not
give support to the idea that the application of concepts is needed for space and time
to be given as intuitions. Two, when Kant refers to ‘a synthesis which does not belong
to the senses’, he could mean by this that it does not belong to the senses because it is
a priori, not because it does not concern to sensibility. And three, it could be that it is
the unity of the formal representation (space being given as the particular object of
geometry) that requires a synthesis, which would fit well with the rest of Kant’s
account. Finally, this is an unclear and controversial footnote at the end of the
B Deduction; it cannot be taken to drive the interpretation. It cannot, for example,
outweigh Kant’s opening the Deduction by saying that ‘The categories of the under-
standing, on the contrary, do not represent the conditions under which objects are
given in intuition at all, hence objects can indeed appear to us without necessarily
having to be related to functions of the understanding . . . for intuition by no means
requires the functions of thinking’ (A89–90/B122–3), as well as all the other textual
and philosophical considerations we have looked at concerning the singularity and
immediacy of intuition.
The conceptualist reading comes at serious cost: it is inconsistent with fundamen-
tal claims about cognition made repeatedly throughout the Critique and never
retracted. It should be accepted only if there is no alternative available. There are
alternatives. I have argued that Kant’s claim that intuitions are singular and imme-
diate representations which give us objects should be understood as saying that
intuitions present particulars to consciousness.
In Chapter 8, I show that the reading of intuition presented here enables us to
understand Kant’s main argument for transcendental idealism, and look at the
implications this has for interpreting his idealism. Intuitions give us acquaintance
with particulars outside us; they are not merely mental images that represent objects;
this fits with the relational account of essentially manifest qualities, given in Chapters
5 and 6. My reading of intuition allows us to take seriously the role Kant says
intuitions play in cognition: that of giving us objects. As we will see, Kant’s idealism
centrally concerns limiting empirical reality to what can be given to us.
8
The Argument for Transcendental
Idealism in the Transcendental
Aesthetic
I Introduction
The final piece in presenting my essential manifestness account is to show how well it
fits with Kant’s main argument for transcendental idealism, which is given in the
Transcendental Aesthetic.1 This chapter presents an interpretation of this argument,
and then evaluates the kind of idealism that follows from the argument. My focus is
primarily on understanding Kant’s strategy—what the argument is supposed to be—
rather than on evaluating, much less defending, it. Kant’s central argument for
transcendental idealism turns on his notion of intuition, and, in particular, the notion
of a priori intuition. I will show that this has implications for how we understand the
nature of his idealism, as well as his explanation of the possibility of metaphysics,
which I discuss further in Chapter 12.
By the end of the Aesthetic, Kant takes himself to have established that space and
time are not features of things as they are in themselves and are merely the forms of
our intuition. He also takes himself to have established the generalised idealist claim
that spatio-temporal objects are mere appearances which do not exist independently
of our experience of them and which do not present us with things as they are in
themselves. There is much debate about how Kant takes himself to establish his
conclusion and even about what the premises in the argument are supposed to be.
Kant presents brief arguments for the claim that our representations of space and
time are a priori intuitions, and he also appeals to the claim that we need to explain
the possibility of cognition of synthetic a priori mathematical propositions about
space and time. He then concludes that space and time do not represent things in
1
His other explicit official argument is the indirect argument in the Antinomies (A405–567/B432–595).
The indirect argument says that on the assumption that our knowledge of the world is knowledge of things
as they are in themselves we find ourselves driven to contradictions when we try to think about the extent
of the world in space and time, the divisibility of matter, and freedom of the will. Kant argues that these
contradictions can be avoided by seeing that our knowledge is only of things as they appear to us, rather
than of things as they are in themselves.
THE ARGUMENT FOR TRANSCENDENTAL IDEALISM 
themselves and are merely mind-dependent forms of intuition; how this conclusion
is supposed to follow is unclear and widely disputed. In my view, the key to the
argument in the Aesthetic is Kant’s notion of intuition. As I will show, if we
understand intuition in the way I set out in Chapter 7—as a kind of representation
which gives us acquaintance with its objects—we can see why Kant takes his
conclusion to follow at the point at which he does and from the premisses he has
presented. As I understand him, Kant thinks that it follows from what an intuition is
(the fact that it involves the presence to consciousness of what it represents) that an a
priori intuition could not present a mind-independent feature of reality. This is why
he takes it to follow from showing our representations of space and time to be a priori
intuitions that they do not present mind-independent features of reality.
Kant’s idealism is closely linked to his answer to the question with which he opens
the Critique: how are synthetic a priori judgments possible? He says that cognition of
synthetic a priori claims is a mystery ‘the elucidation of which alone can make
progress in the boundless field of pure cognition of the understanding secure and
reliable’ (A10). He says that the possibility of metaphysics stands or falls with a
solution to this problem (A8–10; B19; B11–13; Proleg. 4: 257, 260, 377; 29: 794).
Transcendental idealism is supposed to enable us to explain how cognition of
synthetic a priori metaphysical claims is possible. Famously, Kant’s solution is that
cognition of synthetic a priori propositions in metaphysics can be made intelligible if
such propositions are taken to give us cognition only of things as they appear to us
and not of things as they are in themselves (Bxviii).2 How exactly this explanation
relates to Kant’s argument for transcendental idealism is a delicate question. One of
the central points I argue for in this chapter, and in Chapter 12, is that Kant’s
explanation of the possibility of synthetic a priori cognition in geometry is not the
same as his explanation of metaphysics, and that the former, but not the latter, is part
of his argument for his transcendental idealism.
One simple and prominent way of reading of Kant’s argument runs together his
explanation of metaphysics and his argument for idealism. On this interpretation, it
is thought that Kant has a general explanation of the possibility of synthetic a priori
cognition according to which it is in virtue of the fact that these propositions involve
only what our minds bring to the world that we are able to have insight into them, or
to justify them, independently of experience. The thought is that, first, idealism
explains the possibility of cognition of synthetic a priori claims; and, second, the
fact that idealism explains such cognition is the argument that establishes idealism.
Kant starts with an example of cognition synthetic a priori propositions that he takes
to be actual and indisputable: mathematics.3 On this simple reading, the argument
2
See also Bxix and Bxxiii. B16; A114; A126–30; B167; Proleg. 4: 257, 260, 282, 319, 377; D 8: 240.
3
It is sometimes argued that for Kant to assume the truth of geometry in the Critique would be
inconsistent with a description he gives in the Prolegomena of the contrast between his method in that
work and in the Critique. He describes his approach in the Critique as using the synthetic approach, which,
 MANIFEST REALITY
says that this actual knowledge is explicable only on the assumption that it is
knowledge of something our minds are responsible for; therefore, our minds are
responsible for this feature of reality. This same explanation is then taken to show
how metaphysics is possible: we are able to have knowledge of metaphysical synthetic
a priori principles because these are principles our minds bring to the world, so the
knowledge involves only insight into our minds. In my view, this simple story
attributes to Kant a very bad argument, and not one he actually presents. In addition,
it does not pay sufficient attention to the difference between mathematical and
metaphysical synthetic a priori claims in Kant’s account, or to the role of intuition
in Kant’s account of cognition.
Kant’s explanation of metaphysics is outlined in the B Preface in the famous
passage in which he compares his position to the Copernican Revolution in astron-
omy. Here, he introduces transcendental idealism as an experimental hypothesis: if
we assume his position, we will be able to make sense of metaphysical (a priori)
cognition. He describes the hypothesis as the assumption that objects ‘conform’ to
our a priori cognition, as opposed to the claim that our cognition must conform to
objects. He says that this hypothesis ‘would agree better with the requested possibility
of an a priori cognition of them, which is to establish something about objects before
they are given to us’ (Bxvi). He concludes, famously, that ‘we can cognize a priori of
things only what we ourselves put into them’ (Bxviii). This, Kant says, explains the
possibility of metaphysics: ‘after this alteration in our way of thinking we can very
well explain the possibility of cognition a priori’ (Bxix).
Coming, as it does, at the very beginning of the book, the summary argument
presented here might be understood as both an overview of Kant’s account of how
metaphysical knowledge is possible for us and a summary of his argument for
transcendental idealism. In my view, however, it is only the former, and the two
are importantly different. The question of how synthetic a priori judgments are
possible in general is not identical to the question of how metaphysics is possible,
although Kant thinks we must answer the former to answer the latter. And Kant’s
argument for transcendental idealism is different from his explanation of the possi-
bility of metaphysics. As I read him, Kant thinks that we have cognition of synthetic a
priori propositions in mathematics, that the possibility of such cognition requires
explanation (whether or not metaphysics should turn out to be possible), and that
explaining the possibility of this mathematical cognition does not provide a complete
unlike the analytic method of the Prolegomena, does not start by assuming the knowledge in question as
actual. (See, for example, Kitcher 2011.) As I understand him, the question Kant says he addresses
synthetically in the Critique is the question whether and how metaphysics is possible (Proleg. 4: 274). He
clearly thinks, in both works, that there is some actual synthetic a priori cognition that is not metaphysical
(mathematical claims), and is not in doubt, nor in need of vindication from philosophy. However we
understand his distinction between his analytic and synthetic methods, Kant never says that mathematics is
subject to doubt, and he nowhere gives an argument attempting to establish its security. He does, of course,
want to explain the possibility of mathematics; but it is precisely because its security is not in doubt that this
is supposed to be powerful.
THE ARGUMENT FOR TRANSCENDENTAL IDEALISM 
explanation of metaphysical knowledge. The explanation he gives of the possibility of

mathematical cognition is that we have a priori intuition, not the idea that our minds
make objects in such a way as to make mathematical claims come out true. Kant then
takes his idealism to follow from his explanation of mathematical cognition—a priori
intuition—together with the claim that this a priori intuition is the form of our
empirical intuition. He then argues that this transcendental idealism can explain how
metaphysical knowledge is possible, but this is not an argument for his idealism; it
depends on idealism’s having been established. The explanation of the possibility of
metaphysical knowledge is also not that our minds make objects in such a way that
these claims come out as true: rather it is that principles which can be demonstrated
to be conditions of the possibility of empirical cognition can be known to be true of
all spatio-temporal objects if we have shown that spatio-temporal objects are limited
to the conditions of our cognition.
In Section II I look at the problems with the simple reading of the argument
according to which transcendental idealism explains how we are able to justify
synthetic a priori claims by appealing to the thought that our minds ‘make’ them
true, and according to which this explanation of our cognition of synthetic a priori
judgments is also the argument for transcendental idealism.
Section III looks at Kant’s presentation of his argument in the Transcendental
Aesthetic. Kant argues for the claim that our representations of space and time are a
priori intuitions; he then says that two conclusions that follow from this are that our
representations of space and time do not present features of things as they are in
themselves, but are merely the form of our intuition. The problem is to explain is why
the latter follows from the former.
Section IV motivates a particular reading of Kant’s question: how are synthetic a
priori propositions possible? I argue that his concern is not with how we justify or
establish such claims but with the prior question of how it could be the case that such
claims concern objects which can be given to us. Kant considers this to be a
requirement of their qualifying as possible cognition. I argue that Kant’s answer to
his ‘how possible’ question is a priori intuition.
In Section V I argue that if we take the reading of intuition for which I argued in
Chapter 7—the view that intuitions are representations which involve the presence to
consciousness of the objects they represent—Kant’s idealist conclusion follows from
the claim that space and time are a priori intuitions in exactly the way he thinks it
does and at the point he thinks it does, without the need for further premisses. I argue
that it is this feature of intuition, in combination with apriority that, according to
Kant, leads to the result that we have a representation that cannot present us with a
feature of mind-independent reality. While I draw on the account of intuition
defended in Chapter 7, this chapter also provides support for that account, since it
enables us to give an explanation of how the argument for idealism is supposed
to work which fits the way Kant presents it and which avoids standard objections to
his argument.
 MANIFEST REALITY
II A Too-Simple Argument
Kant starts with the claim that synthetic a priori judgments are mysterious. Many
rationalists would not accept this,4 so understanding Kant’s argument requires, as a
first step, explaining what the problem is supposed to be—why Kant takes synthetic a
priori knowledge to be mysterious. In fact, Kant does not provide much argument or
explanation for this. He says in the Prolegomena: ‘it is simply not to be seen how
things would have to agree necessarily with the image that we form of them ourselves
and in advance’ (Proleg. 4: 287; see also Bxvii, A10, B19, and Proleg. 4: 294, 282). He
seems to think that previous philosophers have failed to recognise that we have
cognition of synthetic a priori propositions—sometimes because they have failed to
see the need for the relevant knowledge claims, and sometimes because they have not
recognised them as synthetic and a priori (B19; B20). He seems to think that once it is
recognised that we have cognition of propositions that are both synthetic and a
priori, this will immediately be seen to require explanation.
A simple explanation of why cognition of synthetic a priori claims is mysterious is
that we cannot explain the source of the justification of such claims. While we
supposedly have a general explanation of how analytic a priori claims are justified
(through analysing concepts) and of how synthetic a posteriori claims are justified
(empirically), with respect to cognition of synthetic a priori claims these two sources
of justification are not available, so it seems hard to see how we could justify them.
This account of the mystery gives rise to a simple account of the role idealism is
supposed to play in resolving it: the idea is that Kant introduces his idealism to fill the
justificatory gap. On this simple account, it is because our minds are somehow
responsible for those features of objects which we can know a priori that this
knowledge is possible. Stroud, for example, says that, on Kant’s view, a priori
knowledge of the fundamental structure of the world of experience ‘is possible at
all only if it is somehow knowledge of us, or of what “we” bring to the “raw material”
of “sensible impressions” ’ (2000b: 229). He says that according to Kant, a priori
knowledge is in some sense ‘contributed by us’ and ‘therefore directly available to us
by the operation of “pure reason” with complete certainty’ (2000b: 238, my italics).
According to this simple argument, we need no explanation of how we justify claims
concerning what comes from our minds: the fact that it comes from our minds is an
explanation of the source of the justification. In a slogan: we can have a priori
knowledge of what our minds make. There are a number of objections to the simple
argument. I will mention five problems which show that this reading attributes to
4
Lawrence Bonjour, for example, argues that Kant’s belief in the mysterious nature of synthetic a priori
knowledge is born of his hostility to rationalism, and that ‘a viable, non-skeptical epistemology, rather than
downgrading or rejecting a priori insight, must accept it more or less at face value as a genuine and
autonomous source of epistemic justification’ (Bonjour 1998: 92). Bonjour puts pressure on the claim that
synthetic a priori knowledge, in particular, is mysterious by questioning the claim that analytic necessity is
not mysterious. A similar argument is made by Stroud (2000b).
THE ARGUMENT FOR TRANSCENDENTAL IDEALISM 
Kant a very weak argument. I then explain why, in my view, it is not the argument
that Kant presents.
One problem with the simple argument is that it makes Kant vulnerable to a long-
standing objection to his argument: the so-called ‘neglected alternative’. Originally
specifically about space and time, the objection can be more broadly stated by saying
that Kant assumes that synthetic a priori principles must either reflect the way our
minds are or reflect the way the world is in itself, but neglects the possibility that both
could be true. The idea that we have insight into certain principles because they are
principles which come from our minds and not from the world does not rule out the
possibility that the world might correspond with these principles. Another way to put
the point is to say that transcendental idealism includes the claim that things in
themselves are not spatial but it is unclear how this could be established by the simple
argument. Even if we accept that geometrical knowledge could be explained only by
assuming that our minds ‘make’ objects in such a way that geometrical claims come
out as true, nothing seems to follow from this about the spatial status of things as they
are in themselves.5
A second problem with the simple version of the argument is that, as Bonjour
argues, it seems to undermine itself in a way parallel to that in which the verification
principle is self-defeating: the claim that we can know a priori of objects only what we
ourselves put into them is neither analytic nor empirical.6 It must therefore be
synthetic and a priori. This means, Bonjour argues, that Kant must explain it in
the only way in which he allows such knowledge to be explicable, which requires him
to say that the cognitive structure of our minds is responsible for the fact that the
cognitive structure of our minds is responsible for the structure of objects. It is
doubtful that this makes sense, and it clearly generates a vicious regress (Bonjour
1998: 24).
My third objection is that if idealism were introduced to close an explanatory gap,
the argument would be vulnerable to other possible explanations of how this gap
could be closed: other explanations of how synthetic a priori knowledge could be
justified. Kant does not present his account as one of a number of possible explan-
ations of how such knowledge could be justified, and he says that his explanation
should not merely ‘earn some favour as a plausible hypothesis’ (A46/B63). Rather, he
sees it as the only possible explanation of the possibility of cognition of synthetic a
priori judgments. This worry is exacerbated by the fact that it seems that Kant himself
has an alternative account of the source of the justification of synthetic a priori
5
As Hogan points out, the neglected alternative objection should strike us as more than a little
surprising, given the aim of Kant’s argument: ‘Kant purports to establish that space and time are merely
subjective forms of sensibility, that is, not also properties of things in themselves. Critics respond that he
has overlooked the possibility that space and time may also be properties of things in themselves.
Something is evidently amiss’ (Hogan 2009a: 357).
6
The verification principle states that all meaning statements are empirically verifiable, but the principle
itself does not seem to be empirically verifiable. A similar objection is raised by Bennett (1966: 17).
 MANIFEST REALITY
metaphysical claims: his transcendental arguments. Roughly, a transcendental argu-

mentative strategy is one which defends some doubted claim by showing it to be a
necessary presupposition of something which is not doubted. While Kant does not
use the term ‘transcendental arguments’ (and most of his arguments are neither
directed to Cartesian-type scepticism nor concerned with the conditions of coherent
thought, as are some contemporary attempts to use the strategy7), he does employ
this kind of argumentative strategy. In particular, Kant thinks we can show that
certain synthetic a priori metaphysical claims are conditions of the possibility of
experience and empirical knowledge.8 If transcendental arguments give us a method
of establishing synthetic a priori claims, then we have a way of distinguishing those
that are justified from those that are not. Further, this method seems to give us
insight into how we are able to know these claims: we justify them by seeing that they
are conditions of the possibility of experience (or of some specific feature of experi-
ence). If we can show that, for example, the principle of causation is justified as a
condition of the possibility of empirical knowledge, it hardly seems mysterious that
we can know that everything of which we can have empirical knowledge is subject to
it. At the very least, it is not obvious that there must be a further explanation of our
having such knowledge and that such an explanation must be accessible to us.9
Independently of his explanation of the intelligibility of synthetic a priori claims in
general, Kant needs a strategy for working out which synthetic a priori metaphysical
claims are justified. He provides this with his transcendental arguments. However,
once he has such a strategy, it seems that the justification of such claims (the way we
establish them) is explained. Any interpretation of Kant’s strategy must have some-
thing to say about how his transcendental arguments relate to his general argument
for idealism. The problem with the simple account of the argument for the idealism is
that the former seem to make the latter redundant.
My fourth objection follows from the fact that, according to the simple account of
the argument, the unintelligibility of synthetic a priori cognition is supposed to
disappear once we see that our minds are responsible for the features of objects
that we can cognize a priori. The problem with this is that it seems to assume, without
7
See Stern (1999a) for an introduction to the contemporary debate, and Parts I and II of Schaper and
Vossenkuhl (eds.) (1989), for some discussion of transcendental arguments in Kant.
8
As I understand Kant’s transcendental strategy, his arguments (almost all) assume that we have
empirical knowledge and try to demonstrate something about its conditions. They are therefore not
responses to Cartesian-type scepticism; nor are they concerned with the conditions of coherent thought.
Stroud objects to this kind of reading, saying that ‘if would-be transcendental reflection is allowed to start
from the fact that we know certain things, there would be no need for further enquiry’ (Stroud 2000a: 210).
As I see it, Kant assumes that we have empirical knowledge and sees the aim of his transcendental enquiry
as being to show that this empirical knowledge requires that we have some a priori knowledge of the objects
of which we have empirical knowledge. If we are trying to establish a priori knowledge which is in doubt,
starting by assuming empirical knowledge is not begging the question. See Ameriks (2003: Ch. 1) and Bird
(2006) for detailed accounts of why Kant’s primary concern in the Critique is not with responding to
Cartesian scepticism.
9
See Walker (1985: 20) and Stroud (2000b: 222).
THE ARGUMENT FOR TRANSCENDENTAL IDEALISM 
explanation, that there is no mystery about the source of the justification of a priori
claims concerning what our minds bring to the world. It is not clear why this should
be so: it is not clear why we should have a priori insight into to what our minds are
responsible for.10 Against this, it might be thought that the assumption of our having
insight into our own minds is something Kant takes to be unproblematic. This might
seem to be supported by his saying that ‘reason has insight only into what it produces
according to its own design’ (Bxiii) and that since ‘I have to do merely with reason
itself and its pure thinking, to gain exhaustive acquaintance with them I need not
seek far beyond myself, because it is in myself that I encounter them’ (Axiv).
However, Kant also clearly thinks that there are aspects of our minds which are
‘blind’ and of which ‘we are seldom even conscious’ (A78/B103), and he thinks that
we do not know ourselves as we are in ourselves—we know neither mind nor matter
as they are in themselves.11 This makes it harder to attribute to him the thought that
our minds’ being responsible for certain features of the world gives us unproblematic
insight into these features. What he thinks reason has unproblematic insight into are
the principles that reason requires for cognition; whether such principles give us
insight into objects is precisely the point that requires explanation.
My fifth and, in my view, most serious, objection, is that the simple argument does
not pay enough attention to the important differences between the status of math-
ematical and metaphysical synthetic a priori claims in Kant’s account. In addition to
simply not fitting the text, this difference means that the explanation of synthetic a
priori cognition postulated by the simple argument attributes to Kant an extremely
weak argument for his transcendental idealism. The problem is that Kant cannot
argue for idealism on the basis of the claim that it would explain metaphysics, since
he does not regard metaphysics as an established, secure starting point, but neither
can he establish the generalised idealism he needs in order to explain metaphysics on
the basis of the simple reading of the argument from geometry. Though the simple
argument (it seems to me) is widely held, commentators do not often make fully
explicit the different ways in which mathematical and metaphysical cognition would
need to feature in it. This may lead to not paying sufficient attention to just how
problematic it would be for Kant to attempt to establish his transcendental idealism
and his explanation of metaphysics on the basis of the simple argument with respect
to mathematics. Perhaps this is because many commentators are doubtful about
whether the argument is any good, but also don’t see an alternative. Once we pay
attention to the Kant’s concern with the conditions of cognition, and not just
knowledge or justification, we will see that Kant has an alternative argument. Before
10
See Van Cleve (2002: 38).
11
In the Deduction Kant says that inner sense ‘presents even ourselves to consciousness only as we
appear to ourselves, not as we are in ourselves’ (B152–3), and in the Paralogisms, he argues that ‘if we
compare the thinking I not with matter but with the intelligible that grounds the appearance we call matter,
then because we know nothing at all about the latter, we cannot say that the soul is inwardly distinguished
from it in any way at all’ (A360).
 MANIFEST REALITY
looking at this, it is important to bring out explicitly just how weak the simple
argument is as a way of establishing the transcendental idealism Kant needs to
explain metaphysics.
It is important that, for Kant, there are a number of differences in the status and
justification of mathematical and metaphysical synthetic a priori claims. First, unlike
his transcendental arguments for synthetic a priori metaphysical principles, Kant
does not argue in the Transcendental Aesthetic that geometry gives us conditions of
the possibility of experience (or empirical knowledge). Second, Kant has different
accounts of how we establish or justify metaphysical and mathematical synthetic a
priori claims: mathematical claims are established through proofs that make use of
construction in pure intuition, whereas metaphysical claims can be established only
by being shown to be conditions of the possibility of empirical knowledge. Third,
Kant opens the Critique assuming that while mathematics is a secure body of
knowledge, metaphysics is in trouble, and that there is serious doubt whether there
are any metaphysical claims that can be established. This has implications for how
the argument relating idealism to the explanation of synthetic a priori claims should
be understood to work.
Remember that the simple argument says that our being able to establish synthetic
a priori claims can be explained by the idealist hypothesis that our minds ‘make’
objects in such a way that these claims come out as true (so the knowledge involved is
insight into our own minds), and that the fact that idealism would explain our
knowledge of such claims is what establishes idealism. However, given Kant’s starting
point, it cannot be that we start with two groups of synthetic a priori claims,
mathematical ones and metaphysical ones, then we postulate for each the explan-
ation that our being able to have insight into these claims (or our being able to see
that all objects are in necessary agreement with these claims) would be explained by
our minds making the claims in that group true of the world, and then finally take
this to establish that the world is in fact dependent on our minds with respect to these
two groups of claims. The argument cannot work like this because we do not start
with two groups of established claims; with respect to metaphysics, unlike geometry,
we do not have the knowledge in question, and we cannot use the explanation the
simple reading offers of geometrical knowledge to establish the generalised idealism
needed to explain our knowledge of metaphysical claims. The simple reading says
that we can explain our knowledge of geometry on the assumption that objects
depend on our minds for their geometrical properties, but the dependence of objects
on our minds for their geometrical properties will not establish their dependence on
our minds with respect to, for example, their standing in relations of causal necessity.
On the one hand, we cannot appeal to synthetic a priori metaphysical claims to
establish idealism with respect to these claims (the dependence of objects on our
minds with respect to these claims) since, unlike geometry, they are in doubt. But, on
the other hand, if the argument for idealism is that it explains synthetic a priori
knowledge we actually have, this argument will not establish the dependence of
THE ARGUMENT FOR TRANSCENDENTAL IDEALISM 
objects on minds in any respects other than are needed to explain mathematical
knowledge. In order to reach the desired conclusion, the proponent of the simple
argument would need to say that our knowledge of geometry can be explained only
on the assumption that our minds shape objects in terms of whatever principles our
minds require for knowledge of objects, rather than on the more specific assumption
that our minds are responsible for geometrical features of objects. This would be
an extremely peculiar explanation of geometrical knowledge and it is not one
Kant gives.
According to the simple reading, the idealist explanation postulates something like
the following: if the spatio-temporal objects of our knowledge are things which do
not exist apart from the possibility of our being able to know them, then reason’s
principles (conditions of the possibility of knowledge) can be known to be true of
these objects. The problem is that this neither establishes that reason’s principles are
true of the spatio-temporal objects of our knowledge nor does it establish idealism. It
says only that if our knowledge of these principles were knowledge of objects, the best
explanation of this would be that the object conformed to our knowledge. But we
have not yet established that we know them. Of course, once we have established the
relevant form of idealism (i.e. the thesis that spatio-temporal objects do not exist
independently of our being able to know them), we can show that principles which
are conditions of the possibility of empirical knowledge of spatio-temporal objects
give us insight into the nature of these objects. But this is not an argument for
idealism. It is an explanation of how conditions of empirical knowledge give us
insight into the objects of knowledge that itself depends on idealism. So while it is
clear that Kant thinks that his idealism plays a crucial part in the explanation of our
knowledge of synthetic a priori metaphysical claims, it cannot be that he takes his
idealism to be established on the basis of its explaining our knowledge of these
metaphysical claims. The generalised idealist claim that spatio-temporal objects do
not exist independently of the possibility of our cognizing them simply would not
follow from the claim that our knowledge of geometrical principles can be explained
by the hypothesis that our minds make these (geometrical) principles true of spatio-
temporal objects. And postulating idealism as a way of explaining metaphysical
claims which are not in fact known cannot be a way of establishing idealism.
Kant thinks that synthetic a priori metaphysical claims can be justified by being
shown to be conditions of the possibility of experience. He thinks that we are able to
have insight, through reason alone, into a priori principles which are the conditions
of the possibility of empirical cognition (such as, Kant thinks, the principle of
causation and the principle of conservation). However, the transcendental proofs
which establish these a priori principles enable us to establish only the limited,
conditional conclusion that they are true of the objects of which we have cognition,
and not that they are true of objects in general. If the conditional claim were our only
concern, then we would not need to add the idealism: a transcendental proof that a
principle is a condition of the possibility of empirical cognition and knowledge is
 MANIFEST REALITY
enough to establish that the principle holds of the things of which we can have
empirical cognition and knowledge. Kant’s idealism enables him to explain how our
insight into principles of cognition is also insight into the objects of cognition: this is
possible because the objects of cognition are merely appearances, i.e. things which are
essentially objects of experience. However, this is not the argument which establishes
the idealism. Rather, the idealism has already been established, and is what enables
Kant to make this argument. Further, the kind of idealism that enables Kant to
convert conditional claims about the conditions of cognition into claims about
spatio-temporal objects (an idealism that asserts that objects are not independent
of the possibility of our cognizing them) could not be established on the basis of the
postulation that our minds making geometrical claims true would enable us to
explain our insight into geometrical claims. And Kant does not present this expla-
nation of geometry.
The problems I have discussed arise from seeing Kant as invoking transcendental
idealism (the dependence of objects on our minds) to explain how we can justify or
establish synthetic a priori claims. This way of reading Kant was perhaps encour-
aged by the translation of Erkenntnis as ‘knowledge’ rather than ‘cognition’, which
may have led commentators to have read Kant’s problem in terms of contemporary
concerns about justification, rather than paying attention to his specific concerns
with cognition. Significantly, Kant does not in fact start by asking how synthetic a
priori judgments are justified or established but with the simpler but perhaps (from
a contemporary perspective) stranger question of how such judgments are possible.
I will argue, in Section IV, that once we pay attention to Kant’s account of cognition
we can see that his concern is with a question that is prior to the justificatory one:
how is it possible for synthetic a priori propositions to concern objects with
which we can have acquaintance (and therefore to qualify as cognition)? In
Section III, we will see that the argument Kant presents in the Aesthetic is not
the simple argument.
III A Missing Step?

I now turn to the presentation of the argument in the Aesthetic. I discuss the
argument with respect to space. In the first edition Kant presents five short, num-
bered arguments for the claim that our primary representation of space is an
intuition, not a concept, and is a priori, not empirically derived. Arguments 1–3
aim to establish that our representation of space is a priori;12 4 and 5 that our primary
12
Argument 1 says that representing objects in space is the way we represent them as distinct from us
and each other, so our representation of space could not be derived from our representation of distinct
objects. Argument 2 says that we can abstract objects from our representation of space, but cannot abstract
space from our representation of objects. Argument 3 says that geometrical claims contain necessity and
that this requires that our representation of space is a priori.
THE ARGUMENT FOR TRANSCENDENTAL IDEALISM 
representation of space is an intuition, not a concept.13 Together, Kant takes these

Arguments to show that our representation of space is an a priori intuition. Imme-
diately after this, without further argument, Kant presents two ‘conclusions’ which
purportedly follow from this: ‘(a) Space represents no property at all of any things in
themselves nor any relation of them to each other’; and ‘(b) Space is nothing other
than merely the form of all appearances of outer sense, i.e., the subjective condition of
sensibility, under which alone outer intuition is possible for us’ (A26). Conclusions
(a) and (b) express the transcendental ideality of space.14 Our question is how this
conclusion is supposed to follow from the claim that our representation of space is an
a priori intuition.15
It is noteworthy that of the five arguments in the first edition for the claim that our
representation of space is a priori and an intuition, only Argument 3 seems to
concern cognition of synthetic a priori propositions (in geometry). Significantly,
Argument 3 does not say that our knowledge of geometry requires ideality to explain
it (much less that our minds make objects in such a way that geometrical claims come
out true), but rather that the necessity of geometrical claims requires that our
representation of space is a priori. This suggests that, like the first two arguments,
Argument 3 aims to establish that our representation of space is a priori. The other
arguments appeal to features of the way space is presented in our experience. It is
13
These arguments turn on the claim that space is represented as singular (Argument 4) and that it is
immediately given (Argument 5), together with the claim that intuitions are singular and immediate
representations through which objects are given.
14
Allison regards conclusions (a) and (b) as representing Kant’s main argument for transcendental
idealism. He says that these claims lead to what he calls the ideality thesis (Allison 2004: 118). However,
Kant presents (a) and (b) as conclusions following from the previous arguments, not as arguments for
space being an ‘objectivating’ or ‘epistemic’ condition. It is true that, on their own, Kant’s conclusions (a)
and (b) fall short of the complete transcendental idealist position, since Kant has not yet generalised his
position to talk about all objects of experience being mind-dependent appearances and has not yet argued
that no knowledge of things in themselves is possible, but it seems clear that they do correspond to the
transcendental ideality of space, since they say that space represents no property of or relation among
things in themselves, and that space is nothing but merely the form of our intuition.
15
Further, Kant takes his argument in the Aesthetic to establish his generalised idealism—i.e. not just
that the objects of our experience are dependent on our minds for their geometrical properties or their
spatio-temporal form, but that spatio-temporal objects are mind-dependent appearances, things which do
not exist apart from the possibility of our having knowledge of them. Kant therefore needs to move from
the claim that space and time are transcendentally ideal to the claim that objects in space and time are
transcendentally ideal. He appears to assume that this step follows without further argument. Having
concluded that space and time do not present us with features of things as they are in themselves and are
merely the a priori form of our intuition, he says that ‘absolutely nothing that is intuited in space is a thing
in itself, and that space is not a form that is proper to anything in itself, but rather that objects in themselves
are not known to us at all, and that what we call outer objects are nothing other than mere representations
of our sensibility’ (A30/B45). What is crucial in understanding why Kant takes this to follow is that he
thinks not just that space and time are a priori intuitions, but that they are the a priori forms of our
intuition—the forms in which the matter of sensation is arranged, in order for us to have empirical
intuitions. Kant thinks that the form in which we represent particulars is something which does not exist
apart from the possibility of our representing it. What is presented in this form is therefore limited to the
possibility of what we can represent: to our possible experience, and, in particular, to what can be given to
us or be presented to us in intuition.
 MANIFEST REALITY
sometimes thought that they are supposed to be arguments for the claim that space is
a condition of the possibility of experience of objects (which would fit with the role of
synthetic a priori claims in the simple argument), but Kant does not appear to argue
this, and the arguments seem instead to go the other way round. In Argument 1, Kant
argues from (not for) the claim that space is the way we represent things as distinct
from us to the claim that our representation of space is not empirical (is not derived
from our experience of distinct objects). Argument 2 is supposed to establish the
apriority of our representation of space by appeal to the claim that, although we can
abstract objects from our representation of space, we cannot abstract space from our
representation of objects. Arguments 4 and 5 are supposed to show that our
representation of space is an intuition on the basis of, first, the way we represent
its singularity (as a whole whose parts are limits, rather than through first represent-
ing the parts and putting them together); and, second, the way we represent its
infinity (as something immediately given not constructed). Together, these argu-
ments are supposed to show that our primary representation of space is a priori, not
empirical, and that it is an intuition, not conceptual.
The conclusion of Arguments 1–5 is that our representation of space is an a priori
intuition. In the first edition, without any further argument, Kant presents his
transcendental idealist conclusions as following immediately from this. He says it
follows from the claim that our representation of space is an a priori intuition that
‘(a) Space represents no property at all of any things in themselves nor any relation of
them to each other’; and ‘(b) Space is nothing other than merely the form of all
appearances of outer sense, i.e., the subjective condition of sensibility, under which
alone outer intuition is possible for us’ (A26). This has seemed puzzling to many
commentators. Arguments 1–5 are supposed to establish claims about our represen-
tation of space, while the conclusions (a) and (b) seem to concern space; it is not clear
how the latter follows from the former. Many philosophers have thought that Kant
needs further argument to go from the claim that space is the a priori form of our
intuition to the claim that space is merely the a priori form of our intuition.16 The
idea that space is the form in which we need to arrange sensations to represent
objects does not seem to rule out the possibility that objects, in themselves, could
have this form. This is the famous ‘neglected alternative’ objection: the idea is that
showing that space is the a priori form of our intuition does not show that it could
not, in addition, be the form that reality has independently of us. Seeing space as the
structure in which we organise the sensory input in our representation of the world
seems compatible with realism about space.17
16
See, for example, Setiya (2004: 67), Hatfield (2006: 76, 82), and Guyer (2006: 58, 63).
17
Even more strongly, Guyer says ‘If we somehow know a priori that we can only perceive objects
distinct from ourselves in space, indeed in three dimensional Euclidean space, why isn’t the explanation of
our success in perceiving some particular outer object precisely that it really is spatial, indeed three-
dimensional, quite apart from our representing it as such?’ (2006: 63).
THE ARGUMENT FOR TRANSCENDENTAL IDEALISM 
The idea that there is a step missing in the argument might seem to be supported
by the alteration Kant makes in the second edition. In B, the arguments for the claim
that our representation of space is a priori and intuitive are grouped under the
heading ‘The Metaphysical Exposition of the Concept of Space’, and the Third
Argument in the A version, which concerns geometry, is taken out of this section
and is developed at more length, in the section called ‘The Transcendental Exposition
of the Concept of Space’, which is located between the Metaphysical Exposition and
the transcendental idealist conclusions (a) and (b). This may be taken to suggest that
it is considerations about geometry that are supposed to take us from the claim that
space is the a priori form of our intuition to the conclusion that space is merely the a
priori form of our intuition, rather than a feature of reality as it is in itself. However, if
we pay attention to what Kant actually argues for in the Transcendental Exposition,
he simply presents an independent argument for the idea that our representation of
space is a priori and is an intuition.
In the Transcendental Exposition, Kant says that geometry provides us with
synthetic a priori cognition of space, and asks what the representation of space
must be for such cognition of it to be possible. His answer is that our representation
of space must be pure (that is, a priori) and must be an intuition, because ‘from a
mere concept no propositions can be drawn that go beyond the concept, which,
however, happens in geometry’ (B41). The main difference between the Transcen-
dental Exposition and Argument 3 in the first edition (which it replaces) is a
developed statement of Kant’s reason for thinking that our representation of space
must be an intuition. In contrast, Argument 3 in the A version argues only for the
need for our representation of space to be a priori (to explain the necessity of
geometrical claims). In the B version, Kant has more clearly brought out the idea
that synthetic a priori propositions require a priori intuition; geometry is synthetic a
priori knowledge of space, so if it is to explain how geometry is possible our
representation of space must be intuitive and a priori. Like the arguments in the
Metaphysical Exposition, the conclusion is that our representation of space is a priori
and is an intuition. This does not add a new argument that would show how we get
from the idea that our representation of space is a priori and is an intuition to Kant’s
conclusions (a) and (b).
At the end of the Transcendental Exposition, Kant asks: ‘how can an outer
intuition inhabit the mind that precedes the objects themselves?’ He answers: ‘Obvi-
ously not otherwise than insofar as it has its seat merely in the subject, as its formal
constitution for being affected by objects and thereby acquiring immediate repre-
sentation i.e., intuition of them, thus only as the form of outer sense in general’
(B41). Here we do have the move to ideality, but we do not have a new account of the
argument that is supposed to get us there. As in the first edition, Kant seems to take
the ideality of space to follow from the claim that the representation of space is an a
priori intuition. Nowhere in the Argument does Kant say that geometrical claims
provide conditions of the possibility of experience, and nowhere does he say that our
 MANIFEST REALITY
insight into geometrical claims can be explained by our minds’ being responsible for
the geometrical nature of objects. He simply says that an outer intuition that precedes
objects can only be something in the subject.
The simple argument we considered in the previous section reads Kant’s explan-
ation of our cognition of metaphysical synthetic a priori claims back into the
argument in the Aesthetic. It is relatively uncontested that Kant argues that we can
know conditions of the possibility of empirical knowledge to be true of objects if the
objects of our knowledge are appearances that do not exist independently of our
being able to have knowledge of them (this is his explanation of metaphysical
knowledge). The simple reading attributes to him a similar thought with respect to
geometry: the thought is that geometrical knowledge is knowledge of the conditions
our minds need for knowledge of objects, and it is because objects do not exist
independently of our minds that we are able to have this knowledge. The proponent
of the simple reading may argue that although in his statement of the argument Kant
does not actually say that the source of the justification of mathematical claims is that
our minds are responsible for the spatial structure of objects, attributing this thought
to him enables us to explain his argument. And since Kant says that metaphysical
cognition can be explained once we see that objects conform to our knowledge, it
might be thought not a stretch to bring a similar thought into the interpretation of
the Transcendental Exposition. However, Kant does not actually present this argu-
ment, and, as we have seen, it would be a very weak argument which would fail to
establish his general idealist conclusion. We would have reason to attribute this
argument to Kant by reading it into passages in which he does not actually present
it only if no other strategy were available to make sense of his argument. In my view,
there appears to be a step missing in Kant’s argument, and no alternative to the
simple argument as a way of closing the gap, only if we fail to pay attention to Kant’s
account of the nature of intuition, and the role of intuition in cognition.
IV How Geometry is Possible: What Kant

Does Not Argue in the Aesthetic
In this section I argue that Kant’s basic ‘how possible’ question in the Transcendental
Aesthetic does not concern how we justify synthetic a priori propositions, but rather
how they can have objects which are given to us. This is, in my view, the fundamental
reason Kant takes synthetic a priori cognition to be mysterious and in need of
explanation. It has implications for our understanding both of the argument for
idealism and of Kant’s explanation of the possibility of metaphysics.
A number of commentators have argued that the Transcendental Aesthetic needs
to be re-read in the light of considerations which are presented later.18 For example,
18
For example, Longuenesse (1993).
THE ARGUMENT FOR TRANSCENDENTAL IDEALISM 
Michael Friedman’s important work explains the distinctive role of our pure intu-
ition of space in geometry in terms of the role that constructions in pure intuition
play in geometrical inference (against the background of the logic available to Kant);
he argues that this needs to be understood in the light of the Transcendental
Analytic.19 This emphasis is useful when considering Kant’s account of geometrical
cognition as a whole, but it can distort our reading of the Aesthetic, in which Kant is
not trying to give a complete explanation of geometrical cognition as a whole. Ideally,
we should have a reading that enables the basic argument of the Transcendental
Aesthetic to function at the point at which it is presented. In the Aesthetic, Kant’s
discussion of geometry is concerned with one specific question: what must the
representation of space be for it to be possible that there is a science that determines
the properties of space synthetically yet a priori? Re-reading the Aesthetic in the light
of considerations presented later in the Critique informs our understanding of Kant’s
‘how possible’ question in more than one way: it enables us to understand how Kant
sees geometrical cognition as a whole, but, in addition, looking at the explanations of
geometrical cognition Kant develops in later parts of the Critique can clarify what he
is not concerned with in the Aesthetic.
In addition to what he argues in the Aesthetic, there are at least three further parts
to Kant’s account of geometry in the Critique. First, in the Axioms of Intuition, he is
concerned with the question of how mathematics has what he calls objective validity.
Second, in the Transcendental Deduction Kant argues that the transcendental unity
of apperception is a condition of all cognition, and therefore a fortiori of geometrical
cognition, and he uses this to explain how we manage to represent space as the
unified object studied by geometry. And third, in the Doctrine of Method, Kant
presents his account of how mathematical concepts are defined and what this implies
for how geometrical inference works. Since Kant presents his account of how
mathematics has objective validity, how we represent mathematics as the object of
geometrical cognition, and the nature of mathematical concepts and inferences much
later in the Critique, it is reasonable to conclude that these are not his concerns when
he asks how mathematics is possible in the Aesthetic.
In the Axioms of Intuition, Kant aims to demonstrate that geometry applies to the
actual objects of experience and is not a mere thought-game: that it has objective
validity and applies to the objects of experience. In the Aesthetic, he aims to explain
something much less: how geometry is possible.
Geometry is cognition and, as such, it involves concepts. Since the Aesthetic
abstracts from concepts, it clearly cannot give a complete account of geometrical
cognition; it is concerned only with one ingredient in cognition. As cognition, math-
ematics is subject to the conditions that all cognition is subject to—including the
19
Friedman (2012). His perspective is that of explaining how the faculties of sensibility and imagination
yield necessary and universal knowledge. This is an important question, but not one to which Kant
provides a complete answer in the Transcendental Aesthetic.
 MANIFEST REALITY
transcendental unity of apperception and the conditions of this unity. Kant’s explan-
ation of how geometry is possible in the Aesthetic is not the same as, and is presented
prior to, his complete account of how we have cognition of space as a unified object.
In the Doctrine of Method, Kant presents an account of how mathematical
concepts are defined and of the status of mathematical axioms, which he contrasts
with philosophy (A712/B74–A738/B766). His aim is to explain why, unlike math-
ematics, philosophy cannot start with definitions, axioms, and demonstrations. He
argues that synthetic a priori claims in geometry require that we go beyond the
concept ‘to properties that do not lie in this concept but still belong to it’ (A718/
B746). Kant explains how we do this with his account of the way mathematical
concepts are defined, which involves the ‘construction of concepts’, where to con-
struct a concept means ‘to exhibit a priori the intuition corresponding to it’ (A713/
B741).20 Explaining Kant’s idea of a construction that both exhibits its object and is
general in the way required to support the necessity of mathematical claims is difficult.
For our purposes here, we need simply to note that his account of geometrical concepts,
and their construction in intuition, is central to his account of geometrical inference
but is not part of the Aesthetic. This suggests that his ‘how possible’ question in the
Aesthetic does not concern the nature of geometrical inference, and therefore that
Kant’s concern is not with a justificatory gap. Further, this makes it clear that Kant does
not think that all synthetic a priori propositions are justified in the same way: the way in
which mathematical concepts are defined is, Kant thinks, specific to them, and cannot
explain how inference works in metaphysics.
So, paying attention to what Kant adds to his account of mathematical cognition in
later parts of the Critique tells us something about what he is not looking for in the
Aesthetic; this therefore clarifies the ‘how possible’ question that is presented and
answered there. He is not asking for a demonstration that mathematics is true of all
objects we can experience: this requires the arguments in the Principles. He is not
giving an account of how space becomes represented as the object that is the subject
of geometrical cognition: this requires bringing our representation of space under the
transcendental unity of apperception. Finally, he is not looking for an explanation of
the nature of geometrical inference, or of how mathematical claims can be justified:
this requires his account of mathematical concepts.
I suggest that Kant’s ‘how possible’ question in the Aesthetic is concerned with
something prior to all these questions and fundamental to Kant’s account of
cognition: how it is possible for mathematical claims to have (to be about) given
20
On Lisa Shabel’s account of this, ‘[m]athematical concepts are thus given through synthetic defin-
itions, which prescribe a rule or pattern for constructing a corresponding intuition. Geometric concepts in
particular provide us with the rule or pattern for constructing sensible intuitions of the spatial magnitudes
of objects of outer sense’ (2006: 99). Her account fits with my account of intuition as presentations to
consciousness: ‘the procedure and result of all mathematical construction is, for Kant, fundamentally
ostensive: to construct a mathematical concept one necessarily exhibits an intuition that displays its
features manifestly’ (2006: 101).
THE ARGUMENT FOR TRANSCENDENTAL IDEALISM 
objects—objects which are (or could be) present to consciousness. Kant thinks that
cognition requires given objects: objects with which we can have acquaintance. He
opens the Aesthetic with the claim that the representations through which we are
given objects are intuitions and that we humans need to be affected by objects in
order for them to be given to us in intuition (A19/B33). If we put these claims
together, we can immediately see why Kant thinks that cognition of synthetic a priori
propositions is mysterious, and why he does not take the existence of the mystery to
require explanation: we can have cognition only where objects can be given to us, but
the representations that give us objects involve objects affecting us, so are not a priori.
If their objects cannot be given to us, synthetic a priori claims will not qualify as
cognition. How cognition of such claims is even a possibility is therefore a mystery.
Such cognition seems to lack one of the ingredients it needs to be possible as
cognition: acquaintance with the objects of cognition. Thus, the first, and central,
problem for synthetic a priori cognition is not how we justify synthetic a priori
claims,21 but how such claims can have given objects: how they can be about anything
that is present to consciousness, and therefore can qualify as cognition.22
This understanding of Kant’s ‘how possible’ question readily fits with the answer
he presents to his question: a priori intuition. At the end of the Aesthetic, we are not
in a position to give a complete account of cognition or, a fortiori, of mathematical
cognition. We are not, therefore, in a position to explain how space becomes an
object of cognition, how geometrical inferences work, or how geometrical claims are
justified. Yet Kant says that we do have one part of the answer to the question of how
synthetic a priori knowledge is possible, and what we have is the idea of a priori
intuition. Since it is intuition, what it presents us with—its ‘object’—is immediately
given. Since it is a priori, it presents us with something independent of experience,
unlike ordinary (empirical) intuitions, which require their objects to affect us.
A priori intuition therefore explains how it is possible for synthetic a priori claims
to concern something given to us. Further, this explanation does not merely ‘earn
some favour as a plausible hypothesis’ (A46/B63); the only way objects are given to us
21
Asserting that synthetic a priori claims cannot be justified might also be thought to require further
explanation and argument, not least, against rationalists who think they can be justified.
22
See Brittan (1978; 2006: 231), who argues that the point of Kant’s notion of a priori intuition is not to
justify mathematical proofs, but to ensure that mathematical claims have objects, and therefore can be
objective: for Kant, ‘there is no objectivity without objects, there are no objects without reference, there is
no reference without intuition’ (Brittan 2006: 229). He argues that, for Kant, singular reference requires
intuition, and a priori intuition is required in mathematics to show how mathematical claims are possible,
i.e. to provide an account of how their subject terms manage to refer (Brittan 2006: 229). Arguing against
those who see the role of a priori intuition as evidential, or as necessary for proofs, Brittan says that ‘the
fundamental issue for Kant has little to do with “proof”. It is, rather, whether one can “determine” the
object of singular reference in mathematics in a purely conceptual or descriptive fashion’ (Brittan 2006:
228). Further, he argues that it is not clear that the argument in the Aesthetic depends on or gives support
to any particular body of geometrical knowledge, since the Transcendental Aesthetic seems to establish
only very general topological features of space, and the ‘metricization’ of space is established only in the
Analytic (Brittan 2006: 227).
 MANIFEST REALITY
is in intuition, and so the only way a priori claims can concern given objects is if we
have a priori intuition. A priori intuition is therefore required for cognition of
synthetic a priori claims to be possible. On this reading, the explanation of the
possibility of knowledge of synthetic a priori claims in geometry is not that such
objects are ‘made’ by the mind in a way which guarantees that they hold, but rather
that we have a priori intuition.23 As I shall argue, Kant then thinks that idealism
follows from his answer—from the notion of a priori intuition.
V Intuition and A Priori Intuition

In this section, I put together the pieces argued for so far in this chapter and in
Chapter 7. An interpretation that explains Kant’s presentation of his argument would
be one that shows why Kant thinks that it is in the nature of intuitions that (for us) an
intuition cannot both present a mind-independent thing and be a priori. It would
follow from such a reading of intuition that showing that our representation of space
is an a priori intuition just would be showing that it does not present us with things as
they are in themselves and is merely the form of our intuition.24 I argue that the
account of intuition presented in Chapter 7 provides this: it is precisely because he
holds that intuitions involve the presence of the objects they represent that Kant
thinks that an a priori intuition does not present us with something mind-
independent.
In the Metaphysical and Transcendental Expositions Kant argues for the claim
that our representations of space and time are a priori and intuitive, and he does not
give us any reasons for thinking that an a priori intuition could not represent
something mind-independent; this, I think, is what makes it seem as if his conclusion
does not follow at the point he thinks it does. However, he has presented us with
material before these sections—in particular, his account of what intuition is and the
role it plays in cognition—and the arguments in the expositions should be read with
this in mind.25 If we assume the reading of intuition I have given in Chapter 7 and
23
See also A178/B746. Van Cleve interprets the argument from geometry as saying truths about
geometry follow from what we can construct only if they do not describe things as they are in themselves
(2002: 36). This is not something Kant says in the Aesthetic (his account of construction in intuition
is presented towards the end of the Critique, in the Doctrine of Method). On my reading, Kant thinks
that truths about geometry can follow from what we can construct only if we have a priori intuition
(A713/B741).
24
This reading is suggested briefly by Daniel Warren (1998), who argues that ‘in general the mere fact
that a representation has its origin in us has no idealist consequences. But when that representation is an
intuition, then, according to Kant, it does—and that is because intuition, insofar as it is what Kant calls an
“immediate” representation, guarantees the presence of its subject’ (Warren 1998: 221). See also Setiya
(2004).
25
As Marcus Willaschek (1997) points out, Kant’s transcendental idealist conclusions appear to be
insufficiently justified if we expect them to follow from the Metaphysical Exposition alone, but Kant might,
not unreasonably, expect us to take the argument of the Metaphysical Exposition together with all the
previous material introduced in the Aesthetic. Willaschek argues that in the Aesthetic Kant introduces
THE ARGUMENT FOR TRANSCENDENTAL IDEALISM 
put it together with the arguments in the Metaphysical and Transcendental Expos-
itions, we can see why he takes his idealist conclusion to follow at the point at which he
does. As I understand him, Kant has an account of intuition according to which a
representation could not be a priori and an intuition and present a mind-independent
feature of reality. He then argues that our representation of space is an a priori
intuition.26 It follows that our representation of space does not present us with a
mind-independent feature of reality. If we grant his premisses, Kant’s conclusions will
follow at exactly the point that he presents them as following.
In Chapter 7, I argued that, for Kant, intuitions are representations which involve
the presence to consciousness of the things they represent. Here, I suggest that it is
because Kant sees intuitions in this way that he thinks that an a priori intuition does
not represent something mind-independent. As I understand it, Kant’s line of
thought is as follows:
(1) An intuition is a representation which presents its object to consciousness.
(2) What a priori intuition represents is present to consciousness—because it is
intuition. (1)
(3) Objects which are independent of us can be present to consciousness only
through affecting us.27
(4) An a priori intuition does not involve anything affecting us—because it is a
priori.
(5) Therefore what a priori intuition presents us with is not something which
exists independently of our representing it (or of our being able to represent
it). (2, 3, 4)
(6) Our representations of space and time are a priori intuitions.
(7) So our representations of the structure of space and time do not present us
with a mind-independent feature of reality: they are merely the forms of our
intuition and do not represent things as they are in themselves. (5, 6).
intuition as a kind of representation which is essentially caused by the object it represents. It follows from
this, he argues, that an a priori intuition could not represent a mind-independent feature of reality
(because, being a priori, it could not be caused by such a feature).
26
Actually, the Metaphysical Exposition is more complicated than this, since it wants to show not just
that our representation of space is an a priori intuition, but that space is the pure form of sensible intuition,
i.e. the a priori form that organises empirical intuition. Thus the argument proceeds not just by arguing
that the representation of space is a priori and is not a concept, but also by showing that space plays a
certain role: that of enabling us to be presented with empirical particulars in empirical intuition. Warren
(1998) argues that in the Metaphysical Exposition Kant is not trying to show that space is a condition of
representing distinct objects, but rather that our representation of space is a priori. While I agree with this,
it is entirely compatible with taking the claim that space is necessary in order to represent distinct
particulars to be a premiss of the argument. On this reading, Kant thinks that the structure in which we
represent distinct particulars must be a priori; he takes as a further premiss the claim that space is the way
we represent distinct particulars; it therefore follows that our representation of space is a priori. I discuss
this in more detail in Allais (2009: 409–12).
27
‘Us’ refers to finite creatures with sensible intuitions.
 MANIFEST REALITY
I take (1) to be part of Kant’s account of what intuition is, as I argued in Chapter 7.
I take (4) to be relatively uncontroversial, or at least uncontroversial as a claim about
what Kant would think, though I say more about it later. (2) follows from (1). (5)
follows from putting together (2), (3), and (4). (6) is the conclusion of the arguments
in the Metaphysical and Transcendental Expositions, and (7) follows from (5) and
(6); in other words, it follows from putting the conclusion of the Metaphysical and
Transcendental Expositions together with the account of intuition that Kant gives
before them. (7) corresponds to Kant’s idealist conclusions (a) and (b): ‘(a) Space
represents no property at all of any things in themselves nor any relation of them to
each other’; and ‘(b) Space is nothing other than merely the form of all appearances
of outer sense, i.e., the subjective condition of sensibility, under which alone outer
intuition is possible for us’ (A26).) Just as in Kant’s presentation of the argument, (7)
follows from (6). The claim that requires the most explanation is (3). My aim is
simply to make it plausible that Kant would have accepted it. If this claim is part of
his starting point, then once we put his assumptions together with his arguments in
the Metaphysical and Transcendental Expositions, his transcendental idealist con-
clusions, (a) and (b), follow at exactly the point at which he thinks they follow. On
my reading, it is because Kant thinks that intuitions involve their objects being
present to consciousness that he thinks that a priori intuition cannot present us
with a feature of mind-independent reality. This is why he takes it to be the case that
showing our representations of space and time to be a priori intuitions suffices to
show that they do not represent any property of things as they are in themselves.
In support of the idea that Kant’s argument should be understood in this way,
I note that in explaining his drawing of his transcendental idealist conclusions (a)
and (b), Kant does not appeal to the general impossibility of a priori cognition of
things in themselves, but rather appeals specifically to the idea that we cannot intuit
features of things a priori (A26/B42). Further, we can find him making precisely the
argument I am attributing to him, in the Prolegomena. He says that
An intuition is a representation of the sort which would depend immediately on the presence
of an object. It therefore seems impossible originally to intuit a priori, since the intuition would
have to occur without an object being present, either previously or now, to which it could refer,
and so it could not be an intuition. . . . There is therefore only one way possible for my intuition
to precede the actuality of the object and occur in an a priori cognition, namely if it contains
nothing else except the form of sensibility, which in me as subject precedes all actual impressions
through which I am affected by objects. (Proleg. 4: 282)
According to this argument, it is because intuitions involve the presence of their

objects that it seems prima facie impossible to intuit a priori. But Kant then explains,
what at first seems impossible is possible, provided that what the intuition presents is
something that is merely ‘in’ the subject and not a feature of something independent
of the subject’s representing it. Our claim (3) above, readily makes sense of this.
Notice that Kant’s argument here specifically concerns a feature of intuition, not a
THE ARGUMENT FOR TRANSCENDENTAL IDEALISM 
priori cognition in general (as the simple argument, with its concern for justification,
would require). In fact, intuition is contrasted in this respect to concepts. Kant says:
‘Concepts are indeed of the kind that we can quite well form some of them for
ourselves a priori . . . without our being in immediate relation to an object’ (Proleg. 4:
282). My suggestion is that it is because of this crucial difference between concepts
and intuitions—that intuitions do and concepts do not essentially involve the
presence to consciousness of their objects—that a priori intuition cannot present a
mind-independent feature of reality.
As I have presented Kant’s line of thought so far, a crucial claim is that the object
that an a priori intuition represents cannot be mind-independent. Explaining this
claim requires saying something about the fact that space and time (what a priori
intuition presents us with) are not objects in any obvious sense. They are certainly not
objects in the sense in which Kant will later go on to explain the notion of an object in
general (a causally unified subject of properties, which is in causal interactions with
other objects) in the Transcendental Deduction. When we talk of the a priori form of
intuition as presenting an object, ‘object’ should be understood in an extremely
general sense, according to which the term simply refers to whatever it is that a
priori intuition represents/presents. According to Kant, what it is that a priori
intuition presents us with is space and time. Space and time are the structures or
forms within which empirical intuitions are arranged, and which set limits on what
we can construct and on what we can perceive. Kant’s idea is that since the presence
to consciousness of this structural or formal aspect of our experience is independent
of anything affecting us, it does not present us with a mind-independent feature
of reality.
It might be objected that even if we grant that the structure that is present to us in
our a priori intuition is something mind-dependent, it would not follow that it
cannot represent a feature of mind-independent reality (this would generate a version
of the neglected alternative objection). In responding to this, what is crucial is the
immediacy of intuition: intuitions are not representations which point beyond
themselves to some object they picture or describe; they do not represent indirectly.
Rather, the object presented is a constituent of the intuition. Something that repre-
sented its object without its object’s being present would not be an intuition, but
rather a mere image. It is because Kant thinks that intuitions do not represent their
objects in this way that his conclusion follows. Space is not merely pictured in our a
priori intuition but it is actually present in it. Since this presentation is independent
of anything affecting us, Kant thinks that what is presented cannot be something that
exists independent of us. This is why he takes showing our representation of space to
be an a priori intuition to show that space represents no property at all of things in
themselves and is nothing other than merely the form of our intuition.
A worry about Kant’s argument is how he can be entitled both to assert that we
have no knowledge of things in themselves and to say that things in themselves are
not spatio-temporal. Both in relation to this and in relation to the neglected
 MANIFEST REALITY
alternative, it is important to notice that the claim that things in themselves are not
spatio-temporal is not exactly what Kant asserts in the Aesthetic (although he does
provide argument for it in the Antinomies).28 Rather, he says that space and time
represent no property of things in themselves or relations between things in them-
selves (‘Der Raum stellet gar keine Eigenschaft irgend einiger Dinger an sich, oder sie
in ihrem Verhältnis auf einander vor’ A26/B42). He says that we have representations
of space and time that play a fundamental role in our experience of objects, and that
these representations do not present us with mind-independent features of reality.
This is not a positive claim about the nature of things as they are in themselves (that
they are not spatio-temporal), but instead a claim about what is presented to us: our
representations of space and time do not present us with mind-independent features
of reality. It is compatible with this that there should be some structure in things as
they are in themselves; the point is that this structure is not what it is that is present to
us when we represent space and time. Here again, the idea that intuitions immedi-
ately present their objects makes sense of Kant’s conclusion: it enables us to say that
even if there were something like space and time in mind-independent reality, this
something would not be that of which our representation of space is a representation,
since it would not be that which is present to us in a priori intuition. The space that
we represent—the space that structures our experience of objects, the space that is
present to us—is not a feature of mind-independent reality, and it does not present us
with a feature of things as they are in themselves. Kant’s claim that we cannot know
things as they are in themselves follows from his view that we cannot have cognition
without something presented to us in intuition, together with the claim that the form
of our intuition does not present us with things as they are in themselves. It is not,
therefore, in tension with the claim that space does not present us with things in
themselves, but rather depends on it.
28
It might also be that Kant’s argument based on incongruent counterparts is supposed to establish this.
Incongruent counterparts, or ‘enantiomorphs’, are objects which are mirror images of each other: two
objects which have qualitatively identical intrinsic spatial relations among all of their parts, but which
could not be made to occupy the same part of space, such as left and right-handed gloves in three-
dimensional space. Kant was interested in incongruent counterparts throughout his critical writings, but
the purposes to which he put them developed and changed. In his 1768 essay ‘Concerning the Ultimate
Ground of the Differentiation of Directions in Space’, Kant argues that the existence of incongruent
counterparts proves the existence of absolute space, against the Leibnizian relational account, because the
difference between incongruent counterparts is not one which can be captured by a relational account of
space. Throughout his critical writings Kant continues to take the existence of incongruent counterparts to
show that space cannot be simply a system of relations, but he ceases to think that it establishes the
existence of absolute space, and takes it to show other conclusions. In his 1770 Inaugural Dissertation, Kant
takes incongruent counterparts to help establish the difference between intuition and concepts as ingre-
dients in cognition and our need for both of them, since, he claims, incongruity cannot be captured using
concepts alone, and ‘can only be apprehended by pure intuition’ (Inaugural Dissertation 2: 403; see
Buroker 1981: 74). In the Prolegomena, Kant again takes incongruent counterparts to support the idea
of intuitions and concepts as essential and distinct ingredients in cognition (Proleg 4: 285–6), but he also
goes beyond the argument of the Inaugural Dissertation, arguing that since something is represented in
intuition that could not be captured conceptually, the objects of the senses cannot be noumena in the
positive sense, but are merely intelligible entities which could be fully cognized by an understanding alone.
THE ARGUMENT FOR TRANSCENDENTAL IDEALISM 
My interpretation depends on attributing to Kant the thought that the only way an
object that exists independently of our representing it could be directly present to
consciousness would be through affecting us. An objection to this reading is that it
might seem to attribute to Kant an argument that begs the question against the
rationalists, who would not start with this assumption. However, nice as it would be
to be able to attribute to Kant an argument which reaches his conclusion while
starting only with premisses that a rationalist like Leibniz would grant, this does
not seem to me to be possible at many points. Among what are indisputably
Kant’s premisses are the claims that cognition requires both concepts and
intuitions and that concepts are essentially general and cannot give us objects.29
These are not premisses shared by the rationalists: Leibniz, for example, thinks
that concepts can uniquely individuate. Since at least some of what are indisput-
ably Kant’s premisses are not shared by the rationalists, it cannot be an over-
whelming objection to my interpretation that it involves attributing to him an
assumption which is not shared with the rationalists. The premiss I am attrib-
uting to Kant makes sense of his argument, and there is evidence that he holds it:
there is evidence that Kant thinks that objects which exist independently of our
representing them can come to be present to consciousness only by affecting us.
In the Transcendental Exposition, immediately after arguing that only if our
representation of space is a priori and an intuition can we explain how geomet-
rical cognition is possible, he says:
Now how can an outer intuition inhabit the mind that precedes the objects themselves, and in
which the concept of the latter can be determined a priori? Obviously not otherwise than
insofar as it has its seat merely in the subject, as its formal constitution for being affected by
objects and thereby acquiring immediate representation, i.e., intuition of them, thus only as
the form of outer sense in general. (B41)
29
Though this may be contested, it seems to me that Kant also assumes as a premiss that intuition
must have an a priori form. Similarly, he does not argue for the claim that representing things spatially is
the way we represent them as distinct; he simply states this. At the point at which he introduces the claim
that the form of intuition is a priori Kant simply states that the form in which sensations are structured
cannot be derived from sensation (A20/B34). It is not clear whether this is supposed to be his argument
for the claim or whether the argument is supposed to be given later on, perhaps in the Metaphysical
Exposition. Willaschek suggests that we understand this claim in the light of the arguments in the
Metaphysical Exposition. He argues that the claim that that in which alone sensation can be ordered
cannot be given in sensation is justified only if (1) no sensation organises itself and (2) there are global
ordering features which apply to all sensations (Willaschek 1997: 550). He argues that this is what Kant
intends to establish in the Metaphysical Exposition. For example, in the second argument of the
Metaphysical Exposition, Kant says that we can abstract what we know about bodies from what we
know of space, but we cannot abstract space from our representation of bodies. A problem with
Willaschek’s reading is that, on the face of it, the aim of the Metaphysical Exposition is to establish
that space coincides with the a priori form of our intuition, taking the claim that our intuition has an a
priori form to be already established. In other words, the arguments of the Metaphysical exposition are
directed towards the claim that our representation of space has global structuring features which
mean that it is an a priori form of our intuition, but not the prior claim that intuition must have an a
priori form.
 MANIFEST REALITY
Here, Kant does not say, as the simple argument would require, that the idea that our
minds determine objects’ geometrical properties explains our insight into these
properties. Rather, he says that an outer intuition can be a priori, and therefore
precede the objects themselves, only if it is something that is only in the subject.30
This is exactly the thought my argument requires.
There are more ways of responding to rationalists than by starting with premisses
they accept and showing that these premisses lead to a conclusion the rationalists do
not accept. Kant’s response could take the form of presenting a complete picture that
is explanatorily more powerful than the rationalists’. Kant is, of course, not alone in
thinking that it is mysterious how rational insight can be a source of knowledge of the
world; further, he does not simply deny this, but also has an account which explains
why there is a mystery. His explanation is that cognition requires acquaintance with
its objects, and he thinks that we have no account of how this can occur without
objects causally interacting with us. The rationalist may assume that we can have
knowledge of non-empirical objects (such as God and Cartesian souls) through mere
conceptual description; Kant denies this. If the rationalist agrees that acquaintance is
needed for cognition, they owe us an account of how it is possible for us to have the
acquaintance with non-empirical objects (objects which do not affect our senses). It
is not enough simply to seem to have rational insight; we need an account of how this
apparent insight succeeds in being about anything—how it connects with objects.
Kant thinks that we cannot have cognition of any objects, and a fortiori of non-
empirical objects, through mere conceptual description, and he thinks we cannot
explain how things that are independent of us could be present to consciousness
without their affecting our senses. As Kant sees it, the rationalist makes a priori
claims about objects that exist independently of our representing them but lacks an
explanation of how we achieve acquaintance with such objects.
30
The reading of the argument presented here has implications for the much debated issue of how
much Kant’s argument depends on his belief that the geometry of space is Euclidean and that this is
knowable a priori. My reading of the argument does not take Euclidean geometry to be a condition of the
possibility of experience, nor does it take our having specific (Euclidean) a priori knowledge of the
structure of space to be a condition of the possibility of experience. Rather, what is both a condition of
the possibility of experience and a condition of the possibility of geometry is that there is an a priori form of
our intuition. The argument for transcendental idealism is therefore compatible with but not dependent on
Kant’s belief about physical space being Euclidean. How Kant’s account might relate to what we now know
about non-Euclidean geometry is a complex question; I note briefly on two possibilities. First, he might
question whether the objects described by non-Euclidean geometries can really be presented to us and
constructed in intuition. If they cannot, Kant could take this as grounds for denying that such geometry is
fully cognition, as opposed to a kind of play with concepts. Alternatively, he might argue that in addition to
the a priori form of our intuition, we can in fact have other a priori, formal intuitions (non-conceptual
representations of structures which are presented to us independent of experience of objects), including an
a priori intuition corresponding to non-Euclidean geometrical concepts, which would enable us to perform
constructions in intuition corresponding to these concepts. Finally, it might be possible to combine either
of these options with the thought that the form of our intuition is in fact non-Euclidean (that our
representations of objects in experience are not Euclidean). It might be that there is one geometry that is
fully cognition (or that actually describes space), and this is the one that corresponds to the form of our
intuition (how we actually represent objects), but that Kant is wrong about which geometry this is.
THE ARGUMENT FOR TRANSCENDENTAL IDEALISM 
In my view, Kant starts with the claims that intuitions are representations that
involve the presence to consciousness of the objects they represent and that objects
that are independent of us become present to consciousness by affecting us. He takes
this to entail that an a priori intuition could not present us with a mind-independent
feature of reality. He then argues that our representation of space is a priori and is an
intuition. Given the previous assumptions, it follows, just as Kant takes it to, without
further argument, that ‘(a) Space represents no property at all of any things in
themselves nor any relation of them to each other’ and ‘(b) Space is nothing other
than merely the form of all appearances of outer sense, i.e., the subjective condition of
sensibility, under which alone outer intuition is possible for us’ (A26/B42). While, on
this account of the argument, some of Kant’s key premisses have been insufficiently
established (and may be rejected by the rationalist), the claims which are left
unargued for are exactly those for which Kant himself provides no argument, and
Kant’s conclusions follow from his assumptions at exactly the point at which he says
they do.
VI Implications for the Mind-Dependence of Space

I have presented an account of Kant’s argument for his idealism which exactly fits the
way he presents it and explains why he takes his conclusions to follow at the point at
which he does. On my reading, the argument turns on the claim that what an
intuition presents us with is present to consciousness, together with the claim that
an a priori intuition presents us with its object independently of anything affecting
us. I suggested that Kant thinks objects that exist independently of our representing
them get to be present to consciousness by affecting us, so he takes it that something
that is present to consciousness without anything affecting us does not exist inde-
pendently of our representing it. Since space and time are a priori intuitions, they do
not exist independently of our representing them. It might be thought that this
argument supports a phenomenalist account of transcendental idealism, because it
might be taken to imply that the object an a priori intuition presents exists merely in
consciousness. However, Kant’s account of the mind-dependence of space in fact fits
better with the essential manifestness account.
Our question is with which of two forms of mind-dependence are better supported
by Kant’s argumentative strategy. What intuition specifically gives us is the presence
to consciousness of its object; the argument for mind-dependence is specifically
based on the thought that something that is present to consciousness independent
of anything affecting us does not exist independent of the possibility of this presence.
What follows from this is a mind-dependence according to which the object pre-
sented does not exist independently of the possibility of its being present to con-
sciousness, not a mind-dependence according to which the object exists literally in
the mind. It is crucial to keep in mind here that something that is present to
consciousness is not thereby a merely mental state (a modification of an inner state
 MANIFEST REALITY
of a subject). Something that is outside us but does not exist independently of the
possibility of its being present to consciousness is dependent on our minds but not
merely in our minds. Essentially manifest qualities present us with things, but do not
present us with things as they are independently of their possible presence in
conscious experience. They are neither merely in individual subjects’ minds, as
modifications of subjects’ inner states, nor are they theoretical constructions out of
actual and possible inner states. They are mind-dependent relational qualities of
objects. To say that space is the form of intuition is to say that it is the form in which
what is presented to consciousness is arranged. To say that space is merely the form
of intuition is to say that it does not exist apart from the possibility of its being
presented to consciousness. This does not mean that it exists literally in minds, but
rather that there is nothing more to space (the space that we represent) than what can
be presented to us.
Not only does the essential manifestness account of the mind-dependence of space
follow from the argument, it also has an overall better fit with Kant’s account of
space. Neither a straightforward Berkeleyan account nor a sophisticated phenomen-
alist fits straightforwardly with Kant’s account of space.
Kant’s account is importantly different from Berkeley’s in the following respect:
Berkeley thinks space is constructed from features of ideas in us. Kant argues that our
representation of space could not arise in this way: it could not be abstracted from
our experience of particulars in space, because it is a requirement of perceiving
distinct particulars. This is one of Kant’s arguments for the apriority of our repre-
sentation of space (Argument 1 of the Metaphysical Exposition). For Kant, truths
about space are not just truths about what is actually in subjects’ minds, or about
what we have actually perceived, and neither are they truths about what can be
extrapolated from or constructed on the basis of what is in subjects’ minds. They are
truths about what could possibly be presented to subjects like us.
There are ways other than Berkeley’s of seeing our representation of space as a
phenomenalist construction; it could be seen as some kind of theoretical construction
that is necessary for experience of objects, rather than a construction drawn from our
experiences of empirical objects. The problem with attributing this kind of position
to Kant is that he precisely denies that our primary representation of space could be a
construction. He says that we represent space as an infinite, immediately given, single
structure that is present both in our experience of objects and when we do pure
mathematics. Our original representation of space is a structure or form that is
present to consciousness immediately, without boundaries (rather than being con-
structed out of bounded parts put together), that provides limits on what it is possible
for us to perceive and what it is possible for us to construct. Kant thinks that
constructing a representation of an infinite space would require first representing
parts of space and then running through and synthesising these parts. In contrast, he
holds that we represent parts of space by representing limits to the immediately given
singular infinite structure that is presented in intuition, and that representing the
THE ARGUMENT FOR TRANSCENDENTAL IDEALISM 
parts of space in this way is subsequent to an immediate representation of space as

one. The crucial point here is that Kant’s account of our representation of the infinity
of space is precisely not that we construct such a representation: the infinity of space
is immediately given, not constructed. Kant thinks that what an a priori intuition
presents is limited to what can possibly be present to our consciousness; he does not
argue that it is limited to what we can construct. On the contrary, he thinks that what
can be presented to us places limits on what we can construct. Thus, neither the
Berkeleyan account of space as a construction out of our ideas of empirical objects,
nor the idea of space as a theoretical construction of the sort a phenomenalist might
appeal to fits with Kant’s account of space as an intuition, and the way this features in
his argument for his idealism.
The essentially manifest view also fits with the way Kant talks about the role of
pure intuition in mathematics. Kant thinks that space is present to us both in our
experience of objects outside us and when we produce constructions in pure intuition
in mathematics.31 He does not say that the objects mathematical claims concern—the
objects presented to us in pure intuition—exist only as ideas in our minds. Rather, he
thinks that mathematical claims are limited to what it is possible for us to construct
in intuition (in a representation with which we have acquaintance), and that the
limits on what we can construct are given by the form of our intuition. It is crucial to
Kant’s account that a construction in intuition is not a theoretical (conceptual)
construction: it essentially involves an object that is present (exhibited, manifestly
displayed). The objects of mathematical claims can be presented to us, Kant thinks,
because we have a pure intuition that is present to us, and he thinks that mathem-
atical truth is limited to what can be presented in pure intuition. Like truths about
space, Kant thinks that mathematical truth does not go beyond what can possibly be
constructed in intuition, and this means that it is limited to something that can be
manifested to consciousness. Again, this fits with the idea that Kant’s idealism is a
denial of experience-transcendence: that the objects of our knowledge are essentially
things that can be manifest to us.
So far I have considered Kant’s direct argument for transcendental idealism in the
Aesthetic; the account I have presented fits well with his indirect argument in the
Antinomies. In the Antinomies, Kant argues that if we assume that our knowledge is
of things in themselves (things which are entirely independent of us and ontologically
fundamental and self-sufficient), we will find that when we try to think about the
extent of the world in space and time we are unable to avoid running into contra-
dictions. We will find ourselves having conclusive reasons to assert that the world has
a beginning (or boundaries) in space and time and that it has no boundaries but
rather is infinite in space and time, as well as equally good reasons to assert that the
31
However, the latter depends on the former for its ‘objective validity’. It is only because we actually
have experience of spatio-temporal objects that the form of intuition genuinely presents us with something,
and is not a mere thought-entity.
 MANIFEST REALITY
matter contains simple indivisible parts, and that matter is infinitely complex. Kant
argues that his transcendental idealism enables him to avoid these contradictions: he
says that what makes avoiding the contradictions possible is seeing that the spatio-
temporal objects of our knowledge are not metaphysically fundamental (are not
things as they are in themselves) and are only mind-dependent appearances which do
not exist apart from the extent of possible experience. His explanation requires both
that appearances are mind-dependent and that there is a way things are in themselves
(something metaphysically fundamental). Further, Kant’s explanation postulates a
specific kind of mind-dependence: one which limits the extent of spatio-temporal
objects to what can be present to or given in our experience—what can be given to us
in intuition. Kant says that we can avoid contradiction if we see that the world in
space and time is neither infinite nor finite—it does not have a definite extent, and he
invokes his specific form of mind-dependence to explain how this can be possible.32
Crucially, to explain the world’s not having a determinate extent he does not appeal
to the claim that an infinite world in space could not be constructed; it is hard to see
why an infinite world in space and time could not be constructed. This means that
seeing space as a phenomenalist construction would not explain why the world in
space and time does not have a determinate extent. Rather, he says that an infinite
world in space could not be given to us; it is not something that could be present in
experience. Kant thinks that neither an infinite extent of the world in space nor a
boundary to space can be given to us; since the world in space is limited to what can
be given to us, it does not have a determinate extent. It is specifically a kind of mind-
dependence that limits truths about space and time, and about the world in space and
time, to what can be given to us in intuition that enables us to explain how it could be
that the world does not have a determinate extent in space.
Conclusion to Part Two

Part One showed that making sense of Kant’s idealism requires a form of mind-
dependence that does not involve existence literally in the mind. In this part of the
book I have presented an account of mind-dependence—essential manifestness—
which fits extremely well with the way Kant presents his idealism, his account of the
relation between appearances and things in themselves, his account of the role of
intuition in cognition, and his argument for his idealism. On this reading, Kant’s
idealism is not fundamentally about concepts carving up the world, nor about minds
shaping objects in such a way that we can justify synthetic a priori claims, but about
givenness. I complete my account of the role of concepts in Kant’s position, and of his
explanation of the possibility of metaphysics, in the final part of the book.
32
A482–3/B510–11; A483/B511; A492/B521; A498–9/B526–7; A505/B533; A517/B545.
PART THREE
Kant’s Idealism and his Realism
9
Relational Appearances
I Introduction to Part Three

In Part One I presented textual and philosophical evidence for the claim that Kant
holds spatio-temporal appearances to be mind-dependent and also holds that things
have natures in themselves that ground their mind-dependent appearances. I also
showed how the (unstable) extremes of interpretation in the literature are responses
to legitimate textual concerns, but neither is able to do the text justice. In Part Two
I explained Kant’s claim that spatio-temporal reality does not extend beyond our
possible experience of it in terms of the claim that it is essentially manifestable (rather
than existing in the mind). Kant holds that something is empirically real only if it is
possible for it to be presented to us in an intuition: if it is possible for us to have
acquaintance with it. Kant also thinks that spatio-temporal reality is incomplete in
a way which could make sense only if it were not metaphysically fundamental,
and that it is grounded in something mind-independent which we do not cognize.
The position I attribute to Kant involves a complex combination of realism and
idealism, metaphysical and epistemological concerns; this is the subject of the final
part of the book.
Kant’s idealism is bound up with his realism in two central ways. One, Kant says he
combines transcendental idealism with empirical realism: empirically real objects are
both genuinely mind-dependent and also exist outside us in space. Two, Kant’s
account of the mind-dependent nature of empirical reality requires that there is
something that exists in itself, entirely independent of us. On the one hand, he thinks
that we can characterise the objective world that is the object of study of science with
essentially manifest qualities only—without cognition of any features of reality that
are non-sensible and that exist independent of any relation to minds like ours. On the
other hand, he thinks that we cannot have a complete, coherent ontology without
being committed to there being something entirely independent of us, and that the
way we think about the mind-dependence of appearances requires this. Empirical
reality is not ontologically fundamental, is essentially incomplete, and is entirely
relational; Kant does not think it makes sense to think that reality per se could be like
this. This combination of idealism and realism is my subject in the next two chapters.
The current chapter shows how the particular kind of mind-dependence Kant
ascribes to empirical reality is compatible with a robust empirical realism that sees
 KANT ’ S IDEALISM AND HIS REALISM
appearances as the objects of study of science. I do this by examining the similarities

(and differences) between Kant’s account and three contemporary positions: mean-
ing-theoretic anti-realism, anti-realism in philosophy of science, and structural
realism in philosophy of science. These comparisons also enable us to see why
Kant thinks this kind of mind-dependence requires things in themselves that ground
appearances. Kant’s commitment to things in themselves grounding appearances is
the subject of Chapter 10. I argue that Kant takes his account of empirical reality to
require the existence of intrinsic natures that ground appearances.
The final way I explore the combination of idealism and realism in Kant’s position
is by examining his argument in the Transcendental Deduction of the categories
(Chapter 11) and the implications of this for his explanation of the possibility of
metaphysics (Chapter 12). The Deduction is sometimes taken as the centre of the
Critique (this is suggested by Kant himself (Axvi–i)). Kant’s aim in the Deduction is
to vindicate our use of the a priori concepts he calls the categories, by showing that
they are conditions of the possibility of empirical cognition. The Deduction can be
(and has been) taken to be central to Kant’s idealism in a number of ways. Kant’s
emphasis on the role of a priori concepts in experience can be understood as
expressing a kind of conceptual idealism which holds that reality is divided up by
our concepts. Kant’s claim that the categories are necessary for us to have ‘relation to
an object’ or for anything ‘to be an object for me’ can be thought to be expressing an
idealist thought that what counts as an object depends on our concepts. Against this,
I argue, in Chapter 11, that the central argument in the Deduction is epistemological;
this, in my view, partly explains the appeal of deflationary interpretations that see
Kant’s central concerns as epistemological. Kant does have epistemological concerns
at the centre of the Critique.
Although I read the central argument in the Deduction as epistemological,
Kant’s final aim in this section is closely related to his idealism, since his argument
for the claim that all spatio-temporal objects are subject to the categories requires
his idealism. The role the idealism plays here is central to his explanation of the
kind metaphysics he thinks is possible for us, which is my subject in Chapter 12.
Understanding the delicate relation between Kant’s idealism and his explanation of
metaphysics requires seeing the different roles intuitions and concepts play in the
account. As I read him, Kant’s proof that the categories apply to spatio-temporal
objects does not depend on showing that they are necessary for us to be presented
with perceptual particulars (to be given objects in intuition) and does not depend
on showing that they are necessary for us to have relation to an object (referential
thoughts). Rather, the conclusion of the epistemological argument that shows
that the categories are conditions of the possibility of empirical concept application
(and therefore necessary for thought to have relation to an object) can be
extended to all spatio-temporal objects because the idealism already established in
the Aesthetic says that spatio-temporal objects are dependent on the possibility of our
cognizing them.
RELATIONAL APPEARANCES 
This chapter is concerned with showing how Kant’s particular account of the
mind-dependence of appearances is compatible with a robust empirical realism and
also requires the existence of a mind-independent reality that grounds appearances.
In Section II, I look at Kant’s position in the light of a version of anti-realism that
rejects experience-transcendence, and in Section III, I compare it to a form of anti-
realism in philosophy of science that denies that we have knowledge of unobservable
features of reality. These comparisons enable us to show how limiting our knowledge
to qualities that can be presented in experience can be compatible with thinking of
the world as objective: the compatibility of empirical realism and transcendental
idealism. In Section IV, I compare Kant’s account of appearances with a different
contemporary position in philosophy of science: structural realist positions which,
like Kant, hold that science gives us knowledge only of relational features of reality. In
Section V, I argue that Kant’s structural realism about empirical reality is connected
to his idealism, but also potentially separable, and that it provides him with inde-
pendent reason for thinking there must be a way things are in themselves.
II Essentially Sensory Qualities and an Objective

world: The Rejection of Experience-Transcendence
As we have seen, Kant suggests that his idealism can be understood by comparing the
status of the spatial qualities of objects of which we have empirical knowledge with
the mind-dependent status of so-called secondary qualities like colour. In his pres-
entation of the analogy, Kant explains that transcendental idealism includes ‘among
mere appearances the remaining qualities of bodies, which are called primarias:
extension, place, and more generally space along with everything that depends on
it (impenetrability or materiality, shape, etc)’ (Proleg. 4: 289). However, many
philosophers think that we cannot characterise or conceive of an objective world
with secondary qualities alone. Locke, for example, opposes primary qualities to
qualities which are understood in essentially sensory terms and thinks that we need
primary qualities to make sense of there being objects (an objective world). Kant, in
contrast, thinks that qualities can be (in a sense) essentially sensory, yet still play
some of the central roles for which Lockean primaries are thought to be needed in
giving us a conception of an objective world: featuring in the causal explanations that
give us our idea of what it is for a material object to exist, featuring in causal
explanations in science, and enabling us to conceive of existence unperceived. On
Kant’s account we can account for causal explanation at the level of experience and in
science, as well as for existence unperceived, with essentially manifest qualities only.
An account of secondary qualities which identifies them with experiences of
subjects, or ideas that exist only inside subjects, will, arguably, not enable them to
play the role of thinking about existence unperceived, if thinking of objects as existing
unperceived requires thinking of them in terms of qualities that they have when they
 KANT ’ S IDEALISM AND HIS REALISM
are not being perceived.1 Some arguments for the secondary status of qualities like
colour are based on the claim that such qualities do not feature in causal
explanations—including causal explanations of visual experiences of colours. On
these views, the very reasons we have for thinking that colours are mind-dependent
are reasons for thinking that not all qualities of objects could be like this and for
thinking that we could not give an account of an objective world in terms of
secondary qualities only. This provides a good reason not to use such accounts of
the secondary status of colour to explain Kant’s analogy. In contrast, I argue that this
problem does not arise for the essentially manifest view.
Manifest qualities are qualities of objects that can be presented to us in experience;
essentially manifest qualities are qualities whose existence is not separable from the
possibility of their being presented to us in experience. However, this does not mean
that they exist only in particular events of being perceived, much less as states in
subjects. Although objects do not have essentially manifest qualities independently of
the possibility of our perceiving them, like manifest qualities generally, objects have
their essentially manifest qualities even when we are not currently experiencing
them. We can explain our perceiving such a quality in terms of the object’s having
the quality, together with our being suitably situated and suitably receptive. This
enables us to think about objects as existing outside us and distinct from us (of
thinking of an objective world) even though the qualities through which we under-
stand and engage with the objects cannot be divorced from a relation to us.
In addition to the fact that essentially manifest qualities can exist unperceived, they
can also feature in causal explanations. Unlike error theories such as the subjectivist
dispositional view of colour, and like primitivist objectivist views, the essentially
manifest view allows colour to feature in ordinary causal explanations at the level of
experience and causal explanations in the sciences, and it allows that we can
manipulate colours. When we say, for example, that the pink paint was produced
by mixing white paint with red paint, we are giving a causal explanation, involving
objects in the world and the ways their properties interact with each other. Similarly,
explanations within biology may tell us, for example, that animals use the colours of
objects as ways of discriminating, for example, when fruit is ripe. Unlike the
primitivist objectivist, the essentially manifest view of colour holds that the properties
that we manipulate when we manipulate colours are features of the way things look,
and the causal explanations in which colours feature involve qualities which can be
understood only in terms of things’ looking certain ways to conscious subjects.
It is crucial to see that essentially manifest qualities are qualities of objects. If it
were thought that truths about colour were truths about the experiences of subjects,
generalising the status of colour to all properties that we experience would threaten
causal explanation and empirical knowledge. However, the essentially manifest view
1
See Strawson (1963), Evans (1985a), and Campbell (1997).
RELATIONAL APPEARANCES 
of colour does not hold that claims about colours are claims about the experiences of
subjects; they are about colours. To say that mixing red paint with white paint
produces pink paint is to say something about properties of things in the world,
how they interact, and ways in which they can be manipulated. The essentially
manifest view, however, holds that the properties referred to in the claim can be
grasped only from the point of view of visual experience and that it does not make
sense to attribute them to objects as they are independently of the possibility of their
perceptually appearing to subjects like us. I return to the issue of causal explanation
in the next section.
How exactly to formulate the mind-dependence of essentially manifest qualities—
the idea that they have an essential relation to visual experience—is a tricky question.
In Chapters 5 and 6, I suggested that one way to express this is in terms of the denial
of experience-transcendence: the thought is that as essentially visual qualities, colours
do not have natures which have more to them than can be presented in experience;
they do not have hidden, non-visible natures. Further, this is not a contingent fact
about colour. Kant’s idealism has in fact been compared with versions of anti-realism
which deny the existence of verification-transcendent reality.2 Such forms of anti-
realism are different from phenomenalist forms of idealism in that they see objects as
existing outside us, whether we experience them or not, just as realists do. They
therefore account for our understanding of physical objects, of causal explanation in
science, and of existence unperceived in the same way realists do. However, the anti-
realists add to this the idea that objects have nothing more to them than what it is
possible for us to know of them (or warrantedly assert about them). The notion of
verification-transcendent reality is the notion of a state of affairs for which we could
have no evidence, or no reason for thinking that it was the case. The claim that there
is no verification-transcendent reality amounts to saying that there is nothing of
which we could not, in principle, have knowledge (or evidence, or grounds for
asserting, etc.).3 The realism to which this view is opposed involves a commitment
to a notion of truth under which a statement can be true even if we could not know
this. What is crucial here is that the anti-realist does not see objects as existing merely
in mental states, but rather is committed to the same external objects the realist
picture involves. The anti-realist simply denies that this reality could contain any-
thing of which we could not have evidence (or experience, or warrant for believing,
depending on the specific version of anti-realism).4 Kant’s position is importantly
different from this kind of anti-realism in some respects, but there are similarities
2
This comparison has been suggested by a number of philosophers. See Stevenson (1983), Walker
(1983), Posy (1984; 1983), Putnam (1987; 1981), Rogerson (1993), Van Cleve (1999), Moran (2000),
Kroon (2001), and Abela (2002) for rejection of the comparison.
3
Another way of expressing the point is to say that facts are evidence-dependent (Van Cleve 1999): for
any fact there is something which could, in principle, be recognised by us as evidence for it or would give us
grounds for thinking that it is the case.
4
See Wright (1996: 119).
 KANT ’ S IDEALISM AND HIS REALISM
between the positions which make comparison between them helpful in explaining
Kant’s account of empirical reality. Kant’s idealism about empirical reality can be
expressed as the denial of experience-transcendence. Kant has a particular account of
this, because of his notion of experience: spatio-temporal reality does not transcend
what can possibly be presented to us in an intuition.
Because of its association with Michael Dummett and Crispin Wright, and in
order to distinguish it from the very different kind of anti-realism which features in
philosophy of science, I will call the kind of anti-realism that rejects verification-
transcendence British anti-realism. British anti-realism is primarily a theory of
meaning or a theory of truth which denies that we could have a genuine notion of
truth for statements that goes beyond the grounds we could give for asserting them. It
is associated with two central claims: that to grasp an assertoric sentence is to know
under what conditions it would be warrantedly assertable and that there is no
verification-transcendent truth, or verification-transcendent reality.5 One way the
latter claim is understood in terms of the rejection of bivalence: where there is no way
for us to know which of two contradictory statements is true, there is no fact of the
matter as to which is true.
There are two obvious respects in which Kant’s position is radically different from
British anti-realism: first, transcendental idealism is not a theory of meaning;6 and
second, Kant is committed to there being a way things are in themselves. Some of
those commentators who see Kant’s idealism as a form of anti-realism understand it
as a theory of meaning7 (transcendental idealism has also been seen as, or as closely
connected to, a theory of meaning of a more straightforwardly verificationist type8).
At least part of the basis for such interpretations is Kant’s saying that concepts with
respect to which we cannot experience an instantiation are empty, or have merely
logical significance, and lack content or significance in some sense. For example, he
says that
If a cognition is to have objective reality, i.e., to be related to an object, and is to have
significance and sense in that object, the object must be able to be given in some way. Without
that the concepts are empty, and through them one has, to be sure, thought but not in fact
cognized anything through this thinking, but rather merely played with representations.
(A155–6/B194–5)9
The sense in which Kant holds concepts without intuitions to be empty and lacking
in sense (Sinn) and significance (Bedeutung) is a difficult issue, but while Kant
5
See, for example, Dummett (1978c; 1978d; 1991; 1993) and Wright (1980; 1992; 1996).
6 7
See Van Cleve (1999) and Abela (2002: Ch. 4). Posy (1983; 1984).
8
Bennett (1966) and perhaps P. F. Strawson (1966). P. F. Strawson sometimes formulates what he calls
Kant’s principle of significance as a claim about meaning: ‘the principle that there can be no legitimate, or
even meaningful, employment of ideas or concepts which does not relate them to empirical or experiential
conditions of their application’ (P. F. Strawson 1966: 16).
9
See also Bxxv–xxvi, A51/B75, A90/B122, B148–9, A139/B178, A146/B186, A219–B266–7, B288–9.
RELATIONAL APPEARANCES 
certainly thinks that there is something problematic about the use of concepts whose
instantiations we cannot be given in experience, the problem cannot be that they are
straightforwardly meaningless. The claims of traditional transcendent metaphysics
are never regarded by him as being meaningless: his discussion about which side of
the Antinomies we would prefer to fight on if we were to choose purely on the basis of
interest clearly assumes that neither side is lacking in sense (A466/B494–A474/
B502), and one of his aims in the Third Antinomy is to demonstrate that the idea
of libertarian freedom is coherent despite the fact that it is not something we could
cognize.10 Further, some of the (verification-transcendent) ideas of transcendent
metaphysics play a key role in Kant’s ethics (belief in God, an immortal soul, and,
most importantly, freedom of the will), which they could not do if they were
meaningless.11 Kant thinks both that things in themselves are entirely verification-
transcendent and that we can talk meaningfully about them, so, unlike British anti-
realism, transcendental idealism is not a theory of meaning.
The second (closely related) difference is that British anti-realism denies the
existence of verification-transcendent reality; in contrast, this is exactly what Kant
asserts with his commitment to things in themselves. In this respect, Kant’s position
is further away from anti-realism than is the kind of realism to which anti-realism is
standardly opposed. The typical realism to which the British anti-realist is opposed
simply holds it to be the case that we have no guarantee that we can have knowledge
of everything that exists; Kant thinks that we have a guarantee that we cannot have
knowledge of all of reality. However, this does not mean that the anti-realist rejection
of verification-transcendence could not be helpful for understanding part of Kant’s
position. Kant is an idealist about appearances, so although his position as a whole
cannot be like anti-realism, the comparison with British anti-realism is helpful in
understanding part of his position—his idealism about things as they appear to us.
Kant does not agree, in general, with the claim that we can, in principle, have
experience of everything that exists, but he thinks that this claim is true of things
as they appear to us. The empirically real world does not extend beyond our possible
experience: if something is (empirically) real, we could have experience of it.
A third difference from British anti-realism follows from the fact that transcen-
dental idealism is not a theory of meaning: Kant’s reason for rejecting experience-
transcendence is different to that of the British anti-realists. But however it is
argued for (whether or not it is based on a theory of meaning), the rejection
of verification-transcendence (or of experience-transcendence) can give us a way of
characterising a kind of mind-dependence.12 The reason the British anti-realist
10
He says that transcendent ideas ‘serve, in part, to restrain the understanding’s arrogant claims,
namely, that (since it can state a priori the conditions of the possibility of all things it can cognize) it has
thereby circumscribed the area within which all things are possible’ (CJ:168, also B167).
11
And Kant thinks there are ways we can cognize these things through practical reason.
12
Whether the rejection of verification-transcendence or experience-transcendence is necessarily a
form of idealism is contested in debates around British anti-realism. Wright calls himself an anti-realist
 KANT ’ S IDEALISM AND HIS REALISM
denial of verification-transcendence is thought to express a relation of mind-dependence

is that it asserts a necessary connection between what exists and what can be known
or justifiably asserted by minds like ours. What is important here is the order of
dependence13 and whether the denial of experience-transcendence is thought to be
contingent. For example, while a primitivist objectivist about colour might hold that
it contingently happens to be the case that colours are accessible to us only through
sight; the essentially manifest view holds that it is essential to what colours are that
they are visual qualities, or qualities which could be seen by us. On this view, the
existence of colour is not independent of the possibility of its being present in a
possible visual experience. This means that the existence of colour is not independent
of the possibility of conscious experience, so colour is a mind-dependent feature of
objects (although it is not something that exists only in minds or only in particular
events of being perceived). This view of colour can be argued for on grounds other
than a theory of meaning. For example, someone might hold that colour is an
essentially manifest property on the basis of the claim that we cannot, in principle,
access colour through more than one sensory modality and the claim that there are
truths about colour relations that can be understood only from the point of view of
visual experience, and therefore that colour is essentially a feature of how things look.
Here, we do not work from a theory of meaning which leads to the rejection of
experience-transcendence. Rather, we start with the idea that colour is mind-
dependent and introduce the idea of rejecting experience-transcendence to capture
the particular kind of mind-dependence at issue. This, I suggest, is a helpful way of
thinking about Kantian appearances.
Anti-realists who link what exists to what we can verify (or can have warrant for
asserting, etc.) need to give an account of what counts as verification (or warrant).
This gives rise to different versions of anti-realism. Kant’s central idea is not
verification-transcendence (his concern is not with what we can justifiably assert)
but experience-transcendence and, as we saw in Chapter 6, Kant has a specific
account of what it takes for something to be part of possible experience. First, it
must be causally connected to something that is actually present to us in perception.
Second, crucially, it must be something which could, in principle, be given in or
present in a possible perceptual experience—something which could be manifest to
us, or with which we could be directly acquainted. It must be possible for it to be
given to us in an empirical intuition. Without the possibility of being experienced in
this sense, something is not part of empirical reality. So the sense in which empirical
reality is not experience-transcendent is that everything that is empirically real is
and denies that his position is idealist (Wright 1996: 54–5). Van Cleve argues that the view that facts are
evidence-dependent is idealist if it is understood as saying that nothing is ever true unless there is a truth-
maker that someone actually recognises. On the other hand, he thinks the view is not idealist if it requires
only that truth-makers be recognisable—not that anyone actually has recognised them (Van Cleve 1999:
219). For detailed argument that British anti-realism is a form of idealism, see Walker (1995).
13
See Wright (1992: 86–7), Campbell (1994: 216).
RELATIONAL APPEARANCES 
something which could, in principle, be given to or present in a conscious experience

of subjects like us. Kant says that the existence of distant stars to be encountered in
‘world-space’ just means that ‘in the course of an empirical progress in space I could
encounter them’ (A496/B524; A494–5/B522–3).14
Understanding Kant’s idealism about appearances in terms of the rejection of
experience-transcendence is strongly supported by the way he resolves the Anti-
nomies. Kant thinks that reason inevitably leads us into contradiction when we try
to think about the extent of the world in space and time and about the divisibility
of matter, and he argues that transcendental idealism enables us to avoid these
contradictions. Kant’s solution to the Antinomies—and, in particular, the first two
Antinomies—is crucial for understanding his idealism: the claim that his account
dissolves the Antinomies is one of his arguments for his position. We thus need to see
what kind of mind-dependence would play the explanatory role that Kant thinks his
idealism plays in dissolving the Antinomies. In Chapter 4 Section III, I argued that
the way Kant invokes transcendental idealism in his resolution of the Antinomies
supports the claim that his position is metaphysical in two respects: it supports
thinking that appearances are mind-dependent; and it supports thinking that appear-
ances do not exhaust reality but rather are based on something more fundamental
(things as they are in themselves). Kant says that it is precisely because transcenden-
tal idealism sees appearances as mind-dependent and as not ontologically fun-
damental that it enables us to avoid being committed to contradictory claims
about the extent of the world in space and time. I argued that this shows that
deflationary accounts which do not see Kant as having these metaphysical
commitments will not be able to give an account of how the position is able to
do the work for which he invokes it in resolving the Antinomies. Crucially, it is a
very specific kind of mind-dependence that is needed to do this work: the idea
of rejecting experience-transcendence. Kant says that where a claim transcends
the conditions of possible experience it cannot be true of appearances, and he
says that the fact that the existence of the world in space and time does not go
beyond what it is possible for us to cognize of it is what enables us to resolve the
Antinomies.
In resolving the first Antinomy, Kant wants to explain how apparently contradict-
ory claims about the size of the world could both be false. He says that this is possible
if the claims are asserted of something which does not exist as a determinate totality
(A505/B533): we can make sense of saying both that it is false that the world has
infinite extent in space and time and that it is false that the world has a finite extent in
space and time if the empirically real world does not have a determinate extent
14
In arguing for his realist interpretation, Paul Abela claims that Kant’s position does not involve mind-
dependence because it does not appeal to what we actually will experience (Abela 2002: 243–4; A496/
B524). However, Kant asserts that it is essential to something’s being empirically real that it could be
experienced: it is the fact that it could be experienced that makes it count as part of what is empirically real.
 KANT ’ S IDEALISM AND HIS REALISM
in space and time.15 But this is, on the face of it, a puzzling claim, so the next step is to
explain how it is possible for the spatio-temporal world to exist without having a
determinate extent—to exist without existing as a determinate totality. Kant says that
what enables us to make sense of this is the mind-dependence of appearances: it is
precisely because the things that we experience are appearances (which, ‘as mere
representations are not given at all if I do not achieve acquaintance with them’)
whose existence depends on the progress of possible experience, rather than things in
themselves, that we can explain how it can be that they can fail to have a determinate
extent in space and time. It is ‘because the world does not exist at all (independently
of the regressive series of my representations), it exists neither as an in itself infinite
whole nor as an in itself finite whole’ (A505/B533). Crucially, the kind of mind-
dependence Kant appeals to here is the rejection of experience-transcendence. The
world in space and time does not extend beyond the possibility of our cognizing it;
the world as a determinate totality (whether an infinite or a finite totality) is not
something we could cognize; therefore the world in space and time does not exist as
either a finite or an infinite totality.
Asserting that the world does not exist as a determinate totality is similar to a
central part of British anti-realism: the rejection of bivalence. One way in which
British anti-realism is formulated is precisely in terms of the rejection of the idea that
every claim must have a determinate truth value. However, Kant’s position is not best
understood as a rejection of bivalence. He does not deny that each opposed propos-
ition (‘the world is finite in space and time’ and ‘the world is infinite in space and
time’) is either true or false, but rather says that for the mathematical Antinomies
both theses and antitheses are false (of the world of appearances), and for the
dynamical Antinomies both are true, one of the world of appearances and one of
the world as it is in itself (and with the third and fourth Antinomies, that both could
be true).16 There is no reason to think that Kant rejects bivalence generally, and in
fact the reverse is more plausible (see A502–4/B530–2 and B601; see also Abela 2002:
Ch. 4). Despite this, he does, like the anti-realist, deny that the spatio-temporal world
15
Another place where Kant rejects the idea that the world of appearances exists as a determinate
totality is in his account of systematic unity as a goal of science. He thinks that the goal of a complete,
unified science is necessary to drive empirical science (A651/B679; A569–70/B597–8), and also that it is, in
principle, unachievable. He says that the ideal of systematic unity is ‘only an idea (focus imaginarius)—i.e.,
a point from which the concepts of the understanding do not really proceed, since it lies entirely outside the
bounds of possible experience’ (A644/B672; see also A644/B672 and A669/B697–A704/B732). Kant says
that ‘reason cannot think this systematic unity in any way other than by giving its idea an object’ but that
‘such an object cannot be given through any experience’ and ‘is a mere idea, and is therefore not assumed
absolutely and in itself as something actual, but rather is taken as a ground only problematically’ (A681/
B709; A648/B676; A671/B699; A647/B675).
16
Further, while Dummett holds that where we could not have grounds for asserting one of two
contradictory statements, neither is true, he also thinks that no statement could be recognised as
determinately neither true nor false—in other words, we could not have a determinate example of the
failure of bivalence (Dummett 1978e: 35). Kant, on the other hand, clearly identifies a number of pairs of
contradictory statements both of which he thinks are unknowable and not true of appearances.
RELATIONAL APPEARANCES 
could transcend what we can experience or cognize of it, and he denies that it exists as
a determinate totality. This denial—this formulation of mind-dependence—is pre-
cisely what he takes to enable him to resolve the Antinomies. The role this plays in his
resolution of the Antinomies supports understanding his idealism as the rejection of
experience-transcendence, (rather than as a form of phenomenalism), and it sup-
ports a particular version of the denial of experience-transcendence: the denial of the
empirical reality of that which cannot be manifested to or given to us. His objection
to the existence of appearances as a complete totality is not that a complete totality
could not be constructed but that it could not be given. He says that the world as a
whole is an object that can never come before one, cannot be given in any possible
experience (A483/B511), and that ‘in the empirical regress there can be encountered
no experience of an absolute boundary’ (A517/B545; see also A482–3/B510–11 and
AA492/B521). A sophisticated phenomenalist allows as empirically real what can be
constructed out of possible experience; this kind of view would not do the work Kant
requires his idealism to do in resolving the Antinomies, since he does not provide
reasons why the size and extent of the world could not be theoretically constructed.
Rather, he appeals to what can be given: what can possibly be manifest to us.
Empirical reality is limited to what can be given to or presented to us, to what we
can have acquaintance with; neither a boundary in space, nor the world existing as an
infinite totality, can be presented to us. They are therefore not empirically real.
We have seen that understanding the idealism of appearances in terms of the
rejection of experience-transcendence exactly fits the way in which Kant appeals to
the mind-dependence of appearances in resolving the Antinomies. Further, this kind of
mind-dependence enables us to see how Kant could think both that empirically real
properties exist unperceived and feature in causal explanation, while also thinking that
they are, in a sense, essentially sensory. More strongly, Kant thinks that we can under-
stand something as part of the empirically real world only if it is possible for it to be
presented to a conscious sensory experience. This point leads us to a comparison between
Kant’s idealism and a different form of anti-realism: anti-realism in philosophy of science.
III Anti-Realism and the Manifest Image

As we have seen, Kant thinks that we can have knowledge only of essentially manifest
features of reality; empirical reality contains nothing that, in principle, cannot be
experienced by us. As I understand him, Kant rejects the thought of there being a
conflict between the so-called manifest image and the so-called scientific image. This
can be explained by comparing Kant’s position to another kind of anti-realism: one
which features in philosophy of science. The broad family of positions called anti-
realist in philosophy of science are different from those which feature in philosophy
of language; I will refer to them as American anti-realism (following Papineau 1996b: 5).
One way of understanding American anti-realism is as a response to a radical
change in the fundamental properties postulated by the physical sciences. In a
 KANT ’ S IDEALISM AND HIS REALISM
Lockean account, primary qualities are both qualities that can be accessed in ordinary
perceptual experience (and that feature in our causal explanations of interactions
between the ordinary objects of experience) and also the explanatory properties
appealed to by science. The very small particles of matter which, Locke holds, are
responsible for the observable properties we experience things as having are under-
stood as having properties like shape, size, and arrangement comparable to those we
access in perception. In this respect, fundamental physics has completely changed.
The explanatory properties postulated by fundamental physics are no longer prop-
erties that feature in our perceptual experience. One way of understanding this
situation is to say that the properties and objects postulated by physics are unob-
servable; American anti-realism is a response to this. American anti-realism rejects
the claim that scientific theories should be understood as committed to the existence
of unobservable or theoretical objects or properties (Papineau 1996a; van Fraassen
1984). Rather, it says that science describes and predicts observable things and
cannot be regarded as making literally true claims about unobservable entities.
American anti-realists may deny that there are unobservables, deny that we can
give content to the idea of unobservables (here we have some similarity to Dummet-
tian anti-realism), or deny that we can have knowledge of them.
The realism to which the position is opposed claims that the aim of science is to
achieve true theories, as opposed to merely predictively successful ones, that there is
no principled reason to think that this aim will not be achieved, that contemporary
science is at least approximately true, and that the theoretical terms of contemporary
science are genuinely referential (Leplin 1984; Papineau 1996a). Realists think that
we ought to be committed to the existence of theoretical entities postulated by science
even though they are not observable. American anti-realists do not deny that the
physical world about which our theories are developed is independent of our
awareness of it: the position is not a form of idealism (although it could be combined
with idealism). It can be viewed as a form of scepticism, but it is more like a version of
epistemic humility.17 The position does not deny that there is a fact of the matter
about that which science aims to describe, so the position is very different from
British anti-realism which, as we have seen, denies that our judgments can answer to
a world that goes beyond our capacity to verify assertions about. On the contrary, it is
because science answers to a world which goes beyond our powers in this way that
American anti-realists place limits on what we can know through science. American
anti-realists think that scientific theories cannot be regarded as literally true of the
unobservable world they seem to describe; at the same time, they think that scientific
theories can be used to predict the course of events in the observable world.18 We
cannot know the truth about unobservables but we can have knowledge about
17
At least in the positions I discuss here—there is an enormous variety of positions here, and I do not
speak for all of them.
18
See Papineau (1996b: 4).
RELATIONAL APPEARANCES 
observable objects and we can use scientific theories to make predictions about these.
As one theorist puts it, the most that scientific theories can do is ‘save the phenom-
ena’ (van Fraassen 1984).
The idea that science describes something different from the world of our ordinary
perceptual experience could be put in terms of there being an opposition between
what is called, in Sellars’s term, the ‘manifest image’ and what he calls the ‘scientific
image’.19 The ‘manifest image’ refers to the world as presented in sense experience,
while the ‘scientific image’ refers to a supposedly completely different understanding
of the world—a world thought, in Locke’s time, to consist of colourless, extended
solid particles and thought, today, to consist of things which have no resemblance to
anything we understand experientially and which can be described only mathemat-
ically. Like American anti-realism, Kant’s account rejects this contrast: for Kant, the
world that science studies is the manifest world, and properties can be understood as
being empirically real only if they are manifestable.
Just as with the British anti-realist notions of experience, verification and warrant,
there are different ways of understanding the idea of what is ‘observable’, which will
lead to different versions of American anti-realism. A worry that might be raised
about American anti-realism is whether we can draw a principled distinction
between entities that are observable and entities that are, in principle, unobservable.
One way in which we might be thought to observe or experience electrons, for
example, is by causally interacting with them.20 As we have seen, Kant has a specific
account of what is required for something to be part of possible experience or
possibly perceivable. First, it must have causal implications for something we directly
perceive; second, there must be a way it could be presented in empirical intuition. He
thinks that everything that is empirically real is perceivable in this sense, and that
science gives us knowledge only of things which are observable in this sense. For
Kant, the things of which we have experience are the things that are described by
science, what we know through science counts as part of possible experience, and
neither science nor perceptual experience gives us knowledge of an unobservable
reality. Both give us knowledge only of essentially manifest features of reality. Both
direct perceptual experience and science gives us knowledge of the same objective
world, and in the same way: through interacting with it. Perceptual experience does
not give us knowledge of a subjective world, and science does not reveal that the
appearances of things are illusory or are properties of mental states. The manifest
image presents us with empirical reality, and is what science studies.
As we saw in Section II, manifest qualities and essentially manifest qualities can
both feature in causal relations. Kant thinks that there being causal connections
between things means, at least, that events are connected in law-governed ways (he
19
Sellars (1962). While this contrast is presented by Sellars, I am not concerned with the details of his
account here.
20
See Hacking (1984).
 KANT ’ S IDEALISM AND HIS REALISM
also thinks that this involves causally unified objects which are in law-governed
interaction with other such objects).21 This means that causal explanation in science
neither requires postulating hidden essences, nor (as we will see in the next section)
involves insight into intrinsic properties of things in virtue of which they have their
powers. Our insight into something’s being a cause is simply a matter of seeing that it
falls under a necessitating law, rather than being a matter of seeing how a certain
causal power has to follow from some inner nature. Causal explanation in science is
not about gaining insight into inner natures, but rather postulating necessitating
relations between things we observe. Such explanations can become increasingly
complicated because they can be captured by complex mathematical descriptions,
and because they can involve explaining a thing’s properties by explaining the
interactions of its parts. However, neither of these features need be taken to involve
postulating unobservables. Kant thinks that penetrating into the inner properties of
things involves giving further explanations involving empirically real properties, at a
finer level (A278/B334). This means that it involves properties which could, in
principle, be present in a possible experience, rather than theoretical, unobservable,
or hidden inner natures. Both the abstractness and the complexity of mathematical
descriptions and mathematical predictions of relations between appearances may
give us the impression of insight that goes beyond manifest features of reality.
However, on this account, the things the abstract mathematical descriptions describe
are manifestable, perceptible features of reality. Kant thinks that scientific claims
describe empirical reality only to the extent that they describe essentially manifest-
able qualities and that science does not require the postulation of anything
unobservable.
It might be argued that if we allow that what mathematical scientific claims
describe is grounded in things as they are in themselves, then Kant will be forced
to admit that we have some (mathematically describable) knowledge of things as they
are in themselves. If things in themselves ground appearances and mathematical
descriptions apply to appearances, then it seems that our mathematical descriptions
of appearances will map onto things in themselves and this might be thought to give
us some knowledge of them. In responding to this argument it is important to keep in
mind that Kant does not rule out our being able to say anything at all about things in
themselves; we can know certain formal claims about them. For example, if appear-
ances are appearances of things, then we know something about things in knowing
their appearances: we know that things appear to us in these ways. What Kant is
21
Kant expresses the principle of causality as saying that ‘[e]verything that happens (begins to be)
presupposes something which it follows in accordance with a rule’ (A189) or ‘[a]ll alterations occur in
accordance with the law of the connection of cause and effect’ (B232), and says that ‘[t]he schema of
the cause and of the causality of a thing in general is the real upon which, whenever it is posited, something
else always follows. It therefore consists in the succession of the manifold insofar as it is subject to a rule’
(A144/B184).
RELATIONAL APPEARANCES 
concerned to rule out is our having specific knowledge of the mind-independent

natures of things which ground the appearances we experience. That mathematical
descriptions in some sense map onto things in themselves does not mean that they
give us any insight into the natures things have in themselves. Consider our example
of colour, understood as essentially manifest. Mathematical descriptions of relations
among red perceptual appearances would map onto the disjunctive mind-independent
grounds of red appearances, but if it is the red perceptual appearings that are the
input to the equations, then this is what we are describing mathematically. Kant
thinks that mathematisation enables science to achieve new levels of precision in
describing relations between appearances but that this does not give us insight into
unobservable intrinsic natures. In Kant’s view, our understanding of what features in
the mathematical equations comes from our experience of properties which are
presented to us in experience.
We have noted that one significant difference between Kant’s account and Ameri-
can anti-realism is that while, like Kant, the latter deny that we can have knowledge
of unobservables, unlike Kant they do not (typically) think that the properties of
which we can have knowledge are essentially observable. They are not (typically)
idealists. There may also be differences between the positions with respect to the
understanding they have of the existence of an unknowable aspect of reality. There
are different versions of American anti-realism. If American anti-realism is under-
stood as claiming that science aims to describe an unobservable reality but does not
succeed in doing this, then it differs from Kant’s position. Kant does not think of
science as aiming to give us knowledge of noumena: the subject matter of science is
empirical reality. In Kant’s position, things in themselves are not theoretical entities
postulated by science; rather, scientific explanations involve manifest reality only and
science has no need of things in themselves. However, what is common to (some
versions of) American anti-realism and Kant’s position is the idea that there is
something unobservable that underlies what we can have knowledge of (the natures
things have in themselves) and that we cannot have knowledge of this unobservable
aspect of world, but rather can have knowledge only of what is, in principle,
perceivable.
IV Relational Appearances
We have considered an objection that says that physics tells us at least something
about things in themselves because its mathematical descriptions map onto whatever
it is that grounds appearances. In the other direction, the mathematical nature of
physics is taken by some philosophers to support thinking that it does not give us
knowledge of the intrinsic natures of things. The thought is that to the extent that
physics is mathematical, it gives us knowledge only of formal and relational features
of reality, and not of the intrinsic nature of what is in the relations. This has been
used to argue for the ‘inscrutability of matter’ thesis, which is a version of the claim
 KANT ’ S IDEALISM AND HIS REALISM
that we cannot know the nature of things as they are in themselves.22 In holding that
physics gives us knowledge only of relational features of reality, this position is closely
related to the position in contemporary philosophy of science called structural
realism.23 Structural realism has both an ontological and an epistemological variety.
Epistemological structural realism claims that science gives us knowledge only of
structural features of reality and that we cannot have knowledge of the intrinsic, non-
structural nature of reality. Worrall explains it as claiming that ‘it is a mistake to
think that we can ever “understand” the nature of the basic furniture of the universe’
(Worrall 1996: 162). Ontological structural realism agrees that we can have know-
ledge only of structural features of reality, but adds that this is all there is: there is no
more to reality than the structural nature revealed by science.24 Kant’s position has
some similarities with both of these positions: Kant thinks that we have knowledge
only of relational features of reality and that empirical reality is entirely relational. It
should be noted that Kant sometimes speaks interchangeably of powers and forces,
and calls both relations. He does not draw systematic distinctions among these
notions and my discussion follows him here. What is important for the discussion
is what they all have in common: they are not categorical grounds. The examples
Kant gives of the relations we know are spatial relations, motions (B67), and forces
(B67, A265/B321), and he talks of knowing matter in terms of its effects, which he
then describes as knowing outer relations (A277/B333). He opposes relations to
‘inner determinations’ (A283/B339), ‘that which is internal to the object in itself ’
(B66–7), and what is absolutely internal (A277/B333), and denies that we have any
knowledge of the latter. He says that ‘a persistent appearance in space (impenetrable
extension) contains mere relations and nothing absolutely internal’ (A284/B340),
and that place, shape, contact, and motion are all outer relations (A274/B330).
The idea that we need categorical grounds in addition to relational qualities is
sometimes motivated by the claim that such properties are central to our under-
standing of what it is for something to be a material object. However, as Langton
argues, many of the properties that give us an understanding of objects both at the
level of experience and in science are powers (Lockean tertiary qualities) rather than
categorical grounds of powers.25 This thought may be inadvertently displayed by the
22
See Eddington (1946), Russell (1967), Foster (1982), Lockwood (1989), and G. J. Strawson (2002;
2003).
23 24
Worrall (1996). Ladyman (1998).
25
See Langton (1998: Chs. 7 and 8). See also Smart (1963: 72), P. F. Strawson (1980: 280), Howard
Robinson (1982: 109), Blackburn (1990: 62–5), and Jackson (1998: 23). Locke says of secondary and
tertiary properties that they ‘are only powers to act differently upon other things, which powers result from
the different modifications of those primary qualities’ (1690: II, viii, 23). He sometimes seems to take the
view that the properties we know of objects are the secondary and tertiary ones: ‘most of the simple ideas
that make up our complex ideas of substances, when truly considered, are only powers, however apt we are
to take them for positive qualities: v.g. the greatest part of the ideas that make up our complex idea of gold
are yellowness, great weight, ductility, fusibility, and solubility in aqua regia, etc., all united together in an
unknown substratum; all which ideas are nothing else but so many relations to other substances, and are
not really in the gold, considered barely in itself, though they depend on those real and primary qualities of
RELATIONAL APPEARANCES 
following two quotations from philosophers who are explaining the role of primary
qualities in giving us our conception of what it is for something to be a material
object. Bennett says that ‘a thing’s having any specific primary quality consists in its
relating to many other things in specific ways’ (Bennett 1971: 100, my italics). And, in
discussing a primitive mechanics, Peacocke says: ‘What is it for an object to be a
material object, for it to be of material composition? I suggest that for something to
be a quantity of matter is for changes in its state of motion to be explicable by
mechanical forces acting on it, and for its changes of motion to exert such forces’
(Peacocke 1993: 172).26 If our grasp of what an object is at the level of experience
requires understanding how it relates to other things, and how its state of motion will
be changed by and will change other things, this requires powers (or, more generally,
relational properties) rather than categorical grounds of powers.
One of Locke’s reasons for thinking that we need categorical properties is thinking
that they play a fundamental explanatory role in science and enable us to understand
why things have the relational properties they have. The development of science has
led many philosophers to question this thought. Blackburn, for example, argues that
‘science finds only dispositional properties, all the way down’ and that physics cannot
give us what is categorical, so ‘any conceivable improvement in science will only give
us a better explanation of causal powers’ (Blackburn 1990: 63). Blackburn claims that
this is all science can do, firstly because the theoretical terms of a science are defined
functionally, in terms of their place in a network of laws; and, secondly, because the
concepts that physics uses—for example, energy and entropy—cover changes of
states, and ‘such concepts in effect tell us what is the same about a changing system,
in terms precisely of its powers and dispositions’ (Blackburn 1990: 63–4).27 On
its internal constitution, whereby it has a fitness differently to operate and be operated on by several other
substances’ (1690: II, xxiii, 37). Langton uses the idea of primary qualities understood in opposition to
powers to give her own reading of Kant’s secondary quality analogy: she says that his analogy should be
read as making all the qualities of which we can have knowledge powers. However, the way Kant
introduces the analogy makes it clear that it is intended to capture some kind of mind-dependence,
describing secondary qualities as those ‘which may be said to belong, not to the things in themselves, but to
their appearances, and to have no proper existence outside our representation’ (Proleg. 4: 289). His
intention is to compare properties like shape and size with properties which are in some sense per-
ceiver-dependent, properties which are essentially understood in perceiver-involving terms.
26
Similarly, Shoemaker says that ‘if we could indicate all of the ways in which the having of this
property could contribute to the casual powers of things, we would have said all there is to know about the
intrinsic nature of this property’ (Shoemaker 2003a: 249–50). He argues for this as follows: if, per
impossible, there were properties whose possession made no difference to the causal powers of the things
that have them, there could be no way for us to know about them, and no reason for us to believe in their
existence. Note that the position I ascribe to Kant is not that there are properties whose possession makes
no difference to causal powers, but rather that we cannot know the natures of things in virtue of which they
have the properties they have.
27
Similarly, P. F. Strawson says: ‘it seems that our search for the properties of the categorical base must
finally lead to the undeniably theoretical properties which physics assigns to the ultimate constituents of
matter–perhaps force, mass, impenetrability, electric charge. But these properties seem themselves to be
thoroughly dispositional in character. . . . If it seems true of the sensory properties in general that they all
dissolve together under reflective pressure, into dispositions, this seems even more certainly true of the
 KANT ’ S IDEALISM AND HIS REALISM
Kant’s account science gives us knowledge only of relational features of reality,

and does not describe or require categorical grounds of powers (something
non-relational).
In the Critique, Kant presents his most detailed discussion of the idea that
appearances contain only relations in the appendix to the Transcendental Analytic:
‘On the amphiboly of concepts of reflection through the confusion of the empirical
use of the understanding with the transcendental’ (A260–92/B316–49). The
Amphiboly section is one in which Kant’s direct concern is responding to, and
criticising, Leibniz’s position, and one problem that this creates is that it may not
always be completely clear which parts of the text represent Kant’s position and
which are descriptions of a Leibnizian position that he rejects. However, this section
clearly expresses Kant’s commitment to the entirely relational nature of empirical
reality. Kant says of matter:
What pertains to it internally I seek in all the parts of space that it occupies and in all effects
that it carries out, and which can certainly always be only appearances of outer sense.
I therefore have nothing absolutely but only comparatively internal, which itself in turn
consists of outer relations. Yet the absolutely internal in matter, according to the pure
understanding, is a mere fancy, for it is nowhere an object for the pure understanding; the
transcendental object, however, which might be the ground of this appearance that we call
matter, is a mere something about which we would not understand what it is even if someone
could tell us. For we cannot understand anything except that which has something corres-
ponding to our words in intuition. If the complaints “That we have no insight into the inner
in things” are to mean that we do not understand through pure reason what the things
that appear to us might be in themselves, then they are entirely improper and irrational;
for they would have us be able to cognize things, thus intuit them, even without senses.
(A277–8/B333–4)
Kant thinks that what is given to us in outer intuition contains nothing that is
absolutely inner or non-relational. He argues that as science penetrates further into
the inner natures of things it merely discovers relations at a deeper level; it never
reaches anything absolutely intrinsic or non-relational. Kant thinks that science does
not need absolutely intrinsic properties, that complaints about lacking them are
“physical” properties which are held to constitute their categorical base. Properties do not become less
dispositional simply by becoming more theoretical’ (P. F. Strawson 1980: 280). Smart argues that ‘all the
properties which science ascribes to physical things seem purely relational’ (Smart 1963: 72). Howard
Robinson claims that science ‘sees the basic constituents of the material world as being purely dispositional
entities which are characterised solely by reference to their ability to act upon and influence things in their
vicinity’ (Howard Robinson 1982: 109). For discussion, see also Ellis and Lierse (1994: 32) and Jackson
(1998: 23). A physicist writing for the lay person describes mass as ‘the property of an object that leads to its
gravitational force, and to the fact that force is necessary to change the velocity of an object’
(Chandrasekhar 1998: 224), and says that electric charge is ‘just a compact way of saying that the electron
has a property which makes it exert a force, which has nothing to do with its mass, on another electron’
(Chandrasekhar 1998: 63). Here, these basic properties of the constituents of matter are described purely in
terms of their effects.
RELATIONAL APPEARANCES 
improper, and that absolutely inner grounds are something whose nature we would
not understand even if someone could tell us about it. Intrinsic, non-relational
grounds play no role in scientific explanations of things’ powers.
Kant’s view that matter, as appearance, consists only of relations or forces is
fleshed out in greatest detail in his Metaphysical Foundations of Natural Science, in
which he says that he applies the doctrines of the Critique to the concept of matter
(MFNS 4: 474–6, 478) and in which he puts forward his dynamical theory of
matter.28 Kant explains the space-filling property of matter, impenetrability, in
terms of opposed forces rather than solidity, saying that attractive and repulsive
forces constitute the essence of matter.29 On this account, the fundamental bedrock
of scientific explanation is force, and we can never get beyond this to anything
categorical, to anything which would ground or explain forces or powers. He says
that ‘everything real in the objects of the outer senses, which is not merely a
determination of space (place, extension, and figure), must be viewed as moving
force’ (MFNS 4: 523).
On the Lockean account, primary qualities ground relational properties or
powers, and part of what this means is that they explain them: a thing’s primary
properties explain why it has the powers that it does. On this account categorical
grounds give us insight into why things have the relational properties (powers)
they do. Kant, in contrast, thinks that scientific explanation will bottom out in
fundamental forces which cannot be further explained (MFNS 4: 502, 513, 524,
534). As Daniel Warren, puts it, Kant’s dynamical theory of matter involves giving
up a certain kind of explanatory ambition: explaining why things have the relations
they do by seeing how things’ relational properties follow from their intrinsic or
categorical properties.30 Kant thinks that science will never give us this kind of
insight. Warren argues that for exponents of seventeenth-century mechanism all
properties of matter were to be explained in terms of fundamental properties which
characterise a thing ‘apart from its relations to others, or more particularly, apart
from its causal relations to others’ (Warren 2001a: 94). The mechanists aim to
28
See Langton (1998: Ch. 8) and Warren (2001a; 2001b) for helpful discussions of Kant’s dynamical
theory of matter.
29
See Harré and Madden (1975: Ch. 9), who see Kant’s dynamical theory of matter as one of the three
significant historical precursors to Faraday’s field theory. They argue that the materialist and the dynamist
agree that what we know of matter we know only by virtue of forces it exerts, but the materialist wants
substratum on which to base the forces, whereas the dynamist does not.
30
Two other helpful discussions of these issues are Smith (1990) and Wilson (1992). Smith argues that
the distinction between primary and secondary qualities was ‘an integral part of the most significant feature
of seventeenth-century thought: the break with Aristotelianism’ (Smith 1990: 225) and that the distinction
was part of the search for a genuinely explanatory science. He comments on the dynamical accounts of
Boscovich, Kant, and Faraday, who explain the space-filling property of matter in terms of force, that
‘[s]uch a line of thought will be seen as momentous in its implications when it is borne in mind both how
far an outlook such as this is from the quest for a fully intelligible intrinsic characterisation of the material
world that dominated the seventeenth century’ (Smith 1990: 247).
 KANT ’ S IDEALISM AND HIS REALISM
make forces comprehensible in terms of fundamental categorical properties; for

example, solidity is supposed to be the ground of a power: the property objects
have which explains their being impenetrable. Kant thinks that science bottoms out
in fundamental forces which cannot be further explained.
That Kant gives up the ambition of comprehending how things’ powers flow from
their intrinsic natures is important for his account of what it is for something to be a
cause. He thinks that understanding empirical causes involves seeing interaction
between spatio-temporal objects as falling under laws, but also that we are not able to
comprehend how one thing necessitates another, or how a thing’s powers flow from
its intrinsic nature. This is why, as noted in Section III, understanding something as
an (empirical) cause, for Kant, involves seeing it as being in law-governed inter-
actions with other things, rather than being a matter of having insight into how its
nature necessitates its powers. As Warren argues, when Kant says that we cannot
comprehend the fundamental forces of nature, what he means is that we cannot see
how they flow from anything else—not from forces, because they are, ex hypothesi,
fundamental, and not from the inner determinations of things, because we do not
know these (Warren 2001a: 107–8). This means that ‘[a]t some fundamental level we
will have to renounce the project of explaining how something with certain primary
qualities produces such and such effects’ (Warren 2001a: 95–6). Kant says that the
mechanists have the advantage that they can explain things in terms of properties we
can comprehend, like shapes, whereas on his dynamical theory of matter we simply
cannot get further than interacting forces and can say nothing to make explicable
why there are these and no other forces (MFNS 4: 524–5). He says:
it lies altogether beyond our horizon to comprehend original forces a priori with respect to
their possibility; all natural philosophy consists, rather, in the reduction of given, apparently
different forces to a smaller number of forces and powers that explain the actions of the former,
although this reduction proceeds only up to the fundamental forces, beyond which our reason
cannot go. (MFNS 4: 534)
That the possibility of the fundamental forces should be made conceivable is a completely
impossible demand; for they are called fundamental forces precisely because they cannot be
derived from any other, that is, they can in no way be conceived (4: 513; see also 4: 524, 534)
Since Kant thinks that science neither uses nor needs non-relational qualities and
that empirical reality has only relational features, we might think that he has an
entirely relational ontology, like ontological structural realists.31 In my view, how-
ever, although Kant thinks that we cannot cognize categorical non-relational features
of reality, that we do not need such features for knowledge of an objective world and
for science, he thinks that we cannot have a complete ontology with relations only.
This is the topic of Chapter 10.
31
Hanna (2006) takes this view. For discussion, see also Esfeld (2001) and Moore (2001).
RELATIONAL APPEARANCES 
V Relations and Idealism

It is clear that Kant thinks both that empirical reality is entirely relational and also
that it is mind-dependent, but it is not clear how these two parts of Kant’s account
relate to each other.32 Although, as we have seen, there are a number of places in the
Critique where Kant says or implies that everything that is empirically real consists
only of relations and contains nothing absolutely intrinsic, he does not mention this
view in his initial presentation of transcendental idealism, his overview of it in the
B preface, or in his arguments for the position. There are different views on how his
relational understanding of empirical reality relates to his idealism. Jill Vance
Buroker, for example, sees Kant’s claim that we can know only relations as following
from his view that we cannot know things in themselves, which, in turn, she sees as
following from the claim that we can have knowledge only of spatial things (Vance
Buroker 1981: 129). In contrast, Langton sees the argument as going in the other
direction and attributes to Kant the view that it is because we know only causal
powers, which are relational, that we cannot know things as they are in themselves,
which are intrinsic and non-relational (Langton 1998). (I discuss this in detail in
Chapter 10.) In my view, although Kant’s relational theory of matter and his idealism
are related, they are separable: one could accept either the relationality of space and
spatial objects or the ideality of space on its own, without accepting the other. Kant
thinks that spatial properties and properties which are known through the inter-
action of things in space are relational not intrinsic, and he thinks that our cognition
of spatial objects is structured by an a priori form of intuition which does not present
mind-independent reality. This means, in my view, that Kant has independent
reasons for thinking that we cannot know reality as it is in itself.
Kant first, briefly, introduces the idea that we know only relations towards the end
of the Transcendental Aesthetic in the second edition. After having argued that space
and time do not present us with features of mind-independent reality, and that the
spatio-temporal objects we know are merely representations and do not present us
with things as they are in themselves, Kant suddenly introduces a new claim about
appearances, without making it entirely clear how exactly this relates to the previous
claims: he says that everything we cognize consists in relations. He says:
everything in our cognition that belongs to intuition (with the exception, therefore, of the
feeling of pleasure and displeasure and the will, which are not cognitions at all) contains
nothing but mere relations, of places in one intuition (extension), alteration of places (motion),
32
Warren gives an account of this by explaining Kant’s idea that we can know only powers partly in
terms of a detailed analysis of Kant’s category of reality; he argues that ‘on Kant’s view, a sensible property
can be subsumed under the category of reality only if it can be regarded as a power to produce effects of a
certain sort’ (Warren 2001b: xiv). He maintains that the category of reality, as applied to objects of the
senses, always involves an intensive magnitude, and that we can represent an intensive magnitude as a
determinate magnitude only by reference to its causal effects. Thus, the real properties of objects of the
senses must be regarded as causal powers (Warren 2001: 13–31).
 KANT ’ S IDEALISM AND HIS REALISM
and laws in accordance with which this alteration is produced (moving forces). But what is
present in the place, or what it produces in the things themselves besides the alteration of place,
is not given through these relations. Now through mere relations no thing in itself is cognized,
it is therefore right to judge that since nothing is given to us through outer sense except mere
representations of relation, outer sense can contain in its representation only the relation of an
object to the subject, and not that which is internal to the object in itself. (B66–7)33
The passage is extremely compressed. When, at the start of the passage, Kant says
that everything we cognize contains nothing but relations, this does not seem to
relate to his idealism, but rather to the idea that what we cognize are forces and
motions. He then says that no thing in itself is cognized through mere relations,
which implies that to cognize a thing in itself would be to cognize its properties
without reference to other things. Kant then seems to move very quickly, and without
argument, from the claim that we know only relations to the claim that we know
things only as they are in relation to subjects. Kant introduces the passage as
‘confirmation of this theory of the ideality of outer as well as inner sense, thus of
all objects of the senses, as mere appearances’ (B66), indicating that it is supposed to
support the idea that we can know only mere appearances. This might suggest that
the ideality of things as they appear to us and the fact that we cannot have knowledge
of things in themselves is supposed to follow from the claim that we know only
relations. On the other hand, the passage is compatible with seeing the order of
explanation as the other way round: if Kant takes it to be a consequence of cognizing
only spatial-temporal appearances and not things as they are in themselves that we
cognize only relations, then pointing out that we cognize only relations could provide
support—or as he puts it ‘confirmation’—of his theory.
In the Amphiboly, Kant’s reason for saying that we can know relations only and
that science never reaches what is categorical seems to be that we can cognize objects
we are given in outer intuition and that outer intuition gives us spatial objects. In
other words, he takes it that the relationality of appearances follows from their
spatiality (and does not seem to argue for this). He says that having knowledge of
the absolutely inner nature of things would require knowing things without their
being given in sensible intuition (A277–8/B333–4), because what is given in sensible
intuition are spatial objects, and these are relational. He says that the
inner determinations of substantia phaenomenon in space, on the contrary, are nothing but
relations, and it is itself entirely a sum total of mere relations. We know substance in space only
through forces that are efficacious in it, whether in drawing others to it (attraction) or in
preventing penetration of it (repulsion and impenetrability); we are not acquainted with other
properties constituting the concept of the substance that appears to us in space and which we
call matter. (A265–6/B321–2)
33
In the parallel comments on time, Kant suggests that that which contains nothing but relations is the
form of intuition (B67). A little later, he says that an appearance property is one which depends on the kind
of intuition of the subject in relation to the given object (B69).
RELATIONAL APPEARANCES 
[T]hat which is inner in their state cannot consist in place, shape, contact, or motion (which
determinations are all outer relations). (A274/B330)
He argues that if our cognition did not require intuition, and we could know things
through mere concepts, we could then know things’ inner natures. Since our cogni-
tion requires intuition, it limits us to spatial objects, and ‘a persistent appearance in
space (impenetrable extension) contains mere relations and nothing absolutely
internal, and nevertheless can be the primary substratum of all outer perception’
(A284–5/B340–1).
In these claims, Kant simply seems to assert that spatial properties are relational.
If Kant thinks that empirical reality is relational because it is spatial, then its
relationality is linked to his idealism but separate from it. It is linked because
Kant thinks that space is ideal and thus that objects known in terms of their spatial
relational properties are ideal. However, it is also separable, since it is not the
ideality of space that Kant invokes to explain the relationality of empirical reality,
but merely the fact that our knowledge is limited to spatial objects. This implies
that he would think that empirical reality was relational even if he were not an
idealist,34 so long as our knowledge is limited to spatial objects. Thus, it may be that
Kant’s view that appearances are relational is over-determined: they are relational
because essentially manifest qualities exist in relation to possible perceivers,
and they are relational because we know only spatial relations and forces. Both of
these claims follow, Kant thinks, from the claim that our cognition is limited to
spatial objects.
Appearances are relational in two senses. However we understand the connection
between these two, both tell us something about the way Kant thinks of things as they
are in themselves, because the notion of things as they are in themselves is opposed to
relations in both of these senses. Things as they are in themselves are things as they
are independently of their perceptual appearing to us, and things’ intrinsic or
absolutely inner natures are the natures they have independently of the relations
they are in to other things. Kant thinks such categorical grounds are not a part of
manifest experience and are not needed by science, but he retains a commitment to
the idea that there are categorical grounds or intrinsic natures.
In the classic Lockean account, there are four central features of primary qualities
that enable them to play their role in characterising or conceiving of an objective
world:
(1) They are observable qualities of physical objects, accessed in ordinary percep-
tual experience, which give us our understanding of what it is for something to
be a material object.
(2) They feature in causal explanation in the physical sciences.
34
As in Langton’s (1998) account.
 KANT ’ S IDEALISM AND HIS REALISM
(3) They are contrasted with essentially phenomenal properties—properties

which are understood in essentially sensory ways.
(4) They are contrasted with powers; they are the categorical grounds of powers.
The Lockean account maintains that properties which fill all of these roles are
necessary for understanding and thinking about an objective world; further, it holds
that the same properties play all of these roles.35 Both of these claims can be
contested. For example, it may be argued that our conception of the physical world
requires only relational properties and that we do not need anything categorical
(ontological structural realists hold this). And if it is granted that we need something
that plays each of these roles, it may be disputed whether the same properties fill each
role. As we have seen, Kant agrees with Locke in thinking that all these roles must be
filled, but he disagrees with Locke that the same properties fill all of them. He thinks
that properties grasped in sensory terms can play a role in causal explanation in
ordinary experience and in science, that these are in fact the only properties that
feature in causal explanations accessible to us, and that science neither reaches nor
needs categorical properties. However, in my view, Kant still thinks that categorical
grounds of powers are needed: he thinks that an ontology of only manifest, relational
appearances would be incomplete, as such properties are not ontologically funda-
mental or self-sufficient. This is my subject in Chapter 10.
35
Whether Locke thinks that we actually experience primary properties and can comprehend
how secondary and tertiary properties follow from them is controversial. See Wilson (1979: 144) and
Ayers (1981).
10
Intrinsic Natures
I Introduction: Intrinsic Natures

The previous chapter was primarily concerned with Kant’s understanding of empir-
ical reality. Kant thinks that the empirical world of which we have knowledge is
entirely relational and contains only things that can be presented to us in possible
perceptual experiences. It contains nothing categorical, intrinsic, or non-relational.
This chapter is concerned with the role of things as they are in themselves in Kant’s
account. I argue that he thinks that things have intrinsic, non-relational, categorical
natures which ground their relational appearances.
As we saw in Chapter 9, one central reason for thinking that we cannot make
sense of an objective world without primary qualities is based on the thought that
powers, or relational properties more generally, require categorical properties that
ground them. This is clearly part of Locke’s conception of primary qualities.
Lockean primary qualities are the categorical grounds of powers, play explanatory
roles in science, and can be presented to us in experience (but are not essentially
sensory qualities). Kant, as we have seen, holds that we can cognize only essentially
manifest qualities of reality and does not think that science needs or uses categor-
ical, non-relational qualities. However, I will argue that he agrees with Locke in
thinking that relations must be grounded in something non-relational—and that
this role is filled by the way things are as they are in themselves. On Kant’s account,
as opposed to the Lockean account, this grounding role is made much murkier by
the fact that Kant thinks that categorical grounds play no explanatory role in
science or in causal explanation at the level of experience: this makes it harder to
see why Kant should think that we need such grounds, or even what it means to say
that things in themselves ground appearances. As I understand him, Kant thinks
that this murkiness is an unavoidable part of our epistemic situation. He thinks
that not only can we not know things’ intrinsic natures, we cannot understand how
they ground appearances, or even what kind of relation this grounding might be.
However, I argue that this does not rule out our being entitled to say that the way
things are in themselves grounds appearances. As I will argue, Kant thinks we
cannot have a complete, coherent account of the status of appearances without a
commitment to things in themselves, even though we cannot understand the way
in which things in themselves ground appearances.
 KANT ’ S IDEALISM AND HIS REALISM
As we saw in Chapters 1, 3, and 4, there is textual and philosophical evidence

throughout the Critique that Kant thinks that there is a way things are in themselves.
In this chapter I examine a further piece of textual and philosophical evidence for
this: Kant thinks that relations require something non-relational. He discusses this
primarily in the section ‘On the Amphiboly of Concepts of Reflection through the
Confusion of the Empirical Use of the Understanding with the Transcendental’
(A260–92/B316–49), the section where he presents his relational account of empir-
ical reality in the greatest detail in the Critique. I argue that this section provides
further support for thinking that Kant has a commitment to there being a way things
are in themselves. Along with Daniel Warren and Rae Langton, I also argue that it
also gives us a way of explaining this commitment: Kant thinks that things have
intrinsic natures that ground their relational features.1 Thus, transcendental idealism
includes the claim that we cannot have knowledge of the intrinsic nature of reality.
In the Amphiboly section, Kant clearly states his view that appearances are
entirely relational. Because he argues that empirical reality (and the science that
describes it) is entirely relational, some take this section to show that he has an
entirely relational ontology. He also seems both to argue for the existence of things
in themselves and to characterise them by saying that there must be something
absolutely non-relational that grounds appearances—that things have intrinsic
natures. However, whether the latter is really part of Kant’s view is complicated
and controversial. In the Amphiboly Kant is specifically concerned with criticising
and rejecting views he attributes to Leibniz. Kant does not always explicitly state
when he is presenting a Leibnizian view he entirely rejects and when he is present-
ing claims he agrees with but criticising the conclusions Leibniz draws from them.2
He reconstructs a line of thought he attributes to Leibniz, with respect to which he
rejects the conclusion but clearly accepts at least some of the premisses. He does not
explicitly say whether he accepts all the premisses but thinks Leibniz’s conclusion
does not follow from them or whether he also rejects some of Leibniz’s premisses.
This leaves it controversial which of the premisses Kant attributes to Leibniz he
accepts, which makes it problematic to rest an interpretation on this section alone.
This is a difficulty both for those who want to use this section to show that Kant has
a commitment to intrinsic natures and for those who think the section shows that
he has an entirely relational ontology. I will argue that there is a clear reading of the
1
Daniel Warren argues that the Amphiboly section shows that Kant’s characterisation of things in
themselves is not entirely negative: it is the idea of things as they are independently of their relations to
other things, including us (Warren 2001b: 38). Rae Langton interprets transcendental idealism as claiming
that we cannot have knowledge of the intrinsic nature of reality and that intrinsic natures are causally inert
(Langton 1998). I reject the second part of her interpretation but I agree that transcendental idealism
includes the former claim.
2
Langton, who draws heavily on this section, has been criticised for making extensive use of texts that
occur in the context of Kant’s criticism of Leibniz and in which it is not clear that the views discussed are in
fact Kant’s. See Bird (2000: 106) and Rosefeldt (2001: 267).
INTRINSIC NATURES 
Amphiboly, compatible with Kant’s critique of Leibniz, on which Kant is commit-

ted to there being something non-relational and that this straightforwardly fits with
the way he talks in the rest of the Critique about things in themselves and the role
they play. Therefore, I take this section to help us in understanding Kant’s com-
mitment to there being a way things are in themselves.
Before looking at the argument in the Amphiboly I note two possible problems
with understanding Kant’s commitment to things in themselves as asserting that
things have intrinsic natures which ground their relational appearances. The first
problem is that he says not only that we know things only as they appear to us, but
also that we know relations between things only as these relations appear to us. Kant
primarily uses the expression ‘things as they are in themselves’ in contrast to ‘things
as they appear to us’: in this sense, the term is contrasted not with all relational
properties but, more specifically, to certain relations to us. It is compatible with this
that the way things are in themselves includes relations between things as these
relations are in themselves, independently of their appearing to us. This means that it
is imprecise to say that ‘things as they are in themselves’ refers to things’ intrinsic
natures, if this is understood in opposition to all relational properties.3 However,
I will argue that Kant’s reasons for thinking that there must be a way things are as
they are in themselves centrally include his reason for thinking that things must have
intrinsic, non-relational natures (though our ignorance of things in themselves may
also include ignorance of relations between them as they are in themselves). This
means that we can use the idea of intrinsic natures to examine his arguments for
things in themselves and how to understand their grounding role, so long as we keep
in mind that Kant also goes further: not only does he claim that we cannot know
things’ intrinsic natures, he also claims that we cannot know relations between things
in themselves independently of the way these relations appear to us.
The second problem with attributing to Kant the idea that we cannot know things’
intrinsic natures is that there is no agreement among philosophers about how to
3
For discussion of this problem in Langton’s account, see Allais (2006) and Langton (2006). In addition
to this problem with talking of intrinsic natures, there is a further terminological difficulty with Langton’s
account. She identifies things in themselves with substances and claims that nothing in the phenomenal
world really counts as substance (Langton 1998: 20, 56). Since Kant denies that the substances we know
have something absolutely inner, Langton is, at least, using the term ‘substance’ differently from Kant. Kant
thinks that, like all the categories, we can legitimately use the category of substance only in relation to
something that is given in space and time. And he also argues that it is a condition of the possibility of
experience that the objects of experience are persisting things (substances) which are made up of
permanent (conserved) stuff (substance). If we use the category of substance independently of the
conditions of our sensible intuition we will have merely the notion of something which is a subject of
predicates, but no idea what this applies to, what is being said of a thing when we apply this concept, or in
virtue of what features of things we should apply it. (For a helpful discussion of unschematised categories,
see Warren 2001b: 1–11.) However, it seems to me that this is a terminological, not a substantial, problem
with Langton’s account. While her terminology is not Kant’s, we could restate her account in his terms:
instead of saying that substances are things with intrinsic properties, we can simply say that things have
intrinsic properties or intrinsic natures.
 KANT ’ S IDEALISM AND HIS REALISM
understand the idea of intrinsic properties,4 and Kant does not present a precise
account of this or of the nature of relations. He speaks simply of absolutely inner
properties, which he opposes to relations and forces. However, it seems to me that the
terminology of intrinsic natures is useful and that there is no unproblematic vocabu-
lary here. In my view, rather than talking of intrinsic properties, the simplest and
most intuitive way of expressing Kant’s thought that things have non-relational
natures would be to say that things have a way they are in themselves, a way they
are independent of their relations to other things and to subjects. But the expression
‘things as they are in themselves’ is what we are trying to interpret, and it is of course
laden with the interpretative history of the first Critique, including the widespread
understanding of ‘things in themselves’ as a kind of non-sensible, non-spatio-temporal
things: positive noumena. In this context, the idea of intrinsic, non-relational natures is
helpful. Using what Humberstone calls the intuitive idea of intrinsic properties—
properties a thing has in and of itself (Humberstone 1996: 229)—helps us understand
what it is that Kant thinks we are lacking in not having knowledge of things as they are
in themselves. It gives us some further insight into his reason for denying us this
knowledge and into the philosophical role things as they are in themselves play in his
account.
II Outer and Inner: Relations in the Amphiboly

The Amphiboly section clearly expresses Kant’s commitment to the entirely rela-
tional nature of empirical reality. What is controversial is whether, in addition, the
section supports attributing to Kant the view that things have intrinsic or inner
natures we cannot know. An alternative reading is that the Amphiboly section merely
says that noumena understood in the positive sense (about whose existence Kant is
agnostic) would, if they existed, be things that had an absolutely inner nature.5 On
this reading, in other words, Kant diagnoses why Leibniz thinks there are things with
inner natures but does not himself express a commitment to things having inner
natures. This is a delicate question. In my view, one part of what Kant argues in this
4
See, for example, Van Cleve (1995), Humberstone (1996), Hawthorne (2001), Langton and Lewis
(2001), Lewis (2001), Marshall and Parsons (2001), Sider (2001), Weatherson (2001), and Shoemaker
(2003a; 2003b).
5
It is not completely clear to me what Warren’s view on this issue is. On the one hand, he says that Kant
‘is not denying that things have inner properties, i.e., that things have properties apart from all their
relations. These inner properties simply do not figure in our knowledge’ (Warren 2001b: 45). On the other
hand, he describes the idea of the absolutely inner as an idea in Kant’s technical sense. (Warren 2001b: 58).
In Kant’s technical sense, an idea is a concept which we cannot help using, and which we need to use to
unify experience, but which cannot be known to correspond to anything actually existing. An example is
the idea of complete systematic unity which we use to organise scientific knowledge, but which does not
correspond to something actually existing in empirical reality (A644/B672; A681/B709). Kant thinks that
the idea of intrinsic natures plays a regulative role in guiding science, in leading us always to look for more
fundamental explanations of the behaviour of the parts of matter. In my view, however, he is also
committed to things having intrinsic natures, independently of us.
INTRINSIC NATURES 
section is that the way we understand positive noumena (to the extent that we do) is
by thinking of objects that have entirely (only) inner natures (objects which are not
spatial) and he says that we can give content to the idea of such objects by thinking of
them as things that have only mental properties. Leibnizian monads, as simple, non-
extended soul substances, fit this description. Kant argues that we cannot have
knowledge of such objects—we cannot even know whether there are such objects—
and he diagnoses what he thinks is the mistaken line of argument that led Leibniz to
hold that we can. However, he also asserts that relations require non-relational
grounds and therefore that there is something that grounds the relational appear-
ances of which we have knowledge. As I read him, Kant argues that the Leibnizian
mistakenly takes the correct thought that we are committed to something that
grounds relational properties, together with the correct thought that we have know-
ledge only of spatial (and therefore relational) properties, and incorrectly concludes
from this that there are monads (things with only inner natures, things that are not
spatial). The Leibnizian’s mistake is to take the first two claims to show that there are,
in Kant’s terms, noumena in the positive sense. Kant rejects the conclusion: he rejects
the claim that we can show that there are non-spatial objects with entirely inner
natures. However, in my view, he asserts the two premisses that the Leibnizian takes
to lead to this conclusion: that we know only relations and that relations must be
grounded in something non-relational. As I will argue, he is committed to the
position that the things that appear to us have intrinsic natures—ways they are
independently of their relations to us and to other things.
In the Amphiboly, Kant criticises ways in which we might be tempted to take
relations between certain opposing concepts to give us insight into objects. He says
that since our cognition is limited to things that are given to us intuition, we cannot
simply take analytic relations between concepts, on their own, to give us insight into
the objects of our knowledge, but must instead consider these concepts in relation to
the forms of our sensible intuition. Otherwise, although our thoughts might lead to a
conceptually consistent system (such as the Leibnizian monadology), it would not be
one of which we could have knowledge, since one of the conditions of cognition
would be missing.
The opposed concepts Kant discusses are identity and difference, agreement and
opposition, inner and outer, and form and matter. Kant says that we can understand
these concepts as logically opposed without worrying about whether their objects are
intelligible or sensible, but if we want to use them to give us insight into objects, we
must consider the objects that fall under the concepts with respect to how we cognize
them, in other words, with respect to our spatio-temporal forms of intuition (A269/
B325). For example, with respect to the concepts of identity and difference: consider
objects x and y which have exactly the same determinations (properties known
through concepts). Kant thinks that if we were to abstract from our form of
sensibility we would conclude from these objects falling under all the same concepts
(having the same properties) that they are the same thing. This is the principle of the
 KANT ’ S IDEALISM AND HIS REALISM
identity of indiscernibles. But once we bring in the spatial form of intuition, which is
a requirement of our cognition, we can see that even if no distinction is to be found in
our concepts of the things, they could be none the less be distinguished by their
position in space.6 Thus the principle of the identity of indiscernibles does not hold
for the spatio-temporal objects of our knowledge. Spatial objects x and y could have
the same properties (could fall under all the same concepts) without being the same
object, so long as they are differently located. So, Kant says, thinking about the
opposed concepts of identity and difference merely in terms of the content of the
concepts, without considering the spatial form of the objects of our experience, will
lead us, mistakenly, to think that the principle of the identity of indiscernibles gives
us insight into objects. If our access to objects were purely conceptual, then knowing
of objects x and y that they had all the same determinations would enable us to
conclude that they are the same object. However, because our access to objects is not
purely conceptual, but requires objects to be given to us in intuition, we cannot draw
this conclusion with respect to the objects of our knowledge. Kant thinks that if we
want to use these concepts legitimately we must either abstract completely from an
object (as in pure logic) or else must understand them in relation to the objects which
are given to us in intuition (A279/B335).
The important discussion for our present purposes concerns Kant’s treatment of
‘the inner and the outer’ (A265/B321), because here Kant considers the thesis that we
have knowledge only of relations and the thesis that relations require something non-
relational. There are three claims in this discussion we can attribute to Kant without
much controversy. First, he clearly says that our knowledge is limited to relations and
cannot reach anything absolutely non-relational or absolutely inner. Second, he says
that it is a conceptual truth that relations require something non-relational. And,
third, he argues that we cannot assert that the objects of our experience are appear-
ances of monads: simple thinking things that do not causally interact with each other.
The Leibnizian, Kant thinks, mistakenly thinks that the third claim follows from the
first two. What is controversial is whether, in this section, Kant is committed to
thinking that things in themselves have intrinsic natures which ground the relations
we can know. Given the complications in working out what part of the argument
contains Kant’s own commitments, I will quote three important passages at length.
In the first passage, Kant says that
In an object of the pure understanding only that is internal that has no relation (as far as the
existence is concerned) to anything that is different from it. The inner determinations of
substantia phaenomenon in space, on the contrary, are nothing by relations, and it is itself
6
This assumes that an object’s spatial position is not simply a further series of conceptually described
properties. This is central to Kant’s account, since he thinks that our primary representation of space is an
intuition and that there are features of the way objects are presented to us that are not captured by
conceptual determinations. Spatial locations, in Kant’s account, are primarily presented to us through
direct acquaintance, not descriptive criteria.
INTRINSIC NATURES 
entirely a sum total of mere relations. We know substance in space only through forces that are
efficacious in it, whether in drawing others to it (attraction) or in preventing penetration of it
(repulsion and impenetrability); we are not acquainted with other properties constituting the
concept of the substance that appears to us in space and which we call matter. As object of the
pure understanding, on the contrary, every substance must have inner determinations, and
forces that pertain to its inner reality. Yet what can I think of as inner accidents except for those
which my inner sense offers me?—namely that which is either itself thinking or which is
analogous to one. Thus because he represented them as noumena, taking away in thought
everything that might signify outer relation, thus even composition, Leibniz made out of all
substances, even the constituents of matter, simply subjects gifted with the powers of repre-
sentation, in a word, monads. (A265–6/B321–2)
Here Kant reconstructs a line of argument which, he thinks, led Leibniz to the
position that reality consists of monads. As I read him, Kant says that it is because,
on the one hand, we are committed to the view that things have intrinsic (non-
relational) natures but, on the other hand, we have knowledge only of spatial,
relational properties and are presented with inner properties only in inner sense,
through thinking, that we mistakenly think we are entitled to conclude that
the spatial appearances of our knowledge are appearances of monads with inner,
mental natures.
Since Kant is here criticising Leibniz’s reasoning, it might be thought that none
of the passage presents Kant’s view. This point is delicate. Kant thinks that the
comparisons Leibniz makes are correct if we abstract from our sensible intuition:
they are correct in terms of logic alone. Further, it is clear, and is confirmed in the rest
of the section and elsewhere (for example, B67), that the claim that ‘[w]e know
substance in space only through forces that are efficacious in it’ is Kant’s view. Kant
also thinks that we have no representations of anything inner other than what is
offered in inner sense. It is clear, therefore, that he is committed to some of the
premisses he attributes to Leibniz. Our question is what he is committed to when he
says that ‘[a]s object of the pure understanding, on the contrary, every substance
must have inner determinations, and forces that pertain to its inner reality’. This
could be taken to commit him merely to saying that the idea of positive noumena
(which, so far as we know, may or may not exist) is the idea of things which have
inner determinations, or, on the other hand, to the claim that all objects, as we
understand them in thought, have absolutely inner determinations or non-relational
properties. The passage seems consistent with either reading, though it more natu-
rally expresses the second. If Kant’s claim were that a special kind of object—a
noumenon in the positive sense—must be thought as having an entirely inner
(non-spatial) nature we would expect him to say something like ‘things which are
merely objects of a pure understanding have inner natures’ as opposed to saying that
every substance, when thought through the understanding alone, must be thought as
having inner natures. I will argue that the second reading is more plausible. As
I understand him, Kant is not rejecting the claim that everything relational requires
 KANT ’ S IDEALISM AND HIS REALISM
something non-relational; rather, he is arguing that we cannot conclude from this

conceptual thesis that reality consists of simple soul substances.
Kant repeats his reconstruction of Leibniz’s argument a few pages later, saying:
the Leibnizian monadology has no ground at all other than the fact that this philosopher
represented the distinction of the inner and outer merely in relation to the understanding.
Substances in general must have something inner, which is therefore free of all outer relations,
consequently also of composition.7 The simple is therefore the foundation of the inner in
things in themselves. But that which is inner in their state cannot consist in place, shape,
contact, or motion (which determinations are all outer relations), and we can therefore
attribute to substance no other inner state than that through which we internally determine
our sense itself, namely the state of representations. This completes the monads, which are to
constitute the fundamental matter of the entire universe, the active power of which, however,
consists merely in representations, through which they are properly efficacious merely within
themselves. (A274/B330)
Here again it might be argued that it is not entirely clear whether Kant himself is
committed to saying that substances must have something inner or whether he
thinks that this is something one would say only after making the mistake that the
Amphiboly is concerned with—failing to think about concepts in relation to the
sensible form of our intuition. The general outline of Kant’s diagnosis of Leibniz’s
error is clear: Kant argues that Leibniz goes wrong in drawing ontological conclu-
sions from relations between concepts without considering the cognitive faculty
through which the objects falling under the concepts are cognized, in other words,
without bringing in intuition. He thinks Leibniz mistakes the analytic claim that
the outer requires the inner for insight into the objects of our knowledge, and
wrongly takes it to show that the objects of our knowledge have entirely inner
(non-spatial) natures.
That Kant thinks something absolutely inner, or something non-relational, is
required as a matter of a conceptual truth can be seen in the final, complex, passage
from the Amphiboly I will discuss. Kant says that
According to mere concepts the inner is the substratum of all relation or outer determinations.
If, therefore, I abstract from all conditions of intuition, and restrict myself solely to the concept
of a thing in general, then I can abstract from every outer relation, and yet there must remain a
concept of it, that signifies no relation but merely inner determinations. Now it seems as if it
follows from this that in every thing (substance) there is something that is absolutely internal
and precedes all outer determinations, first making them possible, thus that this substratum is
7
In Langton’s discussion, she translates this as ‘Substances in general must have some intrinsic nature,
which is therefore free from all external relations’ (A274/B330); she says that she avoids ‘inner’ and ‘outer’
because of their spatial implications (Langton 1998: 18n). I follow Guyer’s and Wood’s more literal
translation, as it avoids other problems: the difficulty involved in giving a precise characterisation of the
distinction between intrinsic and extrinsic properties and the different ways in which the notions are used
in contemporary philosophy.
INTRINSIC NATURES 
something that contains no more outer relations in itself, consequently that it is simple . . . All
this would be correct, were it not that something more than the concept of a thing in general
belongs to the conditions under which alone objects of inner intuition can be given to us, and
from which the pure concept abstracts. For these show that a persistent appearance in space
(impenetrable extension) contains mere relations and nothing absolutely internal, and never-
theless can be the primary substratum of all outer perception. Through mere concepts, of
course, I cannot think something external without something inner, for the very reason that
relational concepts absolutely presuppose given things and are not possible without these. But
since something is contained in the intuition that does not lie at all in the mere concept of a
thing in general, and this yields the substratum that cannot be cognized through mere
concepts, namely a space that, along with everything that it contains, consists of purely formal
or also real relations, I cannot say that since without something absolutely inner no thing can
be represented through mere concepts, there is also nothing outer that does not have
something absolutely internal as ground in the things themselves that are contained under
these concepts and in their intuition. For if we have abstracted from all conditions of intuition,
then of course there remains nothing in the mere concept except the inner in general, and its
relation in that, through which alone the outer is possible. But this necessity, which is grounded
only on abstraction, does not obtain in the case of things insofar as they are given in intuition
with determinations that express mere relations without having anything inner at their ground,
since these are not things in themselves but simply appearances. And whatever we cognize only
in matter is pure relations (that which we call their inner determinations is only comparatively
internal); but there are among these some self-sufficient and persistent ones, through which a
determinate object is given to us . . . It is certainly startling to hear that a thing should consist
entirely of relations, but such a thing is also mere appearance, and cannot be thought at all
through pure categories; it itself consists in the mere relation of something in general to the
senses. (A282–5/B338–41, my italics)
Kant’s starting point is what we can say of objects if we think of them merely in terms
of the understanding, through concepts alone. At this point, he is concerned with
conceptual relations. The key claims here are his assertions that ‘[a]ccording to mere
concepts the inner is the substratum of all relation or outer determinations’, that
‘[t]hrough mere concepts, of course, I cannot think something external without
something inner and for the very reason that relational concepts absolutely presup-
pose given things and are not possible without these’, and that ‘without something
absolutely inner no thing can be represented through mere concepts’. It is clear that
he is committed to there being a conceptual connection between the outer (the
relational) and inner (the intrinsic, the non-relational). The conclusion he wants to
resist is that this conceptual relation can be used to establish the claim that the objects
of our knowledge have simple, mental natures.
One way of understanding this discussion would be to see Kant as saying that the
claim that the outer requires the inner, or that relations require something non-
relational, is merely a claim about analytic entailments among concepts, and gives us
no insight at all into metaphysics. This can then be taken to be consistent with seeing
Kant as having an entirely relational ontology, and not being committed to things in
 KANT ’ S IDEALISM AND HIS REALISM
themselves. It might also be thought to fit with his aim in the Amphiboly, which is to
argue that what we take ourselves to be able to infer from mere conceptual relations
does not give us insight into objects until we consider these conceptual relations in
relation our intuition. Apparent support for this reading is given by Kant’s account of
the difference between logical possibility and real (or metaphysical) possibility
(A244/B302, Bxxvin), which implies that we cannot simply read metaphysics off
logic or conceptual possibilities. As we have seen, Kant thinks that we cannot infer
that something is (what he calls) really possible from its being logically possible. He
says ‘that the non-being of a thing does not contradict itself is a lame appeal to a
logical condition, which is certainly necessary for the concept, but far from sufficient
for its real possibility’ (A244/B302). We cannot infer from the lack of contradiction
in a concept that it is the concept of a really possible object. Similarly, Kant thinks
that we cannot infer from the fact that logic does not connect two things that there is
no necessary connection between them; something may not be logically necessary,
yet there may still be another kind of necessary connection. This may be taken to
support the thought that although Kant does say that it is a conceptual truth that
relations require something non-relational, or the outer requires the inner, we should
not take this as telling us anything about metaphysics. However, this reading
misunderstands what follows from Kant’s account of the relation between logical
and real possibility. Crucially, something’s being logically possible is not enough
(Kant thinks) to show that it is really possible, but something’s being logically
impossible is enough to show that it is really impossible. If logic says that something
is not coherent then it is not any kind of possibility: Kant says that the principle of
non-contradiction is ‘the general though to be sure only negative condition of all our
judgments whatsoever’ (A150/B189). In the Amphiboly, Kant clearly thinks that it is
a logical truth that relations presuppose something non-relational. The idea of there
being relations without there being something non-relational simply does not (Kant
thinks) make sense; it is not any kind of possibility. Although Kant thinks that we
cannot read metaphysics off logic (a concept may contain no logical contradiction yet
not be a concept of a really possible thing, and logic may not connect two things
which metaphysics sees as necessarily connected), logic does inform metaphysics;
analytic entailments between concepts rule out claims denying them.
If Kant thinks that logic commits us to thinking that relations require something
non-relational, then his criticism of Leibniz cannot be that it is because Leibniz failed
to consider objects in relation to the conditions of our cognition (the requirement for
something given in outer intuition) that he concluded that relations require some-
thing non-relational. Rather, Kant’s argument is that by failing to consider objects in
relation to the conditions of our cognition, Leibniz wrongly takes the conceptual
truth that relations require something non-relational to give us insight into the
spatio-temporal objects of our knowledge and to entitle us to assert that, since they
must have inner natures, and their inner natures are not spatial, their inner natures
must be simple souls. Kant does not say that Leibniz’s error lies in thinking that he is
INTRINSIC NATURES 
entitled to assert that things have an entirely inner nature; rather, he argues that the
error lies in taking this claim to give us the kind of substantial knowledge of the inner
natures of things that would be required to assert that they are simple soul
substances.
Kant thinks that it is a conceptual truth that relations require something non-
relational, yet he asserts of the spatial objects of our knowledge that they contain
nothing absolutely non-relational. This might be thought to be problematic: Kant
seems to say that an analytic claim fails to hold of appearances. In my view, this
apparent problem in fact shows that Kant is committed to the thesis that things have
entirely non-relational, intrinsic natures. Kant thinks both that logic requires that
independent things have absolutely inner natures and that the objects of our know-
ledge do not have absolutely inner natures. The fact that an analytic claim appears not
to hold of the objects of our knowledge would be a problem if the spatial objects of our
experience were metaphysically fundamental and ontologically self-subsistent—if
they were things as they are in themselves. Rather than taking his relational analysis
of appearance to show that relations are all there is, Kant takes his idealism, with its
relational analysis of appearances, to explain how appearances could be entirely
relational, despite the fact that relations require relata. The problem is how an analytic
claim could fail to hold, and the solution is his idealism: the solution works because it
tells us that when the analytic claim is applied to the spatio-temporal objects of our
knowledge it is not applied to something metaphysically fundamental which exists
complete on its own. The claim that spatio-temporal objects contain nothing non-
relational is compatible with the claim that relations require something non-relational
if spatio-temporal objects are not all that exists. This explanation works only if Kant
has a commitment to things in themselves that have intrinsic properties, and if
appearances are not metaphysically fundamental and complete.
Kant says that it is ‘startling’ to be told that a thing consists of mere relations
(A285/B341). It is startling because it appears to involve asserting something that
does not make sense. He thinks that we can explain away what seems not to make
sense (the existence of things with no inner natures) once we see that the objects of
our knowledge are mere appearances and therefore not ontologically fundamental.
This enables us to show that the claim that the objects of our knowledge consist
entirely of relations is not the (as Kant thinks) problematic claim that everything that
exists consists entirely of relations. As with his appeal to idealism in the context of the
mathematical Antinomies, Kant thinks that we are able to explain something that
appears not to make sense by seeing that the objects of our knowledge are mind-
dependent appearances and therefore not ontologically complete or fundamental.
With respect to the Antinomies, Kant thinks that this enables us to make sense of the
claim that the world in space and time does not have a determinate extent. In the
Amphiboly, he thinks that it enables us to make sense of the claim that the spatio-
temporal world contains relations only. Both these claims would (he thinks) be
problematic if we asserted them of reality per se, or of something ontologically
 KANT ’ S IDEALISM AND HIS REALISM
fundamental, but both can be explained if we see that the objects of our knowledge
are mind-dependent appearances.
At this point, it might be thought that a problem I discussed at the end of
Chapter 3 resurfaces: the problem of making sense of Kant’s referring to appearances
and things in themselves as in some sense the same things. In the Amphiboly, Kant
asserts that the spatial objects of our knowledge have no inner natures. The problem
is that if we think that appearance are appearances of things which have intrinsic
natures, and that the same things which appear to us have a way they are in
themselves, then it might be thought that Kant has to say that spatio-temporal
appearances do have intrinsic natures. Can we say both that the objects of our
knowledge have no intrinsic natures and that the objects of our knowledge are in
some sense the same things as things which do have intrinsic natures? Similarly, as
we saw in Chapter 3, Kant sometimes says that appearances represent things them-
selves (B164; A252) but also denies that appearances represent things as they are in
themselves (A44/B62; A270/B327; A276/B332). When making sense of how Kant can
claim both that appearances do not contain anything absolutely inner and that they
are grounded in things’ absolutely inner natures, it is important to remember that his
claim is not that we can pick out objects with intrinsic natures and objects with
relational natures and then identify them as the same objects. Kant thinks that things
have intrinsic natures but that these natures are no part of what we have empirical
cognition of, or of what science describes. They are thus not part of empirical reality.
Kant thinks that no matter how far science goes in breaking things up and explaining
their inner natures, it will always do this in terms of relational properties, and that
nothing non-relational will ever be given to us, or form part of what we can cognize,
and further, that science (empirical knowledge) has no need for non-relational
properties. In this sense, the objects studied by science do not have intrinsic natures.
At the same time, he thinks that the relational appearances of which we have
knowledge are not ontologically complete or self-subsistent and are grounded in
something non-relational which we cannot cognize. We are not asserting an identity
relation between things that have no inner natures (spatio-temporal objects) and
things that have non-relational natures. Rather, we are saying of things that their
entirely relational spatio-temporal appearances exhaust their empirical reality but are
also grounded in non-relational natures they have independently of us. Though
things have inner natures, spatio-temporal objects do not have inner natures because
they are merely relational appearances of things.
Because of the complications concerning the way Kant presents Leibniz’s view, the
Amphiboly section is not the clearest section of the Critique and perhaps it cannot be
taken to support any one interpretation conclusively on its own. However, some
things are clear and uncontroversial in this section. Kant thinks that the empirically
real world contains nothing non-relational; that conceptual analysis commits us to
thinking that relations require something non-relational; that objects thought merely
through the understanding would be thought to have non-relational, inner natures;
INTRINSIC NATURES 
and that we cannot conclude from these claims that the objects of our experience
have inner natures that are simple soul substances. The controversial point is
whether he thinks, in addition, that the objects of our experience are appearances
of things which have absolutely non-relational, inner natures. We have seen three
considerations which together give strong grounds for attributing this view to him.
One, a commitment to things’ having intrinsic natures is compatible with Kant’s
aims in the Amphiboly: his critique of Leibniz and his view of the connection
between logic and metaphysics. Two, nowhere in the Amphiboly section does Kant
deny that there is a way reality is in itself that grounds appearances; if this were his
view, it seems plausible that he would have made this explicit, given the centrality in
the Dialectic of his account of transcendental metaphysical concepts we unavoidably
posit but which do not correspond to objects of which we have knowledge. Finally,
Kant’s commitment to things having a way they are in themselves is in fact needed to
make sense of how he can reconcile the claim that we are conceptually committed to
something non-relational with the claim that the objects of our knowledge contain
nothing non-relational.
III Causally Inert Intrinsic Natures?

I have argued that Kant’s transcendental idealism includes the claim that we cannot
know the intrinsic natures of things, which ground their relational appearances. So
far, the view I am attributing to Kant is similar to Rae Langton’s interpretation. She
argues that transcendental idealism should be understood as saying that we cannot
have knowledge of the intrinsic nature of reality. Langton goes further in thinking
that we can say more to characterise the way Kant understands the intrinsic natures
we cannot know and to reconstruct his reasons for thinking that we cannot know
them. In particular, she attributes to Kant the idea that things have causally inert
intrinsic natures which we cannot know. I will argue that this positive characterisa-
tion of the intrinsic natures we cannot know is not Kant’s view in the Critique.
Langton contends that Kant’s distinction between noumena and phenomena
should be understood in terms of a distinction between intrinsic and extrinsic
properties,8 that this corresponds to a distinction between causally inert properties
and causal powers, and therefore that Kant’s claim that we cannot know things as
they are in themselves should be understood as the thesis that we cannot know
8
Langton’s account brackets, and does not attempt to explain, the sense in which Kantian appearances
are mind-dependent. She makes this explicit, but she sometimes still speaks as if her account is a complete
characterisation of Kant’s distinction between noumena and phenomena. In the places where she does
discuss Kant’s idealism, she indicates that Kant’s idealism could lie in the fact that his views about things in
themselves may imply that ultimate reality is not physical in nature (Langton 1998: 207). Clearly, this
cannot be the point of Kant’s idealism about appearances. As Langton points out, it would be incompatible
with Kant’s claim that we do not have knowledge of things as they are in themselves, and Kant clearly sees
this (A351–61). And while Kant allows for the possibility that ultimate reality is in some sense mental, his
idealism does not concern things in themselves, but appearances.
 KANT ’ S IDEALISM AND HIS REALISM
things’ causally inert inner properties. She argues that Kant thinks that causal powers
are not reducible to intrinsic properties9 and that this means that intrinsic properties
are causally inert. She further argues that this explains how Kant gets from the claim
that we need to be affected by things in order to have knowledge of them to the claim
that we cannot have knowledge of things in themselves. She attributes to Kant the
following thoughts: we can know things only through their affecting us (through
knowing their powers); things’ powers cannot be reduced to their intrinsic qualities;
and, in fact those powers have no dependence of any sort on things’ intrinsic
properties. Therefore our knowledge of things’ powers tells us nothing about their
intrinsic natures.
Langton starts with the intuitive idea that a thing must have some intrinsic
properties because it must have a way that it would be if other things did not exist:
things must have properties which are compatible with loneliness. I agree with her
that Kant holds this. She then argues for attributing to Kant a more demanding
conception of intrinsicness, one which is compatible with both loneliness and law-
lessness. On this conception, intrinsic properties are those which an object could have
if it were the only thing that existed and irrespective of what the laws of nature were,
or even whether there were laws of nature. It is the second criterion which captures
the irreducibility of causal powers. She further explains the claim that causal powers
are extrinsic and irreducible in terms of superaddition—by thinking about what God
would need to do to create powers. The idea is that to create the objects with their
intrinsic natures requires one act by God; to give things their powers requires a
separate, additional creative act by God. In creating objects with all their intrinsic
natures, God has not determined the powers things have.10 A secular way of putting
the point is simply to say that things’ causal powers are in no sense dependent on
their intrinsic properties; things have both causal powers and unrelated causally inert
properties. As Langton explains the view:
[f]acts about intrinsic properties place no constraints at all on facts about causal powers. God
superadds . . . to the monads powers of relating to each other, and this creative act is entirely
“arbitrary”, and can be omitted or not omitted at God’s pleasure (Langton 1998:118,
my italics).
There are a number of problems with attributing the superaddition view of the
relation between powers and intrinsic properties, and with it the lawlessness criterion
of intrinsicness, to the critical Kant. I will briefly discuss the following problems: one,
9
Langton’s claim that Kant rejects the reducibility of relations generally has been criticised by Van
Cleve, who says that if her argument works, it establishes only the irreducibility of causal relations (Van
Cleve 2002: 225). In fact, this is all her argument needs. See Langton (2006: 184).
10
Langton attributes the superaddition view to Locke as well. There is some debate about this; see
Wilson (1979; 1982), Ayers (1981), and Langton (1998). Against the superaddition reading of Locke, Ayers
argues that Locke’s talk of the ‘want of a discoverable connection’ (IV, iii, 28) expresses a merely epistemic
point and that Locke’s talk of superaddition and God’s good pleasure is meant to mark contingency and
our ignorance (Ayers 1981: 225–6).
INTRINSIC NATURES 
Kant claims that things in themselves ground appearances; two, there is no textual
evidence for Langton’s view in the Critique; and three, Langton’s interpretation is
inconsistent with Kant’s view of the relation between logic and metaphysics—she
attributes to him an argument which draws a metaphysical conclusion from a logical
claim, in just the way he argues we cannot do.
The first problem is that the claim that intrinsic properties place no constraints at
all on things’ powers is hard to reconcile with Kant’s claim that things in themselves
ground or cause appearances. Kant does not just say that things have powers and
that, in addition, they have intrinsic natures; rather, he says that the way things are in
themselves is responsible for their relational appearances. As we have seen, he
frequently speaks of things in themselves as the ‘unknown ground’ of appearances
(A379; A360), and the ‘non-sensible cause’ of representations (A494/B522), and he
says that things in themselves affect us (A44/B61; see also A190/B235, A288/B344,
A393, and A496/B524) suggesting that the way things are in appearances is somehow
the result of the way things are in themselves.11 These would be very strange ways to
express the view that things have properties which have no implications at all for
their powers. In the passage in which Kant first introduces the claim that appearances
are relations, he says that ‘everything in our cognition that belongs to intuition . . .
contains nothing but mere relations . . . But what is present in the place, or what it
produces in the things themselves besides the alteration of place, is not given through
these relations’ (B66–7, my italics). Here he says that something in itself produces the
alteration of place.
The second problem for Langton’s reading is that it simply lacks textual support.
Langton’s textual basis for attributing the irreducibility thesis and the superaddition
version of laws to Kant is largely based on two early works, the New Exposition (1755)
and Thoughts on the True Estimation of Living Forces (1757).12 She does, in addition,
appeal to a number of passages in the Critique, but these do not support her reading.
Some of the passages she appeals to merely support the idea that Kant thinks things
have intrinsic or inner natures. This I agree with but it falls short of the claim that
these intrinsic natures are causally inert and unrelated to things’ causal powers. Most
of the other passages she appeals to are, in my view, concerned with Kant’s claim that
11
See Ameriks, who argues that ‘transcendental affection . . . cannot be either simply phenomenal or
simply non-relational’ (Ameriks 2003: 157).
12
For a discussion of Langton’s reading of the pre-critical works, see Watkins (2005: Ch. 2). Watkins
argues that Kant’s concern is to argue for ‘physical influx’ (real causal influence between substances as
opposed to Leibnizian pre-established harmony) and that his argument is not based on the reducibility of
relations. Kant’s aim is to oppose Leibniz’s view that it is not metaphysically possible for finite substances to
act on each other, and he argues that in a causally isolated substance change could not be explained and is
not possible. This is not the same as saying that the properties that causally isolated substances have place
no constraints on their powers or on what relations they could be in. Watkins argues that even if Kant, at
this stage, thinks that relational properties do not reduce to intrinsic properties, it does not follow that
intrinsic properties must be completely irrelevant to their relational properties, since what relations objects
are in will affect the ways in which they can act on each other.
 KANT ’ S IDEALISM AND HIS REALISM
cognition requires intuition, and not with the reducibility of relations, or the inert
nature of intrinsic properties.
For example, she draws on a passage discussing community at the end of the
Principles (B293–4), which she interprets as asserting that substances conceived as
entirely independent things could not be in community (Langton 1998: 135; see
B293–4). The passage comes from a section in which Kant is claiming that the
categories do not give us insight into objects without outer intuitions. His concern
is with conditions of cognition, and his argument is not the metaphysical claim that
things which have only intrinsic properties cannot interact, but that we cannot
cognize interaction (and persistence, and alteration) without space. He says that
community cannot be comprehended through mere reason, but requires intuition—
and in fact outer intuition in space. Kant argues that it is because Leibniz thinks of
substances as entities conceived by the understanding alone (known only through
concepts) that he is unable to explain how we can cognize community. The passage
therefore does not show that Kant thinks objects’ intrinsic properties are causally
inert, but rather that he thinks we could not cognize objects as being in causal
community without cognizing them as spatial.
Similar problems arise with a passage Langton appeals to from the Third Analogy,
in which Kant argues that we could not perceive the coexistence of isolated sub-
stances (Langton 1998: 136; see B258–9). She reads Kant as saying that the mere
existence of substances which are capable of existing on their own is insufficient for
interaction among them. But this metaphysical claim is not Kant’s concern here;
rather, his concern is with the idea that the coexistence of substances could not be
cognized by us if they were not in mutual interaction. He says that if there were
isolated substances which were not in causal interaction, we could not know them to
coexist; he does not claim that substances have properties which have no implications
for their causal powers.
Langton sees Kant’s criticism of Leibniz’s use of the principle of the identity of
indiscernibles in the Amphiboly as an argument for the irreducibility of relations. She
reads Kant as saying that if relations were reducible, there would be no difference in
spatial properties without a difference in intrinsic properties (Langton 1998: 134–5;
see A272/B328). She then attributes to him the claim that there are differences in
spatial properties without differences in intrinsic properties; therefore relations are
not reducible. Kant’s concern in rejecting the principle of indiscernibility is, however,
not to argue for the irreducibility of relations, but rather to reject the Leibnizian idea
that we can cognize objects through concepts (descriptive criteria) alone, without
intuition (acquaintance). As we have seen, the argument Kant makes in the section is:
if we could cognize object through concepts alone, then objects x and y with the same
properties would be the same object; we cannot cognize objects through concepts
alone, but require them to be given to us in intuition; therefore, objects x and y with
all the same properties can be distinct objects. The argument is not about whether
spatial properties are reducible to intrinsic properties, but about the difference
INTRINSIC NATURES 
between cognizing objects conceptually and cognizing objects in a way that requires
both concepts (descriptive criteria) and intuitions (acquaintance).
The third problem with Langton’s reading is that it is inconsistent with Kant’s view
of metaphysics.13 Like Langton, and unlike deflationary interpreters, I read transcen-
dental idealism as a partly metaphysical position, so the problem is not, in my view,
simply that she attributes metaphysical claims to Kant. More specifically, the prob-
lem is that her argument for attributing the irreducibility thesis to the critical Kant is
inconsistent with his view of the relation between logic and metaphysics. Her
argument is that Kant thinks that laws of nature are not logical consequences of
the intrinsic natures of substances and therefore that facts about intrinsic properties
place no constraints at all on facts about causal powers (Langton 1998: 118). She
points out that Kant thinks that impenetrability does not follow logically from
solidity; she argues that this means that the connection between them is not neces-
sary, and therefore that objects could have had the same intrinsic properties they
actually have while having completely different causal powers. But this argument
ignores Kant’s distinction between logical possibility and real possibility (A244/B302;
Bxxvin). Kant clearly thinks that the former is not enough to establish the latter: real
possibility is given by coherence with space, time, and the principles of experience. In
particular, Kant thinks that we cannot do metaphysics by thinking about logical
possibility only. When we are talking about things in themselves all we have is logical
possibility, but this does not entitle us to draw conclusions about what is really
possible. The fact that impenetrability does not follow logically from solidity does not
tell us that it is really possible to have impenetrability without solidity. In basing
metaphysical conclusions on logical possibility, Langton is doing the kind of meta-
physics which Kant explicitly rules out (which is not to say that he rules out all
discussion of metaphysics).
Langton’s position is meant to be a reconstruction and not an argument Kant
explicitly presents. She takes positions Kant may have held in earlier works and uses
them to explain how Kant gets to one of his conclusions in the Critique. This is a
legitimate strategy, and the argument she attributes to Kant is interesting. Unfortu-
nately, the textual evidence she finds in the Critique supports only the claim that
things have intrinsic natures, not the further claim about the irreducibility of
relations, much less the more radical ‘no implications’ claim, and her central argu-
ment for irreducibility turns on a move which ignores Kant’s distinction between
logical and real possibility. Finally, a reconstruction of an argument which might
explain an author’s conclusions gains interpretative power if the author otherwise has
no argument; this is not the case. Kant has an explicit argument for the claim that we
cannot know things as they are in themselves, which we examined in Chapter 8. He
thinks that we can cognize things only as presented through forms of intuition and he
13
See Carr (2000: 110) and Van Cleve (2002). Langton argues that there is textual support for her
reading in the Critique, in the Amphiboly and the Analogies.
 KANT ’ S IDEALISM AND HIS REALISM
argues that our a priori forms of intuition do not present us with features of mind-
independent reality. It is not simply because we need to be affected by things that we
cannot know them as they are in themselves, but because being affected by things will
not get us cognition of these things unless our representations are arranged in the a
priori form of intuition, and because (Kant thinks) a priori intuition cannot present
IV Intrinsic Grounds
Along with Langton, I have argued that Kant’s transcendental idealism includes the
claim that we cannot know the intrinsic natures of things. In my view, Kant thinks
that things have primary, categorical, non-relational, intrinsic natures and that we
cannot know these natures. Langton has a specific account of intrinsic properties—
that they are causally inert and have no implications for things’ causal powers. I have
argued that it is textually and philosophically problematic to attribute this view to
Kant. Langton’s interpretation attributes to Kant more knowledge of things as they
are in themselves than he allows: knowledge that intrinsic properties are causally
inert would be metaphysical knowledge of things in themselves. However, it might be
thought to be a problem for my position that if we have no knowledge at all of
intrinsic properties, and of the nature of the relation between them and powers, the
upshot of saying that things have intrinsic natures that ground their appearances
seems to become unclear. This is my concern in the final section of this chapter. As
I understand Kant’s position, there are two reasons for thinking that there are
intrinsic, non-relational natures. One, they are part of our conception of what it is
for a thing to exist: we cannot make sense of the idea that something could exist and
have only relational qualities (this is common to my reading and Langton’s). Two,
intrinsic, non-relational properties ground relations; it is at least partly in virtue of
things’ intrinsic natures that they have the relational properties that they do. In my
view, Kant is committed to the thesis that things in themselves play both these roles,
and he does not think we could make sense of the essentially manifest, essentially
relational nature of appearances without this. As I have already noted, Kant does not
have a precise account of relational properties, and he talks interchangeably of
relations, powers, and forces; I follow him in this.
The first part of the conception of intrinsic properties that I attribute to Kant is
that they are part of what we are committed to when we think of a thing existing. As
Locke says of the primary qualities ‘[w]e have by these an idea of the thing as it is in
itself ’ (1690: II, viii, 23). However, for Locke categorical or primary properties give us
an understanding of what it is for there to be an object by giving us insight into the
nature of this object, and why it has the relational properties it has. This cannot be the
role that intrinsic, non-relational qualities play for Kant, since Kant thinks we do not
know anything about things’ intrinsic natures or about how their intrinsic natures are
responsible for their having the relations they do. The thought that non-relational
INTRINSIC NATURES 
qualities form an essential part of our conception of what it is for there to be an object
can be understood in a more minimal sense: the idea is simply that a thing cannot
exist and have only relational properties (whether or not we are able to have
knowledge of non-relational properties).14 This thought is expressed by Langton’s
loneliness criterion: she says that ‘a substance is the kind of thing that can exist on its
own: it can exist and be lonely. But nothing can exist without having properties. If a
substance can exist on its own, it must have properties that are compatible with
existing on its own. If a substance can be lonely, it must have properties that are
compatible with loneliness’ (Langton 1998: 19, 22).
Another way of putting this thought is to say that there must be a possible
description of a thing—a way that it is—in a world in which it is the only thing
that exists. The problem with ascriptions of powers, or relational properties more
generally, is that they tell us about things in terms of other things: ascribing a power
to a thing involves saying what it can do to other things (often with additional
reference to environmental conditions, and possibly to causal triggers). If we want to
describe a thing in the world in which it is the only thing that exists, we cannot
describe it in this way. Suppose that an object x has as its only property the bare
power to repel y-type objects. Imagine a world in which x is the only object that
exists. It seems that there is no way to characterise x in this world: there is no
description of an object which has as its only property an unmanifested power in the
world in which it is the only thing that exists. We can describe it only in terms of what
it would do if other things were present; there is no description of the nature of the
object as it is in the world that it is in, but only counterfactual descriptions of what
would be the case if it were in a different world. Blackburn says ‘[t]o conceive of all
the truths about a world as dispositional, is to suppose that a world is entirely
described by what is true at neighbouring worlds, and the result is that there is no
truth anywhere’ (Blackburn 1990: 64). Thus, a reason for thinking that things must
have intrinsic, non-relational properties is that we simply cannot understand what it
would be for a thing to exist and yet for there to be no possible description of it in the
world in which it is the only thing that exists.
This line of thought could perhaps be resisted by denying that there are objects
which could exist on their own, without being in relations to some other things.
Perhaps an object like x, whose only property is the power to repel y objects, could
exist only in a world in which y objects exist. Against this, John Foster argues that
without an understanding of x and y objects as having a way that they are, intrin-
sically, independent of their powers, we will be unable to characterise what it is that
the power to repel y is a power to do. We will therefore be left without a character-
isation of either things or their powers. He says that ‘if all the fundamental powers
were construed in this way [as bare powers], there would be no physical items in
14
See Foster (1982: 68–9), Evans (1985a), Van Cleve (1995: 102–3), Langton (1998: 19, 22), and Smith
and Stoljar (1998: 85–111).
 KANT ’ S IDEALISM AND HIS REALISM
terms of whose behaviour the content of the powers could be specified, and conse-
quently . . . in the last analysis, there is nothing which the powers are powers to do’
(Foster 1982: 68). He argues as follows. Assume that the fundamental particles are
identical atoms which have a number of causal powers, and suppose that resistance is
the essential property of these atoms. If powers are not grounded, each atom is
simply (barely) a mobile centre of impenetrability or resistance. If we want to know
what it is to which mobile centres of impenetrability are impenetrable, the only
answer can be that they are impenetrable to other spheres of impenetrability. Foster
argues that this renders us unable to explain what impenetrability or resistance is:
when we ask what resistance is the power to do, the answer is simply to resist
resistance.15 This, he says, cannot be an answer, as it appeals to the notion of resistance,
which it is supposed to explain. The idea is that we cannot understand what resistance
is if we can characterise it only as the power to repel the power to repel the power to
repel . . . and so on.16 But resistance was, by hypothesis, the only thing that was
supposed to characterise the objects, so if we cannot understand resistance, we cannot
characterise the objects at all. Foster concludes that if each thing is fully characterised in
terms of facts about other things, there will be no facts about any things; we can make
sense of the idea of there being things, and of things having powers, only by supposing
that at least some powers are grounded in categorical features of things (Foster 1982:
68–9).17
We find a similar argument in Hume, who thinks that the Lockean primaries of
motion and solidity are not in fact categorical properties,18 but rather properties
which are understood in terms of effects on other properties. He says that instead of
explaining objects by thinking of such properties as their fundamental properties, ‘we
utterly annihilate all these objects’ (1739 40: 228). The objects are annihilated because
nothing primary or categorical is left. He says that we cannot characterise an object
only in dispositional terms:
Add to this, that, properly speaking, solidity or impenetrability is nothing, but an impossibility
of annihilation, as has already been observ’d; For which reason, ’tis the more necessary for us
to form some distinct idea of that object whose annihilation we suppose impossible. An
15
The point of this argument is not to deny that resistance could, perhaps, be metaphysically
fundamental and not further explainable; it is that it must be at least be characterised in terms of what it
does to something.
16
Similarly, as Smith and Stoljar argue, someone who wants to characterise Bill’s political views by
telling us that they are to the left of Bob’s, and to characterise Bob’s as being to the left of Ollie’s, and Ollie’s
as being to the left of Al’s, and so on, will never succeed in telling us what anyone’s political views are
(Smith and Stoljar 1998: 103–4).
17
An opponent may argue that the centres of impenetrability can be characterised as properties of
space. Whether this view involves something non-relational will depend on whether or not space is
understood substantivally. See Lucas (1976), who argues that a pure field ontology is possible only for a
substantivalist about space.
18
Hume argues that our understanding of solidity is merely in terms of impenetrability, the idea that
‘two objects, which being impell’d by the utmost force, cannot penetrate each other’ (1739 40: 228).
INTRINSIC NATURES 
impossibility of being annihilated cannot exist, and can never be conceived to exist, by itself,
but necessarily requires some object or real existence, to which it may belong (1739 40:
229–30).19
The thought is that objects cannot be characterised entirely in terms of non-categorical

properties, because ways of affecting other objects cannot exist on their own. This
conception of objects is contested: the idea that a thing should be able to exist and be
lonely will be rejected by ontological structural realists. However, my aim is not to
establish it, but to bring out the philosophical point of this conception of intrinsic
properties and to argue that it is plausible to attribute it to Kant.
So far, the suggestion has been that a reason to think of things as having intrinsic
properties is the claim that describing a thing in terms of its powers cannot be a
complete description of an object: an independent thing (object, substance) is
something over and above the ways it would interact with or relate to other things.
Langton’s reading of Kant contains a further commitment about how these intrinsic
properties should be understood: the idea is that things have properties which are
distinct from their powers, properties which are not powers, and which have no
implications for things’ powers. The claim that things cannot be fully described in
terms of actual and possible relations to other things does not have to be understood
in this way. One alternative is to say that powers are not distinct from those
properties which things have as they are in themselves, but rather correspond to
distinct ways of describing or knowing things’ intrinsic properties. This kind of view
is presented by Foster, who argues that a power’s description is always a description
of a thing in terms of other things rather than a description of the thing itself, and
calls this an opaque description (Foster 1982: 87). Daniel Warren makes a similar
point in using a contrast between what he calls ontological priority and epistemo-
logical priority. On this account, the ontological priority of inner properties is the
idea of a thing’s having properties which are compatible with its having no outer
relations. The epistemological priority is the idea that inner properties cannot be
understood through the relations that things can be in: through grasping things’
powers we do not grasp the intrinsic natures in virtue of which things have their
powers.20 Warren argues that ‘if we can characterise an inner property merely as . . . a
power to produce certain effects, then it lacks epistemic priority’ (Warren 2001b: 51).
He attributes this view to Kant. On this reading, to say that we know relational
19
Jackson says that ‘holding that the nature of everything is relational cum causal . . . makes a mystery of
what it is that stands in the causal relation’; he thinks that we should, therefore, acknowledge that it is
possible that ‘a large part (possible all) of the intrinsic nature of the world is irretrievably beyond our reach,
but that all the nature we know supervenes on the (mostly or entirely) causal cum relational nature that the
physical sciences tell us about’ (Jackson 1998: 24). Harré and Madden think that the explanatory bedrock of
science seems to be forces or fields of potential, but they acknowledge that it is possible that there is
something further underlying this (Harré and Madden 1975: Ch. 9).
20
Langton expresses a similar point by saying that powers are conceptually relational, since a power
ascription describes a thing in terms of other things (although she argues that Kant sees powers as
relational in a stronger sense) (Langton 1998).
 KANT ’ S IDEALISM AND HIS REALISM
properties and not powers is not to say that in addition to their powers things have
properties which have no implications for their powers, but rather that we do not
grasp things’ natures transparently in knowing their powers. Another alternative is to
say that having a causal power is a second order property—that of filling a certain
causal role—which is ‘realised’ by a first order intrinsic property.21 Considerations
motivating this view include the fact that powers can be variably realised, and that
things can have different powers in virtue of the same intrinsic nature. Suppose that
object M has the power to nourish A and the power to poison B, and has both these
properties in virtue of its intrinsic nature EFG. For example, bread has the power to
nourish us in virtue of its chemical composition; the same chemical composition
might be responsible for bread’s being inedible to other creatures. Suppose that
objects M and N each has the power to nourish A, but M has this in virtue of
intrinsic nature EFG, whereas N has it in virtue of intrinsic nature UVW.22
In order to understand Kant’s view that things have a way they are in themselves
which grounds appearances, we do not need to choose between these accounts,
rather, we need two theses which are common to a number of different accounts of
the relation between powers, dispositions, and intrinsic natures. The first has already
been discussed: the claim that objects cannot be fully described by describing their
powers. The second, to which I now turn, is the claim that objects have their powers
in virtue of their intrinsic natures.23 Another way of expressing this idea is to say
that intrinsic natures ground powers.24 This idea is expressed by Harré and Madden,
who say:
A power is specified in terms of an effect which is not part of the description of the nature in
virtue of which the power is possessed. (Harré and Madden 1975: 11).
To ascribe a power to a thing is to say something about what it will or can do, but it is not to
assert any specific hypothesis about the nature of that thing. To ascribe a power to a thing
asserts only that it can do what it does in virtue of its nature, whatever that is. It leaves open the
21
See Smith (1977: 445), Prior, Pargetter, and Jackson (1982), Ellis and Lierse (1994: 33), and Martin
(1994).
22
Structural realists could argue that these powers are grounded in further powers.
23
See G. J. Strawson (1987: 260) and Harré and Madden (1975). Both these theses are compatible with a
variety of views about how exactly powers relate to intrinsic properties: they can be held by those who think
powers are intrinsic properties known opaquely, those who think that powers are second order role
properties which are realised by intrinsic properties, and a number of other views. The second thesis is,
however, at odds with Langton’s Kant, who holds that objects have causally irrelevant intrinsic natures and
superadded bare powers and who therefore does not think that objects have their powers in virtue of their
intrinsic natures.
24
The idea of grounding has recently been subject to a lot of philosophical attention, with philosophers
defending the idea that it is a legitimate and useful metaphysical notion. See, for example, Schaffer (2009),
Rosen (2010), and Clark and Liggins (2012). Clark and Liggins argue that grounding is a metaphysical
notion, involving asymmetric dependence between relatively more and relatively less metaphysical fun-
damental things or classes of facts, which corresponds to the locution ‘in virtue of ’. Rosen argues that ‘the
project of rehabilitating the grounding idiom is analogous, as I see it, to the project begun in the 1960s for
the rehabilitation of traditional notions of necessity and possibility—a project that is now more or less
complete, and whose value is beyond dispute’ (Rosen 2010: 34).
INTRINSIC NATURES 
question of the exact specification of the nature or constitution in virtue of which the thing,
person or material has the power (Harré and Madden 1975: 87).
To ascribe a power is to ascribe a disposition to a specific form of behaviour, together with an
unspecific reference to the nature or constitution of the thing or material concerned (Harré
and Madden 1975: 92).
The claim that intrinsic properties ground powers does not require thinking that
things have properties which are not powers (which are causally inert).25 One way of
understanding the claim is that intrinsic natures explain causal powers. This, as we
have seen, is part of the role of Lockean primary qualities: they enable us to
understand why objects have the causal properties they do. However, this cannot
be the point for Kant, since, on his account, we have no knowledge of things’ intrinsic
natures, including of how they ground things’ relational appearances. I argue that we
can still make some philosophical sense of the idea that objects have their powers in
virtue of their intrinsic natures. To illustrate this, I start with a case in which we do
have knowledge of a ground of a power.
Suppose we allow, for the moment, that chemical composition is an intrinsic
property,26 and that it is in virtue of its chemical composition that bread has the
power to nourish us (and the power to poison some other creatures). This is different
from saying that the intrinsic nature (the chemical composition) and the power are
the same property, and also different from saying that the object has a causally inert
intrinsic property and, unrelatedly, a power. Part of what the claim expresses is the
thought we have already discussed: the idea that a thing cannot simply be its possible
effects on and relations to other things. In describing bread as having the power to
nourish us, we do not describe the bread as it is in itself (its chemical composition),
rather, we describe it in terms of effects on us (and, perhaps, relevant environmental
factors, including relations to other powers). A complete description of bread cannot
describe it simply in terms of its effects on us. In addition to this, we have the thought
that the chemical composition is responsible for the power: it is because bread has its
chemical composition that it has its power to nourish us.
The idea that intrinsic natures ground powers is not that they causally explain
powers. There are causal explanations involving the chemical properties of bread and
their interaction with human bodies, but the relation between the chemical property
and bread’s power to nourish us is not a relation of one thing’s causing another. If the
25
Shoemaker argues that all of an object’s intrinsic properties are causal powers (Shoemaker 2003a:
115; see also 2003b: 332). His central use of ‘intrinsic’ is in opposition to what are called mere Cambridge
properties, the idea being to distinguish metaphysically significant properties of an object, such as its causal
powers, from the indefinitely many relational properties which seem to be connected only artificially to the
thing, such as its being 50 miles from a burning barn. See Shoemaker (2003b: 332). See Humberstone
(1996) for some of the difficulties of giving a principled way of drawing this distinction.
26
It is relatively intrinsic, though Kant’s view is that when we investigate it we find that the chemical
composition, too, can be characterised only in terms of powers, and therefore in terms of effects on
other things.
 KANT ’ S IDEALISM AND HIS REALISM
relation is not causal, it might be wondered what the upshot is of saying that the
chemical composition of the bread grounds its power to nourish us. In contemporary
literature on the topic, grounding is understood as a non-causal notion that captures
a way in which one kind of thing or class of facts explains another, where this
explanation involves the dependence of something less metaphysically fundamental
on something more fundamental.27 Talk of grounding is said to pick out an order of
dependence and ontological priority. So the idea that bread’s power to nourish us is
grounded in its chemical composition does not mean that its chemical composition
causes the power, but rather captures the idea that the chemical composition is a
relatively more fundamental feature of bread than its power to nourish us, on which
its power to nourish us depends.
Understanding the claim that intrinsic natures ground relational properties
(including powers and dispositions) as capturing an order of ontological priority
and dependence is useful for understanding Kant. However, in the contemporary
literature on grounding, some of those who defend the useful and intuitive nature of
this relation also appeal to the thought that grounding is a relation in which one thing
(property, class of facts, etc.) explains another.28 Kant, of course, thinks that we do
not and cannot explain how things in themselves ground appearances. There is a
reasonable concern that if we give up on the idea that their intrinsic natures explain
their powers we might lose our grip on the idea that intrinsic natures ground powers,
or that things have their powers in virtue of their intrinsic natures. In response to
this, I argue that if we can make sense of the idea of grounding in the case in which we
have insight into the way the ground explains the power, we can carry at least part of
this account over into cases in which we do not understand how the ground explains
the power. In the case where we do not understand how the ground explains the
power we still have the idea that there is an order of dependence, a relation of relative
metaphysical fundamentality, and that the ground is responsible for the power.
Consider a case where our ignorance is contingent and may be temporary. Suppose
an object has a power and we do not yet know what aspect of its intrinsic nature
explains its having this power. If we allow for the idea of grounding in cases where we
grasp how an intrinsic property explains a power, we can also allow for the idea in the
context in which we do not yet know what grounds the power. What this commits us
to is thinking that there is some more fundamental nature on which the power
depends, which, if we knew about it, would explain why the thing has the powers it
has. We can continue to maintain this commitment even if it turns out to be
contingently true that we will never understand the relation between the ground
and the power. If intrinsic natures ground powers in cases in which we happen to be
able to understand how they are responsible for powers, intrinsic natures can still be
thought to ground powers in the case where we never manage to understand this. We
27 28
Schaffer (2009), Rosen (2010), and Clark and Liggins (2012). Clark and Liggins (2012).
INTRINSIC NATURES 
do not lose our entire grip on the grounding relation in this case. Even though we
lack the specific explanation, we still have the idea that the thing has its powers in
virtue of its intrinsic nature, the idea that the intrinsic nature is relatively metaphys-
ically fundamental, and that the powers depend on it. These relations of dependence
and fundamentality do not cease to exist just because we happen not to reach the
explanation. Kant’s position adds to this that the limitation on our explanations (on
our grasp of the relation between intrinsic natures and relational features) is not
contingent. This addition does not undermine the features which have been central
to the grounding relationship we have appealed to in the cases in which we do
understand the grounding relation: dependency and fundamentality. We can even
allow that the grounding claim still includes a commitment to there being something
which explains why things have their relational natures; but this explanation is of a
sort which is beyond our capacity to grasp.
I started this section with the idea that a description in terms of powers or relations
cannot be a complete description of an object: a thing must have a way that it is
independently of its actual and possible relations to other things. We added to this
the idea that powers are ultimately grounded in properties things have in themselves.
Kant’s view can be understood by taking these two claims and adding to them the
idea that we can have knowledge only of relational features of reality, and cannot, in
principle, have knowledge of the fundamental non-relational natures in virtue of
which things have the relational features they do. Powers can of course be grounded
in more fundamental powers, as the case of bread and chemical composition
illustrates, and Kant allows that science gives us increasing knowledge of this kind
of grounding. However, he thinks that there cannot be relations all the way down;
there must ultimately be intrinsic natures which ground powers, but we cannot know
these intrinsic natures.
If some relations are grounded in non-relational natures, but science is limited to
relational explanations, then there will be a fundamental grounding relation which
we cannot explain at all and cannot even specify in physical terms. The notion of
grounding here seems extremely unsatisfying.29 However, while it is unsatisfying, it is
not completely mysterious, because we can get some grasp of it through the examples
of grounds of which we do have knowledge. These examples enable us to specify
some formal features of the grounding relation, such as the fact that it involves
dependence and metaphysical fundamentality; these features are common to the
grounding relation both in cases in which we grasp the way one property explains
another and in cases in which we do not. Although we have no idea how things’
intrinsic natures ground their powers, we still have the following ideas: things as they
are in themselves are more metaphysically fundamental than things as they are in
29
As Foster notes, ‘there is a natural reluctance to postulate physical properties which are empirically
inscrutable and not even transparently specifiable in physical terms’ (Foster 1982: 68).
 KANT ’ S IDEALISM AND HIS REALISM
relation to other things (including relations to us); and things appear to us as they do,
in part, because of the way they are in themselves.
While the position I attribute to Kant contains some mystery, none of the
alternatives is free from problems. Kant’s position results from combining a certain
(fairly common) view of science and the explanations it gives (that they are essen-
tially and only relational) with an intuitive idea of what it is for a thing to exist (that a
thing cannot be fully described by saying things about its effects on other things). An
alternative to his position would be to abandon this intuitive idea and say that, if our
best science does not invoke non-relational properties then there is no reason to
think that anything non-relational exists: this is the position of ontological structural
realism.30 As we have seen, this view is thought by many to be unintuitive and
philosophically problematic. Kant thinks that it is a conceptual truth that relations
require something non-relational, so he does not think we can give up this part of the
position. In addition, he thinks that taking the claim that science gives us knowledge
only of relations as a reason to think that relations are all there is would involve
mistaking science for metaphysics, something which he thinks we have no justifica-
tion in doing.
Some of those who argue that Kant has an entirely relational ontology think that
attributing to him the idea that relational features of reality are grounded in some-
thing non-relational would be attributing to him a transcendent metaphysical claim
of the sort he rejects. In my view, in contrast, he would take the argument that leads
to the view that reality is entirely relational to be making an illegitimate, transcendent
metaphysical claim: if we take an entirely relational ontology to follow from the thesis
that science is entirely relational we will be doing transcendent metaphysics, by
mistaking science for metaphysics. This is Kant’s account of the way empiricists do
transcendent metaphysics, and he thinks it has as little justification as rationalist
transcendent metaphysics, which makes claims about non-spatio-temporal objects.
In his critique of transcendent metaphysics he argues, against the rationalists, that we
cannot have knowledge of non-spatio-temporal, non-sensible objects, objects which
are not presented to us in intuition; at the same time, he argues, against the
empiricists, that we should not take science to give a complete account of reality.
The fact that we cannot know something through science could be taken to show that
this thing does not exist only if we had reason to think that we can cognize everything
that exists. This would require a form of anti-realism that applies to all of reality (not
just empirical reality) and it is not Kant’s view. The claim that things in themselves
ground appearances does not involve asserting any insight into the nature of non-
spatio-temporal objects: we have no insight at all into the specific nature of this
30
Another alternative is to agree that science gives us only relations, and that reality can’t consist of
relations only, but think that we can have knowledge of intrinsic properties that are given in some way
other than scientific explanations. Kant does not seem to consider this example, because he seems to
assume that spatial properties are relational.
INTRINSIC NATURES 
relation or into the nature of things in themselves. In contrast, taking the claim that
science gives us knowledge only of relations to support thinking that relations are all
that exist involves taking what we can have knowledge of to exhaust reality. Kant
thinks that this is something we have no justification in doing. And as we have seen,
Kant himself simply does speak of things in themselves as causing, producing, and
grounding the relational appearances of which we can have knowledge. He says:
the transcendental object that grounds both outer appearances and inner intuitions is neither
matter nor a thinking being in itself, but rather an unknown ground of those appearances that
supply us with our empirical concepts of the former as well as the latter (A379–80, my italics).
The sensible faculty of intuition is really only a receptivity for being affected in a certain way
with representations . . . which, insofar as they are connected and determinable in these rela-
tions (in space and time) according to laws of the unity of experience, are called objects. The
non-sensible cause of these representations is entirely unknown to us. (A494/B522; see also
A44/B61, A190/B235, A251–2, A288/B344, A393, and A496/B524)
That the view I am attributing to Kant contains some mystery (as do the alternatives
to his position) is not a reason to think it is not his position. It is not as if we have no
idea at all what could be meant by a grounding relation in this context, and the view
of grounding I have attributed to Kant is consistent with his key arguments and
claims in the Critique. Further, in addition to explaining a part of transcendental
idealism, this discussion is of broader philosophical significance, since Kant’s con-
clusion follows from assumptions which are more widely held than is transcendental
idealism. The claim that we cannot know the intrinsic nature of reality follows from
accepting both that empirical cognition and science characterises things in terms of
powers or relations (or, more generally, that the only properties of things we can
know are powers or relational) and that at least some powers must be grounded in
properties which are not powers.31 From these two claims, it follows that science
cannot explain all of reality, which, I have argued, is a central part of Kant’s
As we have seen, Kant distinguishes between things as they appear to us and things
as they are in themselves, and says that things in themselves ground appearances. He
also says that we cannot cognize things as they are in themselves, which includes
saying that we can have no understanding of how they ground appearances. I have
argued that there is a straightforward interpretation of this which has both a
metaphysical and an epistemological component, which does not involve attributing
to Kant a belief in the positive noumena he repudiates, and which makes sense of his
claim that things in themselves ground appearances. Kant thinks that things have a
nature, a way they are, independently of their relations to other things including
relations to us and our cognition, which is more metaphysically fundamental than
31
See Foster (1982: 68–9), Evans (1985a),Van Cleve (1995: 102–3), Langton (1998: 19, 22), and Smith
and Stoljar (1998: 85–111).
 KANT ’ S IDEALISM AND HIS REALISM
the relational appearances which we cognize, and on which relational appearances

depend. In this chapter we have seen one part of Kant’s realism: his commitment to
intrinsic natures that are entirely independent of us. In Chapter 9, we saw another
part of his realism: his empirical realism which holds spatio-temporal objects to exist
outside of us and independent of particular instances of our experiencing them. Kant
takes both of these realist parts of his position to be inseparable from his idealism. It
is because appearances are merely relational that they must be grounded in some-
thing other than appearances. Empirical reality is real: spatio-temporal objects exist
outside us, and independent of particular events of our experiencing them, and their
specific natures are not up to us. However, empirical reality (the world of objects in
space and time) is not ontologically fundamental (it is entirely—doubly—relational)
and is not complete. It does not exist as a determinate totality. Kant thinks that this
would not make sense if empirical reality were all that existed: reality is complete and
determinate. Similarly to British anti-realists, he thinks that the spatio-temporal
world we cognize is not experience-transcendent; it does not exist independent of
the possibility of our experiencing it. But like realists he thinks that limiting reality
itself to the conditions of our cognition is entirely unjustified. Like American anti-
realists, he thinks that we should not think of science as describing a non-observable
reality; rather, the world that science describes is the manifest world of human
experience. Like epistemological structural realists, he thinks that science gives us
knowledge only of relational features of reality, and tells us nothing of the intrinsic
natures that ground these relations. Science is thus, in principle, an incomplete
account of reality and its descriptions do not give us ultimate insight into why things
have the relational natures they do.
11
The Transcendental Deduction
Relation to an Object
I Introduction
I have argued that Kant’s idealism should be understood as saying that our cognition
is limited to essentially manifest features of reality, that he thinks such features are
empirically real (not merely in our minds or constructions out of what is merely in
our minds), and that he thinks they are grounded in a way things are independently
of us, which we are unable to cognize. I have argued that Kant takes his idealism to
follow from considerations concerning the need for intuition in cognition together
with considerations concerning the nature of a priori intuition, and that empirical
reality is limited to what can, in principle, be presented to us in an intuition. This
chapter adds the final piece to my account of Kant’s idealism and his realism, as well
as to the role of intuition in his idealism, by examining (part of) his argument in the
Transcendental Deduction of the categories, in order to evaluate the role of a priori
concepts in Kant’s idealism. In the Deduction, Kant argues that certain a priori
concepts, which he calls the categories, necessarily apply to the objects of experience.
This argument is key to the account he will then present of how metaphysics is
possible. Both his explanation of the role of the categories and the kind of meta-
physics he takes to be possible are tied up with his idealism; this chapter and
Chapter 12 present my account of how this works.
Kant argues in the Deduction that the application of the categories is necessary for
our thought to have what he calls ‘relation to an object’. This can be read as
expressing the idealist thought that what counts as an object depends on our
concepts. I argue against this reading. In support of my view that Kant’s idealism is
based specifically on concerns to do with intuition, I argue that his central argument
in the Deduction is epistemological. I argue that his concern with ‘relation to an
object’ in the Deduction is not about what it takes for us to have experience of
perceptual particulars (to be presented with objects), but about the conditions of
referential thought, and that the problem he raises with respect to referential thought
is one he would take to be a problem also for realists. I show that seeing the central
argument of the Deduction as epistemological is compatible with seeing Kant as
having a robust form of idealism, and with seeing the Deduction as central to Kant’s
 KANT ’ S IDEALISM AND HIS REALISM
project in the Critique. In giving an account of ‘relation to an object’ this chapter also
completes the argument in Chapter 7, where I argued that intuitions present us with
perceptual particulars. There I pointed out that one of the central reasons commen-
tators have for denying this account of intuition is Kant’s talk about the categories
giving us relation to an object: giving an alternative account of what he is concerned
with here is therefore necessary to defend my account of the role of intuition. In the
final chapter I then explain how I take the central, epistemological argument in the
Deduction to relate to Kant’s idealism and his explanation of the possibility of
metaphysics. This completes the argument presented concerning cognition of syn-
thetic a priori claims in Chapter 8.
This chapter looks at only part of the argument in the Transcendental Deduction
of the categories. My aim is not to give a complete account of this enormously
difficult, dense, controversial, and notoriously obscure argument. In particular, I say
very little about a (the?) central feature of the argument in the Deduction: the role of
self-consciousness (what Kant calls the transcendental unity of apperception). I do
not engage with the enormous literature on the Deduction and the enormous
number of ways in which the strategy, aim, starting point, and premisses of the
argument have been understood. My limited aim is to say enough about the argu-
ment in the Deduction, as I understand it, to give an account of the role of a priori
concepts in Kant’s idealism. My focus is on how to understand Kant’s talk of ‘relation
to an object’ and the role the categories play in providing ‘relation to an object’.
In the Transcendental Deduction of the categories, Kant’s aim is to vindicate
our use of certain fundamental a priori concepts, such as the concepts of substance
and cause. He is not trying to show that we do use these concepts (which he thinks
is not in doubt), but that we are entitled to use them; that our use is justified
(A84–5/B116–18). His strategy is to show that they are conditions of the possi-
bility of experience (Erfahrung) of objects.
In addition to vindicating the categories, the Deduction also provides part of the
answer to the question with which Kant opens the Critique—whether and how
metaphysics is possible. Metaphysics, Kant thinks, consists of synthetic a priori
propositions. The first part of Kant’s answer to the problem of metaphysics is that
synthetic a priori cognition is possible if we have a priori intuition. As an intuition, it
gives us its object; since it is a priori, it can give an object to us independently of
anything affecting our senses. This explains how synthetic a priori mathematical
propositions can concern objects that are given to us, and can thus qualify as
cognition. The objects such propositions are about are given to us in pure intuition:
they are constructions in pure intuition. This is not enough to explain the possibility
of metaphysical synthetic a priori propositions, however, because such propositions
do not concern objects that are given to us in pure intuition. The problem with these
metaphysical claims is that they relate to objects ‘without any conditions of sensibil-
ity’ but, unlike mathematical claims, they ‘cannot exhibit any object in a priori
intuition’ (A88/B120). These claims will not qualify as cognition unless their objects
THE TRANSCENDENTAL DEDUCTION 
can be given; since their objects cannot be given purely in a priori intuition, they
must be given in empirical intuition. This means we need an explanation of how a
priori concepts relate to empirical objects that are given in empirical intuition. And
Kant says that a transcendental deduction of the categories is indeed the ‘explanation
of the way in which concepts can relate to objects a priori’ (A85/B117).
In explaining how a priori concepts can relate to objects given in empirical
intuition, Kant demonstrates that our use of the categories is justified. At the same
time, he establishes and explains the kind of metaphysics that is possible for us and
rules out every other kind of metaphysics. Kant argues that a priori concepts are
justified as conditions of the possibility of empirical cognition and can be justified
only in this way. This enables us to show how one kind of metaphysics is possible: we
can cognize synthetic a priori principles containing the categories if they are condi-
tions of the possibility of empirical cognition. It also rules out other kinds of
metaphysics: a priori concepts can be used for cognition only of objects that are
given to us in intuition, and this means that we cannot use them for cognition of such
objects as Leibnizian monads, Cartesian souls, or God. Against the rationalists, Kant
argues that a priori concepts cannot give us knowledge of a special kind of (non-
spatio-temporal) object; against the empiricists, he wants to show that they do play a
role in our knowledge of empirical objects. He wants to show both that this role is
necessary for cognition of spatio-temporal objects and that it is confined to cognition
of spatio-temporal objects.
A central part of Kant’s strategy for showing that a priori concepts are conditions
of the possibility of experience of objects is to argue that they are rules for what he
calls the synthesis of intuition. Synthesis is a central notion in the Deduction.
Synthesis is a way of unifying representations. Kant thinks that all concepts are
rules of synthesis (rules governing ways in which we unify representations), and he
argues that synthesising what is given to us in intuition in a way which is governed by
the categories is a condition of the possibility of empirical cognition of objects. He
argues that pure or a priori synthesis introduces a unity to intuition without which it
is not possible to cognize an object and without which our thought lacks what he calls
‘relation to an object’. So, Kant’s aim is to show how a priori concepts can relate to
empirically given objects and his way of doing this is to show that the application of a
priori concepts, as a priori rules of synthesis, is a condition of thought’s having
relation to such objects. Kant thinks that this is the only way we can vindicate the
use of a priori concepts, and it follows from this that where a priori concepts are not
used as rules for the synthesis of empirical intuitions they do not give us cognition of
any objects.
There is nothing about the argument in the Deduction that is clear or uncontro-
versial, including Kant’s aim, his assumptions, and his premisses. Kant wants to
argue that a priori concepts are conditions of the possibility of experience. To
understand his starting point and his aim we need to know what he means by
‘experience’, what a priori concepts are, and what he means by ‘relation to an object’
 KANT ’ S IDEALISM AND HIS REALISM
when he says that the categories are necessary for us to have relation to an object. To
begin to understand his argumentative strategy, we need to understand the role of
synthesis, and we need to see why he thinks we need a priori synthesis.
II What is Experience?
Kant’s use of ‘experience’ (Erfahrung) is technical.1 He says that experience is
empirical cognition: ‘the categories . . . serve only for the possibility of empirical
cognition. This, however, is called experience’ (B147). Experience, in this sense, is
not something that animals have (assuming that animals do not have concepts in
Kant’s specific and demanding sense), and it is not mere phenomenal consciousness.2
So he is not attempting to show that the categories are necessary conditions of having
any kind of consciousness, or of having an inner life, or of perception, but just that
they are conditions of empirical cognition of objects.
In the Deduction, as I read it, Kant starts with the assumption that we do have
experience (empirical cognition) and he does not try to demonstrate this.3 The aim of
his transcendental argumentative strategy is to vindicate a priori representations that
are in doubt, by showing them to be the condition of something that is not in doubt:
in this case, our having empirical cognition in which we apply empirical concepts to
objects—or, in Kant’s term, experience. So the argument is not addressed to the
sceptic about the external world.4 If the argument is addressed to any sceptic, this
sceptic is the empiricist who thinks we can account for empirical knowledge without
a priori concepts, and not a Cartesian sceptic who doubts our knowledge of the
existence of objects outside us.
Kant’s concern in the Deduction is with the conditions of one specific aspect of
empirical cognition: thought or empirical concept application. He thinks that cog-
nition has two ingredients, concepts and intuitions. The Transcendental Aesthetic is
concerned with intuition (sensibility) and its a priori conditions. The Transcendental
1
See Tolley (forthcoming) for a detailed account of this.
2
Towards the end of the Deduction in the first edition, Kant says that without the categories we would
have no experience, and merely a blind play of representations, which would be less than a dream. In my
view, the way to understand this is that the perceptual experience (in our, not Kant’s, sense of ‘experience’)
of non-concept-possessing animals is in some ways less than the dreams of ordinary adult humans.
Although dreams are disordered and illogical, they involve objects that are thought-using concepts—
objects that are recognised as horses, trains, people, etc.
3
He argues that we have experience in the Refutation of Idealism.
4
Here I follow Karl Ameriks (2003) and Graham Bird (2006), among others. This starting point is
controversial, and some commentators see the argument in the Deduction as starting from far more
minimal assumptions. One argument in favour of this alternative is the distinction Kant draws in the
Prolegomena between the analytic and synthetic method, where the former does, and the latter does not
assume the truth of certain claims in order to investigate their possibility (Proleg. 4: 274–5). As
I understand him, the question Kant says he addresses synthetically in the Critique is the question of
whether and how metaphysics is possible (Proleg. 4: 274). He opens the Critique with the claim that we have
experience, and he discusses mathematics as an established body of knowledge; he never says it is in doubt,
and nowhere gives an argument in defence of its security.
THE TRANSCENDENTAL DEDUCTION 
Analytic is concerned with concepts (the understanding) and their a priori conditions.
In the Aesthetic he says that a priori intuitions are the condition of being given objects
in empirical intuitions. Thus, we should expect that the Deduction will be concerned
with the a priori conditions of applying empirical concepts to the objects given in
intuition. And indeed this is how Kant expresses his aim in the Deduction. He says that
‘it is already a sufficient deduction of them and justification of their objective validity if
we can prove that by means of them alone an object can be thought’ (A96); thought, for
Kant, essentially involves concepts. He says that the categories are not conditions of
being given objects in intuition (A89–90/B122–3) and that they are not conditions
under which objects are intuited (A93/B125), but rather conditions ‘under which alone
something can be, if not intuited, nevertheless thought as object in general’ (A93/
B125–6); he also claims that the categories ‘are nothing other than the conditions of
thinking in a possible experience just as space and time contain the conditions of the
intuition for the very same thing. They are therefore also fundamental concepts for
thinking objects in general for the appearances and they therefore have a priori
objective validity’ (A111). ‘Thinking’ an object is, in Kant’s view, precisely what
concepts enable us to do. Showing that the categories are conditions of thinking objects
therefore amounts to showing that they are conditions of applying concepts to objects.
As we saw in Chapter 7, conceptualists see Kant as arguing that the application of
the categories is necessary for intuitions to present us with objects. This may seem to
be supported by such claims as that without the categories nothing is possible as
‘object of experience’ (A93/B126), that applying the categories is necessary for
anything ‘to be an object for me’, and that the unifying done by categorically-
governed synthesis is an objective condition ‘under which every intuition must
stand in order to become an object for me’ (B138; see also B137). However, Kant
opens the Deduction saying that the categories are not necessary for anything to be
given to us in intuition, and he equates showing that without the categories nothing is
possible as ‘object of experience’ (A93/B126) with showing that the categories are
necessary in order to think objects. Thus, when he talks about what is necessary for
something to be an ‘object of experience’ or to ‘be an object for me’, he is not talking
about what is necessary to intuit an object but about what is necessary for us to be in a
position to cognize an object and, specifically, about the conditions of one aspect of
cognition—concept application or thought about objects.5
5
He says that the Deduction will show that the categories are conditions ‘under which every intuition
must stand in order to become an object for me’ (B138). On my reading, intuitions give us objects, but
must be brought under the categories before these objects can be cognized as objects and can, in this sense,
become an object for me. The moderate conceptualist thinks that intuitions cannot play their role as
intuitions (giving us objects) without being brought under concepts, and therefore reads ‘become and
object for me’ as something like ‘present me with a distinct object/perceptual particular’. The problem with
this is that it is an extremely unnatural reading of the claim, given that the way Kant expresses himself
implies that we have intuition independently of its enabling anything to be an object for me. On the
moderate conceptualist reading what Kant should rather have said is that the categories are conditions
under which representations must stand in order to really be intuitions, and become an object for me. This
is not what he says.
 KANT ’ S IDEALISM AND HIS REALISM
III What are A Priori Concepts?

In order to understand what Kant thinks the categories are and what their role is, we
need to understand the entire argument of the Deduction. We can make a start,
however, by looking at what Kant thinks concepts are and what he thinks it means to
say that a representation is a priori.
Concepts, for Kant, are essentially general, mediate representations of objects
and are essentially constituents of judgments. The essential generality of concepts
means, Kant thinks, that they pick out features of things that are common to more
than one thing; they do not uniquely pick out individuals. Kant says: ‘In every
judgment there is a concept that holds of many, and that among this many also
comprehends a given representation’ (A68/B93). He says that concepts are ‘predi-
cates of possible judgments’ (A69/B94), that ‘the understanding can make no other
use of these concepts than by judging by means of them’ (A68/B93), and that
judgments involve ‘the unity of the action of ordering different representations
under a common one’. The mediacy of concepts means that they do not put us in
direct contact with objects. Whereas intuitions present us with objects—they give
us acquaintance with them—the difference concepts make to our experience is to
enable us to make judgments in which we think about things as having general
features. Kant thinks that being able to have general thoughts, on its own, does not
put us in touch with any objects, does not enable us to pick out particular (unique)
objects, and does not guarantee that there are any objects corresponding to our
concepts. In judging, ‘instead of an immediate representation a higher one, which
comprehends this and other representations under itself, is used for the cognition
of the object, and many possible cognitions are drawn into one’ (A69/B94).
Concepts introduce structured thinking into our representation of the world; this
is why concepts are needed for cognition, ‘which is a whole of compared and
connected representations’ (A97).
The generality of concepts, and the connection between this and their being
essentially constituents of judgments, can be understood in terms of Gareth Evans’s
‘generality constraint’. Evans says that
when we say that a subject’s understanding of a sentence “Fa” is the result of two abilities (his
understanding of “a” and his understanding of “F ”), we commit ourselves to certain predic-
tions as to which other sentences the subject will be able to understand; furthermore, we
commit ourselves to there being a common, though partial, explanation of his understanding
of several different sentences. If we hold that the subject’s understanding of “Fa” and his
understanding of “Gb” are structured, we are committed to the view that the subject will also be
able to understand the sentences “Fb” and “Ga”. (Evans 1982: 101)
[I]f a subject can be credited with the thought that a is F, then he must have the conceptual
resources for entertaining the thought that a is G for every property of being G of which he has
a conception. (Evans 1982: 104)
THE TRANSCENDENTAL DEDUCTION 
The idea is that having a concept, or applying a concept to an object in a judgment,

requires having the capacity to combine it with other concepts in different thoughts
and these thoughts involve being able to think of the same object’s having other
properties and other objects’ having this property. A subject has not grasped the
concepts in the judgment ‘a is F ’, if she is not also able to grasp such thoughts as that
‘b is F ’, ‘a is G’, as well as ‘a could be H ’, ‘c could be F ’,6 and so on. The creature which
lacks concepts lacks the possibility of these combinatorial thoughts. Having a concept
is not just a matter of being disposed to react in certain ways or having certain
dispositions to discriminate or group, or form subjective expectations; it requires
having the capacity to combine it with other concepts in judgments. This is crucial to
Kant’s argument in the Deduction.
One of the basic kinds of judgment Kant is concerned with is basic subject–
predicate judgments in which we attribute properties to objects. Properties are
essentially general; a property is something which more than one object can have.
The generality of properties goes together with the generality of concepts: where
I am able to attribute a property to an object by applying a concept to that object,
I must also be able to think of that object as having other properties, and I must
also be able to think of that property as potentially belonging to other things. Kant
says that ‘if I think of red in general, I thereby represent to myself a feature that (as
a mark) can be encountered in anything, or that can be combined with other
representations . . . A representation that is to be thought of as common to several
must be regarded as belonging to those that in addition to it also have something
different in themselves; consequently they must antecedently be conceived in
synthetic unity with each other’ (B133–4n). Thus, my being in a position to
think ‘this thing is red’ requires that I am also in a position to think that other
things are red (in other words, to categorise this thing with the other things that fall
under the concept), and that I am also in a position to think of this thing as having
other properties (to represent its redness as combined with, for example, its
roundness, both properties of one object). Applying the concept ‘red’ to an object
(attributing this property to an object) thus involves at least two different ways in
which we must unify representations, and grasp their unity: I represent this thing
as combined or unified together with other things which fall under the common
property, and I represent the properties of this thing as combined or unified
properties of one thing.
These ways of unifying representations are examples of what Kant calls synthesis:
‘the action of putting different representations together with each other and com-
prehending their manifoldness in one cognition’ (A77/B103). Kant introduces syn-
thesis in the context of his discussion of logic, in which he presents the idea of
transcendental logic and uses this to explain the categories. He says that general logic
6
Kant thinks that modal thinking is present in basic empirical judgments.
 KANT ’ S IDEALISM AND HIS REALISM
contains rules for thinking about objects which abstract from any relation to an
object (A55/B79). He then introduces the idea of transcendental logic as a logic
which concerns a priori rules for thoughts that have relation to objects (A57/B81).7
Thus, transcendental logic can be thought of as a logic for referential thought.
Thought, on its own, cannot (in Kant’s view) have relation to an object, since
concepts cannot give us objects. Thus, a priori rules of thought that succeeds in
having relation to an object are a priori rules of thinking about what is given in
intuition (A62/B87). Intuition presents us with particular things and particular
instances of properties but does not enable us to have thoughts about them; it
presents us with a mass of unclassified, unordered individuals, individuals that are
not grouped with each other and are not grasped as complexes or as having
properties. Having thoughts about a manifold given in intuition requires, according
to Kant, that it be ‘gone through, taken up, and combined in a certain way in order
for a cognition to be made out of it’ (A77/B103). This is synthesis. Synthesis is
something a subject does which combines a manifold that is given in sensibility and
which is necessary for applying concepts.
There is debate about exactly how synthesis relates to the understanding and
whether it always involves concepts. Kant usually attributes synthesis to the imagin-
ation and not to the understanding,8 and there may be forms of synthesis that do not
involve concepts.9 However, it is clear that Kant thinks that the converse dependence
holds: concept application requires synthesis. He thinks that applying a concept
requires both unifying representations under a rule and grasping them as unified
under this rule: concepts are rules governing the way we synthesise. He says that ‘to
bring this synthesis to concepts is a function that pertains to the understanding, and
by means of which it first provides cognition in the proper sense’ (A78/B103).
Concepts, Kant thinks, always involve grasping a way in which a manifold has
been synthesised or unified, grasping a unity in synthesis (A78/B104; A90/B122;
A103–4).
The categories are a priori or pure concepts. Kant says that representations are not
pure if sensation is contained in them; they are pure if there is no sensation ‘mixed
into’ them (A50/B74). He identifies the pure part of representations with their form
and says that pure concepts give us the ‘form of thinking of an object in general’
(A50–1/B74). Kant also says that necessity and universality are the ‘sure criteria’ of
apriority (B4). For Kant, all concepts are rules; categories are strictly universal rules.
7
He introduces this as a provisional idea, since he has not yet shown that there are a priori concepts
which actually relate to objects: ‘In the expectation, therefore, that there can perhaps be concepts that may
be related to objects a priori, not as pure or sensible intuitions but rather as acts of pure thinking . . . we
provisionally formulate the idea of a science of pure understanding and of the pure cognition of reason, but
means of which we think objects completely a priori’ (A57/B81).
8
A78/B103; A120; A118; A119; A123; A124; B151.
9
For example, it is unclear whether the synthesis of apprehension (the first of the threefold synthesis
Kant discusses in the A Deduction) involves concepts (A99–100), and it is not clear whether the
imaginative associations Kant allows to animals should be called synthesis.
THE TRANSCENDENTAL DEDUCTION 
All concepts represent ways in which things are unified, connected together, grouped,
and related to each other; a priori concepts represent things as being unified,
connected, grouped, and related to each other with necessity. All concepts are rules
of synthesis; the categories are rules of a priori or pure synthesis. A priori synthesis
organises and unifies objects in a way that involves necessity. That the categories have
as part of their content necessity and universality is the reason Kant gives for saying
that we could never empirically vindicate our entitlement to use them (A91–2/
B123–4). Although Kant’s aim in the Deduction is to vindicate the categories, his
argument concerns none of the specific concepts that make up his list of categories.
He does not present an argument, for example, for the need for the concept of cause
or the concept of substance. Rather, he argues for the need for a priori synthesis,
where what distinguishes a priori synthesis is the unification of representations in
terms of necessary and universal connections.
Before presenting his argument in the Deduction, Kant derives his list of twelve
categories from a list of what are supposed to be basic forms of judgment in general
logic. The clue to the list of fundamental categories, Kant says, is that the same
function that unites concepts in judgments also gives unity to the synthesis of
intuition (A79/B105). So transcendental logic uses the same forms as pure logic:
whereas in pure logic these are merely ways of unifying concepts, in transcendental
logic they are also ways of unifying or synthesising representations of objects that can
be given to us in intuition. For example, one of the fundamental forms of judgment is
the subject–predicate form in which a predicate concept is unified with a subject
concept. Kant thinks that, corresponding to this form, there is a way of unifying the
manifold of intuition in such a way that something is regarded as a subject of
properties—a way of unifying intuition such that we represent an object as a subject
of properties. He says that ‘[t]hrough the category of substance . . . if I bring the
concept of a body under it, it is determined that its empirical intuition in experience
must always be considered as subject, never as mere predicate’ (A95/B129) and that
the categories ‘are concepts of an object in general, by means of which its intuition is
regarded as determined with regard to one of the logical functions for judgment’
(A95/B128). Kant says that categorial synthesis is necessary to determine an object
for our intuition.
IV What is ‘Relation to an Object’?

Kant says that he will vindicate the categories by showing that they are necessary for us
to have ‘relation to an object’. Considered on its own, there are a number of different
things that could be meant by ‘relation to an object’, and a number of different ways in
which mental representations can have (or lack) ‘relation to an object’—including
being caused by an object, presenting an object, depicting an object, and so on. In the
Deduction, Kant has something much more specific than this in mind.
 KANT ’ S IDEALISM AND HIS REALISM
It is important to keep in mind that we should be looking for an account of what

Kant means by ‘relation to an object’ according to which it is something different
from what he means by being given an object. Kant says that giving us objects is
the role of intuition and is something concepts can never do. He states this at the
beginning of the Analytic (A50/B74), in the Deduction (A92/B125), and again at
the end of the Analytic, in the section on the distinction between noumena and
phenomena (A239/B298). He explicitly says at the beginning of the Deduction
that the categories are not conditions under which anything is given to us in
intuition, and that, even if appearances did not conform to the conditions the
understanding needs to think them, we would still be given objects in intuition
(A89–90/B121–3).
The conceptualist reading of intuition sees Kant’s concern in the Deduction as
being with what it takes for my conscious experience to present me with distinct
particulars. Conceptualists see Kant’s talk of having relation to an object, something’s
becoming an object for me, something’s becoming possible as object of experience, or
determining an object for thought, as all being about what it takes to have a conscious
representation of an objective particular. In Chapter 7, I gave a number of reasons for
not reading Kant in this way, not least how hard this makes it to make sense of his
claim that intuition gives us objects. By seeing the concern of both the Deduction and
the Aesthetic as being with perceptual particulars, the conceptualist reading, in my
view, fails to see what is distinctive in both sections, and fails to account for Kant’s
specific concern with cognition (rather than perception). I have already argued that
the conceptualist reading neglects Kant’s account of the role of intuition in giving us
acquaintance with objects. Similarly, in focusing on conditions for conscious experi-
ence or perception of outer particulars, the conceptualist reading, in my view,
neglects Kant’s specific concern with the generality of concepts and the conditions
of referential thought.
Crucially, Kant’s concern in the Deduction is specifically with the idea that
concepts could lack relation to an object. In particular, he thinks that there is a
prima-facie problem with seeing how a priori concepts could relate to objects. He
wants to determine what it takes for them to have such relation. Kant sometimes
speaks interchangeably of thought’s having relation to an object and of representa-
tions having objective validity or objective reality.10 He says that objective reality, or
relation to an object, is something the categories do not have outside of their relation
to possible experience. Kant says that concepts without intuitions are empty and even
that they lack ‘sense and significance’ (B148) and are ‘entirely empty of content’
(A239/B298). This is a tricky point since, despite this use of words, Kant thinks that
transcendent metaphysical concepts (such as the concept of monads) are not entirely
without meaning, although they are defective as cognition. And Kant contrasts a
10
A109; A139/B178; A155/B194; A239/B240; A242/B291; A246/B302; A289/B345.
THE TRANSCENDENTAL DEDUCTION 
priori concepts with a priori intuitions, saying that as conditions of the possibility
objects being given to us, space and time do not represent anything at all beyond
what can be given to us through the senses, but the categories are free from
this limitation (B148). Their meaning is not tied to the spatio-temporal objects
of the senses, and they can be used to produce coherent thoughts about non-
spatio-temporal objects. However, he says that this extension
does not get us anywhere . . . For they are then merely empty concepts of objects, through
which we cannot even judge whether the latter are possible or not—mere forms of thought
without objective reality—since we have available no intuition to which the synthetic unity of
apperception, which they alone contain, could be applied, and that could thus determine an
object. (B148)
Similarly, he says:
[T]o think of an object and to cognize an object are not the same. For two components belong
to cognition: first, the concept, through which an object is thought at all (the category), and
second, the intuition, through which it is given; for if an intuition corresponding to the concept
could not be given at all, then it would be a thought as far as its form is concerned, but without
any object, and by its means no cognition of anything at all would be possible, since as far as
I would know, nothing would be given nor could be given to which my thought could be
applied. (B146; similarly, see B149)
In both these passages, the problem with using a priori concepts apart from
intuition is not that they are empty in the sense of having no meaning at all, but
that we do not know whether there are any existing objects corresponding to these
concepts, or even whether there could be such objects. Kant thinks that we do not
even know so much as whether they are concepts of really possible objects. Kant
holds that concepts never put us directly in touch with or present us with objects,
cannot guarantee that there are any objects to which they apply, and do not
uniquely individuate or pick out their objects. His account of how we remedy
this lack is through having objects given to us in intuition. Intuitions present us
with objects: this guarantees that there is an object and it does so by putting us into
mental contact with this object.
I suggest that Kant’s concern here is with a kind of referential thought: thought
that succeeds in connecting with an object. Suppose I dream about a man living in
Australia who wears red socks every day and teaches Anthropology. Suppose further
that there is exactly one man living in Australia who wears red socks every day and
teaches Anthropology. This coincidence, one might think, is not enough to make my
dream about that particular man. My dream lacks the right kind of connection to the
man to be about him. Kant’s concern is with how we establish the kind of connection
needed for thoughts to refer to objects. He thinks that conceptual thought on its own
cannot connect to objects, which is why it cannot constitute cognition proper: it lacks
relation to an object. If I put together a series of concepts in a judgment and it just
 KANT ’ S IDEALISM AND HIS REALISM
happens that there is in fact an object which falls under all these concepts, that is not
enough, in Kant’s view, to make my thought about this object—it will not have
relation to the object. On this view, if there are in fact things which fall under the
concept of a Leibnizian monad but we have no way of being acquainted with them,
we are not in a position to use the concept of a monad in successful referential
thoughts—in thoughts that succeed in having relation to these things. The concept of
a monad therefore lacks objective validity for us even if there happen to be things
corresponding to it, because we are not in a position to use it to refer to these things.
Not only do we not have knowledge of monads, we cannot cognize them: we cannot
even succeed in having referential thoughts about them. Kant thinks that for thought
to have objective validity or relation to an object we require the possibility of
acquaintance with the object—the possibility of being given the object in intuition.11
However, mere acquaintance is clearly not sufficient for having a referential thought
about an object because acquaintance (intuition) is not, Kant thinks, any kind of
thought. To have successful referential thoughts it is necessary that I am presented
with perceptual particulars; in addition, I must have a way of thinking about these
particulars. There must be an object that is given to me, and I must be in a position to
apply concepts to this object: I must have a way of bringing my general, conceptual
representations into relation to the objects that are given to me in intuition. Kant’s
concern in the Deduction is with how this relation is achieved.
Kant starts the Deduction with the claim that there is a problem seeing how (non-
mathematical) a priori concepts can succeed in referring, or relating, to objects.
Unlike mathematical concepts, our a priori metaphysical concepts are not given
objects in a priori intuition. This means that these concepts will lack relation to
objects unless they apply to objects which are given to us in empirical intuition. Kant
says that the objective validity of the categories rests on their applying to possible
intuitions (A289/B345), that the objective reality of the categories requires not just
intuitions ‘but always outer intuitions’ (B291), and that ‘without the conditions of
sensible intuition, the synthesis of which they contain, the categories have no relation
at all to any determinate object, thus they cannot define one, and consequently they
do not have in themselves any validity of objective concepts’ (A246/B302). His
concern is therefore to show how a priori concepts can relate to objects given in
empirical intuition. He starts, as I read him, by assuming that we do have successful
empirical concept application (we succeed in using empirical concepts in thoughts
that have relation to an object); if he can show that applying a priori concepts to the
objects presented in empirical intuition is a condition of applying empirical concepts
to them, he will have shown that a priori concepts have relation to objects.
11
This is more demanding than just requiring a causal connection. However, Kant does not require that
we have actual acquaintance with all the things we can successfully think about; rather, it must be possible
for us to have acquaintance with an object, and the object must be causally connected to one with which we
have direct acquaintance.
THE TRANSCENDENTAL DEDUCTION 
So as I read the Deduction, Kant’s concern is to explain how a priori concepts can
be brought into relation to objects given in empirical intuition and his strategy is to
argue that this is a condition of any concept’s being brought into relation to the
objects that are given to us in empirical intuition. The objective validity of the
categories consists in their being used to think about objects that are given to us,
and this objective validity is demonstrated by showing that using the categories is a
condition of having any successful referential thought about such objects. Kant
argues that without the application of a priori concepts, empirical concepts would
not amount to experience and ‘all relation of cognition to objects would also
disappear’ (A111; see also A121).12 We may imagine that there is a possible situation
in which we have successful empirical concept application but do not apply a priori
concepts, but this is not in fact a possible situation. So although he starts with a
problem for a priori concepts, in fact his concern is also with what it takes for
empirical concepts to have relation to an object, since he wants to show that the
former is a condition of the latter. As we saw in the previous section, the categories
are concepts that unify representations with necessity and universality. Thus, we
need to see why Kant thinks that we could not represent objects with the unity and
connections required to bring them under empirical concepts (to give empirical
concepts relation to an object) unless we brought them under concepts that connect
them with necessity and universality: we need to understand why he thinks we need a
priori rules of synthesis to determine an object for our empirical concepts.
V Why Do We Need A Priori Synthesis?

In both editions of the Critique there are two central parts to Kant’s case. First, he
argues that synthesising the manifold of intuition with the necessity and universality
that the categories bring is necessary to determine an object for thought to which
empirical properties could be attributed. He says that a priori synthesis determines an
object for our representations and that without this thought lacks relation to an
object. Second, Kant thinks that this synthesis is bound up with the capacity to be
self-conscious—the capacity to be aware of oneself as the subject of one’s thought,
which, according to Kant, essentially involves being aware of oneself as a necessarily
unified subject of thought. How Kant establishes both these ideas is unclear and
disputed, as is the relation between them. He sometimes seems to claim that the
synthesis that enables us to determine an object for thought is required for us to be
(potentially) aware of a unified self; on the other hand, he sometimes appears to say
that being able to recognise my representations as necessarily belonging to one
consciousness is required for me to be in a position to represent unified objects.
And he also sometimes seems to equate the two, saying that the synthesis that unifies
12
This interpretation clearly requires that we start with the assumption that we have empirical
cognition.
 KANT ’ S IDEALISM AND HIS REALISM
representations as belonging to one consciousness is the synthesis that unifies

representations of objects (A105; A108; A121; B139). My concern here is with
understanding the idea that categorical synthesis is needed to determine an object
for thought, and I do not look at the connection between this and self-consciousness
(I am concerned only with one part of the argument in the Deduction).
As we saw in Chapter 7, one way Kant’s account of synthesis in the Deduction
could be understood is as something like what contemporary cognitive psychology
refers to as ‘binding’. Understanding synthesis in this way leads to a particular
account of Kant’s problem in the Deduction and of his solution: the problem is
seen as that of giving an account of how we can group together sensations to present
us with distinct perceptual particulars, and the solution requires conceptually gov-
erned organisation of the sensory input. The idea is that we have a mass of sensory
input and without applying concepts we have no way of drawing boundaries around
some parts of it, or of determining which bits of sensory input hang together, so as to
represent distinct, discrete particulars. On this reading, the role of both a priori
intuition and a priori concepts is to enable us to have perceptual acquaintance with
objects: the idea is that determining an object for thought is binding sensation so that
we can be presented with a distinct perceptual particular. Thus, on this reading, Kant
is understood to be arguing that the categories are necessary for relation to an object
because they are necessary for us to be given an object in intuition.
Among the serious problems we have seen with this reading is the fact that Kant
has an account of something like binding in the Aesthetic, he thinks that it is
necessary for us to be given objects and he thinks that it is not conceptual. Assimi-
lating the role of a priori concepts to the role of a priori intuition by seeing both as
necessary for perceptual particulars leaves out what is in fact distinctive in each case.
In Chapter 7 I argued that the conceptualist reading fails to take seriously Kant’s
account of the role of intuition and the role of a priori intuition; here I will argue that
it also fails to take seriously what it is that concepts add—the generality of thought.
The judgmental structure that concepts introduce is not mere binding or association
(something Kant thinks concept-lacking animals have).13 Kant’s aim is to show that
the categories are conditions of representing objects with the kind of unity, order, and
structure that is required to have structured, general thoughts about those objects.
Interpreting the basic role of categorically governed synthesis as producing unified
perceptual particulars leaves Kant with a weak and unmotivated argument for the
central part of his case: the need for necessity. The fundamental question of the
Deduction is why we need a priori synthesis: why do we need to organise and unify
the manifold of particulars given to us in intuition in ways that represent things as
connected with necessity and universality? On the face of it, it seems that an
empiricist could agree with Kant’s claims that we unify what is given in intuition
13
Kant explains subjective necessity in terms of the idea that ‘we may expect similar cases (just as
animals do)’ (CPR: 13; see also C 11: 52, 314).
THE TRANSCENDENTAL DEDUCTION 
by bringing it under conceptual rules and that we need to do this to have cognition,
but reject the claim that we need synthesis governed by a priori concepts or that we
need to represent necessary connections. Kant thinks sensation is ordered in the a
priori forms of intuition to produce empirical intuitions, and he thinks that both we
and other animals have empirical principles of association—ways in which we group
together what is presented to us in intuition. He argues that such association is
insufficient for us to have an object of thought. Yet it is unclear why such association
should be thought insufficient to have perceptual particulars, and if this is Kant’s
strategy, his arguments are weak and proceed by an extreme and implausible
dichotomy. His arguments oppose a synthesis that connects things with necessity
to a combination that is entirely arbitrary, haphazard, and random. But these are not
the only alternatives. As Graham Bird says, Kant speaks of conceptual unity variously
as a way of avoiding haphazardness, arbitrariness, or merely accidental connections,
but ‘what is left out between the haphazard and the necessary is a stable regularity’
(Bird 2006: 343). Kant insists that the kind of unity we would get through stable
regularity, or what he calls the empirical affinity of the manifold, is not enough for
our concepts to have ‘relation to an object’. He says that an object ‘is regarded as that
which is opposed to our cognitions being determined at pleasure or arbitrarily rather
than being determined a priori, since insofar as they are to relate to an object our
cognitions must also necessarily agree with each other in relation to it, i.e., they must
have that unity that constitutes the concept of an object’ (A104). However, as Bird
points out, his argument for this seems to be that without a necessary unity we have
merely arbitrary or haphazard connections. Why should the only options be, on the
one hand, that our cognitions are determined ‘at pleasure or arbitrarily’ and, on the
other, that they are determined a priori and with necessity? It is hard to see why
empirical principles of association, together with the spatio-temporal form of intuition,
could not give us perceptual particulars, or why such particulars should be regarded as
arbitrary, haphazard, or accidental. In my view, it is only if we pay attention to the
generality of thought, and the corresponding generality of properties, that Kant’s
opposition between necessary connections and chaos can be made sense of.
Kant says that
Unity of synthesis in accordance with empirical concepts would be entirely contingent, and,
were it not grounded on a transcendental ground of unity, it would be possible for a swarm of
appearances to fill up our soul without experience ever being able to arise from it. But in that
case all relation of cognition to objects would also disappear, since the appearances would lack
connection in accordance with universal and necessary laws, and would thus be intuition
without thought, but never cognition. (A111)
He says that without the synthesis corresponding to a priori concepts (i.e. containing
necessity and universality) sensibility would yield appearances ‘but no objects of an
empirical cognition’ (A124). Note that Kant does not say that without the synthesis
that corresponds to the categories we would have a mere blooming buzzing mass of
 KANT ’ S IDEALISM AND HIS REALISM
sensation. We would have appearances: objects of empirical intuition. What we

would lack is the possibility of thinking of the appearances as objects and therefore
of having empirical cognition.14 He says that the unity that empirical concepts
introduce, a unity that lacks necessity, is not enough for thought to have relation
to an object, but that without this we could still have appearances. In my view, we can
make sense of this if we understand him to be concerned specifically with the kind of
unity that is involved in attributing properties to objects in thoughts, which, he
thinks, corresponds with the unity of a judgment and is different from merely having
distinct perceptual particulars. As we have seen, the generality of thought means that
thinking of an object as having a property requires being able to think of the same
object as also having other properties, and the same property as potentially belonging
to other objects. These features of property attribution require, Kant thinks, a way of
thinking about the unity of the object (the subject of properties) that is more than
simply the unity involved in having a perceptual particular.
In the Deduction in the first edition, Kant starts by presenting three different ways
in which we need to unify what is given in intuition. This threefold synthesis is often
read as presenting Kant’s account of what is needed to produce (outer) intuitions
(perceptual presentations of outer particulars), but in fact, in this section, Kant talks
about synthesising intuitions, not synthesising sensation to form intuitions. This
implies that we already have intuitions (singular and immediate representations that
give us objects), and something further is being done to enable us to think about these
intuitions as presenting objects. As I understand Kant, he thinks that outer intuition
presents us with a manifold of separate, distinct, spatially located, bounded things,
which are not yet grouped or classified, are not grasped as unified complexes of parts,
are not grasped as unified subjects of properties, and, a fortiori, are not grasped as
having properties. Kant thinks that this manifold of singular perceptual units, as well
as the manifold in each intuition, needs to be synthesised for cognition of objects to
be possible—for us to be in a position to make judgments about objects in which we
attribute properties to them. Kant argues first that we could not represent the
manifold in an intuition as manifold without a synthesis that enables us to apprehend
a thing as a unified complex (A99–100). His concern here is not with what it takes to
represent a thing as unified in the sense of being presented with a single distinct
thing, but with representing it as a unified complex—representing the manifold in it
as manifold. Without this synthesis we can only represent a single, distinct thing.
Kant then argues that we need to have a way of reproducing the unity of the object so
that we represent it as unified over time. He says that these two ways of representing
14
Kant immediately goes on to say: ‘Now I assert that the categories that have just been adduced are
nothing other than the conditions of thinking in a possible experience, just as space and time contain the
conditions of the intuition for the very same thing. They are therefore also fundamental concepts for
thinking objects in general for the appearances, and they therefore have a priori objective validity, which
was just what we wanted to know’ (A111).
THE TRANSCENDENTAL DEDUCTION 
unified objects are related: I cannot represent a unified object at a time without being
in a position to represent a unified object over time. And he says (without much
argument) that these ways of representing unity somehow require necessity: con-
sciousness of the necessary unity of the synthesis of the manifold of intuition (A103).
As I read it, the unity Kant is concerned with here is specifically that which is
involved in thinking of an object as a unified subject of a number of properties. As we
have seen, this is a condition of concept application, because applying a concept to an
object requires being able to think of the object as also having other properties. Kant
thinks that this involves a number of connected abilities: representing a unified
subject of properties at a time, representing a unified subject of properties over
time, and representing properties as general attributes potentially belonging to
other objects, in addition to the one under consideration. Kant thinks that this
complex set of capacities is needed to have thoughts in which we attribute properties
to the objects with which we have acquaintance. He argues that it depends on
representing intuitions as connected with necessity and universality and that without
this we would have perceptual particulars but no objects of thought. In other words,
he is arguing that synthesising the manifold with necessity is a condition of thinking
of objects with the unity that is needed to make judgments in which we attribute
properties to objects: bringing concepts into relation to objects given in intuition.
The idea of necessary connections or a necessary unity is clearly crucial to the
argument, but Kant does not give much argument for it at this point. He says that
representing a thing as unified over time would not be possible if the object’s properties
were constantly changing and that this reproduction requires an a priori ground of
necessary unity (A101). Kant’s empiricist opponent seems entitled to wonder why it is
not enough that the object’s properties are in fact not constantly changing—i.e. why
sufficient empirical regularity (what Kant calls the empirical affinity of the manifold) in
what is presented is not enough for our principles of association to enable us to
represent unified objects over time. In my view, we cannot answer this objection if
we see Kant, at this point, as concerned with distinct perceptual particulars; rather, his
concern is with the specific way we need to think about the unity of an object to
attribute properties to the object in judgments. As I read Kant, he thinks that we
organise the sensory input into perceptual units using our forms of intuition (space and
time), as well as biologically determined principles of association. To understand the
Deduction we need to see why he thinks empirical concepts could not be successfully
applied to the individuals presented by intuition unless we bring these individuals
under our a priori concepts: why having referential judgments in which we attribute
properties to objects requires representing these objects as unified with necessity.
VI Spontaneity and Generality

We have looked at one key feature of Kant’s account of concept application: the
generality of thought. The other key feature is what he calls the spontaneity of
 KANT ’ S IDEALISM AND HIS REALISM
thought. I will argue that it is the combination of the generality and the spontaneity
of thought that creates the problem Kant is concerned with in the Deduction. As we
have seen, his opening problem is showing how a priori concepts can be brought into
relation to objects given to us in empirical intuition, and his strategy is to show that
this is a condition of empirical concept application: of empirical concepts being
applied to objects in judgments. This means that there must be a general problem
with empirical concept application that a priori concepts are needed to solve. As
I read him, Kant thinks that the possibility of empirical concept application is
threatened by the combination of the generality and the spontaneity of thought,
and he thinks the problem can be solved only by introducing the specific unity
involved in categorially governed synthesis. We have seen that the essential generality
of concepts means that applying concepts to an object involves grasping the object as
the unified subject of a number of properties which, since they are general, could also
be properties of other things.15 The next thing to understand is his account of the
spontaneity of thought.
Kant thinks that the way we combine objects in grouping them under predicates is
not a feature of reality that is given in intuition: there are indefinitely many possible
ways of combining objects and properties, corresponding to possible predicates
objects could fall under, which are compatible with what is given in intuition. This
is the spontaneity of thought. In the second edition of the Deduction, Kant argues for
synthesis by saying that combination cannot be given through the senses (B129), and
in both editions he says that concepts are functions of spontaneity (A51/B75; A68/
B93; B129). Like the phrase ‘relation to an object’, the term ‘combination’ could mean
a lot of different things; we need to understand Kant’s specific use of it in the context
of his argument, rather than starting from what the term might suggest to us
independently of this context. In my view, we will make more sense of the argument
if we do not assume that ‘combination’ refers to binding, or that it is a general term
that covers any organising work done by the mind. As we have already seen, Kant has
an account of binding in the Aesthetic that does not involve concepts. He says that
combination is something specific: the representation of the synthetic unity of the
manifold (B130–1). When he speaks of combination not being given, Kant says that
this applies both to the combination of intuitions and the combination of concepts
(B129)—not, notice, to the binding or ordering of sensation that is needed to for us to
have empirical intuitions (singular and immediate representations that give us
objects).
15
The idea here is not that ordinary subjects making empirical judgments must be making reflected
judgments containing such concepts as ‘property’ and ‘subject’. In saying that the subject must be able to
think of more than one object as having the same property, the idea is that the subject must be able, for
example, to think of redness as something that could belong to other objects, and not just this one red
thing. Without this, the subject is not thinking of redness as a property; with it, she is, whether or not she
is self-consciously aware of thinking in this way by making judgments using the reflected concept
of a property.
THE TRANSCENDENTAL DEDUCTION 
Kant’s claim that combination is not given (the spontaneity of thought) must be
understood in a way that fits with his allowing intuition to give us appearances
(objects, distinct things outside us), and that makes sense of his claiming that without
a priori synthesis combination would be so haphazard, arbitrary, and chaotic that we
would not be able to cognize anything as an object (would not be able to have
empirical judgments about objects) and ‘all relation of cognition to objects would
also disappear’ (A111). This is not, in my view, a concern about what it takes to have
(outer) perceptual particulars; rather, Kant’s concern is that once we have (outer)
perceptual particulars, there are indefinitely many ways these things could be
grouped or combined in being regarded as falling under properties.16 I suggest that
this can be understood through consideration of deviant, gerrymandered ‘properties’
(mere Cambridge properties) and deviant, gerrymandered ‘objects’, such as mereo-
logical sums which do not correspond to what we ordinarily think of as objects.
While such a reading might be thought anachronistic in using contemporary ter-
minology for the problem, it seems to me to fit well with precisely the issue Kant
raises: that there are indefinitely many different ways things could be combined in
being regarded as falling under concepts or as having properties, and that we need a
way of distinguishing what we count as ‘real’ properties and ‘real’ objects. This is
precisely what gerrymandered concepts and properties represent: the indefinite ways
of classifying appearances (the manifold of intuition) that do not correspond to what
we ordinarily think of as concepts and properties; this indefinite possibility of
combination is the problem with which Kant is concerned. I will argue that this
reading makes sense of his account of the origin of the problem as something that
arises specifically for concept application (from the combination of spontaneity and
generality) as well as his solution (that there is a special group of higher-level, formal
concepts that govern the attribution of empirical concepts). In my view, Kant’s
concern is that unless we can exclude gerrymandered properties and objects, we
will not have a unified object of thought, which means that we will not be able to
think of anything as the subject of more than one property, which means we will not
be able to attribute properties to anything—empirical concept application will not be
able to get off the ground. The idea is that the spontaneity of thought allows for an
indefinite combination of gerrymandered properties and objects and this under-
mines the presentation of a unified subject of properties that is required for the
generality of thought. This is a problem which occurs even when we have distinct
perceptual particulars, and, as I read him, Kant thinks it can be solved only by the
synthesis that corresponds to a priori concepts.
So to understand Kant’s problem, as I read it, we need to understand the idea of
gerrymandered or deviant properties and deviant objects, and to see why (Kant
thinks) they threaten the requirements of the generality of thought. Peter Geach
16
Kant is also read this way by Walker (1985).
 KANT ’ S IDEALISM AND HIS REALISM
introduced the idea of mere-Cambridge change to talk about a change in a predicate

applying to an object which we do not intuitively take to correspond to a genuine
change in the object.17 For example, if a barn starts burning 50 miles away from an
apple, the predicate ‘50 miles from a burning barn’ comes to apply to the apple,
where it did not before, but intuitively we do not think that there has been a change
in the apple’s properties. Corresponding to mere-Cambridge changes, there are
indefinitely many mere-Cambridge predicates which objects could be regarded as
falling under. Gerrymandered properties can also be understood through consid-
eration of properties such as grue,18 or through the idea, made famous by
Wittgenstein’s so called rule-following considerations, that for any apparent rule
or way of categorising things there are indefinitely many ways of continuing to
combine or categorise which can be made consistent with it.19 Although these are
all contemporary ways of explaining the problem, the thought captured by the idea
of Cambridge change is one that fits with Kant’s concern: recognising that objects
can be classified in all sorts of ways that do not correspond to what we think of as
their properties. Kant famously says that any concept can be used as a predicate
logically, whether or not it corresponds to something that is actually a property,20
and he says that combination is not given. As I understand him, Kant’s claim that
combination is not given can be understood as saying that how we group objects in
regarding them as falling under a common property is not presented in perception;
there are indefinitely many predicates corresponding to possible ways we could
group or classify the objects that are given in intuition. That all red objects are
grouped together in terms of their redness is no more a given or perceptually
presented feature of the world than is the grouping together of objects that have the
property of being red if round and green if square. Both are possible ways of
unifying a manifold of intuition. Crucially, this point is fully consistent with
thinking that intuition presents us with perceptual particulars: being presented
with a red apple in intuition does not determine whether or not we will group it
together (unify it) with other things by bringing it under the concept ‘red’, the
concept ‘is 50 miles from a burning barn’, or the concept ‘round if red and square if
green’. All these classifications are compatible with perceiving a red apple. If we
simply see Kant’s concern with combination as with what determines our strategies
of binding it becomes hard to see how the categories solve a different problem to
that solved by the forms of intuition, how we would still have appearances (objects
of empirical intuition) without the categories, and why he thinks that the problem
could not be solved by sufficient empirical regularity.
17
Peter Geach (1969: 71).
18
Something is grue iff either it has been examined before time t and been found to be green, or it has
not been examined before t, and is blue.
19
The idea that Kant has concerns that are similar to Wittgenstein’s is argued by David Bell (1987).
20
In his critique of the ontological argument (A598/B626).
THE TRANSCENDENTAL DEDUCTION 
Corresponding to deviant properties, there are indefinitely many possible deviant

objects that could be regarded as subjects of properties. We could, for example,
synthesise together the legs and surface of the table as one object, or we could regard
the table-top and the lamp attached to it as one object, or any number of stranger
combinations, involving ‘temporal parts’, such as alternative time-slices of the table-
top and the fly buzzing over it. Kant does not talk about crazy mereological sums but it
is clear that he thinks that the combination of spontaneity and generality threatens to
produce a situation in which there is no determinate object of thought, and thinking
about the possibility of deviant objects is helpful in understanding this. The possibility
of deviant properties and the possibility of deviant objects are really part of the same
issue: what counts as a property and what counts as a subject of properties, and how
we group properties together in regarding them as belonging to an object.
In order to understand Kant’s claim that we are threatened with a situation in
which there is no determinate object of thought, it is helpful to see why he does not
take being given an object in intuition to be the same as having a determinate object
of thought: why having objects given in intuition (being presented with distinct
perceptual particulars) leaves the object of thought underdetermined. Suppose I am
visually presented with an apple, but I do not grasp it as an apple, or even as a body;
I simply see a spatially located, bounded, shaped, patch of colour of a certain size.
There are different things that I could regard as a subject of properties: I could judge
‘that object is red’ or ‘that redness is round’. I could regard the red insect that is on
top of the apple as part of the spatially bounded red object, and therefore think that
the object has changed when the insect flies away. I could regard the part of the apple
that is located in a shadow, and therefore appears a different colour, as not being a
part of the apple. Crucially, thinking of the apple-plus-insect as a subject of proper-
ties, thinking of just the apple as a subject of properties, and thinking of the redness
itself as a subject of properties are all possibilities compatible with what is given in
intuition (perceiving the bounded, located red particular outside me). Further, these
possibilities are fully consistent with the idea that binding of sensation is necessary
for us to be presented with a spatially bounded, located, round red thing outside us—
that binding is necessary for us to perceive distinct particulars. Once we have such a
perceptual particular, there is still a further step required for us to be in a position to
think of something as a subject of properties, and there is more than one way of
unifying (synthesising, grouping) the manifold of intuition (perceptual particulars)
to do this. It might be thought that it is our (perhaps biologically determined)
associative binding strategies that determine that it is the apple to which we are
paying attention, rather than a gerrymandered object. As I read him, Kant’s point is
that this is indeed part of the picture, but that this is not enough to explain what it
takes for us to be in a position to think about an object. Having a determinate object
of thought involves having something you are thinking of as a unified subject of
properties, and merely having acquaintance with an object is not enough to explain
the capacity to think about an object in this way.
 KANT ’ S IDEALISM AND HIS REALISM
The idea, as I understand it, is that without a way to rule out gerrymandered
concepts we are unable to rule out deviant objects, and this means that we do not
have a determinate object of thought, which makes it impossible to apply any
concepts (including deviant ones). The generality of thought means that in order
to think of something as having a property a subject must be in a position to think of
it as a subject of other properties as well, and must be in a position to think of the
property in question as potentially belonging to other objects. This, I submit, is what
Kant thinks we cannot do in the chaotic situation in which there is no limit to what
can count as a property, and correspondingly no limit to what can count as a subject
of properties. Investigating an object’s properties requires a way of ruling out
properties that do not belong to it. It also requires a way of thinking of those
properties that do belong to it as unified in belonging to one object. We could put
the point by saying that we need some account of the unity relation properties have
when they are united in an object and we need an account of property identity.21 Any
number of gerrymandered concepts can be attributed to an object if combined with
the right gerrymandered objects. This means that in a situation in which we have no
way of excluding gerrymandered properties and objects (no way of determining what
counts as a real property and what counts as a real object), there is no way to
investigate what properties any object has, because any result we get will fall under
a possible way of combining things. We will never be able to discover that a property
attribution is wrong, and therefore learn what properties an object has. Kant thinks
this is a situation in which property attribution lacks objectivity: we have only
subjective tendencies to combine, but nothing that enables us to investigate the
world and discover that one way of combining is correct and another is not. This
is a situation in which empirical concept application cannot get off the ground, and
in which there is nothing that counts as an object of thought. We would really have
simply haphazard, arbitrary, rhapsodic connections—the kind of chaos, at the level of
thought, that Kant is trying to avoid. We cannot, Kant thinks, solve the problem
empirically, because this would require being able to determine empirically which
properties belong to an object, but this is precisely what we are not in a position to do
without a determinate object of thought.
One way of seeing why the unity of the object is necessary for thinking of more
than one property as belonging to the same object is the fact that properties that are
unified as properties of the same object interact with each other, or have implications
for each other, which in turn have implications for the overall properties of the
object. For example, the implications for an object’s overall properties of its being
round to it will depend on its other properties, and the upshot of attributing
roundness to two objects with very different other properties (being made of concrete
21
These are the terms in which Sydney Shoemaker (2003a; 2003b) presents what seems to me to be a
similar problem.
THE TRANSCENDENTAL DEDUCTION 
or being made of smoke) will be different.22 This means that understanding what it is
to attribute a property to an object involves understanding how this property will
interact with (have implications for) the object’s other properties; this requires seeing
these properties as having the particular unity that is involved in their hanging
together as properties of one object. Further, understanding what it is for an object
to have a property involves having an idea of how the object would affect and be
affected by other objects—how it could be changed—which means that attributing a
property to an object involves a conception of the unity of the object over time as well
as at a time. (As we saw in the discussion of the threefold synthesis, Kant holds that
thinking of an object as a unified complex requires thinking about its unity at a time
and over time). This point is easiest to see with properties that are powers. Attrib-
uting a causal power to an object involves thoughts about how the object would
interact with (affect and be affected by) other things in certain circumstances—
thought about how that very thing would change over time in specified circum-
stances.23 This involves a thought of the ‘it’ (the unified subject of properties, unified
at a time and over time) that would do the changing. This means that attributing
powers to objects, representing an object as a unity at a time, and representing an
object as a unity over time are capacities that are bound up with each other.
22
A similar, but less abstract, line of argument is given by Shoemaker, who argues that, for example,
giving something the property of being knife-shaped will have very different implications for its causal
powers depending on whether it is made of steel, butter, or a cloud of invisible gas. As Shoemaker says, ‘if
all I know about a thing is that it has this property [being knife-shaped], I know nothing about what will
result from its presence in any circumstances’ (Shoemaker 2003a: 212), and I thus do not have a grasp of
the upshot of attributing the property to any object. He argues that in order to understand the property that
you are endowing an object with when you make it knife-shaped, you need to have some idea about how
being knife-shaped will interact with other properties and therefore what powers it will endow a particular
thing with. Like Kant’s argument, Shoemaker’s turns on the essential generality of properties. It is because
of this that being able to attribute a property to an object requires being able to attribute the same property
to other objects as well. Thus we need an idea of property identity: we need a way of determining what
counts as different objects’ having the same property. However, the same property will have different
implications for the overall causal properties of different objects, because of the ways it interacts with
different objects’ other causal powers. Shoemaker argues that this means that we cannot attribute
properties to objects without the idea of causation—of properties having causal implications for each
other. See Dickie (2011: 17).
23
This argument is put forward by Shoemaker, who says ‘The attribution of a particular property will
not have a determinate truth value unless the identity conditions for things having that property are
already fixed; for the attribution of a property to a thing will imply propositions about what will happen to
that thing under certain circumstances, and these propositions will have a determinate truth value only if it
is already settled under what circumstances a future happening is to count as an episode in the history of
that thing. But if this is so, the unity conditions for nonsimultaneous property instantiations can be no
more a matter for conventional decision than the unity conditions for simultaneous property instanti-
ations’ (2003b: 253). He uses this to argue against the idea that gerrymandered objects and properties are
on an equal ontological footing to ordinary objects and properties. Gerrymandered objects do not have the
kind of causal continuity that is needed to attribute to them the momentary properties they seem to have:
attributing causal properties to them at a time will have implications for what will happen to them over
time that will be false. Ordinary continuants are, while gerrymandered series are not ‘histories of entities
that are the subjects of the properties instantiated in the stages’ (2003b: 258).
 KANT ’ S IDEALISM AND HIS REALISM
Kant thinks that we could not learn from experience how things actually happen to
be grouped together without having some way of determining, as a starting point,
what counts as subject of properties to investigate empirically and what counts as
more than one property belonging to the same object. His solution to this problem is
that we have an a priori conception of an object in general which determines what we
regard in thought as an object—a unified subject of properties. The categories are
higher-level ways of organising our conceptual thought that determine what counts
as an object and thereby provide an a priori framework which makes possible
objective conditions of correctness for concept application. Kant says that the
categories are ‘concepts of an object in general, by means of which its intuition is
regarded as determined with regard to one of the logical functions for judgments’
(B128–9). He says that
the function of the categorical judgment was that of the relationship of the subject to the
predicate, e.g., “All bodies are divisible.” Yet in regard to the merely logical use of the
understanding it would remain undetermined which of these two concepts will be given
the function of the subject and which will be given the function of the predicate. For one
can also say: “Something divisible is a body.” Through the category of substance, however, if
I bring the concept of a body under it, it is determined that its empirical intuition in experience
must always be considered as subject, never as mere predicate; and likewise with all the other
categories. (B128–9)
Bringing something under the concept of substance determines that it is what is

being thought of as the subject of properties. The content that the categories
introduce is entirely formal, and central to this is the idea of representing necessary
connections or a necessary unity. Properties are unified in belonging to one object by
being represented as related to each other with necessity. So Kant’s solution is that
the possibility of empirical concept application involves thinking of necessary con-
nections: thinking objects as having a unity that involves necessary connections
between properties thinking of and objects as relating to other objects in ways
governed by necessary connections. He says that
the concept of body serves as the rule for our cognition of outer appearances by means of the
unity of the manifold that is thought through it. However, it can be a rule of intuitions only if it
represents the necessary reproduction of the manifold of given intuitions, hence the synthetic
unity in the consciousness of them. Thus in the case of the perception of something outside us
the concept of body makes necessary the representation of extension, and with it that of
impenetrability, of shape, etc. (A106)
Kant says that the possibility of objective judging (concept application) depends on
thinking of there being necessary connections. He says of the ‘copula is’ that
this word designates the relations of the representations to the original apperception and its
necessary unity, even if the judgment itself is empirical, hence contingent, e.g., “Bodies are
heavy.” By that, to be sure, I do not mean to say that these representations necessarily belong
THE TRANSCENDENTAL DEDUCTION 
to one another in the empirical intuition, but rather that they belong to one another in virtue
of the necessary unity of the apperception in the synthesis of intuitions . . . Only in this way
does there arise from this relation a judgment. (B142)
We have not looked at the relation between the synthesis according to the categories
and the unity of consciousness, but, leaving this aside, we can use the above
discussion to make sense of a puzzling feature of this passage. Kant wants to say
that empirical judgments in which contingent properties are attributed to things
require thinking that the properties hang together with some kind of necessity. How
can the attribution of the properties to the objects both be contingent and involve
necessity? The properties do not necessarily belong together—they are contingent.
Yet making a judgment in which the properties are attributed to the object requires
thinking of them as governed by a conception of an object as a unified subject of
properties which have necessary implications for each other. Although this involves
putting the point in a less abstract way than Kant does in the Deduction (and in a way
which he thinks we are not in a position to do until after the Schematism and the
Principles), we could explain the idea by saying that thinking of an object as a subject
of properties involves thinking of it as a causal unity, at a time and over time, and
as being in law-governed causal interaction with other (causally unitary, space-
occupying) objects. Part of what it means for an object to be a causal unity is that
its properties interact with each other in ways governed by necessary connections, so
have causal implications for the object’s overall properties. On this account, seeing
properties as causally unified in this way just is what it is to see them as belonging to
one object and it is required for us to start investigating an object’s properties by
doing things to it. Among the indefinitely many ways in which the manifold could be
synthesised, the basic way in which we determine what we regard as an object (a
subject of properties) and what we regard as a property is by using the concept of
substance, which (Kant argues later) includes the idea of a space-occupying, causally
unitary subject of properties which persists through time, which is an aggregate of
parts, and is made of stuff which is conserved, and which is in causal interaction with
other substances.
The idea is not that we have to know whether something is in fact a unified subject
of properties to start investigating it.24 We could pick out a perceptual particular
(something with relatively coherent spatio-temporal boundaries) which, on investi-
gation, turns out not to be an object. A plant together with the rock on which it is
24
In addition, there is a way in which a creature which lacks concepts can investigate an object. It can
grab it and do things to it, and it may form subjective expectations of ways in which things will respond.
These capacities might involve some kind of proto-conception of causation, but on the Kantian picture we
are considering, this would not be the true concept of causation, as it would not put the creature in a
position to think general thoughts—to think of the object as having a property which it also thinks of other
objects as having. It would be mere Humean association. According to the reading I am considering, this
kind of association could put the creature in a position to act on the object, but it could not put it in a
position to have general, inferentially related thoughts about the object.
 KANT ’ S IDEALISM AND HIS REALISM
growing might appear as a thing with a unified spatial boundary, even though it is
ultimately discovered not to have the kind of causal unity that is involved in being an
object. We discover this by investigating its properties, including the way in which its
various properties relate to each other, and the principles of their unity. But doing
this, Kant thinks, requires starting with the idea that properties of an object have
necessary implications for each other, and that the thing causally interacts in law-
governed ways with other things. It is only because we start with the idea of an object
as a causal unity that we are able to investigate the plant-and-rock’s properties and
find out that we are not dealing with one object. Similarly, the particular implications
properties actually have for each other are contingent and learned from experience.
We do not know in advance of experience how an object’s properties interact with
each other, what implications this interaction has for how the object will interact with
other things and what kinds of changes it can survive. Kant’s position is that learning
all these things (empirically) requires being in a position to think of the thing as a
unified subject of properties and that it is the a priori concept of an object in general
which enables us to do this.
One way of reading Kant, associated with conceptualism about intuition, is to see
his strategy in the Deduction as being to argue that organising the sensory input
using a priori concepts is a requirement of having mental representations that
present us with distinct particulars in a world outside us. I have argued that rather
than being about the conditions of perceiving distinct particulars, Kant’s concern is
with what it takes to have a determinate object of thought, which is something more
than merely having acquaintance with a particular. On my reading, it is specifically
features of concept application (the generality and spontaneity of thought) rather
than the need for perceptual binding that generates the problem he is concerned with
in the Deduction. I have suggested that Kant thinks that the requirements of the
generality of thought (that attributing a property to an object requires being able to
think of it as one among a number of properties that are unified in belonging to that
object, and being able to think of the same property as potentially applying to other
objects) seem to be threatened by the spontaneity of thought (there being indefinitely
many ways in which the manifold of perceptual units given in intuition could be
classified). The combination of these two features of concept application threatens to
undermine our capacity to attribute empirical concepts to the objects given in
empirical intuition, because it threatens a situation of chaos in which we have no
determinate object of thought. His solution to this is that a priori concepts, as higher-
order ways of organising the ways things can be classified, determine what counts as
objects of thought, thereby making it possible for thought to have relation to an
object. Determining an object for thought is not a matter of discriminating a
particular out of the mass of sensation, but determining, of the particular presented
to intuition, how it can be thought of with the unity required to be thought of as a
subject of properties. As I read him, Kant’s transcendental argumentative strategy
starts off with the assumption that we do have successful empirical concept
THE TRANSCENDENTAL DEDUCTION 
application. He argues that this would not be possible without the application of a
priori concepts containing necessity and universality. This shows that these a priori
concepts are able to be brought into relation to objects given in empirical intuition,
and thereby shows that the categories do not lack relation to an object.
VII Relation to an Object and Idealism

I look in more detail at the implications of my account of the Deduction for Kant’s
idealism in Chapter 12. As a transition to this, in the final section of this chapter
I explain why, on my reading, Kant’s account of what it takes for thought to have
‘relation to an object’ is fundamentally epistemological rather than being an expres-
sion of idealism. Kant’s argument for the claim that we need a priori synthesis might
be taken to be based on his idealism; in addition, his claim that the categories give us
relation to an object might be understood as an expression of his idealism. The
argument might be thought to be based on idealism because it might be assumed that
Kant needs to invoke necessary connections to determine an object for thought only
because he is an idealist who does not allow that it is the way the world is that
determines what counts as an object. And it might be thought that the claim that the
categories give us relation to an object expresses idealism because it expresses
the thought that what counts as an object depends on our concepts. However, the
account of the argument I have presented is neither based on idealism nor an
expression of idealism. As I understand him, Kant’s concern with relation to an
object (both what it is and what it requires) is epistemological and is based on
problems which he would take to be problems for a realist. The problem is not
with what it takes for something to be an object (to exist) but with what it takes for us
to be in a position to think of anything as an object, and giving an account of the
requirements of referential thought can also be taken to be a problem by a realist. As
Kant explicitly says at the beginning of the Deduction, his concern is to argue that a
priori concepts determine an object not in terms of its existence, but ‘as an object of
cognition’, because, as he says, only through these concepts is it possible to cognize
anything as an object (B125). Thus, saying that the application of a priori concepts
gives our thought ‘relation to an object’ need not be understood as expressing the
idealist thought that our concepts determine what objects there are and what objects
are like; it could be taken as expressing the idea that we need to think of objects
as unified with necessity in order to have thoughts in which we attribute properties
to objects.
In opposition to this, the view that central claims in the Deduction depend on
(and/or express) Kant’s idealism might be taken to be supported by the following
dark passage:
We find, however, that our thought of the relation of all cognition to its object carries
something of necessity with it, since namely the latter is regarded as that which is opposed
 KANT ’ S IDEALISM AND HIS REALISM
to our cognitions being determined at pleasure or arbitrarily rather than being determined a
priori, since insofar as they are to relate to an object our cognitions must also necessarily agree
with each other in relation to it, i.e., they must have that unity that constitutes the concept
of an object.
It is clear, however, that since we have to do only with the manifold of our representations, and
that X which corresponds to them (the object), because it should be something distinct from all
of our representations, is nothing for us, the unity that the object makes necessary can be nothing
other than the formal unity of the consciousness in the synthesis of the manifold of the
representations. Hence we say that we cognize the object if we have effected synthetic unity in
the manifold of intuition. But this is impossible if the intuition could not have been produced
through a function of synthesis in accordance with a rule that makes the reproduction of the
manifold necessary a priori and a concept in which this manifold is united possible. (A104–5)
The claims that we have to do only with the manifold of our representations, that the
X which corresponds to them is ‘nothing for us’, and the unity that the object makes
necessary can be nothing other than the formal unity of the consciousness in the
synthesis of the manifold of the representation, could be taken as expressing idealism.
On this reading, the thought is that it is because we have no access to the world as it is
in itself (the independent object is ‘nothing for us’) that what counts as an object for
us is dependent on us. Kant is understood in this way by P. F. Strawson, who reads
him as saying that we need the categories because they are a kind of surrogate for the
way that, for a realist, objects constrain our thought (1966: 91). Similarly, Cassam
argues that Kant needs to invoke transcendental synthesis because the idea that the
object itself is ‘nothing for us’ follows from his idealism (Cassam 1987: 367).
In support of seeing the argument of the Deduction as depending on Kant’s
idealism, it could be argued that the Deduction might reasonably be expected to
assume what Kant thinks he has already established in the Aesthetic. And at certain
stages in the argument Kant appeals to the claim that we know only appearances (e.g.
A108, A111).25 However, assuming what is established in the Aesthetic does not
mean assuming that we make an a priori conceptual contribution to experience: it
could be that the argument of the Deduction depends on idealism with respect to the
forms of intuition, but not with respect to the forms of thought. And it could also be
that the argument of the Deduction depends on the idea that we know appearances
25
This is also assumed by Sebastian Gardner, who says that the Deduction ‘is an account of how the
world is (must be) constructed conceptually on the assumption that the fundamental conceptual features of
the objects of our cognition derive from our mode of cognition, rather than being determined by how
things are independently of our subjectivity’ (Gardner 1999:120, cf. 118. See also Ward 2006: 52). Gardner
cites a number of passages which he says show that the argument of the Deduction depends on
transcendental idealism (Bxvii; A92–4/B124–7; A95–7; A111; A114; A128; B163–4; B166–8; A139/B178;
A180–1/B223–4; Proleg. 26–30). While I agree with Gardner that these passages link the Deduction to
transcendental idealism, this does not establish his account of the order of explanation; in my view, that
order is the reverse of what he presents it as being. Apart from A111, these passages all express the idea that
transcendental idealism is needed to explain synthetic a priori metaphysical knowledge, and therefore
presumably could not be used as the premiss of an argument aimed at establishing that we have such
knowledge, or vindicating one of the components of such knowledge—a priori concepts.
THE TRANSCENDENTAL DEDUCTION 
only (manifest features of reality), without its depending on the mind-dependence of

appearances (that appearances are essentially manifest features of reality—mere
appearances).
A serious objection to making the argument of the Deduction depend on idealism
with respect to the categories is that this would establish the conclusion of the
argument too easily, and render the result unsatisfying. The argument would be
too easy, because if we start with the idea that experience is a function of our a priori
concepts, all we need is Kant’s derivation of the table of categories and we will already
have the result the Transcendental Deduction aims to establish. Kant thinks that we
have certain fundamental forms of judgment from which we can derive fundamental
concepts we use in thinking, and he thinks that it does not follow from this that such
concepts actually apply to anything; they could be merely subjective conditions of
thinking that lack objective validity. But if we start by assuming that the objects of
experience are a function of our fundamental concepts, then it seems we can already
conclude that these concepts apply to the world that we experience. Kant would not
need the tortuous argument of the Deduction if he started the section with the idealist
assumption that the objects of our experience are a function of the subjective
conditions of thinking: this would already show that the subjective conditions of
thinking apply to objects.
On Strawson’s and Cassam’s readings, it is because Kant is an idealist who thinks
that the world does not determine what we regard as objects that he needs to invoke a
contribution by minds (transcendental synthesis) to group (synthesise) features of
the manifold of intuition to determine what counts as objects for us. On this reading,
a realist would be in a position to claim that it is the way the world is that determines
that the roundness, redness, and squishiness of the tomato all belong to one object,
and thus that the way the world is provides the constraints on our thoughts that we
need for objective representation—to represent objects. In my view, this is a mis-
reading of Kant’s problem with concept application in the Deduction. The problem
raised by gerrymandered properties (by the spontaneity of thought) is not about what
it takes for an object to exist, but about what it takes for us to think about an object.
Kant takes both the spontaneity and the generality of thought to be features of
concept application in general, not expressions of idealism. The spontaneity of
thought means that the way objects are given to us does not determine how we
group or classify them, so does not put us in a position to have a determinate object of
thought; this would be true even if it is realist, mind-independent objects that are
given to us. There is a parallel here with the fact that the existence of medium-sized
spatio-temporal objects does not solve the binding problem: the problem of how the
brain manages to organise sensory input in order to represent unified objects. When
cognitive psychologists are trying to understand how the brain manages to unify
sensory input about colour and shape so as to enable us to represent coloured, shaped
objects, the solution is not simply to point out that there exist coloured, shaped
objects, independently of us. The problem is how we manage to represent these
 KANT ’ S IDEALISM AND HIS REALISM
objects, and realism does not make this problem go away. Similarly, as I read him,
Kant thinks that the existence of medium-sized spatio-temporal objects would not
solve the problem of explaining how we can be in a position to think of anything as
being an object. Kant thinks that merely being affected by objects (having sensations)
is not enough to have perceptual particulars, and he thinks that merely having
acquaintance with perceptual particulars is not enough to be in a position to think
of anything as an object.
Even Kant’s claim that the object itself is ‘nothing for us’ can be taken to express
something that Kant would regard as a problem for a realist. He is concerned with the
problem of what it takes to think about something as an object—as objective, as
distinct or independent from me. He says that the object ‘is regarded as that which is
opposed to our cognitions being determined at pleasure or arbitrarily’ (A104):
thinking of something as an object requires representing its properties as hanging
together in a way which is not arbitrary, or simply a result of the way we combine, a
way that is not simply up to us. In order to think of something as distinct from us we
need to represent its properties as combined in ways that are not up to us: we need to
represent it as something distinct from us that constrains how we combine those
properties. The problem, Kant thinks, is that we cannot succeed in thinking of
something as distinct from us and as constraining our thought simply by combining
a property we learn from experience with some further property we learn from
experience. For example, it would not solve our problem to have a representation
of a number of features (redness, roundness, etc.), and, in addition, have a further
representation of another property, such as the empirical principle of unity of this
kind of object. Kant thinks that for any further property of the object we can
represent, we can ask: what unifies it with the other properties? He says that adding
a further representation would ‘simply raise anew the question: How does this
representation in turn go beyond itself and acquire objective significance in addition
to the subjective significance that is proper to it as a determination of the state of
mind?’ (A197/B242). If we simply add further representations of which we can ask
the question whether, and in virtue of what, they are unified with the object, we have
not solved the problem of how anything becomes a unified object of thought. And if
we do not have a way of thinking about the object as the thing that constrains what
properties we can attribute to it, we are not in a position to think of it as having
properties at all.
His solution is that the way we represent a number of properties as belonging to an
object is not by adding some further property but by the way in which we represent
properties as united: with necessity. Representing an object as a causal unity—
representing its properties as causally unified with each other—is representing its
properties as having connections that involve necessity. This, Kant thinks, is what is
required for us to achieve a representation of something as independent of us,
distinct from us, and constraining our thought. Since all our specific representations
of an object’s particular properties are learned from experience, the way we represent
THE TRANSCENDENTAL DEDUCTION 
the unity of the object is through our use of a purely formal, non-empirical concept of
an object in general (something independent of me) which has no positive content
(it is merely something in general = x); it simply represents necessary unity. This is
why ‘the unity that the object makes necessary can be nothing other than the formal
unity of the consciousness in the synthesis of the manifold of the representations’
(A105). He says that the concept of a transcendental object = x provides relation to
an object and contains nothing but necessary unity (A109). It provides relation to an
object because it is only through bringing things under the non-empirical concept of
an object in general (representing things as combined with necessity) that we are
able to think about anything as an object other than us; it is only thus that we can
have an object of thought, something to which empirical concepts can be applied.
The a priori formal representation of the object enables us to unify our represen-
tations in the way that is necessary for any concepts to be brought to apply to objects
given to us in intuition. It provides thought with relation to an object because it
makes it possible for concepts to be applied to objects. Representing things as falling
under the categories (or the non-empirical concept of the transcendental object)
does not make it the case that there are objects, but it makes it the case that it is
possible for us to (conceptually) represent something as an object, and in this sense
it confers on our empirical concepts relation to an object, i.e. objective validity
(A108–9; A379/B350).
My concern in this chapter has been with understanding Kant’s account of relation
to an object in the Deduction and with his argument for thinking that the categories
are necessary for this.26 I have argued that in the context of the Deduction ‘relation to
an object’ is something thought has when it succeeds in referring to objects. For
thought to have relation to an object requires that there are things that are given in
intuition: things with which we have acquaintance. Further, we must be in a position
to bring the things given to us in intuition under concepts. Concepts lack relation to
an object when they cannot be applied to something given in intuition. Kant thinks
that to be in a position to apply concepts to things given to us in intuition we need to
think of them as existing distinct from us, and constraining our thought. The way we
do this is by regarding them as unified with necessity. We can fail to refer to an object
because nothing is presented to consciousness (because we do not have acquaint-
ance), and we can fail to refer to an object because nothing is singled out as an object
of thought. These are not the equivalent requirements and successful referential
thought requires both. He thinks the conceptual component is achieved by the way
we represent unity conceptually—through necessary connections.
26
I have said nothing about a further, central part of his case, which is that representing this unity
relation is essentially related to the possibility of self-consciousness.
12
The Possibility of Metaphysics
I Introduction
I have argued that Kant’s claim that a priori concepts are needed for thought to have
relation to an object does not depend on his idealism, and I have presented an
account of his argument for his idealism according to which it is not based on the
postulation that our minds shape objects in such a way as to make synthetic a priori
claims about objects come out as true. However, there is a close relation between
Kant’s transcendental idealism and his explanation of the only kind of metaphysics
he thinks is possible for us: his vindication of the categories and principles as
applying to all spatio-temporal objects. This delicate relation is the subject of my
final chapter.
We need to move slowly and carefully in getting right the relation between Kant’s
account of empirical cognition, a priori cognition, and his transcendental idealism. We
first need to see that Kant’s idealism is based specifically on concerns about apriority, not
the receptivity of empirical intuition or the spontaneity of empirical concept application;
I discuss this in the rest of this section. We then need to see that a priori concepts and a
priori intuitions do not play exactly parallel roles in Kant’s position; I discuss this in
Section II. Kant’s argument for his idealism is based on the apriority of our form of
intuition and he does not argue that the categories introduce further mind-structuring;
rather, he argues that the use of the categories for cognition is limited to the idealism
established in the Aesthetic. In Section III, I examine the relation between Kant’s idealism
and his vindication of metaphysics. As I argued in Chapters 7 and 11, Kant does not
vindicate the categories by arguing that they are needed for us to have a unified object
presented to us in experience, rather, they are needed for us to have a unified object of
thought. On the account of the argument I have presented, the role of the categories
in enabling us to have a unified object of thought is neither based on nor an
expression of idealism. However, it is related to his idealism as follows: once Kant
has established both that the categories are conditions of empirical cognition (the
central argument in the Deduction) and that spatio-temporal objects are limited to
the conditions of our cognition (the argument for idealism in the Aesthetic), he is
able to conclude that the categories (and, correspondingly, the principles contain-
ing them) apply to all spatio-temporal objects. The categories can be known to
apply to all spatio-temporal objects not because they are conditions of being given
THE POSSIBILITY OF METAPHYSICS 
objects in intuition (they are not ways of carving particulars out of a sensory mass)
and not because they are necessary for thought to have relation to an object (to
have determinate objects of thought to which empirical concepts can be applied),
but because spatio-temporal objects are limited to the conditions of our cognition
and the categories are conditions of our cognition. Thus, the final conclusion of the
Deduction (that all spatio-temporal objects are subject to the categories) does
require idealism, because the idealism established in the Aesthetic enables Kant
to convert the conditional conclusions of his epistemological arguments concern-
ing the conditions of empirical cognition to claims about spatio-temporal objects.
The first step is to pay attention to the fact that Kant’s idealism is based on specific
concerns with apriority, as opposed to receptivity and spontaneity as general features of
our cognition. Kant thinks that our intuition is characterised by the receptivity of our
sensibility: we needed to be affected by things to have them presented to our con-
sciousness. He thinks that our understanding is characterised by spontaneity: what is
presented to us in intuition does not determine how we classify things. Both receptivity
and spontaneity create problems that need to be solved by our minds before we can
have cognition of objects. Merely being affected by objects (having sensations) is not
enough to have perceptual particulars. Merely having acquaintance with perceptual
particulars is not enough to put us in a position to classify these particulars using
(general) concepts. Kant thinks that we need to bind (not his term) or order the mass of
sensory input (sensation) in order to represent perceptual particulars: things that are
distinct from and outside of ourselves and other things. He argues that applying
concepts to the particulars presented to us in intuition involves unifying the manifold
of intuition (and the manifold in intuitions) and grasping it as having been unified;
Kant calls this synthesis. I have argued that it is a misreading of Kant to confuse
binding and synthesis: to think that the unity introduced by conceptual thought is
needed for perceptual particulars. Further, in my view we will misunderstand him if we
think that his idealism follows merely from receptivity and/or spontaneity. He has been
taken in this way. P. F. Strawson (1966) and Langton (1998) take Kant’s idealism to
follow simply from receptivity: the fact that we have to be affected by things to have
knowledge of them. Walker (1985) argues that Kant’s idealism follows from spontan-
eity: the claim that the ways we combine things in bringing them under concepts is not
given by the world. These are examples of what Ameriks (1990) has called ‘short’
arguments for idealism: arguments which appeal to general features of cognition, and
bypass Kant’s specific concern with a priori knowledge.
As I have argued, there is a way of reading both Kant’s concern with (what we now
call) binding and his concern with synthesis that shows that there are questions here
that would not be solved by realism. Cognitive psychology investigates how the brain
manages to take received sensory input about both roundness and redness from, say,
a tomato, and integrate these to enable us to perceive a round red tomato, rather than
simply registering that redness and roundness are both present. This problem is not
solved by the realist insistence that it is in fact the way the world is that determines
 KANT ’ S IDEALISM AND HIS REALISM
that redness and roundness come together in the tomato. The problem concerns how
we get to represent this. Recognising that perceiving a round red tomato is an
achievement that requires active work by the mind by no means undermines the
idea that the tomato is round and red, independently of us. And Kant does not in fact
argue simply on the basis that sensation needs to be organised for his idealism; he
argues on the basis of the claim that space is the a priori form of our intuition. In
other words, on Kant’s account, it is not simply the need for binding that leads to
idealism, but specific features of the way we achieve binding. Similarly, Kant does not
argue that idealism follows from the fact that conceptual combination is not given by
the world. The spontaneity of thought means that the world does not determine
whether we classify objects under concepts such as green, or concepts such as grue,
and this is the case even if, in fact, grass is green and not grue. From the claim that
combination is not given Kant does not argue for idealism, but rather for the need for
a priori synthesis. Kant thinks all concept application, not just a priori concepts,
involves synthesis; his notion of synthesis concerns what it is to organise and order
objects in thought, rather than expressing a general idea of mind making objects.
Rather than looking at receptivity and spontaneity in general, we need to see the
specific roles of a priori intuitions and a priori concepts.
II A Priori Intuitions and A Priori Concepts

In the Aesthetic, Kant argues that our representations of space and time are a priori
intuitions, that they are mind-dependent, and that they do not present us with things
as they are in themselves. He takes it to follow from this that objects in space and time
are mind-dependent. One way of reading his complete account is that a priori
concepts then introduce further ‘mental structuring’, so appearances are mind-
dependent because they depend for fundamental aspects of their nature (their
structure) on both our a priori intuitive and our a priori conceptual contributions.
Thus, Kant’s position can be associated with a kind of conceptual idealism which
holds that our conceptual divisions represent only the way our minds work, and
which sees the world as carved into objects by our concepts. I will argue, in contrast,
that the idealism fully depends on the role of a priori intuitions, that a priori concepts
do not introduce further mind-dependence to the objects of experience, and that
seeing the argument of the Deduction as epistemological is fully compatible with
seeing transcendental idealism as a robust idealist position.
Kant’s summary presentation of his argument for transcendental idealism in the
B Preface might be thought to present a straightforward parallel between the contri-
bution of a priori intuition and that of a priori concepts in ‘constituting’ the objects of
our cognition. He first says that
If intuition has to conform to the constitution of objects, then I do not see how we can know
anything of them a priori; but if the object (as an object of the senses) conforms to the constitution
of our faculty of intuition, then I can very well represent this possibility to myself. (Bxvii)
THE POSSIBILITY OF METAPHYSICS 
He then says that the same applies with respect to concepts:

because I cannot stop with these intuitions, if they are to become cognitions, but must refer
them as representations to something as their object and determine this object through them,
I can assume either that the concepts through which I bring about this determination also
conform to the objects, and then I am once again in the same difficulty about how I could know
anything about them a priori, or else I assume that the objects, or what is the same thing, the
experience in which alone they can be cognized (as given objects) conforms to those concepts,
in which case I immediately see an easier way out of the difficulty, since experience is itself a
kind of cognition requiring the understanding, whose rule I have to presuppose in myself
before any object is given to me, hence a priori, which rule is expressed in concepts a priori, to
which all objects of experience must therefore necessarily conform, and with which they must
agree. (Bxvii)
Here Kant seems to say that the same kinds of considerations apply to a priori
concepts as apply to a priori intuitions, and he concludes that ‘we can cognize a priori
of things only what we ourselves put into them’ (Bxviii), implying that both a priori
intuition and a priori concepts are something that we ‘put in’ to objects. This suggests
that the Copernican Revolution applies in roughly the same way for a priori concepts
and a priori intuition: just as space and time are merely ways in which we structure
experience, so too the categories structure the objects of our experience and do not
represent the way things mind-independently are.
However, in the same preface, Kant also summarises his view in a way which
suggests that the idealism is established in the Aesthetic, and the point of the
argument in the Deduction is to limit our a priori concepts to the mind-dependent
appearances that are presented in space and time, not to introduce another layer of
mind-dependent structuring. Kant says:
In the analytical part of the critique it is proved that space and time are only forms of sensible
intuition, and therefore only conditions of the existence of things as appearances, further, that
we have no concepts of the understanding and hence no elements for the cognition of things
except insofar as an intuition can be given corresponding to these concepts, consequently that
we have cognition of no object as a thing in itself, but only insofar as it is an object of sensible
intuition, i.e. as an appearance; from which follows the limitation of all even possible specu-
lative cognition of reason to mere objects of experience. (Bxxv–xxvi)
As Kant presents the position here, it follows from the ideality of space and time that
the spatio-temporal objects we experience are mere appearances: the objects we
experience are determined by the a priori forms of our intuition. The impossibility
of having knowledge of things as they are in themselves through the categories
follows from the fact that all our concepts (a fortiori, our a priori concepts) require
objects given in intuition. The ideality is established through the role of space and
time, and the further argument with respect to the categories serves to limit them to
the ideality of spatio-temporal objects, not to introduce further mind-dependent
structuring. I will argue that there are a number of reasons for reading Kant’s position
 KANT ’ S IDEALISM AND HIS REALISM
in this way, and that this is compatible with still thinking that the idealism plays an
essential role in the complete vindication of the categories and the explanation of the
possibility of metaphysics. The summary in the B Preface which seems to suggest the
alternative view should be read, I argue, as an account of Kant’s explanation of
metaphysics, which is based on his idealism, rather than as an account of his
argument for his idealism.
If we look at the details of the arguments put forward in the Aesthetic and the
Analytic, as well as the conclusions Kant draws from them, a priori intuitions and a
priori concepts do not play parallel roles.1 At the end of the argument in the
Aesthetic, Kant’s explicit conclusion is that space and time are merely the a priori
forms of our intuition, and that they present no features at all of things as they are in
themselves (A26/B42). He does not make this kind of claim about the categories,
saying rather that the categories cannot be used for cognition of any objects other
than the spatio-temporal appearances that are given to our intuition (A246/B303).
Rather than saying that the categories represent no features of things as they are in
themselves, Kant says that the categories are limited to objects that can be given to
our spatio-temporal intuition. The claim that the categories represent no features of
things as they are in themselves would be hard to make sense of, because they are
supposed to be derived from the basic forms of thinking, and because Kant in fact
says that we can use pure categories to have mere thoughts about things which are
not objects given in experience. He says that ‘Once I have pure concepts of the
understanding, I can also think up objects that are perhaps impossible, or that are
perhaps possible in themselves but cannot be give in experience, since in the
connection of their concepts something may be omitted that yet necessarily belongs
to the condition of a possible experience (the concept of a spirit)’ (A96). Unlike his
stronger conclusion with respect to the forms of intuition, Kant does not say that the
categories represent no property of things in themselves, but rather that they do not
enable us to cognize anything about objects other than the objects that are given to
our senses.
The different conclusions Kant draws with respect to the a priori forms of intuition
and a priori concepts are exactly what we should expect, on the account of the
1
For helpful discussion of the lack of parallel between the categories and a priori intuition, see Thomas
Pogge (1991) and Eric Watkins (2002), who both say that this is something to which commentators have
not paid enough attention (Pogge 1991: 493; Watkins 2002: 196). Prauss (1974) is one commentator who
does comment on this issue; he seems to see it as a question of presentation, and as a hangover from the
Inaugural Dissertation, in which Kant sees the objects presented to us in sensibility as mind-dependent, but
still thinks that reason can give us knowledge of reality in itself (Prauss 1974: 59, 59n12). However, the fact
that the full critical position goes beyond the Inaugural Dissertation in denying that reason can give us
cognition of things in themselves is compatible with more than one account of the role of the categories: it
could be that the categories add an extra ‘layer’ of mind-dependence, in that they structure objects in a
parallel way to the way the a priori forms of intuition do, or it could simply be that the categories are
limited to the mind-dependent objects presented to us by sensibility (which still takes us beyond the
argument of the Inaugural Dissertation).
THE POSSIBILITY OF METAPHYSICS 
argument for the idealism I presented in Chapter 8. Kant’s reason for saying that the
categories can be used for cognition only of spatio-temporal appearances is that, he
thinks, without relation to sensible intuition, the categories lack objects (A246). Kant
says that space and time are conditions of the possibility of objects being given to us
and are valid only for objects of the senses, but that the categories are free from this
limitation. Their meaning is not tied to the spatio-temporal objects of the senses, and
they can be used to produce coherent thoughts about non-spatio-temporal objects.
However, this extension ‘does not get us anywhere . . . For they are then merely empty
concepts of objects, through which we cannot even judge whether the latter are
possible or not—mere forms of thought without objective reality—since we have
available no intuition to which the synthetic unity of apperception, which they alone
contain, could be applied, and that could thus determine an object’ (B148; see also
A248/B304, A236/B296, and A246/B303). The point about the categories is not that
they could not be used to think about things in themselves, but that without what is
given to us in intuition they do not enable us to have cognition of anything. In
contrast, with respect to a priori intuition Kant thinks that the kind of representation
that it is entails that it could not present a mind-independent feature of reality. He
says that ‘[a]n intuition is a representation of the sort which would depend imme-
diately on the presence of an object. It therefore seems impossible originally to intuit
a priori, since then the intuition would have to occur without an object being present’
(Proleg. 4: 282). With respect to a priori concepts, this is not the case; ‘[c]oncepts are
indeed of the kind that we can quite well form some of them for ourselves a priori
(namely, those that contain only the thinking of an object in general)’ (Proleg. 4: 282).
Rather, what follows from their being concepts is that they cannot have objects unless
they are related to something given in intuition: ‘these [concepts] still require, in
order to provide them with signification and sense, a certain use in concreto, i.e.,
application to some intuition or other, by which an object for them is given to us’
(Proleg. 4: 282). Therefore, the categories are limited to what is presented to our
intuition.2 We can coherently think about things in themselves, and, arguably, we
cannot think about them without thinking that they are subjects of predicates and
subject to ground-consequence dependence (even though we have no knowledge of
what noumenal subjects are like, or of what causation would be like in things in
themselves). The limitation on a priori concepts is not that it is impossible for them
to represent features of something entirely mind-independent, but that they never
succeed in referring to anything entirely mind-independent, because they do not
supply their own objects and the only objects that are given to us are sensible
appearances.
This fits with the account of the argument in the Deduction that I presented in
Chapter 11. As I read him, Kant’s problem with what it takes for concepts to have
2
Similar points are made by Watkins (2002: 197).
 KANT ’ S IDEALISM AND HIS REALISM
relation to an object concerns the a priori requirements of successful referential

thoughts. Kant argues that the categories are necessary conditions of empirical
concepts being applied to objects presented to us in intuition, and that this shows
that the categories apply to the objects presented to us in empirical intuition. In
other words, we show that the categories have relation to an object by showing
that they are the condition of any concept’s having relation to an object. Without
the application of the categories, nothing would count as an object of thought for
me, and therefore my empirical concepts would lack relation to an object. The
categories give us relation to an object in that without them we would not be in a
position to think of anything as an object, not because they make it the case that
anything is an object. This is an epistemological argument, which is a response to
an empiricist who thinks we could account for cognition of objects using only
materials derived from experience, and therefore without the idea of necessary
connections. He says: ‘the representation . . . does not produce its object as far as
its existence is concerned, the representation is still determinant of the object
a priori if it is possible through it alone to cognize something as an object’
(A92/B125).
On my reading, Kant’s transcendental idealism is entirely established in the
Aesthetic, so the mind-dependence of empirical objects comes from the mind-
dependence of a priori intuition, and the categories are then limited to being used
to cognize the mind-dependent appearances presented to us in intuition, rather than
introducing further mind-structuring. This is not a deflationary reading: it is com-
patible with thinking that a full-blown, robust idealism is established in the Aesthetic.
However, it is in agreement with part of what motivates deflationary readings: the
thought that one of Kant’s concerns in the Deduction is with epistemological
concerns about the a priori conditions of cognition.
Saying that the idealism is fully established in the Aesthetic would not mean that
nothing is added to the position by the Analytic. The Analytic adds two important
pieces to Kant’s case. First, it furthers his argument as to why we cannot have
knowledge of noumena. Showing spatio-temporal objects to be mind-dependent
appearances might not rule out all possible knowledge of things in themselves if
knowledge through pure concepts of things in themselves were a possibility. The
Deduction and the following section on ‘Noumena and Phenomena’ rule this out.
Not only are the objects that are given to our senses mind-dependent appearances,
further, we cannot cognize any other kinds of objects using the categories, so our
knowledge is limited to spatio-temporal appearances, and we can have no knowledge
of intelligibilia. Second, it completes his explanation of how synthetic a priori
metaphysical knowledge is possible, by showing how synthetic a priori claims can
concern objects presented to us in empirical intuition. It is in the explanation of the
possibility of metaphysics, in my view, that Kant’s argument in the Deduction relates
most centrally to his idealism. As I will argue, seeing part of the argument of the
Deduction as epistemological, and as neither based on nor expressing idealism, is
THE POSSIBILITY OF METAPHYSICS 
compatible with still thinking that there is an important connection between Kant’s
idealism and his full vindication of the categories.
III The Possibility of Metaphysics

In Chapter 8, I argued that Kant’s explanation of the possibility of synthetic a
priori judgments in general is not the same as his explanation of the possibility of
metaphysical synthetic a priori judgments, and that idealism relates to the two in
different ways: as I understand it, transcendental idealism follows from the general
explanation of the possibility of cognition of synthetic a priori judgments (a priori
intuition), and leads to an explanation of the possibility of metaphysics. Further,
I have argued that Kant’s central argument for the claim that the categories are
needed for thought to have relation to an object is epistemological. However, as
I read his argument, transcendental idealism is needed for the final stage of Kant’s
argument in the Deduction: his showing that the categories apply to all spatio-
temporal objects. This then enables us to show how the a priori principles can
apply to all spatio-temporal objects, and how principles which are established
merely as conditions of cognition can be known to be true of all (spatio-temporal)
objects: ‘thus the possibility of as it were prescribing law to nature and even making
the latter possible, is to be explained’ (B160). This a priori knowledge of conditions
of the possibility of objects of cognition is the only kind of metaphysics that is
possible for us.
According to the account I presented in Chapter 8, Kant’s primary question with
respect to synthetic a priori judgments is how it is possible for them to concern
objects with which we have acquaintance, and therefore to qualify as cognition. He
thinks that cognition requires acquaintance, that intuition is what gives us this, and
that intuition typically requires objects affecting us. This makes it seem hard to see
how synthetic a priori claims could qualify as cognition. Kant’s answer to his ‘how
possible?’ question is a priori intuition: because it is intuition it presents us with
something singular and immediate; because it is a priori it can provide a singular
and immediate object for a priori claims. He thinks that this explains the possibility
of geometry: it is possible for synthetic a priori mathematical propositions to
qualify as cognition if we have an a priori intuition that presents their objects.
Kant takes idealism to follow from his account of a priori intuition: as I read him,
he thinks that objects that are present to consciousness without anything affecting
us could not be independent of us. Kant then argues that space and time are a priori
intuitions and that they are the a priori forms of our intuition. He takes it to follow
from this that our representations of space and time do not present us with things
as they are in themselves, are merely the forms of our intuition, and therefore that
objects in space and time are not things that exist apart from the possibility of their
appearing to us.
 KANT ’ S IDEALISM AND HIS REALISM
At the end of the Aesthetic, Kant thinks he has established the transcendental
ideality of space and time and therefore of objects in space and time. He also says we
have one part of the answer to the question of how synthetic a priori cognition is
possible: this part of the answer is a priori intuition, which explains how synthetic a
priori mathematical claims can have given objects. We do not yet have a complete
explanation of how we could have metaphysical knowledge. Like mathematical
claims, metaphysical claims are synthetic and a priori. Like all synthetic claims,
metaphysical claims require objects given to us in intuition (objects with which we
can have acquaintance), but unlike mathematical claims they do not have objects
which are given to us independently of experience, purely in a priori intuition. They
‘cannot exhibit any object in an a priori intuition on which to ground their synthesis
prior to any experience’ (A88/B120). Metaphysical claims are more problematic than
mathematical claims, because we cannot justify them by appealing to experience but
neither can we explain their possibility in the way we explain mathematical know-
ledge, which can exhibit its objects a priori.
There are two further steps to Kant’s vindication of metaphysical claims. First,
Kant argues, in the Deduction and Principles that certain a priori concepts and a
priori principles containing these concepts are conditions of the possibility of
empirical cognition. These conditional arguments, in my view, are independent of
idealism, but are an important step in Kant’s vindication of metaphysics. Showing,
for example, that the application of a priori concepts is a necessary condition of
empirical concept application is something that could be accepted by a realist; the
argument (if successful) establishes the conditional claim that, if we can succeed in
applying empirical concepts to an object, the categories apply to it. This conditional
part of the argument is important because it shows how it is possible for a priori
metaphysical concepts to relate to objects. These concepts are a priori, yet they do not
have any objects that are presented in pure intuition. The only way they could relate
to objects, therefore, is if they relate to objects given in empirical intuition. As we
have seen, Kant aims to demonstrate this by showing that a priori rules of synthesis
of what is given to us in intuition are the condition of the possibility of conceptual
thought about objects; they can be known to apply to objects given to us in intuition
because they are the condition of any concept’s being successfully used to think about
an object given to us in intuition. Since Kant assumes that we have experience
(empirical cognition, empirical concept application), this shows that the categories
do not lack relation to objects (they can succeed in being brought into relation to
something given in intuition), but it does not yet establish the stronger claim Kant
wants to establish: that all spatio-temporal objects are subject to the categories.
A similar point can be made with respect to the principles. Kant thinks that we can
vindicate the principles by showing them to be conditions of the possibility of specific
features of empirical cognition, just as the categories are vindicated as conditions of
the possibility of empirical concept application. A strategy that vindicates certain
principles by showing them to be conditions of the possibility of empirical cognition
THE POSSIBILITY OF METAPHYSICS 
enables us to establish the conditional claim that such principles are true of all the
objects which we can cognize (if we can cognize something, it will, for example, be
subject to the causal principle, and anything we can cognize will be made of
substance that is conserved).3 If we have a sound transcendental argument showing
that a principle is a condition of the possibility of empirical knowledge, then we know
that this principle is true of all the objects of which we can have knowledge.
Kant thinks further explanation is needed because he wants to go a step further. He
does not just want to establish the conditional claim that synthetic a priori principles
are true of the objects we can cognize, but that they are true of spatio-temporal
objects. This is the point at which transcendental idealism makes possible the
explanation of metaphysics. Idealism enables him to make this further move because
it enables us to convert conditional claims about what we need for cognition into
claims about spatio-temporal objects. In particular, the form of idealism that will do
the work Kant needs here is not phenomenalism, but rather an idealism that limits
spatio-temporal objects to our possible experience (or cognition) of them—exactly
the form of idealism I have attributed to him. Kant is able to go beyond a conditional
claim about a priori concepts being necessary conditions of empirical concept
application to the claim that spatio-temporal objects are all subject to the categories
because he has already shown that spatio-temporal objects are mind-dependent
appearances that are limited to the conditions of our being able to cognize them.
Similarly, he is able to go beyond conditional principles about the objects that we can
cognize to principles that apply to spatio-temporal objects because he has shown that
spatio-temporal objects, and space and time themselves, are limited to the conditions
of our cognizing them. Once we have established that spatio-temporal objects do not
exist apart from the possibility of our experience (cognition), it will follow that
principles which are conditions of the possibility of experience (cognition) hold of
spatio-temporal objects. Thus, idealism enables us to explain how cognition of
synthetic a priori metaphysical claims is possible.
On this reading, the role of idealism is not to explain how the categories apply to
objects: this is the job of the epistemological argument that shows the categories to be
conditions of the possibility of empirical concept application. Rather, the role of
idealism is to enable us to convert a claim about the conditions of empirical cognition
to a claim about spatio-temporal objects. This is possible if we have shown that
spatio-temporal objects do not exist independently of the possibility of our cognizing
them. This shows that idealism can be seen to be integral to Kant’s explanation of the
possibility of metaphysics without being seen as a premiss of his transcendental
arguments concerning the conditions of the possibility of empirical cognition. As
3
Kant says that ‘if one will look back on our proof of the principle of causality, one will become aware
that we could prove it only of objects of possible experience . . . and indeed we could prove it only as a
principle of the possibility of experience, hence of the cognition of an object given in empirical intuition,
and not from mere concepts’ (B289).
 KANT ’ S IDEALISM AND HIS REALISM
I read him, Kant does not argue that it is because the categories are conditions of the
synthesis of the manifold of intuition that they apply to all spatio-temporal objects.
Rather, he first argues that the categories are a priori rules of synthesis that are
needed for empirical concept application. This shows that the categories do not lack
relation to an object. Since the argumentative strategy is to show that the categories
are necessary for empirical concept application, the argument, if successful, shows
that they apply to all objects that we can cognize. However, because he has already
shown that the objects that we can cognize are spatio-temporal appearances that are
limited to the conditions of our cognition, we can conclude that the categories apply
to all spatio-temporal objects.
The same point applies to the argument concerning the principles. As we saw in
Chapter 8, a simple reading of Kant’s argument for his idealism sees him as saying
that idealism explains our knowledge of synthetic a priori claims because it enables us
to see how we could have access to such claims. The idea is that our access to, or
insight into, what our minds are responsible for is unproblematic, so postulating the
explanation that our minds are responsible for making certain synthetic a priori
claims true of the world enables us to see how we can access such claims. On this
reading, transcendental idealism explains the possibility of synthetic a priori know-
ledge, and the fact that it explains this knowledge is how we establish transcendental
idealism. I argued against this reading on a number of grounds, including the fact
that it attributes to Kant a very weak argument, and one he does not actually present.
We can establish transcendental idealism by arguing that it explains our knowledge
of synthetic a priori claims only if we start with claims of which we actually have
knowledge. However, at the point at which Kant argues for transcendental idealism,
he does not take himself to have established any synthetic a priori metaphysical
claims and regards metaphysics as still being in doubt. In my view he does start with
the assumption that we have knowledge of synthetic a priori mathematical claims,
but even if mathematical knowledge could be explained on the basis of the idea that
our minds make objects in such a way that mathematical claims come out true, this
would not establish that our minds make objects in such a way that metaphysical a
priori claims come out true.
My reading of the argument, in contrast, says that the general explanation of the
possibility of synthetic a priori cognition is a priori intuition. This leads to idealism.
Independently of idealism, we can establish (against the empiricist) claims about the
a priori conditions of empirical knowledge. When we put the arguments establishing
these conditional claims together with the conclusion of the Aesthetic, we can explain
the possibility of metaphysics. Synthetic a priori conditions that apply to all the
objects of which we have experience can be known to apply to all spatio-temporal
objects, because we have shown that spatio-temporal objects are not things which
exist independently of our experience. Idealism does not explain our knowledge of
geometry, but rather follows from the explanation of geometry (a priori intuition);
idealism is not established by, but rather makes possible, the explanation of
THE POSSIBILITY OF METAPHYSICS 
metaphysics. Idealism is not what shows that a priori concepts and principles apply
to the objects of our knowledge, but rather, it enables us to show that a priori
concepts and principles that are conditions of empirical knowledge apply to all
spatio-temporal objects.
As many commentators have pointed out, successful transcendental arguments
concerning conditions of the possibility of experience seem to establish conditionally
necessary claims, and these claims do not look particularly mysterious. If we have a
successful transcendental argument for the claim that every event has a cause, then
we have shown that we can have empirical knowledge only of events which have
necessitating causes.4 Given this, there does not appear to be much mystery about
our knowing that every event that we can experience will have a necessitating cause.
Because of this, Guyer has argued that Kant’s concern must be not just with
conditional necessities but with absolute necessities. As Guyer sees it, Kant’s claim
is not that, necessarily, if we are to perceive an object then it is spatial and Euclidean,
but that if we can perceive an object then it is necessarily spatial and Euclidean (1987:
361–3). A problem with this reading is that attributing to Kant the view that the
objects that we perceive are spatial and Euclidean in all possible worlds is not
compatible with Kant’s belief that the necessities in question are not analytic.5
Further, as Ameriks points out, knowledge of absolute necessity is in fact incompat-
ible with, rather than uniquely explained by, transcendental idealism: ‘Kant
repeatedly claims that we cannot make absolute modality claims about phenomenal
features; thus the Fourth Antinomy indicates that we need to be agnostic about
saying that the world is absolutely necessary or that it is absolutely contingent’
(Ameriks 1992: 337; see A593–4/B621). In my view, Guyer is right to think that
Kant does not take transcendental idealism to follow from our knowledge of the
conditional necessities that his transcendental arguments establish. However,
I disagree with his view that Kant argues for his idealism on the basis of claiming
that we have knowledge of unconditional necessities. Rather, the point is that
establishing idealism (independently) enables us to explain how knowledge of con-
ditionally necessary principles of knowledge is in fact knowledge of necessary truths
about spatio-temporal objects. It is not that we have knowledge of unconditional
necessities, and therefore need to invoke idealism to explain this. Rather, once we
have established idealism, we can explain how our knowledge of conditional claims
(all the objects we can know are subject to the principles of causation and conserva-
tion) can be taken to have unconditional status (all objects are subject to these
4
This is of course a big ‘if ’, and I have given no account of any of Kant’s specific transcendental
arguments; my concern is with how his general argumentative strategy relates to his idealism.
5
Kant explicitly says that the claim that ‘all things are next to each other in space’ is valid only
conditionally—given that these things are taken as objects of our sensible intuition (A27–8/B43–4). Kant
also says that ‘neither absolute nor relative determinations can be intuited prior to the existence of the
things to which they pertain, and thus be intuited a priori’ (A26/B42), suggesting that his concern is not
with knowledge of absolute necessity, but with cognition of synthetic a priori propositions more generally.
 KANT ’ S IDEALISM AND HIS REALISM
principles). We can make this move so long as we restrict it to spatio-temporal

objects, which have been shown not to exist independently of the possibility of our
cognizing them. This is exactly how Kant presents his move from a conditional to an
unconditional claim in the Aesthetic.
The understanding of the argument that I have presented has implications for the
way we understand the idealism. Transcendental idealism may be associated with a
conceptual idealism which holds that our concepts carve up reality in a way that
determines what objects there are. It is also associated with the idea that metaphysics
is possible because it is nothing more than insight into our minds, and, in specific,
ways in which our minds shape objects. However, neither with respect to mathem-
atical nor with respect to metaphysical synthetic a priori claims does Kant explain
this cognition by saying that our minds make objects in such a way that the claims
come out true. With respect to mathematics, he says that the explanation of the
possibility of this cognition is that its objects can be presented in a priori intuition.
With respect to metaphysics, we have claims which are established as conditions of
the possibility of empirical cognition; we can assert that they are true claims about the
spatio-temporal objects of our cognition because we have already established that
spatio-temporal objects are mind-dependent and do not exist independent of the
possibility of our cognizing them. Our insight into these claims is not a matter of
insight into how our minds make things, but insight into the conditions of cognition.
Such insight becomes insight into objects, if we have established that the objects in
question do not exist independently of the possibility of our having knowledge of them.
Thus, we can take the idealism seriously, can see Kant as asserting genuine mind-
dependence about appearances, and can see this as playing a central role in his
explanation of metaphysics, without understanding his idealism as saying that our
minds shape objects according to the categories. The categories (and principles) are
conditions of cognition which can be known to be true of spatio-temporal objects
because such objects are not independent of the possibility of their being cognized by us.
We have seen that Kant’s idealism combines epistemological and metaphysical
concerns, as well as realism and idealism. His position is realist in that he holds that
(empirically real) objects exist outside of our minds, and in that he holds that they are
grounded in a reality that is entirely independent of us. Further, as we saw in the
Deduction, he thinks that we cannot think about things as objective—of an object as
something distinct from me that constrains my thought—without referring our
representations to something we think of as entirely independent of us. The thought
of something entirely independent of us does not feature in Kant’s account only
because he thinks that a relational ontology could not be complete; it also features in
thoughts that achieve objectivity. At the same time, however, he thinks that such
thoughts do not enable us to cognize anything as it is independent of the possibility of
being presented to us, things as they are in themselves.
My reading of Kant’s idealism sees it as turning, centrally, on the idea of what can
be given to or present in a conscious experience (the argument in the Aesthetic). Kant
THE POSSIBILITY OF METAPHYSICS 
thinks that cognition requires the possibility of acquaintance with its objects and that
intuitions are the representations that give us acquaintance. Intuitions immediately
present their objects to consciousness. Kant thinks that the way we are presented
things as outside of and distinct from ourselves and each other is by representing
them as spatio-temporally located. He thinks that this is not a matter of representing
objects as falling under certain descriptive criteria, but rather as representing them in
a structure or form that is immediately present to us and that has features, such as
being oriented, that relate it to a subject. Kant thinks that representing this structure
or form enables us to represent things as outside and other than us, but also that it is
essential to this structure that it does not present us with a view from nowhere, or
with something that exists independently of the possibility of its being present to
consciousness. He thinks that it does not make sense to think of this structure as
existing independently of the possibility of its being presented to us; what it is is a
structure in which things are (possibly) presented. And he takes this mind-dependence
to apply to what is presented in this structure.
Kant also thinks that achieving cognition requires more than acquaintance: it
requires the possibility of conceptual thought, and this in turn requires thinking of
something independent of us that constrains our thinking. Since he holds that we
never have acquaintance with anything as it is in itself, achieving this thought cannot
be a matter of putting together representations that are given to us, but rather, is
achieved by representing what is given to us as governed by necessary connections. In
representing what is given as governed by necessary connections we are able to
represent an objective world, but what we connect in this way is always something
with which we can have acquaintance, and which is limited to the mind-dependent
structuring conditions of acquaintance. The conditions of thought require that we
organise thought around the idea of something independent of us; the conditions of
acquaintance limit what we can be acquainted with. We can cognize only what is
essentially manifest; we cannot think of the world coherently without thinking of it as
containing more than what is essentially manifest. The thought of something entirely
independent of us is necessary, but does not enable us to cognize anything as it is
entirely independent of us.
IV Freedom of the Will

The synthetic a priori claims about spatio-temporal objects I have been discussing so
far constitute Kant’s metaphysics of experience. Although he thinks that these are the
only metaphysical claims with respect to which theoretical cognition is possible, he
also thinks that there is a way in which transcendental idealism enables us to solve
transcendent metaphysical problems. In keeping with the fact that Kant’s position
has both metaphysical and epistemological components, part of the solution is
understanding what kinds of things are beyond the limits of our (theoretical)
cognition, and part is understanding the nature of that which we can cognize. Of
 KANT ’ S IDEALISM AND HIS REALISM
all the metaphysical problems transcendental idealism is supposed to solve, provid-

ing a resolution to the free will problem would be the biggest pay off. While this topic
is beyond the scope of this book, I will make some brief comments about the
possibilities my interpretation of transcendental idealism opens for understanding
Kant’s strategy for dissolving the free will problem. These comments are speculative
notes about avenues for research, not argued conclusions about how to understand
Kant on this topic.
Kant has what is now called an agent-causal libertarian account of freedom of the
will. He thinks that having freedom would involve a different kind of causality to the
causality of nature. Where the causality of nature involves events falling under laws
which shows that they have to follow from previous states of the universe, the
causality of freedom would involve a causal capacity to initiate a new causal chain
that is not a determined function of previous states of the universe and the laws of
nature. Kant thinks that if there is only the causality of nature, freedom is impossible.
He thinks he has proved that all events in space and time are subject to deterministic
causation (in the Second Analogy); his aim is to show how we can still allow that the
alternative causality of freedom is possible. In other words, he wants to show that
agent causal libertarian freedom is not ruled out by deterministic causation (rather
than by appealing to indeterminism).
Kant does not think that we can have theoretical cognition of the causality of
freedom. His resolution of the free will problem is situated in his critique of
transcendent metaphysics: he thinks that freedom is a transcendent metaphysical
idea. So we cannot (using theoretical reason) cognize the causality of freedom. In
terms of theoretical reason, we cannot even really understand what the causality of
freedom would be. We can characterise it very abstractly as the capacity to initiate a
new causal chain or negatively as a causality that is not the causality of nature, but
this gives us no positive insight into what this alternative causality is.6 Kant’s aim
with respect to freedom in the Critique is simply to show that it is possible that the
causality of freedom exists.
6
However, we do, Kant thinks, have practical cognition of freedom, in at least two ways. Our belief in
the moral law commits us to believing in freedom, and practical reason gives us cognition of the causality of
freedom. Understanding Kant’s account of practical cognition of things of which we cannot have
theoretical cognition is complicated and not the topic of this book; I will simply make one comment
about it here. As we have seen, cognition is not the same as knowledge, and is not distinguished by having
some specified kind of warrant or justification, but rather by the kind of representation of objects it is able
to give. This means, in my view, that it is a mistake to think of practical cognition of freedom as involving a
different way of justifying claims about freedom; rather, practical cognition gives us a different way of
representing or giving content to the causality of freedom. We cannot even understand this causality
theoretically, but we are able to have a representation of it as a form of causality in the moral law. The idea,
as I understand it, is that the only positive understanding we have of the causality of freedom is that it
involves a capacity to initiate actions in ways that are governed by higher order rational principles, and, in
particular, higher order rational principles that rational agents are committed to independently of their
particular projects, goals, and desires (and in virtue of thinking of themselves as acting for reasons).
THE POSSIBILITY OF METAPHYSICS 
Kant thinks that all events in space and time are subject to deterministic causality
and that our actions are events in space and time; he wants to show that this does not
rule out its being possible that our actions are due to the causality of freedom.
Transcendental idealism is supposed to be what enables him to show this. We have
seen two extremes in interpreting transcendental idealism: deflationary interpret-
ations which see him as distinguishing between two ways of considering objects; and
extreme metaphysical interpretations which see him as a noumenalist and a phe-
nomenalist. Correspondingly, there are two extreme strategies for understanding
how transcendental idealism is supposed to help with the free will problem. Corres-
ponding to merely epistemological readings we have the idea of two different points
of view we can take on agency: the point of view of empirical causal explanation; and
the point of view of freedom and morality. A problem with this is that it is hard to see
how it explains the possibility of freedom as an alternative kind of causality. Corres-
ponding to extreme metaphysical interpretations we have the idea of timeless
noumenal agency being free from the constraints of deterministic causation in
space and time. A problem with this is that it makes freedom utterly mysterious
and seems to divorce it from the empirical actions for which we want to be able to
hold people responsible. I suggest that my moderate metaphysical interpretation
points to ways in which transcendental idealism helps with free will without simply
saying that something utterly mysterious is possible: how it is supposed to actually be
a solution.
I have argued that Kant’s idealism about appearances is not a form of phenom-
enalism. It is hard to see how phenomenalistic idealism about spatio-temporal
appearances would help us to show that events in space and time are subject to
deterministic causation and also could be subject to the causality of freedom. In
contrast, I think my interpretation does point to a way of making sense of this. The
crucial point, as I see it, is a number of different ways in which Kant takes science to
be, in principle, incomplete. The anti-realism I have attributed to Kant involves
denying that the spatio-temporal empirical reality we cognize exists as a determinate
totality. It does not have a first moment in time and it does not have a most
fundamental micro-physical level. Although, Kant thinks, science is driven by a
search for systematic unity or a complete theory of everything, he also thinks that
systematic unity is, in principle, not something we could reach.7 The world of
appearances does not exist as a determinate totality, so could not exist as a system-
atically unified totality. So the causal explanations science gives can never reach the
7
He says that the ideal of systematic unity is ‘only an idea (focus imaginarius)—i.e., a point from which
the concepts of the understanding do not really proceed, since it lies entirely outside the bounds of possible
experience’ (A644/B672; see also A644/B672 and A669/B697–A704/B732). Kant says that ‘reason cannot
think this systematic unity in any way other than by giving its idea an object’ but that ‘such an object
cannot be given through any experience’ and ‘is a mere idea, and is therefore not assumed absolutely and in
itself as something actual, but rather is taken as a ground only problematically’ (A681/B709; A648/B676;
A671/B699; A647/B675).
 KANT ’ S IDEALISM AND HIS REALISM
completeness science seeks, even though it is the attempt to reach this completeness
that drives science. In addition, Kant thinks that science gives us knowledge only of
relational features of reality and not of the intrinsic nature that grounds these
relations. He also thinks that the very strategy we have for proving the universality
of deterministic causality limits the application of the proof: we can prove it to be true
only of those things of which we can have empirical knowledge and have no
justification at all for asserting it of everything that exists. In my view, all these
ways in which science does not give us complete cognition of reality are helpful in
moving towards understanding Kant’s solution of the free will problem.
My thought, for further exploration elsewhere, is that what this picture enables
him to show is that we can take seriously the idea that events in space and time have
causes that fall under laws of nature without thinking that this entails that there is
only one possible future or that free action would involve breaking the laws. This
would mean that his account allows that it is true that every event of which we can
have empirical knowledge falls under a scientific law, but that it is not true that this
implies that nothing in the empirical world could have turned out differently to the
way it did. The idea is as follows. We think that deterministic causation threatens
freedom of the will when we are in the grip of a certain metaphysical picture: one on
which there are complete laws of physics which describe the most fundamental level
of reality and which, if we knew them, would tell us that there is only one possible
future consistent with the laws of nature and the initial situations of the most
fundamental particles (or strings or whatever the most fundamental level is). In my
view, part of the point of transcendental idealism is to undermine the grip of this
metaphysical picture. We think that this metaphysical picture follows from taking
science and laws of nature seriously because we wrongly think that science could, in
principle, give a complete account of the fundamental intrinsic nature of reality. The
picture makes us think that there are laws which would be broken by free actions,
because these actions would not be consistent with what the most fundamental laws
say has to follow from the initial arrangement of matter. But when something
happens that is not predicted by a law this entails that a law has been broken only
if the law is the only causal/explanatory factor in play: for example, laws of classical
mechanics predicting the interactions of billiard balls are not broken when someone
picks one of the billiard balls up, because the laws simply have nothing to say about
whether that will happen. The laws make predictions about what will happen on the
assumption that no other explanatory factors are in play, but it is not a part of the
laws of classical mechanics that they are the only thing that explains why anything
happens. On Kant’s account, saying that something falls under a law is not saying
that there is a complete explanation in principle accessible to science of why it had to
happen, given some previous state of the universe; (empirical) causal explanations, in
his view, can never be complete, sufficient explanations and it is never the case that
the laws are the only thing that explains what happens. There is no complete, final
science, there is no causal closure under physical law, and there is no fundamental
THE POSSIBILITY OF METAPHYSICS 
micro-physical level at which the ‘real’ causality is happening.8 Kant’s account is

therefore fully compatible with emergent causation and top-down or whole-part
causation,9 which is, arguably, required by his view in the Critique of Judgment that
biology can never be explained in terms of mechanistic laws of physics, at least partly
because it uses concepts which require explaining the parts of a creature in terms of
the whole, rather than treating the parts as explanatorily prior (CJ 5: 389, 400, 413).
Kant says:
if the effects are appearances, is it also necessary that the causality of their cause, which (namely
the cause) is also appearance must be solely empirical? Is it not rather possible that although
for every effect in appearance there is required a connection with its cause in accordance with
laws of empirical causality, this empirical causality itself, without the least interruption of its
connection with natural causes, could nevertheless be an effect of a causality that is not
empirical, but rather intelligible? (A544/B572; see also A536–7/B564–5; A538/B566)
Free actions would involve a noumenal ground, and therefore a causality we cannot
cognize (through theoretical reason), which grounds the events which happen in
space and time, and which fall under laws of nature. The law-governed interaction of
objects in space and time makes action possible: it makes it possible to manipulate
things in the world. But this does not mean that appealing to the law is the most
fundamental explanation of why the event happened, or that there was only ever one
possible future. Saying that something falls under a law also does not mean saying
that there is some complete final set of laws, which show how this event had to follow
from an initial position, or from what happened at the most fundamental level; the
extent of reality that science cognizes does not contain a most fundamental level or a
first moment. Clearly, this is only the beginning of a sketch in the direction of
understanding Kant’s account of the free will problem, but it does, I think, show
that a moderate metaphysical interpretation of transcendental idealism opens up the
possibility of understanding this in a way that is different from either merely
deflationary ‘two points of view’ solutions or merely the assertion of a mystery.
Keeping in mind Kant’s limited goals (he is not trying to prove that we have freedom,
but only to show that we do not need to take science to lead to metaphysical
conclusions that rule freedom out), this seems to me to be a promising avenue
for research.
My interpretation of transcendental idealism has been focused on giving an
account of the sense in which appearances are mind-dependent and of what the
claim that we cannot cognize the things in themselves on which appearances are
grounded amounts to, and it has been focused on the first Critique. There is much
8
It is not even clear that he can think that there is a determinate totality of laws, since he thinks that
systematic unity is not an actual feature of empirical reality.
9
For example, laws governing only the micro-level movements of particles will still be in force when a
particle is moved around because it is inside the body of a moving animal, but may not be the only thing
explaining where the particle ends up; laws at the level of biology may tell part of the story.
 KANT ’ S IDEALISM AND HIS REALISM
more to the Kantian position than this, and there are a number of ways in which
bringing it in relation to other works will require development of the position. In
addition to developing the account of free will that I have gestured towards, two other
important issues are Kant’s account of the self and his account of the concept of life
in his philosophy of biology, which has implications for how we think about our
knowledge of the empirical world and the claim that we cannot know non-relational
features of reality. There is much to be said about all these topics in relation to
transcendental idealism; I see my account as laying the groundwork for this, by
presenting a coherent account of the transcendental idealism presented in the first
Critique, on the basis of which these further topics could be developed.
Bibliography
Abela, Paul (2002). Kant’s Empirical Realism. Oxford: Oxford University Press.
Adams, Robert Merrihew (1994). Leibniz: Determinist, Theist, Idealist. New York and Oxford:
Oxford University Press.
Adams, Robert Merrihew (1997). ‘Things in Themselves’, Philosophy and Phenomenological
Research, 57(4): 801–25.
Adickes, Erich (1924). Kant und das Ding an Sich. Berlin: Pan Verlag.
Allais, Lucy (2004). ‘Kant’s One World’, The British Journal for the History of Philosophy,
12(4): 655–84.
Allais, Lucy (2006). ‘Intrinsic Natures: A Critique of Langton on Kant’, Philosophy and
Phenomenological Research, 73(1): 144–69.
Allais, Lucy (2007). ‘Kant’s Idealism and the Secondary Quality Analogy’, Journal of the
History of Philosophy, 45(3): 459–84.
Allais, Lucy (2009). ‘Kant, Non-conceptual Content, and the Representation of Space’, Journal
of the History of Philosophy, 47(3): 383–413.
Allais, Lucy (2010). ‘Transcendental Idealism and Metaphysics’, Kantian Yearbook, 2: 1–31.
Allison, Henry (1973). ‘Kant’s Critique of Berkeley’, Journal of the History of Philosophy,
11: 43–63.
Allison, Henry (1976). ‘The Non-Spatiality of Things in Themselves for Kant’, Journal of the
History of Philosophy, 14: 313–21.
Allison, Henry (1983). Kant’s Transcendental Idealism. New Haven and London: Yale
University Press.
Allison, Henry (1989). Review of Kant and the Claims of Knowledge by Paul Guyer, The
Journal of Philosophy, 86: 214–21.
Allison, Henry (1996). ‘Transcendental Idealism: A Retrospective’, in Henry Allison, Idealism
and Freedom: Essays on Kant’s Theoretical and Political Philosophy. Cambridge: Cambridge
University Press.
Allison, Henry (2004). Kant’s Transcendental Idealism, rev. and enlarged edn. New Haven and
London: Yale University Press.
Allison, Henry (2006). ‘Kant’s Transcendental Idealism’, in Graham Bird (ed.), A Companion
to Kant. Oxford: Blackwell.
Allison, Henry (2007). ‘Debating Allison on Transcendental Idealism’, Kantian Review, 12(2):
24–39.
Ameriks, Karl (1982a). ‘Recent Work on Kant’s Theoretical Philosophy’, American Philosoph-
ical Quarterly, 19(1): 1–24.
Ameriks, Karl (1982b). ‘Current German Epistemology: The Significance of Gerold Prauss’,
Inquiry, 25: 125–38.
Ameriks, Karl (1990). ‘Kant, Fichte, and Short Arguments to Idealism’, Archiv fur Geschichte
der Philosophie, 72: 63–85.
Ameriks, Karl (1992). ‘Kantian Idealism Today’, History of Philosophy Quarterly, 9(3): 329–42.
Ameriks, Karl (2000). Kant’s Theory of Mind, new edn. Oxford: Clarendon Press.
 BIBLIOGRAPHY
Ameriks, Karl (2003). Interpreting Kant’s Critiques. Oxford: Clarendon Press.

Ameriks, Karl (2006). Kant and the Historical Turn. Oxford: Clarendon Press.
Aquila, Richard (1974a). ‘Brentano, Descartes, and Hume on Awareness’, Philosophy and
Phenomenological Research, 35: 223–39.
Aquila, Richard (1974b). ‘Kant’s Theory of Concepts’, Kant-Studien, 65: 1–19.
Aquila, Richard (1983). Representational Mind: A Study of Kant’s Theory of Knowledge.
Bloomington: Indiana University Press.
Armstrong, David (1997). ‘Smart and the Secondary Qualities’, in Alex Byrne and David
Hilbert (eds.), Readings on Color Volume 1: The Philosophy of Color. Cambridge, Mass.: MIT
Press.
Ayers, Michael (1981). ‘Mechanism, Superaddition, and the Proof of God’s Existence in
Locke’s Essay’, The Philosophical Review, 90(2): 210–51.
Ayers, Michael (1982). ‘Berkeley’s Immaterialism and Kant’s Transcendental Idealism’, in
Godfrey Vesey (ed.), Idealism Past and Present. Cambridge: Cambridge University Press.
Baldner, Kent (1988). ‘Causality and Things in Themselves’, Synthese, 77: 353–73.
Baldner, Kent (1990). ‘Is Transcendental Idealism Coherent?’, Synthese, 85(1): 1–23.
Barker, S. F. (1969). ‘Appearing and Appearances in Kant’, in Lewis White Beck (ed.), Kant
Studies Today. Illinois: Open Court.
Bauer, Nathan (2012). ‘A Peculiar Intuition: Kant’s Conceptualist Account of Perception’,
Inquiry, 55(3): 215–37.
Baum, Manfred (1992). ‘Kant on Pure Intuition’, in Phillip Cummins and Guenter Zoeller
(eds.), Mind, Ideas and Objects. Atascadero, Calif.: Ridgeview Publishing.
Beck, Lewis White (1978). ‘Kant’s Strategy’, in Lewis White Beck, Essays on Kant and Hume.
New Haven and London: Yale University Press.
Beiser, Frederick (1992). ‘Kant’s Intellectual Development: 1746–1781’, in Paul Guyer (ed.),
The Cambridge Companion to Kant. Cambridge: Cambridge University Press.
Bell, David (1987). ‘The Art of Judgement’, Mind, 96: 221–44.
Bell, David (1999). ‘Transcendental Arguments and Non-Naturalistic Anti-Realism’, in
Robert Stern (ed.), Transcendental Arguments: Problems and Prospects. Oxford: Clarendon
Press.
Bennett, Jonathan (1966). Kant’s Analytic. Cambridge: Cambridge University Press.
Bennett, Jonathan (1971). Locke, Berkeley and Hume: Central Themes. Oxford: Clarendon
Press.
Bennett, Jonathan (1974). Kant’s Dialectic. Cambridge: Cambridge University Press.
Berkeley, George (1975a). An Essay towards a New Theory of Vision, in Michael Ayers (ed.),
Philosophical Works. London: J. M. Dent.
Berkeley, George (1975b). ‘A Treatise Concerning the Principles of Human Knowledge’, in
Michael Ayers (ed.), Philosophical Works. London: J. M. Dent.
Berkeley, George (1975c). Three Dialogues between Hylas and Philonous, in Michael Ayers
(ed.), Philosophical Works. London: J. M. Dent.
Bird, Graham (1962). Kant’s Theory of Knowledge. London: Routledge & Kegan Paul.
Bird, Graham (1996). ‘Critical Review: Recent Books on Kant’, Philosophical Quarterly,
46(183): 226–37.
Bird, Graham (2000). Review of Kantian Humility by Rae Langton, The Philosophical Quarterly,
50(198): 105–8.
BIBLIOGRAPHY 
Bird, Graham (2006). The Revolutionary Kant. Illinois: Open Court.

Blackburn, Simon (1989). ‘Manifesting Realism’, Midwest Studies in Philosophy, 14: 29–49.
Blackburn, Simon (1990). ‘Filling in Space’, Analysis, 50(2): 62–5.
Boghossian, Paul, and David Velleman (1997). ‘Color as a Secondary Quality’, in Alex Byrne
and David Hilbert (eds.), Readings on Color Volume 1: The Philosophy of Color. Cambridge,
Mass.: MIT Press.
Bonjour, Laurence (1998). In Defence of Pure Reason. Cambridge: Cambridge University Press.
Breitenbach, Angela (2004). ‘Langton on Things in Themselves: A Critique of Kantian
Humility’, Studies in History and Philosophy of Science, 35: 137–48.
Brittan, Gordon (1978). Kant’s Theory of Science. Princeton: Princeton University Press.
Brittan, Gordon (2006). ‘Kant’s Philosophy of Mathematics’, in Graham Bird (ed.), A Com-
panion to Kant. Oxford: Blackwell.
Broakes, Justin (1997). ‘The Autonomy of Color’, in Alex Byrne and David Hilbert (eds.),
Readings on Color Volume 1: The Philosophy of Color. Cambridge, Mass.: MIT Press.
Brook, Andrew (1997). Kant and the Mind. Cambridge: Cambridge University Press.
Brook, Andrew (1998). ‘Critical Notice of Kant’s Intuitionism: A Commentary on the Tran-
scendental Aesthetic by Lorne Falkenstein’, Canadian Journal of Philosophy, 28 (s1): 247–68.
Brouwer, Luitzen Egbertus Jan (1913). ‘Intuitionism and Formalism’, Bulletin of the American
Mathematical Society, 20(2): 81–96.
Buchdahl, Gerd (1969).‘The Kantian “Dynamic of Reason”, with Special Reference to the Place
of Causality in Kant’s System’, in Lewis White Beck (ed.), Kant Studies Today. Illinois: Open
Court.
Buchdahl, Gerd (1974).‘The Conception of Lawlikeness in Kant’s Philosophy of Science’, in
Lewis White Beck (ed.), Kant’s Theory of Knowledge. Dordrecht: Reidel.
Buchdahl, Gerd (1989). ‘Realism and Realization in a Kantian Light’, in Eva Schaper and
Wilhelm Vossenkul (eds.), Reading Kant. Oxford: Blackwell.
Buchdahl, Gerd (1992). Kant and the Dynamics of Reason. Essays on the Structure of Kant’s
Philosophy. Oxford: Blackwell.
Buroker, Jill (1981). Space and Incongruence: The Origin of Kant’s Idealism. Dordrecht:
D. Reidel Publishing Co.
Buroker, Jill (2006). Kant’s Critique of Pure Reason. Cambridge: Cambridge University Press.
Byrne, Alex, and David Hilbert (1997a). ‘Colors and Reflectances’, in Alex Byrne and David
Hilbert (eds.), Readings on Color Volume 1: The Philosophy of Color. Cambridge, Mass.: MIT
Press.
Byrne, Alex, and David Hilbert (1997b). Readings on Color, Vol. 2: The Science of Color.
Cambridge, Mass.: MIT Press.
Campbell, John (1994). Past, Space and Self. Cambridge, Mass.: MIT Press.
Campbell, John (1997). ‘A Simple View of Color’, in Alex Byrne and David Hilbert (eds.),
Readings on Color Volume 1: The Philosophy of Color. Cambridge, Mass.: MIT Press.
Campbell, John (2002a). Reference and Consciousness. Oxford: Clarendon Press.
Campbell, John (2002b). ‘Berkeley’s Puzzle’, in Tamar Szabo Gendler and John O’Leary
Hawthorne (eds.), Conceivability and Possibility. Oxford: Oxford University Press.
Carr, David (2000). ‘Review of Kantian Humility’, International Philosophical Quarterly, 40(1):
109–10.
Cassam, Quassim (1986). ‘Necessity and Externality’, Mind, 95: 446–64.
 BIBLIOGRAPHY
Cassam, Quassim (1987). ‘Transcendental Arguments, Transcendental Synthesis and Tran-

scendental Idealism’, Philosophical Quarterly, 37: 355–78.
Cassam, Quassim (1999). ‘Self-Directed Transcendental Arguments’, in Robert Stern (ed.),
Transcendental Arguments: Problems and Prospects. Oxford: Clarendon Press.
Cassam, Quassim (2000). ‘Rationalism, Empiricism, and the A Priori’, in Paul Boghossian and
Christopher Peacocke (eds.), New Essays on the A Priori. Oxford: Clarendon Press.
Chandrasekhar, B. S. (1998). Why Things Are the Way they Are. Cambridge: Cambridge
University Press.
Child, William (1994). Causality, Interpretation and the Mind. Oxford: Clarendon Press.
Clark, Michael, and David Liggins (2012). ‘Recent Work on Grounding’, Analysis, 72(4):
812–23.
Collins, Arthur (1999). Possible Experience. Berkeley and Los Angeles: University of California
Press.
Crane, Tim (2005). ‘The Problem of Perception’, in Edward N. Zalta (ed.), The Stanford
Encyclopedia of Philosophy (Spring 2005 edn.), http://plato.stanford.edu/archives/spr2005/
entries/perception-problem. Accessed January 2014.
Cummins, Robert (1968). ‘Kant on Outer and Inner Intuition’, Nous, 2: 271–92.
Cummins, Robert (1979). ‘Substance, Matter and Kant’s First Analogy’, Kant-Studien, 70:
149–61.
Dickie, Imogen (2011). ‘The Sortal Dependence of Demonstrative Reference’, European
Journal of Philosophy, 22: 34–60.
Dryer, Douglas Poole (1966). Kant’s Solution for Verification in Metaphysics. London: George
Allen & Unwin.
Dummett, Michael (1978a). ‘Wittgenstein’s Philosophy of Mind’, in Michael Dummett, Truth
and Other Enigmas. London: Duckworth.
Dummett, Michael (1978b). ‘Wittgenstein’s Philosophy of Mathematics’, in Michael Dum-
mett, Truth and Other Enigmas. London: Duckworth.
Dummett, Michael (1978c). ‘The Philosophical Basis of Intuitionist Logic’, in Michael Dummett,
Truth and Other Enigmas. London: Duckworth.
Dummett, Michael (1978d). ‘The Justification of Deduction’, in Michael Dummett, Truth and
Other Enigmas. London: Duckworth.
Dummett, Michael (1978e). ‘Introduction’, in Michael Dummett, Truth and Other Enigmas.
London: Duckworth.
Dummett, Michael (1991). The Logical Basis of Metaphysics. London: Duckworth.
Dummett, Michael (1993). ‘Realism and Anti-realism’, in Michael Dummett, The Seas of
Language. New York: Oxford University Press.
Eddington, Arthur (1946). The Nature of the Physical World. Cambridge: Cambridge University
Press.
Ellis, Brian, and Caroline Lierse (1994). ‘Dispositional Essentialism’, The Australasian Journal
of Philosophy, 72(1): 27–45.
Esfeld, Michael (2001). Review of Kantian Humility by Rae Langton, Erkenntnis, 54(3): 299–403.
Evans, Gareth (1982). The Varieties of Reference, ed. by John McDowell. New York and
Oxford: Oxford University Press.
Evans, Gareth (1985a). ‘Things Without the Mind’, in Gareth Evans, Collected Papers. Oxford:
Clarendon Press.
BIBLIOGRAPHY 
Evans, Gareth (1985b). ‘Molyneux’s Question’, in Gareth Evans, Collected Papers. Oxford:
Clarendon Press.
Ewing, Alfred (1934). Idealism: A Critical Survey. London: Methuen.
Falkenstein, Lorne (1995). Kant’s Intuitionism. Toronto: Toronto University Press.
Falkenstein, Lorne (2000). ‘Langton on Things in Themselves’, Kantian Review, 5: 49–64.
Falkenstein, Lorne (2006). ‘Kant’s Transcendental Aesthetic’, in Graham Bird (ed.), A Com-
panion to Kant. Oxford: Blackwell.
Forster, Eckart (1989). ‘How Transcendental Arguments are Possible’, in Eva Schaper and
Wilhelm Vossenkul (eds.), Reading Kant. Oxford: Blackwell.
Foster, John (1982). The Case for Idealism. London: Routledge & Kegan Paul.
Foster, John (2000). The Nature of Perception. Oxford: Oxford University Press.
Friedman, Michael (1992). ‘Causal Laws and the Foundations of Natural Science’, in Paul
Guyer (ed.), The Cambridge Companion to Kant. Cambridge: Cambridge University Press.
Friedman, Michael (2012). ‘Kant on Geometry and Spatial Intuition’, Synthese 186: 231–55.
Frith, Chris (2007). Making up the Mind: How the Brain Creates our Mental World. Oxford:
Blackwell.
Gardner, Sebastian (1999). Kant and the Critique of Pure Reason. London: Routledge.
Garve, Christian, and Johan Georg Feder (2004). ‘The Göttingen Review’, in Immanuel Kant,
Prolegomena to Any Future Metaphysics, ed. and trans. by Gary Hatfield. Cambridge:
Cambridge University Press.
Geach, Peter (1969). God and the Soul. London: Routledge & Kegan Paul.
Genone, James (2014). ‘Appearance and Illusion’, Mind, 123(490): 339–76.
George, Rolf (1981). ‘Kant’s Sensationism’, Synthese, 47: 229–55.
Ginsborg, Hannah (2006). ‘Empirical Concepts and the Content of Experience’, European
Journal of Philosophy, 14(3): 349–72.
Ginsborg, Hannah (2008). ‘Was Kant a Nonconceptualist?’, Philosophical Studies, 137(1):
65–77.
Golob, Sacha (2011). ‘Kant on Intentionality, Magnitude, and the Unity of Perception’,
European Journal of Philosophy, onlinelibrary.wiley.com/journal/10.1111/%28issn%
291468-0378/earlyview.
Goodman, Nelson (1954). Fact, Fiction and Forecast. London: Athlone Press.
Grier, Michelle (2001). Kant’s Doctrine of Transcendental Illusion. Cambridge: Cambridge
University Press.
Grier, Michelle (2009). ‘Kant’s Critique of Metaphysics’, in Edward Zalta (ed.), The Stanford
Encyclopedia of Philosophy (Summer 2009 edn.). http://plato.stanford.edu/archives/sum2009/
entries/kant-metaphysics/. Accessed 6 December 2011.
Griffith, Aaron (2010). ‘Perception and the Categories: A Conceptualist Reading of Kant’s
Critique of Pure Reason’, European Journal of Philosophy, 20(2): 193–222.
Grüne, Stefanie (2009). Blinde Anschauung. Frankfurt am Main: Vittorio Klostermann.
Guyer, Paul (1980). ‘Kant on Apperception and A Priori Synthesis’, American Philosophical
Quarterly, 17(3): 205–12.
Guyer, Paul (1987). Kant and the Claims of Knowledge. Cambridge: Cambridge University
Press.
Guyer, Paul (1992a). ‘Introduction: The Starry Heavens and the Moral Law’, in Paul Guyer
(ed.), The Cambridge Companion to Kant. Cambridge: Cambridge University Press.
 BIBLIOGRAPHY
Guyer, Paul (1992b). ‘The Transcendental Deduction of the Categories’, in Paul Guyer (ed.),
The Cambridge Companion to Kant. Cambridge: Cambridge University Press.
Guyer, Paul (2006). Kant. London and New York: Taylor & Francis.
Guyer, Paul (2007). ‘Debating Allison on Transcendental Idealism’, Kantian Review, 12(2):
10–24.
Guyer, Paul, and Ralph Walker (1990). ‘Kant’s Conception of Empirical Law’, Proceedings of
the Aristotelian Society, 64: 221–42.
Hacker, Peter (1989). Appearance and Reality. Oxford: Blackwell.
Hacking, Ian (1984). ‘Experimentation and Scientific Realism’, in Jarrett Leplin (ed.), Scientific
Realism. Berkeley and Los Angeles: University of California Press.
Hanna, Robert (2001). Kant and the Foundations of Analytic Philosophy. Oxford: Oxford
University Press.
Hanna, Robert (2005). ‘Kant and Nonconceptual Content’, European Journal of Philosophy,
13(2): 247–90.
Hanna, Robert (2006). Kant, Science and Human Nature. Oxford: Clarendon Press.
Hardin, C. L., ‘Reinverting the Spectrum’, in Alex Byrne and David Hilbert (eds.), Readings on
Color Volume 1: The Philosophy of Color, Cambridge, Mass.: MIT Press.
Harman, Gilbert (1997). ‘Explaining Objective Color in Terms of Subjective Reactions’, in Alex
Byrne and David Hilbert (eds.), Readings on Color Volume 1: The Philosophy of Color,
Cambridge, Mass.: MIT Press.
Harper, William (1986). ‘Kant on the A priori and Material Necessity’, in Robert Butts (ed.),
Kant’s Philosophy of Physical Science. Dordrecht: Reidel.
Harré, Rom, and E. H. Madden (1975). Causal Powers: A Theory of Natural Necessity. Oxford:
Blackwell.
Harrison, Ross (1982). ‘Transcendental Arguments and Idealism’, in Godfrey Vesey (ed.),
Idealism Past and Present. Cambridge: Cambridge University Press.
Haslanger, Sally (1989). ‘Persistence, Change and Explanation’, Philosophical Studies, 56: 1–28.
Hatfield, Gary (1990). The Natural and the Normative: Theories of Spatial Perception from
Kant to Helmholtz. Cambridge, Mass.: MIT Press.
Hatfield, Gary (2006). ‘Kant on the Perception of Space (and Time)’, in Paul Guyer (ed.), The
Cambridge Companion to Kant and Modern Philosophy. Cambridge: Cambridge University
Press.
Hawthorne, John (2001). ‘Intrinsic Properties and Natural Relations’, Philosophy and Phe-
nomenological Research, 63(2): 399–404.
Hintikka, Jaakko (1969). ‘On Kant’s Notion of Intuition’, in T. Penelhum and J. J. MacIntosh
(eds.), The First Critique: Reflections on Kant’s Critique of Pure Reason. California: Wads-
worth Publishing Company.
Hinton, J. M. (1973). Experiences. Oxford: Clarendon Press.
Hogan, Des (2009a). ‘Three Kinds of Rationalism and the Non-Spatiality of Things in
Themselves’, Journal of the History of Philosophy, 47(3): 355–82.
Hogan, Des (2009b). ‘How to Know Unknowable Things in Themselves’, Noûs, 43(1): 49–63.
Horstmann, Rolf Peter (1989). ‘Transcendental Idealism and the Representation of Space’, in
Eva Schaper and Wilhelm Vossenkul (eds.), Reading Kant. Oxford: Blackwell.
Howell, Robert (1973). ‘Intuition, Synthesis, and Individuation in the Critique of Pure Reason’,
Noûs, 7: 207–32.
BIBLIOGRAPHY 
Huenemann, Charles (1983). ‘A Note on the Argument for the Non-Spatiality of Things in
Themselves’, Kant-Studien, 84(3): 381–3.
Humberstone, I. L. (1996). ‘Intrinsic/Extrinsic’, Synthese, 108: 205–67.
Hume, David (1968). A Treatise of Human Nature. ed. by L. A. Selby-Bigge. Oxford: The
Clarendon Press.
Hume, David (1988). Enquiry Concerning Human Understanding, ed. by Anthony Flew.
Illinois: Open Court.
Humphrey, Ted B. (1973). ‘The Historical and Conceptual Relations between Kant’s
Metaphysics of Space and Philosophy of Geometry’, Journal of the History of Philosophy,
11: 483–512.
Jackson, Frank (1975). ‘On the Adverbial Analysis of Visual Experience’, Metaphilosophy,
6: 127–35.
Jackson, Frank (1998). From Metaphysics to Ethics: A Defence of Conceptual Analysis. Oxford:
Clarendon Press.
Jackson, Frank, and Robert Pargetter (1997). ‘An Objectivist’s Guide to Subjectivism about
Color’, in Alex Byrne and David Hilbert (eds.), Readings on Color Volume 1: The Philosophy
of Color. Cambridge, Mass.: MIT Press.
Jauernig, Anja (forthcoming). How to Think about Things in Themselves—An Essay on
Kant’s Metaphysics and Theory of Cognition. Oxford and New York: Oxford University
Press.
Johnston, Mark (1996). ‘Is the External World Invisible?’, Philosophical Issues, 7: 185–98.
Johnston, Mark (1997). ‘How to Speak of the Colors’, in Alex Byrne and David R. Hilbert
(eds.), Readings on Color Volume 1: The Philosophy of Color. Cambridge, Mass.: MIT Press.
Johnston, Mark (1998). ‘Are Manifest Qualities Response-Dependent?’, The Monist, 81: 3–43.
Johnston, Mark (2004). ‘The Obscure Object of Hallucination’, Philosophical Studies, 120(1–3):
113–83.
Justin, Gale (1974). ‘On Kant’s Analysis of Berkeley’, Kant-Studien, 65: 20–32.
Kant, Immanuel (1768). ‘Concerning the Ultimate Ground of the Differentiation of Directions
in Space’, in Immanuel Kant, Theoretical Philosophy, 1755–1770.
Kant, Immanuel (1992). Lectures on Logic, ed. and trans. by J. M. Young. Cambridge:
Cambridge University Press.
Kant, Immanuel (1992). A New Elucidation of the First Principles of Metaphysical Cognition
(Nova Dilucidatio), in Theoretical Philosophy, 1755–1770, ed. and trans. by D. Walford and
R. Meerbote. Cambridge: Cambridge University Press.
Kant, Immanuel (1992). On the Form and Principles of the Sensible World, Inaugural Disser-
tation, in Theoretical Philosophy, 1755–1770, ed. and trans. by D. Walford and R. Meerbote.
Cambridge: Cambridge University Press.
Kant, Immanuel (1992). Physical Monadology, in Theoretical Philosophy, 1755–1770, ed. and
trans. by D. Walford and R. Meerbote. Cambridge: Cambridge University Press.
Kant, Immanuel (1996). Critique of Practical Reason, in Practical Philosophy, trans. and ed. by
Mary Gregor. New York: Cambridge University Press.
Kant, Immanuel (1996). Critique of Pure Reason, trans. by Werner Pluhar. Indianapolis, Ind.:
Hackett Publishing Company, Inc.
Kant, Immanuel (1996). Groundwork of the Metaphysics of Morals, in Practical Philosophy,
trans. and ed. by Mary Gregor. New York: Cambridge University Press.
 BIBLIOGRAPHY
Kant, Immanuel (1997). Lectures on Metaphysics, trans. and ed. by Karl Ameriks and Steve
Naragon. New York: Cambridge University Press.
Kant, Immanuel (1998). Critique of Pure Reason, trans. and ed. by Paul Guyer and Allen
Wood. New York: Cambridge University Press.
Kant, Immanuel (1999). Philosophical Correspondence, ed. and trans. by Arnulf Zweig. New
York: Cambridge University Press.
Kant, Immanuel (2000). Critique of the Power of Judgment, ed. by Paul Guyer, trans. by Paul
Guyer and Eric Matthews. New York: Cambridge University Press.
Kant, Immanuel (2002). Metaphysical Foundations of Natural Science, in Theoretical Philoso-
phy after 1781, trans. and ed. by Henry Allison and Peter Heath. New York: Cambridge
University Press.
Kant, Immanuel (2002). ‘On a Discovery, according to which any new Critique of Pure Reason
is made Superfluous through an Older One’, in Theoretical Philosophy after 1781, trans. and
ed. by Henry Allison and Peter Heath. New York: Cambridge University Press.
Kant, Immanuel (2002). ‘What Real Progress has Metaphysics Made in Germany Since the
time of Leibniz and Wolff?’, in Theoretical Philosophy after 1781 , trans. and ed. by Henry
Allison and Peter Heath. New York: Cambridge University Press.
Kant, Immanuel (2004). A Prolegomena to any Future Metaphysics that will Present itself as a
Science, trans. and ed. by Gary Hatfield. Cambridge: Cambridge University Press.
Kant, Immanuel (2012). Thoughts on the True Estimation of Living Forces, in Eric Watkins
(ed.), Kant: Natural Science. Cambridge: Cambridge University Press.
Keller, Pierre (1998). Kant and the Demands of Self Consciousness. Cambridge: Cambridge
University Press.
Kitcher, Patricia (1990). Kant’s Transcendental Psychology. New York: Oxford University Press.
Kitcher, Patricia (2011). Kant’s Thinker. Oxford: Oxford University Press.
Kitcher, Philip (1974). ‘Kant and the Foundations of Mathematics’, Philosophical Review,
84: 23–58.
Kitcher, Philip (1986). ‘Projecting the Order of Nature’, in Robert Butts (ed.), Kant’s Philoso-
phy of Physical Science. Dordrecht: Reidel.
Kroon, Frederick (2001). ‘The Semantics of “Things in Themselves”: A Deflationary Account’,
The Philosophical Quarterly, 51(203): 165–81.
Ladyman, James (1998). ‘What is Structural Realism?’, Studies in History and Philosophy of
Science, 29: 409–24.
Ladyman, James, and Don Ross, with David Spurrett and Jon Collier (2007). Every Thing Must
Go: Metaphysics Naturalised. Oxford: Oxford University Press.
Land, Thomas (2011). ‘Kantian Conceptualism’, in Günther Abel and James Conant (eds.),
Rethinking Epistemology. Berlin: de Gruyter.
Land, Thomas (forthcoming). ‘The Nonconceptualist Reading of Kant and the Transcendental
Deduction’.
Langton, Rae (1998). Kantian Humility: Our Ignorance of Things in Themselves. Oxford:
Clarendon Press.
Langton, Rae (2006). ‘Kant’s Phenomena’, Philosophy and Phenomenological Research, 73(1):
170–85.
Langton, Rae, and David Lewis (2001). ‘Marshall and Parsons on “Intrinsic” ’, Philosophy and
BIBLIOGRAPHY 
Leplin, Jarrett (1984). ‘Introduction’, in Jarrett Leplin, Scientific Realism. Berkeley and Los
Angeles: University of California Press.
Lewis, David (2001). ‘Redefining “Intrinsic” ’, Philosophy and Phenomenological Research,
63: 381–98.
Locke, John (1961). An Essay Concerning Human Understanding, ed. by J. W. Yolton. London:
J. M. Dent.
Lockwood, Michael (1989). Mind, Brain and the Quantum. Oxford: Blackwell.
Long, David (1982). ‘Kant’s Pragmatic Horizon’, American Philosophical Quarterly, 19(4):
299–313.
Longuenesse, Beatrice (1998). Kant and the Capacity to Judge. Princeton: Princeton University
Press.
Longuenesse, Beatrice (2005). Kant and the Human Standpoint. Cambridge: Cambridge
University Press.
Longuenesse, Beatrice (2006). ‘Kant on A Priori Concepts: The Metaphysical Deduction of the
Categories’, in Paul Guyer (ed.), The Cambridge Companion to Kant and Modern Philoso-
phy. Cambridge: Cambridge University Press.
Lucas, J. (1984). Space, Time and Causality. Oxford: Clarendon Press.
Marshall, Colin (2013a). ‘Kant’s One Self and the Appearance/thing-in-itself Distinction’,
Kant-Studien, 104(4): 421–41.
Marshall, Colin (2013b). ‘Kant’s Appearances and Things in Themselves as Qua Objects’,
Philosophical Quarterly, 63(252): 520–45.
Marshall, Dan, and Josh Parsons (2001). ‘Langton and Lewis on “Intrinsic” ’, Philosophy and
Phenomenological Research, 63: 347–51.
Martin, C. B. (1994). ‘Dispositions and Conditionals’, The Philosophical Quarterly, 44(174): 1–8.
Matthews, H. E. (1982). ‘Strawson and Transcendental Idealism’, in R. C. S. Walker (ed.), Kant
on Pure Reason. Oxford: Oxford University Press.
Maund, Barry (2002). ‘Color’, in Edward Zalta (ed.), The Stanford Encyclopedia of Philosophy
(Fall 2002 edn.). http://plato.stanford.edu/archives/fall2002/entries/color/.
McDowell, John (1994). Mind and World. Cambridge: Cambridge University Press.
McDowell, John (1998a). ‘Having the World in View: Sellars, Kant, and Intentionality’, The
Journal of Philosophy, 95(9): 431–92.
McDowell, John (1998b). ‘Singular Thought and the Boundaries of Inner Space’, in John
McDowell, Meaning, Knowledge and Reality. Cambridge, Mass., and London: Harvard
University Press.
McGinn, Colin (1980). ‘Truth and Use’, in Mark Platts (ed.), Reference, Truth and Reality.
London: Routledge & Kegan Paul.
McGinn, Colin (1983). The Subjective View. Oxford: Clarendon Press.
McGinn, Colin (1996). ‘Another Look at Colour’, The Journal of Philosophy, 93: 537–53.
McLear, Colin (2011). ‘Kant on Animal Consciousness’, Philosophers’ Imprint, 11(15): 1–16.
McLear, Colin (2013). Essays on Kant on Perception and Cognition, dissertation, Cornell.
McLear, Colin (2015). ‘Two Kinds of Unity in the Critique of Pure Reason’, Journal of the
History of Philosophy, 53(1): 79–110.
McLear, Colin (forthcoming). ‘Kant on Perceptual Content’, Mind.
Melnick, Arthur (1973). Kant’s Analogies of Experience. Chicago and London: University of
Chicago Press.
 BIBLIOGRAPHY
Mill, John Stuart (1979). An Examination of Sir William Hamilton’s Philosophy, in John
Robson (ed.), Collected Works of John Stuart Mill, Vol. 9. Toronto: University of Toronto
Press.
Miller, George (1973). ‘Kant and Berkeley: The Alternative Theories’, Kant-Studien, 64:
315–35.
Milmed, Bella (1969). ‘Possible Experience and Recent Interpretations of Kant’, in Lewis White
Beck (ed.), Kant Studies Today. Illinois: Open Court.
Moore, Adrian (1992). ‘A Note on Kant’s First Analogy’, Philosophical Quarterly, 42(169):
480–5.
Moore, Adrian (2001). Review of Rae Langton, Kantian Humility, Philosophical Review, 110:
117–20.
Moran, Dermot (2000). ‘Hilary Putnam and Immanuel Kant: Two Internal Realists’, Synthese,
123(1): 65–104.
Nagel, Gordon (1983). The Structure of Experience. Chicago: University of Chicago Press.
Naragon, Steve (1990). ‘Kant on Descartes and the Brutes’, Kant-Studien, 81(1): 1–23.
Papineau, David (ed). (1996a). The Philosophy of Science. Oxford: Oxford University Press.
Papineau, David (1996b). ‘Introduction’, in David Papineau (ed.), The Philosophy of Science.
Parsons, Charles (1992). ‘The Transcendental Aesthetic’, in Paul Guyer (ed.), The Cambridge
Companion to Kant. Cambridge: Cambridge University Press.
Paton, H. J. (1936). Kant’s Metaphysics of Experience. London: George Allen & Unwin.
Peacocke, Christopher (1993). ‘Intuitive Mechanics, Psychological Reality and the Idea of a
Material Object’, in N. Eilan, R. McCarthy, and B. Brewer (eds.), Spatial Representation.
Oxford: Blackwell.
Pendlebury, Michael (1995). ‘Making Sense of Kant’s Schematism’, Philosophy and Phenom-
enological Research, 55(4): 777–97.
Pendlebury, Michael (1996). ‘The Role of Imagination in Perception’, South African Journal of
Philosophy, 15(4): 133–8.
Pendlebury, Michael (1998). ‘Intentionality and Normativity’, South African Journal of Phil-
osophy, 17(2): 142–51.
Pendlebury, Michael (1999). ‘Sensibility and Understanding in Perceptual Judgements’, South
African Journal of Philosophy, 18(4): 356–9.
Pereboom, Derk (1988). ‘Kant on Intentionality’, Synthese, 77: 321–52.
Pettit, Philip (1998). ‘Noumenalism and Response-Dependence’, The Monist, 81: 112–32.
Pippin, R. B. (1974). Review of Erscheinung by Kant by G. Prauss, Journal of the History of
Philosophy, 12: 403–5.
Pippin, R. B. (1976). Review of Kant und das Problem der Dinge an sich by G. Prauss, Journal of
the History of Philosophy, 14: 374–8.
Pippin, R. B. (1982). Kant’s Theory of Form. New Haven and London: Yale University Press.
Pogge, Thomas (1991). ‘Erscheinungen und Dinge an sich’, Zeitschrift Für Philosophische
Forschung, 45: 489–510.
Posy, Karl (1983). ‘Dancing to the Antinomy: A Proposal for Transcendental Idealism’,
American Philosophical Quarterly, 20: 81–94.
Posy, Karl (1984). ‘Kant’s Mathematical Realism’, The Monist, 67(1): 115–34.
Prauss, G. (1971). Erscheinung bei Kant. Berlin: Walter de Grunter.
BIBLIOGRAPHY 
Prauss, G. (1974). Kant und das Problem der Dinge an sich. Bonn: Bouvier Verlag Herbert
Grundmann.
Prauss, G. (1981). ‘The Language of Appearances and Things in Themselves’, Synthese, 47: 313–52.
Prior, E. W., R. Pargetter, and Frank Jackson (1982). ‘Three Theses about Dispositions’,
American Philosophical Quarterly, 19: 251–7.
Pritchard, H. A. (1909). Kant’s Theory of Knowledge. Oxford: Clarendon Press.
Putnam, Hilary (1981). Reason, Truth and History. Cambridge: Cambridge University Press.
Putnam, Hilary (1983). Realism and Reason. Cambridge: Cambridge University Press.
Putnam, Hilary (1987). The Many Faces of Realism. La Salle, Ill.: Open Court.
Putnam, Hilary (1994). ‘Sense, Nonsense, and the Senses: An Inquiry into the Powers of the
Human Mind’, The Journal of Philosophy, 91(9): 445–517.
Robinson, Hoke (1994). ‘Two Perspectives on Kant’s Appearances and Things in Themselves’,
Journal of the History of Philosophy, 32(3): 411–41.
Robinson, Howard (1982). Matter and Sense. Cambridge: Cambridge University Press.
Robinson, Howard (1994). Perception. London: Routledge.
Rogerson, Kenneth (1993). ‘Kantian Ontology’, Kant-Studien, 84: 14–15.
Rosefeldt, Tobias (2001). Review of Kantian Humility by Rae Langton, European Journal of
Rosefeldt, Tobias (2007). ‘Dinge an Sich und Sekundäre Qualitäten’, in Jürgen Stolzenburg
(ed.), Kant in der Gegenwart. Berlin: Walter de Gruyter.
Rosen, G. (2010). ‘Metaphysical Dependence: Grounding and Reduction’, in B. Hale and
A. Hoffmann (eds.), Modality: Metaphysics, Logic, and Epistemology. Oxford: Oxford
University Press.
Rosenburg, Jay F. (1980). One World and Our Knowledge of it: The Problematic of Realism in
Post-Kantian Perspective. Dordrecht: Reidel.
Russell, Bertrand (1967). The Problems of Philosophy, repr. Oxford: Oxford University Press.
Sacks, Mark (2000). ‘Transcendental Arguments and the Inference to Reality: A Reply to
Stern’, in Brigitte Sassen (ed.), Kant’s Early Critics. Cambridge: Cambridge University Press.
Sassen, Brigitte (2000). Kant’s Early Critics: The Empiricist Critique of the Theoretical Philoso-
phy. Cambridge: Cambridge University Press.
Schaffer, Jonathan (2009). ‘On What Grounds What’, in David Manley, David Chalmers, and
Ryan Wasserman (eds.), Metametaphysics: New Essays on the Foundations of Ontology. New
York: Oxford University Press.
Schafer, Karl (forthcoming). ‘Kant’s Conception of Cognition and our Knowledge of Things-
in-Themselves’.
Schaper, Eva, and Wilhelm Vossenkuhl (1989). Reading Kant: New Perspectives on Transcen-
dental Arguments and Critical Philosophy. Oxford: Blackwell.
Schrader, George (1968). ‘The Thing in Itself in Kantian Philosophy’, in Robert Paul Wolff
(ed.), Kant: A Collection of Critical Essays. London: Macmillan.
Scott-Taggart, M. J. (1969). ‘Recent Work on the Philosophy of Kant’, in Lewis White Beck
(ed.), Kant Studies Today. La Salle, Ill.: Open Court.
Sellars, Wilfrid (1962). ‘Philosophy and the Scientific Image of Man’, in R. Colodny (ed.),
Frontiers of Science and Philosophy. Pittsburgh: University of Pittsburgh Press.
Sellars, Wilfrid (1963). ‘Empiricism and the Philosophy of Mind’, in Wilfrid Sellars, Science,
Perception and Reality. London: Routledge & Kegan Paul; New York: The Humanities Press.
 BIBLIOGRAPHY
Sellars, Wilfrid (1967). ‘Some Remarks about Kant’s Theory of Experience’, The Journal of
Philosophy, 64(20): 633–47.
Sellars, Wilfrid (1968). Science and Metaphysics: Variations on Kantian Themes. London:
Routledge & Kegan Paul.
Sellars, Wilfrid (1969). ‘Kant’s Views on Sensibility and Understanding’, in Lewis White Beck
(ed.), Kant Studies Today. La Salle, Ill.: Open Court.
Senderowicz, Yaron (2005). The Coherence of Kant’s Transcendental Idealism. Dordrecht:
Springer.
Setiya, Kieran (2004). ‘Transcendental Idealism in the Aesthetic’, Philosophy and Phenomeno-
logical Research, 68(1): 63–88.
Shabel, Lisa (2006). ‘Kant’s Philosophy of Mathematics’, in Paul Guyer (ed.), The Cambridge
Companion to Kant and Modern Philosophy. Cambridge: Cambridge University Press.
Shoemaker, Sydney (1997). ‘Phenomenal Character’, in Alex Byrne and David R. Hilbert
(eds.), Readings on Color Volume 1: The Philosophy of Color. Cambridge, Mass.: MIT
Press.
Shoemaker, Sydney (2003a). ‘Causality and Properties’, in Sydney Shoemaker, Identity, Cause
and Mind, rev. edn. Oxford: Clarendon Press.
Shoemaker, Sydney (2003b). ‘Identity, Properties, and Causality’, in Sydney Shoemaker,
Identity, Cause and Mind, rev. edn. Oxford: Clarendon Press.
Sider, Ted (2001). ‘Maximality and Intrinsic Properties’, Philosophy and Phenomenological
Research, 63(2): 357–64.
Smart, J. J. C. (1963). Philosophy and Scientific Realism. London: Routledge & Kegan Paul.
Smit, Houston (2000). ‘Kant on Marks and the Immediacy of Intuition’, The Philosophical
Review, 109: 235–66.
Smith, A. D. (1977). ‘Dispositional Properties’, Mind, 86: 439–45.
Smith, A. D. (1990). ‘Of Primary and Secondary Qualities’, The Philosophical Review, 99(2):
221–54.
Smith, A. D. (2002). The Problem of Perception. Cambridge, Mass.: Harvard University Press.
Smith, Michael, and Daniel Stoljar (1998). ‘Global Response-Dependence and Noumenal
Realism’, The Monist, 81: 85–111.
Snowdon, Paul (1992). ‘How to Interpret “Direct Perception” ’, in Tim Crane (ed.), The
Contents of Experience. Cambridge: Cambridge University Press.
Snowdon, Paul (2005). ‘The Formulation of Disjunctivism: A Response to Fish’, Proceedings of
the Aristotelean Society, 105(1): 129–41.
Stang, Nick (2014). ‘The Non-Identity of Appearances and Things in Themselves’, Noûs, 48(1):
106–36.
Stang, Nick (forthcoming). ‘Adickes on Double Affection’, Proceedings of the Eleventh Inter-
national Kant Congress.
Stang, Nick (forthcoming). ‘Who’s Afraid of Double Affection?’, Philosophers’ Imprint.
Stephenson, Andrew. ‘Kant on the Object Dependence of Intuition and Halucination’,
MS. The Philosophical Quarterly Advance Access published February 19, 2015. http://pq.
oxfordjournals.org/content/early/2015/02/18/pq.pqu100.full.pdf?keytype=ref&ijkey=
93Xzm1jrP7zXbzJ.
Stern, Robert (ed.) (1999a). Transcendental Arguments: Problems and Prospects. Oxford:
Clarendon Press.
BIBLIOGRAPHY 
Stern, Robert (1999b). ‘On Kant’s Response to Hume: The Second Analogy as Transcendental
Argument’, in R. Stern (ed.), Transcendental Arguments: Problems and Prospects. Oxford:
Clarendon Press.
Stevenson, Leslie (1983). ‘Empirical Realism and Transcendental Anti-Realism’, Proceedings of
Strawson, Galen (1987). ‘Realism and Causation’, Philosophical Quarterly, 37: 253–77.
Strawson, Galen (1996). The Secret Connexion. Oxford: Clarendon Press.
Strawson, Galen (2002). ‘Knowledge of the World’, Noûs, 36(s1): 146–75.
Strawson, Galen (2003). ‘Real Materialism’, in L. Antony and N. Hornstein (eds.), Chomsky
and his Critics. Oxford: Blackwell.
Strawson, P. F. (1963). Individuals. New York: Doubleday.
Strawson, P. F. (1966). The Bounds of Sense. London: Methuen.
Strawson, P. F. (1976–7). ‘Scruton and Wright on Anti-Realism’, Proceedings of the Aristotelian
Society, 77: 15–21.
Strawson, P. F. (1980). ‘Reply to Evans’, in Z. Van Straaten (ed.), Philosophical Subjects.
Oxford: Clarendon Press.
Strawson, P. F. (1981). ‘Perception and its Objects’, in G. MacDonald (ed.), Perception and
Identity. London: Macmillan.
Strawson, P. F. (1989). ‘Sensibility, Understanding and Synthesis’, in E. Foster (ed.), Kant’s
Transcendental Deductions. Stanford, Calif.: Stanford University Press.
Strawson, P. F. (1994). ‘The Problem of Realism and the A Priori’, in Paolo Parrini (ed.), Kant
and Contemporary Epistemology. Dordrecht: Kluwer Academic Publishers.
Strawson, P. F. (1997). ‘Kant’s New Foundations of Metaphysics’, in Strawson, Entity and
Identity and Other Essays. Oxford: Clarendon Press.
Stroud, Barry (1982). ‘Transcendental Arguments’, in R. C. S. Walker (ed.), Kant on Pure
Reason. Oxford: Oxford University Press.
Stroud, Barry (1999). The Quest for Reality: Subjectivism & the Metaphysics of Colour. Oxford:
Oxford University Press.
Stroud, Barry (2000a). ‘The Goal of Transcendental Arguments’, in Stroud, Understanding
Human Knowledge. Oxford: Oxford University Press.
Stroud, Barry (2000b). ‘The Synthetic A Priori in Strawson’s Kantianism’, in Stroud, Under-
standing Human Knowledge. Oxford: Oxford University Press.
Stroud, Barry (2000c). ‘The Allure of Idealism’, in Stroud, Understanding Human Knowledge.
Thompson, Manley (1973). ‘Singular Terms and Intuition in Kant’s Epistemology’, Review of
Metaphysics, 26: 314–43.
Tolley, Clinton (2013). ‘The Non-conceptuality of the Content of Intuitions: A New
Approach’, Kantian Review, 18(1): 107–36.
Tolley, Clinton (2014). ‘Kant on the Content of Cognition’, European Journal of Philosophy,
22: 200–28.
Tolley, Clinton (forthcoming). ‘Kant on the Distinction between Perception and Experience’.
Treisman, A. (2003). ‘Consciousness and Perceptual Binding’, in Axel Cleeremans (ed.), The
Unity of Consciousness: Binding, Integration and Dissociation. Oxford: Oxford University Press.
Turbayne, C. (1955). ‘Kant’s Refutation of Dogmatic Idealism’, Philosophical Quarterly,
5: 225–44.
 BIBLIOGRAPHY
Valberg, Jerome, J. (1992). The Puzzle of Experience. Oxford: Oxford University Press.
Van Cleve, James (1979). ‘Substance, Matter and Kant’s First Analogy’, Kant-Studien,
70: 149–61.
Van Cleve, James (1992). ‘The Argument from Geometry, Transcendental Idealism and Kant’s
Two Worlds’, in P. Cummins and G. Zoeller (eds.), Minds, Ideas and Objects: Essays on the
Theory of Representation in Modern Philosophy, North American Kant Society Studies in
Philosophy 2. Atascadero: Calif.: Ridgeview.
Van Cleve, James (1995). ‘Putman, Kant, and Secondary Qualities’, Philosophical Papers, 24(2):
83–109.
Van Cleve, James (1999). Problems from Kant. New York and Oxford: Oxford University Press.
Van Cleve, James (2002). ‘Receptivity and Our Knowledge of Intrinsic Properties’, Philosophy
and Phenomenological Research, 65: 218–36.
van Fraassen, Bas (1984). ‘To Save the Phenomena’, in Jarrett Leplin (ed.), Scientific Realism.
Berkeley and Los Angeles: University of California Press.
Walker, R. C. S. (1971–2). ‘The Status of Kant’s Theory of Matter’, Synthese, 23: 121–6.
Walker, R. C. S. (1978). Kant. London: Routledge & Kegan Paul.
Walker, R. C. S. (1983). ‘Empirical Realism and Transcendental Anti-Realism’, Proceedings of
Walker, R. C. S. (1985). ‘Synthesis and Transcendental Idealism’, Kant-Studien, 76(1): 14–27.
Walker, R. C. S. (1989). The Coherence Theory of Truth. London and New York: Routledge.
Walker, R. C. S. (1990). ‘Kant’s Conception of Empirical Law’, Proceedings of the Aristotelian
Society, supp. 64: 242–58.
Walker, R. C. S. (1995). ‘Verificationism, Anti-Realism, and Idealism’, European Journal of
Walker, R. C. S. (1999). ‘Induction and Transcendental Arguments’, in Robert Stern (ed.),
Transcendental Arguments: Problems and Prospects. Oxford: Clarendon Press.
Walker, R. C. S. (2010). ‘Kant on the Number of Worlds’, British Journal for the History of
Philosophy, 18(5): 821–43.
Walker, R. C. S., and Paul Guyer (1989). ‘Kant and the Claims of Reason’, Philosophical Quarterly,
39(156): 373. Ward, Andrew (2006). Kant: The Three Critiques. Cambridge: Polity Press.
Warren, Daniel (1998). ‘Kant and the A Priority of Space’, Philosophical Review, 107: 179–224.
Warren, Daniel (2001a). ‘Kant’s Dynamics’, in Eric Watkins (ed.), Kant and the Sciences.
Warren, Daniel (2001b). Reality and Impenetrability in Kant’s Philosophy of Nature. New
York: Routledge.
Watkins, Eric (2002). ‘Kant’s Transcendental Idealism and the Categories’, History of Philoso-
phy Quarterly, 19(2): 191–215.
Watkins, Eric (2005). Kant and the Metaphysics of Causality. Cambridge: Cambridge Univer-
sity Press.
Weatherson, B. (2001). ‘Intrinsic Properties and Combinatorial Principles’, Philosophy and
Westphal, Kenneth (1997a). ‘Noumenal Causality Reconsidered: Affection, Agency and Mean-
ing in Kant’, Canadian Journal of Philosophy, 27: 209–45.
Westphal, Kenneth (1997b). ‘Affinity, Idealism, and Naturalism: The Stability of Cinnabar and
the Possibility of Experience’, Kant-Studien, 88: 139–89.
BIBLIOGRAPHY 
Westphal, Kenneth (1998). ‘Buchdahl’s “Phenomenological” View of Kant: A Critique’, Kant-

Studien, 89: 335–52.
Westphal, Kenneth (2001). ‘Freedom and the Distinction between Phenomena and Noumena:
Is Allison’s View Methodological, Metaphysical, or Equivocal?’, Journal of Philosophical
Research, 26: 593–621.
Westphal, Kenneth (2004). Kant’s Transcendental Proof of Realism. Cambridge: Cambridge
University Press.
Wilkerson, T. E. (1976). Kant’s ‘Critique of Pure Reason’: A Commentary for Students. Oxford:
Clarendon Press.
Willaschek, Marcus (1997). ‘Der transzendentale Idealismus und die Idealität von Raum und
Zeit’, Zeitschrift für philosophische Forschung, 51: 537–64.
Willaschek, Marcus (2001). ‘Affektion und Kontingenz in Kants transzendentalem Idealis-
mus’, in R. Schumacher (hrsg.), Idealismus als Theorie der Repräsentation. Paderborn.
Willaschek, Marcus (2008). ‘Kant on the Necessity of Metaphysics’, Akten des X. Internatio-
nalen Kant-Kongresses. Berlin and New York: Walter de Gruyter.
Wilson, Kirk (1975). ‘Kant on Intuition’, The Philosophical Quarterly, 25: 247–65.
Wilson, Margaret (1974). ‘On Kant and the Refutation of Subjectivism’, in Lewis White Beck
(ed.), Kant’s Theory of Knowledge. Dordrecht: Reidel.
Wilson, Margaret (1979). ‘Superadded Properties: The Limits of Mechanism in Locke’, Ameri-
can Philosophical Quarterly, 16(2): 143–50.
Wilson, Margaret (1982). ‘Superadded Properties: A Reply to M. R. Ayers’, The Philosophical
Review, 91: 247–53.
Wilson, Margaret (1984). ‘The “Phenomenalisms” of Berkeley and Kant’, in Allen Wood (ed.),
Self and Nature in Kant’s Philosophy. New York: Cornell University Press.
Wilson, Margaret (1992). ‘History of Philosophy Today and the Case of Sensible Qualities’,
The Philosophical Review, 101(1): 191–243.
Wood, Allen (2007). ‘Debating Allison on Transcendental Idealism’, Kantian Review,
12(2): 1–10.
Worrall, J. (1996). ‘Structural Realism: The Best of Both Worlds?’, in David Papineau (ed.), The
Philosophy of Science. Oxford: Oxford University Press.
Wright, Crispin (1980). Wittgenstein on the Foundations of Mathematics. London: Duckworth.
Wright, Crispin (1992). Truth and Objectivity. Cambridge, Mass.: Harvard University Press.
Wright, Crispin (1996). Realism, Meaning and Truth. Cambridge: Polity.
Yolton, J. W. (1984). Perceptual Acquaintance from Descartes to Reid. Oxford: Blackwell.
Yolton, J. W. (1996). Perception and Reality: A History from Descartes to Kant. New York:
Cornell University Press.
Yolton, J. W. (2000). Realism and Appearances. Cambridge: Cambridge University Press.
Young, Michael (1994). ‘Synthesis and the Content of Pure Concepts in Kant’s First Critique’,
Journal of the History of Philosophy, 32(3): 331–57.
Index
Abela, Paul, 54n, 215n Campbell, John, 105n, 110, 113, 115
acquaintance, 14, 34, 147–8, 155–9, 163, 168, categories, 18, 68–9, 148, 150, 155, 162–3,
177, 193, 200, 214, 217, 236n, 246–7, 168–9, 173–4, 233n, 259–302
270–2, 279, 284, 288–9, 291, 297–8, 303; a priori concept of, 148
see also intuition Cassam, Quassim, 287
actuality, 139–40 cause, 48–9, 68–9, 209–11, 219–20, 226, 227n,
Adams, Robert Merrihew, 42 243–54, 283–4, 301–2, 304–7; see also
Allison, Henry, 7, 22n, 77–88, 95–6, 187n Analogies of Experience; categories
Ameriks, Karl, 34, 95n, 149n, 291, 301 change, 48, 156, 223, 278–9, 281, 284
Amphiboly, 62, 68, 73, 96, 224, 228–9, mere Cambridge, 278
232–243, 246–7 coexistence, 246
Analogies of Experience, 48–50, 139–40, 246, 304 cognition
analytic, 33–4, 44, 70, 86, 178n, 180–1, 235, as discursive (requires both intuition and
238–40, 301 concepts), 13–15, 86–7, 145–75, 193, 199,
anti-realism, 256, 305 269–70; see also acquaintance; concept;
in science, 217–21, 258 epistemic condition; intuition;
meaning theoretic, 211–19, 258 transcendent metaphysics
Antinomies, 57, 90–7, 176n, 203–4, 213, empirical, 261–2, 290–1, 298–9; see also
215–17, 241–2, 301, 303–8; see also experience; science
conditioned; free will; transcendent practical, 304n
metaphysics theoretical, 61, 303–4
appearances cognitive processing, 108–10
as essentially manifest, 13, 129–44; see also Collins, Arthur, 17n, 45–6, 125n
non-phenomenalist idealism colour, 73–4, 116–27, 130–9, 144, 209–11,
as grounded by things in themselves, 32–3, 214, 221
65–70, 74–6, 95–7, 129–32, 220–6, 229–33, as accessible via only one sense, 121
236–43, 245, 248–58; see also ground as essentially manifest, 103–4, 116–24, 130–9,
(grounding) 144, 210–11, 214, 221; see also qualities,
as identical to things in themselves, 27–30, 60, primary and secondary; secondary quality
70–5, 242 analogy
as in us, 21–3, 133–4 idealist dispositional account of, 118–21,
as objects, 135 126–7, 210
as mere representations, 23–5 primitivist objectivist account of, 118, 120–1,
as relational, 69–70, 221–43 124, 126, 210, 214
as requiring a connection with actual scientific dispositionalist account of colour,
perception for their existence, 25–7 119–20
interpreted neutrally, 20–1, 25 scientific objectivist account of, 118
see also things in themselves community, 246; see also categories
Axioms of Intuition, 172, 191–2 concept, 145–75; see also intuition; judgment
a priori, 257–89, 292–302
Berkeley, George, 12, 37, 44–8, 50–7, 102, 126n, as distinct from intuition, 145–75,
129, 134, 202–3; see also Refutation of 292–7
Idealism as empty, 152–3, 212–13, 268–70
Bird, Graham, 35, 86, 273 as general, 146–7, 199, 264–84
binding, 169n 170, 272, 276–9 as mediate, 146–7
Blackburn, Simon, 223, 249 as spontaneous, 275–84, 287, 290–2
Breitenbach, Angela, 85 as thought about objects, 146–7, 154–6,
Brittan, Gordon, 193n 263, 265
Boghossian, Paul, 120n empirical application of, 261–3, 270–1,
Bonjour, Laurence, 180n, 181 275–7, 280–4, 298–300
 INDEX
concept, (cont.) Golob, Sacha, 171n

problematic, 64–5, 67–8 Grier, Michelle, 90n
role in cognition, 145–75; see also cognition, ground (grounding), 90, 119–22, 130–2, 222–6,
as discursive 229–31, 233, 235, 242, 248–58, 307; see also
conceptualism, 148–75, 263, 268, 284; see also appearances, as grounded by things in
non-conceptualism themselves
argument for, 151–3, 165 Grüne, Stefanie, 173
moderate, 149–50, 153–4, 161–2, 164, 173–4, Guyer, Paul, 38–9, 50, 301
263n
strong, 149–50, 153–4, 159, 160n, 161–2 hallucination, 12, 41–2, 107–8, 110–12, 156
conditioned, 90–3 Hanna, Robert, 70
conservation, 185; see also substance Hume, David, 48, 250, 283n
Copernican Revolution, 5–6, 178, 293
idealism
Descartes, René, 54–6, 102, 134; see also extreme, see phenomenalism
perception, Cartesian account of; external non-phenomenalist, 13, 22–7, 37–58, 114–24,
world skepticism; transcendent 132–6, 201–4, 209–17, 296; see also
metaphysics appearances, as essentially manifest;
disposition, 118–20, 123, 223, 249–50; see also Refutation of Idealism
forces (powers) identity, 60, 71–5, 119–20, 130, 132, 154, 209,
direct realism, 39–41, 102–12, 116, 135–7; 242, 280–1; see also appearances, as
see also relational account of perception identical to things in themselves
Doctrine of Method, 191–2 and difference, 235–6
Dryer, Douglas Poole, 16n of indiscernibles, 73, 108, 236, 246,
Dummett, Michael, 216n impenetrability, see solidity
incongruent counterparts, 163–4, 198
empiricism, 26, 87–8, 140, 256, 261–2, 272, 275, indirect realism, 39, 43, 104–5, 112–13; see also
296, 300; see also rationalism; transcendent perception, Cartesian account of
metaphysics inner and outer, 222–5, 234–41, 251; see also
epistemic condition, 79–84, 86–7, 94–5; see also Amphiboly
transcendental idealism, deflationary sense, 55–7, 150, 228
interpretation of intelligibilia, 7, 10, 35, 59–76, 93, 97, 296;
ethics, 3, 68, 152, 153n, 213, 305 see also noumenalism, transcendent
Evans, Gareth, 264 metaphysics
existence, 69; see also categories intuition, 145–75, 194–201
experience, 262–3 a priori, 163–8, 186–201, 292–8
possible (possibility of), 47, 137–44, 214–17, as blind, 152–3
219–20, 261, 299; see also realism, as distinct from concepts, 145–75, 292–7
empirical; possibility as distinct from sensation, 14, 109, 146–7,
external world scepticism, 3, 35, 55, 159, 149, 154, 160–1, 172
182, 262 as giving (presenting) objects, 146–7, 153–5,
158–60, 162, 167, 170, 266, 274, 276
Falkenstein, Lorne, 149n, 156n, 157n as guaranteeing the existence of its objects,
forces (powers), 34, 118–20, 220, 222–31, 234, 154, 157–8, 167; see also external world
237, 243–55, 257, 281 skepticism
magnetic, 47–8 as immediate, 146–75, 187, 197, 274, 276,
Foster, John, 249–51 297, 303
free will, 3, 5–6, 19, 95–6, 176n, 213, 303–8; as singular, 146–75, 187, 274, 276, 297
see also Antinomies; transcendent intellectual, 61, 63, 81, 84, 88
metaphysics manifold of, 150, 169–72, 271–9, 282–4,
Friedman, Michael, 191 286–7, 289, 291, 300
role in cognition, 11, 13, 145–75, 178, 268,
George, Rolf, 161n 272; see also acquaintance; cognition, as
geometry, 174–5, 181, 183–94, 200n discursive, objective validity
Ginsborg, Hannah, 168 irreducibility, 244–7
God, 5–6, 14, 21, 44–5, 61, 67–8, 79, 84, 87–9,
129, 200, 213, 244, 261; see also Jackson, Frank, 120n, 251n
transcendent metaphysics judgment, 157, 172–74, 264–5, 267, 274, 283
INDEX 
Land, Thomas, 173 Paralogisms, 24, 46, 55, 102

Langton, Rae, 139n, 222, 223n, 233n, 238n, Paton, H., 17n
243–9, 251, 252n perception
law (necessary connection), 48–9, 96, 123, 144, adverbial account of, 39–43
214, 219–20, 223, 226, 240, 244–5, 247, animal, 149, 151, 154–5, 169n, 175, 262,
267, 273, 275, 282–5, 303–4, 306–7; see also 265–6, 272, 283
cause Cartesian account of, 12–13, 105–8, 110, 118,
Leibniz, 26, 75, 198n, 199, 224, 232–8, 240, 128, 134
245n, 246; see also Amphiboly; monad; disjunctivist account of, 105, 107–8, 112
transcendent metaphysics intentional object account of, 39–43
Locke, John, 34, 50, 73, 84n, 103, 125–6, 135, Kant’s account of, 151
209, 218, 222n, 223, 225, 229–31, 248; non-veridical, 111–12, 114–16, 122, 137
see also qualities, primary and secondary; relational account of, 12–13, 103–11, 113–17,
secondary quality analogy 123, 133–5
representationalist account of, see perception,
magnitude, 171; see also categories relational account of
McDowell, John, 152n phenomena, 61, 63–4, 66, 69, 96, 236, 243, 268,
McLear, Colin, 165n, 166n 296; see also appearances; noumena
mentalism, see phenomenalism phenomenalism, 7–12, 22–7, 30, 35, 37–58,
metaphysics 78–9, 90, 92, 103, 109, 115, 126–7, 129,
as distinct from mathematics, 177–9, 132–3, 135, 137–8, 142–3, 201–4, 211, 217,
183–4, 192 299, 305
of experience, 303 sophisticated, 42, 45, 47, 49, 52, 56, 202, 217;
possibility of, 14, 177–8, 208, 260–1, 290–303 see also perception, adverbial account of;
transcendent, 6, 67, 89–90, 213, 256, 303–4 perception, intentional object account of
Mill, John Stuart, 50 possibility
monad, 7, 14, 21, 61–2, 73, 148, 235–8, 244, 261, actual, 142
268, 270; see also Leibniz logical, 62, 70, 140–2, 240, 247; see also
concept, problematic
neglected alternative, 83, 181, 186–90, 197–8 real, 62, 64, 70, 140–2, 240, 247; see also
non-conceptualism, 165, 167, 173–4; see also objective validity; experience, possible
conceptualism (possibility of)
non-phenomenalist idealism, 13, 17, 22, 27, Postulates of Empirical Thought, 139–40
41, 58 powers, see forces (powers)
noumena, 7, 51, 60–76, 307 Prauss, G., 249n
as a boundary concept, 64; see also possibility, properties
logical; concept, problematic; things in deviant, 277–80; see also object, deviant
themselves, as a merely limiting concept, essentially manifest, see qualities, essentially
negative conception of, 60–5, 88 manifest
positive conception of, 60–5, 89 intrinsic and extrinsic, 72–3, 75, 221–58
noumenalism, 7–11, 59–76, 92–3; see also mere Cambridge, 253, 277–8
intelligibilia; noumena; things in relational, 9, 73, 113, 115, 121–4,
themselves 221–58
object qualities
deviant, 277–80; see also properties, deviant essentially manifest, 117–18, 122, 128, 202,
relation to an, see objective validity 207, 209–10, 217, 219–21
supersensible, 30, 32, 35, 70, 84; see also primary and secondary, 44–5, 47, 84,
intelligibilia; noumena, positive conception 102–3, 105, 121, 125–44, 156, 209–10,
of transcendental object = x, 287–9; see also 218, 222–3, 225–6, 229–31, 248, 254;
substance; noumena, negative conception of see also colour; secondary
unified object of thought, 271–85; see also quality analogy
appearances, as identical to things in
themselves rationalism, 26, 88–90, 180, 199–201, 256, 261;
objective reality, see objective validity see also empiricism; transcendent
objective validity, 147, 191, 259–89 metaphysics
ontology, 5–6, 88, 207, 226, 230, 232, 239, 256 reality, 227; see also categories
 INDEX
realism as grounding appearances, 32–3, 65–70, 74–6,

empirical, 207, 212–13, 215, 217, 220–2, 224, 95–7, 129–32, 220–6, 229–33, 236–43, 245,
226–7, 229, 232, 258 248–58
transcendental, 79, 86, 88 as intrinsic natures (non-relational), 5, 27, 30,
epistemological structural, 34, 222–6, 258 33, 72–3, 75, 85, 121, 208, 220–6, 229,
ontological structural, 230, 251, 256 231–58
Refutation of Idealism, 12, 52–5, 102, 156, our inability to cognize, 27–33, 196; see also
159, 262 cognition, as discursive; transcendent
relations, see properties, relational metaphysics
representation, 12–13, 25, 41–4, 74–5, 109–10, transcendent metaphysics, 5, 6, 67, 78, 89, 97,
115–18, 128, 133, 156–7, 168–75, 271–5, 152, 153, 213, 256, 304
288, 302–3; see also appearances, as mere Transcendental Aesthetic, 17, 21, 44, 52, 88,
representation; intuition; perception, 91, 94, 96, 133, 146, 148, 153, 155, 163–70,
relational account of 174, 176–204, 208, 227, 262–3, 268,
and idealism, 56–8, 132–5 272, 276, 286, 291–4, 296, 298,
as presentation, 13, 25, 104; see also intuition, 300, 302
as giving (presenting) objects Transcendental Analytic, 168–75, 191, 224,
Rosefeldt, Tobias, 17n 262–3, 268, 294, 296; see also
Russell, Bertrand, 20, 158–9 Transcendental Deduction
transcendental argument, 182, 184–5, 299, 301
science, 18, 47, 74, 90, 118–19, 132, 135, 207–12, Transcendental Deduction, 18, 29, 146, 150,
216n, 217–32, 234n, 242, 251n, 255–8, 154–5, 162–3, 168–75, 183, 191, 197, 208,
305–7 259–302
Shoemaker, Sydney, 253n, 281n as epistemological, 259, 285–9, 296
secondary quality analogy, 103, 125–44, 209 transcendental idealism,
Senderowicz, Yaron, 69 argument for, 186–201
sensibility (receptivity), 63–4, 75, 87–8, 148, as an alternative to ontology, 5–6, 8; see also
151, 155, 156n, 174–5, 235, 262, 266, 273, transcendent metaphysics
290–2; see also cognition, as discursive; as bare empirical realism, 10, 26, 28, 59–60,
intuition, space and time as a priori 62, 65–70, 75–6, 91, 131
intuitions, understanding as deflationary, 7–10, 28, 30, 59–60, 62, 65,
solidity, 222, 225–6, 229, 247, 250 71, 75–6, 77–97, 130–1, 208, 215, 247,
soul (immortal), 6–7, 14, 21, 29, 61, 67–8, 89, 296, 305
173, 183, 200, 213, 261, 273 as empirical realism, see realism, empirical
space and time as a priori intuitions, 102, 162–7, as noumenalist, see noumenalism
186–9, 191, 194–204, 236, 292, 297 as phenomenalist, see phenomenalism
Smith, A. D., 121 in terms of ‘two-world’, ‘two-aspect’, and
Stephenson, Andrew, 156n, 157n ‘one-world’, 8–9
Strawson, P. F., 7, 11n, 86, 286–7 simple argument for, 178, 180–6, 197, 200,
Stroud, Barry, 182n 300; see also neglected alternative
substance, 48–9, 68, 233n, 237–8, 245n, 246–51, transcendental unity of apperception, 170,
282–3; see also Analogies of Experience; 191–2, 260, 283
categories; impenetrability
superaddition, 244–5, 252 unconditioned, 90–5; see also transcendent
synthesis, 150, 155, 161–2, 165–70, 172–5, metaphysics
261–2, 265, 267, 271–87, 289, 291–2, understanding (generality and spontaneity),
298, 300 172–5, 237–9, 242–3, 246, 266, 275–85,
synthetic a priori, 5–7, 14, 18, 52, 69, 176–94, 290–2; see also cognition, as discursive;
260–1, 286, 290–303 categories; concept; judgment; noumena;
sensibility
things in themselves, see also appearances; unity
noumena; transcendental idealism conceptual, 165–6, 171, 261, 273–4
as a merely limiting concept, 28, 32, 64–5, 67 intuitional (aesthetic), 165–6, 171–2, 274
as causally inert, 243–8, 253 systematic, 90–1, 216n, 234n, 305, 307n
as identical to appearances, 27–30, 60,
70–5, 242 Van Cleve, James, 27n, 38–9, 50, 73, 126n, 194n
INDEX 
Warren, Daniel, 195n, 225–6, 232, 234n, 251 Wood, Allen, 69

Watkins, Eric, 245n world, as a determinate whole, 67, 89–91, 93–5,
Westphal, Kenneth, 161n 204, 215–17, 241, 258, 305, 307; see also
Willaschek, Marcus, 159n, 199n transcendent metaphysics

Lucy Allais - Manifest Reality - Kantâ S Idealism and His Realism (2015, Oxford)

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Lucy Allais - Manifest Reality - Kantâ S Idealism and His Realism (2015, Oxford)

Загружено:

Авторское право:

Доступные форматы

OUP CORRECTED PROOF – FINAL, 8/7/2015, SPi

To Geoffrey and Bernardine Barlow

was published in ‘Kant, Non-Conceptual Content and the Representation of Space’,

A Note on References to Historical Texts xi

Part One. Textual Evidence and an Interpretative

Part Two. Manifest Reality

Part Three. Kant’s Idealism and his Realism

A Note on References to Historical Texts

A/B Critique of Pure Reason

I Introduction to Part One

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

Kant is not an idealist. (1999: 2)

Second, James Van Cleve:

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

who agree that he is committed to the existence of things in themselves, there is

This passage has been taken to support an interpretation of Kant’s transcendental

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

interpret transcendental idealism as any kind of metaphysical position would be to

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

different ways of considering objects.14 With this messy terminological back-

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

The existence of strong considerations in favour of both sides as well as serious

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

assumptions we bring to reading it, in particular assumptions about the nature of

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

aims is to show how fruitful it is for the interpretation of transcendental idealism if we

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

that the philosophical work for which he invokes transcendental idealism—the

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

language, I argue that Kant’s idealism can be understood as rejecting experience-

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

II Textual Evidence for Idealism

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

Kant’s use of the term ‘representations’ is seen by many commentators as pro-

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

III Things in Themselves

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

TOWARDS A MODERATE METAPHYSICAL INTERPRETATION 

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

WHY KANT IS NOT A PHENOMENALIST 

Some interpretations try to avoid understanding Kantian phenomena as distinct

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM

WHY KANT IS NOT A PHENOMENALIST 

 TEXTUAL EVIDENCE AND AN INTERPRETATIVE PENDULUM