Академический Документы
Профессиональный Документы
Культура Документы
The Art of the Impossible
S
Prize: One hundred dollars to the first person who identifies a picture of a
logical impossibility. I may be willing to pay more for the painting itself.
This finder’s fee is simply for pointing out the picture. Let me explain more
precisely what I seek.
1 Illegal Pictures
There is a genre of children’s picture puzzles that is marked by the question
‘What is wrong with this picture?’ Well, that goat does not belong in the
library.That clock is mirror-reversed. Ostriches do not fly. . . . The job of the
viewer is to spot the incongruities.
An impossible picture features a nomic incongruity—a violation of a law.
There are many pictures that depict scientifically impossible situations. René
Magritte’s Collective Invention features a reverse mermaid: woman from foot to
waist, fish from waist to gills.
An impossible situation need not involve an impossible object. Many of
Magritte’s paintings feature ordinary objects in impossible relationships. Zeno’s
Arrow simply shows a huge rock that fails to be gravitationally related to the
earth. Actually, all ‘impossible objects’ involve impossible relationships. For
An ancestor of this chapter was presented at the University of Saskatchewan. I thank Karl Pfeifer,
Walter Sinnott-Armstrong, and the editors of this volume,Tamar Gendler and John Hawthorne,
for comments and imaginative suggestions. I thank Milton Katz for permission to reprint one of
his figures, and István Orosz for permission to reprint his drawings.
Gendler-09 2/10/02 12:45 PM Page 338
2 Pseudo-Pictures
Those who regard pictures as sentences are often unclear about whether impossi-
ble pictures actually qualify as sentences. Linguists say that a language is a set
of sentences defined by a vocabulary and a grammar specifying how the words
can be combined into sentences.Therefore, ungrammatical sentences are not
part of the language.Thus, if one characterizes impossible pictures as ungram-
matical sentences of the picture language (Huffman ),then one should not
count them as pictures.This seems harsh. Kennedy attempts a compromise:
Combining incompatible words makes an ‘impossible’ sentence, a sentence that can
have no direct referent in reality.An example is ‘Colorless green ideas sleep furiously.’
Gendler-09 2/10/02 12:45 PM Page 341
The sentence is grammatical—it is not nonsense like ‘furiously sleep ideas green
colorless.’A drawing,too,can show impossible things,things that cannot have a direct
equivalent in reality. (Kennedy : )
If ‘Colorless green ideas sleep furiously’ expressed an impossibility, then it
would have a negation that expresses a necessary truth. But Chomsky regards
‘It is not the case that colorless green ideas sleep furiously’ as equally meaning-
less. He takes his most famous utterance ‘Colorless green ideas sleep furiously’
to illustrate the fact that a meaningless sentence can conform to the grammar
of a language. He thinks that the sentence violates semantic rules. By contrast,
Chomsky (correctly) thinks that contradictory statements fully conform to all
rules of the language. They merely express propositions that are necessarily
false. A grammar that fails to generate contradictory English sentences is
an inadequate grammar. Grammaticality cannot be a necessary or sufficient
condition for possibility.
Many of those who reject the idea that pictures are sentences will still be
inclined to regard meaningless pictures as failed attempts at picturing. Happily,
prize-seekers need not take sides. Contradictions are meaningful. If there were
literally a language of outlines, contradictory pictures would be sentences
within that language.
The argument is sound, because I am thinking of two women who make the
premises true and because the argument is valid. Some of our beliefs are
demonstrative. Demonstratives cannot be reduced to qualitative descriptions.
Hence, pictures can play an essential role in forming the objects of belief.
Nevertheless, I am not interested in the contradiction ‘The color of her hair
is——and is not——’. Although the image plays a role in constituting this
demonstrative contradiction, the image is not doing any logical work.
I am not trying to raise the standard of representation to an impossible
height. If I thought a picture of a logical impossibility were impossible, then I
would feel safe in posting a large prize. In fact, I expect to pay the $ finder’s
fee. I may even wind up paying someone who does not actually believe that
it is possible to picture a logical impossibility. For all he needs to do is to
persuade me.This conditional proof can exploit my concession that conceptu-
ally impossible pictures are possible.The prize could be won simply by demon-
strating the following hypothetical: If there are pictures of conceptual
impossibilities, then this is a picture of a logical impossibility.
The issue for me is the step from conceptual impossibility to logical impossi-
bility. Prize-seekers will find it useful to see what standard of evidence I have
applied to the acceptance of conceptual impossibilities.
4 Historical Background
On the basis of introspection, the British empiricists believed that ideas have
pictorial properties. A speaker uses sentences to describe his mental images.
The pictorial mode of representation is epistemically prior to the discursive
mode.Nevertheless,the empiricists imposed an important,famous limit on the
expressive scope of pictures. David Hume writes:
Tis an establish’d maxim in metaphysics,That whatever the mind clearly conceives includes
the idea of possible existence, or in other words, that nothing we imagine is absolutely impossible.
We can form the idea of a golden mountain, and from thence conclude that such a
mountain may actually exist.We can form no idea of a mountain without a valley,and
therefore regard it as impossible. (–: )
Photographs can only be of actual objects. But drawings can prove the poss-
ibility of uninstantiated objects. A mathematician can convince an engineer
that a larger cube can pass through a smaller cube by drawing a smaller cube
with a diagonal tunnel. (A cube with a -meter face has a diagonal equal to the
square root of meters.) The proof works even if no one bothers to build the
perforated cube.
If drawing X demonstrates the possibility of X, then we appear to have a
quick proof that it is impossible to draw an impossible object. Drawing an
impossible object would show that it is possible for an impossible thing to exist.
Contradiction.Therefore, it is impossible to draw an impossible object.
This proof is sound.But only when read de re (as referring to a thing and then
reporting a feature of it). For instance, the de re report ‘The discoverer of the
largest prime number is being drawn as a winner of the Fields Medal’ entails
that the discoverer of the largest prime number exists. But some depiction is
de dicto (as concerning a representation). For instance, the de dicto report ‘In the
picture, the discoverer of the largest prime number is receiving the Fields
Medal’ does not entail that there is a discoverer of the largest prime number.
Nor does it entail that it is possible for someone to discover the largest prime
number.Any person who earns the $ finder’s fee will be giving me a de dicto
report. He will not be claiming to have discovered an impossibility that has
secured the attention of a faithful portraitist.
5 Requirements
Philosophical tradition and common sense converge on what counts as an
acceptable depiction of the logically impossible. None of the requirements
below are intended to indulge personal idiosyncrasies.
. No Equivocation
In Taxicab geometry all squares are round squares (Krause ). In this form
of non-Euclidean geometry, distance is measured by how a taxi travels on a
Gendler-09 2/10/02 12:45 PM Page 345
there is a set for every property. Nor Ludwig Wittgenstein’s Tractarian belief
that there is a decision procedure for all logical truths (Fogelin : ch. ).
Each philosopher contradicted himself. But none were ‘of two minds’.
Those who believe that anything can be depicted also believe an indivisible
contradiction.To see why, first note that some pictures depict other pictures.
For instance,Watteau’s L’Enseigne de Gersaint features an art merchant selling his
merchandise. Here is a logical truth: there is no picture that depicts all and only
those pictures that do not depict themselves. If this picture depicts itself, then it
does not depict itself.But if it does not depict itself,then it must be amongst the
pictures it depicts. Contradiction. James F. Thomson (: ) discusses a
whole family of contradictions that have this logical form.
This logic exercise proves decisively that there are logically impossible
depictions.Artists are imaginative people. But imagination is not a resource for
evading logical limits. My $ fee can still be earned, because I want only a
picture that depicts a logical impossibility, not a picture that is itself logically
impossible.
For the sake of administrative ease, I will pay a $ bonus for an indivisible
contradiction.The assumption that all impossible figures are divisible into self-
consistent components is commonly made by philosophers—for instance,
Max Cresswell ().The assumption is made uniformly by psychologists.
In impossibles, each part is ecological, but the combination of the parts violates
nature.They could not exist, so they are imaginary, but the fact that they are imagin-
ary does not make them impossible. To make an imaginary object, parts are
combined in possible ways.The combination can be possible but be a combination
that does not exist. For example, there is nothing about surfaces and air spaces that
rules out a horse with a horn, like a unicorn. Nature has not seen fit to evolve
unicorns,but it could do so without contravening its own ways with surfaces and air.
The parts of a unicorn are ecological.The combination of parts breaks no laws of
solidity.In language,one may claim ‘I saw a unicorn,a horse with a horn.’In language,
as in pictures, to be imaginative is to combine familiar parts in possible but novel
ways,whereas to be impossible is to combine the parts in novel ways that violate rules
of nature. (Kennedy : –)
All mathematical analyses of impossible figures have conformed to the idea that
impossible figures are built from possible parts. For instance, Diego Uribe has
analyzed an infinite class of impossible figures as jig-saw puzzles of just thirty-
two equilateral triangles consisting of special bar elements (Ernst b: ).
This is the basis for software (available free over the Web) that enables you to
mechanically construct impossible objects by manipulating these triangles.But
if we take the analogy with language seriously, we should doubt that these
Gendler-09 2/10/02 12:45 PM Page 347
analyses exhaust the stock of impossible figures. All natural languages can
express infinitely many indivisible contradictions. If it is possible to pictorially
represent a contradiction, then it should be possible to pictorially represent an
indivisible contradiction.
An indivisibly inconsistent picture would side-step the problem of distin-
guishing inconsistency from doubt. Consider ‘continuity’ errors in movies. For
instance,in the last ten minutes of Mission:Impossible ,secret agent Ethan Hunt
is riding a motorcycle in a chase scene.The last two digits of his license plate
shift from to .This production error does not make the movie inconsist-
ent about the license plate number. Instead, the conflicting depictions merely
create doubt whether the license plate ends with or .Whenever the con-
tradiction is divisible, there is the opportunity to interpret the scene in this
uninteresting way. Depiction of an indivisible contradiction would avoid
this hitch.
virtue of reflections on their design rather than by execution. But only those
who are inducted into special fields of science and philosophy magnify this
power.
In any case, if subjectivist ‘depiction’ sufficed, then my oldest son Maxwell
would deserve the $ reward.At age three, he loved the color green. In fact,
he loved it to the exclusion of all other colors.Maxwell became a green maxim-
izer. He drew a picture of a ‘green all over rainbow’ with a single green
crayon. Since a rainbow must be multicolored, no scene could match my son’s
description.
The problem with Maxwell’s picture is that it does not reveal what it would
be like to see a uniformly green rainbow. In artistic contexts,‘depict’ is used in
a way that allows failure.When students take art classes, they want to learn how
to render objects perceptually. Techniques such as drawing in perspective
capitalize on the running start we all have from folk optics.The students already
know how to depict objects discursively via pure stipulation (or stipulation plus
an ancillary stick figure).Suppose the art instructor says,‘Drawing is not as hard
as it looks.All you need to do is to decide what your marks on the canvas are
intended to represent.Then, presto, you are done.’The art students will rightly
demand a tuition refund.
One of René Magritte’s most famous pictures,TheTreason of the Pictures ( This
is not a pipe) consists of a picture of a pipe along with the caption ‘This is not a
pipe’.Peter Strawson might be tempted to say that this is not a depiction of any-
thing.According to Strawson (: –), a statement of the form P & not-P
says nothing because the ‘not-P’ merely cancels out the P. Others interpret the
picture as making the point that the picture of the pipe is not itself a pipe.This
illustrates a standard alternative to viewing a picture as depicting an impossibil-
ity: one attributes an ambiguity.To forestall this attribution of an ambiguity,
suppose the caption had instead been ‘This is not a picture of a pipe’.Would
Magritte then have pictured a contradiction?
Well, maybe. But it would not be the kind of picture I seek. I want the con-
tradiction to be within the picture, not between the picture and its caption.
I am not forbidding the kind of illocutionary variety that Wittgenstein alludes
to when he notes that a picture of a boxer can be used to report or instruct or
inquire. But I do forbid examples in which the content of the picture plays no
role.For instance,the picture plays no role in the pictorial conundrum (inspired
by Peter Geach ()) shown in Figure . Here is the enigma: Some pictures
are well-titled, in that they accurately describe the picture. Other pictures are
ill-titled, because they are descriptively inaccurate. But now consider Ill-titled.
If Ill-titled is ill-titled, then its title accurately describes the picture, and so
Ill-titled is well-titled. But if Ill-titled is well-titled, then the picture’s title fails to
Gendler-09 2/10/02 12:45 PM Page 349
⬙
.⬙ .⬙
⬙ ⬙
moving. Absent the intervention of a censor, we see the rocks both ways at
once. Some psychologists have interpreted this as an example of seeing the
logically impossible:
although the after-effect gives a very clear illusion of movement, the apparently
moving features nevertheless seem to stay still! That is, we are still aware of features
remaining in their ‘proper’ locations even though they are seen as moving.What we
see is logically impossible! (Frisby : )
Tim Crane () thinks this shows that concepts cannot be part of perception.
One of the standard tests for ambiguity is the contradiction test. If a competent
speaker believes that x is F and x is not G,then F and G must have distinct mean-
ings M that is, express different concepts. However, in the waterfall illusion, the
speaker is inclined to believe that the rock is moving and not moving.The only
way to retain the contradiction test and deny ambiguity is to abandon the
assumption that concepts are involved in the speaker’s visual judgment.
D. H. Mellor boggles at how a judgment can be inconsistent if it does not
involve concepts. Just what could be the contradiction? A contradiction is a
proposition, so necessarily involves concepts. He goes on to deny that there is
any tendency to believe a contradiction.The waterfall illusion simply involves
two inclinations that cancel out:
We could,however,be inclined to believe that Fa,while also being inclined to believe
that ~Fa.And that, I submit, is what happens in the Waterfall Illusion.There isn’t sim-
ply, as Crane claims,‘a contradiction in the one content of one attitude’. Rather we
are conscious of seeing that a moves while also seeing that it doesn’t.One of these two
perceptual experiences gives us the corresponding belief, say that a doesn’t move,
which then suppresses the rival inclination to believe that it does. (: )
Mellor is proposing a divide-and-conquer solution.There is merely disagree-
ment between two self-consistent perceptual experiences.
I disagree with Crane and Mellor.The rocks are perceived inconsistently,but
it does not follow that the observer perceives a contradiction.The observer sees
ordinary rocks via an inconsistent homuncular process. Such inconsistent
processes are common.What is uncommon is our awareness of the inconsist-
ency. Only in atypically simple situations do we notice incoherences that are
systemic to experience.
Our visual system’s ambivalent reaction to the Necker cube is often said to
illustrate the system’s insistence on consistency. However, the elongation
sequence suggests that consistency is negotiable. Moving top to bottom,
the inconsistent interpretation begins as a weak alternative to the consistent
alternatives. But as the elongation increases, the inconsistent interpreta-
tion becomes the dominant interpretation.Thus the inconsistent interpreta-
tion prevails even though the observer is being primed on consistent
interpretations.
A consistent interpretation is always logically available. Any ‘impossible
figure’ can be interpreted as a consistent drawing by treating the drawing as a
two-dimensional assembly of lines or as a conglomerate of distinct pictures.
With opposite deviousness, one can also interpret any possible figure as an
impossible figure. It is just a matter of connecting consistent dots in an incon-
sistent way (see Fig. ).
Logical availability does not imply psychological availability. Our visual
system is cognitively impenetrable. It cannot be modified to accommodate
the discovery of new possibilities. For instance, topologists have acquired an
excellent algebraic understanding of four-dimensional objects.They can even
calculate an impossible object that would be perceived by beings who can
perceive four-dimensional objects (Kim ). But they cannot visualize the
objects and so cannot grasp the depiction at first hand.
Gendler-09 2/10/02 12:45 PM Page 357
x
–
6 Why I am Optimistic
My picture of the impossible starfish is composed of five Penrose triangles.
Unlike the truncated pyramid, the Penrose triangle stimulates dissonance.
Unlike the elongated Necker cube,the dissonance tracks genuine incoherence.
There really is something awry in the picture.The vertices are each possible,
but not co-possible. One sees this without relying on labels or captions.The
inconsistency is within the picture itself.
Is inconsistency too abstract a relation to be ‘in the picture itself ’? The worry
becomes less pressing when one dwells on the range of properties to which
perceivers are sensitive.When the objects in view are fewer than four, we are
Gendler-09 2/10/02 12:45 PM Page 359
original orientation, hold the photograph above your head with the light side
closest to you.)
The artist’s intention can also ensure a fact of matter. But what if the artist
wants us to interpret his depicted staircase as going both up and down? Even if
we accept the artist’s authority on the matter, I would object that this is the
wrong kind of impossible picture. Just as the ambiguity of the Jastrow duck-
rabbit is internal,the inconsistency of a prize-winning picture must be internal.
The Penroses themselves seem unconcerned about the distinction between
a picture that looks as if it depicts an impossibility and a picture that really
depicts an impossibility. Lionel Penrose delighted in the construction of little
staircases and ramps that look impossible when photographed (or viewed with
one eye) from the appropriate angle. I like impossible construction projects
too—see ‘The impossible plumber’s son’ (Fig. ).The top pipe segments are
actually about a foot apart.The impossible boy darting through my carefully
staged scene is actually my two-year-old, Zachary Sorensen.
Gendler-09 2/10/02 12:45 PM Page 361
The linguistic analogue of Katz’s figure is ‘Some bats are not bats’. The
sentence has a reading that is a contradiction about a certain kind of winged
mammal and a reading that is a contradiction about a certain piece of baseball
equipment.Although the sentence is contradictory under all disambiguations,
the sentence is not itself a contradiction. Similarly, Katz’s figure is not itself
inconsistent. It is merely inconsistent under each of its disambiguations.
A Penrose triangle could be composed of smaller Penrose triangles.Thus an
impossible figure can be composed of other impossible figures.The linguistic
analogue of these figures is a conjunction such as ‘Some electrons are not
electrons and it is not the case that some electrons are not electrons’.This state-
ment is contradictory at two levels.At the base level it is contradictory by virtue
of having a contradictory conjunct. But it is also contradictory in virtue of its
overarching ‘P & ~P’ form.
The hierarchy could be continued upwards indefinitely. Could it extend
endlessly downward? A negative answer is implied by those who think that all
impossible figures can be ultimately decomposed into possible figures. But
perhaps there could be impossible ‘gunk’. Gunk is infinitely divisible matter.
A Penrose triangle might be composed of smaller and smaller Penrose triangles,
ad infinitum.
9 Nonepistemic Conceiving
Fred Dretske () focused much attention on the distinction between epi-
stemic seeing (‘seeing that’),which entails belief,and nonepistemic seeing,which
is neutral with respect to belief. I can nonepistemically see a three-legged frog
even though I believe it is a hallucination. I nonepistemically see object O just
in case there is an appropriate causal connection between object O and myself.
Gendler-09 2/10/02 12:45 PM Page 364
must be able to put the counter-example in his own words.In practice,we have
a rich network of tests for conceiving.
10 Depicting Inconsistencies
Stories can be told in a way that guarantees consistency. But they need not be.
Graham Priest’s memorial tribute to the para-consistent logician Richard
Sylvan,‘Sylvan’s Box’, is designed to be a story in which a contradiction plays
an essential role. The centerpiece of the tale is an empty box that contains
something. In accompanying commentary, Priest explicitly rejects the sugges-
tion that the characters are mistaken about the box or that the box is empty in
one sense but not another.That would be a poor tribute indeed! The aptness
and poignancy of the tale evaporate if we render his story consistent through
the usual devices.
Fiction works like belief. If someone sincerely says that he believes that p,
then that is strong evidence he believes that p. If the author sincerely says that p
is part of his story, then that is strong evidence that p is part of his story. Both
principles are defeasible, because we are also guided by indirect principles of
belief attribution that may trump the speaker’s authority.
The story-teller pretends to be reporting his beliefs. So anything that can be
believed can be part of a story.Priest does not believe that Sylvan had an empty
box that contained something. But he could believe it.A contradiction can be
true in a story in the same way that a contradiction can be true in a belief
system M that is, the contradiction is affirmed in the system. Just as the conse-
quences of p cannot be inferred from ‘Peewee believes that p’, so they cannot
be inferred from ‘In Peewee’s story, p’.
I am not sure whether Priest could convey the belief in a purely pictorial
fashion. However, I am impressed with his resources. If, as I believe, experience
is systematically inconsistent, and if, as I believe, there is no obstacle to incon-
sistent fiction, then visual depiction of a logical inconsistency seems feasible.
Escher successfully pretends to be faithfully recording geometrically impossi-
ble scenes. Geometry and logic seem comparably abstract. So there does not
seem to be any artistically relevant difference between the two.But when I per-
sonally try to arrive at the depiction of a logical impossibility, I draw a blank.
a blank canvas. (Perhaps you have seen reproductions at your local art-supply
store.) As suggested by its title, the work depicts a universe in which nothing
exists. Logic forbids the empty universe. For the universal quantifier has exis-
tential import:‘Everything is F’ entails ‘Something is F’.And ‘Something is F’
means ‘There exists something which is F’.Thus the law of identity,‘Everything
is identical to itself ’ entails ‘Something exists which is identical to itself ’.
My picture vacuously depicts the empty universe in all relevant detail.True, it
relies heavily on its title for interpretation, but it is otherwise independent of
discursive elements.
One reservation about awarding the prize to myself is the controversy that
surrounds the claim that ‘Nothing exists’ is logically false. Most people believe
that they can imagine an empty universe (either straight off or by subtracting
until nothing remains).They take conceivability to be powerful evidence for it
being genuinely possible.They are also influenced by a methodological claim:
logic should only supply the structure for what exists and ought not to commit
us to the content. Existence is a contingent property. So it is contingent
whether anything at all exists.
This opposition to counting ‘Nothing exists’ as a logical falsehood does not
persuade me.But I am not going to fight it.I present ‘The empty universe’only
to exempt it as a prize-winner.I want a substantive depiction of a logical imposs-
ibility. No tricks! Limiting cases will be returned to the sender.
Bradley, M. J. (),‘Building Home Plate: Field of Dreams or Reality’, Mathematics
Magazine, : –.
Braus, Ira (), ‘Retracing One’s Steps: An Overview of Pitch Circularity and
Shepherd Tones in European Music, –’, Music Perception, (): –.
Burge,Tyler (),‘Epistemic Paradox’, Journal of Philosophy, (): –.
Crane,Tim (),‘The Waterfall Illusion’, Analysis, : –.
Cresswell, M. J. (),‘A Highly Impossible Scene’, in R. Baurele, C. Schwarze, and
A. von Stechow (eds.), Meaning, Use and Interpretation of Language (Berlin: De
Gruyter), –.
Dretske, Fred (), Seeing and Knowing (Chicago: University of Chicago Press).
Dwyer, F. M. (), ‘Adapting Visual Illustrations for Effective Learning’, Harvard
Educational Review, : –.
Ernst, Bruno (a), Adventures with Impossible Figures (Norfolk: Tarquin
Publications).
—— (b), Optical Illusions (New York:Taschen; trans. Karen Williams, ).
Fogelin, Robert (), Wittgenstein (London: Routledge and Kegan Paul).
Gendler-09 2/10/02 12:45 PM Page 367
Russell, Bertrand (), Human Knowledge: Its Scope and Limits (New York: Simon
and Schuster).
Ryan,T.A., and Schwartz, C. (),‘Speed of Perception as a Function of Mode of
Representation’, American Journal of Psychology, : –.
Shepherd, Roger (),‘Circularity in Judgments of Relative Pitch’, Journal of the
Acoustical Society of America, : –.
Sorensen, Roy (), Thought Experiments (New York: Oxford University Press).
—— (), ‘Modal Bloopers: Why Believable Impossibilities are Necessary’,
American Philosophical Quarterly, () ( July): –; repr. in Patrick Grim,
Kenneth Baynes, and Gary Mar (eds.), The Philosopher’s Annual (Atascadero,
Calif.: Ridgeview, ), vol. xix.
—— (),‘Ambiguity, Discretion, and the Sorites’, Monist, (): –.
Stalnaker, Robert (), Inquiry (Cambridge, Mass.: MIT Press).
Strawson,Peter F.(),Introduction to Logical Theory (NewYork:John Wiley & Sons).
Thomson, James F. (), ‘On Some Paradoxes’, in R. J. Butler (ed.), Analytical
Philosophy (New York: Barnes & Noble), –.
Tidman, Paul (),‘Conceivability as a Test for Possibility’, American Philosophical
Quarterly, (): –.