Werner Callebaut - Evolutionary Epistemology

EVOLUTIONARY EPISTEMOLOGY
SYNTHESE LIBRARY
STUDIES IN EPISTEMOLOGY,
LOGIC, METHODOLOG Y, AND PHILOSOPHY OF SCIENCE
Managing Editor:
JAAKKO HINTIKKA, Florida State University, Tallahassee
Editors:
DON ALD DA V IDSO N, University of Californiil. Berkeley
GABRltL NUCHELMANS, University of Leyden
WESLEY C. SALMON, University ofPittsburgh
VOLUME 190
EVOLUTIONARY
EPISTEMOLOGY
A Multiparadigm Program
with a complete
Evolutionary Epistemology Bibliography
Edited by
WERNER CALLEBAUT
Limburgs Universitair Centrum, Belgium &
Rijksuniversiteit Limburg. The Netherlands
and
RIK PINXTEN
Rijksuniversiteil Gem. Belgium
D. REIDEL PUBLISHING COMPANY

A MEMBER OFTHE KLUWER
ACADEMIC PUBLISHERS GROUP
DORDRECHT / BOSTON / l.ANCASTER / rOKYO
Library of Congress Cataloging in Publication Dat.

Evolutionary epistemology.
(Synlbese library; v. 190)
Bibliography: p.
Includes index.
1. Knowledge, Theory of-Congresses. I. CaUebaut, Werner.
II. Pinxten, Rik. III. Title: Evolutionary epistemology bibliography.
N. Series.
BDl61.E85 1987
121
87-20629
ISBN-I3: 978-94-01G-8260-0
e-ISBN-I3: 978-94-009-3967-7
DOl: 10.10071978-94-009-3967-7
Published by D. Reidel Publishing Company,

P.O. Box 17,3300 AA Dordrecht, Holland.
Sold and distributed in Ibe U.S.A. and Canada
by Kluwer Academic Publishers,
101 Philip Drive, Assinippi Park, Norwell, MA 02061, U.S.A.
In all other countries, sold and distributed

by Kluwer Academic Publishers Group.
P.O. Box 322, 3300 AH Dordrecht, Holland.
All Rights Reserved
1987 by D. Reidel Publishing Company, Dordrecht, Holland

Softcover reprint of the hardcover Ist edition 1987
No part of the material protected by this copyright notice may be reproduced or
utilized in any form or by any means, electronic or mechanical
including photocopying. recording or by any information storage and
retrieval system, without written permission from the copyright owner
TABLE OF CONTENTS
vii
PREFACE
PARTI:BACKGROUND
CALLEBAUT and RIK PINXTEN
Evolutionary
Epistemology Today: Converging Views from Philosophy, the
Natural and the Social Sciences
IL YA PRIGOGINE / The Meaning of Entropy
HENR Y C. PLOTKIN / Evolutionary Epistemology and the
Synthesis of Biological and Social Science
RENE THOM / Epistemology of Evolutionary Theories
CECILIA M. HEYES / Cognisance of Consciousness in the Study of
Animal Knowledge
WERNER
57
75
97
105
PART II: EVOLUTIONARY APPROACHES

TO SCIENCE AND TECHNOLOGY
Selection Theory and the Sociology of
Scientific Validity
139
LOUIS BOON / Variation and Selection: Scientific Progress Without
159
Rationality
KARIN KNORR CETINA / Evolutionary Epistemology and
Sociology of Science
179
GERHARD VOLLMER / What Evolutionary Epistemology Is Not
203
ANDREW J. CLARK / The Philosophical Significance of an
Evolutionary Epistemology
223
LINNDA R. CAPORAEL / Homo Sapiens, Homo Faber, Homo
Socians: 1 echnology and the Social Animal
233
DONALD T. CAMPBELL /
PART III: THE PIAGETIAN APPROACH

CHRISTIANE GILLIERON /
Evolutionary?
Is Piaget's "Genetic Epistemology"

247
TABLE OF CONTENTS
vi
ARTHUR I. MILLER /
The Genesis of Atomic Physics and the
Biography of Ideas
CLAUDE LAMONTAGNE /
267
Sensorimotor Emergence: Proposing a
Computational "Syntax"
APOSTEL
/
Evolutionary Epistemology.
Epistemology. History and Neurology
LEO
283
Genetic
311
PART IV: EXTENSIONS AND APPLICATIONS

and DANI DE WAELE / The Exchange of Genetic
Information Between Organisms of Distinct Origin Can Play an
Important Role in Evolution
JEAN PAUL VAN BENDEGEM / Fermat's Last Theorem Seen as an
Exercise in Evolutionary Epistemology
FERNAND VANDAMME / Language and Evolutionary or Dynamic
Epistemology
PHILIPPE V AN PARIJS / The Evolutionary Explanation of Beliefs
JEF SCHELL
327
337
365
381
PART V: BIBLIOGRAPHIES
and WERNER G.
Evolutionary Epistemology Bibliography
DONALD T. CAMPBELL, CECILIA M. HEYES,

CALLEBAUT /
405
GENERAL BIBLIOGRAPHY
433
INDEX
451
PREFACE
This volume has its already distant or1g1n in an international conference on Evolutionary Epistemology the editors
organized at the University of Ghent in November 1984. This
conference aimed to follow up the endeavor started at the
ERISS (Epistemologically Relevant Internalist Sociology of
Science) conference organized by Don Campbell and Alex Rosenberg at Cazenovia Lake, New York, in June 1981, whilst injecting the gist of certain current continental intellectual
developments into a debate whose focus, we thought, was in
danger of being narrowed too much, considering the still
underdeveloped state of affairs in the field.
Broadly speaking, evolutionary epistemology today consists of two interrelated, yet qualitatively distinct investigative efforts. Both are drawing on Darwinian concepts,
which may explain why many people have failed to discriminate
them. One is the study of the evolution of the cognitive
apparatus of living organisms, which is first and foremost
the province of biologists and psychologists (H.C. Plotkin,
Ed., Learning, Development, and Culture: Essays in Evolutionary Epistemology, New York, Wiley, 1984), although quite
a few philosophers - professional or vocational - have also
felt the need to express themselves on this vast subject
(F.M. Wuketits, Ed., Conce ts and Approaches in Evolutionary
Epistemology, Dordrecht Boston, Reidel, 1984). The other
approach deals with the evolution of science, and has been
dominated hitherto by (allegedly) 'naturalized' philosophers;
no book-length survey of this literature is available at
present.
Having explored the already overwhelming literature
labeled 'evolutionary epistemology' (see the comprehensive,
up-to-date bibliography by Campbell, Heyes and Callebaut at
the end of this volume), we felt that on the whole, little
had been accomplished in terms of either dependable theory or
relevant application. As Ron Giere was to put it at the Ghent
conference,
what evolutionary epistemology seems to be
lacking today i.s a really good Kuhnian exemplar, a paradigmatic problem solution. Our vaulting ambition with the conference was to lay the necessary groundwork for the transformation of evolutionary epistemology from a rather heterogeneous
collection
of
more or less wild and more
or
less
interesting speculations into a theoretically sound and empivii
viii
rically fertile,
PREFACE
multi- or even interdisciplinary yet scien-
!!!! research program. A number of first-rate scientists had
accepted to join our project; here was our chance to prove

its timeliness and feasibility.
Needless to say, things actually turned out rather differently than we had expected. We wanted the conference to be
critical of existing approaches to evolutionary epistemology
as well as constructive. To succeed in the second undertaking,
a number of current intellectual developments,
hitherto unexplored by evolutionary epistemology, were scrutinized:
(i) the suggestions for an evolutionary theory of knowledge,
including science, arising from current transformations in
the natural sciences, in particular far-from-equilibrium
thermodynamics;
(ii) current micro-sociological approaches to the study of
science and technology; and finally
(iii) genetic epistemology in the Piagetian tradition.
In this way, we hoped to shed new light on questions such as
"How come human mathematics seem to fit the world"? (Piaget!
Apostel), or "How does science accomplish referential competence for its ideas as distinguished from mere pragmatic
success?" (Campbell), or "What is the survivalistic advantage
of evolving devices to test the honesty, the veridicality,
and the pragmatic usefulness of information exchanged between
creatures?" (Neil Tennant). As this volume is apt to show,
substantial progress was actually made at a number of fronts,
either at the conference itself or during the subsequent
period of rumination and the writing up of publishable
papers. (Contributors were invited to criticize each other's
papers, which many have accepted. This procedure involved a
good deal of shipping back and forth comments to the distant
places our authors live in or moved to. It also postponed
this publication. However, we also experienced the great joy
of coming closer to one another.) On the other hand, the
conference also made it abundantly clear that the intellectual distances still to be covered before the advocates (in
some sense or other) of evolutionary epistemology will be
able to sing in unison, are considerable.
The organization of the material presented here reflects
the topics which the conference dealt with rather accurately.
(To ensure overall coherence, two papers presented at the
conference were not retained for publication in this book.
For the same reason, two papers by authors who did not talk
at the conference are included.) Part I ('Background'), one
PREFACE
ix
could say, is basically about the metaphysical underpinnings

and preconceptions of evolutionary epistemology both as
science and philosophy. The five papers gathered here deal in
one way or another with mechanism as the metaphysics of
classical, 'Newtonian' science, which also impregnates Darwinian and (even more obviously so) neo-Darwinian thinking.
Werner Callebaut and Rik Pinxten's discussion of converging
views from philosophy, the natural and the social sciences
puts evolutionary epistemology in its proper setting. Apart
from discussing the significance of the alleged demise of
mechanism (see in particular Plotkin's paper) and interpreting a number of hot issues in evolutionary epistemology
from the prespectives of the philosophy of biology and
anthropology, the introductory paper also emphasizes the
importance of devising cognitive equivalents of the biological distinction between phylogeny and ontogeny, thus paving
the way for the incorporation of a Piagetian perspective into
evolutionary epistemology (which is the subject of part III).
Prigogine and Thom's views on the metaphysical and epistemological significance of recent developments in the study of
complex physical systems (in particular, thermodynamic systems), though contradictory in certain respects, both allow
to bridge the gap between 'hard' physics and the biological
and cultural study of the evolution of mankind. According to
Ilya Prigogine, human existence can now "appear to us as one
of the most striking realizations of the basic laws of
nature" (p. 73). For Renl~ Thom, phenomena as divergent as the
birth of the universe with the Big Bang, the evolution of
life on earth and the diachronic evolution of language can be
fruitfully put on a par epistemologically. Reformulating the
old positivist demarcation criterium in terms of his own
concept of "pregnance", he concludes that "Science differs
from magic insofar as pregnance propagation can be submitted
to constraints expressed by quantitative 'laws' which can be
verified". (p. 103). Determinism will have to be weakened
(Thom) , if not given up altogether (Prigogine) in certain
respects. Henry Plotkin, wishing to account for "the continuing Balkanization of biology and the social sciences",
probes into the 'metaphysics of process' which the physicist
David Bohm claims to have located beyond mechanism. Cecilia
Heyes' paper, which concludes the first part, is a more downto earth attempt to find out whether the student of animal
learning can do without intentional language and methodology,
as mechanism would seem to imply.
Part II ("Evolutionary approaches to science and tech
PREFACE
nology") consists of six papers. Donald Campbell's contribution is a plea for 'selection theory' as a (psychological and
sociological, i.e. naturalistic) theory of justification for
beliefs,
"near
at
scientific and other. He argues that convergence is

hand
between the corrigible
justificationism
of
mainstream Anglo-American analytic philosophy and that predominant variety of evolutionary epistemology which passes
the justificatory buck to biological evolution" (p. 139). In
addition,
he
criticizes the "erroneous emphasis on continu-
ity" in the intellectual genealogies used by historiographers
of science taking an evolutionary view, as well as the exces-
sive ('Panglossian') adaptationism evolutionary-epistemological approaches often share with biological orthodoxy. Andrew
Clark's paper usefully complements Campbell's by pointing out
that evolutionary epistemology will never rebut the traditional skeptic (a point endorsed by Campbell) and suggesting new
questions (as opposed to those of the traditional faoundationalist enterprise) the evolutionary epistemologist might
rightfully ask and try to answer; e.g. the question whether an
evolutionary account of mind can be reconciled with a
corre-
spondence view of truth (yes according to Campbell, no

according to Clark), the question of cognitive universals, or
the question of the epistemic accessibility of the 'Ding an
sich'. Gerhard Vollmer elaborates the differences between the
two evolutionary epistemology programs mentioned before - the
biological ("evolutionary epistemology") and the philosophical ("evolutionary philosophy of science") program - in great
detail, thus preparing the way for an answer to the crucial
question: "What, i f anything, does evolutionary epistemology
contribute to philosophy of science?".
Whereas Vollmer
remains tributary, on the whole, to a rather traditional
epistemological viewpoint (grossly simplified: positivism cum
Popper cum Kuhn),
Knorr Cetina and Boon assess
evolutionary
epistemology as a theory of science from the perspective of

the radical relativism characteristic of the sociology of
science of the last decade, which stresses the local character of most, if not any, knowledge increments. Louis Boon
tries
to
demonstrate
that "A conception of
science
as
blind, hierarchically structured selection process no longer

needs rationality as the driving force of science" (p. 176).
Karin Knorr Cetina argues,
pace Campbell, that "discovery in
science may be better understood as an editorial process

rather than as a process of conceptual innovation, and hence
must be seen as a phenomenon rather unlike biological mutation" (p. 183). Linnda Caporael's paper, which rounds off
PREFACE
xi
part II, ventures to explain why it is that it seems easier

to design technology than to make decisions about how to use
it, by arguing that "we are neither Homo sapiens, nor even
Homo faber, but rather Homo socians, and that technology has
been hostage to a sociality that represents the wisdom of
past selective environments" (p. 234).
Part III is a an attempt - the first in the literature
we are aware of - to systematically incorporate the geneticepistemological approach as developed by the Geneva school of
Jean
Piaget into evolutionary epistemology.
Christiane
Gillieron's paper is an attempt to answer the question "Is
Piaget's 'genetic epistemology' evolutionary?" by means of a
painstaking analysis of texts documenting Piaget's own intellectual development from his 1918 novel Recherche to his
mature views on the "circle of the sciences": psychology
mathematics - physics - biology. To the external selection of
ideas, Gillieron opposes Piaget's view that knowing is not
perceiving - that it is "anticipating, formalizing, separating by self-reflection" (p. 265). Arthur Miller's paper
applies an amended Piagetian view (he also uses ideas from
Gestalt psychology) to an episode in the history of science:
the genesis of atomic physics. Claude Lamontagne wants to
provide the first elements of an answer to the question what
the process of reflective abstraction ("abstraction r~fle
chissante") really amounts to in terms of a sketch of a
computational "syntax". Understanding reflective abstraction
is of paramount importance if one wants to make sense of the
very concept of development. In the paper which concludes
this part, Leo Aposte1, Piaget's longtime collaborator, commenting on the previous three papers, tries to convey an
overall picture of the present relation between genetic and
evolutionary epistemologies. He points to a number of convergences and incompatibilities, and concludes by offering three
specific recommendations to workers in both fields.
The fourth part of this volume comprises some extensions
and
applications of evolutionary epistemology.
Writing
against the background of genetic engineering, biologists Jef
Schell and Dani de Waele challenge the neo-Darwinian orthodoxy by arguing that the transfer of independently evolved
genetic information from different types of organisms is an
important phenomenon in natural evolution itself. If warranted, this result is likely to influence the way in which
the differences between biological and sociocultural evolution are to be conceptualized (see Campbell's paper). Next
Van Bendegem, in an attempt to write an epistemologically
PREFACE
plausible history of the various ways in which mathematicians

have tackled Fermat's last theorem over time, expresses the
need to transform and enrich the historiographic framework
sketched in Lakatos' Conjectures and Refutations by freely
adopting concepts from evolutionary epistemology. In the
third paper, Fernand Vandamme, after having outlined some
general ideas about dynamic, evolutionary and developmental
explanation,
develops
some of his own views
on
linguistic
development and evolution. In the final paper Philippe Van

Parijs, writing from a sociological perspective, applies his
own
view on evolutionary explanations to the explanation
of
beliefs. He thus hopes "to go some way beyond the (pretty

unhelpful and desperately ambiguous) standard distinction
between Darwinian and Lamarckian mechanisms" (p. 398).
Acknowledgements
The conference was financed by grants from the Belgian
National Science Foundation, the Belgian Ministry of Education and the Faculty of Letters and Philosophy of the
University of Ghent. We are grateful for their support.
Finally, we want to thank all those who helped to bring
this book about, in particular Josette Thijs and Marleen
Vara, for the excellent job they did on the word processor.
The Editors.
REFERENCES:
See the Evolutionary Epistemology Bibliography and the General Bibliography at the end of this volume. For reasons of
identification,
some
items have been marked with an
they are all to be found in the second bibliography.
~-sign;
Part I:
Background
EVOLUTIONARY EPISTEMOLOGY TODAY:

CONVERGING VIEWS FROM PHILOSOPHY,
THE NATURAL AND THE SOCIAL SCIENCES
Werner Callebaut
Limburgs Universitair Centrum &
Rijksuniversiteit Limburg
Rik Pinxten
Rijksuniversiteit Gent
Evolutionary epistemology (EE henceforth) exists. Or
does it? Beginning with the speculations of many legitimate
or illegitimate founding fathers in
the nineteenth century,
an imposing array of studies somehow relevant to EE has

seen the light (see the comprehensive bibliographies in
Campbell, 1974a, and at the end of this volume). Yet on
closer inspection one gets the feeling that on the whole,
little has been accomplished to date in terms of dependable
theory or relevant application. As Ronald Giere neatly put
it at the conference out of which this volume grew, what
seems to be lacking most "is a number, or even one
really good Kuhnian exemplar, i.e. an important problem,
recognized as such,
solved by the clever application of EE;
an exemplar on which one could model further solutions of

important problems".
In this introductory chapter, some of the reasons for
this unsatisfactory state of affairs will be discussed (cf.
also the papers by Plotkin, Knorr Cetina and Apostel). But
our aims with this book go beyond the merely critical. In an
attempt to boost EE, we will confront
thematic and methodological ideas from EE as it is
usually
understood,
viz.
primarily
descriptive inves-
tigation of cognition (broadly

conceived)
drawing on
Darwinian concepts;
the suggestions for a dynamic and/or evolutionary theory
of
knowledge, including science, arising from current
transformations
in
the
natural
sciences
(faF-
from-equilibrium thermodynamics in particular);

- insights from the microsociology of science; and finally,
genetic
epistemology
in
the
Piagetian
W. Caiiebaul and R. PinXlen (edJ.). Rvo/ulwnary EpistemoloRY, 3-55.

1987 bv D. Reidel PublLlhinRCompany
tradition.
w. CALLEBAUT AND R. PINXTEN
It seems imperative to emphasize at the very outset

that EE as
the attempt to account for cognition as a
biological phenomenon
in
animals
and
humans "by a
straightforward extension of the biological
theory of
evolution to those aspects or traits of animals which are
the biological substrates of cogn1t1ve activity" (Bradie,
1986, p. 403) should be distinguished from EE as an
attempt to also model scientific change - a cultural
phenomenon on evolutionary-biological lines. As Campbell
(this volume) stresses, "(a)ll evolutionary epistemologists
find
biological
evolution
via
natural
selection
epistemologically
relevant",
but
a majority goes no
further and rejects,
or is non-commital on,
natural
selection analogues
in understanding
the
history of
science. Some of the implications of this distinction, to
which Vollmer's paper is wholly devoted, will occupy us
later on (see especially section 3.2.). Here we only
observe that both programs
are more or less equally
represented in this volume, and that several authors have
achieved
a
more
careful
and
more
balanced
appreciation of the dichotomy ("literal vs. metaphorical
extension of evolutionary biology") supposedly implied by
their existence than was customary until very recently.
Before delving into the heart of the matter, it will
be useful to put the current fascination for EE in proper
perspective. We contend that it cannot be dissociated from
a spreading conviction that the sciences themselves are
gradually becoming "evolutionary" in scope and method.
The structure of the paper is as follows:
In the first section, the call for an EE will be brought to
bear on (a growing awareness of) changes in the metaphysics
underlying any scientific endeavor; changes which are often,
but
not
necessarily
adequately,
summarized in the
catch-word "Mechanism is not enough". A full-fledged EE will
have to do more than coping with Quine's (philosophical)
challenge of a naturalized epistemology (Quine, 1969") and
(scientific)
bio-cognitive
program
implementing =the
suggested by Lorenz' and Campbell's identification of the
evolution of life with a knowledge process
already two
formidable tasks in themselves. In addition, we will argue,
EE must come to grips with the emergence of an "evolutionary
paradigm" in the physical sciences and in the sciences of
man.
EVOLUTIONARY EPISTEMOLOGY TODAY
Next (section 2) we will inquire into the concepts of

cognition/knowledge and evolution appropriate for a viable
EE. Here our analysis will depart from previous treatments
of EE (e.g. Campbell, 1974a; Vollmer, 1985; Bradie, 1986)
mainly in its
emphasis on the
necessity
to devise
evolutionary-epistemological equivalents of the biological
distinction between phylogeny and
ontogeny. It is chiefly
at this juncture, we maintain, that genetic epistemology in
the Piagetian tradition may become relevant to EE.
In section 3 we will scrutinize the applicability of
the evolutionary paradigm to knowledge in general and to
scientific knowledge in particular. Among other things it
will be argued,
pace Popper but on social-epistemological
lines, that a viable EE cannot possibly be an "epistemology
without a knowing subject" (nor, for that matter, an
"epistemology without
an object to
know").
Even if
Prigogine's views on far-from-equilibrium thermodynamics
offer no causal or explanatory principle rival to natural
selection, the "embedded" character of dissipative systems
(Prigogine, this volume), so we will maintain, may inspire
us in our search for an epistemology taking into account the
intervention of the subject in the object (cf. T. Nagel,
1986) .
In section 4 on the forces of evolution, the Darwinian
approach will be contrasted with Piaget's peculiar brand of

evolutionary theory and the prospect for EE of an amended
theory of evolution, not subject to the flaws of current
neo-Darwinism, will be briefly explored.
Finally,
in section
5,
we
will mention some
consequences for EE of the units of evolution controversy in
the philosophy of biology which usually remain unnoticed.
1. EMERGENCE OF THE EVOLUTIONARY PARADIGM
IN THE SCIENCES AND IN PHILOSOPHY
1.1. Evolutionary science and its metaphysics: beyond mecha-
nism?
tree," Bertold Brecht declared in his fragmentary Notizen

zur Philosophie, "knows man at least to the extent that it
knows carbon dioxide". And he went on: "To man, the use of
oxygen is part and parcel of his knowledge of the tree.
The notion of knowledge must therefore be enlarged." For
"A
W. CALLEBAUT AND R. PINXTEN
"life itself is a knowledge process. I recognize a tree

because I am living myself." (Brecht, 1929-1941/1967; p.
139; translation and italics ours) (1).
In a sense, the success of modern science, epitomized
by the
reversible
laws
of
classical
physics (see
Prigogine's paper), has been made possible only through a
repression of these basic facts: that the individual knowing
subject is a living organism, and that collectively, man is
an evolving biological and cultural species. In nature,
irreversible processes abound. Life is among them, and so is
cognition.
"In the very act of knowing," Rene Thom writes,
"something fundamentally irreversible is always involved.
It is only by discarding ['en gommant'] this irreversible
aspect that physicists have been able to say that phenomena
are subject to reversible laws". And he goes on to discuss
the apparent counterexample of celestial mechanics:
"If
celestial
mechanics gives rise
to a
phenomenology, it is because we see the planets
there, and we can see them
because they are
illuminated by the sun. Thus it is through the
coupling
of celestial
mechanics (which is
reversible in principle) with the fundamentally
irreversible phenomenon of solar radiation (the
transformation of the gravitational energy of
the sun into light energy) that we can see the
planets, and that, consequently, the reversible
phenomenon
becomes a phenomenology." (Thorn,
1983, pp.36-37; translation ours.)
Time and again,
the decline of the "Greek",
or "Galilean",
or "Newtonian", or "Cartesian" approach to science (2), and

the withering away of mechanism as a metaphysical paradigm
have been announced fiercely. Sometimes, the emergence of a
new, "evolutionary" paradigm is prophesied (e.g., Jantsch,
1980; Capra, 1982; Prigogine & Stengers, 1984; Depew & B.H.
Weber, 1985; R. Weber, 1986).
Organicism.
But what could such an alternative
outlook be
like? Very few candidates seem to exist to
date. More than a century ago, Charles Saunders Peirce
already intimated the possibility of an alternative to the
linear-hierarchical mode of
thinking exemplified by the
method accredited to Descartes (3) in his metaphor of the
chain and the cable:
"Philosophy
sciences in
ought
to
imitate
its methods,
so far
the
successful
as to proceed
only from tangible premises

which can be
subjected to careful scrutiny, and to trust
rather to the multitude and variety of its
arguments than
to the conclusiveness of any
one. Its reasoning should not
form a chain
which is no stronger than its weakest link, but
a cable whose fibers may be slender, provided
they are sufficiently numerous and intimately
connected." (Peirce 1868, quoted in Wimsatt,
1981a, pp.124-125.)
What the "cable" stands for here is, of course, first of all
an epistemological maxim ("robustness"
Wimsatt, 1981 a).
But underneath lies (lurks, some would say) what is usually
called the organicist world view. (For a recent application
in a critique of the "molecular Darwinism" of the hypercycle
theory of Eigen and Schuster (1977/1978),
see Wicken
(1985).) A version of it was at the center of Ludwig von
Bertalanffy's preoccupations when he set out to develop
systems theory (Gray & Rizzo, 1973). The persistent warnings
of mechanists in biology notwithstanding,
some brand of
organicism
is
usually
also
offered
as
panacea
for
the problems
of science and society
by ecologically
inspired thinkers such as Fritjof Capra (Capra, 1982) and by
other
latter-day holists.
And finally,
in a rather
intriguing but illuminating reversal of the roles that are
traditionally assigned to mechanistic vis-a-vis organicist
thinking in the various sciences,
a form
of anti-mechanism
also
pops
up
in
Margaret
Boden's
plea
for
an
anti-reductionist reading of artificial intelligence, when
she endorses a "computational approach to life and mind
( ... ) entirely compatible with notions of human freedom"
and opposes it to "the mechanization of the world-picture
brought about by the
natural sciences", in which she
includes the life sciences (Boden, 1984, p. 317) (4).
Mechanism, many scientists and philosophers agree
nowadays, is not enough (5). But, we are inclined to ask, do
we really have good reasons to cast our fate to an
organicist alternative which on closer inspection turns out
to be defined in an essentially negative way (holism vs.
reductionism,
synthetic
vs. algorithms, etc.)?
VS.
analytical
method, heuristics

After
(vitalistic
all,
hasn't this been
biology,
psychological
interactionism,
organicist
conceptions
tried out before

dualisms such as
of society, etc.),
with dubious results to say the least? The conciliatory

view ("Mechanism and organicism are compatible, for both
metaphysical world views have their own
distinct ways
leading to reliable knowledge and their own areas of
rewarding application") is sometimes justified by appealing
to a philosophically questionable generalization of quantummechanical complementarity to all spheres of scientific
knowledge (see already Lindenberg & Oppenheim, 1974). To our
mind, however, this position will beg the question as long
as we do not know for certain what complementarity (proper
or generalized) implies, epistemologically speaking, and as
long as we have not grasped the deeper meaning (provided
there
is
any)
of
"holism",
"organicism",
etc.
(cf.
Kaufmann, 1974). Elsewhere, one of us (Callebaut,

1983,
ch. 7) has argued that the reductionism-vs. -" .. " debate in
analytical philosophy of science, past and current, is
heavily biased toward the first alternative, as positive
characterizations of concepts essential to any alternative
position,
such as "emergence",
science", are mostly wanting.
or Itautonomy of a domain of
Methodological mechanism.
In terms of the Hegelian
concept of Aufhebung, the proper alternative to mechanism
ought to point to something transcending its mere negation.
After all, Newtonian
science can boast of quite some
achievements! Where shall we look for such an animal? Robert
Brandon, elaborating ideas previously defended by Marjorie
Grene (Grene, 1974) characterizes mechanism as a methodological (as contradistinguished from an ontological) thesis,
viz. as the search for mechanisms which in fact explain how
the phenomena under investigation are produced. Invoking one
or several mechanisms he takes to be necessary for any
science to be able to model the processes it wants to
investigate:
"To model a process is to offer some more or
less
plausible
hypothesis concerning the
mechanism underlying the process.
Thus any
process
capable
of being
modelled
is a
mechanistic process." (Brandon, 1985, p. 346.)
Mechanisms may thus consist of springs and gears, but also
of, say, "small peripheral populations and geographic isolating barriers" in the case of evolutionary biology (ibid.).
Most
importantly,
mechanistic
explanations
do
not
necessarily explain wholes in terms of their parts; thay may
also be "given in terms of entities more inclusive
than the
explanandum entities" (p. 347). Thus mechanism is not to be

conflated with reductionism.
The potential advantages of this unusual construal of
mechanism seem rather obvious. (For a forceful statement of
the familiar view that mechanism entails reductionism, see
in particular Plotkin's paper in this volume.) It would
allow one to remain a methodologically impeccable scientist
or naturalized
philosopher (being
a mechanist) while
embracing a multi-level ontology whose
attractivity it
would be difficult to contest (cf. infra). It would even
make "empirically based forms of holism" (p.348) look
acceptable (6). Thus it looks like an elegant solution to
our problem.
Yet it is unclear, at least to the present authors,
whether Brandon's very liberal definition of mechanistic
processes as
"any process(es) capable of being modelled"
will do the job. It only seems to push the problem one step
further, since we do not possess a satisfactory theory
(model?) of what models are or should be. Considering their
manifold uses and functions (see most notably Apostel,
1961), it would prima facie seem implausible that all sorts
of models could reduce to one basic sort. Would this leave
us with various brands of mechanism, then? Even if these
epistemological issues were resolved, the average scientist
would presumably continue to go for the "point-at-able
entity or set of entities whose behavior is highly predictable" (Plotkin, this volume), which, psychologically,
seem to be the hallmark of the mechanism of "real" modern
science. (Thus, Campbell, 1987, self-consciously describes
his EE as a "general program to mechanize and physicalize
all aspects of believing and knowing".) Is the presence of
this "point-at-able" quality also guaranteed in the case of
explanations of parts in terms of wholes? As long as we do
not know this to be the case with certainty, the question
which prompted this digression
"What is wrong with
mechanism?" -, we take it, will remain actual.
But even if anti-mechanism were a necessary ingredient
of the evolutionary approach, it would not be sufficient to
capture the latter adequately. Let us therefore look at
another feature of evolutionary explanations now.
10
1.2. Scientific explanation and historicity

Only in the nineteenth century was time really discovered,
to borrow Toulmin and Goodfield's apt phrase
(Toulmin &
Goodfield, 1965). It is but natural to try to locate the
evolutionary character of the physical, the biological and
the sociocultural sciences in the historical character of
the processes they investigate.
Organized complexity and predictability. - Yet again
it is our impression that we do not really know, in 1987,
what the "emergence of the evolutionary paradigm", now
defined in terms of features of the study of the trajectories of systems, exactly amounts to. At which level(s)
of analysis are its distinct characteristics to be found?
Prigogine's school, whose views are now adopted by many
others, locates them in the essential historicity of the
description of the regimes of dissipative systems. On this
analysis (see especially the remarks on systems with fractal
at tractors in Prigogine's paper), the evolutionary paradigm seems to be related primarily to the possibility of
dealing scientifically with the problems of organized complexity (Warren Weaver), i.e. problems concerning systems in
which the interrelations of the numerous (micro-)elements
can condition the system's (macro-) behavior in decisive
ways (cf. Simon, 1969; Wicken, 1984), making the latter
(much) less predictable. This is a property not found in
simple
chaotic
Newtonian
systems
systems studied in
equilibrium
thermodynamics,
nor,
for
that
matter,
classical equilibrium
whose
behavior,
at
in the
or nearcertain
levels of description, is quite regular, and therefore often

predictable (Wimsatt, 1986a).
As usually understood, an evolutionary theory is
historical in the sense that knowledge of the past is
necessary to predict the future; a view expressed by, e.g.,
Gustav Bergmann and David Hull (Levine & Sober, 1985, p.
305). Jon Elster thus distinguishes between ontological
and epistemological (social) "hysteresis" (Elster, 1976;
1983, ch. 1): Whereas it cannot be the case that ontically,
the past impinges on the future beyond the traces it has
left in the present, scientific explanations may well have
to rely on knowledge of the past, namely whenever adequate
and sufficient knowledge of the (traces of the past left in
the) present is lacking. Note that the standard models of
neo-Darwinian evolutionary theory do
not satisfy this
epistemological condition of historicity: With only a few
II
exceptions, in these models the past impinges on the future

only insofar as the past has ~ffected the present. In a deep
sense, therefore, the Modern Synthesis must be considered as
"Newtonian" (Sober, 1984, ch. 1; Kitcher, 1985, pp. 43-50)
rather than "evolutionary" (cL Depew, 1986). On the other
hand,
when Prigogine
(this
volume)
emphasizes that
dissipative systems, as opposed to the systems studied by
classical
mechanics,
"may
forget
perturbations",
he
is
obviously making an ontological claim.

Beyond the covering-law view of explanation: but how?Are evolutionary explanations simply weaker than the deterministic (respectively statistical) explanations which the
logical-empiricists'
model
of
deductive-nomological
(respectively inductive-statistical) explanation patterns
unsuccessfully tried to capture, in that their predictive
power is limited to trivialities? (See Thom's paper on the
"necessity of incorporating time in individual beings" and
on the "residuum of indetermination" in
any evolutionary
theory.) In light of the work of Mary Williams (M. Williams,
1973, 1982) and others on the structure of evolutionary
explanation and prediction (see the review in Thompson,
1986), such a view seems no longer justified (cf. Bunge,
1978). It would be still more tributary to an old-fashioned
instrumentalist
view of explanation
than most postpositivist philosophers of science (or many scientists for
that matter) are now willing to concede. Alternatives such
as a realist interpretation have been proposed (e.g. Jensen
& Harre, 1981; Bunge, 1983), but they cannot on the whole be
considered as convincing either (Callebaut, 1988).
Is it a "built-in
temporal
asymmetry"
which
makes a theory evolutionary (Levine & Sober, 1985)? If so,
evolutionary biology (endorsing Dollo's law: "evolution is
irreversible") and the "new physics" of dissipative systems
might turn out to be not so different after all (Prigogine &
Stengers, 1979, 1984; Brooks & Wiley, 1986; Depew, 1986)
(but see section 3.1.). Some of the spirit of "real
evolutionary thinking" (see Mayr, 1982, for a substantiation
of
this phrase)
is certainly
captured
in
Stephen
Toulmin's useful distinction between ("bad") evolutionist
and ("good") evolutionary thinking (Toulmin, 1972), which
Van Parijs (1981) has reminded us of.
Whereas an evolutionary perspective is centered around the
underlying
mechanism(s), an evolutionist perspective emphasizes the
overall direction of change:
12
"An
evolutionist
consists
in
development,
looking
as
perspective
at
progress,
history
as a
essentially
( ... )
as
succession of
stages of increasing complexity or perfection.

Its explanatory claims are often restricted to
spelling out a logic of development ( ... ). An
evolutionary perspective, on
the other hand,
focuses on ( ... ) mechanisms of selection between
actual (as opposed to potential) alternatives."
(Van Parijs, 1981, pp. 51-52.)
The "natural selection" mold is but one of the conceivable,
possible
and
even existing patterns
of evolutionary
explanation,
reinforcement ("the direct selection of the
features
to
be
explained
within the
entity they
characterize") being another, and
important, one (Van
Parijs, 1981 and this volume).
Variational explanation.
What, then, does your
archetypical evolutionary explanation, the Mother Scheme
from which all particular evolutionary explanation patterns
in some sense derive, or ought to derive, look like? Following Lewontin (1983), Elliott Sober has proposed to oppose
variational explanations to developmental explanations. The
latter lay down a sequence of stages through which life
forms, being preprogrammed, are constrained to pass (cf. Van
Parijs'
"evolutionist" explanation); Levins & Lewontin
(1985, p. 86) call this "transformational theory", e.g.
Lamarck's theory of evolution or Freud's and Piaget's
theories of psychic development "derived from theories
of
embryological development of the nineteenth
century". In
contrast, variational explanations crucially hinge on the
fact that the individual organisms
composing, say, the
population
under
study,
vary. Whereas for
Lamarck,
population change was a consequence of
individual change,
"Darwin took his question to have
an
irreducibly
population-level character":
population change
is the
consequence of individual stasis plus individual selection
(Sober, 1984, p.150). Selection in itself does not imply
evolution, even when it is the only evolutionary
force at
work;; stability (heritability) assumptions are also required
(see sections 2.1. and 4). Sober convincingly argues that
the variational paradigm is compatible with
the idea of
endogenous constraints on evolution, which we will consider
in section 4.1. (8). Natural selection is the "obvious
prototype" of variational explanation; but
according to
Sober,
"(t)he grip of the
variational
paradigm
on
EVOLUTIONARY EPISTEMOWGY TODAY
13
evolutionary thinking goes deeper than the Darwinian commitment to the historical hypothesis that natural selection is
the preeminent force of evolution" (p. 153). Thus drift
could be thought of as a process of "random selection"(!):
"Sampling error
may transform a population without any of
the organisms in it changing at all. ( ... ) When drift
modifies the composition of the population in this way, it
is not because the individual organisms change but because
they vary" (ibid. ) .
Unfortunately,
apart from
this
one
additional
example, Sober's claim that (all?) "other evolutionary
forces are
conceptualized
answering
"What-for?"-
in
the
same
way"
(i.e.
in
terms of variational
explanation) remains entirely
programmatic. We insist on making this point because a case
can
be made for
distinguishing between
evolutionary
explanations of a functional or teleological
kind
questions
and
evolutionary
explanations
that do not involve function or teleology.
The theory of evolution by natural
selection clearly
belongs to the first category, the essence of which can be
rendered as follows:
"Put
cryptically,
trait A's
existence is
explained in terms of what A does. More fully,
A's existence is explained in terms of effects
of past instances of A; but not just any
effects: we cite only those effects relevant to
the adaptedness of possessors of A." (Brandon,
1981, p. 103.)
On the other hand, it is also clear that "(t)here is nothing
teleological about the theory of evolution by random drift
or theories of speciation" (ibid.). Hull, we take it, would
agree: Biologists such as Eldredge and Gould (1972) are
claiming that
species
are
static
systems, possibly
homeostatic systems.
However,
"no mechanism has
been sug-
gested for the production of

such homeostasis. Thus,
functional explanations of species development
are as
questionable as traditional functional
explanations in
sociology" (Hull, 1982, p. 314) (7).
But maybe we are just haunted by a chimera here: Maybe
all grand explanation schemes such as those proposed by
Hempel, Oppenheim and other positivists are bound to fail,
including Sober's candidate. One could then be compelled
to conclude, for instance, that "real-life" explanations can
at best capture certain causal mechanisms, which are always
as
so meticulously
and necessarily context-dependent,
14
documented by Lewontin (1974), Wimsatt (1976), Cartwright

(1983)
and
others
in
the
last
decade
or
so.
(Unfortunately, the issue of causality cannot be pursued
here; but see Thorn's paper.) Even someone not prepared to
go so far would probably agree that the neo-Darwinian theory
of evolution
is
not monolithic,
in terms
of
explanatory
mode, but "is best thought of as a supertheory or collection

of theories". (Brandon, 1981, p. 103).
Teleology: so what? - To round off our discussion of
evolutionary explanation, a few words on the implica tions
A decade ago,
of using teleological language are in order.
Steve Gould observed that
"It is still
unfashionable,
in biological
circles,
to
use
such
words
as 'design',
'purpose', or 'teleology'. Since final cause is

so indispensable a concept in the elucidation of
adaptation,
produce
conscious
and
since
a well-designed
natural
intervention
selection
structure
of
can
without any
God's super-human
wisdom or the sub-human intelligence of the

animal in question, one would think that these
terms would again be admitted into orthodoxy.
Evidently, however, in our choice of words, we
are still fighting the battle
with theologians
that we won in deeds almost a century ago."
(Gould, 1976, p. 97n.)
Considering the recent revival of Creationism, Gould's last
sentence may have been overly optimistic. But our concern
here will rather be with the use of teleological language in
EE.
The overwhelming majority of evolutionary biologists
seem
to agree that "each adaptation
organization that is relative to

environment", and that the relation
and the environment is Ilene of

not
requiring a contracausal
constitutes a
form of
some aspect
of the
between the adaptation
goal-or end-directedness tt ,
or finalistic framework
(Plotkin, 1982, pp. 4-5, and references therein). Even
those few philosophers who take the teleology of organisms
to be "intrinsic"
in the sense of irreducible to a
non-teleological, yet explanatorily adequate account, as a
reductionist such as Ernest Nagel (E. Nagel, 1977) would
rather have it - are careful enough to disentangle organic
teleology both
from
intentional
teleology (involving
15
mentalistic purpose) and from the (non-intrinsic) teleology

of artifacts. Or at least they claim to be able to do so
(Jacobs, 1986; cf. Woodfield, 1976).
Likewise, what most,
if
not
all
evolutionary
epistemologists aim at is, as Campbell
puts
it,
"the
explanation of order and the explanation of
fit (of one
system to another) that is achieved without
external
guidance". Just as Darwin "increased the evidence of design,
but
undid
the
argument
to
designer",
the evolutionary
(of the second kind; cf. supra) trying to

understand scientific change - to stick to this one example
ep~stemologist
does "not
need
progress
to
a known-in-advance
goa111,
which is
what e.g. Kuhn (1970, pp.171-173) was rejecting.
Rather,
"( ... ) we can have the niche-filling model in
which - however meandering science is -, if the
nature of the physical world is involved
in one
of the the many feedback processes only one

you can have local niche-filling, belief-environment matching, which does not need to involve
So
this progress to a goal known in advance, but

does explain why a local specialty in science
has competent
beliefs
about
its
special
referents". (Campbell) (9).
far so
good.
Cyberneticians
and
systems
theorists-
whether they accept neo-Darwinian biology or not - have long

taken for granted that non-teleological descriptions of
teleological achievements are
feasible.
Conversely, we should add, it is also trivially
possible to
translate
non-teleological
(say, causal)
descriptions into teleological ones (see e.g. Mesarovic,
1964; Samuelson, 1975; or Sachsse, 1979). However, the
trouble with both kinds of translation is that we want
explanation, not mere description. Sachsse (cf. already
Craik, 1943) points out that a deterministic ontology seems
to imply, on the one hand, that causality and finality
mutually entail one another (every initial state inevitably
leads to one final state), but on the other, that teleology
is impossible (no final state being reachable from different
initial states). One might be tempted to accept a definition
of teleology in terms of the production of negentropy, where
this is not taken to imply a decrease of entropy (Buchel,
1982), to avoid Sachsse's paradox; the more
so when
considering that most extant philosophical
analyses of
teleology (which are usually in terms of the prevalence of
16
preferred states, closed feedback loops, and programs

Hull, 1974", p.l03} are not entirely satisfactory from a
strong reductionist's point of view (Rosenberg, 1985, pp.
52-68) .
Be that as it may; it looks as if teleological
explanation "is unavoidable in biology for contingent and
nonconceptual reasons", namely because "(t)he
world
is
just
much
more
complicated
than
provincialists [=
anti-teleological reductivists - w.e. & R.P.] have allowed"
(Rosenberg, 1985, p. 65). But accepting this verdict is a
far cry from suggesting that the issue of teleology can be
resolved, not by trying to get rid of it but by tackling it
the other way around, as Ho and Saunders (1984,
p.6) do
when they write that "Ultimately, mechanism and teleology
may be one and the same, being part and parcel of the
properties of matter of which living things are composed".
On a higher plane, a second autonomy issue can also be
raised. In a paper we already mentioned, Margaret Boden has

argued for the autonomy of AI and has justified the use of
computational concepts providing explanatory power "over and

above that of the more basic theories in the life sciences,
while being entirely compatible with them" by pointing to

the
absence
of the
concepts of representation and
intentionality - central to AI - from the vocabulary of the
life sciences (Boden, 1984, p.317). Thus she perceives a
gulf between the computational approach
compatible with
"human freedom" - and the mechanism of the natural sciences.
How does EE relate to this?

The issue of determinism and its relation
to free
will need not concern us here (10). Regarding the absence of
the concept of representation from biological vocabulary,
we take it that one of the things that motivate at
least
some biological evolutionary epistemologists (minimally, the
"cognitivists") is precisely the ambition to develop such
notions,
or something more or less
equivalent
to
them.
Likewise, intentionality is a hotly

debated topic in,
say,
cognitive
ethology.
Heyes' paper (this volume)
contains an incisive discussion of both issues at the level
of animal studies. As to the role of intentionality in
science, see Hull (1982) and Callebaut (1984).
17
1.3. The symbiosis of science and EE

Some of the contributions to this volume (most notably
those by Prigogine, Thorn, and Van Parijs) directly or
indirectly address one or more of the questions related to
evolutionary explanation or its ontology. The time probably
is not ripe yet for definitive answers. But this should not
prevent us from already appreciating the way in which EE is
likely to interact with the sciences which are "turning
evolutionary"
(cf.
note 5). As Griesemer (1984) has
observed, quite a few ideas presented under the label "EE"
are metaphysical rather than epistemological in nature. The
"German-Austrian branch" of EE is most outspoken in viewing
the latter
as an endeavor signalizing a return, on a higher
intellectual plane, to a situation which existed in

pre-Kant ian or even pre-Cartesian philosophy, before the
knowing subject had superseded metaphysical considerations
as a "first philosophy" (e.g. Vollmer, 1985, pp. 320-322;
but also Shimony, 1970). In Bernhard Irrgang's words: "EE is
prepared to bring ancient and modern First Philosophy into
line with one another by means of the idea of a "Hypothetical Realism" and to solve their open problems in this
way." (Irrgang, 1986, p.104; translation ours.)
with
some
notable
exceptions
such
as
the
epistemological views of Descartes and Leibniz (who were
also first-rate scientists, at least as mathematicians) or
Kant (who in a way anticipated physical field theory), the
epistemologies of the past tended to reflect scientific
achievements post factum. This was most certainly true, of
course, of analytical philosophy of science in pre-Kuhnian
days (11). If EE is really what it claims to be, namely
the product of two converging intellectual
advances
the "biologizing" of epistemology and the "epistemologizing"
of biology - (cf. Plotkin, 1982; Bradie, 1986), then the
prospect of a real symbiosis of science and philosophy may
be near. Apart from the broader cultural implications of
such a "new alliance", there is a message for
epistemology proper here. In the past,
philosophies of science
have often been unconsciously
moulded
after their
"idols",
scientific
or
other.
(The
subject
is
always
infected by its object, of which it unwittingly assimilates

certain characteristics - Vandenbrande, 1979.) In a situation of symbiosis, which is potentially more symmetrical,
the two intellectual communities and cultures - the scientific and the
"meta-scientific" culture - would be able, so
18
we surmise, to more consciously harmonize and thus to better
take advantage of one another's "nested hierarchies of

selectors", which are primarily social (cf.
Campbell,
1979; Boon, this volume) and which harbor the mechanisms
warranting intellectual progress.
But for the time being, let us keep our feet planted
firm on the ground and assess what EE looks like at present.
2. THE RELATIONSHIP BETWEEN COGNITIVE EVOLUTION

AND COGNITIVE DEVELOPMENT
2.1. Evolution and development
All evolutionary epistemologists seem to agree with the
general statement that the phenomenon of knowledge is a
product of evolution. Most will agree with the position that
evolution
itself
is
continuous
"knowledge
process"
(Campbell, 1974a). Agreeing on these statements, however,

does not get us very far, since the terms in them remain
ambiguous: What do we mean by "evolution" and by "knowledge"
in EE? Even a short survey of the field reveals that there
is little or no agreement on their meanings. Let us try to
clarify
this point.
Evolution by natural selection generalized.
On
Richard Lewontin's widely accepted analysis, evolution by
natural selection can be elucidated by means of a scheme
involving three essential principles:
(i)
Phenotypic variation:
different individuals in a
population
have different morphologies, physiologies, and
behaviors;
(ii)
have
differential fitness: different (phenotypic) organisms

different rates of survival and
reproduction in
different environments; and
(iii) heritability of fitness: there is a correlation

between parents and offspring in the contribution of each to
future generations (cf. Lewontin, 1970, p. 1). Lewontin
took these general and purely formal principles to be
necessary and also sufficient for evolution to occur. It
has been argued subsequently that it is a necessary but not
a sufficient set of conditions for a type of entity to act
as a unit of selection (Wimsatt, 1981b;
Lloyd, 1986),
which should be supplemented with appropriate definitions
of units of selection (Brandon & Burian, 1984;
Sober,
19
1984),
But
as well as of levels of
such
worries
underdeveloped
need
not
stage of EE.
Lewontin's account does
not,
selection (Brandon,
concern
There
is
an
rations
to
the
mere
in
1982).
the present,
Most important is to
see that
and need not invoke anything
related to the material structures

lutionary processes.
increasing
Waddingtonian
biologists
approach to evolution,
us
actually realizing evo-
awareness
among
post-
that a
"genetic bookkeeping"
reducing all phenotypic conside-
"statistical
abstraction" of mean
phenotypic fitness (G.C. Williams, 1966, p.33; Dawkins,

1976, 1982) is much too narrow; see Wimsatt (1981b), Arthur
(1984), Sober (1984), and Bechtel (1976, part III: "Incorporating
development
biology
into
the
evolutionary
synthesis") for critical appraisals.
It is in the vein of such criticism that David Hull,
in line with Lewontin's general analysis of evolution, has
proposed the concept of an interactor, i.e. "any entity
which directly interacts as a cohesive whole with its
environment in such a way as to make replication differential", to complement Dawkins' genic-reductionistic concept of a replicator, i.e. "any entity which passes on its
structure largely intact through successive replications"
(cf. Plotkin's paper in this volume).

Elsewhere, one of us (Callebaut, 1986) has assessed
the potential of this generalized, "cybernetic" conceptual
machinery for EE. Here we want to stress only that in the
light of such an analysis, development should never be taken
to
be
synonymous to
evolution,
as
is
sometimes done in
discussions of EE (e.g. Thagard, 1980). It also follows that

it is misleading to say, for instance, that the "proper"
level of evolutionary processes of variation and selection
is that of the alleles, or to suggest that the use of
evolutionary concepts at other levels of organization (e.g.
the genome, certain social groups or conceptual systems,
etc.) is improper (e.g. von Schilcher & Tennant, 1984,
esp. p. 26; cf. Bunge, 1983, p. 58). We will return to this
issue in section 4.
On the other hand, it is true that biological evolution is "like all historical processes in not being fully
determined by law. Those laws that hold are statistical in
nature" (von Schilcher & Tennant, 1984, p.74). Thus probability (weak causation), an irreversibility (Dalla's law,
justifiable only by means of a statistical argument) that
does not preclude openness (Lewontin, 1978), and maybe even
20
"creativity" (Mayr, 1982, p.591; Popper, 1982, p.174),

provided it can be defined naturalistically, could be constituent meanings of evolution.
Ontogeny, phylogeny, and the Kantian aprioris. - By
means of blind
an
increasing
or " ... "
variation and selective retention,
wel1-adaptedness
to
the
environment
is
achieved. It is by the latter characteristic that evolution

can itself be recognized as a knowledge process (e.g. Plotkin, 1982). (Cf. Thorn, 1975, p.359, who dates back model use
to the earliest prehistory, with roots "au tnHonds de
l'organisation biologique elle-meme" (12).)
However, students of knowledge processes have differed
in their basic definitions. We here confront the first
opposition between the Piagetian and the EE perspective on
knowledge. Piaget's focus (e.g. in his 1982) is on the
ontogenetic cognitive process coming about through maturation and gradual development of concepts and thought
procedures in the course of a lifetime. On the other hand,
Lorenz (1973), Campbell (1974a) and many
others have
defined knowledge as a phylogenetic process, thus focusing
on evolution and not on development. If we are profitably to
use Piaget's approach for the further elaboration of an EE,
we have a problem here.
Especially Lorenz (1962) and some other Austrian and
German scholars (e.g. Riedl, 1984a; Kaspar, 1984) have been
working on an issue which will come to bear, we conjecture,
on the said opposition. In these works the status of the
Kantian a prioris is discussed. In the evolutionary model
one cannot,
of course,
set apart the most basic notions of
knowledge (Kant's a prioris) as innately given and not

themselves subject to evolution. On the other hand, these a
prioris obviously have a stable and organizing status in our
system of knowledge. They contain very basic insights into
the structure of the
sort of
environment (Vollmer's
"mesocosm" - Vollmer, 1983, 1985) in which knowledge proved
to be a good adaptation (Lorenz, 1973; Campbell, 1974a).
Lorenz's solution,
which echoes a view one already finds in
Herbert Spencer,
is to reduce the status of Kant's a
prioris to that of ontogenetic a prioris, themselves the
products of phylogenetic evolution. As Wuketits captures it
in his "second postulate" of EE:
"Innate dispositions are the outcome of natural
selection;
they are the products of selective
mechanisms, which, among all 'innate products',
21
favor and stabilize the one which best copes

with the conditions of living and surviving."
(Wuketits, 1984c, p. 6.)
An important effect of this reinterpretation of Kant's a
prioris is that a fallibilistic pos1t10n, at the most
fundamental
epistemic level, becomes possible (Lorenz's
(1973) learning theories; Campbell's trial and error already
in 1960), in line with
Quine's attractive revisability
thesis (see Campbell's paper and especially the references
to Sober therein). This is one of the reasons why we object
to Wuketits' move which consists in "immunizing" Kant's (and
Kantian)
epistemology
and epistemological
metaphysics
against EE by interpreting the former as prescriptive and
the latter as merely descriptive (Wuketits, 1986, p.200).
Thus the fertile dialectic of an EE both presumptively
descriptive of knowers (and the world to be known) and
hypothetically normative, as called for by Campbell (e.g.
1987), would never get a chance to evolve (13).
A multiple-level model of evolution. - The work of
Plotkin and Odling-Smee seems to offer further material
enabling genuinely to understand learning in an evolutionary
perspective
(cf.
Boyd
& Richerson, 1985). Individual and
cultural learning are conceived here

as two distinct
knowledge- or information-gaining processes in the context
of
a model of
organisms adapting
central nervous system
of
to
an
the individual
environment. The
organism
is the
storage site in terms of which the level of individual

learning is defined. The storage site for the next-higher
level of cultural learning is

population sharing a non-genetic
the "culture pool" of a

channel of communication
(Plotkin & Odling-Smee, 1982, p. 45).

What we have here looks like a subtle metaphorization
of the evolutionary program: (i) The a prior is of knowledge
are seen as
phylogenetic products, and (ii) the new
adaptations by learning at the level of the individual or
the group (somehow parallel to
the genetic level of
biological adaption) are checked and retained selectively in
accordance with lower-level evolution.
However, the fact that learning is conceived
here in
a "nested hierarchy" of knowledge-gaining processes (Plotkin

& Odling-Smee, 1982) clearly goes beyond the traditional
evolutionary frame. In their (1979) paper the same authors
presented the theoretical foundation
Phylogenesis is the
"primary"
information-gaining process
in a biological
22
system. "Variable epigenesis" (cf. Waddington)
be a second
such
process.
they take to
Learning processes - individual
and cultural - presuppose both these levels.

There is a hierarchy, such that no information of the
third or fourth level can contradict that of the first level
over generations. Thus the operation of Campbell's (1974c)
"downward causation" is confined. Moreover, there is a clear
subdivision of
functions: "learning provides the fine
behavioral tuning that cannot be supplied by either primary
or secundary referents acting alone" (Plotkin & Odling-Smee,
1979, p. 28).
We conjecture that this integration of l:arning in

the global framework of variation and selection 1S a good
working hypothesis for the evolutionary epistemologist.
Integrating genetic epistemology. - The contradiction
between
EE
and
Piaget's
genetic
or
developmental
epistemology
(cf.
section
1.2.)
could
consequently
disappear, provided we interpret Piaget as speaking about
the logic of "finely tuning" evolutionary products. We
should emphasize that this is not to say that all knowledge
should be viewed as the outcome of either a phylogenetic or
an ontogenetic process. Such a dichotomy would be "merely
the old nature-nurture issue in a different guise" (Plotkin,
this volume). To the contrary, in Plotkin's model of

evolution, knowledge is multi-layered and has multiple
origins:
"knowledge is (1) a web of relationships, which

(2) find partial expression in genotypic and
phenotypic structure ( ... ), and which (3) is the
outcome of the
operation of
a number of
processes, these processes themselves represent-
ing
knowledge in the form of relationships."
(ibid. )
As evolutionary epistemologists, we will have to study and
integrate Piaget's model at this level in a more systematic
way than was done in the past. Leo Apostel, a long-time
student of Piaget, offers the following reasons for taking
genetic epistemology much more seriously than most advocates
of EE usually do:
"the Piagetian brand of EE has the unique
distinction: a)
of being based on an adequate
field of observation and experiment (as is also
the case
for Lorenz's EE);
b) of
being
expressed by means of algebraic tools which
attempt to capture development as such (and here
23
only Waddington's EE approaches this degree of

precision); c) of using an explicit theory of
evolution of a particular brand, quite explicitly defended (even if it is doubtful); d) of
being applied systematically to a great number
of fundamental concepts in the natural sciences
(the biological, social
and human sciences
being still mainly absent from
the Geneva
experimental scene)." (Apostel, 1980, p. 129.)
In order to convince the sceptic,
Apostel's partisan claims
ought to be substantiated (see Gillieron's

paper and
Apostel's reply). In the light of detailed studies of
Piaget's views on biology (Piaget, 1967, 1974, 1976) such as
Sophie Haroutunian's (Haroutunian, 1983), we may safely
conclude that Piaget's "third way" beyond Darwinism and
Lamarckism, to the extent that this position has ever been
articulated unambiguously, is highly problematic (to say the
least). But this verdict does not automatically render
suspect genetic epistemology
as
such,
as authors
such as
Toulmin (1972) have sometimes implied. The contributions in

part III take up the challenge of its confrontation with EE
for the first time (cf. also Plotkin and Vollmer). It will
be
seen that starting a dialogue between EE and genetic
epistemology
(to which evolutionary epistemologists often
pay lip service) turns out to be rather more difficult than
one would expect.
2.2. Genotypes and phenotypes; knowledge and action
A problem closely related to that of the distinction
between
evolution and development is the problem of
defining evolutionary-epistemological equivalents of the
original
biological
concepts of a
"genotype" and a
"phenotype".
If genotypic
and phenotypic levels of
cognitive change can be clearly delineated, then it becomes
possible to assign "cognitive evolution" and "cognitive
development" to their proper places.
The trouble is that evolutionary epistemologists seem
to have thought little about this topic, for whatever
reasons (Wimsatt,
1981b,
167-173;
Bradie, 1986, pp.
411-413); and if they have done so, they have not reached
substantive conclusions, so it seems. However, we think this
situation is going to change soon (see Callebaut, 1986, for
a more detailed account).
24
Knowledge and action.

the way in which a similar
Let us first take a look at

problem has been tackled in
another domain,
that of evolutionary economics. According
to Kenneth Boulding,
one of the basic differences between
organic and "social" (we would rather say: sociocultural)

evolution has to do with the fact that
"( ... ) the genetic information which produces
biological artifacts
is
contained
in the
organisms themselves.
The
genetic information
which produces human artifacts is contained in

human beings, human organizations, and material
artifacts which are different from the ones
produced. With the advent of the human race
something that might almost be called 'super
sex' came into evolution, whereas biological
evolution never got beyond
two sexes. The
production of human artifacts is 'multiparental', in the sense that the genetic information
which
organizes
the
production
artifacts is contained not just in
artifacts,
but in
artifacts of great
themselves,
of human
two other
very
large
numbers of
variety
human
beings
blueprints,
libraries,
computers,
and so on. This undoubtedly is the main reason

why
the
development
of
human
artifacts
enormously speeded up the pace of evolution,
just as the development
of biological sex
speeded it up, simply because of the enormous
increased
potential not only for change and
variety
but
also for
the
production of
artifacts." (Boulding, 1981, p. 16.)
As an economist, Boulding takes (the products of) human
action as the focus of his attention;
they are his
phenotypes. We think EE would benefit from the adoption of
this idea. For not only is cognition a kind of internal
action (as Piaget has always insisted); but, the proof of
the pudding being in its eating, the ultimate test of the
reliability, nay of the adequacy of our ideas can only be
their application in our behavior or action. It has always
struck us as rather odd that evolutionary epistemologists,
who treat hierarchies of selection levels at great length
(e.g. Boon, 1983; Campbell; Plotkin and Odling-Smee), spend
so little attention on the relation between knowing and
acting. (For a contrasting approach, see the view of an
anthropologist such as
Caporael (this volume) or Irons
25
(1979,
p.9),
who writes that
"(i)deas, beliefs and
sentiments, from the point of view of behavioral biology,
are important only to the extent that they influence
behavior".) After all, even the most rarefied forms of
adaptation remain subject in principle to the inexorable
test of selection. That is the kernel of the very idea of a
"nested hierarchy", albeit that in most cases - and happily
so - it will be our ideas that "die in our stead" (Popper).
This issue should be disentangled from the rather
different
questions
concerning
the
realist vs.
instrumentalist (etc.) character of scientific knowledge
and the nature of the "non-ultimate" selection criteria for
scientific belief and belief change.
Contrary to what many
evolutionary epistemologists would maintain, EE does not
"prove" the truth of scientific realism as it has come to be
defined
of late (van
Fraassen,
1980;
Churchland &
Hooker, 1985; Paller & Campbell, 1987). For a defense of the
mildly agnostic view that "given the natural possibility of
alternative life-styles, needs, capacities and cognitive
structures it makes
no sense to
identify our ideal
scientific model of reality with the ultimate nature of the
world-in-itself" (Clark,
1986, p.1S8), see Clark (1983,
1986).
But let
us return to Boulding's
proposal. His
definition of a
sphere of human
tl
gene pool" of human artifacts ("the whole

knowledge in brains, books, and computers
spread over the face of the earth") is certainly too vague

to allow us to evaluate unambiguously its deficiencies and
potential merits. Nor is it very clear how, on Boulding's
model, genotypes become phenotypically expressed.
Finally, his model of sociocultural evolution (cf.
Boyd & Richerson, 1985 on the variable number of "cultural
parents ll ,
and
Campbell,
this
volume,
on "selective
cross-borrowing")
lends
itself
to
an interpretation
altogether different from his (in terms of "super sex") as
well. Thus Hirshleifer (1977, p. 14) was able to maintain
with equal justification that in the sphere of economics,
"no
genetic recombination
himself
offers
pretation.
More
no
cultural evolution,
could
account for
good
is
involved"
reasons to
importantly,
he fails to
the
fact
at
like
so
many
specify the
that
all. Boulding
prefer his own inter-
we
seem
writers
on
force(s) that
to
observe a
continuous speeding up,

rather than the coming to a halt,
of sociocultural evolution (14). "Super sex" runs counter to
the
intuition
of
Mendelian
and
transmission geneticists
26
according to which, other things being equal, the blending

of characters resulting inevitably from such a mode of
inheritance will slow down and eventually call to a halt the
evolutionary dynamics. (For an unorthodox rebuttal of this
dogma, see De Waele & Schell, this volume.) This is why
neo-Darwinians such as Ernst Mayr put so much emphasis on
diversity in their definition of evolution ("changes in the
diversity and adaptation of populations of living organisms" - Mayr, 1976, p.1). From this it does not follow, of
course, that Boulding is wrong. It only follows that ifhe
is right, we can expect mechanisms or forces barring the
blending of characters to be at work in cultural evolution
which have no equivalent in biological evolution proper, or
which at least differ significantly from the segregation
mechanisms known in genetics. Cultural genotype concepts
should be defined liberally enough to account for such
mechanisms or forces (Callebaut, 1986).
Action rule/action. - Another economist whose work is
relevant in this context, the one who has maybe devoted most
energy
to the careful delineation
of genotypic and
phenotypic levels in human action and cognition, is Sidney
Winter. He reminds us that
one of the characteristics of
the "tandem mechanism of variation and selection" (Mayr) is
that
"the environment does not act directly
on genes
or genotypes. It acts upon the actual historical

individuals, whose characteristics and behaviors
are determined in part
by the environment
itself". (Winter, 1975, p. 97.)
Thus a genetically determined trait
which confers superiority only in environments that never occur will not be
favored by selection pressures. This is why Winter, whose
primary interest
is in the theory of the business firm,
distinguishes between phenotypic organizational actions and
the
genotypic
gies ll ,
or,
"decision
rules",
"action rules", "strate-
most adequately, the "routines lt of an organiza-
tion. "What the environment operates on, and rewards and punishes, is not the rule but the actions evoked
from the
rule by variables in the environment itself" (ibid.).
Disregarding here the methodological problems which
the actual identification of the genotypic behavior routines
involves (they
certainly are not
identical with the
routines as formally
described by, say, an organizational
chart), a potentially very useful idea seems to emerge: One
of the ways (but certainly not the only way!) in which the
27
relation between cultural genotypes and phenotypes can be

defined is by means of the action rule/action distinction
(cf. Harre, 1981).
Knorr
Commenting on an earlier version of this
correctly
remarked
rules poses the problem that

from the rules;
and that
maintain,
it
seems
us
to
does
that
actual
so regularly
indeed
to
in
paper, Karin
reliance
practice
fact,
"as
on
some
is
rendered
why
be
the
we
would
insist
that
"naturalistically".
An
of us
and systematically". This

weakness
social-philosophical accounts of evolution we are
This
action
m.ay deviate
of
"rule-following"
important
step
most
aware of.
in
be
this
direction has been taken by R.R. Nelson and Winter (1982) in

the development of Winter's views on the organizational
genotype from his earlier ideas (Winter,
1964) to the
analysis of organizational routines in terms of predictable
behavior. It remains to be seen whether the steps artificial
intelligence is taking towards such a naturalization (and
which R.R. Nelson and Winter are aware of) will allow to
make sense of older but extremely relevant philosophical and
psychological concepts bearing on this topic, such as
Michael Polanyi's "tacit knowledge" (see e.g. Kuhn, 1970).
Philosophers may be reminded here of the transition,
in
utilitarian
ethics,
from
the
traditional
"act
utilitarianism", which evaluates the merits of every human
action considered in isolation, to the "rule utilitarianism"
of Richard Brandt and others, which applies the criterion of

utility not to individual acts but to the general rules
postulated to govern these acts. Rule utilitarianism turns
out to fit existing moral practices far more better than
does
tradional
utilitarianism.
(This
becomes
clear
immediately if one considers a specific example such as the
morality of lying.)
Methods vs. "theseg'. - Nicolas Rescher, in a critique
of the evolutionary theories of science (e.g. Toulmin's or
Campbell's)
which rely on what he labels as "thesis
pragmatism", i.e. the view that "propositions are acceptable, i.e. qualify as true, if their adoption is maximally
success-promoting", has proposed an analogous move from
objectto
meta-level in epistemology.
He calls it
"methodological pragmatism". I t is based on the view that "a
proposition is said to be acceptable, i.e. to qualify as
true, if it conforms to an epistemically warranted criterion
or method, and a method is warranted if its adoption as a
generic principle for propositional acceptance is maximally
28
success-promoting". The latter he takes to be

that it rules out "gratuitious success"; false
superior in
beliefs can
sometimes be practically or "survivalistical1 y" efficacious,
while true beliefs can be counterproductive in these ways as

well. This move, Rescher thinks, allows us "to have it both
ways":
to
avoid
"occultism"
as
well
as
"the
rational
impotence of an inability to account for the actual course

of scientific progress" (Rescher, 1977, p. 159) .
Attractive as it looks at first sight, Rescher's
proposal as
it stands seems to us flawed in various
respects. To begin with, seen in light of post-postpositivist (i.e.
post-positivist
and post-historicist)
theories
of
science,
his two-level view
of science seems
difficult to defend; see, e.g., McMullin (1983) on the

all-pervading role of epistemic values in theory appraisal,
and especially Shapere (1984, 1987) on the many sorts of
scientific belief (items of scientific "domains") open to
revision.
It is
implausible that
one would ever
obtain a
neat dichotomy of the kind envisaged by Rescher. Most

importantly,
Rescher' 5
"methods
pragmatism"
remains
non-naturalistic, being a model not of natural but of
rational selection; i.e.
"( ... )
a matter not of biological but
rationally prefe7en~ial
elimination
historical
transm1SS10n
preference on the basis

tions,
in
the
present
owing
to
of
of
reasoned
of purposive consideracase
considerations
oriented
towards
welfare
and affective
well-being" (Rescher, 1977, p. 133).
It shares this question-begging feature (non-naturalism)
with Toulmin's (1972) model, in which a "cunning of reason"
is invoked to avoid the tautology that "might is right:
selective survival
is success" (Lakatos). (We will return
to this issue in section 3.2.) On a more general level
(cultural evolution) than the one considered by Rescher
himself (scientific evolution), we will retain the idea
that teaching methods (instead of teaching specific ideas
or behaviors) is an important type of "replication by
selective borrowing" (Campbell).
On the other hand, if the genotype-phenotype distinction from biology is appropriately generalized (one of us
has attempted this in terms of a self-simplification of
the phenotype in the genotype, which functions essentially
as an economizing device), it becomes possible to envisage
a nested hierarchy of genotype-phenotype distinctions on
29
which different (sorts of)

selection
forces can act
(Callebaut,
1986). Thus, the limitations of ll.escher's
dichotomous model could be transcended.
Degrees of generative entrenchment. - To round off our
discussion of evolution and development we would like to
attract the reader's attention to the work of Wimsatt
(1986b), in which the notion of generative entrenchment is
used to explain a number of
features of evolution and
development,
of earlier
consequences
such as the greater evolutionary conservatism

development
stages,
the
more
pe:vasive
of
earlier
of Itfrozen accidents",
and
and
modifications,
the
persIstence
the greater taxonomic, morphological
functional generality of earlier development features,

the rarity of "revolutions". wimsatt's abstract "devemodel is inspired by Simon's
lopmental lock ll
work on the
heuristics of problem solving (see especially Simon, 1969,

ch.4 on "nearly decomposable" problems/systems). wimsatt
treats developmental programs - phenotypically characterized
as solutions to the problem of building an organism (or
any other
kind of "growing"
system)
suited to its
environment, starting from a single cell (the zygote). In
terms of the developmental lock model, a character is
generatively entrenched in proportion to the number of
"downstream" features which depend on it in development.
Apart from allowing to
view "innateness"
as a degree
property related to
the degree
Wimsatt's approach
therto
unexplored
also provides us with attractive, himeans to generalize

the biological
genotype/phenotype
analogues as well.
the
physical
of generative entrenchment,
distinction so as to apply to cultural

Thus, it allows to consider features of
environment as generatively
entrenched in a
developmental program, which are clearly not genetically

innate or otherwise inborn. Considering the importance for
any EE of a certain degree of environmental stability
(however transient), in the absence of which negative
feedback
would be impossible (15), this seems to be a most
desirable feature. More generally: What from the biological
point of view is considered as its weakness when compared to
more specific models of development such as certain genetic
regulatory systems
the generality of Wimsatt's approach,
due to the fact that it is independent of the actual

mechanics of development (Burian, 1986, p. 221) - we take to
be its strength for the evolutionary

epistemologist's
purposes. Let us make clear what we mean by this.
30
Plotkin has warned against attempts to establish an

(antic)
identity across biological and cultural evolution,
as for instance Ho and Saunders (1984) are tempted to do,
and urged us rather to look for an identity of process and
explanatory modes.
For him, "the principal problem for
theory in biology is the distinction between the genotype
and the phenotype", and that distinction he uncatholically
regards as "an absolute partition: information does not move
back from the phenotype into the genotype". It suffices to
realize that when we move to the evolution of culture, these
constraints will have to be loosened considerably; for here,
one expects the genotype-phenotype interaction
to be very
dynamic indeed (16). Wimsatt's developmental lock model
seems to us to be especially suited to cope with various
kinds of relaxation one would like to consider (see, e.g.,
Boyd & Richerson, 1985), including those pertaining to the
important fact of exosomatic evolution (cf. GeorgescuRoegen, 1971, ch. 1).
3. THE EVOLUTION OF KNOWLEDGE AND OF
SCIENTIFIC KNOWLEDGE
Knowl~dge
and information.
Before we address our next
questlOn, a word of our use of the terms "knowledge" and
"information" is in order. The philosophically-minded reader
will have
remarked that we have been extremely liberal in
our deployment of this vocabulary, using the concepts of
"knowledge" and "information" interchangeably. This is a
deliberate choice: While awaiting a verdict in the heated
debates concerning the proper use
of "knowledge" and
"information" (Dretske, 1981; Engels, 1984), we want to
allow for the possibility that, say, animals do have
knowledge (even without beliefs (!): Carrier, 1980), as well
as for other possibilities that are anathema to the adepts
of knowledge as "justified true belief". Campbell's and
especially Plotkin's and Heyes' papers have useful things to
say on this topic.
3.1. Does knowledge evolve?

While we all agree that knowledge (as the product of cognition) is the result of some evolution or other, and that
evolution itself is a "knowledge process", we are seldom
explicit
about the applicability
of the evolutionary
31
paradigm to knowledge. We can only become clear on this

point by
differentiating
between
various
levels of
knowledge.
BVSR.
Among contemporary epistemologists, Campbell
probably did most of the work on the hierarchical nature of
knowledge.
Campbell (1960, 1974a, 1974b, 1977) proposes
an uncompromizing model of "blind-variation-and- selectiveretention" (BVSR) which is compatible with and is hence
enlarged to encompass the following constituents (c~. Brewer
& Collins, 1981):
A
nested hierarchy
of
{"vicarious"} selectors
(Campbell,
1974a: ten levels) going from more or
less automatic or subconsious processes to the very
sophisticated processes of science.
(ii):
Entitativity:
obvious entities
("natural kind"
specimens)
present themselves unambiguously to the
(i):
perceiver.
His
evolutionarily
adapted
knowledge building in a certain environment

for
"necessary"
recognition
of
these
systems of
will make
entities
(Campbell, 1973).
(iii): Triangulation: truth and validity are dependent on
the
perspective(s) of the knower. Thus, Campbell
(1974b) pleads in favor of "multiperspectivalness" in
stressing at once a clear epistemological

relativism and a need for interdisciplinary research.
science,
(iv):
(v):
Pattern matching and contextuality are genuine

features of
human perception and cognition. On our
understanding, this emphasis allowed for his later
shift to the empirical social study of science, viz.
the "tribal model" of science (Campbell, 1977, 1979).
Moderate reductionism: the hierarchy is not reductive
in the classical positivist sense. By introducing the
notion of "downward causation",
he explains that "the
laws of the higher-level selective systems determine

in part the distribution of lower-level events and
substances" (Campbell, 1974c, p. 180).
This reversal
introduces more constraints or
more internal
regulation ("vicarious selectors") on the processes of

information gaining, leading to the characterization of the
basic process of evolution of knowledge as "nonprescient"
or "unjustified variation" (Campbell, 1974b, p. 151). To our
knowledge,
this more
"cognitivist"
interpretation of
evolution is taken up only by those scholars who want to
32
integrate learning and development into the evolutionary

perspective
as supplementary levels
(e.g.
Plotkin &
Odling-Smee, 1982; Boyd & Richerson, 1985).
Dissipative systems and catastrophes. - Is evolution
a
multivariate process?
In
the latter
case,
wouldn't we
benefit from considering dissipative structures (Prigogine

& Stengers, 1984)? Aspects of evolution could then be seen
as the creation of new forms through processes of shifting
away from previously established equilibria. In terms of
Campbell's model, a nested hierarchy of selectors would
constraint the dissipative structuring. But this is pure
speculation today, since in its present state, Prigogine's
theory does not allow us to speak about such phenomena. To
cope with them we would need a theory of the behavior of
aggregates of dissipative structures, which is not available
(yet). For the time being, we can only marvel at the

complexity of the animal or human brain, and share in
Prigogine's awe when he ponders
that
"the dynamical
complexity of the human brain cannot be an accident" or that
it "must have been selected for its very instability"; and
we can only reiterate his question: "Is biological evolution
the history of dynamical systems leading to the dynamical

instability which we find in the brain, and which may be the
basic
ingredient
of
creativity
characteristic
of
human
existence?" (Prigogine, this volume).

To assess the relevance - actual or potential - of the
"new physics" for EE, we will need a picture of the theory
of dissipative structures, of catastrophe theory and of
their mutual relationship more precise than those we are
acquainted with. Is catastrophe theory a part of the theory
of dissipative structures,
as Prigogine implies (e.g.
Nicolis & Prigogine, 1977, p.74), or vice versa, as Thorn
clearly thinks (e.g. Thorn 1974, pp.24-25 on the "suppleness"
of his approach). Thorn seems to claim
that only his
approach is explanatory,
as it alone allows for the
introduction of the notion of causality. But at the same
time he
also pretends to offer a theory which is less
deterministic than Prigogine's, as it uses local deterministic features only.
At the basis of Thom's theory is his classification

theorem.
It states that only a limited number of catastrophe types (seven) can occur. In contrast, Nicolis and
Prigogine (1977, p. 74) notice that the exceptional richness
in the behavior of the solutions of the systems they are
interested in is accomplished "at a
considerable price;
no
33
general
has
been
classification
achieved
as
of the solutions
of
now".
What
of such equations
are
we
to
infer
from such a statement? The field of applied catastrophe

theory shows considerable fragmentation, and it has been
asked by Thorn himself whether his approach is really a
theory after all. Supposing catastrophe "theory" to be part
and parcel of Prigogine's "theory" (granting that this may
be a counterfactual statement), can it then be said that the
dissipative
structures
paradigm
does
not
offer
a
full-fledged theory either (Prigogine himself would readily
agree), as it also branches out into various scientific
disciplines (using different mathematical formalisms) and
as it also has multifarious domains of application? Put
differently: Are there any positive criteria which would
allow us to decide whether a specific problem (and of course
we are thinking here of cognitive evolution and development
in the first place) is suitable for treatment from Prigogine's or Thorn's point of view? More specifically, how are
Thorn's "potential functions" and "discontinuity" to be
interpreted in the domains which concern us here? Does
speaking of a statistical approach to discontinuity (Prigogine) make sense?
These are all questions awaiting further examination.
For the time being, we shall have to refer the interested
reader to a
recent and highly controversial reinterpretation of evolutionary
theory in quasi-thermodynamic terms
(Brooks & Wiley, 1986; cf. also Wicken,
1984)
and to
Depew's assessment of its (evolutionary-) epistemological
implications (Depew, 1986) for a partial answer to some of
these questions, which are not further discussed in this
volume.
Reintroducing subject and object into epistemology. A sceptic unimpressed by the arguments referred to above may
continue to maintain that dissipative structures per se do
no tell us anything "about adaptive organization, or about
the selection necessary to produce it, and thus nothing
about the most striking and central feature of biological
organization", as Wimsatt once put it (cf. Depew, 1986). To
him we would like to point out that
Prigogine's and Thorn's
theories can also be explored from a completely different
point of view.
In an important sense, EE
or at least one widely
publicized version of it,
namely Popper's -
has
been
an
"epistemology without a knowing subject"

(Popper,
1979,
ch. 3). The feasibility of an epistemology of disembodied
34
knowledge on Popperian lines

crucially
hinges on the
acceptability of his "Three Worlds"
program. We take it
that by now it should have become clear that this program
is an utter failure (e.g.
Bechtel & Richardson, 1983;
Callebaut, 1983, ch. 7, and references therein; Church,
1984). Once this is realized,
a
thoroughly
social
epistemology becomes inescapable (cf.
Flohr,
1985). For
only individual
even when it is granted that ultimately,
animal or human minds (and individual
computers of the
future?)
can really
know
(Callebaut,
ibid.;
cf.
Campbell,1987), the epistemologist will have to study the
various "social
system vehicles"
(Campbell, 1979, 1986)
influencing individual belief and (scientific)
knowing
processes (cf. Richards, 1981).
Since Campbell and especially Boon and Knorr Cetina treat these processes in some
detail in their respective papers,
we can refrain from
further discussing them here.
On the other hand,

justification
that in a
varieties of EE are also,
one could claim with equal

different sense, most extant
implausible as this may seem,
lIepistemologies without an object to know".
They share this
characteristic with most epistemological thinking in the

analytic tradition, which has always been wary of anything
with
a
"metaphysical"
flavor.
What
we
mean
is
basically that epistemologists have been busy building
epistemologies without bothering in the least about the
structure of the world their theories and models .of knowing
and knowledge were supposed to apply to. We have caught a
glimpse of the limitations of such
an approach when
discussing functional explanation in evolutionary biology
and epistemology (section 1.2.). As in so many domains,
Campbell's work is novel here in that it signals a crucial
departure from this anti-metaphysical tradition: It reintroduces
the
"world"
pole
in
the
object-subjectrep:esentation triad which constitutes the doxastic/epistemlC process (cf. also Simon, 1957, ch. 2, an early
discussion of the role of the structure of the environments
as complexity reducers for the problem-solving organisms
living in them).
Summarizing, we can say that in contemporary epistemology,
one witnesses a growing awareness (i) of the crucial
role of pragmatic (functional) determinants of the process
of cognition, i.e. of the structuring role of the individual
35
and collective subject, and (ii) of the structuring role of

the
object (of
"Nature
herself"
according
to the
hypothetical realist Campbell).
Against this background, it can again be seen why
Prigogine's and Thorn's
views
on
the
self-organization of
matter (of which we are a part and with which we interact,

also when perceiving and knowing) and on the implications of
irreversibility
"embedded"
for
system in
human
cognition
(which
Prigogine's sense)
may
is
also
an
in principle
provide us with important

clues for investigating the
possibilities and (probably even more so) the limitations of
human understanding in a world that is essentially evolving
in an "open" and maybe even in a "creative" way (cf. Levinson, 1982c).
To round off this section, let us return to the
question whether knowledge is evolving. No one taking the
fact of biological evolution seriously can consistently
deny that in some sense at least, knowledge is bound to
evolve (although this
discovery has been a historical
feat). For as Andy Clark put it,
"Nature must be red not
just in tooth and claw, but in instinct and des ide also if
the
teeth
and claws are
to be put
to good use"
(Clark, 1986, pp. 151-152). EE, of course, lives off this
discovery.
Nevertheless,
biologists
and
other
hitherto,
"more
emphasis
there
adherents
has
is
of
been
sentiment
biological
placed
among
EE that
upon the
epistemological problems than upon the evolutionary ones",

whereas biology also "has a great deal to say about how
knowledge is acquired and
organized" (Ghiselin, 1981,
p. 269). Keeping this caveat in mind, let us now look at
some of the claims adherents of the second EE program - the
evolution of science program - have been making.
3.2. Does scientific knowledge evolve?
Two opposite
discussion.
and
(i)
more
views provide a neat
frame for
beginning our
von Schilcher and Tennant deny that evolution,

in particular evolutionary fit, is a relevant
characteristic of science:
"Although there may have been some change, it

is
difficult
to
make out
a case why
particularly good scientists should have been
36
reproductively more successful. The important

shifts in cogn1s1ng since the birth of culture
have taken place at a more formal and symbolic
level." (von Schilcher & Tennant, 1984, p.189.)
(ii) In contrast, Oeser displays a fairly naive view
of
science when he states that the scientific method
"surpasses
life- and species-preserving functions, by
serving objective knowledge" (Oeser, 1984, p. 151). In his
unconstrained
enthusiasm
he
refers
to the "third
evolution" of scientific thought, and credits it with the
potential
to
have
altered
and
reversed
the
phylogenetically conditioned innate perception apparatus,
especially in the ancient Greek era. More in particular,
this "third evolution" he takes to imply "that the human
perceiving apparatus does not follow a natural adaptation
process but rather places itself outside this process" (p.
152).
The
scientific method
would
thus
allow for
self-transgression: "This stage does not only transcend the
biological evolution of plants and animals but even the
socio-cultural evolution of man, with consequences that
cannot be foretold as yet" (ibid.).
We are convinced that no genuine EE (and certainly
not a
thoroughly naturalized one
cf. infra) could
subcribe to all of
Oeser's points (his non-determinism,
his reversal of evolutionary processes etc.). On the other
hand, von Schilcher and Tennant's remark clearly shows that
they are confusing (as many others before them) the first
and
the
second
EE
programs.
The
alleged
assumption
of evolutionary epistemologists
that human
beliefs, scientific or other, must be somehow selected for
their relevance to human survival is a persistent target of
the critics of both the narrower evolution of science
program and of evolutionary approaches
to culture in
general. (Cf. Elster, 1983, p. 61, on the role of the
fitness function in evolutionary biology and its absence
elsewhere.) Although it is true that the literature on the
evolution of science program abounds with examples of superficial "biologizing" (as well as with other forms of sloppy
reasoning), we agree with Campbell (this volume) that the
"( ... ) mistaken expectation that EEs must assume that
scientific beliefs are selected by relevance to human
survival does not accurately describe any actual evolutionary theory of scientific development".
37
Beyond literal and metaphorical extension. - For an

interesting attempt to explain some important aspects of the
institutional structure of science (such as the recognition
of
original
contributions,
the punishment
of fraud and
other aspects of self-policing) in functionalist sociological terms, see Hull (1978). This is one of the very few
examples we know of an attempt to consistently extend the
functional
"logic"
of
evolutionary
explanation (cf.
section 1.2.) to the aspect of culture called science, which
is neither prone to dogmatic reductionism (despite its misleading title, referring to sociobiology) nor to facile
metaphorization.
In his (1982), Hull has explicitly pointed out that
there is a third route open to EE: building more general
theory. A crucial difference between, on the one hand, a
direct analogical or metaphorical extension of the form:
(1) biological exemplar -> model of culture (e.g. science)
and, on the other hand, building a more general model (or
theory, etc.), which is then respecified, of the form:
(2) biological exemplar
(e.g. science)
->
general model
->
model of culture
is that in the two-step process of (2), the relation '->' is

non-transitive
and
therefore
unobjectionable
from a
formal-logical point of view (see Callebaut, 1986, for
details). We will return to this issue in section 5.
Another problem we confront in this context is that
different notions of "science", "scientific knowledge" and
"scientific method" are used without being made sufficiently
explicit and
unambiguous. For Oeser, the rationalistic
view on scientific method of analytic epistemologists still
holds. That is to say,
he has no room for the fundamental
criticism of empirical epistemologists: Neither the "social"
and "opportunistic" scientist of Knorr Cetina (1981) nor
the "tribal" scientist of Campbell (1979)
do have a place
in Oeser's system. We think microsociology of science can
provide a good antidote here (see Boon's and Knorr's papers)
by offering valuable information on what science is "really"
about (17) and even more by redeeming us from dogmatic
slumber.
38
I/Evolutionary" does not imply "naturalizedl/. - Oeser I s

position is a good illustration of a recent remark of Dan
Dennett's:
"( ... ) while striking the Naturalistic Pose is

welcome as it is easy,
actually doing
naturalized philosophy has proved difficult - indeed a very
as agreeable and
unnatural act for a philosopher to perform".
We indeed want
to argue here that it should by no means be taken for

granted that an EE will de facto also be a naturalized
epistemology (cf. our comments on Rescher in section 2.2.).
This will become clear immediatp.ly: If, following Kornblith
(1985, pp. 1-3), we take the naturalistic approach to
epistemology to consist in the view that the question "How
ought we to arrive at our beliefs?" cannot be answered
independently of
the question "How do we arrive at our
beliefs?" (actually a rather weak version of naturalism),
then a proponent of EE such as Popper, whose antihistoricism,
antipsychologism, antisociologism... are known to any
philosophy undergraduate, clearly falls out.

One
Kornblith's
can,
of
course,
definition,
question
the
adequacy
of
e.g. by pointing out that natura-
lism comes in many varieties and grades (R.J. Nelson, 1984,

p. 175 ff.). We have indicated already why Rescher's and
Toulmin's EEs cannot,
listic;
i.e.
the methodologically
happens
is
methods which,
natural
on our
view, be considered natura-
when we take the latter to be synonymous with

monistic view that
whatever exists or
susceptible
to
explanation
"through
although paradigmatically exemplified in the
sciences,
are
continuous
from
domain
to domain
( ... )"
(Arthur
Danto).
Robert Richards, discussing
Campbell's BVSR model of scientific knowledge acquisition,
fell in the same anti-naturalistic trap when he wrote:
"Ideas are selected and retained by men for a
variety of explicit and implicit reasons: power,
passion,
reason
enjoy.
inertia,
derangement, stupidity, and
are all determiners of the ideas men

But when a mechanism modeled on natural
selection depicts acquisition of ideas primarily
for reasons other than reason, then it is not

a model of knowledge acquisition." (Richards,
1977, pp. 500-501.).

The reasons why all these authors ultimately resort to
Putnam's
idea that "Reason can't be naturalized" (see
Vollmer, 1985, for a rebuttal) are manifold. But they
usually seem to revolve around

unfamiliarity
with,
criteria
the non-existence of, or

to define
humans' (and
39
animals') "minimal rationality" (Cherniak, 1986) in a nonvicious way (i.e. in a way not already presupposing rationality).
Here
again
we
think
the
evolutionary
epistemologist can
learn a great deal from the way
certain economists have dealt with the problems of limited
rationality (Callebaut, 1983; for a recent review see Simon,
1983).
Towards an integral epistemology? - A few quotations
will make clear we are still far away from the "integral
epistemology"
some EE enthusiasts claim is already coming
of age:
(i) According to Campbell (e.g. 1974a, 1974b, 1977), science
is
first and foremost
a particular
system of
"winnowing"
those features allowing for "testable knowledge" (1974a,

p. 434). The selective forces are thus taken to be at work
at the level of data and of hypothesis testing
(cf. the
positivists' "justification context"
as long as it is
understood that it is scientific groups themselves, and not
philosophers, who do the justifying):
"What is characteristic of science is that the
selective system which weeds out among the
variety of
conjectures
involves deliberate
contact with the environment through experiment
and quantified prediction, designed so that outcomes quite independent of the preferences of
the investigator are possible." (ibid.)
(ii) In a remarkable comparison of Kant's and EE's problems
and notions, Vollmer (1983') tries to convey the conviction
that
evolutionary
epistemologists
have
firm
and
empirically founded views on the evolutionary status of

cognitive structures and
of
basic
concepts (Kantian
categories).
It is one of the
most
straightforward
expositions we have seen on this particular point. However,
the implications this is supposed
to
have
for the
second EE program, which Vollmer (this volume), "despite
all similarities,
Darwinian ll ,
(iii)
analogies and parallels"
remain open to debate.
At still another level of
the complex
regards as "not
phenomenon of
science, we can ask whether the generation of new ideas also
takes place in BVSR terms; in other words, whether the

latter also applies to the contexts of discovery and
pursuit. With Popper, Campbell is one of the few hardliners
here:
40
"Do I really, with a straight face, want to

advocate that
the discovery of new scientific
theories is through an 'unjustified' or 'blind'
variation of theories and a selective retention
process? Yes, I do, implausible as it may seem"
(Campbell, 1974b, p. 152).
He thus explicitly combines the discovery issue (whose
philosophical relevance someone like Popper would deny)
with the EE program. That this combination can be fruitful
and can add to the elaboration of an adult EE is worked out
in some detail in Callebaut (1983, ch. 6; 1984).
However, this focus captures only part of the field:
We do not have a full-fledged EE, able to cope with the
problems of justification or even with less philosophically
laden issues,
such as the historical appearance of the
experimental method (but see Campbell, this volume, for a
fresh start) or the analysis of observational strategies and
norms (cf. Knorr, this volume). In other words, evolutionary
theories of science at present can provide us with little
systematic information about the intrinsic features of
scientific knowledge production (Campbell's ERISS: "epistemologically relevant internalist sociology of science").
4. THE FORCES OF EVOLUTION
With the question of the forces of evolution we reach
a second oppos1t10n between EE and the Piagetian program.
Lewontin (1982a) presented a clear-cut analysis of both approaches, clarifying the issue to a considerable degree (although unnecessarily cautious with respect to the prospects
for an EE). We will use his excellent problem statement
wherever possible in this section.
4.1. The selectionist approach to evolution
Lorenz, Campbell and many others EE scholars remain
loyal to the Darwinian approach to evolution, i.e. they hold
a view of evolution
which is
fully compatible with
neo-Darwinian evolutionary biology and with the modern
synthesis (but see Apostel, this volume). The most pronounced and most debated variant in this
group (see
Richards, 1981) is Campbell's model of blind (or unjustified) variation and selective retention. Here the mechanisms are clear, also when the BVSR "algorithm" is applied
41
to knowledge: Species produce an undirected, non-programmed

set of varying behaviors, concepts etc. in their
struggle
for survival and reproduction; the environment selects out

most and retains only some variants in the next generation.
The model is
so
general as
evolutionary forces
to be capable of
complementing -
dealing with
but never replacing! -
natural selection (recombination, mutation, drift), although

Campbell himself has not extended it in any systematic way
on such lines (but see Boyd & Richerson, 1985) .. On the other
hand, in his more recent work (Campbell, 1977, 1979, 1987,
and this volume) he has successfully taken into account
internal ("structural") factors of evolution.
Richard Lewontin, identifying Campbell's interpretation with the trial-and-error metaphor, objects first of all
that
"organic evolution is not
trial-and-error adaptation"
(1982a, p.157). He
sums up a list of counterexamples and
forgotten cases, such as
hitch-hiking genes and the autopoiesis of organisms (18). He then pleads for a model that
will systematically take into account the dialectical relationship between organism and environment in the process of
evolution.
All this is still at the level of the evolution of
biological organisms and not that of EE proper. Extrapolating the investigation to EE, Lewontin remarks that the
basic structure of the trial-and-error metaphor implies the
unidirectional adaptation of organisms (e.g. scientific
groups) to their environment. Nevertheless, in all knowledge
processes
constantly
and
most certainly
reconstruct
their
in
science, organisms
environment
and
are
restructured by it. Thus, not adaptation to "a fixed

reality" (p. 168), but a dialectical relationship between
organism and environment should be a central metaphor. (For
a forceful defense of the dialectical view, see ch. 3, "The
organism as the subject and object of evolution", in Levins
& Lewontin, 1985.)
The present program of EE, then, covers only parts of
the integral program Lewontin has in mind. On the other
hand, the adoption of a niche-elaboration model of Darwinian
evolution to replace the common assumption of competition
among organisms for single niches may answer in part
Lewontin's criticism. As Herbert Simon remarks,
"The former theory is likely to be considerably
more complicated than the latter, since it must
explain the proliferation of niches as well as
the proliferation of organisms to fill them.
42
Moreover, an important part of each organism's

environment in such a system is provided by the
other organisms that surround it. The very
creation of niches, and the eventual development
of new creatures to fill them, alters the system
in such a way as to allow the development of
still more niches." (Simon, 1983, p. 45.)
We have seen before (section 1.2., comments on teleology)
that EE in fact has already adopted the niche-elaboration
model.
4.2. The "behavior-the-motor-of-evolution" approach
In the
Piagetian
developmental
(or transformational)
approach to change and adaptation a basically different
metaphor is apparent: Instead of blind variation we have
transformations,
and the process is recognized
as one of
"unfolding" (think of Piaget's maturation). It is often said

that Piaget held a more or less Lamarckian view of things
(e.g. Lewontin, 1982a, p. 165, pointing to Piaget's contro-
versial
interpretation
of
Waddington's
as
"collaboration"
by itself un-
objectionable "genetic assimilation"); although according

to Haroutunian (1983), his model was not articulated well
enough
to
allow
to
conclude
so.
In
a
formal
characterization of his posltlon, Piaget himself argued in
favor of Waddington's
"synthesis"
of
Lamarckian and
nea-Darwinian evolution in his theory of "epigenesis" .
Ontogeny and phylogeny are here taken to
be mutually
determining one another (since "epigenesis is the result of
a collaboration between the synthetic activity of the genome
and the environment") in what for Piaget has to be ali
equilibrium model (Piaget, 1982, p. 149). We will have to
keep in mind that the basic problem here resides in the
meaning of
words
activity".
A Piagetian equilibrium model emphasizing trans-
formation is
such
certainly
insufficient to
and "synthetic
explain organismic
evolution, let alone the (dialectical?) evolution of human

knowledge.
Yet Piaget's theory should not be discarded altogether
for assuming a fixed environment. At some level at least,
the parallel between biology and epistemology may be said to
break
down, viz. when science is interested in the
investigation of natural phenomena not or only very super~
ficially subject
to
change
as a
result of human
43
EVOLUTIONAR Y EPISTEMOWGY TODAY
(technological)
some
intervention.
environmental
Negative
stability,
feedback
however
requires
transient, to
paraphrase Campbell; and it seems to be a (contingent)

fact about the world we live in (at least to these authors)
that this stability is, on the whole, greater in certain
objects studied by (natural) science than in the partly
self-created niches of organisms (19).
If we agree in principle with Lewontin's critique that
the
interactive field "organism-environment"
should be the
core of concern of EE, we are nevertheless left with the

problem of delineating the factor "environment" in EE. It is
not
incompatible
with the
program
of
an integral
epistemology to think of it as both the biological and the
cultural environment. In fact, some more or less explicit
ideas have been
voiced by evolutionary
epistemologists on
this point: Campbell's focus on cultural and even spiritual

traditions for the delineation of "knowledge" in the social
sciences (Campbell, 1977) and his "tribal model" of science
(Campbell, 1979) are two partial attemps to flesh out this
factor of ("cultural") environment in our view. On the
other hand, and coming from a rather unexpected direction,
a similar reference to a deep level of cultural traditions
and their impact on the evolution of

the Prigogine school (see especially
1979).
However,
as far as
knowledge comes from

Prigogine & Stengers,
we can see no systematic theory
of culture and no theory of the dialectic relationship is

available. It will be profitable, certainly in the scope of
an
empirical
EE,
to consult both cognitive and ecological
anthropologists about the development of such a model.

4.3. Toward a new synthesis?
Time and again, scholars more or less critical of EE (e.g.

Thagard, 1980; Lewontin, 1982a) have warned against taking
the Darwinian metaphor seriously. Their scepticism is now
also being voiced by authors taking a basically sympathetic
stance vis-a-vis EE, such as David Hull. In a discussion of
the wider implications of nea-Darwinism, Hull recently

warned us that
" ... the fundamentals of evolutionary theory are
currently in a state of flux. Now is not the
time to take a particular interpretation of
biological
evolution and apply
it uncritically
to social evolution." (Hull, 1981, p. 73.)
44
Since no one can seriously deny today that science is essentially a collective phenomenon (see, e.g., Knorr Cetina,
1981,
or Boon, 1983, for hard data), Hull's observation
applies to EE
as well. We will return to this point immediately.
Moreover, Hull warns us - and we think h~s point is
well taken - against overly optimistic expectatlons as to
what an EE will ever be able to accomplish:
"No strictly biological theory of evolution is
going to explain very much about the content of
human
conceptual
systems
because
these
particularities are not the sort
of thing
evolutionary theory is designed to explain.
( ... ) If it cannot make ( ... ) predictions about
the genetic makeup of biological populations,
it certainly will not be
able to explain
comparable changes in societies or conceptual
systems." (Hull, 1982, p. 275.)
Keeping in mind these limitations, we can still rightfully
ask what advocates of EE can learn from biologists, be
they neoDarwinian hardliners or challengers of the
orthodoxy. As Ho and Saunders (1984b, p. 3) remind us,
"(t)here have always been
critics of the neo-Darwinian
synthesis: independent thinkers who steadfastly refused to
lose sight of the fundamental problems of evolution which
the theory does not address". Piaget can be seen as one of
them.
Conrad Waddington and
his
Theoretical Biology
Club, the "arch antisynthesist" Richard Goldschmidt, best
known for his "hopeful monster" idea (Gould, 1983, p. 89),
D'Arcy Thompson and others have raised many disturbing
questions about the modern synthesis. Yet the precipitating
factors for the current turmoil seem to have been primarily
some recent discoveries in molecular biology (Kimura's
"neutral theory" according to which many mutations may be
neutral with regard to natural selection - Kimura, 1983)
and in paleontology (the saltationism controversy - e.g.,
Eldredge & Gould,
1972; Eldredge & Cracraft, 1980).
According to Ho and Saunders, the neutral mutation concept
"presaged the fall from dominance of the genetic theory of
natural selection -
and the concominant return
of theories
on organismic structure and form", while the new phyletic

theory leads to the
conclusion that "the
origin of
species may require other explanations than those offered
within neo-Darwinian orthodoxy" (Ho & Saunders, 1974, p. 4).
Most biologists certainly disagree.
45
The implications of the saltationist view for EE (viz.

the Popper ian variety) have been briefly discussed by Ruse
(1983a) and will not concern us here. But we would like to
stress that the "new evolutionary paradigm" that authors
like Ho and Saunders see emerging may in principle also shed
new light on problems other than the problem of"revolutionary", non-gradual transformations: e.g. the problem of
the asymmetrical treatment of
the organism-environment
interaction
(cf.
supra), the problem of Platonism/essentialism, and even the problem of progress.
(i) The idea that "nature produces everything it can
rather than anything it needs" (Lovejoy's "principle of
plenitude") is
popular with many contemporary natural
scientists
nature).
(cf.
Prigogine
on
the
"over-creativity" of
According to Ho and Saunders (1984,
pp. 5-6), the
nea-Darwinian concept of random variation carries with it an
even stronger ontological presupposition which they label

as a "major fallacy": that "everything conceivable is
possible". This is especially clear in discussions of the
origin of life, "when it is claimed that life is utterly
improbable
without natural selection"; cf. Prigogine &
Stengers' (1984) critique of Monod.
"The argument is usually couched in terms of
the probability
that a functional polypeptide
of
a specific
amino acid
sequence could arise
by
change,
which
is
vanishingly
small
( ... ) Now, the latter supposition contains the
false assumption that we have a universe of
pure numbers devoid of physics or chemistry. In
fact,
experiments under simulated prebiotic
conditions
consistently tell
us
that the
probability space of prebiotic proteins is much
more restricted ( ... ) The same experiments show
that the other assumption implicit in such
formulations - that function is a very rare and
special quality created only as the result of
natural
selection
is also false ( ... )"
(ibid. )
Ho and Saunders therefore vindicate Henderson's famous
thesis of the "fitness of the environment". They point to
the one- sidedness of the Darwinian notion of adaptation
(cf. Lewontin, 1978; 1982a), which misses the reciprocal in
the relationship:
"the
environment
is fit
for the
origination and evolution of organisms". Inherent in the
alternative theory of evolution which the authors represen-
46
ted in
their reader
determinism
want
to
advocate is
the
concept of
in evolution, which implies a "time I s arrow" in
evolution, as well as the adoption of a theory of material

self-organization (cf. Prigogine).
(ii) This brings us to our second point. Ho and
Saunders observe that what most unites the plurality of
opinions voiced in their book is the emphasis on process.
This results, they claim, in "a transcendence of
the
predominantly Aristotelian framework of neo-Darwinism - in
which organisms are explained in terms of essences or genes
to the post-Galilean world view in which relation and
process are primary" (Ho & Saunders, 1984, p. 5.) These are
bold statements which a life-long defender of the orthodoxy
such as Mayr (1982) would certainly reject: consider his
plea for "population thinking" as opposed to the Platonism
inherent in traditional typological thinking.
(iii) Progress.
One of the standard criticisms
voiced against EE
as
a theory
of
scientific evolution is
"that it lacks a concept of progress essential in historical

epistemology" (Thagard, 1980, pp. 191-192; Bradie, 1986, p.
427). Richards (1981) and others have refuted this claim
convincingly, and we will not repeat their arguments here.
But let us make this one general remark (see Callebaut,
1983,
ch. 6 for a substantiation): Not only is the
variation-and-selection
model
of
biological evolution
compatible with a number of notions of progress, but most
authorities on the subject also agree that it is plausible
to actually consider evolution as
progressive,
albeit in a
vague' sort of way and with a global notion of progress

wanting, probably forever. In a new evolutionary frame,
the problem of direction in evolution would
probably also
look different (cf. Thoday, 1975). For Ho and Saunders, it
would
become
"primary
to
evolution",
hence "epistemolo-
gically prior to natural selection" (Ho & Saunders, 1984, p.

5). Evolution would then be considered in need of explanation "by necessity and mechanism with the least possible
appeal to the contingent and teleological" (ibid.); accidental variation and selective advantage would thus be
relegated "to the last resort". Assuming Ho and Saunders'
alternative theory to be true for a moment, one can see that
the gap between the

the allegedly non-
Darwinian view)
progressive evolution of science and

directed
evolution of life (on the
would he narrowed or even closed here.. But
we bet few evolutionary epistemologists are

the price for this advantage.
willing to pay
47
We now leave this vast new subject and turn to that

other Gordian knot of evolutionary theory: the problem of
the units of evolution.
5. THE UNITS OF EVOLUTION
Among biologists,
the agreement between orthodox neoDarwinians
and
more or less
heterodox "Lamarckians"
(Piaget),
"cognitivists" (Waddington, Goodwin), "soloists"
(Sheldrake) etc.
often concerns the acceptability of one
or another unit of selection or,
more generally, of
evolution.
The by now consecrated interpretation of the modern
synthesis has it that the basic material of selection is the
gene (Brandon & Burian, 1984; Sober, 1984), which is or is
not favored
in a
certain
environment
and will be
differentially replicated accordingly.
Working in a tradition of botanists going back to
Bateson,
Sheldrake (1982) was one of many biologists to
propose a systematic metaphorical expansion of the unit of
selection: One needs not, on his view, focus on a material
substratum (such as the chromosome) or stick exclusively to
an informational unit
(such as the allele). One should
rather conceive of a multltude of units, some of which may
even have a very unusual way of reproducing (e.g. his very
controversial "resonance" of units).
A very popular
metaphorical unit in
EE
is a
generalized
concept of information (cf. Lorenz, 1973) or
knowledge (see Plotkin's and Heyes' papers), apparent also
in Campbell's unification of the concepts of "selector"
and "learning system" as elements of hierarchical systems.
The forces of evolution which are accepted are the same as
those in the synthetic theory (cf. section 4), but they are
now exerted on analogues of biological information at
"higher" levels of evolution, viz. culture
(Campbell,
1974a, 1974b; Boyd & Richerson, 1985, p. 3). The original
biological paradigm has been "blown up". There is no harm
in such a procedure, provided it can be justified .(cf.
section 4).
An internal justification has in fact been offered by
independent scholars. The "nested hierarchy of selectors"
presupposes the unity of the biological world, comprising,
e.g., both the amoeba's and (at the other extreme of the
spectrum) the scientist's system of adaptation (Campbell).
48
The systems biologists

Rupert Riedl (e.g. 1984a; cf.
Wuketits, 1986) has gone to great lengths exactly to prove
that.
Starting from a heterodox theoretical tradition
(nonevolutionist transcendental morphology in the Romantic
tradition), his is a plea for the unity of nature and for
the all-pervasiveness of evolution. It is up to biologicts
to
test the strength and usefulness of Riedl's approach,
many aspects of which are also being voiced by Ho and
Saunders (1984).
A second presupposition of Campbell and Lorenz is that
some sort of hierarchical ordering exists between different
levels of information and/or selectors. They arrive at these
hierarchies without leaving the domain of the organisms
involved,
i.e.
without systematically treating
organisms
in their interactions
with their environments. This
critique has been brought forward by Plotkin and Odling-Smee
(1979) and it applies to most of EE. It is a fundamental
critique because it focuses precisely on
the units of
selection in EE. Plotkin and Odling-Smee (1979) claim that
not the knowledge systems (and the selectors) per se are the
proper units of evolution for EE to deal with; rather it
are
the
clusters
of knowledge-cum-environment units
evolving in a
constant
dialectical relationship (cf.
section 2.1.). Their critique is again convincing, we
conjecture, but it is not clear in what way this will
affect the hierarchical organization Campbell and Lorenz
are speaking about. This is a program that needs to be
worked out in the future.
A lot of fuss has been made recently about the claims
of some sociobiologists (especially Dawkins, 1976, 1982) to
be justified to dispense with all units of selection
"higher" than the ("selfish") gene: e.g. the group, the
individual organism of traditional Darwinism, and even the
genome considered as an organized whole (Mayr, 1976, p. 49,
p. 61). According to George C. Williams, who started this
razzamatazz, it is to be conceded that "it is unrealistic to
believe that a gene actually exists in its own world with

no complications other than abstract selection coefficients
and lIIutation rates" (G. Williams, 1966, p. 56). The unity of
the genotype thus seems to invalidate the (one-)locus model
of natural selection
dear to the
"monolithic" geneselectionists. Yet this is not a fatal objection, Williams
claims. For no matter how complicated a gene's interactions
with other genes and with the environment,
49
it must always be true that a given gene

substitution will have an arithmetic mean effect
on fitness in any population. One allele can
always be regarded as having a certain selection
coefficient
relative
to
another
at
the same
locus at any
given
point
in time. Such
coefficients are numbers that can be treated
algebraically, and conclusions
inferred from
one locus be attributed to
the effect of
selection acting independently at each locus."
(ibid. )
This is a daring claim, and a lot of energy has been spent
in recent years in order to refute it. These arguments (see
most notably Wimsatt, 1981b; Sober, 1984; and Lloyd, 1986)
are rather technical and are not directly relevant to our
theme per se. But the upshot of this controversy, which is
essentially about the possibilities and limitations of
reductionistic research strategies, is of paramount interest
to us. In a nutshell, it turns out that the "monolithic"

gene-selectionists' approach as well as related forms
of
uncompromising reductionism
are characterized
of systematic biases that tend

resulting in the reification of
have
been
posited
in
more
by
a number
to reinforce one another

abstract constructs that
or
less
arbitrary manner
(Wimsatt, 1980b).
The lesson EE can learn from this is that the Scylla
of
reductionism and the Charybdis of anti-reductionism
(viz. group- selectionism as an all-pervading phenomenon a
la
Wynne-Edwards)
are
both
to
be distrusted.
Pluralism reigns once again
(cf.
Sober, 1981"), and
advocates of EE can breathe without having to worry too much
(at present!) about the weak anchoring of their concepts in
"hard" biology. Dawkins shows them the way
here,
with
his
(naked) "memes", introduced in the last chapter

of his
bloody (1976) book, which already signalled a departure
from the author's original monolithic gene-reductionistic
stance, further attenuated in the more serene parts of his
(1982) .
If
the
conceptual toolbox
scrutinized
further in the vein
of
of
neo-Darwinism is
Dawkins' work on
"replicators" (Dawkins,
1982) and Hull's on "interactors"
(Hull, 1982), the amended concepts, we bet, will be quite
suitable for the evolutionary
epistemologist's purposes.
Hull's own work on the historical
nature of conceptual
50
systems and of their carrying vehicles - groups of scientists

(Hull, 1982, 1983, 1985) is a first step in the
right direction.
To our mind,
the incorporation of a phenotypic
dimension ("interactors") as well as its anti-essentialist
ontology (species ... as individuals) make Hull's approach
especially suited to
tackle "the major unsolved problems
[of biology - w.e. & R.P.] - in particular, the problems
of development and of cognition"
(Maynard Smith, 1986, p.
v). EE, even if it is not at present a unitary subject and
not much more than a multiparadigm program, may thus come of
age.
ACKNOWLEDGEMENTS
Our warmest thanks go to Leo Apostel, Barbara Frankel, Jens
Robert Maier, Arthur Miller and Neil Tennant for
their oral comments on the "position paper" for the 1984
Ghent EE conference out of which this introductory paper
grew; to Jean Paul Van Bendegem for allowing us to borrow
some ideas from an unpublished manuscript on the relation
between Prigogine's and Thorn's theories; to Aderito Sanches
for remaining one hundred percent a Newtonian psychologist;
and especially to Celia Heyes, Karin Knorr and Henry
Plotkin, whose exemplary homework saved us from many a
mistake. Our very special thanks go to Don Campbell, for
everything.
H~yrup,
NOTES
(1) Since few readers will be acquainted with this unexpected and maybe unsollicited reference relevant to EE
historiography,
we
add
the German
original to our
translation of fragments 5, 6, and 8 from Brecht's "tiber
'Das Ding an sich"': (5) "Der Baum erkennt den Menschen
mindestens so weit, als er die Kohlensaure erkennt." (6)
"Zur Erkenntnis des Baums gehort flir den Menschen die Benutzung des Sauerstoffs. Der Begriff des Erkennens muss also
weiter gefasst werden." (8) "Das Leben seIber ist ein Erkenntnisprozess.
Ieh
erkenne einen Baum,
indem ich selber
lebe." The fragment immediately preceding our quotation is

also nice: (4) "Die Dinge sind flir sich nicht erkennbar,
weil sie flir sich auch nicht existieren konnen" (things for
themselves are not knowable, since for themselves they
cannot exist either},
51
And the missing link,
rather more in
the vein of Otto Neurath's philosophico-political agenda

than
related
to
its
context,
simply
says:
(7)
"Erkenntnistheorie
muss
theory of knowledge must

first place).
vor allem Sprachkritik
be
sein ll (the
a critique of language
in the
(2) We cannot possibly scrutlnlze all these labels

here. Let us only call to mind that Descartes saw his
analytical method as a genuine method of discovery (as did
Galileo),
and
that
his
heuristic
was
to
be
contradistinguished from a synthetic, formal- deductive
exposition, on Euclidian lines, of discoveries already
made. Only later (most notably in Spinoza) came the latter
method to be favored as "a paradigm of the luminously
intelligible" (Bernard Williams).
(3) Cf. note 2. We have in mind here, more than
anything else, Descartes's consecretation of the "chains of
reasoning" approach imputed to Euclid: "( ... ) to carryon
my reflections in due order, beginning with objects that
are the most simple and easy to understand, in order to rise
little by little, as if by degrees, to knowledge of the most
complex [connaissance des plus composes] ( ... )" (Descartes,
1637/1948, p. 54; translation ours).
(4) That AI is increasingly "turning organicist" is
most explicitly documented by the
appearance
of the
"distributed problem solving approach" in computer science
(see,
most
notably,
McClelland
& Rumelhart, 1986).
Callebaut & Van Bendegem (1982), Callebaut (1984) and Clark
(1987)
discuss various
aspects
of
the relationship
between parallel distributed processing and EE. Authors
working in this new field are borrowing ideas from organization theories such as Herbert Simon's, which turn out to
be fertile source models (Fox, 1981).
(5)
In the lecture
delivered
when
he
received the
Nobel Prize in Economic Science, Herbert Simon (Simon, 1979,
pp. 510- 511), a pioneer of the information processing

approach who has always been fascinated
by Newtonian
mechanics as the paradigm for any rigorous science, pleads
forcefully for the adoption of other source models in the
dismal science: "If we wish to be guided by a natural
science
metaphor,
I suggest one drawn
from biology rather
52
than physics ( ... ) Obvious lessons are to be learned from

evolutionary biology ( ... ) We can see the role in science of
laws of qualitative structure, and the power of qualitative
as well as quantitative explanation." Economics has always
put great emphasis on rationality and intentionality assumptions,
inspired by maximum
principles in physics
(Samuelson, 1975). If the "queen of the social sciences" is
turning "evolutionary",
to follow. Cf.
Wimsatt (1986a).
(6)
Van
the other social sciences are bound
Parijs
(1981),
Actually the Brandon-Grene
McKelvey
(1982) and
position departs even
more from the ontological view of mechanism,
which seems to
be inevitably associated with reductionism. Brandon takes

it that "good hard-headed mechanistic methodology" is not
only compatible with, but actually entails the rejection
of "all forms of reductivist metaphysics" (Brandon, 1985, p.
348); he calls this "Grene's insight". Since
compatibility
and entailment are logically independent, we take it that
one could endorse either a weaker or a stronger version of
the mechanism-cum-nonreductivism position.
(7)
The latter point
reflects Hull's qualified
endorsement of Jon Elster's critique of functionalism in
social-scientific explanation. According to Elster (e.g.
1983, ch. 2), genuine functional explanation requires that
the actual operation of feedback loops passing through the
relevant
social
group
(etc.)
is
demonstrated.
"Functionalist" sociologists rarely, if ever, do this. Hull
thinks that Elster's requirement is not overly restrictive,
but that "at least sometimes it can be met by social groups
in their production of conceptual systems" (Hull, 1982, p'.
314).
(8) Sober also stresses - correctly we think - that
"the idea that variation in nature is not only something
that can be explained but is also an explanatory principle
in its own right" (Sober, 1984, p. 135) is at the heart of
what Mayr (e.g. 1976, ch. 3) has called population thinking.
Mayr
opposes
this
to
typological
or
essentialist
thinking in the Platonic-Aristotelian tradition (not to be
confused with what Aristotle thought himself, at least
according to Mayr:
Mayr, 1982, p. 11). The ontology
underlying population thinking has important consequences
53
for EE (and especially for the second EE program: the study

of the evolution of science), even if few evolutionary
epistemologists seem to be aware of this today.
(9) Comments on the "Position paper" of Callebaut and
Pinxten, presented at the Ghent EE conference, 1984.
Fisher (1983).
Also
notice the
(10)
But see
unexpected, almost complete convergence of the views of a
Marxist antireductionist - Richard Lewontin
and an
uncompromising gene- selectionist - Richard Dawkins - on
this issue. According to Lewontin (1982b, p. 183), we are
led to "a definition of freedom within causality". According
to Dawkins (1982, p. 11), "human nervous systems a.re so
complex that in practice we can forget about determinism and
behave as if we had free will".
(11) Let us qualify this bold statement to avoid a
possible misunderstanding. We are not implying that on the
whole, epistemologists have only reflected on (perceived)
past scientific successes. Considering the recent origin of
the gap between science and its philosophy (defined in terms
of, say, the professionalization of the disciplines), this
would be an utterly naive historical statement. But we
want
to
exclude here
all cases of
"false consciousness",
such as Newton's "hypotheses non fingo", contradicting his

own, less empiricist, scientific practice. And, in any case,
"not just the positivists but virtually every other branch
or school of philosophy of science that has developed in
this century and actually before, from about 1850 on, has
been retrospective rather than prospective", i.e. looking
at the "track record" of a theory rather than asking "What
can you do for me tomorrow?"
the kind of question a
practicing scientist would care about most (Thomas Nickles,
interview for Belgian public radio, 1985).
(12)
Recanting his previous views on fit as a
"mapping" of the environment (e.g. in his 1974a), Campbell
now wants "seriously to include only the 'fit' found in
those mappings which are the product of specialized mapping
structures,
presumably selected
by
natural selection
because of [their] adaptive value" (Campbell, 1987; italics
ours).
54
(13) This is not to say that "evolutionizing" Kant's

transcendental idealism does not have its problems; see,
e.g., Bradie (1986, pp. 445-6) on Lorenz's rather peculiar
reading of Kant. Vollmer (1985, chapters V and VII) reviews
various criticisms of EE's Kant interpretation, and offers a
reply. See also Engels (1984).
tricky
(14) We are aware that making such a judgement is a

thing as long as one has not defined units of
evolution,
since it is generation time rather than absolute
time that is of interest here.

(15) Cf. Hull (1982, p. 317) on the appearance of
homeorhesis in science, of "theories zeroing in on the
truth", which cannot stem from the working out of a
program that already contains the information:
"Sp,:,cies
are
forever
chasing
a changing
enV1ronment. The long-standing goal of science
is to discern in this flux some unchanging
regularities. It is these regularities that
scientists attempt to capture in their theories.
( ... ) Scientific theories are constantly being
changed, not because these regularities are
changing (so scientists believe) but because our
understanding of them is changing."
We will return to this issue when discussing the assumption
of a fixed environment (section 4.1.).
(16)
Plotkin,
comments on the "position Paper"
presented by Callebaut and
Pinxten at
the Ghent EE
conference, 1984.
(17) Lest we be accused of disregarding that the
constructivists and adherents of the various strong programs
in the microsociology of science emphatically deny to be
realists, we want to stress that in a major anthology of
their work, one reads: "Sociological studies must attempt
to take science and sociology as they find them ( ... )"
(Barnes & Edge, 1982, p. 147; italics ours).
(18) For an assessment of Maturana and Varela's
autopoiesis- paradigm from an evolutionary-epistemological
perspective, see Lelas (1986).
(19)
The evolution of
55
science is
certainly not only
and not even primarily determined by "external II criteria of

technological success;
at least not in
its present,
"non-finalized" mode, which the late Starnberg group used
to characterize as "naturwiichsig",
i.e.
where wasteful
variation by definition precedes the selection of scientific
"alternatives"
(Schifer,
1983).
In our op,n,on, the
relatively large autonomy ("Eigendynamik") of scientific
evolution might eventually turn out
to be
the main
obstacle hindering the construction
of a full-fledged
model of cognitive evolution after nea-Darwinism.
REFERENCES
See the cumulative
bibliographies at the end
of this
volume. For reasons of identification, some items have been

with an D-sign; they are all to be found in the second
bibliography, containing all the items not listed in the

first "EE bibliography" of Campbell, Heyes and Callebaut.
THE MEANING OF ENTROPY

Ilya Prigagine
Universite Libre de Bruxelles
Let me start with a personal remembrance.

I had the
privilege to meet Erwin Schrodinger at the occasion of the
lectures he gave at Ghent, Belgium (I believe it was in
1938-39).
As you know, Schrodinger was one of the greatest
physicists of our time; he was responsible for the discovery
of one of the basic equations of quantum mechanics. But he
was also quite influential in more than one field; let us
only mention his epoch-making essay What is Life?, issued at
the end of the war. His last book, My War ld View, discussed
one of the basic problems of philosophy, the nature of
individual (read: separate, personal) existence: What may
individual existence mean in the context of the Universe as
a whole?
For a scientist, he took a very strange approach. He
did in fact not take for granted that as a physicist he
could or should use the concepts of physics in order to
speak about his personal world view: He went as far as to
write:
"Not a word is said here of acausality, wave

mechanics, indeterminacy relations, complementarity,
an
expanding
universe, continuous
creation,
etc ... On this I can cheerfully
justify myself: because I do not think that
these things have as much connection as is
currently supposed with a philosophical view
of the world" (Schrodinger, 1960, pp.vii-viii).
In fact, he discussed this problem almost uniquely in the
philosophical
context
of
the
Vedas,
emphasizing a
pantheistic view in which the individual perspective is
reduced to an illusion.
Here we have a man endowed with a
who is not using
deep understanding of modern physics,
science at all in order to deal with the relation between
individual life and
cosmos.
As a matter
of fact,
Schrodinger explains why he is doing so: He is interested in
a basic philosophical problem, and his conviction is that
science is here of no help at all.
This is a strange
conclusion coming from a great physicist.
How is this
,7
W Callebautand R. Pinxten (eds.), EvoiutionaryEpistemoJogy, 57-73.
/987 b}' D. Reidel Publishing Company.
I.PRIGOGINE
58
possible?
Here we are discussing one of the great problems
of philosophy, one which should be connected with our global
view of the world : How could science have nothing to say
about it?
Still, I believe that in a sense Schrodinger was
right.
Classical
science,
even
when
one
includes quantum
theory, gives a very specific picture of physical existence.

The world described by the physics of Isaac Newton, the
world of classical dynamics, is made of material points and
motion: it is a kind of giant c lockwork in which there is no
direction of time, no distinction between past and future.
To a large extent, this is also the case in quantum

mechanics . Instead of material points, we here speak of wave
functions, which satisfy Schrodingerts equation, in which
again there is no distinction between past and future
(Prigogine & Stengers, 1984) (1).

It is obvious that in terms
of these timeless
descriptions,
we cannot even identify the problem of
personal
existence .
For us, individual existence is
time-oriented,
and
as long as
orientation in
included in the basic description of physics,

unbridgeable gap
between
the
ontological
existence and classical science .
time is not
there is an
problem of
The aim of this paper is to emphasize that science

is evolving, and that thirty years after Schrodinger's book
we may take a different view:
Science is now in
a pos1tl0n
to make a posltlve contribution to the basic problems of

philosophy - not to solve them, but at least to enable us to
see them in a new way.
Our century
has
witnessed
great
revolutions
in
science; it started with the formulation of Planck's quantum

theory and Einstein's relativity theory.
These are the
basic new contributions
to the understanding
of the nature
of the universe which have arisen in the first half of the

century.
But in the very last decades, something different
has begun: the emphasis is now on temporality at all levels,
be it at the level of elementary particles, be it at the
level of cosmology and of the global evolution of the
universe, or be it at the macroscopic level to which our
existence belongs.
While classical physics was postulating
that the basic laws of nature were deterministic and
in
which
time-reversible,
we
now
enter
a
world
diversification
irreversibility
play increasingly
and
important roles.
59
In this sense,
we can say that our period is
rediscovering time.
But let us first ask the question: How
was time conceived of at the beginning of the century?
Interestingly enough,
opinion.
there
was in fact some
consensus of
Great thinkers such as Einstein, Bergson or Heidegger,

in spite of their differences, held as a common belief that
time as irreversibility is not and cannot be the object of
science proper. You all know the famous statement by
Einstein,
that time
read: irreversibility
is an
illusion.
This statement has always amazed me: Time is for
us the main phenomenological and existential dimension; to
say that time is an illusion is,
in
a
sense, an
extraordinary expression of faith in symbolic thinking. For
Einstein, irreversibility was an illusion just because his
equations were in
a sense more
'real'
to him than his own
temporal existence. He believed that irreversibility is

subjective - is something that cannot be put into the
equations of physics without destroying their objective
value.
Bergson came to the same conclusions through a
different way.
He took for granted that irreversibility
cannot be taken into account by science because science is
using a
'spatialized'
caricature
of
time;
to
deal with
irreversible and creative time, one had, according to his

view, to use the perspective of 'duration' based on our
intuition.
Finally, Heidegger carne to the same conclusion: For

him, time was necessarily outside science, because the
perspective
of
Being
and
Time
is,
logically and
historically, anterior to the endeavour of modern science.
In fact, the problem of time has been

with us since
the
dawn of rational thinking,
since
the admirable
breakthrough
of
Greek
philosophers
into ontological
problems.
I can only evoke here the profound analysis of
time as developed by Aristotle in the Book IV of his
Physics: "Time is the number of motion, in the horizon of
the earlier and later" (2).
I refer you to the commentary by Heidegger in his
Basic
Problems
of Phenomenology
(Heidegger,
1975).
Aristotle asks:
and later come
"Where does this perspective of the earlier
from?
Is
it
inscribed in
the
nature of
'objective' things, or is it a feature of the 'soul which

counts'?" This was to be a major problem for philosophers
since Aristotle.
Is time only a human illusion or is it a
cosmic property of which man is but one of the realizations?
60
I.PRIGOGINE
We shall encounter this problem several times during our

investigation. I do believe that the crucial feature of our
period is that we are, for the first time in the history of
science,
in
position
to
choose
between
the
two
possibilities. Recent developments in both theoretical and

experimental physics
point out clearly
that time as
irreversibility is an essential ingredient of nature, and
that our existential experience of time is part of a cosmic
experience: Time is no longer separating us from the
physical world into which man is embedded.
How did we come to this conclusion?
The story starts
with the formulation of the second law of thermodynamics,

some 120 years ago: Clausius formulated the second law
introducing a new quantity,
entropy.
The fundamental
importance of entropy is that as a result of irreversible,
time-oriented processes,
the entropy
of the Universe
(considered as an isolated system) is increasing. This was
an amazing statement at a time dominated by classical
physics: The formulation of the second law was the first
occasion where,
into the frame of modern science,
the idea
of a history of the universe was introduced.

Since Clausius, physics is dealing with two concepts
of time: time as repetition and time as degradation. But it
lS obvious that
we have to go beyond this duality. The
negation of time,
or time seen as a
degradation, can never
do justice to the complexity of the physical world. We have

therefore to reach a third concept of time which contains
POSItIve, constructive aspects as well.
You cannot write
history describing only the decadence of the Roman Empire:
you must include as well the foundation of the Roman Empire.
This is true as well for the world as a whole: The world is
structured; our concept of time has to include therefore
both the formation and
the
destruction
which have occurred in the past.

Anyway, thermodynamics provided
historical framework of physics.
of
us
the structures
with
the first
Entropy is increasing for
an isolated system, while for a non-isolated system there

are both production of entropy inside that system and flows
of entropy coming from the outside world. Today, interest is
shifted from isolated systems to non-equilibrium systems
interacting with their surroundings through entropy flows.
Let us emphasize this essential difference with the

description of classical mechanics.
In thermodynamics, we
are dealing with 'embedded' systems: interaction with the

outside world through entropy flow plays an essential role.
61
This immediately brings us closer to objects like towns or

living systems, which can only survive because they are
embedded in their environment.
Non-linearity and
far-from-equilibrium situations are
two closely related factors characteristic of these systems;

their effect is that they lead to a multiplicity of stable
states (in contrast to near-to-equilibrium situations, where
we find only one stable state). New solutions emerge, which

are often
called dissipative
structures, that means:
structures whose very
existence is
due
to non-equilibrium
and non-linearity, and which require dissipation to survive.

Moreover, near bifurcation points which give access to these
new
states,
the
system
has
'choice'
between various
branches; we can therefore expect a stochastic behavior:

near bifurcation points fluctuations play an important role.
The so-called Benard instability is a striking example
of instability in a stationary state giving rise to such a
phenomenon of spontaneous self-organization; the instability
is due to a vertical temperature gradient set up in a
horizontal liquid layer.
The lower face is maintained to a
given temperature, higher than that of the upper. As a
result of these boundary conditions, a permanent heat flux
is set up, moving from bottom to top. For small differences
of temperature, heat is conveyed by conduction, without
convection;
but when the imposed temperature gradient
reaches a threshold value, the stationary state (the fluid's
state of 'rest')
becomes unstable: Convection arises,
corresponding to the coherent motion of a huge number of
molecules,
increasing the rate of heat transfer.
In
appropriate conditions, the convection produces a complex
spatial organization in the
system,
instead
of the
non-coherent state one could expect.
Here we have a first instance of what I would like to
call the "third time" of physics: time as irreversibility
has a constructive role here.
There is a basic difference with mechanics. Suppose
we have some foreign celestial body approaching the earth:
this would lead to a deformation of the earth's trajectory,
which would remain forever: dynamical systems have no way to
forget perturbations.
This is no longer the case when we
include dissipation.
A damped pendulum will reach a
position of equilibrium, whatever the initial perturbation.
Now we can
also understand in quite general terms
what happens when we drive
a system
far away from
equilibrium.
The "attractor"
which dominated the behavior
I. PRIGOGINE
62
(a)
.*
... .
.....
(bl
Attracteur
Dimension: 2 < d < 3
(cl
Figure 1:
Three types of attractors for dynamical systems.
Point attractor (1a); line (limit-cycle) attractor (1b);
fractal (non-integer dimensional) attractor (1c).
63
of the system near equilibrium may become unstable, as a

result of the flow of matter and energy which is imposed to
the system.
We see here how non-equilibrium may become a
source of order: New types of attractors, more complicated
ones,
may appear,
and
give
to the
system "new space-time
properties.
Dissipative systems may forget perturbations. These
systems are
characterized
by
attractors.
The most
elementary at tractors
are points
or lines such as
one may
see on figures 1a and 1b.

On figure 1a, we have a point
attractor P in a two-dimensional space (X1 and X2 may be
concentrations of some species):
whatever the initial
conditions, the system will evolve necessarily towards P.
On figure 1b, we have a line attractor: whatever the initial
conditions, the system will eventually evolve on this line,
called a limit cycle.
But attractors may present a more
complex structure; they may be formed of a set of points
such as on figure 1c.
Their distribution may be dense
enough to permit us to ascribe to them an effective (non
zero) dimensionality.
For example, the dimension of the
attractor on figure 1c may be any real number between 2 and
3.
Following the terminology of Benoit Mandelbrot, one may
say that this is a 'fractal' attractor.
Systems
with
fractal
attractors
have
unique
properties, reminiscent of, for example, turbulence, which
we encounter in everyday experience.
They combine both
fluctuations and stability.
The system is driven to the
attractor, and still, as the latter is formed by so "many"
points, we may expect large fluctuations in the motion of
different points.
One speaks often of "attracting chaos".
These
large
fluctuations are
connected
to a great
sensitivity in respect to initial conditions.
The distance
between neighboring trajectories grows exponentially in time
(this growth is characterized by the so-called Lyapunov
exponents) .
Attracting chaos has now been observed in a series of
situations, including chemical systems or hydrodynamics; but
the importance of these new concepts goes far beyond physics
and chemistry properly. Let us indicate some recently
studied examples.
We know
that the earth climate
has fluctuated
violently over the past. Climatic conditions that prevailed
during the last two or three hundred million years were
extremely different from what they are at present. During
these periods, with the exception of the quaternary era
64
I.PRIGOGlNE
(which began about two million years ago), there was

practically no ice on the continents, and the sea level was
higher than its present value by about 80 meters. A
striking feature of the quaternary era is
the appearance of
a series of glaciations, with an average periodicity of one

hundred thousand
years,
to which is superimposed an
important amount of 'noise'.
What is the source of these
violent
fluctuations,
which have obviously played an
important role in our history?
There is no indication that
the intensity of the solar energy may be responsible.
A recent analysis by C.
and G.
Nicolis (1984) has
shown that these fluctuations can be modelled in terms of 4
independent variables, which form a non-linear dynamical
system leading to a chaotic attractor of dimension 3.1
embedded in a phase space of dimension 4. The variability of
climate could have been thought of as resulting from the
interplay of
large
number
of
variables,
acting
in a
deterministic fashion; it would

then be a situation very
similar to the outcome of the law of large numbers. The new
insight is that it is not so.
The temporal complexity is
only due to 4 independent variables. We may therefore speak
of an intrinsic complexity or unpredictability of climate.
In a quite different field, recent work has shown that
the electrical activity of the brain in deep sleep as
monitored by electro-encephalogram (EEG) may be modelled by
a fractal attractor. A deep sleep EEG may be described by a
5
variables; again, this is very
dynamics involving
remarkable, as it shows the brain acts like a system
possessing
intrinsic
complexity
and
unpredictability
(Babloyantz, Salazar and Nicolis, 1985).
It may be this
instability which permits the amplifications of inputs
related to sensory impression in the awake state.
Obviously, the dynamical complexity of the human brain

cannot be an accident.
The brain must have been selected
for its very instability.
Is biological evolution the
history of dynamical sytems leading to
the dynamical
instability which we find in the brain, and which may be the
basic ingredient
existence?
As
of creativity
pointed
out
irreversibility seems to
the human brain,
be
characteristic of human
by
David
H.
Ingvar,
inscribed in
the structure of
where we see traces of the tripartition of
time in past, present and future; specific areas of the

cortex (the frontal/prefrontal areas) are responsible for
the anticipatory, goal-directed activities, as they offer
support for what he beautifully calls "memory of the
future";
while
65
other areas are
responsible
for recalling
past events (mainly postcentral cortical areas, especially

superficial and deep parts of the temporal lobes) and for
the experience
of
the present (mainly the
primary sensory
projection fields) respectively.

There can be no more
direct proof of the reality of irreversibility than this
temporally polarized structure of the brain (Ingvar, 1985).
We now come to the very core of the problem.

As a
result of the constructive role of irreversibility, we have
to reconsider our interpretation of entropy.
We also have
to understand its meaning at the microscopic level.

Till
now,
we
have considered entropy
macroscopic
context only.
if there
But
role of irreversibility, entropy must

basic level of physical description,
quantum-mechanical.
in
the
is a constructive
also appear at the

be it classical or
The classical interpretation of entropy was given by

Boltzmann. Every physicist is familiar with the celebrated
formula:
s = k log P
which relates entropy S to probability P (k is a universal
constant, called Boltzmann's constant by Planck).
But let
us
then ask:
What is Probability?
Is the use of
probability
an
effect
of
our
ignorance
or
the
very
breakdown of the description in terms of trajectories? As

in the case of the quarrel of hidden variables in quantum
mechanics,
physicists
have
been inclined to
think
that
probability enters the scene in order to

account for our
ignorance of the true microscopic
evolution of a given
system. (This was actually the view of Boltzmann himself.)
It
is in this
perspective
that
Born
said
that
"irreversibility is the effect of the introduction of
ignorance into the basic laws of physics" (Born, quoted in
Denbigh, 1981).
It
is
interesting
to
contrast
the
opinion
of
physicists on this fundamental issue.

For A.S. Eddington,
"The law that entropy always increases, ( ... ) holds ( ... )
the supreme position among the laws of nature", while for

G.N.
Lewis "in several important cases unidirectional time
and unidirectional causality have been invoked, but always
( ... )
in support of some false doctrine" (Eddington, 1948;
Lewis, 1930).
I.PRIGOGINE
66
~b
(a)
(b)
Figure 2.
Evolution of a system in phase space r. s (t)
maps point a on point b (2a); there is a conservation of
volume in phase space (2b).
Science is
67
often as
controversial as
modern art: we
are far from the idea of a monolithic building of knowledge

advocated by positivistic thinkers.
In fact, I would like
to show that this conflict can now be solved: The second law
actually conveys a message concerning the very nature of the
physical world in which we are embedded.
In order to come
to this conclusion, we need to take into account the
progress realized in the last decades in classical dynamics.
A real revolution is going on in this field since the

pioneering work of Poincar
and Kolmogorov: Important
classes
of
dynamical systems are now
known
to
present a
highly unstable behavior: Trajectories originating from two

neighboring points may diverge exponentially as time goes
on;
the
inverse characteristic
called the Lyapunov exponents.

As a consequence,
we
know
time
that
scales
involved are
no calculation done
with any finite precision can predict the long-term behavior

of such dynamical systems. Our knowledge about the state of
the system is necessarily about a given, finite number of
digits, and so we are only in a position to evaluate the
probability of finding ulterior states of the system in some
given region of phase space.
Taking unstable systems into account
thus leads
necessarily to the introduction of probability.
We may
associate a computer calculation to the evaluation of the
outcome of a coin toss.
What, then will the computer
predict?
If we have an infinite
info~ation
conditions, the computer will

probability one the outcome of
about initial
enable us to predict with

the coin toss.
But in the
finite information case, any computation will give us no
more certainty about the outcome than we get by expressing
the actual outcome of coin tossing in the usual terms of
probability theory.
This is so regardless of the precision of the
initial data. It is only in the limit of infinite precision
that the computer may get
rid of probabilistic concepts
and give the
exact
outcome.
But
neither physical
experiments, nor any computer evaluation are made with
infinite precision.
The situation is quite different from
what intuition would say: We would have guessed that we come
nearer to the exact deterministic result if we increase the
precision. But it is not quite so.
There remains an
insuperable gap between the behavior of dynamical systems
with
finite precision and the
deterministic behavior
corresponding to infinite precision.
68
J.PRlGOGINE
For
stable dynamical systems,
the situation is
different: increasing the precision, we come closer and
closer to the prediction of the deterministic dynamics.
We
can
expect
that
probability
and, through
Boltzmann's equation, entropy, will be the expression of
dynamical instability and not, as assumed by classical
physics, the mere expression of our ignorance.
Let us describe more finely the difference in behavior
of an idealized dynamical system in which the evolution can
be depicted by trajectories, and the 'real' dynamical
evolution, corresponding to a system about which we have
finite information.
As mentioned, physicists have usually
looked at dynamical evolution in terms of trajectories in
an appropriate space, called phase space.
They have been
working on the evolution of sets of points which occupy
some volume zone in phase space.
A characteristic feature
of classical mechanics (conservative dynamical systems) is
that this volume (the 'measure',
in the mathematical
language) is conserved in time.
For unstable dynamical systems the volume in phase
s~ace
will be highly deformed or even broken into small
p1eces.
The destruction of the initial 'simple' volume
gives the appearance of an approach to equilibrium, in which
all the points would be uniformly distributed in the phase
space.
Conservation
of volume
in
phase
space and
conservation of entropy are closely related.
That is the
reason wh~, in classical dynamics, entropy is strictly
constant 1n time.
Initial conditions can be restituted.
Indeed, the fragments of the initial volume could be brought
back simply by inverting the direction of time.
But this description is an idealized one, as it is
based on the traditional idea of trajectories.
As we have
seen, the concept of a given point in phase space loses its
operational meaning for unstable dynamical systems, and we
must show that we
then go into a
quite different
description, the thermodynamical description.
This is a
description in
terms of
processes,
in which the volume is
no longer conserved. But how does one obtain this new

thermodynamical description,
starting from a dynamical
description?
The classical description using probability
distributions p
{Gibbs'
ensembles}
satisfies
the
conservation
of
measure;
finite information,
but we
expect,
to find a description in
in the
terms
case of
of new
ensembles p , which includes irreversibility - in technical

terms, a Markovian process.
69
~ bw/;////gz/d
(0)
(b)
(e)
Figure 3:
The baker transformation.
General operation of
baker transformation on state (a) gives state (c) through
intermediary state (b).
tal
IIII
I'
iI
iii
A,
D
I
(bl
A,
[J ~
B,
e,
B,
e,
Figure 4:
Contrasting evolutions of ensembles of
points in the baker transformation. Contracting, vertical
fibers (4a); dilating, horizontal fibers (4b).
I
I
70
I. PRIGOGINE
The idea which I have followed for a number of years

(Prigogine, 1973), but which has now been established
rigorously for some classes of dynamical systems, is simple
(Martinez & Tirapegui, 1985). We may introduce a relation
between distributions p and p.
p;
where
is a 'deformation'
for unstable
of
dynamical systems.
which can be defined only

We
cannot
go
into more
details here; the essential element about the operator ~

is that it replaces a point (an inaccessible concept) by a
set of points, all compatible with the finite information
given.
In
'delocalized'
other
words,
we
deal
more
ensemble in which no distinction
with
among classes of points.
new
can be made
In the mechanical description, the world appears as a

museum in which everything, including entropy, is conserved.
In the description in terms of Markovian processes, there is
no conservation of volume any more, and the system is driven
towards equilibrium in the future; the new ensemble ~ is

temporally polarized.
We may suggest then,
a simple meaning of the
direction in which time flows.
To grasp this idea, we may
look at some dynamical systems, such as the so-called "baker
transformation".
Here the phase
space
is
square; the
transformation corresponds to the well-known operations of

the baker: flatten, cut, fold, as represented in Figure 3.
The sequence of states for a given point is clearly

deterministic.
But any region, whatever its size, contains
points which refer to trajectories diverging

at each
fragmentation.
We may then introduce the concept of
"contracting" and "dilating" fibers (Figure 4). These are
peculiar
objects
ln phase
space,
whose
fates are
contrasting. Let us suppose that a distribution of states is
concentrated
on
a
vertical
line:
At
each
baker
transformation,
this line is reduced, and would reduce to a
point in the far distant future.

On the contrary, a
distribution corresponding to a horizontal line will be
duplicated, and would eventually occupy the entire phase
space. Clearly, these evolutions correspond to opposite
behaviors: Dilating fibers reach equilibrium in the distant
future,
while contracting fibers correspond
in the distant past.
direction
of
Our
dilating
physical time is
fibers.
Instead
to equilibrium
oriented in the
of
remaining
constant,
71
the extension of
the domain
occupied
by
p (its
support) increases with time, and reaches equilibrium in our

future. Whatever numerical experiment 15 made, we never
p decreases
obtain that the support of
in time.
Contrary
to the world of classical mechanics, we here have a

privileged direction of time. Our time coincides with the
evolution of dilating fibers.
The world of thermodynamics is a world of processes,

destroying and creating information; the extension in phase
space is not
conserved.
Think of the
evolution of
temperature, the inhomogeneity of which

leaving any trace.
We
see the
fundamental
contrast
objects were
with
classical
conservative
undamped pendulum or planetary motion.

would
be
made
only
of
such
disappears without
science, whose
systems such
systems,
there would be no
irreversibility, no life and no human consciousness.
understand
as an
If indeed the world
We now
the message of the second law, and the physical
meaning of entropy accordingly: We are living in a world of

unstable sytems, a world polarized in time.
It
would
be
interesting
to
consider
also similar
developments in quantum mechanics and relativity, but this

is outside the range of a single paper; and as a matter of
fact, much - even most - remains to be done in these fields.

However, I would like to make a few general remarks.
Quantum mechanics as usually presented has a rather odd
dualistic structure.
In the
traditional (Copenhagen)
interpretation,
potentiality;
'actuality'
the
and
to
wave
function
transfer
functions.
understand
is
referred
potentiality
to
into
we need a quantum mechanical measurement, which
eventually leads to the 'collapse'
which a wave
this
function
is
of the wave function, in
replaced by an
ensemble of wave
This leads us to a paradoxical situation: To

quantum mechanics (in which a wave function
remains a wave function, just like in classical mechanics a

trajectory remains a trajectory), we need the collapse of a
wave function, a process which is outside the range of

quantum mechanics proper.
This is the paradox inherent to
the orthodox Copenhagen interpretation.
As in classical
mechanics, irreversibility in quantum mechanics means that
we
have to go from wave
functions
to appropriate,
time-oriented ensembles.
Then, there is no more need for a
collapse of the wave function; it is irreversibility which

leads to the transition from potentiality to actuality.
I. PRIGOGlNE
72
Neglecting this aspect, it would again be man who

introduces
time as irreversibility.
My
friend John
Archibald Wheeler is a great exponent of this approach,
through his 'Observer-Participancy' theory, in which man
creates time through his measurements.
Let
us conclude.
Coming back to Schrodinger's
questions, we can now begin to locate the temporal character
of individual existence in cosmological becoming. Morever,
the theory of irreversible processes relates irreversibility
to specific
features
of
the universe,
such
as dynamical
instability.
In this new frame, when we apply it to open
systems, we can understand the appearance of differentiation
in sufficient far-from-equilibrium conditions, as well as
the evolution of such differentiated regions of space into
forms of existence which become more and more independent of
the outside world.
Science can now distinguish
between
two formulations
of the ontological problem: In the frame of physics, we have

on one side the classical point of view, dealing with states
which are time-symmetrical, propagated through laws which
are also time-symmetrical in time.
On the other side, we
have states characterized
by a
broken time-symmetry,
propagated by laws which are also time-oriented, and for
which the second law of thermodynamics plays an essential
role.
From a logical point of view, there are therefore at
least two possible solutions to the problem of Being and
Becoming. In the first case, any intrinsic temporal element
is eliminated: Becoming is the mere unfolding of Being. The
second solution introduces a temporally broken symmetry in
both
Being and Becoming.
The solution to the ontological
problem is not a merely logical one, however; it involves a
factual element.
Indeed, if we ask for the meaning of
either Being or Becoming, we already introduce through this
questioning an
solution open
time-symmetry.
orientation in
time.
Therefore,
the only
to us is the one associated with broken

In my opinion, it is an important event that
physics permits us today to formulate this distinction.

In the new image of science as it emerges in this
second part of the 20th century, rationality can no longer

be identified with 'certainty',
nor
probability with
ignorance.
At all levels, in physics, in biology (Deneubourg, Pasteels & Verhaeghe, 1985), in human behavior
(Allen, Engelen & Sanglier, 1984), probability and irre-
versibility play an essential role.
We are witnessing a new
convergence between two
73
'visions
of
the
world',
the one
emerging from scientific experience,

the other from our
personal existence, be it through introspection or through
existential experience.
the
Sigmund Freud told us that the history of science is

history of an alienation (3): Since Copernicus we no
longer live at the center of the universe; since Darwin, man
is no longer different from other animals; and since Freud

himself conscience is just the emerged part of a complex
reality hidden from us.
Curiously, we now reach an opposite view.
With the
role of duration and freedom so prevalent in human life,
human existence appears to us as one of the most striking
realizations of the basic laws of nature.
NOTES
(1) For a general overview, see the section "The temporal
evolution
of quantum systems" in Prigogine & Stengers
(1984, p.226ff). I here disregard the controversial problem
of the collapse of the wave function, which would lead us
beyond the formalism of quantum mechanics proper.
(2) Physica, 219b: "arithmos kineseos kata to proteron
kai usteron".
(3) See the end of the 18th Vorlesung: "Zwei grosse

Krankungen ihrer naiven Eigenliebe hat die Menschheit im
Laufe der Zeiten von der Wissenschaft erdUlden mUssen ... "
(Freud, 1915 - 1917).
EVOLUTIONARY EPISTEMOLOGY AND THE SYNTHESIS OF

BIOLOGICAL AND SOCIAL SCIENCE
Henry C. Plotkin
University College London
1. INTRODUCTION
The chapter will bring together some ideas and approaches,
none of which is especially original, in order to try and
answer the question of what a general theory in biology that
makes contact with, and is meaningful to, the social
sciences might look like.
It will be especially concerned
with those elements that are owed to the EE that is defined
in scope by Campbell's review (Campbell, 1974a). This will
be done in two stages.
First, the kind of conceptual road
that I think must be followed if a synthesis is ever to be
achieved will be briefly explored.
The second will touch
upon some of the central issues on which a synthesis stands
or falls.
The reader who is impatient with metatheory may
want to go directly to the second part of the essay.
2. THE METAPHYSICS OF PROCESS AND CHANGE
Many biologists and social scientists, of course, dislike
and distrust theory in direct proportion to its intended
breadth. 'Premature synthesis' is a charitable judgement and
'grand theory' a term of abuse. To the ingenue raised on a
diet of the philosophy of modern physics this is a curious
judgement. What, after all, is the goal of science if it is
not
to provide
an ever smaller
set of explanatory
principles of ever wider compass? Every scientist would
agree with that, albeit grudgingly.
How then, apart from
ignorance, do we account for the continuing Balkanization
of biology and the social sciences?
It is suggested here
that one important reason lies in certain deeply held
beliefs and assumptions about how phenomena in biology are
best explained and which outweigh surface sentiment about
how science in general works.
My argument is that the
thinking of biologists is driven by two contradictory
'forces' - a background and rather hazy notion of generality
on the one hand, and more specific conceptions of the nature
of biological explanation on the other. And by evidence of
75
W. Callebaut and R. Pinxlen reds.), Evolutionary Epistemology, 75-96.
1987 by D. Reidel PubliJhing Company.
H.C,PWTK/N
76
sheer weight of numbers of those who emphasize distinctions

and divisions rather than unity and synthesis, the latter
is much the more potent.
The emphasis on difference
is
pervasive.
It ranges from demarcation disputes between
closely allied disciplines to really major claims
of
difference between evolution and development, between things
social and things biological, between whole organisms and
molecular biology, between ultimate and proximate causes,
between
functions
and
structures,
and
so
on.
The
conceptual lines that are drawn vary widely.

But what is
constant are the claims for the primacy of difference and
distinction over unity and synthesis.
It is, of course, certainly the case that differences

between different kinds of biology do exist.
Systems as
complex as living things will have differences of all sorts
in
them
some are
obvious
and/or
trivial,
others are
metaphysics,
of course,
methodologically based, and others are indeed important.

But is it worth asking what the source of the emphasis on
difference is. I suggest that it is the metaphysics that is
espoused by most biologists.
By metaphysics is meant the
most basic assumptions and premises that scientists have
about the phenomena that they study and what they consider
to be an adequate explanation of them. Most biologists have
By mechanism I
a metaphysics that is based on mechanism.
mean some (perhaps) irreducible, point-at-able entity or set
of entities whose behaviour is highly predictable. And most
biologists believe that they are studying point-at-able
things whose
explanation
is to be
sought
in such
irreducible, point-at-able events. If mechanism is what one
thinks of all the time, then differences, major differences,
become the inevitable focus of ones thinking, and a general
theory does indeed seem to be impossible.
There are,
however, other kinds of metaphysics.
Everybody in
science has
though many of us dislike the notion that we possess any

such thing. The result is often that these most fundamental
and powerful assumptions remain unarticulated and relatively
unknown, even as they are powerfully directing our thinking.
These elementary points are laboured because there is no
hope for any attempt at sketching the rudiments of a general
theory unless the metaphysical basis of the rudiments is
made explicit.
This is because any contradiction between
metaphysical assumptions in different areas will lead to
conflict and a failure of synthesis.
It is this kind of
conflict that so frequently, perhaps always, underlies the
THE SYNTHESIS OF BIOLOGICAL AND SOCIAL SCIENCE
77
arguments between biologists and social scientists. The

current round in that long-standing debate is the argument
that has been going on about sociobiology. If one may be so
impertinent as to pull other people's metaphysics out into
the light of day for them, the cause of the argument becomes
immediately obvious.
The metaphysics of the so-called
vulgar or radical sociobiologists is Democritean.
That is,
it is a reductivist mechanism.
Everybody knows that. On
the other hand, the metaphysics of the social sciences
cannot, by virtue of definition, be reductionist. Nor do
the social sciences have any mechanisms as I have defined

them.
Now you cannot marry a reductivist, mechanistic
metaphysics of biology to a non-reductivist, non-mechanist

metaphysics of social science.
That is why the writings of
the sociobiologists sit so oddly.
As a synthesis it does
not fit and the argument has really been about a clash of
metaphysics.
Now
metaphysics in
correct or incorrect.
fruitful or unfruitful.
science is not
something that is
Rather it is helpful or unhelpful,

It seems to me that any metaphysics
based on mechanism, whatever its advantages may be within
restricted explanatory framework, is neither fruitful nor

helpful in establishing the bases of some kind of general
theory of the biological and social sciences.
As I have
already said, obvious differences in mechanism - say between
evolution and development or between biological and cultural
evolution - these are what have been most inimical to the
development of a more general theory.
I claim no originality whatever in the alternative
metaphysics that I suggest is more appropriate to anyone
entertaining some form
of synthesis between
the social and
biological sciences.
This is the metaphysics of process.
Bohm's essays in Volume 2 of the Waddington series Towards a
Theoretical Biology are an especially fine exposition of
such a view.
For Bohm
the appropriate metaphysics for
is one in which process is fundamental.

That is to say,
'There is no thing
Things,
objects,
relatively
entities
constant
from
are
a
abstractions
process
all science
"'All is process.
in the universe.
of
of
what
movement
is
and
transformation" (pp.42). He goes on to say that the notion

of process is rather empty until one can say something of
its order, and he suggests that this can be seen in terms of
three stages:
"These are:
78
H.C.PWTKIN
(i) Quasi-Equilibrium Process

(ii) Dynamic Process
(iii) Creative Process
As a rule, we tend to begin in a situation
close to equilibrium,
which enables us to
recognize
certain
relatively
static,
or
entities (e.g.
the
constant, features of process.

We give these
features names, and are thus led to regard them
as stable objects or entities (e.g.
as we can
do when looking at clouds).
Then, as we see
that
these
features
are
changing
and
transforming, we seek to explain their relative
stability in terms of a dynamic process of
interaction
of
some basic
shapes of
the clouds are the
results of
movements of molecules of air and water). Still
later, we come to the notion of a creative
process, in which there are no basic objects,
entities or substances, but in which all that is
to be observed comes into existence as a certain
order,
remains relatively stable for some time,
and then passes out of existence (e.g. as

physics now explains the movement of electrons
through annihilation of existing orders and
creation of new orders).
In the metaphysics of
process, creation and transformation of order is
always taken to be the
deepest
and most
fundamental account of the laws of a process".
(Bohm, 1969, p.43)
As both Grene (1969) and Waddington (1969) observed, the
metaphysics of process does not begin with Bohm; and more
recent analyses (Ghiselin, 1985 for example) also emphasize
the importance of thinking about evolution in terms of
process rather than mechanism. But what is important about
Bohm's thesis is his notion of stages of order, which will
be returned to below.
What must be stated here are the
reasons why a metaphysics of process rather than mechanism
is a more appropriate basis for a general theory that seeks
to establish meaningful links between biological and social
sciences.
First, reiterating a point made above, any theory can

only be as wide as those areas which embrace a common
metaphysics.
If mechanism does not supply an adequate
explanatory basis for the social sciences, whereas process
is
habitually accepted as doing
so,
then a COmmon
metaphysics
might be
achieved
interpret biology in
then
synthesis
if
it
process terms.
between
social
79
proves
If this
and
possible to
can
be done
biological science
becomes attainable. Second, the need for theory to come to

grips with the great complexity of living systems is
increasingly recognized (Eldredge and Salthe, 1985; Wimsatt,
1974).
Causal explanations framed in terms of unilevel
mechanisms
are viewed
become a commonplace
most,
phenomena
processes,
with increasing scepticism.
recent
biology
years
that
It has
many, perhaps
development,
cognitive
ecology - are to be understood as highly complex
processes in
complex
in
of
which
structures
elementary
that
enter
constituents
into
give
rise to
higher
order causal
to
give rise to
interactions with other complex structures
yet other entities.

This is the kind of picture that few
biologists would now dissent from. A metaphysics of process
provides the common assumptions,
imagery and metaphor that
accounts for such diverse and complex phenomena.
There is a third reason for adopting a metaphysics of

process that is closely related specifically to the concerns
of EE, central to which is the notion that it is fruitful to
think of living systems as knowledge systems. Knowledge
does
not
take
the
form
of
"thing",
but
is
compounding, dynamic and endlessly changing relationship

between living systems and their
environment and
between
parts of living systems. These complex relationships between
and within living systems and the rest of the world are
wholly complemented by Bohm's metaphysics of
process.
Now a great many people, particularly philosophers, find
this usage of the word "knowledge" objectionable. However,
it seems to me that if one is going to write EE, then it
must be of an unashamed sort. For this reason I must expand

a
little on the way I think evolutionary epistemologists
should use the word knowledge.
I have elsewhere (Plotkin, 1982b) argued for EE as
being a way of thinking about biology,
whatever kind of
philosophy it mayor may not also be.
In that case, I am
frequently asked, why employ a word that has such well
accepted
common~sense
and
philosophical
meanings
describe, say, the eye-spot markings on the wings

of
moth.
How can wing markings on a moth be knowledge?
suppose
bears
about
it
is
because the answer
is
so
to
a
I
obvious that it
repeating, as follows.
I make the same assumptions
the adaptive attributes of phenotypes that most
biologists do,
viz.
that adaptations are defined by their
H.C.PWTKlN
80
seeming end-directed relationships to specific features of

their environment (see Plotkin, 1982b for references and a
fuller treatment).
In the case of the wing markings, the
relevant environmental feature is a
predator with an
aversion to the presence of eyes at close quarters. Such
matching between phenotypic organization and environmental
order
always only partial because of other constraints
determining phenotypic form, and he~ce knowledge is only
ever partial too - can only be the result of the former,
phenotypic organization, being "in-formed" by the latter,
environmental order.
As will be seen in a moment, the
former is part determines the latter and so the process is
dynamic and reciprocal, not simply one of "instructing",
but that is a complicating factor.
Taken in its crudest
form, the argument is that the phenotype is "in-formed" by
its environment.
Indeed this must be the case, for if
there were not such a relationship then adaptations would
not take the relatively stable phenotypic form that they do
and hence would not be effective.
That is, they would not
be adaptations.
As Lorenz (1969) -who trod this same road decades agoobserved, information is the obvious word to use to describe
the end-product of a process of 'in-forming'.
But that
word has been invested with very specific meaning by
engineers and information scientists.
So what does.that
leave us with? Knowledge. And there is point to using that
word
in
this
context
because
knowledge,
as
used
in
connection with all forms of adaptation, reflects the notion

that cognitive capacities also are adaptations, no more or
less so than the wing markings of a moth.
They are defined
by the same criteria of end-directedness and have their
source in the same processes.
And to go half-way, as
Dretske (1981) for example does, and define knowledge as
'information-caused belief' takes us no further than the
philosophers.
It
seriously restricts the
notion of
knowledge and, more importantly, it seriously distorts our
understanding of knowledge by suggesting that it is not to
be seen as an adaptation. Anyone who wants to maintain the
word knowledge as having no biological meaning fails to see
the strength of Piaget's (1971) assertion that "cognitive
mechanisms are extensions of the organic regulations from
which they are derived". Knowledge is a problem in biology,
whatever else it may also be.
81
NDr ShDUld we think Df knDwledge as the DutcDme Df

either a phylDgenetic prDcess Dr an DntDgenetic prDcess.
Such a false dicho.to.my (it is merely the DId nature-nurture
issue in a different guise) is harmful to' EE and symptDmatic
Df an emphasis Dn difference that CDmes frDm the implicit
adherence to' a metaphysics Df mechanism.
I o.ffer the
fDllo.wing: knDwledge is (1) a web Df relatiDnships, which
(2) find partial expressiDn in genDtypic and phenDtypic
structure (this is the stDrage term), and which (3) is the
DutcDme Df the DperatiDn Df a number Df prDcesses, these
prDcesses themselves representing knDwledge in the fDrm Df
relatiDnships.
SDme Df these relatiDnships are within
Drganism, and SDme between Drganismic DrganizatiDn and
Drder.
I
do.
nDt
underestimate
the
envirDnmental
difficulties
Df
establishing
the
metric
Df
these
relatiDnships, but hDwever that is achieved, fDr me thDse
relatiDnships are essential to' a cDncept Df knDwledge, never
the phenDtypic attribute in iSDlatiDn, and never as a
co.nsequence o.f the o.peratio.n o.f just Dne prDcess.
So., fDr example, the sDng Df a chaffinch is knDwledge
insDfar as it is a manifestatiDn Df a web Df relatiDnships
between certain genetic factDrs and their expressiDn as eNS
templates and endDcrine maturatiDn; and between certain
aspects Df envirDnmental Drder, to. wit, seasDnal changes in
light and temperature and the sDng Df cDnspecifics; and this
web Df relatiDnships is established by several prDcesses.
We shDuld nDt fDDl Durselves abDut the simplicity Df the
task and declare Durselves just fDr DntDgeny, say, and rule
DUt the rest.
EE has, I think, great pDtential fDr
synthesizing, but it will nDt do. that by simplifying and
supplying easy answers.
There is a fDurth aspect to. the use Df the wDrd
knDwledge in this CDntext.
It denDtes an active, dynamic
prDperty Df living things. KnDwing, the cDnsequence Df the
DperatiDn Df these prDcesses in the fDrm Df that web Df
relatiDnships established, alters the knDwer and in knDwing
the knower alters the wDrld that is knDwn since knDwing is
always relative to. a living system and never abso.lute. We
are not only epistemDlogical relativists (Campbell, 1977),
we are also. all epistemolDgical activists.
A number Df
eVDlutiDnists and eCDlogists have been critical Df the
"phenotype-as-passive-vehicle-for-genes ll
view that they see
as typical Df neD-Darwinism.
They are quite right in their
criticism.
But thDUgh LewDntin (1982) includes EE in his
strictures, in this instance he is wrong.
If any arm Df
H.C.PWTKlN
82
adaptational theory
stresses
the
dynamic
nature of
adaptations, it is EE.
It is to Piaget especially that we
owe a debt in this regard.
However, this realization
of
the dynamic, mutualistic, dialectic relationship that holds
between phenotypic adaptations and the environment, and
that must include the development of these attributes and
their relationships
within
any
living
system, this
realization is at a very early stage of formulation though
it finds expression at several different
levels
of
analysis.
For example, Patten's (1982) coevolution is cast
in general systems and ecological terms; Lewontin's (1982)
'metaphor of construction' in which individual organisms
are active constructors of their environments is
set at
more traditional level of the organism, as is Hull's

(1980'; 1985) concept of the interactor; and the Piagetian
analysis was set primarily at the level of individual
cognition.
Surely this kind of dynamic mutualism must find
expression in reciprocal social interactions
at the social
psychological, sociological and economic levels of analysis
as well.
A formal analysis of the dynamic nature of the
phenotype, that is, the dynamic ways in which phenotypes
gain and store and utilize information,
is at the heart,
or rather one of the several hearts,
of
a synthesis
between biological and social sciences.
Finally, and perhaps most importantly, thinking in
this way forces a quite different view on the theory of
evolution itself,
which becomes a set of dynamically
interacting processes akin to the kind of conceptions that
Waddington and Piaget wrote about; and really, substantially
different in spirit to current, rather static approaches.
In summary then, we travel a path where the conceptual
paving stones
are process;
where adaptations
are,
to use
another of Campbell's phrases, embodied knowledge; where

such knowledge is to be seen as a web of relationships
established by a number of processes; that these include
cognitive skills; that all knowledge gain is an active,
dynamic process; and that this kind of formulation, very
traditional in many ways, encompasses the notion of natural
selection and its consequences in terms of development and
the propagation of phenotypic attributes. differentially
across generations.
83
3. SOME ISSUES ON WHICH A SYNTHESIS STANDS OR FALLS

There are a large number of fundamental problems facing
any attempt to synthesize theory across the biological and
social sciences.
Just three of these will be briefly
considered here.
3.1.
Different knowledge processes have
characteristics
different temporal
Knowledge is gained, stored and acted upon over a wide range

of time intervals in living systems.
Let us take the lower
limit as the millisecond currency of the neurosciencesthough biochemists may want to talk about intervals that are
an order or two of magnitude briefer - and the upper limit
as the 10 14 milliseconds which was Haldane's estimate of
speciation time - probably not a defensible figure, but no
matter.
Ranged
between these extremes
are processes such
as learning (in the region of 10 5 milliseconds) and the

mammalian immune response (10 7 milliseconds), the accuracy
of the operating times given for these processes also being
questionable, but that is not the point. What is the point
is that we are looking at differences that range over quite
a few orders of magnitude.
It is obvious that in these
complex systems knowledge must be stored in
different
places (by store I refer to the organismic end of the web
of relationships); and it is likely that the storage
devices have characteristics that match the
temporal
operating characteristics of each
particular knowledge
process.
What a general theory must do is marry these
knowledge processes that have these different temporal
operating
framework.
characteristics
into
some
single
coherent
Contemporary evolutionary theory fails to do this.

Its focus is on the long end of the range, on stability over
relatively long periods of time.
G.C. Williams' (1966)
statement that "the selected entity must have a high degree
of permanence" is the classical example of pointing to genes
as the essential and central ingredients in a theory of

evolution.
Now this is fine if one thinks that evolution
is a problem only in genetics.
But if one believes that
evolution is compounded out of phenotypes - concatenations
of individual ontogenies - then genes are simply not enough.
Ghiselin has remarked that the emphasis on genes by most
evolutionary
theorists
is
much
due
to
cultural
84
H.C.PWTKIN
predilection for permanence over change as to anything else.

He is surely partly correct. Another reason is adherence
to the metaphysics of mechanism with the restricting
requirement that the theory be based on a point-at-able
enti!y.
Genes with their molecular structure that ensures
an 1nertness
and
hence
excellent
long-term storage
properties, obviously provide that.
But permanence and
point-at-ableness are not enough.
Knowledge, as has already been said, is about how an
organism,
or part of an
organism,
theory.
Lorenz
of
relates to "what is out
there".
In a world that is ceaselessly changing, in the
metaphysics of process where "creation and transformation of
order is fundamental", "what is out there" is not constant.
Hence the need for knowledge to change also. Thus not only
must knowledge be permanent in the sense of being stored in
a stable form, but it must also be updated, and these are
clearly related properties.
Furthermore it takes time to
update knowledge,
how
much
time
depending
on the
characteristics of the processes of knowledge-gain. During
that time of updating "what is out there" may continue to
change.
This is a well recognized problem in adaptational
spoke
"generational
deadtime"
and
Waddington
of the
"uncertain
futures
problem".
An
especially insightful treatment was by Sommerhoff (1950)
whose analytical biology was cast in terms
of three
adaptational processes - short, medium and long term - in
each of which a "back reference period" limits the accuracy
of an adaptation.
The significance of Sommerhoff's work
lies in his recognizing the need to partition adaptations on
the basis of the temporal characteristics of the processes
forming them.
SOCIOCUl TURAl
I
lEARNING
level 4
level3
IMMUNOLOGICAL
DEVELOPMENTAL
level 2
DEVELOPMENTAL
GENETIC
level,
I
I
knowledge
Figure
GENETIC
5. The levels of cognitive and immune

processes in man. See text for details.
85
Plotkin and Odling-Smee (1979; 1981) took a slightly

different approach to the same
problem.
We modified
Campbell's scheme of a nested
hierarchy of knowledge
processes by markedly reducing the number of levels in any
one hierarchy, increasing the number of possible hierarchies
operating in parallel in living systems, and the result was
a form of multiple-level model of evolution.
Our only
contribution was to point out rather more formally than
others had done what the sampling limitations of each level
in such hierarchy are; and to argue that these sampling
limitations together with accelerated rates of change in the
world are the selection pressures that lead to the evolution
of more subsidiary levels in these hierarchies that operate
on "faster time-bases. Figure 5 shows the appropriate levels
for two different kinds of knowledge processes, which for
simplicity are presented in simple linear rather than
hierarchical form (the whole problem of hierarchies will be
returned to below).
The number of different hierarchies
will vary in different species, as will the number of
levels.
The left-hand side of Figure 5 shows the hierarchy
of cognitive processes in a
species
like man.
The
right-hand side shows the immune system, the fourth level of
which is questionable though it may exist naturally in the
shared immunity of mother and foetus - and "unnaturally" it
is obviously present in communities
of immunized and
vaccinated individuals, the achievement of which is due to
the
interaction
of two different sets
of knowledge
processes, the cognitive and the immunological.
This kind
of 'cross-talk' is a significant feature of cognitive
systems.
Thus, n-level hierarchies of knowledge processes with
each level operating on a faster time-base is the solution
to those two related problems: how to keep knowledge in a
system over a potentially long period of time and in a
relatively stable form, and how to update it as rapidly as
possible.
Both requirements must be fulfilled.
In the
language of evolutionary theory, one unit of selection will
not, cannot, do both.
There must be multiple units of
selection operating
at mUltiple levels
of evolution.
Working out a sound analysis of the relationship between the
different levels of the hierarchy and the rates of change in
the world is a very important task. Plotkin and Odling-Smee
(1979) made a brief attempt but it needs to be done
86
H. C. PLOTKlN
rigorously.
A significant part
of
the analysis
knowing how to partition the experienced

the different levels of each hierarchy.
3.2.
Is there
processes?
some
universal
set
of
world
rests on
relative to
knowledge-gaining
If the scheme of evolution occurring at mUltiple levels in

various hierarchies of knowledge processes is a useful
conceptual
reaches of
device for synthesizing

theory across wide
biology, the next question to ask is whether
there is something in common between levels in terms
of how
they work. Are the processes of evolution, the processes of

knowledge-gain, the same or at least similar?
It may be
pitching it rather high to say that this is a requirement of
a general theory. But surely such elegant parsimony cutting
across conventional dividing
lines would be
learning,
and
an argument in
support of the possibility of a general theory.

The answer to the question is yes
at least a
tentative yes.
The candidate for a universal knowledge
process is much older than the notion of a nested hierarchy
of such processes. It is the evolutionary analogy th~t goes
back well over a century.
The evolutionary analogy ~s the
idea that development in general, but more specifically
understood
problem
if
we
solving
so
on,
adopt the hypothesis
can
that
be
they
better
can be
described and accounted for by processes similar to those of
biological evolution.
meant by the latter.
So one must first understand what is

A number of writers have attempted to
encapsulate the processes that result in evolution
by a set
of principles that can be expressed as a relatively simple

sequence of steps, an algorithm, and most are similar and
look thus: first, there is the generation of variants;
second, the variants are tested (selection is the phrase
favoured by evolutionists); and third, the selected variants
are regenerated. Exponents of the evolutionary analogy
extend these processes to phenomena like learning and even
science. Campbell's blind-variation and selective-retention
algorithm is the most widely known form of contemporary
evolutionary analogies.
Odling-Smee and I rewrote Campbell's set of processes

in order to present explicitly a regenerate phase, and we
called it a g-t-r heuristic (using heuristic not in its
formal sense but with its everyday meaning of
a device that
87
guides learning and discovery). The g represents a generate

phase i.e.
the generation of variants, the t a test or
selection phase
production
of
and
the
further
regenerate
variants.
More
should be called an r-t-r heuristic or
phase
i.e. the
realistically
device
it
or algorithm
because the g phase can only be the very first instance of

variant generation which thereafter becomes a string of
variant regenerations.
The reason for our writing the
algorithm in this form was that it allowed us to emphasize
that the regenerate phase is more, must be more, than simply
the production of variants selected by the last test phase.
We argued that the regenerate phase need not slavishly
follow the previous t phase and that in producing a range of
variants, though some may be similar or identical to those
selected previously, others might be novel. So the ~t-r or
r-t-r
heuristic is
chance
component.
produces
some
of
a mix of
what
conservative
worked
before -
What the ~t-r
pragmatism - it
and a radical or
heuristic
does,
to be
anthropomorphic, is it 'gambles' on the future being similar
to the past -
hence the regeneration
of selected variants.
But just in case it isn't, it hedges its bets by adding some

novel variants.
And to' the extent that the future is not
like the past,
those
novel variants
provide the potential
for new solutions to new problems.

Of course, this mix of chance and constraint occurs at
the genetic level.

Mutation rates vary widely in frequency
by orders of magnitude - both within and between species.
And segregation, gene interactions, gene multiplication,
duplication and so on, are additional devices for generating
novel variants and combinations
of variants.
Natural
selection and the fact of
heredity itself is
the source of
the conservative transmission of already selected variants.

Presumably the ratio of conservative: chance elements is
itself a product of evolution.
But are there similar mixes
at other levels in the nested hierarchy of knowledge
processes?
At the individual cognitive level there is as
yet little if any data because this way of looking at
learning and thinking is still so uncommon.
But studies in
animal learning (reviewed in this light by Changeux et al.
(1984), Richelle (1986) and Staddon (1983)) do suggest that
variation is a central component of the learning process and
that different learning forms do vary along this dimension.
Very conservative learning forms,
where the ratio of
conservative:chance elements is high, is epitomized by
Pavlovian classical conditioning.
The environment "stamps"
88
H.C.PWTKIN
an impression, an association, on the learner which is

little altered thereafter.
The learning has the appearance
of Itinstruction" and its conservative, stereotyped nature is
why
psychologists
have
tended
to
ignore classical
conditioning or to belittle its importance. The Skinnerian
operant is a more flexible learning form because the learner
generates a higher proportion of chance variants and the
ratio of conservative: chance is thus lower;
and some
evidence suggests that the ratio decreases significantly
with perceived changes in the environment.
Here the
learning is more obviously a process of "selection". In
general terms, the lower the ratio the greater the seeming
inventiveness and creativity of the learner, since by
creativity is meant the appearance of the novel and the
unexpected.
Thus, contrary to traditional approaches to
learning theory that emphasized learning's 'instructional'
nature and which could not cope with novel, creative
solutions in learning, the evolutionary analogy can at least
begin to account for creativity. It presents the possibility
of a really general theory of learning, that will deal with
learning forms ranging from the most dull and uninventive to
those that many still feel are quintessentially human. It is
worth pointing out that no other approach to learning comes
anywhere near such a wide explanatory range.
Again, I think that it is a reasonable assumption that
the different ratios that may characterize different forms
of learning have evolved.
And again one of our important
needs 1S to
understand
how
learners
partition the
environment in terms of change and rates of change, with the
evolution of
different
ratios
of conservative: chance
variants matching such rates of change.
A recent development in
evolutionary theorizing,
Maynard Smith's application of the theory of games (Maynard
Smith, 1982) does have an important bearing on this kind of
use of the evolutionary analogy since the r- t- r heuristic is
itself a primitive games theory approach to the analogy.
Maynard Smith accepts that the notion of evolutionarily
stable strategies or phenotypic stable strategies can be
extended to developmentally stable strategies and culturally
stable strategies.
The ESS approach does not contradict
what I am suggesting here,
rather it complements it.
It
tells us how these different ratios (strategies in Maynard
Smith's view) are to be explained in terms of competition
between individuals.
Thus not only might the ESS approach
tell us how different ratios of the analogy have evolved,
EXAMPLES OF HIERARCHIES
LIFE
INHERITANCE
89
(Simon, 1973)
COGNITION
ecosystem
Mendelian genetics
society
organIsm
chromosomes
organisations
tissue
genes
small groups
cell
DNA
individuals
I
I
I
organelle
macromolecule
I
I
I
I
I
I
I
eNS programmes
elementary information
processing units
Figure 6. Structural and control hierarchies. The former are

adapted from Simon (1973).
90
H. C. PLOTKIN
but it provides a potential predictive test of future

evolution of these ratios. However, I resist a capitulation
to the Maynard Smith form of analysis because it leaves out
so much that makes an EE rich:
notably the hierarchies of
knowledge processes and the way in which it may come to
captu:e the meaning of adaptations by relating organismic
organlzation and environmental order by some
dynamic,
reciprocal relationship.
There is something about the
nature of adaptations as having meaning, a semanticity,
that seems to be within the scope of EE and that gives EE
its promise.
3.3. Structural and control hierarchies
The third issue that I want to touch on is how, if at all,
the notion of complexity of organization is dealt with by
biologists on the one hand and social scientists on the
other.
Specifically,
what is meant when the phrase
'hierarchical organization' is used. Hierarchy theory has
become very active recently (see Eldredge and Salthe, 1985;
Vrba and Eldridge, 1984 for example) and there is real
confusion as to the meaning of the term and its various
usages (Grene, 1985).
What follows, therefore, is rather
tentative.
Figure 6 presents three of Simon's (1973) hierarchies,
which, as with Figure 7, are depicted in simple linear form.
Most biologists, if they think of their work in a wider
context at all, tend to do so in terms of the hierarchy
presented on the left of the Figure, that is, the hierarchy
of life.
Hence it is a not unreasonable assertion that
that hierarchy represents the dominant conceptual scheme in
biology - or at least has been so since the early 1950s.
However,
even
a cursory examination of
Simon's
other two
hierarchies, hierarchies of inheritance and cognition, shows

that all three hierarchies share a common feature - with the
exception of the level of 'Mendelian genetics' in the
hierarchy of inheritance, which is a curious anomaly in the
schemes and is presumably a simple error.
This common
feature is that all three are hierarchies of structure,
which is usually defined by the property of 'containment' or
'embeddedness'.
By this is meant that an organism is
literally 'made up' of, composed of, bears a relatively
fixed, if complex, spatial relationship to, its constituent
organ systems; that these in turn are made up of tissues,
and tissues of cells, and so on.
Simon calls this the
91
Chinese Box property. This property is not incidental. The

stability that results from this property is the reason why
structural hierarchies have evolved so widely in living
systems.
But there is another, seemingly different, kind of
hierarchy which is much more prominent in the thinking of
social scientists and which has been badly neglected, with
some honourable exceptions, by the biologists. This is what
Bohm (1969) calls the hierarchy of government, Dawkins
(1976) and Nelson (1973) refer to it as the hierarchy of
connection (as opposed to containment), and Pattee (1973) as
the hierarchy of control.
The lower part of Figure 6
illustrates such a hierarchy using a small part of a
university as the example.
Obviously armies, churches, or
industries would have served equally well.
What the lower
part of Figure 6 tries to do here is decompose the Dean into
two different hierarchies by rotating him/her through 45
degrees and running two hierarchies through that one level,
the Dean, at right angles to each other.
One of these is
the Dean as a part of a structural hierarchy in that the
Dean, together with Provost, departmental heads etc. make
up faculties,
and the
faculties
together
with the
administration comprises the university.
But the Dean can
also be depicted as a part of another hierarchy, a control
hierarchy.
As can be seen by inspecting Figure 6, the
relationship between the levels, and the nature of the
levels themselves, seem to be different in a control
hierarchy, and there is no simple way of mapping these two
hierarchies on to each other. They intersect at the Dean, or
indeed any other member of the control hierarchy that one
wants to depict also as a part of a structural hierarchy,
but they seem not to be the same thing.
The difference is that while a control hierarchy may
be a structural hierarchy as well, or may be partially so,
it need not be.
That is, there need be no containment at
all.
A Dean is not physically made up of his departmental
heads,
nor are the latter made up of
their staff.
Furthermore,
their
is a marked
feature of temporal
succession in control hierarchies in that 'something' is
passed across levels which becomes a part cause of events at
those other levels.
The general issues an order to his
commanders and that order becomes a part cause of their
subsequent behaviour.
And, importantly, though such a
hierarchy is defined by the authority relation between
levels, causation can move upward and downward in a control
H. C. PLOTKIN
92
hierarchy. Common sense tells us this must be so, otherwise

the system becomes uncoordinated and incoherent. Generals
do not merely send orders down the line.
They receive
information from
lower command levels
and such information
to
be
altered
appropriately.
What is the relationship between
structural and
control hierarchies? Are they entirely independent of each
other, as depicted on the left of Figure 7?
Is it simply
the case that every level of a control hierarchy is a part
of a structural hierarchy and no more than that? This is an
causes
their
uninteresting
subsequent
answer,
and
behaviour
intuitively seems to
be wrong.
Does
the
difference between
them
reflect
the old
form-function dichotomy?
This is one view advanced by
Pattee (1973).
Let us consider another approach based on
Bohm's notion of the stages of order.
Let us assume that
Simon is essentially correct in that
structural hierarchies
evolved because of the stability that they confer on complex

systems.
This is Bohm's quasi-equilibrium stage of order
which is the result of the interactions between levels being
very tightly
constrained by
actual spatial proximity
entailed in the condition of embedment
of structural
hierarchies. As evolution proceeds to more complex forms of
organization in some lineages,
there occurs a relaxation on
the requirement that control between the components of

complex systems be maintained by spatial constraint. The
complex systems that evolve, and that coexist side by side
with the original structural hierarchies, have levels that
are less constrained in their interactions which reflects
itself in more flexible, more creative order.
If this is
so, then structural and control hierarchies are closely
related to one another.
Containment is not the central
issue,
except
in
a historical
sense.
The nature
of the
interactions at the interfaces between levels is.

As I
understand it, this is close to Pattee's position, t~ough he
doesn't express it in this way.
It is an encourag1ng view
for the evolutionary epistemologist who defines knowledge in
terms of the dynamical interactions between levels, and
between those levels and the world.
93
a:
::>
f-
::>
--'
o
Q:
Q:
fVI
f-
------- leVel2
lL.
:r
f-
a.
/1
--I~vell
~
,:
~
structure
Quasi-eQuilibrium
creative
BOHM'S STAGES OF ORDER
Figure 7. Some possible relationships between structural and

control hierarchies. See text for details.
The right hand side of Figure 7 depicts this notion in
a particular way.
It assumes that any hierarchy can be
plotted along Bohm's stages of order and some dimension that
traditionally describes a structural hierarchy - the actual
hierarchy plotted here at four levels is an expression of my
prejudice, and not one original with me, that for EE an
interesting hierarchy is one
that links
genetic and
sociocultural levels via intermediate cognltlve levels.
The conventional structural hierarchy, Simon's hierarchy of
life for
example,
is
the
vertical column
at the
quasi-equilibrium end of the Bohm
hierarchy of
qualities,
some single value in
for
instance
the
dimension;
terms of
command
and a control
its structural
structure
in
university, is depicted by the speckled bar that lies in the

horizontal position.
The dimension of 'depth' is arbitrary
there are likely better measures than this. The actual
94
H. C. PLOTKIN
hierarchy shown here is the interesting case of a control

hierarchy whose component levels vary in their relationships
to a set of structural hierarchies. It is just this kind of
case that we need to understand better.
The 'solution' that this Figure offers is almost
certainly incorrect - the use of just two dimensions to
describe any hierarchy is simple-minded, and the dimensions
that have been used are surely wrong. But I am confident of
my more general point.
Virtually everyone seems to be
agreed that hierarchies are the appropriate way of looking
at the complexity of living systems.
My argument is simply
that we must find a way of running together the kinds of
concepts of hierarchy that biologists and social scientists
use if any form of synthesis is to be achieved.
4. DOES EVOLUTIONARY EPISTEMOLOGY HAVE
AN EMPIRICAL CONTENT?
Exception may well be taken to many of the ideas presented
above. Nonetheless, if EE of a different kind, or indeed of
any kind, is to be a candidate for inclusion in a general
theory, then it must fulfil one final requirement.
It must
have empirical implications and the power to lift problems
out of conceptual impasse and turn them towards an empirical

programme. There does come a point when argument, no matter
how elegant, is not enough. EE is fast approaching that

point.
I am not competent to review
all empirical
programmes that are specifically indicated by EE. I will
briefly consider the one area that I am most at home in,
namely learning.
A bit of postdiction first.
One of the most puzzling
aspects of learning is that it is now known with certainty
that animals "know" what it is that they have to learn.
These are the learning predispositions
that no other
approach to learning can explain, and they seem to present
us with a paradox.
But, place learning in the context of a
nested hierarchy of knowledge processes with knowledge
gained a posteriori at a more fundamental level of the
hierarchy being fed to less fundamental levels where it
appears to be a priori
knowledge,
and you have an
explanatory triumph for EE that is largely owed to Lorenz
and also Campbell. (Plotkin and Odling-Smee (1982) provide a
detailed description of
this case.)
This is a real
achievement.
95
Within a more traditional empirical framework there

are a host of other problems that need to be worked on,
ranging from how development and learning are linked, to an
understanding of the way the chance:pragmatic ratio of
generated variants relates to the learning of different
kinds of things.
Another line of study that needs to be
followed if EE is to advance is concerned with learning by
observation and imitation. Not only do we need to know much
more than we presently do about this form of learning, but
because it is so important to the problem of the evolution
of culture, the kinds of questions that must be asked
are
very different from those that were
previously framed
about learning from conspecifics.
We need to know how the
evolution of this form of learning relates to life history
strategies.
We need to know what kinds of
knowledge can
be transmitted between individual learners.
And we need
to know much more about the relationships between learner
and
teacher I
One of the charms of EE is its self-referential,

self-reflective nature.
A science of knowledge is also a
science of Science.
I began with Bohm, and with some
self-reflective musings on observer and observed, objects
and subjects, we can end with him:
"In the metaphysics of process, observer
and observed cannot be
taken
as separate
entities.
Rather, they are only names of
aspects of the total process.
And, indeed, we
see in the notion of the hierarchical structure
that since information is moving upward and
downward at all levels,there is no need for a
separate 'subject' who would be 'doing the
observing'
C... )
at no stage is there a
separate observer and subject.
The observer is
the totality of all that exists, and this is
also the observed.
Indeed, the movement of
information is the dominant order in the whole
process, from which the mechanical order is
abstracted as
'subordinate' ."
CBohm, 1969,
p.58.)
Self-referential systems are notoriously complex as well as
charming, and if one follows the form of synthesis suggested
in earlier parts of this essay to the science of Science
itself, then a number of as yet unresolved problems present
themselves.
If all knowledge is an active process, then so
too is science and the science of
Science.
Does
that mean
96
H. C. PLOTKIN
that the object of our knowing when we do any science

changes, and if so does that mean that science can never be
complete, including a science of Science?
Do we rule out
the possibility of a synthesis right at the beginning? If
that is unacceptable then the rudiments for a synthesis that
I have sketched are
either incomplete
or incorrect.
Perhaps, then, the problem of self-reference should be
listed as one more crucial issue on which the possibility of
a synthesis stands or falls.
ACKNOWLEDGMENTS
I am grateful to a number of the participants of the
International Conference of Evolutionary Epistemology held
at the University of Ghent in November 1984 whose comments,
critical and encouraging, helped to shape this essay. I
especially
appreciated several pages
of detailed and
complicated comments on the self-reference problem that Dr.
Jean Paul Van Bendegem sent me.
The final paragraph is all
that I am able to do in this paper in acknowledgment of the
issues that he raised, which require at least one lengthy
essay on their own.
I am sure that he could write that
piece better than I could.
For several years Celia Heyes
had to suffer my talking aloud about some of these points.
Patient listener and insightful critic that she is, I owe
particular thanks to her.
EPISTEMOLOGY OF EVOLUTIONARY THEORIES

Rene Thorn
Institut des Hautes Etudes Scientifiques

Bures-sur-Yvette
By evolutionary theory we mean any enterprise which seeks to

provide a 'scientific' explanation for a process so global
in nature as to allow of no experiment.
A process, in other
words, which presents a unique character, sui generis;

historically localized in time and place, it is a past
process which may end up in the present but which is not
repeatable. As a rule man has no means at his disposal of
controlling such a process, so that the resources of
'experimentation' are lacking. Let us take three examples:
1) cosmology as a whole (birth of the universe with the Big
Bang and its subsequent evolution up to the present time),
2) the evolution of life on earth with man as its end
product,
3) the diachronic evolution of a particular
language in linguistics. These three examples belong to
widely divergent disciplines (physics, biology, linguistics)
and their scientific statuses appear to be very different.
Nevertheless in each of these cases the fact that it is
impossible to reproduce
the whole process
raises an
epistemological problem. This problem is particularly thorny
for 'demarcationists' who insist on setting forth criteria
to separate scientific theories from
those they deem
unworthy of the name.
There are, I believe, three lessons to be learned from
this particular situation in evolutionary theories. If we
abandon the demarcationist position (once held by Popper),
which
consists in simply declaring
that evolutionary
theories
are
outside
the domain
of science, three
conclusions remain to be drawn from their existence.
1. The scientific analysis of an evolutionary theory

may well be possible, but the more or less formalized
character of the techniques of explanation affects and
graduates the scientific status of the theory.
97
W Callebautand R. Pinxten reds.), Evolutionary Epistemology, 97-104.
1987 by D. Reidel Publishing Company.
98
R.THOM
2. We generally have to forgo integral determinism in

an evolutionary theory (it is all too easy to foresee the
past).
The residuum of indetermination will often be
dismissed
as
pertalnlng
to
the
arbitrariness
contingency) of initial conditions.

3. In spite of the requirements of
(or
the
formalization, we
cannot relinquish the notion of cause, which is intimately

related to the irreversible passage of time. The ideal,
then, is to furnish causal justification for the successful

formalizations of the evolutionary process.
Evolutionary
theories
implying
a non-formalized
causality may be put forward
and have been formulated -,
but their scientific status remains doubtful. One must
therefore be on one's guard against possible ideological
contamination, sometimes revealed by examination of the
'pregnances'
(1)
underlying
the
causalities
under
consideration.
The problems thus raised are difficult questions, for

in point of fact they concern the junction of time and being
in general (Sein and Zeit). A first question arises: can one
attribute to time alone a property of creating the event?
Since time is the 'necessary condition for all experience',

time is present in every fact and cannot consequently
assume, by itself, a determinant role in the apparition of

the event. However one could imagine that time might take on
various qualitative hues; that a sort of temporal ether - a
metaphysical 'Zeitgeist' - might at every moment impregnate
all things, perhaps inducing transformations in their state.
This conception of historicism was rightly criticized by K.
Popper in The Poverty of Historicism (1960). But it must be
seen that time is rarely the abstract, cosmic, universal
time; more often than not it is the time of a being. Now for
a particular being the conception of time as a controller is
less absurd. We have only to think of a machine program in
order to realize that such a conception might sometimes be
perfectly justified (cf. the washing machine that reads its
program from an unwinding tape and carries out the orders
there inscribed). A global conception of history
a
'chronosophy' according to K. Pomian (1984) - may envisage
historical facts as being ordained in
advance by an
omniscient instance; the will of God for example. This would
be to extrapolate to a global order from a structure
possible for a 'local being', and to be guilty of a 'holism'
moreover perfectly sterile. Yet on the particular scale of a
local being,this conception of time as being able to take on
EPISTEMOWGY OF EVOLUTIONARY THEORIES
99
different qualitative determinations is fundamental to a

scientific explanation. When such determinations form a
finite set designated by the letters g1, ... ,gk and when time
is discrete
any history is
a word
constructed
with these
letters: Thus we construct the free monoid, generated by the

alphabet g1 ' gk' which is the universal structure of
algebra. If the determinations are quantitative and spatial,
any history is a trajectory in a continuous space X
and all
the mathematical formalism based on Differential Calculus

can then be applied.
Towards the end we shall come back to this necessity
of incorporating time in individual beings and to the links
between this conception and causality.
All evolutionary theory must of necessity start with a
description of the evolutionary process; the process is a
morphology in
a substratum-space
geometry.
This is
taxonomy.
(The
indispensable
which usually
when
one
includes a
wants
to know
whether two local processes are similar; for the local

repetition of locally similar accidents is the basis of all
'similarity'
vaguer than geometrical

nevertheless founded.)
in question may of
similarity,
In the best of cases,
upon
course be
which
it
is
we come to describe the process
as a spatio-temporal concatenation of typical forms

classified in a finite catalogue. The structural approach
then consists in reducing the arbitrary nature of the
description by applying to these concatenations the most
determinant constraints possible; compulsory associations of
neighboring
forms,
likely
associations,
forbidden
associations.
It is
in
linguistics,
with phonology, that
this formalizing technique has been the most successful.

In this connection it is important to correct the
ready-made opinion according to which structuralism requires
synchronic description and rejects diachrony. As a rule this
is not the case at
all;
one can indeed -
according to the
procedure of Einstein criticized by Bergson - spatialize

time and consider the process as a morphology in space-time.
What structuralism does reject is not time but causality. To
a structuralist the obligatory temporal association between

effect B and cause A is an obligatory concatenation A.B like
any other; in this regard the opinion of Hume, whereby
causality is a habit acquired by the mind confronted with a
necessary sequence of facts, is an example of structuralism
before its time.
100
R.THOM
It will
generally
be
observed
that axiomatic
constraints noted in the associations of forms are not
enough to determine
the process:
they allow neither
foresight from the past nor hindsight from the present. This
can be
seen even in linguistics,
where phonological
constraints of primary articulation have to be completed by
constraints
of
secondary
articulation,
that
is
to say,
neither too 'convergent'
nor too
constraints imposed by a signified context. Even in this

case, linguistic form is defined only up to a paraphrase and
the diversity of language has to be explained by the
contingency of initial conditions (the arbitrary nature of
the Saussurian sign). In cosmology the situation is much the
same. Here the irreversibility of physical time expressed by
the second principle of thermodynamics (in spite of the
reversibility of the laws in physics) is
usually explained by the singularity of initial conditions (the Big
This 'explanation', by its shift onto initial
Bang).
contingency, can be used every time the dynamics of temporal
evolution are shown to be
'divergent'; that is to say, the modes of representation of

the 'real' during the process remain to a considerable
extent invariant (cf. Hamiltonian dynamics in physics). In
these
models,
then,
strict determinism is generally
abandoned,
in
compliance,
(there is no merit in
case
intellectual
moreover,
with
natural ethics
foreseeing the past ... ),
honesty
consists
in
and in this
avoiding
abusive
determinism and formalizations of illusory precision. But

all the same, in so far as it makes it possible to reduce
the arbitrary nature of the description, formalization does
accomplish
a
task which can legitimately
be termed
'scientific'
even
if,
working,
as we are, on a closed
corpus, there can be no foresight of the future.

In fact, though the structural approach may satisfy
rigorous
minds,
it
still
leaves
some
degree
of
unsatisfaction.
Where
do
the
axioms
governing the
concatenation of forms come from? In algebra or in logic one
knows that a system of axioms may be more or less fruitful,
and it is important to be able to recognize which axioms are
useful
or
true,
causality
comes
in
concatenations a~pear
corresponding
aX10ms
according
to
Frege.
This
is
where
once
again;
cause
~
effect
immediately intelligible, and the
by
which
they
are
imposed
seem
necessary. This means that there is a deep ontological link

between the generativity of the formal system and the
temporal succession (additivity of the group (t. In a way,
101
what is expressed by the axioms is the nature of the space

where the action of 'times' develops. For example, if the
action generated by
two transformations x,y
is commutative
x 0 y = yo x this can be interpreted by saying that the

strategies x,y and y,x as represented in the plane Oxy by
translation vectors x,y, bound a rectangle,so that xoy
can be distorted continuously into yo x. Philosophically
speaking: universal time ramifies into a set of times of
beings. A strategy is a word of the free monoid W(gc). This
word W can be considered as a being limited by a beginning
and an end: onr------"1off. (On-off: commands On, Off). In a
sense, universal relations of equivalence between strategy
products describe the nature of the space where these
actions develop: the axioms often present a 'translocal'
character, magical it would seem, which can sometimes be
justified by
an interpretation involving a continuous
background.
From this point of view, if - for the time being - the
structural approach must be opposed to the causal approach,
it is to be hoped that one day the distinction between the
two will fade and disappear. Insofar as a process can be
linguistically described,
it gives birth to a formal
structure, namely the graph which can be attached to any
history, any narrative structure; for each character in the
history a corresponding straight line is projected on the
time axis at its birth and at its death. And all verbal
interaction between actants gives rise to a vertex common to
these paths. Thus we obtain a minimum graph which is the
semantic skeleton of the process. And the topological types
of the interaction vertices are extremely limited (Thorn,
1972). The process also exists in physics; here the actants
are particles, the vertices are interactions (Feynmann
graphs). But in the world of man, a fact always appears as
an irreducible singularity, a morphological accident, the
causes of which are at best defined by connection with a
previous set of facts or of characters. We then find
ourselves, ln our search for the causes of a fact, faced
with a very
common situation;
as we move up
the causality
graph, we find that there are more actants involved and that
the observed facts are wearing thin. This situation could be
described as resulting from a 'diffuse causality', creator
of that fact initially observed. We have here the origin of
a notion important in historical theory, the notion of
tendency. The way the graph converges towards the fact, as
well as to a possible multiplication of facts of the same
102
R.THOM
kind,
may
'tendency',
be
the
manifestation
of
the
advent
metaphysical entity if ever there was
of
one, but
the reality of which can hardly

be doubted. Similar
situations occur in biology concerning the causes of certain
physiological processes, such as the coagulation of blood or
somnolence. This leads to a certain rehabilitation of
Moliere's 'virtus dormitiva' ...
In fact,
the necessity of introducing imaginary
entities
to convey
~ausality
becomes
evident straightaway.
Let a fact A be the cause of an effect B. Considered as

spatia-temporal morphologies, the facts A and B may appear
in two very different modes. Either A and B are temporal
sections (tA <t B ) of the same form C, in which case we are
dealing with a trivial form of causality, that of the
permanence of the object. Or A and B are spatially distinct,
separated by a gap (a) in space.
Here we can infer the existence
of
causal influences
born of (A) and g1v1ng rise to (B). These invisible

influences can materialize in a flow of entities too small
to be perceived, or they may be considered as propagating
entities of a continuous nature, as in the case of physical
fields. This type of entity corresponds to what I have
chosen to call 'pregnances' - objective or subjective (Thom,
1983c). The investment of a (salient) visible form by a
pregnance is the typical fact.
The actantial pattern
associated with a nuclear sentence (in the transitive
sentence SVO) may be considered as a transfer of pregnance
from the subject to the object. The 'tendencies' already
mentioned can likewise be considered as global pregnances.
It seems
to me very
probable
that
all successful
formalizations in evolutionary science have axiomatics which
can be given a dynamic interpretation linked with pregnance

propagation. For example, Jakobsonian binarism in phonology
could be interpreted as the complication of a well of
parabolic potential V = x 2 /2 into a cliffed well where the
excited states are stable and dynamically opposed. In the
same way I have put forward a dynamic interpretation for A.
Greimas's semiotic square, in terms of a cycle of hysteresis
(Thom, 1983b). I believe that all axiom systems important
in algebra can take the same sort of interpretation linked
with some pregnance propagation in a universal space such as
the free monoid formed by the times of beings. (We are
talking here about a reversible pregnance leading to a
relation of equivalence, like the commutative relation xoy =
yo x
described
earlier
on;
quantum
mechanics
are
also
103
defined by a reversible pregnance defined by the wave

function (m;t) which indicates the propensity of the field
to be manifest in a pin-point particle.) But as a rule a
pregnance is irreversible,
going from source-being to
target-being. On reaching the target the investing pregnance
provokes therein changes of state known as figurative

effects. (Often enough, the invested being will in its turn
become source of pregnance, as in the case of disease
contagion phenomena). Because of its general character, this
type of causal analysis can equally well be applied to human
situations described in natural language as
to mathematical
formalisms in mechanics and physics where one material point

exerts a force upon another.
Science differs from magic insofar
as pregnance
propagation
can be
quantitative 'laws'
theories claiming to
'pregnant' concepts,
are called in,
submitted to
constraints
which can be verified.
expressed by
In sociological
account for historical development,

like the class struggle for example,
whose mode of
action is often
difficult to
appraise. In this sense there are, without doubt, degrees in
the scientific quality of
the theories.
In biology, people
are very dubious about the metaphysics of qualities and

object to the imaginary.
So that certain material entities
(like DNA molecules) are supposedly endowed with a power of

organization by remote control, and one falls into what A.N.
Whitehead calls "the fallacy of misplaced concreteness",
where a regression towards magic can be
discerned.
In this
respect it is important to make sure whether the propagating

virtues attributed to such and such a being or concept, can
be so defined as to satisfy scientific consensus, or whether
they are not, even partly, a reflection of the author1s
adhesion to a Kuhnian paradigm or to a sociologically
contingent ideology. Whenever it is a matter of universal
principles - such as the reductionism-holism distinction-
we touch upon
options of
permanent
character (Holton's
Ithemata l
(1973, and the answer may not be clear. But in
general (fortunately for
science), these great principles
have in themselves but a limited fecundity,
and it
is only
their local incarnations (analytical decomposition into

elements, or the holistic unity of a system) which can
result in verifiable descriptions.
104
R.THOM
NOTE
1. Here we use 'pregnance' as the equivalent of the French
'pregnance' with the meaning proposed by R. Thorn in earlier
writings (1983).
COGNISANCE OF CONSCIOUSNESS
IN THE STUDY OF ANIMAL KNOWLEDGE
Cecilia M. Heyes
Cambridge University
1. INTRODUCTION
By presenting simple typologies,

have encouraged the recognition that
epistemology'
is
descriptive
semi-independent enterprises.
In
distinguishes
'evolutionary
several recent authors

the term 'evolutionary
of
at
least
two
this volume, Campbell
'biological evolutionary
theories of science',
epistemology' from
Vollmer contrasts
Lorenzian with Popperian

EE,
and,
in a forthcoming
contribution,
Bradie (1985)
differentiates the EE of
scientific theories (EET) from that of cognitive mechanisms
(EEM).
Although there is some controversy as to whether,
and how, the enterprises can be distinguished in terms of
their objects of explanation,
these authors hold one
principle of classification in common: the literal versus
the metaphorical use of neo-Darwinian theory (1). To this
extent then,
the following view may be
regarded as
representative.
"There
are
two
interrelated but distinct
programs which go by the name 'EE'.
One is the
attempt to account for the characteristics of
cognitive mechanisms in animals and
humans by a
ideas,
culture
straightforward extension
of the biologi:al
theory of evolution to those aspects or tra1ts
of animals which are the biological substrates
of cognitive activity,
e.g., their brains,
sensory systems, motor systems etc.
The other
program attempts to account for the evolution of
scientific
theories
and
in
general by using models and metaphors drawn from

evolutionary biology." (Bradie, 1985, p. 2-1)
Many contributions to the present volume are part of the
latter programme, but this essay focuses on issues which
are central to the aims of EEM or biological EE. In the
paragraph cited,
Bradie
provides
a
somewhat modest
characterisation of those aims.
They may be more boldly
105
W. Callebaut and R. Pinxten reds.), Evolulionary Epistemology, /05-136.

C.M.HEYES
106
stated:
to
substantiate
the
Darwinian
claim
that man's
intellectual capabilities are continuous with those of other
animals by specifying the nature of the continua, and their

source in terms of ecological or phylogenetic factors; and
thereby to provide a coherent framework within which animal
experiments can be used as a means of investigating human
knowledge processes, where practical and ethical constraints
render such a substitution desirable.
Whatever the extent
of a biological evolutionary epistemologist's ambitions, he
is likely to find himself in the predicament of being
vitally dependent upon the data from animal studies for the
fulfillment of his aims,
and
yet unable
to devote
sufficient time either to the collection of such data
or
to keeping abreast of the
relevant literature.
In
qualified
adherence
to
the
'fish
scale
model of
omniscience'
(Campbell,
1969)
then, this
essay
is
intended to provide
an outline
of some contemporary
developments in the empirical analysis of animal knowledge
for those whose interest in biological EE originates in a
different area of specialisation.
Contemporary students of
animal
behaviour
or,
more
specifically,
cognitive
ethologists and animal cognitive psychologists, have their
own theoretical orientations (and associated conceptual
confusions) that only partially overlap with those of the
biological
evolutionary
epistemologist.
Since
these
perspectives influence both the range of animal competences
that are selected for investigation and the manner in which
they are documented, it is necessary for the evolutionary
epistemologist to recognise the direction that empirical
studies of animal behaviour
are taking if he is to make
full and appropriate use of this important data source.
The discussion will focus on the role of consciousness
in the explanation of animal behaviour.
The preceding
sentence is deliberately ambiguous because consideration
will be given to three of the possibilities it confounds:
(i) That members of at least some non-human species are
conscious and that this fact can be used in explaining their
behaviour.
(ii) That the explanatory mode traditionally
associated with the attribution of consciousness (in terms
of reasons and actions) can be profitably applied regardless
of whether the system to which it is applied is in fact
conscious. (iii) That human consciousness is used as a model
for explanations
of animal competence even
when the
ontological
and
epistemological
claims
for
animal
consciousness
are
denied
or
pronounced
irrelevant. This
COGNISANCE OF CONSCIOUSNESS AND ANIMAL KNOWLEDGE
107
issue has been chosen as a focus because, after decades of

obscurity, strenuous attempts have recently been made to
re-establish
its
currency among
students
of animal
behaviour; and because it is fundamental to the traditional
philosophical objections to biological EE.
That is, the
non-existence
or
unique
inaccessibility
of
animal
consciousness
constitutes
for
many
scholars
an
insurmountable obstacle in the path of any attempt to render
coherent the notion of animal knowledge (2).
The first and
second sections of
this essay evaluate
Donald Griffin's and Daniel Dennett's attempts to enrich the

study of animal behaviour by making it
cognisant of
consciousness.
The first deserves attention as an example
of the muddled thinking that is increasingly influential in
this area, and the second provides a partial antidote to the
confusion in the form of a disciplined and coherent analysis
of the problem. However, in the final section I shall argue
that while the evolutionary perspective of both programmes
could, from the point of view of EE, profitably augment
animal
cognitive
demonstrating
any
psychology,
current
neither
utility
for
is
the
persuasive
consciousness in the study of non-human species.
concept
in
of
2. CONSCIOUSNESS AS AN OBJECT OF
EMPIRICAL INVESTIGATION
"What is it like to be an animal? What do
monkeys, dolphins, crows, sunfishes, bees and
ants think about?
Or do non-human animals
experience any thoughts and subjective feelings
at all?
People have always been fascinated by
the
question
of
animal consciousness, because
both pets and wild animals arouse our admiration

and curiosity.
They tempt us to put ourselves
into their skins and imagine what their lives
are like.
But is this possible? Have students
of animal behavior learned enough to constrain
our
speculations
thoughts
and
constructively,
feelings
of
other
about
the
species?"
(Griffin, 1984, p.1.)

So begins the most recent addition to a series of articles
and books by Donald Griffin which constitute a crusade in
defence of the scientific validity of animal consciousness
(Griffin, 1976; 1978; 1982; 1984).
The author is
a
respected biologist, amongst whose achievements has been the
C.M.HEYES
108
original recognition and ana~ysis of echolocation in bats.

His view on animal conSC10usness has been published in
highly reputable sources and he is increasingly cited by
members of his target audience, cognitive ethologists.
This group is united by its interest in explaining, as well
as describing, the behaviour of animals under free-living
conditions, and by its disaffection with behaviourism as a
means to that end.
The passage quoted above introduces most of the
important elements of Griffin's programme: (i) Two questions
are addressed: Are animals conscious, and if so, what are
they conscious of?
(ii) In attempting to answer, attention
is not
confined to a
particular kind
of conscious
experience.
Griffin is concerned with both "thoughts and
subjective feelings" (inferential and perceptual beliefs),
and he often neglects to specify to which category or
sub-category of these he is referring.
(iii) Species from
almost any phylum are construed as serious candidates for
the attribution of consciousness, in a manner that would be
regarded as audacious by any interested philosopher or
scientist of the last century, with the possible exception
of Romanes (1883).
(iv) Griffin allies himself with the
layman, the affectionate pet owner and curious huntsman. He
calls for
a
re-examination of the
issue of animal
consciousness by
biologists and psychologists on the
grounds that their denial or disinterest is in conflict with
the layman's intuitive conviction (3). (v) Of the many kinds
of conscious experience about which Griffin is willing to
speculate, he proposes that "the ability to think about
objects and events, whether or not they are part of the
immediate situation", should be the focus of empirical
investigation.
His own research
largely consists of
accumulating instances of behaviour which can, in his
opinion, be plausibly explained as the outcome of this type
of conscious thinking.
Their hallmark is regarded as
adaptability to changing circumstances, because it is under
these conditions that we humans are most frequently and
intensely conscious of our mental activities.
For example (one chosen at random from a collection

which is impressive in the diversity
of species and
functional categories of behaviour that it samples), Griffin
regards the protective case building behaviour of caddis fly
larvae as 'suggestive of intentional action': He cites
Hansell (1972)
who studied a species of caddis fly,
Lepidostoma hirtum,
that constructs its hard outer covering
COGNISANCE Of CONSCIOUSNESS AND ANIMAL KNOWLEDGE
109
from pieces of leaf which it cuts into rectangles one to two

millimeters in size and binds together with silk secreted
from its own body.
The structure is strengthened by the
staggered arrangement of the pieces, each joint between two
at the side of the case intersecting with the middle of a
roof plate and vice versa.
When Hansell cut away the front
of a completed case, giving it a continuous smooth edge, the
larva would cut leaves into shapes which differed from those
used in the original process of construction, and glue them
into place restoring the staggered arrangement. It was this
flexibility which impressed Griffin.
Many of his other
examples concern communicative behaviour in animals. This
is judged to be important because it is principally through
language that we assess whether other people are conscious,
and learn about the contents of their phenomenological
world.
Griffin's criteria for
determining whether a
communicative act has an intentional component
necessary to
qualify it as language are informal in the extreme (4).

It is only with difficulty that one can decipher from
the morass of anecdotes and polemic the exact nature and
novelty of Griffin's claims.
Oakley (1985) has suggested
that Griffin's definition of consciousness is similar to his
own: "that consciousness can be defined, irrespective of its
sUbjective accompaniments, as synonymous with the use of
mapping
or representational strategies
of information
processing" (p. 148). If this were the case then Griffin's
claim that certain vertebrates are conscious would be
entirely compatible with current opinion in animal cognitive
psychology (discussed in more detail below), but I doubt the
accuracy of Oakley's interpretation. Dominant among his
attempts to express the meaning of 'consciousness' is
Griffin's appeal to layman's intuition. He cites dictionary
definitions that are based on such an understanding, and
endorses
Armstrong's
(1981)
suggestion
that we can
comprehend the meaning of 'consciousness' by contrasting our
states of mind in relation to the actions involved in
driving a car when we are learning, and later when the skill
has been thoroughly acquired.
This strongly suggests that
Griffin is conventional among cognitive psychologists in
referring to "the use of mapping and representational
strategies of information processing" as 'thought', and
reserving the terms 'conscious' and 'conscious thinking' to
describe that which has "subjective accompaniments".
110
C. M.HEYES
In a further attempt to rationalise his views it might

be suggested that Griffin's frequent failure to discriminate
among sub-categories
of conscious
experience magnifies the
member to understand
his companions'
moods, intentions and
that
are
then
controversial quality of his claims.

If he were asserting
that species from many phyla experience pleasure and pain,
but that only certain birds and mammals have conscious
inferential beliefs, then his position would be similar to
that of Lorenz (1963), Churchland or Pollock. Church land
(1979) distinguishes between feeling and jUdging that one
feels, and permits that almost any animal is capable of the
former; while Pollock (1985) assumes that humans are unique
only in their possession of third order "introspective
sensors that sense the operation of some aspects of their
cognitive machinery".
It is clear that Griffin's views are
not compatibly modest when we consider his response to
Humphrey's (1978) thesis that consciousness may have arisen
in human evolution when social groups reached a size and
degree of interdependence that made it important for each
thoughts.
Agreeing that this
describes an important
function of consciousness,
Griffin contests Humphrey's
conclusion that these capacities are confined to certain
primates, and argues that since they would also be very
useful to social insects one may expect to find them there
also.
No, Griffin's claims for consciousness in invertebrate
species are
not just a consequence
of idiosyncratic
semantics.
They result from
the
rejection
of two
traditional 'rules' for the attribution of consciousness.
The first is the 'similar brains' argument which runs
roughly as follows.
Assuming some form of a materialist
theory
of
mind
which
connects
or
identifies
neurophysiological events with conscious experience, and
we
as
humans
conscious,
mind
should be
attributed to others to the extent that their brains

resemble our own.
Griffin challenges this specific version
of the argument from analogy on the grounds that too little
is known about the relationship between neurophysiological
and mental events to eliminate the possibility that a wide
variety of different neural mechanisms could result in (to
use his emergent-materialistic language) the same conscious
experience.
Thus, he could be said to adhere to a 'similar
nervous tissue', but not a 'similar brains' rule, and within
this scheme finds himself able to grant that even ants may
be capable of forming inferential beliefs.
III
A second break with tradition is marked by Griffin's

assertion that instinctive behaviours may have conscious
accompaniments.
He attributes the denial of such an
association to selection of inappropriate introspections
from which to reason by analogy.
To be sure, we are not
always aware of blinking but why, asks Griffin, should we
treat blinking as a paradigmatic human instinctive act?
Some of the complex acts of which we are fully aware may be
instinctive even though we have failed to identify them as
such because of technical difficulties involved in teasing
apart nature and nurture.
Taking a different tack, Griffin
tries to persuade his audience that even if complex human
behaviours are not instinctive, those of other species might
well be, and concludes that the denial of consciously sensed
instinctive
behaviours might
be
based
on 'unwisely
anthropocentric' reasoning.
The goal of Griffin's progr~e is not to prove that

animals are conscious, but to conV1nce students of animal
behaviour that the burden of proof should fall upon those
who deny it; that the existence of animal consciousness
should be the null hypothesis.
Of course, this is an
attempt to reverse a bias which many believe to have been
established by Lloyd Morgan's Canon:
"In no
In
case may
we interpret an action
as the
outcome of the exercise of a higher psychically

faculty,
if it can be interpreted as the
exercise of one which stands lower on the
psychological
scale."
(Lloyd Morgan, 1894,
p.48).
fact, Lloyd Morgan, and even Thorndike in his early
writings,
expressed opinions on animal consciousness not at
all dissimilar from Griffin's.

The latter allies himself
with Darwin and Romanes in their use of anecdotal evidence,
consideration of subjective feelings as well as thoughts,
and belief that instinctive acts could have conscious
accompaniments.
Indeed, in his more speculative moments
Griffin is their contemporary representative, but when he
gets down to identifying the most interesting and persuasive
evidence
for animal consciousness
he homes in on
the same
sort of behaviour as Lloyd Morgan (and, for that matter, as

McDougall & McDougall, 1931, and Tolman, 1926), that is, the
non-habitual utilisation of previous experience. The fate
of the line of enquiry which Griffin seeks to resurrect
after some eighty years was sealed, not by Lloyd Morgan, but
C.M.HEYES
112
by the strict emp1r1c1sm of Watson and his followers. Is

the climate now ripe for Griffin's approach to succeed where
that of the early comparative psychologists failed?
Not surprisingly, Griffin thinks that it is. In a
tract entitled "The Issue of Falsification", he acknowledges
his share in the predicament of underdetermination which
Quine (1963) and others have convinced us is universal:
"I have suggested in earlier chapters that
versatile coping with new challenges provides
suggestive evidence of conscious thinking, but
in every case a behaviorist can argue that a
completely unconscious organism could behave in
the
same
effectively
adaptable
fashion."
(Griffin, 1984, p. 208).
His grounds for preferring the attribution of consciousness
under these conditions appear to be threefold: (i) It is
highly plausible or intuitively true; (ii) "extensive
and
mutually reinforcing evidence"
is accumulating in its
favour, and (iii) other hypotheses, like those concerning
evolutionary lines of descent, are regarded as both useful
and respectable even though they cannot be definitively
confirmed or falsified by current methods.
Concerning the first of these, the problem is that
Griffin's thesis is too plausible. In their classic work
Heider and Simmel (1944) demonstrated that intentions and
awareness are often attributed to filmed neutral objects,
such as squares
and
triangles,
exhibiting patterned
movement.
Similar
attributions
have
been
observed with
respect to natural disasters, sporting events, gambling, and

in a host of other circumstances where the objects of these
attributions could not possibly be conscious within any
coherent sense of the term (Felson and Gmelch, 1979; Lewis,
1963).
These data may well lend weight to some variant on
the thesis
that
the
evolutionary
or1g1n
of human
consciousness lies in the necessity to predict environmental
contingencies (e.g.
Humphrey, 1982), but it should make us
cautious about projecting our experience onto members of
other species. Griffin's empirical programme is utterly
impractical because it does not provide the scientist with
principles by which to exercise that caution.
He gives an
example of what he regards as "unwisely anthropocentric"
reasoning, but no clue about the general criteria that could
be
applied
in making other,
similar judgements; no
indication of when the human case should be the metaphier
and when the metaphrand (Jaynes, 1976).
II3
Griffin's second reason for preferring the attribution

of consciousness could be regarded as a claim that his
theory is robust (Campbell & Fiske, 1959; Wimsatt, 1981a);
that the
accumulating
evidence has been derived by
different methods, based on independent sets of assumptions.
In fact, all the data which Griffin counts as evidence does
so by virtue of the single assumption that an animal is
conscious at a given time to the extent that its behaviour
at that time resembles the conscious activity of a person.
To be sure,
Griffin uses a variety of
criteria of
behavioural
similarity,
but
these
do
not generate
inde~endent
assumptions.
Apparently
aware
of
the
requ1rement for multiple determination, Griffin discusses
the development of techniques for measuring the electrical
activity of the brain with enthusiasm.
Of particular
interest is the work on the human event related potential
(ERP) P300, (Donchin, 1981; Galambos and Hillyard, 1981).
These
experiments
suggest
an
association
between
the
magnitude and latency of P300 potentials and the extent to

which the task performed by the subject involves a complex
conceptual
judgement.
For example,
if subjects are
presented with the name 'Nancy' 20% and 'David' 80% of the
time, and asked to count the number of presentations in the
less frequent category, then the latency difference between
P300 waves in response to frequent and rare stimuli is less
than when the same task is being performed with a range of
girl's names as the 20% category and a range of boy's names
as the 80% category.
However, until an attempt is made to
use this technique and others like it to discriminate
occasions on which a person is consciously thinking from
those on which he is not, and to solve the difficult problem
of finding analogues of the relevant ERP's in other species,
this route to independent assessment of Griffin's claim will
remain inaccessible.
Consideration of Griffin's third reason suggests a

lack of coherence in his approach that even the most
sophisticated electrical recording could not remedy. In
view of the acrimonious debate over adaptationism, the
choice of hypotheses concerning lines of descent as examples
of those which are untestable and yet respectable was a poor
one.
However, in contrast with Griffin's, these are
supported within a sound conceptual framework, that of
evolution by natural selection.
Under the umbrella of
"emergent materialism"
(the nature of
which is left
unspecified), Griffin's 'common-sense' approach attributes
C.M.HEYES
114
causal force to conscious experience.

The way in which he
expresses his views on "the adaptive economy of conscious
thinking" makes this interpretation inescapable. He argues
that
animals
living in diverse
or rapidly changing
environments are likely to be at a selective advantage if
they are able to learn implicitly, to form "concepts and
generalizations" and to consider the likely outcome of
various actions without interacting directly with the world.
This argument
is,
of course, familiar to biological
evolutionary epistemologists (e.g.
Popper, 1972) and forms
the bedrock of animal
cognitive psychology.
However,
Griffin believes that implicit learning may be adaptive by
virtue of its conscious nature, and seeks to defend this
claim in the following way.
"We (humans) perform innumerable complex actions
rapidly,
skilfully
and efficiently without
conscious thought. From this evidence many have
argued that an animal does not need to think
consciously to weigh the costs and benefits of
various activities.
Yet when we acquire a new
skill,
we
have
to
pay
careful
conscious
attention to details not yet mastered. Insofar

as this analogy to our own situation is valid,
it seems plausible that when an animal faces new
and difficult challenges, and when the stakes
are high - often literally a matter of life and
death
conscious evaluation may have real
advantages." (Griffin, 1984, p.41.)
Talking about the function of consciousness
does not
necessarily commit the speaker to Cartesian metaphysics.
For example, in the context of a cybernetic analysis of
cognitive
processes,
Shallice
(1972)
identifies
consciousness with the selector input to the dominant action
system.
When he suggests that the two properties of this
input are "the dual functions of consciousness", he makes it
clear that this is a figure of speech which combines
different levels of analysis.
Griffin recently had an
opportunity to clarify his position on the causal efficacy
of consciousness when,
in an
exchange
of
'Letters to the
Editor',
Harnad (1985)
asked him:
"As
to studying
consciousness, how is one to study an entity that (unlike
say, quarks) is causally superfluous in any theory?"
If a disclaimer such as Shallice's were available to
Griffin, surely this would have been the occasion on which
to use it.
Instead, Griffin mistook Harnad's challenge for
115
an allegation that he (Griffin) denies that consciousness

has physical causes, and reaffirmed his belief that it also
has physical effects, using an analogy which so completely
eschews the mind-body problem that we are given no hint as
to how his claim could be true:
"Some of our conscious thinking (and perhaps
that of other species) entails comparisons of,
and choices among,
the likely outcomes of
various actions. Such thinking may well be no
more causally superfluous than contractions of
cardiac muscle are superfluous for the transport
of oxygen from lung to brain.
Of course
thoughts,
hearts and
hemoglobin
have all
resulted from antecedent causes. But why should
this be selectively taken as
disparage
the
significance
thinking?"
a
of
reason to
conscious
to ask Griffin:
'Why, under some conditions
should we flout scientific tradition and say of an animal
that it has a certain belief?', he would ultimately reply
'Because it is true'. A proposition of this kind might be
acceptable if it had either empirical force or conceptual
coherence, but in failing to demonstrate either, Griffin's
If one were
research does not, even in its own terms, provide

reasons for the investigation of animal consciousness.
good
The response of some philosophers to the problems

exemplified by Griffin's programme has been to reject the
concept of consciousness as otiose in the study of animal
knowledge,
and to
render
the
existence
of animal
consciousness impossible by definition (e.g. Carloye, 1985).
This is not Daniel Dennett's solution; if he were asked:
'Why should we say of an animal that it has a certain
belief?', he would probably reply: 'Because, under some
conditions, it is useful'.
3. CONSCIOUSNESS AS AN INSTRUMENT
Some philosophers
reading
the
impatient either with Griffin,
foregoing
or with my
may
have become
fairly ponderous
discussion of his views.

They will have immediately
recognised many of the objections to Griffin's thesis as old
chestnuts of the 'other minds' debate, and wondered whether
they warrant yet another airing.
I would defend myself by
pointing
out that,
in view
of
the
rising
popularity of
C.M.HEYES
116
Griffin's
issues
approach among biologists and
may not
be
so
familiar
to
psychologists, the
specialists
in other
fields,
and
that the
problems associated
with the
attribution of mind to other persons and to other species
are not completely co-extensive.
However, I heed the
drumming fingers and hope that their pace will slacken as I
make a few comments on Dennett's recent attempt to introduce
some conceptual discipline into the proceedings.
Dennett (1983) addresses himself to the same audience
as Griffin: "the new ethologists [who], having cast off the
straightjacket of behaviourism and kicked off its weighty
overshoes,
are looking about
somewhat insecurely for
something presentable to wear". The latter frequently cites
Dennett as the provider of a philosophical imprimatur for
his views, but (and this is reflected in the absence of
reciprocal citation) Dennett is arguing quite a different
point. The principal difference between the two authors is
that Dennett is apparently making an epistemological claim,
whereas Griffin is making an ontological claim about animal
consciousness.
Applying his theory of explanatory modes or
'stances' (elaborated in, for example, Brainstorms, 1978),
Dennett argues that it is useful under some conditions to
explain animal behaviour using the language of conscious
intention, that is, in sentences characterised by their
referential opacity (6) and apparently descriptive of mental
states
such
as
belief,
desire,
fear, recognition,
expectation etc. 'Apparently' is stressed because Dennett
emphatically denies
that if a
hypothesis couched in
intentional
terms
successfully
predicts
an animal's
behaviour, then that animal must experience the mental
states to which those terms commonly refer when they are
applied to humans. Ironically, this dissociation is so
fundamental to Dennett's argument (it is the means by which
he avoids the dualism inherent in Griffin's approach), that
it can be easily overlooked.
I would draw the attention of
those in doubt to Dennett's endorsement of the application
of the intentional stance to population genetics, embryology
and morphogenesis (1983, Footnote 17), and to one of the
passages in which he states his position explicitly:
" ... my point is that one should not confuse
the
predictive success
of the intentional
stance (in some domain) with confirmation of a
particular
representation
manipulation
hypothesis." (Dennett, 1983, p. 381, emphasis
in the original)
117
Despite the conceptual gulf that separates their claims,

the empirical programmes recommended by Griffin and Dennett
for cognitive ethology are similar
in many respects.
Dennett also emphasises the importance of accumulating
anecdotal reports of phyletically diverse animals' behavior
under free-living conditions, and does not regard this as a
means of disproving hypotheses which might be preferred by a
hypothetical behaviourist. However, while the documentation
of a range of behaviours that can
be
explained in
intentional terms is regarded by Griffin as sufficient
evidence in support of his claim, the same data are only a
starting point for the research Dennett recommends. He is
not content to 'eyeball' a piece of behaviour and decide
casually whether any intentional statement can describe it;
he wants cognitive ethologists
to generate hypotheses
concerning which intentional states, and what order of
intentionality, are appropriate.
To illustrate the kinds of hypotheses he has in mind,
Dennett presents the following alternative accounts of
Seyfarth, Cheney and Marler's (1980) observation that vervet
monkeys exhibit distinct cries in the presence of leopards,
eagles and snakes, and that conspecifics respond to these
sounds by climbing trees, looking up and looking down,
respectively. (The caller is named Tom, and the respondent,
Sam) .
Non-intentional account
Tom (like other vervet monkeys) is prone to
three flavors of anxiety or arousal; leopard
anxiety, eagle anxiety, and snake anxiety. Each
has its characteristic symptomatic vocalization.
The effects on others of these vocalizations
have a happy trend, but it is just tropism, in
both utterer and audience.
First order intention
Tom wants to cause Sam to run into the trees
(and
he has this
noise-making trick that
produces that effect; he uses the trick to
induce acertain response in Sam).
Second order intention
Tom wants Sam to believe that (i) there is a
leopard (ii) he should run into the trees.
C. M.HEYES
118
Third order intention

Tom wants Sam to believe that Tom wants Sam to
run into the trees. (7) (Ibid. p. 346)
It
seems to
such as these
that
"The
be very important to Dennett that hypotheses

should be empirically testable.
He stresses
question
is
empirical.
The
tactic
of
adopting the intentional stance is not a matter

of
replacing empirical investigations with
aprioristic
('armchair')
investigation, but of
using the
stance to
suggest
which brute
empirical questions to put to nature. We can
test the competing hypothesis by exploiting the
rationality assumption." (Ibid. p.347.)
In the case of the hypotheses listed above, Dennett's
commitment to this assumption (that the use of intentional
idioms carries a presupposition of rationality in the
creature or system to which the intentional states are
attributed) leads him to suggest that the non-intentional
explanation could be discounted if Tom silently climbed a
tree upon seeing a leopard when no conspecifics were around;
and that the second-order account would be preferred to the
third-order one if it could be demonstrated that Tom could
make Sam climb a tree by some means other than his cry, e.g.
by imitating the sound of a leopard growling.
There are a couple of things to note in connection
with these examples of the way in which empirical data can
affect the choice of intentional explanations. The first is
that Dennett does not imagine that any of the relevant types
of hypothesis can be readily swept away by 'crucial'
experiments;
the
behaviourist and
the romantic ethologist
alike could 'save' the data by postulating additional

contextual variables or complex beliefs.
In consequence
(and this is the second thing to note), Dennett proposes
that
we
augment
the
selection
procedure
with
a
meta-empirical principle of what I shall refer to as
'limited parsimony': when more than one explanation is
consistent with the data, prefer that which invokes a lower
order of intentionality.
"The mere fact that vervet monkeys apparently
have so few different things they can say holds
out little prospect for discovering any real
theoretical utility for such a fancy hypothesis
as a fourth-order candidate.
It is only in
contexts
or
societies in
which one
119
must rule
out (or
in)
such possibilities as irony,
metaphor,
story-telling,
and
illustration
('second-intention'
uses
of
words,
as
philosophers would say) that we must avail
ourselves of such high-powered interpretations."
(Ibid. p.347.)
From his instrumentalist position, Dennett could not be
advocating that we choose among hypotheses according to
their likely correspondence with events inside animals'
heads.
Rather, the process of selection, and the selection
criteria themselves
(rationality and simplicity) have
presumably been designed by Dennett to maximise the chances
of the intentional stance realising its potential as a
heuristic.
But what is its "theoretical utility" in
cognitive ethology?
Dennett is quite clear about this: its
purpose is to provide descriptions of real life animal
competences that will facilitate the tasks of the artificial
intelligence and information processing specialists. It is
incumbent upon them, and not the 'mentalist', to explain
behaviour in terms of causally active internal states (to
borrow a phrase from Patricia Churchland's commentary).
Dennett is not so clear about how intentional accounts will
assist AI and information processing theory; or rather, the
unique advantages of the intentional
stance are not
specified.
"The
intentional
profile
or
stance
characterization of an animal
or for that
matter, an inanimate system - can be viewed as
what engineers would call a set of specsspecifications for a device with a certain
overall
information-processing
competence.
An
intentional system profile says, roughly, what

information
must
be
receivable,
usable,
rememberable,
transmittable by the system."
(Ibid. p.349.)
Let us consider two kinds of epistemological claim that
Dennett could be making here.
The first is weaker: The
intentional stance is useful because it is, on the one hand,
a very familiar form of explanation owing to its ubiquitous
application in everyday life and, on the other, relatively
novel in the context of ethology. Therefore, it gives the
cognitive ethologist a 'handle' on the problem, it heightens
his motivation by making his research exciting. (In this
case it may be to the advantage of Dennett's cause for him
C.M.HEYES
120
to allow his instrumentalist position to be mistaken, so

that cognitive ethologists believe that they are really
discovering whether animals have mental states.) The result
of all this is that the cognitive ethologist churns out a
lot of data, fleshes out the descriptions of competences
that are already dimly known, and discovers new ones. The
intentional character of these descriptions is not important
in itself.
"Specs" that portray animals as rational and
simple are no more useful to the information processing
theorist than functional descriptions like those generated
by, for example, Sommerhoff's (1969) formal model of goal
attainment, or the computational theory Marr (1982) uses as
the first stage in his analysis of visual processing. On
this reading,
it
does not
matter whether Dennett's
rationality assumption is incorrect (as,
for example,
Elster (1983b) suggests (8, or whether his principle of
limited parsimony is without theoretical foundation. The
function of these criteria is merely to project upon
cognitive ethologists the illusion that they are real
scientists, involved in the explanation as well as the
description of behaviour, and thereby to increase their
output.
The second and stronger version of Dennett's claim is
more interesting (and more palatable to the intellectual
pride of cognitive ethologists), but if we are to accept it
as the version that Dennett would uphold,
then we must note
it reveals some omissions and inconsistencies in his

thesis.
The second possibility: In addition to providing a
'concert party for the troops' (the functions outlined
above), disciplined application of the intentional stance in
cognitive
ethology will yield
descriptions of animal
behaviour that are especially useful to the information
processing theorist. In telling the latter which behaviours
can be construed as rational and, among these, which mental
states and relations among those states are appropriate (in
accordance with a principle of limited parsimony), the
products of the intentional stance have unique advantages
over purely functional accounts.
The following passage
suggests that Dennett is, in fact, committed to this view.
that
"The intentional stance,
the
right
"black
box"
cognitive
interface
however, provides just
between
specialities:
characterization of behavioral and
competences observable
in the field,
121
but couched in language that (ideally) heavily

constrains the design of machinery to put in the
black box." (Ibid. p.350)
But how might intentional accounts "constrain" information
processing theories?
In
a
footnote
following this
quotation we are referred to "How to Study Consciousness
Empirically: Or, Nothing Comes to Mind" (Dennett, 1982)
for
"more detail"
on this point.
That article is
interesting because it indicates an important difference
between
Dennett's
recommendations
for
the empirical
analysis of human and animal consciousness (9), but it does
not begin to answer the question posed, for either group;
it merely re-expresses "( ... ) the hope ( ... ) that if one
a
clear,
detailed
and well-confirmed
can just get
description of
what is represented,
this will force
constraints on hypotheses about how the representing must be
done." (Dennett, 1982, p.196)
A significant omission lies in Dennett's failure to
discuss functional explanation; that which accounts for
behaviour in terms of goals and their means of attainment,
often referred to as 'teleonomic' (Pittendrigh, 1958) to
distinguish it from the teleological language of beliefs and
desires.
Teleonomy is the first port of call for any
student of animal behaviour who has become exasperated by
the complexity of appropriate causal/mechanistic/physical
stance description.
Contrasting intentional accounts only
with those of a hypothetical behaviourist, Dennett ignores
this important middle ground, such that we are uncertain as
to whether in the context of cognitive ethology functional
accounts are
equivalent,
or inferior, to intentional
accounts in their potential to describe what is represented
and, if the latter, precisely what it is that they lack.
There are some philosophers who take such a dim view of
intentional explanation in general, that one would expect
them to oppose Dennett and recommend functional accounts if
physical ones are not available. Rosenberg (1980) and
Churchland both regard explanations in terms of reasons and
actions as the outcome of a theory of persons (including
self) which is deeply entrenched in all human societies,
but which lacks predictive power in the context of everyday
life, let alone in science.
Churchland attributes this to
various flaws in the theory:
"Its comprehension both of practical and factual
reasoning is sketchy at best; the kinematics and
dynamics of emotions it provides is vague and
C.M.HEYES
122
superficial; the vicissitudes of perception
and
perceptual illusion are, in its terms, largely

mysterious; its comprehension of the learning
process is extraordinarily thin; and its grasp
of the nature and causes of mental illness is
almost ni1." (Churchland, 1979, p.114.)
This attitude
is
surprising
given
the naturalistic
inclinations of both Churchland and Rosenberg. It seems
unlikely, to paraphrase an oft repeated battle cry, that
natural selection would leave us with an habitual method of
construing conspecific behaviour that regularly deceives us!
This, however, is an aside (and a rather dangerous one since
it renders a complex issue in a very simple way). The point
is that neither Rosenberg nor Churchland would anticipate
that the intentional stance can provide a
basis for
predicting animal behaviour.
With the help of many ad hoc
hypotheses it may retrospectively satisfy our curiosity but,
in their opinion, it speaks not at all to the future. If
they were right about this, then the intentional stance may
have a less positive effect on the morale of ethologists
than Dennett anticipates, but the second version of his
claim, that intentional accounts are particularly amenable
to information processing analysis, could still be valid.
The development of Rosenberg's view conflicts with this
also.
In brief, Rosenberg (1980) argues that the terms of
any
potential
law-covering
intentional
explanations
(beliefs, desires etc.), cannot be defined independently of
that law. For him, the most significant consequence of this
is not the preclusion of non-circular tests of potential
laws, but the isolation of those laws from generalisations
of other kinds which might explain them, or be explained by
The other kinds of generalisation that Rosenberg
them.
considers concern physiological states of the brain, but his
argument equally implies that intentional terms are isolated
from information theory. The upshot of the argument is that
intentional explanation must be replaced, it cannot be used
as a pathway to, or be encompassed by, accounts that are
consistent with a materialist ontology.
These are complex issues to which I refer the reader,
but
as a
non-philosopher cannot
hope adequately to
penetrate.
However, as a cognitive ethologist and former
member of a research
group that tried to
take the
intentional stance seriously, I can comment on its more
practical aspects with greater authority.
As everyday
experience suggests, and experiments such as those of Heider
123
and Simmel (1944) confirm, it is easy to elaborate a variety

of intentional descriptions of a given behaviour. Dennett
asks cognitive ethologists to select from the array those
which portray the animal's action as rational yet simple.
However, he neglects to tell us what he means by 'rational',
and places ill-specified conditions upon application of the
simplicity criterion. In view of the now inherent ambiguity
of the term 'rational' (e.g.
Black, 1985) it is hardly
adequate for Dennett (1983, p.345) to leave his audience to
deduce its meaning from his usage.
The likely outcome of
such detective work would be a construal of rational
behaviour merely as that which 'makes sense' to some
observer; a matter about which one could expect little
consensus.
The
ambiguity
becomes apparent as he
of
Dennett's
simplicity
discusses how one might
criterion
account for
the behaviour of bees which remove dead conspecifics from

the hive.
He points out that a behaviourist's explanation
in terms of response to oleic acid secreted by the corpse,
is much easier to grasp, easier for the AI specialist to
model, than an account in terms of the bee's recognition
that their sister is dead, belief that her remains are a
health hazard, and desire to avoid its deleterious effects.
If this is so, why is Dennett urging us to seek intentional
explanations in the first place? He may reply that they are
recommended only when the alternative
involving innumerable reflexes,
contextual
principle,
accounts are complex,
conditioned
responses, and
variables.
This
a
satisfying
repast in
but it is not very useful in practice - the very
area in which Dennett seeks to contribute.

Just as Griffin
failed to tell us under what circumstances the analogy from
introspection is valid, Dennett neglects to give us any
substantial guidelines as to when we should throw in the
towel on our attempts at mechanism. Those guidelines he does
provide are
mentioned
in
passing
and
without any
For
example,
he hints that
accompanying rationale.
behaviour, however complex, which is the result of extensive
laboratory training, may be less appropriately described in
intentional terms than novel and spontaneous behaviour
observed in the wild - why? Dennett seems to have imported
several assumptions or inferential rules-of-thumb (e.g.
Lloyd Morgan's Canon, and exclusivity of explanation in
terms of learning
and of intention)
from the early
comparative
psychologist's
consciousness,
and
set
realist
them
down
approach
intact
to
animal
within
his
124
C.M.HEYES
instrumentalist framework.
In their original context they
are comprehensible
in relation to their
history and
connection with other premises, even if we have changed our
views enough to see them as misguided.
Their role in
Dennett's approach is mysterious.
They might have been
assimilated
thoughtlessly,
or
designed to
make the
ethologist feel 'at home', in accordance with the first of
my versions of his claim.
In summary, Dennett's contribution provides a much
needed (and very entertaining) lesson on the topic of
scientific realism for cognitive ethologists and, pending a
more explicit account of how intentional descriptions are to
be unpacked into information processing models, it may
predict cooperation between two much divided disciplines.
However, in its present form it does not achieve its purpose
of alleviating our intellectual insecurity.
Indeed, it may
aggravate the
condition.
Having
been
assured that
intentional explanations are legitimate, we are uncertain as
to why this is the case.
Is it just that they will keep us
working, or do they have a more specific function? If,
ignoring our suspicion of a snub, we start generating
intentional explanations for a given behaviour, we soon
discover that they breed like toads and are beset by a
plague.
Remembering that the search for reductionist
explanations is no longer interesting, and that lower order
intentions are to be preferred to higher order ones, we are
in a quandary about where to begin culling the beasts. As in
our confusion we reach out for the axe of the rationality
assumption, we find that it was made of rubber all along,
and only put there to give us the illusion of strength. It
is perhaps at this moment that the cognitive ethologist
decides to hang up his field glasses, become a cognitive
psychologist, and have nothing further to do with talk about
consciousness or intention.
4. COMPROMISING ON CONSCIOUSNESS
This heading is not just an example of the alliterative
indulgence common among
those
who
employ
the term
'evolutionary epistemology'. As is usual in the final
section of an essay of this kind, I shall tentatively
recommend some compromises (more respectably described as
syntheses) among the programmes reviewed, to promote a more
vigorous approach to the study of animal knowledge. But
125
first,
will point out some compromises that may have
already been made by animal cognitive psychologists, which
are
in
relevant to the evolutionary
their work.
epistemologist's interest
These concern their
consciousness and evolutionary theory.
attitude
toward both
A simple outline of the character, and particularly

the origins, of animal cognitive psychology (ACP) is an
appropriate preface to discussion of both these issues. It
is an extension of human cognitive psychology (based, in
turn,
on information theory and cybernetics), which emerged
as an independent discipline around 1976 (Roitblat et al.,

1984). Subsequently it has probably replaced Skinnerian and
Hullian behaviourism as the dominant approach to the study
of animal behaviour, although the continued strength of the
latter approaches,
evidenced by the numerous journals
devoted exclusively to their data and comment, should not be
underestimated.
A distinguishing characteristic of ACP is
its emphasis upon the
subject's internal organisation
(described in terms of cognitive or symbolic schemata,
templates, maps, structures or representations) in the
determination of behaviour. Edward Tolman (1926, 1951 for a
collection)
who
used
the concept of
"cognitive maps", is
frequently cited as the father of the contemporary cognitive

approach
Hull in
to animal behaviour.
However,
he was similar to
his willingness to postulate unobservable entities

on the condition that they could be operationally defined,
and it is the rejection of this proviso that makes current
ACP seem more akin to its human counterpart than to Tolman's
early work. Day to day, ACPs tend to work in the laboratory
rather than the field, studying a very limited range of
species chosen largely for convenience of housing and
malntenance.
They examine learning and memory functions by
requiring their subjects to perform complex discrimination

and spatial orientation tasks.
Typically, the animals must
press a lever, peck a key or navigate their way through a
maze in order to obtain a food reward. Roitblat (1982)
provides examples of these experiments, and a convenient
description of the
aims of
ACP
which
involve the
specification of a number of features of an internal
representation: (i) Domain
the range of situations or
tasks in which it is used by the animal; (ii) Content - the
aspects of the world that are represented; (iii) Code - the
transformational rules underlying the mapping between the
features of the
representation and
the corresponding
C.M.HEYES
126
features of the represented world; (iv) Medium

the
physical instantiation of the representation;
and (v)
Dynamics - changes in the representation over time.
The following remarks about ACP's attitude toward
consciousness relate principally to the purported codes and
media of representations. The 'party line', contrary to the
views of both Griffin and Dennett, is that ACP has no need
of consciousness as either an object or an instrument of
investigation.
To illustrate: introducing a recent edited
volume on animal cognition, Terrace (1984) states as a
sub-heading that "Animal cognition is not the study of
animal consciousness",
and
goes
on:
"Both
in
human and
animal cognition it is assumed that the normal state of

affairs is unconscious activity and thought". Similarly, in
concluding his book which reviews ACP, Walker (1983) says of
consciousness: "This is a word I have tended to avoid

because it is used in several, distinctly different ways".
While
sympathetic
to
Walker's
confusion
and
resulting
failure to refer to consciousness, we note that he does not

regard his discussion as deficient in consequence. If, as
these
authors
suggest,
representation is not the same
description
as,
of
an
internal
or even derived from, a
description of the contents of consciousness, then what is

it?
The standard but nonetheless controversial reply
resembles that of Chomsky (1980), i.e. it is an "abstract
characterization of the properties of certain physical
mechanisms".
In the case of ACP specifically, the abstract
language is that of information theory.
The first compromise that I will suggest to have been
effected within cognitive
psychology
can
be crudely
expressed in a research strategy: Profess complete lack of
interest in consciousness in order to avoid metaphysical
deep water, but use introspection as a model for explanatory
hypotheses because it a)
is easy,
b) renders those
explanations plausible, and c) apparently leaves open the
possibility of
a
synthesis
between
intentional and
information processing accounts of the operation of mind.
At the heart of this suggestion is the contentious issue of
the extent to which cognitive psychology is parasitic upon
conscious experience; a question which receives subtle and
vigorous discussion in a special edition of The Behavioral
and Brain Sciences (1980, 3:1). The contributors to that
volume sought to determine whether the codes postulated by
cognitivists
(mathematical,
verbal/propositional
and
imaginal) exhaust the conceivable possibilities, and whether
127
they are coherent descriptions of unconscious processes. In

short,
they asked whether cognitive psychology is in
principle based on a confusion among levels of explanation.
My aim is much more modest.
I wish merely to suggest that
some practitioners of cognitive psychology show signs that
they have become confused, and that the extension of the
cognitive approach to non-human species may cast a spotlight
on its potential for misinterpretation.
The confusions I
will point out are consistent with use of the compromise
strategy,
the
intricacies
of
which
can
be
easily
forgotten.
'Sahlins' fallacy' (Maynard Smith) exemplifies the way
in which
the
analogical role that the
products of
introspection play in the strategy can be mistaken.
"The central idea of kin selection is that a
gene A, causing an animal to be more likely to
perform an act X, may increase in frequency in a
population even if act X reduces the individual
fitness (expected number of offspring) of the
animal itself, provided that the act increases
the fitness of animals related to the actor.
Following Haldane, biologists refer to such acts
as
'altruistic'.
It
may require considerable
skill to calculate in just what circumstances

particular acts will evolve.
This has led some
people to commit what may conveniently be called
'Sahlins' fallacy', and to suppose that the
operation of kin selection requires that animals
or even
people,
are able to perform the
necessary calculations."
(Maynard Smith, 1984,
p. 63.)
As an anthropologist, Sahlins (1976) was interested in the

application (or rather, misapplication) of kin selection to
human societies.
If his attention had been focused instead
on animals, he may have been alerted to the possibility that
its language of conscious, mathematical calculation is
metaphorical; that it describes something going on in the
heads of scientists, and not necessarily in those of all

altruistic organisms.
As a means of illustrating Sahlins'
mistake, Maynard Smith cites an example of kin selection in
a bacterial plasmid, but he need not have gone to a species
remote from man in order to make his point.

since
there is a deep rooted assumption among
scientists that
animals are not capable of
complex propositional
or
so
mathematical
thought the application
to
other
species of
128
C.M.HEYES
information processing models formulated to account for

human competencies can illuminate the extent to which they
confound instrumental
and
realist
interpretations of
consciousness, and depend on this for their plausibility.
Animal cognitive
psychologists
enjoy
a greater
opportunity
to
discover
the
media
of
internal
representations than do their human counterparts.
The
relaxation
of ethical
constraints and the
speed of
generational turnover in many species allow them greater
access
to
neurophysiological
function
and
genetic
programming.
It may have been the lure of these rewards
that led Roitblat (1982) to exhibit a confusion between two
meanings of the term 'representation' that remains hidden in
the research of those dealing exclusively with human
behaviour.
As Dretske (1982) points out, Roitblat fails to
appreciate the distinction between "causal" and "information
theoretic" definitions of a representation.

When he
suggested that the size of a mantid's gut represents how
much it has eaten, Roitblat was using the former, in which
some internal state of an organism is a representation of
certain features of the environment by virtue of its having
been caused by them.
This is much easier to understand and
less controversial than the information - theoretic notion
which maintains an independence between abstract codes
and
physical
media
such that one
representation may be
instantiated by different physical substrates.
In the
latter case it is the functions which map features of an
abstract description onto features of the
world that
determine whether the description is a representation.
Thus,
the hypothetical ethologist who becomes a
convert to cognitive psychology in order to avoid the
complexities of dealing with consciousness might well be
disappointed.
In professing such a complex faith, the
members of his new brethren must constantly resist the
temptation to forget the abstract, contingent nature of the
relationship between their doctrines and reality, and to
reify them in the phenomenological and neurophysiological
worlds of their subjects.
However, the cardinal symptom of
conversion would be a decline in the erstwhile ethologist's
attention to evolutionary theory.
It is in their attempts
to specify the domain and content of representations that
ACP's ambiguous attitude toward this becomes apparent.
As honorary or practicing ethologists, both Dennett
and Griffin recommend the investigation of a wide range of
species under
naturalistic
conditions;
that
is, by
129
observation of free-living animals or using laboratory tasks

designed to reproduce the various conditions and demands of
natural ecology.
In contrast, the interest of ACP is
focused on the behaviour of rats, pigeons and chimpanzees,
in situations that bear at best an abstract resemblance to
the contexts of foraging and orientation behaviour in the
wild.
In
consequence,
the
purported
content
of
representations relates to features of
an artefactual
world (e.g.
the temporal pattern of illumination of a set
of colored response keys); and whole domains such as those
of predation,
habitat selection, maternal, sexual and
aggressive behaviour, are neglected.
The extent to which this research strategy betrays a
disregard for evolutionary theory is not easy to divine.
Both Dennett (1984b) and Boden (1984) have pointed out that
ACP and its close cousin AI are apt to generate models which
are biologically improbable. The former describes these as
'cognitive wheels', and a potential example is provided by
Ullman's (1979)
research
on visual
perception which
construes the detection of stationary form as operationally
prior to the detection of movement. But this is only a part
of the problem. In order for ACP to be compatible with
evolutionary theory (and therefore, with EE) it must not
only produce models that could be instantiated somewhere in
the animal kingdom; it must concern itself with the fact and
pattern of instantiation, dealing with diversity resulting
from phyletic and ecological variables.
Some animal cognitive
psychologists, perhaps the
majority, believe that a comprehensive account of animal
cognition can be achieved without reference to evolutionary
~heory.
Others acknowledge that it must eventually play an
1mportant role, and defend the specificity of their current
research on the grounds that it is necessary to develop a
deeper understanding of a few instances in
order to
interpret the
full
range
and
diversity
of animal
competences. In acknowledgment of this long-term goal, they
stress the ecological validity of their experiments; the
resemblance between a laboratory maze and the rat's burrow,
between a variable interval schedule of reinforcement and
natural resource availability. However, the question remains
as to how this future extension is to be achieved. Some
animal cogn1t1ve psychologists seem prepared to blindly
extrapolate
their
theories
to
other
species
and
behavioural
categories,
to
assume
their
underlying
generality in the face of manifest diversity.
C.M.HEYES
The uncertainty of ACP's commitment to evolutionary
thinking can be seen as the outcome of an historical
compromise.
Its ascendancy over behaviourism was achieved,
in part, through the intervention of what is known as the

'biological constraints' approach to the investigation of
animal learning.
In the 1960s and early 1970s a number of
studies (the most frequently cited being that of Garcia and
Koelling, 1966) demonstrated certain discontinuities in the
associative
tendencies
of
different
species
which
considerably weakened already failing confidence
in a
paradigm that had a general process view of learning as its
conceptual core.
These studies helped to give ACP a
'break', but it did not pursue in earnest their emphasis
upon adaptation and consequent diversity.
Instead, ACP
reacted against the empiricism of the behaviourist era,
while retaining many of its prejudices concerning the
generality of the principles of behaviour.
As a means of weaving together the threads of this
essay,
I will
summarise its conclusions and
outline their
implications for future compromises.

Both Dennett's and
Griffin's programmes contain two sorts of recommendations
for
students
of
animal
behaviour;
proposes
a manner
those
relating
to
communication
or
consciousness on the one hand and to evolution on the other.

In so far as each encourages the cooperation of ethology and
ACP
to
produce
penetrating
but
truly evolutionary
explanations of animal competences it is compatible with the
goals of biological EE.
Dennett seeks more actively than
Griffin to promote
this collaboration,
indeed
his
emphasis on
intentional description seems to result from
certain assumptions about the way in which the collaboration
will be effected.
He envisages a group
of
data
collectors (ethologists) delivering their observations
to
a group
of theorists
(cognitive
psychologists)
for
explanation,
and
of
delivery judged to be appropriate under these circumstances

(10). The last decade of research on birdsong acquisition
demonstrates
the
vigorous
potential
of
a
less
caste-ridden collaboration among cognitivists, ethologists
and neurochemists, in which intentional language plays no

part.
(See Kroodsma and Miller, 1982 for a review of the
birdsong literature.)
Regarding
consciousness,
the
two
authors'
prescriptions are rather different.
Griffin wants to know
which animal behaviours are associated with subjective
mental states, while Dennett is concerned with which can be
131
usefully described in those terms.

However, both authors
assume that there is a relevant negative set (that some
behaviours do not reflect mental states/cannot be described
usefully in intentional terms), and assign the task of
identifying these instances to the cognitive ethologist. It
is because neither author provides the criteria necessary
for making this discrimination in practice that I disagree
with Silverman (1983) and Burghardt (1985) who, reviewing
the same literature, concluded by endorsing application of
the concept of consciousness in the
study of animal
knowledge.
The application of consciousness to animals is
inevitably
a
process
of reasoning
by
analogy and
insufficient is known about the model, the human case, to
prevent its benefits from being outweighed by its costs.
Griffin's own research demonstrates that an analogy based on
behavioural resemblance can be extended to the very brink of
panpsychism, and his discussion of the 'similar brains'
argument is one of many which concedes that the contemporary
grasp of neurochemistry can do little to confine such
speCUlation.
Aware of the need for discipline (heightened
beyond that required in other areas of science by an
apparent
human
inappropriately),
predisposition
Dennett offers
to
attribute consciousness
some
specific guidelines
which, upon close examination, assume an arbitrary character
hardly justified by the unsubstantiated hope that they will

facilitate the task of the information theorist.
Although I see little current utility for the concept
of consciousness in the study of animals, I do not imagine,
as Carloye (1985) apparently does, that its associated
epistemological and ontological claims can be evaluated on
an a priori basis.
Reacting against the early excesses of
introspectionism
and
the
traditional
philosopher's
assumption that all thought is conscious, human cognitive
psychology
has
neglected
phenomena
as
objects
of
investigation in a manner which is startling given their
subjective,
intuitive significance.
In view of, for
example,
research
on
subliminal
perc:ption,
the
philosophical
prejUdice
need hardly
cont1nue
to be
influential, and surely in this post-post-positivist age
some of our strictures against introspectionism can be
relaxed. Attention to consciousness in human research, apart
from its intrinsic interest, would provide

basis for testing the claims for animal
reviewed above.
At present little more is
metaphier than about the metaphrand.
the necessary
consciousness
known about the
C.M.HEYES
have
Many
authors (e.g.
provided
questions
detailed
about
human
Silverman, 1983; Burghardt, 1985)

recommendations
consciousness
concerning
that
need
to
the
be
addressed.
These include examination of its ontogeny,
relationship with neural structure and function (using
electrical recording techniques and studies of brain damaged
subjects), and association with problem-solving behaviour.
The
studies
proposed
seek
a
correlation
between
neurophysiological or behavioural events and mental states,
assuming that the subject's verbal report indicates the

occurrence of the latter.
As a post-script, I wish briefly
to contrast Dennett's (1982) heterophenomenological approach
with these proposals, and to note one of its puzzling
features.
Throughout this essay it has been tacitly assumed that
there are two sets of facts to be discovered with respect to
consciousness: whether an animal or person experiences

sUbjective mental states, and whether their behaviour can
profitably
be described
in
those
terms.
Dennett's
metaphysics defies this
convenient distinction between
ontological and epistemological categories in claiming that
the usefulness of intentional language is the only issue to
be addressed; that the reality of mental states consists of
their
predictive
value
as
descriptive
device.
Thus,
within a heterophenomenological investigation, the subject's

verbal report does not reflect his or her mental state; in
some sense it is that state.
This claim is particularly difficult to comprehend
since both
the criteria that
Dennett recommends for
evaluating
intentional
hypotheses
concerning
animal
behaviour,
and the programme he suggests
for the empirical
analysis of human consciousness seem to be consistent with a

simple
realist
approach.
Surely
a
true
more
'heterophenomenology', one which allowed its human subjects
the
status
of
automata,
would
assign
intentional
descriptions (the procedure recommended for animal studies),
or at least compare the utility of those collected from the
subjects with those devised by the experimenter. In this
respect, studies of split-brain patients, hypnosis and
mUltiple personality disorders (e.g.
LeDoux et al., 1979;
Oakley & Eames, 1985, for an overview), are instructive in
their identification of conditions under which there is a
manifest disparity between the two accounts. In short, what
puzzles me is the following.
If there is no such thing as
privileged access to mental states, then why should a
subject's
receive
own
intentional
exclusive
description
consideration?
of
And,
133
their behavior
would
not
satisfactory answer to this question re-open the gulf

between the significance of intentional language describing
animal and human behaviour that Dennett has sought to close?
Whatever the resolution of these particular issues, a
contingent or critical realism is regarded by many as
fundamental to both of EE's enterprises, and to the extent
that Dennett makes students of animal and human behaviour
aware of its intricacies and demands, his
contribution to
the understanding of animal knowledge is of considerable
value.
ACKNOWLEDGEMENTS
I am grateful to Werner Callebaut, Don Campbell, Dave
Oakley, and Henry Plotkin for their guidance during the
preparation of this essay.
This research has been supported
by a Harkness
Fellowship of The Commonwealth Fund of New York.
NOTES
1.
Hull's (1982) tripartite classification contrasts with
those cited in recognising that attempts "to present a
general analysis of evolution through selection processes
which applies equally to biological, social and conceptual
evolution" (p.275) may represent an alternative strategy to
those of reasoning by analogy from, or literal extension of,
neo-Darwinian theory.
This addition provides an important
clarification
of the
motives
of
those evolutionary
epistemologists whose
work is regarded by
Bradie as
anomalous in the context of his typology, and thereby
reminds us that the
explanation
of animal cognitive
processes does not always depend on literal extension of
evolutionary theory.
Whatever its analytic strategy, the
elaboration of a powerful biological EE demands conceptually
sound descriptions of the competences of a broad range of
species, and some factors affecting the supply of these raw
data are the subject of this essay.
2.
I have chosen to use the term 'animal knowledge'
for two reasons: First, surpr1s1ng as it may seem to the
philosopher, 'knowledge' is a relatively neutral term in the
context of the literature I will be discussing. Had I opted
134
C.M.HEYES
for an alternative such as 'cognition', 'information',

'intelligence' or 'computation' I would have implicitly
allied my views with a particular school of thought, in a
fashion
incompatible
with
my
assumed
role
as an
intermediary. However, there are other terms which resist
association with particular conceptual and methodological
frameworks (Walker (1983) finding himself in a predicament
similar to mine, settled on the term 'thinking'), so my
choice has further significance: 'animal knowledge' seems to
me to encapsulate the goals of biological EE. I respect the
tradition which specifies that knowledge equals justified
true belief, recognise that the formula is in the process of
being revised
by epistemologists with
a naturalistic
inclination, and anticipate that the product of their
endeavours, having been informed by animal research, will
permit the evaluation of both human and non-human knowledge
claims.
If the reader requires further evidence of good
faith in this regard, I refer them to a forthcoming essay
discussing the potential of some contemporary externalist
theories of justification for evaluation of the animal
beliefs revealed by laboratory experiment.
3.
Studies by Kellert and Berry (1980) and Burghardt
(1985) go some way toward quantifying the extent of this
conviction.
4.
Tennant
(1984)
exemplifies a more rigorous
approach.
This paper is also pertinent in it recognition
that, owing to their lack of ecological validity and an
asymmetry
in
the
logic
of
deriving
evidence for
intentionality from behaviour, ape language experiments
cannot be used to deny second-order intentionality to the
species concerned.
5.
Refer to Nagel (1974) for the most elegant
exposition of the problem of underdetermination as it
applies to the attribution of animal mind.
6.
A sentence with the property of referential opacity

least one term which,
if it were replaced by
another term referring to exactly the same object, could
result in an alteration of the sentence's truth value.
contains at
135
7. Apologies for such a long quotation, but nobody can

do this sort
of
thing
better
than
a philosopher
(particularly this one), and it is important for the reader
to get a flavour of the kind of explanation Dennett himself
is recommending.
8.
Elster identifies consistency of desires and of
beliefs as a
minimally
necessary
condition for the
attribution of rationality, and argues for the partial
independence of rationality and
intentionality on the
grounds that there are comprehensible intentions that are
based on beliefs which could not all be true and be
believed.
The example he gives concerns a man who hangs a
horse shoe on his door because he believes (inconsistently)
that it will bring luck even to those who do not believe in
its magical properties. A hypothetical example involving an
animal can be drawn from Premack and Woodruff's (1978)
research on deception by the female chimpanzee Sadie. Sadie
believes (i) that a malevolent trainer requires her signal
to decide which of two boxes contains food, and (ii) that
the malevolent trainer can see through the sides of the
boxes to their contents. Sadie pointed to the 'wrong' box
because she wanted the banana and believed (inconsistently)
that the malevolent trainer required her signal to decide
which box contained the food.
If intentional explanations
are plausible even when they portray
the
animal as
irrational, then Dennett's insistence that hypotheses be
selected on this basis loses its foundation.
9. The difference between the investigation of animals
and humans within Dennett's scheme is associated with the
process of procuring a description of "what is represented",
In the human case, this description is provided by the
subject in the form of a verbal report (apparently) on their
conscious experience.
Taking this report seriously is part
of what Dennett calls 'heterophenomenology' because it does
not require the investigator to assume that the subject is,
in fact, conscious; and it consists of asking what must
really be going on inside the person's head for them to
represent it in this way.
Thus, with regard to persons, an
intentional description of a behavioural competence is, in
addition to the competence itself, an important part of the
explicans.
Judging
by the
1983
paper, Dennett is
recommending
something
rather different
to cognitive
ethologists. It is they who must generate the intentional
136
C.M.HEYES
description of what is represented,

not their animal
subjects.
Consequently, the description itself is merely a
heuristic
device (if
not
an explicandum),
and the
behavioural competence stands alone
as that requiring
explanation.
10.
It
is
not
uncommon
for
naturalistic
epistemologists to see the future in terms of a rigid
division of labour
between experimentalists (erstwhile
scientists) and theorists (former philosophers).
Quine
(1969a) provides another example.
Part II:
Evolutionary approaches to science and

technology
SELECTION THEORY AND THE SOCIOLOGY

OF SCIENTIFIC VALIDITY
Donald T. Campbell
Lehigh University
1. INTRODUCTION
So often have I reviewed the 150 year old heresy now called
"evolutionary epistemology" that the harried and hurried
reader needs to be told
redundant update,
something new.
whether this essay
is just another
or whether I, at least, feel that it says
While still a minority position within philosophy, the

field has grown beyond essay-length summary.
Our
600
item bibliography (Campbell,
Heyes, &
Callebaut, this volume) may help someone someday do a
book-length integration, but it is more likely to make the
task seem impossible. Each of the dozen full-length books on
the subject cover ~ut very narrow portions of the field.
Thus this essay will not follow its predecessors (Campbell,
1959, 1974a, 1974b) in attempting an entree into the whole
literature.
What then is claimed to be new? May I include some
hints that need follow-up by more diligent scholars? May I
call attention to important developments I've had nothing to
do with?
1. Convergence seems near at hand between the
corrigible
justificationism
of
mainstream AngloAmerican analytic philosophy and
that predominant
variety of EE which passes the justificatory buck to
biological evolution.
2. EE has been extended to cover our intuitions
about rational inference.
3. Those of us who use natural-selectionist
analogues in describing the development of scientific
beliefs are now so numerous and well-placed that I
feel we can politically afford internal dissension: I
use this freedom to reject our majority, who so
closely model biological evolution that they end up
with
an
erroneous
emphasis
on
continuity
in
intellectual genealogies. My use of "Selection Theory"

in the title flags this point.
139
W Callebaut and R. Pinxten (eds.), Evolutionary r./Jisremology. 139-158.
/987 hy D. Reidel Publishing Company.
D. T. CAMPBELL
140
In
all
adaptationism
4.
must
versions
the
construction
be
of
avoided.
EE,
Panglossian
Nor must we claim to
have answered the skeptics.

5.
While passing the justificatory buck to
biological evolution may do useful epistemological work
for visually generated beliefs about ordinary physical
objects (as in point 1 above), this does not extend to
scientific beliefs.
The limitation can itself be
understood in terms of the biological evolution of
species in which there is genetic competition among
the cooperators. Instead, Selection Theory must provide
a plausible sociology of belief selection in science in
which "the nature of the physical world" might affect
social
of
consensuses
among
physicists.
6. (An epicyclical solution to a problem created
by
my dogmatic adherence
to
Selection Theory):
Selection Theory
emphasizes
the
role
of "retention"
(and hence tradition) fully as much as variation and

selection. Yet the sociology of scientific validity
implicit in the ideology of the scientific revolution
is vigorously anti-traditional. This inconsistency is
explained away by asserting that in cultural evolution,
the domain of beliefs about invisible physical reality
had been preempted for social solidarity functions. If
literal referential competence
in this domain
were to
be achieved, the cultural heritage of such beliefs had

to be rejected.
1. BIOLOGICAL-EVOLUTIONARY EPISTEMOLOGY
(AND MODERN ANALYTIC THEORIES OF JUSTIFICATION)
All evolutionary epistemologists find biological evolution
via natural selection
epistemologically relevant. Most
evolutionary epistemologists go no further, and reject, or
are
non-commital
on,
natural
selection
analogues
in
understanding the history of science.

Quine's
(1969,
1973,
1975)
"epistemology
naturalized,"
uses only biological evolution;
so too
Shimony's
(1970)
"Copernican
epistemology,"
so
too
Churchland (1979), Dennett (1971, 1978), Lorenz (1977),
Piaget (1971, 1980), Plotkin (1982), Rescher (1977), Riedl
(1984b), Rosenberg (1981), Sober (1981a, 1981b), Vollmer
(1975), Wuketits (1983b), Yilmaz (1973), and many others.
SELECTION THEORY AND THE SOCIOLOOY OF SCIENTIFIC VALIDITY
141
For most of these uses of biological evolution, vision

is the prototypical example, with an argument that can be
crudely phrased: if our eyes had misled us, we would not
have survived. Natural selection has instead continually
selected for visual competence.
An important development is
the extension of this argument to rationality itself. Burks

(1977) briefly invokes such an argument for the grounding of
the principles of decision making at several points in his
important comprehensive theory of rational choice (e.g. p.
613). Ellis (1979) gives a greater, albeit more tentative
emphasis: "And the fact that we have this desire and these
intuitions supports the hypothesis that being rational
according to these intuitions has some survival value. For
it would be difficult to explain their existence, given the
scientific view of man, on any other hypothesis" (1979, p.
4-5). "I do not believe that it is just an accidental
by-product of human evolution that we have these ideals
( ... ) Our rationality has almost certainly been a major
factor in our survival as a
species
( ... ) probably
genetically determined. But I do not yet know how Our ideals
of
rationality contribute
to
our survival
as
a species"
(1979, p. 12). Sober (1981a) and Sosa (1983) further extend

the argument.
When biological evolution is invoked in this way for a
philosophical, epistemological argument, there has been no
answer to the skeptics. There has been instead a "begging
of the question of induction."
A product of science
(evolutionary theory) has been used to explain, or even
"justify"
the efficacy of SC1ence.
Quine
has been
especially explicit on this: "( ... ) our question, 'Why is
science so successful' ( ... ) [is to bel taken ( ... ) as a
scientific
question,
open
to
investigation
by
natural
science itself. ( ... ) Individuals whose similarity groupings

conduce largely to true expectations have a good chance of
finding food and avoiding predators, and so a good chance of
living to reproduce their kind. ( ... ) I am not appealing to
Darwinian biology to justify induction. This would be
circular, since biological knowledge depends upon induction.
Rather, I am granting the efficacy of induction, and then
observing that Darwinian biology, if true, helps explain why
induction is so efficacious as it is." (Quine, 1975, p. 70).
"For all of their fallibility,
our innate similarity
standards are indispensable to science as an entering wedge.
They
continue
advances.
to
be
indispensable,
For the advance of
moreover, as science
science depends
on continued
142
D. T. CAMPBELL
observation, continued checking of predictions. And there at

the observational level, the unsophisticated similarity
standards of common sense remain in force." (1975, p. 71).
It is this shift to doing a science of knowing which
make the majority of epistemologists reject the relevance of
all
naturalized
and
biologically-evolutionary
"epistemologies" to the philosophical problem. However,
except for the few modern skeptics, this rejection is
inconsistent with their own positions. Beginning with G. E.
Moore at least, mainstream Anglo-American epistemologists
propose that we start the epistemological quest not with
total doubt, but rather by assuming that we have a great
deal of "knowledge", and that the epistemologist's task is
to explain how such "justified true belief" can be possible.
In the initial assumption of abundant "knowledge", the
problem of induction has been begged.
Various contemporary theories of justification further
illustrate this. Belief is, for some, justified if there are
"good reasons"
for believing, or if belief is "more
rational" than doubt. The "good reasons" and grounds of
rational choice, when followed up in specific instances, all
depend in part on trusted "knowledge." "Reliability theory"
(e.g. Goldman, 1975, 1979) regards belief as justified if
it is the product of "generally reliable" processes, and
thereby
assumes
that
the
validity
(and occasional
invalidity) of prior beliefs produced by the same processes
has
been ascertained.
Harman's justification by "inference
to the best explanation" also turns out to depend upon

assuming other knowledge in making the choice.
Even the modern "foundationalists" such as Chisholm
(1982) and Alston (1976) can be included. They argue that
when one has pushed a justificatory inquiry ("why do you
believe that?") back to "I saw it with my own eyes," it
makes no sense to inquire further. In Alston's terms,
perceptually based beliefs are "self-validating". But none
of these "foundationalists" claim that such justification
produces certainty, or is indefeasible. This all remains a
psychology of justificatory discourse and rationalization
unless one assumes that the beliefs supported
by vision
are generally valid, thereby using induction to justify
induction.
At this point, one can assimilate biological EE into

the modern analytic justification program of "knowledge
equals justified true belief," by arguing that perceptual
judgements based upon "normal" sense organs, operating in
SELECTION THEORY AND THE SOCIOLOGY OF SCIENTIFIC VALIDITY
ecologies
"typical ll
of
biological
143
evolution,
are
justified by their general reliability.

(Goldman [1975,
1979] for one, has made this move, as also has Rescher
[1977], in his "Methcds Darwinism".) So interpreted, the
evolutionary model seems to push the argument
back one more
step, by providing a theory as to

objects given in perception should be
why beliefs in the

in general reliable.
the skeptics'
induction
This does
assumed
provide
not,
in
of course,
provide any
charge that successful
the justification of
richer
grounds
for
better
induction.
answer
has
But
further
it
to
been
may
elaboration,
particularly for those (such as Goldman) invoking a causal

theory of belief and/or ~erception: For example,
the
evolutionary
model,
with
lts attentlon to
past selective
ecologies, provides theoretical grounds for predicting

explaining
beliefs
the
occasions
are misleading.
limited,
on
which
perceptually
Biological emphases
neuro-perceptual
machinery,
and
and
given
upon finite,
upon
metabolic
costs, are also relevant here (Campbell, 1986c).
2. EVOLUTIONARY THEORIES OF SCIENCE

EMPHASIZING GENEALOGICAL CONTINUITY
This type differs from Selection Theory, below, mainly in
the more complete analogy with biological natural selection
which is employed. For Selection Theory, Darwin's natural
selection
inspires
an
retention-and-reproduction"
examples of increased "fit"
abstract
algorithm
"variation-selective-
appropriate
to
all
between one system and another.
Biological
evolution
is
only
one
such
exemplar.
Trial-and-error learning, radar, sonar, computerized problem
solving, and human thought are others. Biological evolution
has specifics which may be biologically necessary but not
part of the abstract algorithm. Many theories of scientific
evolution retain in their analog irrelevant biological
specifics.
A substantial
class
of the
critics
of
evolutionary theories of scientific progress (e.g. Cohen,
1973; Kary, 1982; Putnam, i 983) are criticizing theories of
this type in ways which do not apply to Selection Theory.
While Q'Hear's (1984) mistaken expectation that evolutionary
epistemologies must assume that scientific beliefs are
selected
accurately
scientific
by
contribution
to
human
survival
does
not
describe any actual

evolutionary theory of
development, it is more clearly disarmed by
Selection Theory.
D. T. CAMPBELL
144
In my judgment, the most frequently misborrowed aspect

of biological evolution is its genealogical rigidity. With
some trivial exceptions (Schell & De Waele, this volume;
Lewin, 1985), and subsequent to the great symbiosis founding
the eukaryotes, adaptive advances in biological evolution
have precluded the route of cross-lineage borrowing. But
this is not an essential feature of the V, S, R, & R
formula.
Indeed,
in
cultural
evolution,
selective
cross-lineage borrowing is a major process. This disanalogy
with biological evolution has been noted from the very first
efforts to achieve a formal theory of cultural evolution
(e.g.
Childe, 1951; Ginsberg, 1961; Waddington, 1961;
Campbell, 1965; Boyd & Richerson, 1985).
The intellectual-genealogy emphasis
-"Who studied
with whom?"- fits in naturally with standard practices in
the history of science.
Judging that we evolutionary
epistemologists are now strong enough politically to afford
quarrels among ourselves, I challenge several of my closest
allies
to
overcome their
excessive genealogical emphasis.
This includes Toulmin (of 1972 and 1981, but not 1967),
Richards (1981), and Hull (1982, 1983). Along with this
goes a failure to specify how the differential prospering of
lineages could be due to the greater validity of the beliefs
transmitted (rather
than,
for
example, to political
centrality in high-prestige graduate training programs,
the acceptability of their beliefs
to extrascientific
powers, culture, etc). For me, the crucial advantage of
natural-selection analogs is the argument that the referents
of the beliefs
(e.g.,
"physical reality" or "Nature
Herself") playa role in the differential survival of the
beliefs. A plausible scenario as to how this could occur is
a
necessary aspect of
~ny
epistemologically relevant
evolutionary
personally
theory
doubt
of
that
sc~ence.
an
adequate
Given
this
theory
progress can
be achieved
without
conversions to alien theories which are
some cross-lineage
motivated, in part,
by the borrower's own research results.

The social mechanisms which accomplish
doubt complex.
agenda,
of scientific
this
are no
Here are some possible components: There are
scientists who play the role of fascinated bystanders to the
major theoretical quarrels
of their day,
and who encourage
in their students critical comparisons. Such scholars and

their students may constitute a well-informed neutral jury
attending to the "observational" evidence (laden, to be
sure, with old theories, but not all of which are in
SELECTION THEORY AND THE SOCIOLOGY OF SCIENTIAC VALIDITY
145
immediate contention). Oedipal psychodynamics may playa

part: The public image of science may be such as to recruit
persons who strongly desire autonomy from coercive pressures
as to what they are to believe to be true. It is my
observation that in large research and training programs,
there will be some students who are motivated to avoid local
discipleship.
If their
local leadership
is not too
authoritarian, they may solve their oedipal problem by
loyalty to a theoretical point of view and research program
whose center is at some other university.
Provine (1971) provides examples from the victory of

Mendelian genetics over Pearsonian (see also MacKenzie &
Barnes, 1979; Roll-Hansen, 1983). One is the case of A.D.
Darbishire,
assistant and student of
the influential
Professor Weldon, Pearson's major ally. After humiliating
cross-examination
by
Bateson
about
Darbishire's own
experimental data,
Darbishire converted
to the still
relatively
powerless
Mendelian
position, against all
political
self-interest,
as seen at
the time. This
conversion
seems
to
Morgan first made
an
me
an
exception
to
the
theory of
scientific belief choice

and
belief
adherence which
MacKenzie and Barnes (1979) employ to explain the behavior
of Pearson, Weldon, Bateson, and other participants in this
paradigm change. But to say this is not at all to claim it
as thus exempt from sociological explanation or personal
interests. Given the ideology of science then locally
current
(the
social
customs
of
belief
exchange,
"experiments" as "divination rituals" [Campbell, 1979],
social and cultural indoctrinations producing pain and fear
of sanctions from flagrant dishonesty in suppressing or
misrecording research
observations, etc.). Darbishire's
subjective estimate as to the future consensus among his
scientific colleagues provides a sociological explanation of
his conversion, a sociological explanation in which his own
experimental observations play some role.
The case of another convert in this paradigm change
was different in terms of power relations within the
scientific community. As Provine (1971) reports, T. H.
Mendelian, without
needed to defer to
changed to
become
impressive
early career
as an anti-
any dominant patrons whose views he

for the sake of his career. Then he
a vigorous
pro-Mendelian, again without
that move jeopardizing his prestige or academic tenure.

Historically, this turned out to be a very opportune move in
terms of salary, influence and fame. But it was not
146
D. T. CAMPBELL
opportunistic in terms
contemporary scientific
of established
power
in the
establishment.
He participated,
however,
in a
social
system
employi~g
Itreplications"
as
divination rituals
1n
experimental
its mutual
persuasion
intellectual
persuasion processes. Given this, the F2 variability greater

than Fl, etc. which he found in his early drosophila
research may well have convinced him that
here were
that he could
count
on others
dependable phenomena
replicating.
Such replication
would be
an asset in
for
self-serving
leadership, an asset which he

anti - Mendelian.
career
would
lack
of
if
he remained
3. SELECTION THEORY AND DIFFUSION

In my own intellectual development, it was W. Ross Ashby's
great Design for a Brain (1952) that forced my attention to
this general model for both the fit of one system to
another (the one system's "knowledge" of the other) and for
the increase in internal order. Here was the explicit
analog between natural selection in evolution, non-mnemonic
trial and error exploration in Jennings' Stentor, and
crystal formation through a "trial and error" of molecular
adjacencies, in which stable adjacencies tend to cumulate,
while unstable ones changed. Many others have noted the
analog between trial and error learning, natural selection,
and creative thought (see Campbell, 1974a, pp. 457-458 for
a partial list).
Later, I became aware of Baldwin's
(1909/1980)
extensive similar analysis. Baldwin's term
"selection theory" for this general model is very a propos:
Increases in "fit", adaptation, validity, or order, are not
to be explained by prescient variations, but rather by
selection.
This form of evolutionary
theory
of scientific
progress has also been the subject of many attacks, mainly
triggered by the concept of "random", or "blind" variation.
The critics wish to affirm that in scientific creativity the
variations are intelligent and intentional, not blind (e.g.
Richards, 1977; Thagard, 1980; Siegel, 1980). Let me concede
at once that the initiation of deliberate search needs more
explication, as indeed I noted in my comment on Fouille's
1906
book: "Note that his emphasis on conscience as
providing a
goal directed 'artificial
selection', in
contrast to aimless natural selection, deserves special
treatment." (Campbell, 1974a, p. 457; see also Rosenberg,
147
1981). But while introducing consideration of intentional

problem-solving will explain constraints in the randomness
of search
(introduced by
assumptions
tentatively
about
trusted
the
combination
"already
established"
beliefs)
it
will
of
purpose and
"knowledge"
not
obviate
or
the
wasteful fumbling among alternatives that characterizes all

discovery processes. If one is going beyond already trusted
"knowledge,"
Intelligent
one has
no
choice
but
to
choice represents already
or trusted belief.
More
common
incompatibility
"random"
or "blind '1
creative
the
lS
between
scientists.
assertion
the
variation,
If
one
explore blindly.
achieved "knowledge"
of
critic's
descriptive
connotations
to
and the actual behavior of
pays
full
attention
to my
"nested hierarchy of vicarious variation and selective

retention
processes, II
the
descriptive
disagreement
disappears entirely, since I, joining Poincare and others,
attribute much of this fumbling to unconscious thought
processes, and challenge the critics to come up with an
alternative mechanistic model of thinking and intuition. I
have answered such crltlcs in

great detail elsewhere
(Campbell, 1960, 1974a, 1974b, 1977. See also Gamble, 1983).
It
is
enough
here to present the condensed
nested hierarchy in its 1960 version:
model
1) A blind-variation-and-selective-retention
of the
process
is fundamental to all inductive achievements,

to all
genuine increases in knowledge, to all increases in fit
of system to environment.
2)
The
many processes
which
shortcut
a more
full
blind-variation-and-selective-retention process
are in themselves inductive achievements, containing
wisdom about the environment achieved originally by
blind variation and selective retention.
3) In addition,
such shortcut processes contain in
their
own
operation
blind-variation-and
-selective-retention
process
at
some
substituting for overt locomotor exploration
life-and-death
winnowing
(Campbell, 1960, p. 380.)
of
organic
level,
or the
evolution.
148
D. T. CAMPBELL
4. AVOIDING EXCESS ADAPTATIONISM

Whatever the evolutionary model, our theory of science
should take heed from current developments in biological
evolutionary
theory,
and
avoid
Panglossian
excess
"adaptationism"
(Gould, 1982; Lewontin, 1982), or the
epistemological analog of "bean-bag genetics," in which
atomistic components have only uniform main effects (not
among each other) vis-a-vis the selecting
As an aspect of the same caution, one should
interactions
environment.
expect the fit of theory to the world to be less than

perfect at all times, with some of the imperfection due to
the internal coherence-providing structure of the theory.
There are several routes to this consideration. In

adaptive
processes,
the
internal organizational
all
requirements of the
adapting organization
provide an
internal selective system,
winnowing out disorganizing
variations. The maintenance of this system has priority over

the external selection which leads to improved fit to the
external environment (Campbell, 1987a).
Here is an analogy to provide some imagery: "bean-bag

genetics" implies a fit to the environment akin to that
achieved in sand casting, where the numerous independent
grains of sand collectively provide high fidelity "fit" to

the model being cast. But these grains of sand do not have
the organizational structure that a organism or theory has
to have. A more appropriate model would be the "cast" of a
stone made by the roots of a tree that surrounds it. For
this,
the internal
structure
of
the
roots,
their round
cross-section and branching, set limits to the

the fit of the "mold" to the stone.
goodness of
of
fit between
Within evolutionary
developments
stressing
organism and environment.
homozygosity
environments,
prevalence
loci.
in
in
species
theory
today there are
the
looseness
a number
While many biologists had assumed
stably
adapted
to
their
heterozygosity
at all
Ayala's research, among others, has shown the
wild
populations of
Kimura has emphasized the abundance of
and alleles.
of
But closest to
the spirit
of
neutral genes
my emphasis on
internal, structural selectors is the doctrine of punctuated
equilibria
(Gould,
this doctrine
stable
over
environmental
1982).
emphasizes
long
time
changes
Using
that
spans
occur,
palentological evidence,
organismic
during
and
form may remain
which
contrast
substantial
this
with
149
gradualist-adaptationist
model in
which each organism
closely
tracks the
moving
target
of environmentalopportuni ty.
Piaget's (1971) distinction between assimilation and
accomodation has the appropriate character. Once a child has
achieved
a
structured
schema
for
negotiating with
environmental events, it tends to hang on to it, treating
new types of experience as though they were appropriate to
the old schema, resulting in schema stability at the expense
of poor fit to environmental opportunity. Frequent and
serious errors in expectations eventually motivate a schema
change, to a new one in which more of the environmental
regularities are "assimilated" into the schema. Kuhn's
(1962) concepts of paradigm, normal science, anomaly, and
paradigm change describe a similar sequence of cognitive
revisions.
"Normal
science fl
illustrates the value
of the
cognitive coherence and stability which the paradigm and

dominant theory offer, along with the looseness of fit and
anomalies which are the
price of
the coherence and
stability.
Also
relevant is
the
emphasis
on
levels of
organization,
and the role
of
variation and
selection of
alternative organized
subunits,
as in Simon's (1969)
"architecture of complexity" and "semi-decomposability."
This is equivalent to a stress on hierarchical organization
with a node of selection at each level of organization
(Campbell, 1974c). This model too predicts a fit to the
environment with imperfections due to the organizational
structure.
5. EVOLUTIONATRY PROVIDENTIALISM IN THE

EPISTEMOLOGY OF VISION AND IN THE
SOCIOLOGY OF SCIENCE
It is worth noting that the 17th century intellectual
preoccupation with the mediational mechanics of vision led
to a revival of radically skeptical thought. Taking the
mechanisms of perception to be objects-in-the-world on a par
with the normal
objects of perception
(the sort of
thorough-going realism that this paper advocates for both
vision and
the
sociology of
science) increased the
plausibility of the argument from illusion. The camera
obscura (that closet with a pinhole in one wall, in which
the observer
outside
view
inside
can see an
projected on
upside-down
the far
wall)
image
was
of the
toy for
D. T. CAMPBELL
150
leading intellectuals
from da
Vinci
through Kepler,
Descartes, and Pascal at least (Crombie, 1967). And it
provided compelling awareness that any real-world event, no
matter what its true third dimensionality, that produced a
similar two-dimensional retinal display would be perceived
the same, a preoccupation that persisted in the trompe
l'oeil artistic tradition up to the introduction of the
photograph, and is revived in the Ames distorted room' and
rotating trapezoidal window (Ittleson, 1952). But still more
upsetting to the implicit assumption of clairvoyant vision
was the realization that the fluid and mode of flow of the
animal spirits in the neural tubes was basically the same
for the tactile perception of heat, cold, pain, and touch,
and by extension, for vision and hearing: that is, that the
mechanical intermediaries
representations, that all
did not contain faithful iconic

perceived qualities were in this
sense secondary, constructed-in-the-brain/mind. Descartes

participated in this, both in his model of neural tubes and
by experimenting with the lenses from oxen eyes (Crombie,
1967). He fully articulated the skeptical doubts thus made
plausible, and stepped back from living with this radical
skepticism only by positing that a good God would not have
given us eyes that mislead us. Harre (1980) has reviewed
this combination of skepticism and providential ism, and
finds it repeated not only in Locke and Berkeley, but also
in Hume:
"I have pointed out earlier in this chapter
how importantly a providentialist justification
of the actual condition of man loomed
in the
background to philosophical skepticism. We may
not be able to know very much,
but what
we
do
know is enough for our needs.

The development
of a kind of secular providential ism
seems to
me to have
been an essential
element
in
the
preparation
of
the
conditions
for
the
appearance of evolutionary theories based upon
adaptation.
In a way,
adaptation
through
natural selection is a providentialist notion,
and the idea that men, like other beings, are
adapted
to
their
life
conditions
is a
providentialist thesis." (Harre, 1980, p. 33.)
"Hume's providentialism emerges in his
psychological theory of belief. Clearly, we
would be best constituted for practical life
were we to come to believe in the continuance of
151
just those phenomena which have been regularly

conjoined in the past. Reason, so Hume argues,
cannot make knowledge out of this
but our
psychological constitution can make belief. If
'a belief differs from a fiction only in the
manner of
its
being
conceived'
(Treatise, I,
145), that is in its 'force and vivacity', then

we must look for the source of that 'force and
vivacity'." (Harre, 1980, p. 44.)
To repeat, the stance of the modern biological evolutionary
epistemologist can be epitomized: "Natural Selection would
not have left us with eyes that regularly mislead us."
Thus,
reference to natural selection can be used to
"justify"
visually
supported beliefs in
the formula
"Knowledge is justified true belief," in the weakened
interpretation
of
"justification"
used
by
all
modern
epistemologists except skeptics.

This general program of EE (or of providentialism) can
only with great difficulty, if at all, be extended to the
social processes producing scientific belief. I will give
two brief epitomies of the problem. In the EE program, any
"validity" or usefully competent reference, is attributed to
the selection processes which weed out, rather than to the

competence
of the generation
processes producing the
variations.
We know of
so many selection processes
in the
generation,
publication,
teaching,
and
believing of
scientific truth claims that are irrelevant or inimical to
improving the competent reference of beliefs that it becomes
hard to argue for a dominant role for "Nature Herself" in
the selecting. This is in contrast to the case we can make
for Her role in the biological evolution of the eye and
brain.
A reflexive use of biolog~cal evolutionary theory
provides a complementary perspect1ve. Both Cartesian and
evolutionary providentialists could plausibly say "(God)
(Natural Selection) would not have given us untrustworthy
eyes." But even if they noted that the social system of
SC1ence requires great (albeit selective) trust of fellow
scientists,
neither
the old providentialist
nor the
evolutionary epistemologist
would be apt to
find it
plausible
to argue
that
"(God)
(biological Natural
Selection) would not have given us untrustworthy fellow
scientists." If we can in fact often validly trust fellow
scientists, this is because of culturally evolved and
fragile social systems, not because of innate honesty and
152
D. T. CAMPBELL
objectivity.
Hull (1978)
has raised the issue in a
sociobiological framework, in a pioneering essay that should
have appeared in Social Studies of Science rather than in
The Journal of Animal Behaviour.
The problem is due to the fact that for us vertebrates
there
is genetic
competition
among
the cooperators
(Campbell, 1972, 1975, 1983). For the social insects whose
cooperators are almost completely sterile, one can say:
"(God) (Natural Selection) would not have given an ant
worker untrustworthy scouts." But for us social humans, the
belief assertions or public truth claims we make have
important utilities for us other than optimally guiding our"
own (and our identical twin's or clone's) behavior. We have
selfish (including nepotistic) interests in what others
believe and in what others believe we believe, often
motivating belief assertions that differ from those that
would optimally guide our OWn behavior vis-a-vis the objects
that are nominally the referents of the truth claim. In
addition to, or instead of "valid description," we have an
often conflicting interest in influencing the decisions that
our
listeners
will make.
(This
becomes
a problem
particularly when "secondary groups" rather than "primary
groups" are involved. The problem may not have been acute
for a stage in human social evolution in which inbred tribes
of 100 or so were in intense close competition with other
similar tribes, with individuals unable to change tribes
successfully.
But
it
is
acute
in
all
secondary
organizations,
including
science.)
We
must
J01n
Merton
(1973), Merton & Gieryn (1982), and Zuckerman (1977) in

attempting to understand how the social system of science
inhibits self-serving dishonesty, and as a result, allows
"external reality" to have somewhat more influence in the
social winnowing of beliefs for consensus formation. While
many
philosophers
of
science
now
recognize
the
epistemological relevance of such a sociology (Campbell,

1986b lists some 30), an adequate theory is not yet
available. (See Campbell, 1986a for a preliminary sociology
of scientific validity.)
SELECTION THEORY AND THE SOCIOWGY OF SCIENTIFIC VALIDITY
153
6. CULTURAL EVOLUTION AND METAPHYSICAL BELIEF

"But I was thinking of a plan
To dye one's whiskers green,
And always use so large a fan

That they could not be seen."
Not only was number 6 the most opaque of my initial list of
hints,
this fuller description of it may add little
clarification. While I present it with complete seriousness,
I also sympathize with my unnamed critic who finds it akin
to the White Knight's fan, an inelegant solution to an
arbitrarily self-assigned, otherwise non-existent, problem.
Let me first try to gentle you into the problem: Just
as we need an EE of vision, rationality, and science, so too
we will eventually want an evolutionary

sociology of
knowledge which will encompass not only science but
also
religious, magical and ideological beliefs,
ancient and
modern. A Selection Theory version of cultural evolution
(Campbell, 1965; Boyd & Richerson, 1985) makes it plausible
that a blind variation, selection, blind retention and
reproduction of superstitions could produce "superstitions"
with
an
adaptive
"wisdom"
that
went
beyond
the
understanding of the cultural transmitters, that is, beliefs
whose functions were latent, not manifest. For such an
adaptive cultural evolution to work, respect for tradition
(retention and reproduction of beliefs) would have to be
very great.
Certainly within the area of tools and crafts there
has been a variation-and-selective-retention process leading
to increased competence in manipulating the materials upon
which tools are used and of which they are made. This has
probably been accompanied by the evolution of increased
competence in linguistic reference to these raw materials,
tools, and procedural recipes. I have earlier (1974a, pp.
234-237) identified science as exemplifying and continuing
such cultural evolution,
differing only in the focal
precision of the selection process, and therefore the speed
of the adaptation.
While the focal precision of the
winnowing of beliefs, recipes, and tools is certainly a
major aspect of science, I now believe this characterization
is much
too simplistic.
It seems
contradicted
by,
or at
least fails to account for, the profoundly anti-traditional,

anti-authoritarian,
anti-metaphysical
ideology (if not
D. T.CAMPBELL
154
practice)
of
science
as
contagious
social
movement
appearing in Europe in the 16th and 17th centuries. The

focus on the cultural evolution of technology also omits
attention to the separate social evolution of moral systems
which produce
group-level
effectiveness by inhibiting
individual opportunism. Such moral systems are probably
needed as a part of the answer to Hull's (1978) problem,
discussed
in
the
previous
section.
Let
me
sketch some
fragmentary hunches as to what may be involved:
Consider the cultural evolution of

aspects of the physical world unobservable
beliefs about
by vision and
cosmological
as gravitation
touch,
and
such as past objects and events, the microscopic and
realms,
magnetism,
invisible forces such
and perhaps remote "objects"
such as stars
when unobservable. For Laudan (1981) the most impressive

achievement of the scientific revolution is disciplined
discourse about such hidden physical processes. It is my
conjecture that in the historical period in which science
emerged, cultural evolution had long ceased to produce
(if
it ever had produced) improved referential competence in
the beliefs about the unobservable physical world.
In the explanation that follows, I am returning to the
biological evolutionary problem that lead me (above) to
reject evolutionary providentialism for scientific beliefs.
More importantly, I am calling the attention of our EE
community to the direct relevance of formal theories of
cultural evolution
to our enterprise.
Boyd and Richerson's
Culture and the Evolutionary Process (1985) is my mode of

entry into this literature.
In biological evolution it has often occurred that
organs
evolved under one
intermediate
functions.
route
to
selective
selection
("Preselection"
is
advantage
for
term
quite
often
provide the
different
used
in
describing such species histories, an awkward term because

of its unwarranted connotations of teleology or fore-sighted
external guidance.) Boyd and Richerson (in their Chapter 7)
describe such a shift in function, making possible selection
for group-organizational efficacy in cultural evolution,
even though group selection is precluded in vertebrate
biological evolution by the
genetic competition among
cooperators.
What has been cited above (see Selection Theory) as
cross-lineage borrowing, Boyd and Richerson designate as
"multiple parenting," i.e. the tendency to adopt beliefs
from others in addition to one's biological parents. They
155
propose a non-linear variant to this, a "frequency-dependent

bias," a bandwagon effect, in which individuals of the
learning generation tend to adopt the belief of the largest
plurality or majority among available mentors. Such a
transmission rule would probably speed up the cultural
cumulation of competent beliefs. But in addition, it has a
dramatic side effect. (Their argument is a cultural analogue
to genetic drift in biological evolution.) For small,
isolated communities of belief-sharers, chance can lead to
local pluralities on any given belief. The bandwagon,
plurality-enhancing transmission will in a few generations
push such pluralities to local unanimity. At the same time,
chance initial differences will have pushed other isolated
communities
into
different
internal
unanimities.
This
combination
of intragroup
homogeneity
and intergroup
differences would set the stage for a "group selection" of
beliefs, were some beliefs to be accompanied by greater
group efficacy than others. Such differential propogation
would typically be by cultural diffusion (e.g., selective
borrowing
of
ideologies
from
more
successful
groups,
political imposition of beliefs). Differential fertility

would not necessarily be involved.
(Even before we consider differential group selection,
however, note that the combination of intragroup homogeneity
and intergroup differences has one immediate function in

helping
solve
the problem of cooperation
and group
solidarity. These intergroup differences (along with the
concomitant costume and scarification differences) function
as badges
or
uniforms,
indicating
co-membership in
culturally inherited reciprocal altruism pacts [Brewer,
1981; Campbell, 1979', 1983].)
It seems very likely that important sets of beliefs
that a cultural group-selection of beliefs might select
would be those that tended to curb selfish individual
opportunism, cheating on the social contract, cowardice in
battle, laziness, disloyalty to the social unit, etc. A
casual survey of ancient complex division-af-labor societies
shows
widespread
beliefs
in
rewarding
reincarnations
and/or
occurrence
ancient civilizations
Buddhism
burdened
reincarnations and
in
its
10
afterlives.
and
(Ancient
punishing
Chinese
folk
believers
with both punishing
to 16 specialized hells.) The
of
lavishly wasteful
royal burial customs calls for a functional explanation

because of its near universality. This function might lie in
the heroic demonstration of official belief in an after
156
D. T. CAMPBELL
to
calculate
their
net
hedonic
payoff
with
a
time-perspective extending beyond their current bodily life,
and hence act so as to further group purposes rather than
individual purposes were these to be calculated for a
biologically limited lifetime (Campbell, 1975, 1983).
Today, after four centuries of negotiation between
science and religion, we tend to relegate such beliefs to
the metaphysical or supernatural. I want instead to join a
long-standing anthropological observation
(see Shweder,
1986, pp. 171-176, for a review) and classify these beliefs
(from the point of view of the believer) as being about
invisible physical reality, i.e. as being about forces that
are causally linked to physical observables. They are thus
in the class with
atoms, etc.
gravity,
magnetism, electrical charges,
To summarize: it IS hypothesized that in cultural

evolution,
the
domain
of
beliefs
about
physical
unobservables had been thoroughly
co-opted for social
solidarity functions. The cultural winnowing of such beliefs
would then no longer be producing increased "competence of
reference" (a euphemesism for IIliteral truth", a term which
I find unusable). Thus for any subgroup social system (such
as science) to generate increased competence of reference to
invisible physical forces, it had first to disavow the

larger cultural tradition including political and religious
authority. The ideology of the scientific revolution did
just this, replacing these with the authority of individual
observation of visibles (even where theories of invisible
forces were in contention) and individual intuitions of
rationality (Campbell, 1986a).
So much for the White Knight's fan. May I finish by
trying to make his green beard seem less arbitrary. Quine's
role as an evolutionary epistemologist began for most of us
with a couple of essays in Ontological Relativity (1969',
1969'b). I want to suggest that we should also claim for
ourselves his founding post-positivist essay, "The Two
Dogmas of Empiricism" (1951). In the final dozen paragraphs,
he
says that in
improving our beliefs,
such that we have no choice but
our predicament is
to trust the
great bulk of
our current beliefs, revising as few as possible in light of

recalcitrant observations, but without exempting any of
them from revision. The Quine-Duhem problem can be solved in

no other way. The skeptics are not answered by such
coherence strategies of belief revision, but such strategies
are,
faute-de-mieux,
the best we can do even if we hang on
157
to a correspondence definition of the meaning of
the truth.
I have
Ratio"
(1978)
epitomized
(granting
this
as
the
"99/1 Trust/Doubt
that during a "scientific
revolution" in
physics or chemistry, the ratio of trusted old beliefs to

challenged ones might drop to 90/10). Within science, the
respect for tradition must hold, and increasingly so the
more competent the scientific cumulation.
This, I argue, is
a central part of Selection Theory (or, indeed, any theory

of cumulative increase in adaptive fit or referential
competence that does not make use of external revelations or
clairvoyant knowers).
As such, the 99/1 Trust/Doubt Ratio should hold too in
biological evolution. Indeed, it does. From a physicists'
point of view, the extreme loyalty of gene duplication in
both meiotic and mitotic cell
division
the
outside
is
more remarkable
than the occasional mistake or mutation. None of the alleles

at any of the gene loci are of "proven validity." (Indeed,
concept
ensemble
of
blueprints.)
is
meaningless
proteins
All
must
and
be
the
the
genes
"trusted ll
in
context
that
the
are
of the
their
duplication
process, including neutral and mildly deterious genes that

have not yet been weeded out. The intuitions of evolutionary
biologists
as
to
the
optimal
retention/ variation
(trust/doubt)
ratio
is made clear
by their uniform
disapproval of anything that would increase the human
mutation rate, be it x-ray examinations of the pelvis or
nuclear weapons testing.
That
mutatlon
rate
is very low,
relative to the overwhelming loyalty of duplication.

CONCLUSIONS
EE should adopt the Baldwin/Ashby program of a general

Selection Theory to explain all cases of increased order,
fit to environment, and the competence of reference of
beliefs. This differs substantially from using biological
evolution as the most general model for the evolution of
science. In particular, Selection Theory does not preclude
cross-lineage borrowing,
consensus change.
an essential
aspect of scientific
The
evolutionary
providentialism
that
provides
fallibilist justification to visually supported beliefs in
ordinary physical objects cannot be extended to science, for
reasons of the evolutionary biology of species in which
there is genetic competition among the cooperators. Instead,
D. T. CAMPBELL
158
Selection Theory extended to science will have to describe a

social system of belief exchange in which it is plausible
that the nature of physical reality plays some role in
changes in the scientific community's beliefs about that
reality. Customs of experimentation emphasizing effects upon
observables might play such a role.
Given the nature of the culturally evolved beliefs
about unobserved physical processes, the antitraditional,
antiauthoritarian ideology of the scientific revolution may
be understood
as functional.
However, within the sciences,
particularly the most successful, a 99/1 Trust/Doubt Ratio

for the accumulated corpus of beliefs is required.
An evolutionary model of scientific belief supports
the expectation of only a crude fit at best between belief
and
referent,
rather than Panglossian or clairvoyant
perfection.
We must get to work on an adequate Selection Theory
for the social system of belief-sharing and consensus-change
in science.
VARIATION AND SELECTION:

SCIENTIFIC PROGRESS WITHOUT RATIONALITY
Louis Boon
Rijksuniversiteit Limburg
1. PROBLEMS OF RATIONALITY
Much of the fiercest debate in philosophy of science over
the last decades has concerned the rationality of science.
Indeed it has sometimes seemed that we have two warring
cultures, rationalists and antirationalists, in philosophy
of science.
Why is the rationality of science considered so
important?
The crux is that rationality is held to be the
vehicle of progress in science.
It is through the rational
application of scientific method that
we improve our
knowledge of the world.
If the rationality of science is
attacked,
the very idea of cognitive
progress seems
endangered and the specter of relativism ralses its ugly
head. This seems to be an assumption shared by both parties
to the debate.
The strong program and the ethnomethodology
of
science,
both
highly
critical
of
scientific
rationality, proudly proclaim themselves relativist.
The
philosophical landscape consequently
has become neatly
divided between the forces of darkness and the
forces
of
light: rationalists for progress and anti-rationalists for
relativism.
Is this conjunction of progress and rationality (and
its converse) necessary?
Could we develop a theory of
science in which we would not predicate a privileged
rationality for science and its practitioners and still
remain faithful to the idea of cognitive progress? The
latter can - despite the heroic rhetoric of Paul Feyerabend
hardly be denied when we take a long term or global
perspective. Despite an occasional "Kuhn ian loss", knowledge
has grown and accumulated in terms of the number of facts,
the explanatory power of theories, the precision of our
predictions, or the accuracy and scope of description. Such
a rough concept of progress seems condoned even by authors
who have been branded irrationalist and relativist (Kuhn,
1969, p.170, p.206; Bloor, 1976, p.34).
159
W. Callebaut and R. Pinxten (eds.), Evolutionary Epistemology. 159-/77.
/987 by D. Reidel Publishing Company.
160
L.BOON
Though the concept of cogn1t1ve progress is beset with

intricate philosophical problems - especially regarding the
interpretation of progress: instrumentally or in realist
terms, - the real troublemaker is rationality. More or less
explicitly, starting as far back as Bacon and Descartes,
rationality and scientific method have been considered the
driving force of progress.
This basic assumption has been
confronted with an increasing number of problems.
One of
these has been the fact that a number of contradictory
theories of scientific rationality has been developed.
Lakatos
lists
inductivism,
conventionalism,
critical
rationalism, the methodology of scientific research programs

(Lakatos, 1978). One may now, in view of the recent vogue,
What is rational in one theory may well
add Bayesianism.
be irrational in another. What one views as contributing to
progress the other sees as stagnation.
All these theories
have
their historical 'exemplars'.
Worse, scientists have
used
these
various
methodologies
to
justify their
decisions. Skillfully capitalizing on this fact, Feyerabend
has tiredlessly pointed out: In science, anything goes
(Feyerabend, 1975).
Some philosophers have argued that we should not take
what scientists say
too seriously but
should rather
concentrate on how they act (most notably Lakatos). However,
this would seem to get rationality even deeper into trouble.
The theory that should be preferred according to methodology
A, and which is in fact preferred by a given scientist,
may, as far as the scientist is concerned, be preferred
because he adheres to methodology B.
The better theory is
chosen for the wrong reasons.
What this means can be shown
In their paper on Ptolemy's and
by looking at an example.
Copernicus's theories, Lakatos and Zahar argue that there
were 'objective' grounds for Kepler and Galilei to prefer
Copernicus,
because
even
his
rough
model
of the
Commentariolus had excess predictive power over its
Ptolemeic rival, only to retract this statement immediately
in a footnote: "Note that this statement does not say
whether
and
why Kepler and Galileo
actually became
'Copernicans'" (Lakatos and Zahar, 1978, p. 188). Of course,
this footnote is necessary, because the authors know that
other reasons determined preference in this case.
By restricting attention to the actual 'choice' of
scientists,
progress
and
rationality are effectively
separated.
Implicitly, one must now fall back upon a
'cunning of
reason'
that guides blind or irrational
161
SCIENTIFlC PROGRESS WITHOUT RATIONALITY
scientists to the 'right' theory.
The predicament is clear:
Given modern liberal theories of rationality one must choose
between an ocean of anomalies (scientists adhering to

incorrect theories of rationality) or a cunning of reason.
This situation is, I believe, general: Whatever theory of
rationality one adopts, it will in general be impossible to
show it actually determined a choice of scientists.
If, on
the other hand one takes historical facts seriously, these
will in general refute the theory
of rationality in
question.
Such problems have their roots in the fact that
rationality refers to actions of individuals. As Laudan
sees it, "rationality consists in doing or believing things
because we have good reasons for doing so." (Laudan, 1981b,
p.187). Rationality then becomes both too broad and too
narrow.
Too broad because having good reasons for actions
is not
restricted
because this kind

social nature of
to
scientific
activities.
Too narrow
of rationality will be thwarted by the

science.
Stated differently: Even if
scientists were always rational, the social nature of

science would contextualize reason, for what is rational in
one situation may not be so in another.

Let us look more closely at a variant of this: What
may be rational if a few scientists do it, may be irrational
if all did or what may be rational for limited periods leads
to irrationality when practiced all the time.
Criticism is
an obvious example of this.
If everyone were always
critical, then science would cease to exist.
It is in a
sense too easy to criticize a new theory. Scientists would
cease to work on a theory if they heeded all refutations.
Criticism is necessary for the overall growth of science,
but it must be supplemented by tenacity.
As Kuhn has put
it: Science needs the essential tension of innovation and
conservatism. Stubbornness or dogmatism may seem irrational
on the individual level but may contribute to progress on
the global level.
Precision is another example.
Researchers who try to
increase the precision of their experimental results to test

theories are important for science as a whole. On the other
hand, the growth of science requires a certain amount of
experimental
sloppiness.
Max Delbrlick, the molecular
biologist,
even
formulated a
'principle
of limited
sloppiness' and pointed to cases where scientists were lucky
enough to have been imprecise.

(Cairns et al., 1966, p.
158). So you need both precision freaks in your community
L.BOON
162
and sloppy experimenters. Spreading risks in this way has a

further advantage when it comes to ignoring anomalies, or to
hailing certain results as victories.
One can now say:
Don't trust his work, he is sloppy (or overly precise). It
gives
individuals the leeway for
escaping unpleasant
conclusions with good reasons and for getting on with
research.
If a special rationality is neither the hallmark of
science, nor necessary for its growth, and if furthermore
it leads to the necessity of a 'cunning of Reason', then why
has
its
importance traditionally
been emphasized so
strongly?
I believe that a distinction between local and
global development in sciences is eminently relevant here.
2. LOCAL AND GLOBAL DEVELOPMENTS IN SCIENCE
The global, or - from a time perspective
long term
development of science displays a rather clean structure in
terms of progress.
For instance, the development of
genetics between the rediscovery of Mendel in 1900 and the
discovery of the genetic code in the early sixties gives us
a structured tableau of successful research programs that
elaborate upon the results of predecessors; despite an
occasional
slip,
successive
theories
show correspondence
relations.
On this level we witness the progress, or the
progressive order, that is so characteristic of science:
deeper, more precise explanations of an increasing number of
facts.
The local, short term, development of science refers
to the multitude of traditions, contexts and situational
logics in which individuals or groups grope for strands of
evidence, and invent or improve theoretical interpretations.
Here contexts differ, people negotiate, persuade or are

persuaded.
Local science lacks a clear structure, seems
chaotic, is reigned over by several forms of rationality. It
is this level on which Feyerabend finds arguments to bolster
his slogan that anything goes.
We human beings are typically short-sighted: We tend
to consider
our
immediate
future
or
environment
and to
neglect long term effects.

(This is a very general
phenomenon that may even threaten the future of our race and
planet.) Hence we tend to believe that the overall progress
of science follows immediately from our local actions.
Traditional philosophy of science of the rationalist variety
condones this short-sightedness: By applying scientific
163
method the scientist bring~ about progress.

The global
order is believed to be lmminent within us as rational
researchers.
The order of the system is the order of the
parts. The very same short-sightedness resulted in a denial
of progress when historical research had shown that there
existed a multitude of local orders.
If we
reject
both
relativism
and traditional
rationality, the problem becomes: How does the global order
of science arise out of local chaos?
Stated in this way,
the Theory of Evolution seems a suitable candidate to
present us with a model of the process.
From the disorder
of recombination and occasional mutation, natural selection
creates long term biological order.
As promising as a
natural-selection model of science may seem from this
vantage point, a few caveats must be noted.
Using evolutionary theory to illuminate the theory of
science raises the problem of applying a theory that has
been successful in dealing with the development of life to a
completely different realm.
This implies that one is
applying a number of biological mechanisms to science by
analogy or metaphor and the question arises as to how far
one may go in this respect.
Extending Darwinian theory metaphorically to science
leads, beyond a certain point, to 'biologism'. However, the
contrary evil is unwarranted timidity.
If the analogy is
developed only superficially, the potential of the theory of
evolution is not fully realized; i.e. its heuristic is not
fully utilized.
Now I believe that in the present problem context the
latter has been the case. Evolutionary theorists of science
have been overly cautious and the victims of unnecessary
difficulties as a result of their stress on the rationality
of science.
Popper's evolutionary theory of knowledge
through trial and error, conjectures and refutations, is a
good example (Popper, 1963). Though he is interested first
and foremost in the logic of science, and though he stresses
the social nature of science, he still wants to recover
rationality in science in the form of a readiness to
criticize one's theory. Though Popper believes variation in
science to be blind - in the innocent sense that it is not

certain that a variation will be accepted - he insists that
selection is not blind, but purposeful and conscious thanks
to the rational attitude of scientists. Popper's viewpoint
L. BOON
164
a
good opportunity
for
formulating a few
requlrements of a non-rational model of the evolution of
scientific knowledge.
Progress must not
be coupled
with rationality.
Rather, the global or overall structure is the unintended
result of individual actions.
Of course this must not be
taken to mean that individuals have no goals or purposes or
that they do not apply standards.
The idea of unintended
effects refers to the fact that 'good' intentions are not
sufficient to explain the structure of knowledge on the
global
level.
At
this
level
knowledge
acquires
characteristics that were neither planned nor intended in
local communltles.
There is a discrepancy between the
'plans' of individuals (rationality) and overall effects.
Hence an evolutionary model of science must show how the
'individual' is eliminated, how his ideas are expropriated.
It is because of the social nature
of science that
pres~nts
researchers
may
work
without
consensus
over
theories,
methods and criteria and still all be able to contribute to

the overall growth of science.
The latter is possible only if selection
is a
multi-level affair.
The trouble with a rationalist theory
of science is that it assumes a one-step growth function.
We need a hierarchy of selection levels through which order
arises out of the hustle and bustle of interaction in
communItIes. This generates a third requirement: in an
evolutionary model the objectivity of science is the leading
principle. This objectivity means that science includes its

practitioners.
It is a self organizing system of which we
form only temporary parts, to be replaced by others when we
are no longer useful.
Science is structured through a
hierarchy of social or institutional selection filters
behind the backs of individuals and specific communities.
Indeed as in Neurath's ship metaphor, or in Popper's

metaphor of the cathedral of knowledge, we contribute to a
global structure, without the existence of a blueprint.
Scientists are sleepwalkers or to use Dennett's metaphor for
the mind; scientists form an army of idiots, that only
collectively generate cognitive progress.
I will
divided
now present a broad outline of
such
a model,
into two parts, one dealing mostly with variation

the other mostly with selection.
165
3. VARIATIONS AND HEURISTICS

Evolutionary processes in nature may provide us with a
number of insights relevant to the development of knowledge.
An organism may be viewed as an hypothesis manufactured by
nature, which is subsequently tried out or tested in the
struggle for
survival.
Variations in the organism's
blueprint, brought about by recombination or mutation, are
negligible compared to overall similarities with members of
the same
species.
Each
new
variation
arises
against a
background of fixed structures.

In science similar phenomena can be isolated. New
theories or hypotheses are variations that are subsequently
tested
for
viability.
Even
subsequently,
quite
revolutionary variations
only
overturned
few
our
start in
assumptions.
most
basic
a small area
Copernicus's
conceptions
of
or concern
theory
the
which
cosmos,
originated within astronomy and left intact the bulk of

astronomical
knowledge,
including circular motion and
epicycles.
It took decades before Kepler introduced the
ellipse as a further innovation.
A more recent example is
Max Planck's work on black body radiation from which quantum
theory emerged ever so gradually.
Thomas Kuhn's book on
this work aptly illustrates the way in which a series of
small innovations produced a major revolution by degrees, as
it were (Kuhn, 1978).
The background to this ~radualness and carefulness is
the same in nature and SC1ence: any major change in the
blueprint will be non-viable. Because variations are blind,
i.e. it is not clear whether they will be succesful, the
chances that a really fundamental innovation, one that
leaves no stone unturned,
will he viable are nil.
Too many
matters must then be mutually adapted at the same time.

Furthermore one cannot start with a clean bill, neither in
science nor in nature. The burden of the past always weighs
on the present alternatives.
As Campbell has put it, there
is a doubt - trust ratio in science which will always be to
the advantage of trust
(Campbell, 1977). In science as 1n
life
we
criticize.
much more
must
always
trust
more
than
we
can
doubt or
All this implies that conservatism in science is

important
than
progressiveness
and
that
the
acquisition and accumulation of knowledge will take place

within the confines of research traditions whose course
cannot easily be changed.
166
L.BOON
The revolutionary character of science is in a sense

an optical illusion produced by the bird's-eye view of a
field.
Only
from
a
large
distance
are
there
discontinuities. Researchers will be focused in their work
on the local situation, where they will have to solve a
specific problem.
The global coherence and structure of
their discipline is not their first priority.
Still, they
put forth innovative hypotheses or data within this global
structure. As these innovations produce only slight change,
and remains only in the front line, the bulk of traditions
is unaffected in most cases.
The unproblematic background
knowledge of the tradition forms the basis
of their
problems,
and compels them to develop
variations or
interpretations
that
closely
connect
with
existing
viewpoints, even when they also challenge these. Hence
existing knowledge restricts the range of variations. At the
same time,
however,
it also
contains directions or
guidelines to develop new variations.
The Lakatosian
heuristic is an apt formulation of this latter aspect
(Lakatos, 1978).
The (positive)
heuristic of a
research program
consists in guidelines for further articulation. These
embody the 'promise of future' success: the heuristic rules
of the program provide clues as to how to generate new
variations
when the program is confronted with problems:
"If an interpretation does not survive selection, then try a
new one along the following line: ... ".
Such rules, of
course, neither guarantee success, nor are they an 'ars
inveniendi'; but they may still be helpful directives for
the researcher.
In the natural sciences such directives are often
derived from mathematical theories.
If light is conceived
to be a stream of particles, then the mathematical theories
and techniques used in particle mechanics can be put to use.
In such cases the
logic of
the mathematical theory
determines the nature of new variations.
Analogies are
another heuristic tool.
Starting from the idea that the
atom was in
some sense
a miniature solar system,
Bohr was
able to gradually fill in a

more complicated model.
Evolutionary theory fulfills such a role in the present
essay.
The heuristic of a research program is of immediate
relevance for the nature of variations that are generated in
order to be tested through selection.
It is characteristic
of the evolutionary process both in nature and in science
that
successful
variations
selection environment.
new
167
themselves
variation
influence
the
among predators,
which increases their success in hunting, puts pressure on

the species of prey, which may in turn have repercussions
for still other organisms.
But while in nature a new
equilibrium usually results, it is typical of science that
here, 'external attractors' are more the rule than the
exception.
It is because of this dynamic disequilibrium in
science that research traditions are remorselessly burned up
by the way in which variations change
the selection
environment.
It is worthwhile looking at this in more
detail.
Suppose a new tradition is started in an as yet
uninhabited ecological niche of our
knowledge.
Early
research will orient itself to what is known about the area
in other fields or disciplines.
There is a certain freedom
in the choice of selection environments
to which variations
will be subjected.
In the early days of molecular biology
scientists could orient themselves toward the established
selection environment of biochemistry, but also to selection
mechanisms of a more physicalist nature.
The latter were
more drastic and resulted in dramatic developments.
Early hypotheses on the nature of the new area are
imported from other fields, and tend to be of a rather
general nature in any case.
However, each successful
variation will make later selection more stringent. A new
interpretation will have to take earlier successes into
account, and should preferably be (slightly) innovative,
simpler,
more precise etc.
The
nature
of new variations
will be determined increasingly

by what is already known.
As the niche gets filled, the number of degrees of freedom
fot new variations steadily diminishes.
Those variations
that actually lead to new problems and results will entrench
themselves and become the starting point for new research.
The doubt/trust ratio shifts further and further in the
direction of trust, of accepted background knowledge.
Research from early phases of a tradition remains
influential over such long periods because it is then that
the general
direction of variation and
selection is
determined.
Once started down a specific road, it becomes
difficult to introduce radical change, to retrace one's
steps.
The epigenetic landscape forms a vivid picture of
the irreversibility of development.
L.BOON
168
The positive heuristic becomes of central importance

in this context. A powerful heuristic allows a tradition to
generate a long
series
of
idea that
'sports',
ever
deeper
variations and to
change the selection environment over a longer period. A

tradition with a weak heuristic will quite soon loose this
power.
Hence it will no longer generate new problems and
stimulate
new
research.
The saltationist theory of
evolution as it was developed by Bateson is a good example
of the latter. His critique of Darwinism led Bateson to the
only
large
discontinuous variations,
could have evolutionary relevance (Provine, 1971). Bateson

sought for corroborating evidence and started to collect
specimens of sports. These were collected in his monumental
Materials for the study of Variation (1895). His program,
however, lacked any heuristic except for the rule: "Collect
examples, analyse their genealogy and make plausible their
role in evolution". This quickly led to "more of the same".
Around 1900, his program ran solidly aground; it generated
nothing new.
Bateson was saved by the
rediscovery of
Mendel in 1900 which placed a powerful heuristic at his
disposal: Mendel's factors naturally lent themselves to
conceptualization
independent
could
arise.
as
discontinuous
segregation
Addition
and
or
random
variations.
omission
assortment
of
Through
new forms
factors
were
mechanisms of evolution.
With this new heuristic
he was
able to undertake a long series of experiments
aimed
at
articulating and refining his saltationist evolutionary
program.
Failure after failure was
turned into victory.
Of course with hindsight we now know that the saltationist
program failed and that Bateson's work between 1900 and 1910
contributed primarily to genetics and less to evolutionary
theory.
However, this fact makes no difference to the
historical analysis and diagnosis.
Despite the power of
the Mendelian heuristic, Bateson's research stagnated again
after a number of years, and new traditions such as
Drosophila-genetics took over. This course of events is a
quite general phenomenon.
A strong heuristic assures a program of a great
plasticity, whereas a weak heuristic quickly leads to
rigidity.
Whatever its power however, there always comes a
moment when the heuristic loses its momentum, when it has
burnt up its suggestions.
The interplay of variation and
selection hardens the cognitive network.
All heuristics
SCIENTIFIC PROORESS WITHOUT RATIONALITY
169
have only a finite set of rules to suggest new moves in the

game, and using them with success brings a program to its
natural saturation point.
It was one of Lakatos' defects that he refused to take
this into account.
For him a program could always stage a
comeback.
Given enough ingenuity it was always possible to
get out of a degenerative phase.
This puts too much trust
in imagination and neglects the objective characteristics of
a he~ristic.
The logic of variation and selection burns up
the heuristic till there is no longer room to maneuver.
At that point there are two possibilities.
One is
that there are no more problems left to solve. The program
shuts down with maximum success.
From a scientific theory
it now becomes a tool for engineering.
The other and far
more frequent outcome is that there are still problems left
that have resisted all attempts at solution.
In Kuhn's
paradigm theory this is the moment when a crisis starts.
Sometimes
it is
possible
to
indicate
whether the
degenerative phase has a good chance of being temporary. An
example of this is a situation where there are still
heuristic guidelines, but mathematical theories to put these
rules to work and generate testable variations are as yet
lacking. In such a case there is an objective background to
the idea of staging a comeback.
Hence Lakatos' intuition
that in some cases a regeneration of a program is feasible
is not entirely an inspiration from out of the blue.
Another basis for asserting the possibility of a
comeback is the lack of adequate techniques to measure key
variables.
In such a case there is a discrepancy between
the explanatory theories and the interpretative theories
that are used in a tradition.
Related to this are
situations in which it is not yet known how to 'embody' the
interpretative theory in observational machinery. According
to Max Delbrlick, this was the situation in quantum mechanics
in the thirties.
He estimated that it would take decades
before such machinery would be developed, and so he chose
another field of study, the molecular base of biological
processes, to which he brought his physical viewpoints and
techniques.
However, even when such a temporary breakdown
of a tradition has been overcome it is only a stay of
execution.
Sooner or later, stagnation will again set in
when the heuristic assumes its final trajectory. Being
L. BOON
170
saved by the discovery of a completely new heuristic

perspective is, pace Bateson, quite rare. I already pointed
out that even this new addition did not save him in the end.
Drosophila genetics, the combination of Mendel and
chromosome theory, assumed the momentum of genetics after
1910.
Thomas Hunt Morgan's school moved genetics to a new
biological level, the cell, and worked with a different
heuristic.
This program generated the major breakthroughs
in genetics between 1910 and the thirties.
But again,
success glutted the cognitive landscape and in the thirties
research had become completely 'normalized'.
Mapping loci
on the chromosome, research which had great scientific
relevance in the early days and brought dramatic new
developments, now became routine.
Mapping loci had become
something one could assign a student for his dissertation.
The outcome was predictable.
Two decades of research had found their Waterloo. A
new major problem, such as the material nature of the gene,
could not be answered in its terms.
New programs arose
around classical Drosophila genetics: biochemical genetics
and molecular genetics.
The atmosphere of stagnation - or
of routinised science - is sensed by new, young scientists.
When Watson came to Bloomington he wanted to study with
Drosophila geneticist MUller, but as he says: "I soon found
out that Drosophila's better days were over and that all the
younger
geneticists
were
working
with
micro-organisms"
(Watson in Cairns et al., 1966, p.239).

He decided to
switch to Luria, and later to unravel the structure of DNA
together with Crick.
Such facts point in the same direction: all programs
ultimately collapse.
Their life cannot be indefinitely
extended. The nature of variation-selection processes in
science
determines
the
typical
scientific cognitive
products: models that coordinate an increasing number of
facts in an ever more precise way and which explain more and
predict better than predecessors.
Once the first viable
variations have entrenched themselves, any new variant must
improve upon the successes of the earlier ones without
fundamentally or discontinuously changing the framework
embodied in the first variations.
In most cases these
variation are further restricted by the enormous edifice of
relevant background knowledge constituted by other fields.
If there is selection and if the selection pressure is a
function of the level of development of a tradition - if, in
other words,
every success raises the selection pressure or
171
sets higher standards - then the outcome can be nothing else

but the progress of science as we know it.
The question
then becomes how selective pressure
in
work after rationality as its method has

this I will now turn.
science can
do its
been excluded.
To
4. SELECTION: ENVIRONMENT AND LEVEL

Typical of both nature and science is that more variations
are produced than actually survive.
In science, although
variations are already selected against by the individual
researcher, a rather large number is published in articles.
Most of these are never used by other scientists. Still
fewer form the starting point of new research. Very few end
up in textbooks and handbooks. Only a couple of these are
so important that
mention
in
their originator achieves
historical
works.
Variations
Nobel
fame or
are selectively
repressed. Not all of them survive the selection filters.

Those organisms that can
better cope
with the
challenges of their environment will have more opportunity
to contribute to
the
next
generation;
hence their
characteristics or behaviors will spread in the population.
What counts in nature is
differential reproductive success.
It is a prime measure of fitness.

In science this
reproductive success expresses itself as the extent to which
a variation stimulates and influences new research. The
reproduction of someone's work can be measured by the use
other scientists make of it (citations
are a useful
indicator of this).
To achieve such influence it is
necessary, but not sufficient, to pass selection filters. A
scientific mule may survive
individually, but remains
sterile; it does not contribute to the future.
Selection of theories and empirical data in science
takes place in a number of steps.
There are several
selection levels.
Selection takes place when one passes
from one level to another.
At each level selection is
determined by local considerations.
Because of these
different
selection
pressures,
variations
will
be
transformed on their way to the global level of science.
Each level has its
own institutions
that deal with
selection, and such levels can to a large extend be
identified from the accompanying institutions.
I mention
here: the individual researcher, his laboratory journals,
his
research
group,
internal
reporting, journals, review
172
L.BOON
articles, handbooks, textbooks. I will pay some attention

to the nature of selection that operates on these different
levels.
Individual scientists effect an initial selection of
their ideas an results.
Their perspective on the concrete
problem they are working on is necessarily restricted by
socialization, background-knowledge, and the researchers in
their immediate environment.
The "objective" restrictions
operating in a tradition are strengthened in this way.
Especially the dependence on immediate colleagues has not
been given the attention it deserves.
A dramatic example of the latter is presented by the
following episode in Watson and Crick's research on DNA
(Watson, 1968, ch.26). Watson used tautomeric forms of the
four nucleotides as he had found them in chemical textbooks.
The tautomeric forms he had culled from these canonical
sources suggested a model of like-with-like base pairing.
The accidental presence of a colleague, Jerry Donohue, saved
Watson and Crick from proceeding down a blind alley. As a
specialist in
the
particular
field
in
question, Donohue
could inform them that the textbooks arbitrarily preferred

specific tautomeric forms.
Watson and Crick were using
forms the specialists no longer adhered to.
With the new
correct tautomeric forms, like-with-like pa1r1ng was no
longer feasible. At first this was disappointing. However,
as it turned out, complementary based pairing was not only
possible but would in fact become one of the keys to the
later discovery. Without Donohue their work would have been
delayed for months and perhaps the crucial discovery fallen
to someone else. This episode not only underlines the role
of chance in local context but also the fact that knowledge
on the global level of textbooks is often outdated by the
time it has passed the complete hierarchy of selection
levels.
Solving a research problem is not something often
accomplished in one jump.
Many variations are tried out
before the researcher starts writing
an article.
In
discussions with his colleagues, ideas and proposals are
eliminated.
Watson and Crick had tried out a number of
models before they sent off
their
paper to Nature.
Experimental setups or ideas that don't work are of personal
value only.
They remain within the laboratory. Selection
at
this
level
has
a
strong
verificationist
or
SCIENTIAC PROGRESS WITHOUT RATIONALITY
173
confirmationist
bias.
Only
potentially
positive
contributions, those that fit the tradition, are worthwhile
pursuing.
Results that refute or problematize a tradition start
to become interesting at higher
levels of selection.
However, because of the interconnectedness of cognitive
elements on these levels, isolated refutations or anomalies
will pot carry much weight.
The network harbors sufficient
ploys to parry such an isolated assault.
Only when
anomalies start to heap up does the course of research start
to change, and even then only when a new candidate for
further research has presented itself. Hence, on lower
levels of selection, refutations are uninteresting, while at
higher levels refutation is a long winded, cumbersome and
complicated
process.
Perhaps this
explains why the
methodological ideologies of scientists so often have a
verificationist bias, and why they so often display an
emotional revulsion
for
falsificationism
(Mulkey and
Gilbert, 1981). On the level of the research laboratory and
research group,
the characteristic forms of selection
involve 'indexical' negotiation processes on the viability
of hypotheses or findings.
The constructivism of these
negotiations is itself a variation-selection process through
which researchers are helped to filter the expected signals
from the noise.
Once the article that communicates their
his
work is written,
however,
all traces of these
negotiations are eliminated.
The same goes
for the
negotiations that take place between the
editors and
referees of journals.
Almost all of these will have
vanished from the published version.
In part scientists orient themselves in their field
through
personal
contacts.
For their perspective
on the
front line of research, however, journals are primary. Here

a scientist finds recent results, new interpretations, new
techniques or new standards of precision characteristic of
recent research.
Journals form the most immediate feedback
from the public selection levels to individuals and groups.
On the level of journals, the methodological justification
is central: the scientist must indicate in detail which
design he has used, which apparatus, etc. The reader must
be precisely informed as to what he can and cannot take for
granted when he wants to use published outcomes or ideas in
his own research.
L.BOON
174
Of the
immense
mountain of articles
that are
published, the bulk is never cited or mentioned again (de
Solla Price, 1963, p.49). Only a relatively small number of
articles is used by others in further research and is cited
accordingly.
In fast-growing fields, the life span of
research - the period during which other people refer to it
is short.
The bulk of published articles disappears
behind the horizon within a couple of years.
The immense
volume of research that thus proves its redundancy, shows
that the selection performed by editors and referees is
relatively mild.
The necessary conditions for a piece of
science to be published must be lenient.
These facts
underline what has been said before on the minor role of
negative selection in science.
Only the use that others
make of published research, in their own local problem
situations,
effects
real
selection.
In
science
reproductive success is primary.

Selection is positive.
The "fact that editors and referees
cannot perform a
stringent selection
illustrates
the
blind nature of
selection in science.
As the value of research lies in its use by others,

journals are first and foremost information diffusers.
Whatever active selection is undertaken by jourals is of a
conservative" nature: Results and interpretations are not
allowed to deviate too far from what has already been
accepted.
The risk here is that important innovations are
rejected; the obvious advantage is that this policy enlarges
the opportunities for others to use published work.
The
fact that the history of science has shown a host of
examples of innovations that were neglected at first, only
to be later rediscovered as important, confirms this.
The carefully conservative,
but relatively mild,
selection on the level of journals ensures that the content
of science remains within the general confines of the
program.
The mutual relations within the network, the
interconnections, are not yet in the limelight.
" A more serious concern with coherence is displayed in
review
articles.
These conventionally aim at
overviews of
the state of the art in a

field.
They list those
contributions that have often been the starting point of new
research and place them in a broader framework.
On the
basis of such a broad perspective on the near past, it often
becomes possible to indicate explicitly where the major
problems are
necessary to
175
situated, what sort of equipment will be

deal with them, and what standards will be
required of research in the coming years.
A few words on
the differential role of
journals in
the various sciences are in order here.

Journals and their
specific characteristics can function in this way only in
those areas where the cognitive grid is firm enough for

researchers to build upon each other's work. The firmer the
network, the more important journals become as a means of

communication. In the natural sciences the journal article
has virtually replaced the book

contribution to science.
as
the preeminent creative
Barring a few exceptions, books in
the natural sciences are handbooks or textbooks: systematic

overviews of knowledge for newcomers.
In the social
sciences,
the institution of
but the cognitive structure to
the
natural science is
vehicle for creative
the journal is
allow it
lacking.
to
Hence the book
contributions in
remains a
these fields.
in such books, the whole discipline will be

projected from scratch.
Via congresses,
also present,
function as in
Often
structured or
journals and review articles a number
of variations finally become codified in textbooks. It is

here that traditions or disciplines are endowed with a
balanced,
selection
coherent structure.
After passing a number of
thresholds and a form of downward causation,
influencing other work, here at last unity and stability

arise. The resulting picture of a field as sketched on this
level, often based on a variation-selection process that has
spanned decades, is in some ways misleading. The movement,

the chaotic zig-zagging of the research process, is frozen;
by the time results are incorporated in textbooks, they are
often already outdated.
No wonder active researchers focus
on journals.
Textbooks are for socialization. Students
learn the trade from them, but when they start their own
research in a local situation, they will have to unlearn
what they know to some extent.
Of course students are not the sole public at which
textbooks are aimed. They also have relevance for scientists
who are indirectly linked with a field by the necessity to
borrow some
of
its elements.
As
the logic
of evolution
processes leads increasingly to specialization, textbooks

are the only entrance to a field for outsiders, who are not
able to judge which of the many elements available in the
immense quantity of journal articles are relevant for them.
The risk of using outdated science cannot be evaded.
176
L.BOON
No single individual commands all our scientific

knowledge.
The edifice of knowledge is too immense and too
labyrinthine to be known
by one
person.
Individual
scientists work side by side on a whole of which they do not
even know the contours.
Even in single disciplines or
specialities, order can only be attributed with hindsight.
5. CONCLUSIONS
A conception of science
as
a
blind, hierarchically
structured selection process no longer needs rationality as
the driving force of science. Researchers tend to focus most
immediately on their local context and may well employ all
sorts of methodological criteria to judge data and theories.
They may well be 'intending' to find the truth about the
universe. Nonetheless, such rationality does not determine
the global end-product of their endeavors.
Individual
actions have a limited range only, because of the hierarchy
of selection levels.
Individuals have no power over the
ways in which
other people use and
transform their
ideas.
Among local situations the selection environments
and selection pressures differ.
Again, individuals exert
little else but a marginal control.
It is only the
interdependencies - social and cognitive - among scientists
that keep the totality together.
In an evolutionary theory of science, methodological
rules are no longer the prime movers of progress. Hence the
question whether someone preferred a theory for the right,
rational,
reasons
no
longer
arises.
Structure and
transformation on the global levels no longer need to mirror
events on the local level.
It is precisely because
methodologists have always derived their criteria from the
end products of science and expect these same criteria to be
operative locally, that they must also appeal to a cunning
of reason when it turns out that those rules are disobeyed
locally.
For exactly the same reasons they
had to
presuppose a
solid consensus in
science or -
even worse -
the notion that consensus was a goal of science.

A
variation - selection model makes no such demands. The fact
that locally dominant interpretations are contested globally
has as
a counterpart that
globally
dominant
theories are
contested locally. Consensus plays only a peripheral role in
science.
Whatever consensus there is
in
science
has the
character of Wittgenstein's family-resemblances: At no time

is one cluster or set subscribed to by everybody.
SCIENTIF1C PROGRESS WITIIOUT RATIONALITY
177
Though it has no need of rationality, neither is the

variation-selection model relativistic.
It proceeds from
the fact of the growth of knowledge, of its progressive
development.
That there exists a diversity of local orders
does not lead to the assumption of global disorder. One may
even claim that the development of science gives us an ever
better, ever truer picture of the real world, as long as it
isrealized that the approach towards truth is an unintended
consequence of the research process: Global Virtue from
local vice. The wish to find the truth is not sufficient to
explain our success in approaching it.
EVOLUTIONARY EPISTEMOLOGY AND

SOCIOLOGY OF SCIENCE
Karin Knorr Cetina
Universitat Bielefeld
Department of Sociology
1.
INTRODUCTION
It is uncommon for philosophers to appeal to sociology in

support of an epistemological theory, and it is even more
uncommon for them to recognize and accept sociological
findings when they are unsolicited. Recent sociology and
anthropology of knowledge is a case in point: though some of
us
have
on
theoretical
succeeded
occasion
controversies
in getting
found
with
themselves
embroiled
philosophers,
them to notice
we
in
rarely
our empirical results
(1). EE refers to a relatively recent attempt

to extend
the theory of b'iological evolution to scientific change. It
is the first coherent epistemological program
that is
empirically grounded (2), and under the leadership of Don
Campbell it has shown none of the predeterminations with
which
philosophers
of
science
tend
to
fend off
sociological
inroads upon their territory.
In fact,
Campbell's challenge to sociology of science to come up
with interesting answers to the many problems of scientific
change is serious and sincere. But can we meet it?
And
does not our methodological orientation prevent us from
cooperating with the only epistemology that till now has
seriously espoused sociology of science?
And that, for
better or for worse, has felt "unanimously rejected" by the
new sociology of science?
I think we can do better than that, even though the
benefit for either side may be less than clear. For
example, my own methodological orientation has been more
phenomenological
("constructivist lt )
and
microscopic
than
evolutionists are generally keen on, and little imagination

is necessary for someone to envisage the quarrels which
might arise between a constructivist and an evolutionary
orientation. On the other hand, as Geertz says (1983: 19f.)
there has been an enormous amount of genre mixing in
intellectual life in recent years, and one might go as far
179
W. Callehaut and R. Pinxten (eds.). Evolutionary E.pistem%gy, 179-201.
K. KNORR CETINA
180
as to suggest that this jumbling of varieties of discourse

has added to the success of such writers as Kuhn, Feyerabend
and Foucault.
So, presumably Campbell is on the right
track, and in this paper I shall not hesitate to join the
jumblers and to compromise on my own preferred discourse
orientation.
That
the formula of
mutation
translates easily into a number
and
natural selection
of theoretical vocabularies
is
hardly
deniable.
It
can
apparently accommodate
successfully
Popperian
falsificationism
and
Merton's
normative functionalism (Hull, 1978), interest models (it
can be applied to Pickering, 1980), and, if I understand
Campbell (1986) correctly, hermeneutic interpretation. Of
course, this expansionist success of evolutionary theory can
to some degree be explained by the fact that the analogy is
drawn on a very general level, and it may profit from the
fact that Darwinian concepts have long filtered into lay
discourse
and
social
science
disciplines.
It
is
disconcerting nonetheless, for it seems that the more
phenomena
one covers
by
a trim
conceptually parsimonious
explanation of the social world, the less one is actually

saying about it. And the more easily a conceptual framework
lends itself
to
translation into
other
vocabularies, the
more frequently one might ask what we gain from

translations.
doing these
In a sense what I want to do in this paper is just

that: arguing from my interest in scientific work I shall
ask for which parts of the evolutionary imagery the above
translations appear fruitful in that they provide access to
or appreciation of hitherto little examined aspects of
scientific practice and for which they do not. In essence I
shall argue that the use of the theory of biological
evolution as a model of scientific discovery does not
commend itself to the analysis of scientific work. The
evolutionary model collapses some rather complex processes
which we are only now beginning to understand into singular
events (conceptual mutations), which, themselves unexplained
and unexplainable within evolutionary theory (they are
chance occurrences), are burdened with doing all the telling
there is about scientific innovations.
On the other hand,
the theory of natural selection, I shall argue, sits rather
well with some sociologists' of scientific practice beliefs,
and might even suggest some interesting challenges to more
established science studies concepts.
If the notion of a
lSI
EVOLUTIONARY EPISTEMOLOGY AND SOCIOLOGY OF SCIENCE
conceptual mutation, of a "meme" or of other essential

components of mutation theory comes as close to being an
essentially contested concept as any in the social sciences,
the idea of natural selection, properly reformulated, might
become essentially uncontested in social studies of science
(a few remaining rationalists notwithstanding).
Thus I
believe that the full weight of the evolutionary analogy can
probably not be coherently extended to scientific change on
a
concrete level,
but the natural selection
perhaps even catches its tone and

sociology
of scientific
knowledge
scientific practice results.
part can, and
underscores recent
and
sociology of
2. THE USE OF THE EVOLUTIONARY ANALOGY

AS A DISCOVERY MODEL
Most advocates of EE do not propose

to extend biological
explanation to scientific behavior.
Instead they propose
to use the theory of biological evolution,
or some parts
thereof, as a model for scientific
evolution.
The
first problem for someone reasoning analogically from an
established theory is to identify the analogues in the
system to which the model is to be extended.
In EE,
virtually
everyone
ultimate source
who
writes
equates
of variation in biological
mutations,
the
evolution, with
conceptual innovations.
As an example, consider first Toulmin's theory of
scientific evolution, best summarized in an article he wrote
in the American Scientist in 1967.
As most evolutionary
theorists, Toulmin distinguishes between the generation and
the selective survival of ideas, a distinction which mimicks
the traditional philosophy of science distinction between

the context of discovery and that of consensus formation.
Accordingly, at any given time in scientific thought we have
a pool of scientific innovations and an ongoing process of
rational
selection
among
these
innovations.
Conceptual
variants are
the
products
of
individual scientific
innovators; their merits are collectively judged and their
fate decided by communities of specialists.
At any given
time, intrinsic (or intellectual) and extrinsic (or social)
factors jointly affect conceptual variation. For example,
social factors may limit the occasion for scientists to
speculate freely, and intellectual factors may help them to
focus their theorizing on promising lines of thought.
Selection,
on the other hand, in Toulmin's model is
182
K. KNORR CETINA
ultimately
rational.
In the common
situation where
disciplinary
problem
solving
is
based
upon agreed
disciplinary goals and standards, the selective perpetuation
of conceptual
variants
is
straightforward~y rational
(intellectual).
When scientific work
pertalns to the
"rational frontier" (Toulmin 1972, p.241) where standards of
selection are themselves up for reappraisal, scientists may
have to take "rational bets" (1972, p.246). These may not
be arrived at by accumulations of bare facts of Nature, but
they are nonetheless "objective" in the sense that they are
backed by scientists' collective experience and that they
will be judged, in the long run, in the light of their
actual practical sequels for our understanding of Nature and
not by personal considerations (1972, p.244).
with Toulmin the published but not yet accepted
products of scientific work
constitute
the
pool of
variations.
Campbell's
well-known
(1974)
theory of
scientific evolution differs from
Toulmin's in
a number of
respects.
Most noteworthy, perhaps, is his much broader
perspective on variation and selection which prompts him to
include among conceptual innovations the novel not yet tried
out ideas of laboratory work and among extrinsic factors the

many flaws in human visual perception.
He has also chosen
to be less specific than Toulmin about selection standards,
and wisely so as I shall argue later, despite the critics
who attacked him on this lack of specificity (e.g. Corning
1983, p.247).
Partly because of his broadening of the
discovery
model
to include
and commend
"wild
speculation" and partly because of his preoccupation with
the biases of visual perception (and partly because he wants
to offend), Campbell insists on calling variation "blind"
and "unjustified" (1974, p.147ff.). Where Feyerabend uses
the slogan
"everything goes"
to advocate counter-induction
and the proliferation of theories (1970, p.26), Campbell

employs the somewhat polemical aphorism of blind variation
to refer to the discovery of new scientific theories through
'underjustified' speculation (1974, p.152).
No
modelled
resistance
aspect
upon
than
of
theories
biological
their
of
sociocultural
evolution
specification
evolution
has
caused
of
innovation
more
as
conceptual mutations (e.g.

Corning 1983, p.20S), and no
aspect of Campbell's theory has provoked more criticism than
his calling variations "blind" (e.g.
Hull, 1982). The
arguments against the case for mutation-like innovations in
socia-cultural systems,
scientific or otherwise,
have been
183
discussed in
terms
of
'coupled'
versus
evolution,
Lamarckism
versus
Darwinism
'uncoupled'
and
human
conceptual
seen
intentionality versus the chance occurrence of variations.

These arguments are well-known and need not be repeated here
in detail. In essence, most discussions of the problem boil
down to the claim that sociocultural innovations, including
those of science, are intentionallY,designed to survive the
selection process
which
decides
their fate, whereas
biOlogical mutations arise in a spontaneous, random manner
unrelated to the environmental factors which account for
their selective perpetuation.
Toulmin acknowledges the
disanalogy but seems to think it is unimportant (1972,
p.337), to the consternation of his critics (Hull, 1982,
p.17).
Campbell, more belligerent than Toulmin on this
point, attempts what I consider a rather successful rebuttal
(1977, p.503).
No matter how reasonable and intelligent
scientific inquiry is, and no matter how persistently
scientists strive to come up with conceptual variants that
will survive the selection process, innovation cannot be
explained by the restraints or guidance provided by existing
conceptual systems. Discovery by definition cannot strictly
be a matter of producing desired conceptual innovations on
If it were, scientists would
demand (Hull 1982, p.37f.).
not really go beyond what they already know, but merely
enact a culturally acquired knowledge program. Besides, as
Campbell and others have argued (1974, p.147ff., Corning
the
mere existence
of
restraints
caused by
1983)
goal-directedness and previous
knowledge on conceptual
mutations is not necessarily a disanalogy to biological
evolution in which there exist a number of restrictions
which limit the
range of variations
available to a
popUlation.
Now I agree with Campbell that the purposiveness of
scientists' behavior may not be the key to the disanalogy
between biological mutations and scientific innovations; but
I
also
believe
that
problems
remain
nonetheless.
Intentionality is a notoriously intractable internalistic
concept.
I shall follow a different path, and begin'by
looking at the initial recognition of an innovative idea. I
want to borrow a term used by Campbell (1974, p.155) to
suggest that discovery in science may be better understood
as an editorial process rather than as a process of
innovation,
and
hence
must
be
phenomenon rather unlike biological mutation.
as
K. KNORR CETINA
184
3. DISCOVERY AS AN EDITORIAL PROCESS

As everyone
knows,
concepts are
constructs: they construe
the world in a specific (selective) way. But they do so not

only with respect to their propositional content. They also
come with certain illocution-like pretensions; for example,
when they arise in scientific work, with the pretension of

being a "good idea". When we refer to a conceptual event as
a "good idea" or simply as an "idea", we usually mean it is
a fortunate thought potentially valuable in solving the

problems at hand or in opening up new problem solutions. We
do not mean that the thought was irrelevant, or
that it
was a nuisance and suggested troubles.
Hence with respect
to innovations, the point is not only that they are

accidents in the sense that they are unforeseeable and
cannot
be produced at
mutations),
mutations,
which
but
will (and hence are

that
perceive
they
and
are,
like biological
unlike biological
fortunate accidents, by which I mean occurrences
scientists
also
interpret
(preselect)
as
felicitous
opportunities
for
success.
Thus for an
occurrence to enter the selection processes of experimental
trial and disciplinary IIconsensus formation ll ,
it must have
been preselected, first by being recognized as an 'idea' or

an opportunity for success, and second by being judged
feasible under the circumstances and worth to be tried out.
Primary imputation (3), as one might call the recognition of
something in terms of certain implications, refers to this
process of preinterpretation.
The point here is that
a
linguistically coded
occurrence acts as an instruction to
react in
terms of the
coded formulation,
and thus has real consequences in
practical action. If the thought that crosses my mind comes
through as a worry, I shall do what I can to prevent it from
becoming true. If it comes through as a promising new idea,

I shall want to pursue it, and create the circumstances
which will help it to work out. Thus there is no such thing
as a pool of unclassified conceptual mutations waiting for
later selection processes to take hold.
A "bad" or
"irrelevant" idea may go unnoticed or stimulate repairs, but
it would seem to have no chance to become part of the pool
of variants
available for (experimental)
found successful scientists to
selection. I have
pride themselves
to be good
at just that: the recognition of the potential for success

of a novel problem, concept or technique.
What was wrong
with a lot of unsuccessful scientists, one of them told me,
185
was that "they were not dumb, they just worked on the wrong
things" (Knorr Cetina, 1981, p.74). The clues one gets from
such scientists refer to their ability to discriminate,
which initially means to recognize an occurrence as an

opportunity for success. In this connection it is good
to
remember that scientific 'innovations' have many origins,
among which novel thoughts by individual scientists are

but
one.
Another
well known
source of scientific
innovations
are
unexpected
experimental
outcomes
(anomalies), which may be as prosaic as the smell, the feel,

or the look of a substance.
still other sources lie within
the local environment, as when someone unrelated to a
project makes the right comment at the right time or when
an apparatus suggests a particular problem solution. In all
these cases, it is clear that what is at stake is a
scientists'
ability to recognize the potential of an
occurrence.
In other words, the achievement here is one of
discrimination and
selection,
otherwise qualified variation.

as
and
not
of
unjustified or
To repeat, the linguistic designation of an occurrence
"innovative"
or
as
an
"idea"
invariably
involves a
process of imputation which,

within
the
imagery of
evolutionary theory, would seem to be more appropriately
rendered by a selection model than by mutation theory.
since there is no such thing as a
pure, objective,
once-and-for-all scientific innovation irrespective of its

designation as an innovation, there is also no such thing as
a proper equivalent in scientific

work of biological
mutations.
I have referred to the process of imputation as
editorial, and I now want to add two observations which
extend its scope.
First, interpretativity (selectivity)
does not stop with initial designation.
Scientific works
are
construed
as
novelties
through
editing
practices
throughout their laboratory life and in print, not to

mention the reconstructions by posterity,
such as by
historians
of
science.
They
are
also
of course
deconstructed as novelties, often by their own originators
whose work moves on, and by the uses of these novelties.
Users
exemplify
sophisticated
literary
skills
in
highlighting their own originality while de-emphasizing the
contribution of those upon whose work they draw. Of course
on occasion it may also serve the purpose of a user to
'rediscover'
and applaud an
earlier
original inventor in
order to outfight a contemporary who aspires to the honor of

being first in a priority struggle - as Brannigan (1979) has
K. KNORR CETINA
186
shown with
the "rediscoveryll
respect to
of Mendells laws.
These strategies vary, but they would all seem to imply that
novelty is the outcome of locally and historically variable
- and reversible - imputations of novelty negotiated between
participants from within and without science.
In essence
this is what attribution theory predicts and what has been
established in at least one extensive historical case study
(Brannigan, 1982).
In accord with calling the recognition
of an occurrence as
a
potential
innovation primary
imputation, we might refer to these later designations as
secondary imputations.
Yet the discovery process, and this is my second

observation, is only partially captured by attribution
arguments.
into
Construal as
all aspects of
novel
scientific
or
interesting penetrates
inquiry:
scientist may
choose a technical instrument because it is novel and rarely

available, seek out literature not generally known in his or
her area,
give preference to procedures associated with the
frontiers of knowledge,
or buy apparatus perceived
him or her an edge over the competition.
to give
Thus construal as
novel is not just a matter of simultaneous (primary) and

negotiated (secondary) imputation.
It is also a matter of
articulating the product in and through scientific work.
This aspect of the editorial process is more akin to the
writing of a work itself than to a publishers' cosmetic
touch-up operations. If the initial editing of a thought as
an innovative
idea
is
primarily a
matter
of "seeing-asll
(primary recognition), and if secondary editing involves

mainly processes of negotiated imputation (attribution), the
third aspect of the editorial process I have mentioned might
best be described as a matter of work-articulation. Recent
empirical studies
related to the subject
imply that
discovery
think
that
is
an
an editorial
process in all
understanding
of
the
three senses. I
editing
strategies
involved amounts to a theory of scientific innovation.

Now to call discovery in science an editorial process
invokes the text analogy.

It shifts the idiom from
intentionalist approaches taking their
lead
from the
psychology of mind to a phenomenalist perspective content
with reading the surface of things.

Innovations are not
seen as unaccountable hot s~ots,
little explosions of
spontaneity and creativity,
~n
otherwise drab everyday
routines. Rather they are seen what participants "read' and
"write" (articulate) them to be in the discourse and text
building of scientific work. Almost automatically, then, the
187
vehicles of scientific 'innovation' become situational and

social processes, observable from the outside and prima
facie not so different, except in topic of course, from
other processes in everyday life.
Mind-creativity
is
a
phenomenon
intractable depth.
But if we avoid it
scientific
discovery as
an
of
by
editorial process,
seemingly
describing
do
we not
loose the phenomenon of "real" conceptual innovation? The

problem with
this question, I think, is that it identifies
scientific innovation with certain combinatorial skills
participants may display in taking advantage of their
environment, experience or education.
These skills, I
submit, may account for a certain percentage of conceptual
variations,
but they
customarily
call
do
not account
for
"discovery", or
scientific "innovation".
Why not?
Let me expound the
thesis by a brief excursion into the metaphor or analogy
theory of scientific innovation.
As has been argued (Black, 1962), much of what we
scientific
innovation involves reasoning
from analogy, that is it results from bringing to bear

pre-existing and pre-tested concepts from a context in which
they have worked in the past upon a less understood problem
area.
Now the crux of the matter in reasoning from analogy
is not that an idea is new but that it is old; it is a
pre-tested solution in a context similar to the one under
current
investigation.
In
evolutionary
terminology, the
content of the variant is not novel; what is novel is the

connection made between hitherto unrelated problem contexts
and the insertion of a familiar idea into a context to which
it had not previously been applied. According to a number of
authors,
this
"displacement of concepts",
as it is
appropriately called, may be the single most important
source of our conceptual extension of knowledge (Hesse,
1970; Schon, 1963; Mulkay, 1974). If this is true, and in
my own limited laboratory experience I find reasoning from
analogy quite common, then we have here a phenomenon that is
eminently based upon cultural learning and transmission and
very unlike biological innovations.
After all, biological
mutations, whatever else they are, are not transfers of
pre-existing and pre-tested genes from one context to
another (though with the rise of biotechnology, they may
become so).
But be that as it may, the point I wish to make is
that displacements of concepts based on metaphor and analogy
are routine features of scientific (and everyday) reasoning.
188
K. KNORR CETINA
It can be shown (Knorr Cetina, 1981, ch.3.6) that they occur

as frequently in the case of "blind alleys", "degenerative
probleQl-shifts", or simply failures to make something work
as they do in
"innovative"
successes. Thus, a theory which
accounts for scientific innovation solely in terms of

metaphor and analogy is at the same time a theory of
scientific failure and mistake.
It does not discriminate
between the different fates analogy-based ideas encounter in
the process of research and in scientific publication. The
metaphor and analogy theory of conceptual displacement is at
best a theory of the creation of variants. It fails to take
into account the editorial processes through which variants
become stylized as scientific innovations.
To conclude, I think the example of the metaphor and
analogy theory suggests that it might pay to distinguish
between cognitive and other mechanisms which account for the
creation
of
variants
and
processes
which
generation of "discoveries" and "innovations".
explain
the
If we equate
the latter with the former, we not only disregard the larger
part of the picture discovery represents in scientific
practice,
but
we also misconstrue
a
phenomenon as
exclusively mental in which mental factors may play only a
small and in any case
a non-decisive
role.
Innovations do
not lie at the beginning, but rather at the end of processes

which include retroactive imputations on the basis of the
consequences of previous events.
In other words, they are
the result of
participants' "ethno-epistemology" work,
whereas variations are not.
paper.
matter
I shall return to
this point at the end of this
But first I want to stress that it is quite another
to explain
variation than it
is to explain
mind-creativity and scientific 'innovation'.

Consider that
scientists do not have to wait for scientific 'mutations'.
The scientific literature, and the rest of culture, provides

a pool of stored up variations ready to be called forth at
any time by those who can make the appropriate connection.
Moreover, many variations may come about through utterly
uncreative activities,
for example through adapting a
technique or other finding to local conditions, to the
special demands of a new project, or to personal knowledge
and expertise.
Thus, variations may have more to do with
the local requirements of knowledge production and with the
variability of discourse communities than with anything
akin to individual "creativity" or "innovation",
189
4. THE USE OF EVOLUTIONARY THEORY

AS A MODEL OF SELECTIVE RETENTION
I shall now turn to the use of the theory of biological
evolution as a selection model.
As Campbell once said,
there may
be great
conceptual gain in
moving from
teleological causation to natural selection explanations of
teleological achievement (1974, p.145). These gains, I
think, have not been sufficiently exploited by evolutionary
epistemologists.
In order to appreciate the potential of
evolutionary theory as selection theory, let me first recall
what natural selection is not.
I shall draw heavily on
Corning's summary of Darwinian evolution (1983, p.30ff.).
First natural selection is not a mechanism or a
principle or an agent extrinsic to the relationship between
organisms and their environment. Rather, "natural selection"
is a shorthand way of characterizing a vast array of dynamic

processes by a term derived from the "artificial selection"
of plant and animal breeders
which inspired Darwin's
concept.
Whereas artificial selection can be seen to be
teleological in
the sense that it
involves external
selecting agents who deliberately select, with an eye on the
future, the traits for which they are breeding (Nagel, 1979,
p.303), natural selection is "a non-teleological process in
which the multifarious
their
interactions
environment can produce,
between
organisms and
without external direction,
the ...
emergence of functional design (or adaptation) in
nature".
Phrased differently, the term natural selection
refers to a variety of factors which in an given context are
responsible
for
causing
differential
survival
and
reproduction among (genetically) variant individuals, and
for absolute changes in numbers and diversity of different
populations over time (Corning 1983, p.30-31). Recall the
textbook example
of industrial
melanism
as recently
summarized by Corning (1983, p.31).
Before the industrial
revolution, a light colored strain of the moth (Biston
betularia) predominated in the English countryside over the
darker, "melanic" form.
Apparently, when the light moth
rested on tree trunks it was almost invisible against the
then light trees, while the dark form "stood out" and was
preyed upon more frequently by birds.
With the industrial
revolution, smog from factories blackened the tree trunks in
industrial areas, and the relative visibility of the light
and the dark form was reversed.
Now the darker moth became
190
K. KNORR CETINA
the one less subject to predation, and in due course

increased in frequency relative to the lighter strain of the
species.
Now natural selection is not located in any single

factor that led to the reversal; neither in the human
purposes responsible for the industrial revolution, nor in
the perceptual abilities of the birds that switched from the
darker to the lighter variety of moth; neither in the smog
that blackened the tree trunks nor in the moths' wing
pigmentation.
Rather, it refers to a IIconfiguration of
factors" that combined, in a specific context, to produce
differential survival and reproduction
rates.
It might be
described as a sort of vector sum (Wright, 1934) or, less

mathematically speaking, as the upshot of a number of
external and internal influences such
as the resting
behavior
of
the moth,
its
polymorphism
for melanic wing
coloration, factors named only recently such as an augmented

fertility
in
the
melanic
form
and
differential
susceptibilities to variations in air temperature and sulfur
dioxide levels, as well as of course industrial pollution

and the feeding habits and visual detection system of the
birds.
Thus the notion of natural selection must be
disaggregated,
and if this is done
a
multitude of
interacting local factors can be seen to contribute to it.
Now I am dwelling upon this point in order to remind
us that natural selection is not a sculptor but a historical
process
which bears a number of
resemblances to historical
selection processes in social life.
Concretely, the process
of natural selection can be characterized as follows:
(1) It is local, in the sense that differential selection

among
populations or individuals happens
within, and
relative to particular environments.
(2)
In addition to being local,
the process
is contextual
in the sense that the specific constellation of factors

involved in a particular selection effect is historically
contingent and cannot be predicted concretely in advance.
(3) Furthermore, the process is open-ended and without

inherent direction.
It is not propelled forward in one
direction
by
some
law-like
principle which
accounts for
evolutionary progression
although retrospectively, of
course, certain past trends may be discerned.
To these characteristics on which there seems to exist
a broad consensus among evolutionists (see Jacob 1977;
Corning 1983,
that
ch.2). I would like to add a fourth, which is
191
(4) From the point of view of selection theory, the process

needs variations, but does not require us to adopt a model
which locates the

mutations.
origin
of
these
variations
in genetic
There are other aspects of evolutionary theory, of

course, on which there is little consensus among biologists,
such as the question whether the process is incremental or
includes what Mayr (1975) called "genetic revolutions" (see
also Gould & Eldredge, 1977). However, these, aspects need
not concern us here, for whatever the outcome of the
respective controversies among geneticists, it is unlikely
to change the basic premises of evolutionary selection
theory outlined above.
The question for us is rather
whether' the respective factors
can also
be seen, to
characterize selective retention in science and whether the
resulting model offers a plausible alternative to currently
predominant concepts of scientific consensus formation. I
think it does.
To argue my point, I will begin with the
first premise.
5. SELECTIVE RETENTION AS A LOCAL ACTIVITY
Natural
selection
is
local
in
the
sense
that
situation-specific effects such as the relation between the
wing coloration of a moth and the color of tree trunks in
the
English
countryside
become
the
cause
of
transgenerational continuities and changes (Corning 1983,
p.32). Phrased differently, causally efficacious effects in
biological evolution are inherently bound up with the
environment in which they arise.
They do not normally hold
for all members of a species
across different .environments,
regarding social structures.
Cultural environments no less
nor
natural disasters notwithstanding
, for all
environments irrespective of their inhabitants.
Now in
cultural evolution environments are not necessarily defined
by a shared locale or by physical proximity. Instead, they
might be defined in terms of frequency of interaction and
accessibility, a social structural phenomenon dependent upon
participants' awareness and access strategies. These are
not, however, to be confused with outsiders' (say social
scientists' or philosophers') similarity classifications
than natural environments must be "lived" ("transacted") in
order to be effective.
To portray them adequately within
theory, the analyst has to adopt what Edge once called "a
radically participant-centered perspective".
want to
192
K. KNORR CETINA
claim that scientific disciplines, considered by Toulmin

(1967, 1972) and others as relevant selection environments
in the sciences, are based on historically rooted similarity

classifications, which are reflected in administrative and
educational
structures
but
have
little
to
do
with the
working environments within which technical choices are

routinely made.
These
working
environments
are not
constituted by the content of disciplinary textbooks. Nor

are they identical with administrative units divided by
disciplines such as the university departments or other

organizations which constitute a scientist's institutional
affiliation.
Furthermore,
the concept
of a working
environment does not refer to an identity question.
It has
nothing
intrinsically
to
do
though
it
may
at times
coincide
with
scientists' official disciplinary or
professional self-identification. In fact, what constitutes
these environments is entirely an empirical question. It
cannot be
(e.g.
decided by
disciplinary)
in everyday life.
simply adhering to
the institutional
pre-classifications which we encounter
For one thing disciplines are clearly too
large and encompass too many specialties for their members

to be relevantly involved, or even know about, each other's
selection decisions (4).

But if the discipline is something of an institutional
artifact
specialty
about the
and in any case too large,
communities which
have
what about the smaller
dominated
our thinking
social and intellectual organization
of science,
and not
only since
Kuhn (1962)?
The problem with
specialties, I think, is that like scientific disciplines
they
are
not
properly
speaking
"organizations",
goal-oriented
co-operative
little
systems,
explicit systemic goals and superordinate

They lack the property of being "wholes"
that is, with
controls (5).
by virtue of
systematic ("emic"!(6
characteristics such as organized
action in concert or political superstructures. Instead,
they are wholistic constructs, from the analyst's Petrarcan,
"scenic"
perspective,
from which the fissured landscape of
specialty members'
local
dealings
appears
like the
integrated cosmos of a nature painting.
The design of the
various
parts
of
the
human
body
is
never
for
long
independent of the whole.

As someone once said, "When I
decide to go to dinner almost everyone of my roughly ten
trillion cells is constrained to go with me" (7). But when
the member of a scientific specialty publishes a paper, no
one else is constrained to take notice, let alone to react.
193
It is only if some participants decide to react for their

own local reasons - some of which may have to do with each
other
that a "collective" outcome based on the sum or
interaction of these actions becomes possible. But like the
selective survival of populations iIi biological evolution,
this is an aggregate effect, not a purposeful, knowing and
organized achievement of contributing participants.
There is, then, no system or superstructure in a
specialty which makes communally organized selections. This
lack of a superordinate structure leaves us with the actual
working relationships, the arenas of interaction in which
scientists are involved, as potential candidates for working
environments.
Are these the "problem areas", "research
networks"
or
"research
areas"
in
terms
of
which
some
sociologists of science attempt to

redefine specialty
communities (e.g.
Mulkay, Gilbert and Woolgar, 1975)? I
think not quite.
For although these notions have the
advantage of bringing the concept of a specialty community
down to the size of scientific reference groups known to
participants, they are nonetheless inadequate in that they
latch on to
rather than to render problematic - the
culturally presupposed distinction between science and other
institutions.
In other words, they are built on the
assumption that the environments we are looking for must be
internal
to scientific disciplines,
or in any
case to the
scientific community at large.

But is this internalism
justified?
Recall that
evolutionary epistemologists, among others, have been keen
to admit that "extrinsic" factors playa role in scientific
innovation.
For example, Toulmin considers the production
of
conceptual variation
to be
affected
jointly by
intellectual ("intrinsic") and social ("extrinsic") factors
(1972,
ch.3).
To these
Campbell (1977)
has added
psycho-physiological and cultural factors such as those
associated with the many biases of visual perception. On
the other hand, Toulmin (but not Campbell) proposes a
rational
interpretation
of
conceptual
selection.
Considering intellectual disciplines as "organized, rational
enterprises"
(which
conceptual selections
have
denied),
are primarily or
he
presumes
that
"normally" made Ifin
the light of" relevant disciplinary considerations (1972,

p.222ff.). Now with this kind of reasoning Toulmin has many
philosophers on his side.
Until recently, philosophers
maintained that questions of discovery must be distinguished
from the context of validation, and considered the latter to
194
K. KNORR CETINA
be subject to some version of the rational interpretation.

In
the
meantime,
however, the distinction has been
challenged on a number of grounds from within philosophy and
from without (Knorr Cetina, 1981, p.7ff., 1982; Nickles,
1980).
One reason for rejecting the distinction refers to
where selection processes in science are located. For it
can be argued, and I have done so elsewhere in detail, that
the context of discovery is itself the locus of scientific
selection,
the place where the products
of previous
scientific work
become
singled
out
and selectively
incorporated into new scientific inquiry. It has often been
maintained, correctly I think, that the value of scientific
results lies in their usefulness for further res~arch, that
is in the degree to which other scientists decide to built
upon these results in their own research.
But
where would
these
research decisions
take place if
not in the
laboratory, that is in the very context of scientific
discovery?
What is consensus formation if not a proces of
selective affirmation of knowledge claims in and through
further research?
Surely it is not a process of opinion
formation disengaged from actual laboratory decisions. And
even if such a process existed, it would count for little,
for it is quite clear that disengaged opinions have an
oblique, if any, relationship to action. What counts with
respect to the final survival of a knowledge claim is
plainly scientific practice, and this takes place at the
various sites of actual research.
Now if, as I am arguing, the process in which

conceptual variations are produced is at the same time the
process in which previous variations

reproduced
or
"weeded
out",
then
are either selectively

all
factors
which
admittedly affect the production of variations must also

affect their
selective
retention.
Hence we cannot,
following Toulmin, determine the "context of discovery" to
be impure with respect to the origin of factors which enter
participants' discovery considerations and simultaneously
keep the selection context 'clean'. In terms of our previous
notion, there is only one working context of inquiry which
encompasses both discovery and selection, and which in all
likelihood is neither completely 'inside ,. the scientific
community nor external to it.
This does not rule out, of
course, the possibility that one dimension prevails at a
given time to the near exclusion of the other, at least
within the general constraints imposed by the historicity of
human affairs and by anthropological constants. But it does
rule out
internal
wholesale
specialty
considerations.
relegation
of. selection
environments
and
195
processes to
"disciplinary"
Elsewhere I have called these working environments of

"trans scientific"
and "transepistemic lt : They
potentially transcend the scientific community in that they
include relations of dependence and interactions between
scientists and members of a number of 'neighboring groups'
which are somehow engaged or engageable in the pursuit of
relevant inquiries (I have in mind users, administrators,
journal editors, funding agents, toolmakers, the industry
serving science, and others). I use the notion epistemic to
refer to considerations presumed to promote the truth-like
character of scientific results. Transepistemic arguments
traverse and transcend these considerations, and they play,
as one can show (Knorr Cetina, 1982, 1984, ch. 5.11) a
prominent part in scientist's technical decisions. Working
environments are potentially larger than problem areas and
research networks precisely in the sense that they transcend
disciplinary boundaries and "intellectual" considerations.
But they are also potentially smaller in that they include
scientists
only
"working"
relationships of interaction
and influence
actively entertained by relevant participants.

In other
words, the fact that selection decisions may be made by
scientists is not relevant - what is relevant is that these
decisions are embedded in, and hence influenced by, working
environments which encompass participants in different roles
and occupations and different discourse provinces of the
social world.
6. SELECTIVE RETENTION AS AN OPPORTUNISTIC ACTIVITY
Evolutionary theory predicts that internal and external
selection pressures will arise in natural environments, and
that those individuals or groups which are equipped to
meet the challenge or those that organize effectively in
response will be selectively favored over those that do
not.
It does not predict concretely
what selection
pressures will arise or what traits and measures will have
selective value, for the simple reason that these will
all
depend on other factors within the historical interaction
systems (Wright, 1980) of specific selection environments.
What predictions evolutionists might wish to make with
respect to specific processes of selection must be based on
concrete analyses of
local cDnfigurations and cannot be
196
K. KNORR CETINA
derived from the general theory of

natural selection
as a process
evolution. By specifying
in which situation-specific
effects cause transgenerational continuities and change,

the theory of biological evolution also tells us
what it
cannot - without case analysis - further specify,
and
where historical analysis must supplement nomothetic theory.
By dubbing selective retention in science "local" I
have prepared the ground for considering it as contextual in
exactly the same
What
sense as natural
selection
scientific
factors
product
over
matter.
in
another
selection is contextual:
the
cannot
choice
be
of
one
concretely
predicted for the

simple reason that it
depends on
historically
idiosyncratic
configurations
and
interpretations which are not governed by
any single
law-like principle or mechanism.
But
of course the
respective
outcomes
can
be
analyzed retrospectively
(historically), and educated guesses about some future
selections may be
possible
from
contemporary
analysis of
networks
of related
working contexts (transscientific
fields). It might also be noted that from the point of view
of scientific work, contextually contingent operations are
opportunistic in the sense that scientists do not only take
account of
and passively react
to
'contextual' pressures.
They also actively structure their environment in terms of

work-opportunities (and impediments).
One might even say
that
through interpretation,
rearrangement and active
participation,
create
scientists
these
environments with
whatever initial material they have at hand. Opportunism,

of course, is a matter of flexible, ad hoc reaction to
perceived
opportunities
as
they
present
themselves
in
unpredicted ways. As an inquiry strategy, it reinforces and

sustains the presumed contextual contingency of particular
selection processes. And it is of course the most important
reason for insisting that working environments should be
outlined from a participant-centered perspective.
Now if I am right in asserting a strong analogy
between
selection
in
biological
evolution and selective
retention in science with regard to the local and contextual

properties of these processes, this means that single factor
selection theories in philosophy and sociology of science
such as rationalistic interpretations or interest models are
either wrong or tautologous ("interests are whatever prompts
scientists to choose a particular theory").
This is a
strong assumption, but it has much evidence to its support.
My proposal also means that the process of selective
197
retention is as open-ended in science as it appears to be in

biological
evolution.
Again,
this is not a trivial
consequence of what seemed to be the innocent proposal that
selection in science is local and contextual. To be sure,
the assumption of open-endedness gains support from authors
such as Kuhn, but evolutionary epistemologists have not
embraced Kuhn's conclusions.
Nor have their critics. In
fact, one of the strongest disanalogies perceived between
natural selection
and scientific selection is that the
latter
is
seen
to
be
(rationally)
organized and
goal-directed, whereas the former appears to be blind. The
argument runs somewhat like the conjections advanced against
the use of evolutionary theory as a model of scientific
discovery.
For example, Rescher (1977, p.133) describes
selective retention in science to be rational, and contrasts
"rational selection" with natural selection (8). "Rational
selection is a process of fundamentally natural selection both are simply devices for elimination from transmission.
But their actual workings differ, since elimination by
rational selection is not telically blind and bio-physical,
but rather preferential/teleological and overtly rational."
In a
similar
argument,
Elster (1979, pp.9,87)
elaborates
the disanalogy
between
natural
and "rational"
selection by distinguishing between locally and globally

maximizing machines.
As Hull (1982, p. 33f.) points out, "a
system is only locally maximizing if it can react only to
innnediate
conditions".
Natural
selection
"is
at
best
locally maximizing" while, as Elster says, 'the capacity for

global maximization is a unique and distinctive feature of
man'.
Hull seems to
agree with these
authors in thinking
that the key to the difference between biological and

sociocultural
selection
is
to
be
found
in human
intentionality, but he nonetheless believes that the picture
which emerges from reading the primary literature of science
is "not
very
directional"
(1982,
p.38).
I think
intentionality is an inherently individualistic concept
which has nothing to do with the aggregate outcomes of
sociocultural selection.
As Weber proposed, intentions
refer
to
actions.
the meaning an actor attaches to his or her

A society,
a state or a scientific discipline do
not have intentions.
And from the fact that individuals are

even rational actors, it does not
follow that the actions of a number of individuals will, in
the aggregate, lead to purposively organized, let alone
intentional,
perhaps
globally maximizing
selections.
As most social scientists
198
K. KNORR CETINA
will agree,
states
it
and even
is
difficult to
maintain
smaller units such
that societies,
as universities change
through processes of globally maximizing rational selection.

Why, then, is it more plausible to assert that science,
scientific
disciplines
or specialist
communities are
governed by such processes?
Aggregates can be globally
maximizing,
or less strongly put, can have organized
supra-individual
outcomes
if
they
are
structured,
effectively cooperating systems with superordinate controls.
Or if all individuals in the aggregate are altruistic
optimizers of the common good and, with respect to their
selection strategies, exact copies of each other. But, as I
have argued before, scientific disciplines or communities
are not cybernetic systems.
Nor do we believe, after many
years of Parsons-criticism, that individuals in whatever
aggregate share enough goals, values, reasoning strategies
and information in order for the sum of their independent
efforts to result in just what the globally maximizing
In fact, evolutionary theory
outcome was supposed to be.
does not support the notion that
individuals seek global
outcomes.
Instead, it favors a notion of context-embedded
circumspection in which self-interest
and self-related
consideration play no small role.
My argument then is that given human intentionality
and purposiveness it makes sense to analyze scientists'
selections in terms of local reason, but that in the
aggregate scientific selection 's blind. Global outcomes
have the character of retrospectively adopted consequences
which arise behind the back of agents from non-communal,
often inadvertent local choices.
However, this formulation
still neglects the fact that like discovery, selective
retention in science is a double-headed (self-referential)
phenomenon.
It includes the process of selection-making as
well as participants' selective editing of this process
through recording and formating practices. These can be seen
at work not only in the textbooks and teachings of science,
but
also in
scientists'
biographies,
notebooks and
laboratory accounts, to the degree to which it is often
difficult to distinguish selection from the editing of
selection.
However,
unlike
my
ethnomethodological
colleagues who will not agree with me on this, I still think
the distinction is useful, for the following reason: In the
last analysis, all selection appears to be edited once it is
made reportable, but not all selection appears to be
"up-graded"
in
the process to
sununary
record.
As an
199
illustration, think of the video-recording of a process.

CLearly the camera edits the picture through lighting,
camera angle, lens used and other means, but it does not
normally
give us a summary record
of
occurrences (its not
doing so is precisely the problem of video-analysis). On

the other hand,
indicators derived from questionnaire
answers usually provide a highly edited and aggregated
account of the phenomena to which they refer.
If we refer
to this aggregation as a change in format, we might say that
selective retention in science appears to be procedurally
open-ended and opportunistic and editorially directional and
closed.
7. SELECTION THEORY NEEDS VARIATION, NOT INNOVATION
I want to end these speculations about selective retention
by a brief note on variation, which in a way brings me back
to the discovery model from which I started. The claim was
that from the point of view of selection theory, the process
of
selective
retention
needs
variations,
but
does
not
require us to adopt a model which locates the origin of

these variations in some equivalent of genetic mutations.
With regard to science this has the advantage of avoiding
virtually all of the objections raised against the use of
evolutionary theory as a model of scientific discovery,
including my own. For one thing, it is quite another matter
to account for the origin of variants than to explain
scientific innovation in terms of the nearly unaccountable
spontaneous generation which characterizes the birth of
mutations.
A theory of "variations" can accomodate the
claim that much of our "creative" extension of knowledge is
apparently based upon the selective mobilization of old
ideas by means of metaphor and analogy.
It can accomodate
the claim, which I think is implied by contemporary research
and theoretical work,
that there exists a tremendous
reservoir of variations in science which mainly derive from
the fact that scientists are not
very interested in
replication and
much more
components of technical work
interested in
varying the
in order to adjust to local
circumstances and enhance the probability of technical

success.
And it can accomodate the claim that in science,
the process in
at the same
are either
which conceptual variations are
time
the process in
selectively reproduced or IIweeded
this does not
produced is
which previous selections
hold for biological evolution.
out", whereas
Finally, if,
K. KNORR CETINA
200
as
have
argued,
understood as
process of
"discoveryll
a process
of
in
science
is
selective
editing
than
novelty generation,
the consequences
editing
of
this
In
short,
my
argument
say with
variations,
phenomenon
perception of variations.
theory needs to
makes
as a
no differences
with respect to a theory of scientific variation, whereas it
does deal a further blow,
I think, to a theory of
mutation-like scientific innovation. Variation theory might
simply include the point that designated novelties result
from various processes of
this
better
is
regard
for the
that
and explore
status and
whatever selection
to scientific 'mutations'
should be construed as a theory about scientific variation.

Variations can be explained in ways compatible with the
objections raised against the use of
evolutionary theory as
a model of scientific innovation, and accomodate our current

logical, psychological
scientific innovation.
and
sociological
understanding of
NOTES
1. There are exceptions of course. See for example Hesse
(1980), Nickles (1980), and Giere (1984).
2. Although it is not the first attempt to develop an
EE. Don Campbell has made me aware of the fact that Simmel,
as early as 1895, wrote a paper on the relationship between
selection theory and epistemology (Simmel, 1895).
3.
Compare Hesse's (1974)
recognitionll,
use of
the
term "primary
4. This has been argued by Whitley (1978). It is fair

to say that Toulmin (1967, part 4) sometimes refers to
specialities when he talks about "disciplines".
5. For the definition
Corning (1983, p. 210).
of
a cybernetic
system, see
6. Ethnoscientists distinguish between "ernie" (from

phonemic), participant centered, and "etic" (from phonetic)
intercultural approaches. See Pike (1967, p. 37ff.).
7. See Corning (1983, pp. 99-100).
8.
201
The following quotes and arguments are cited after
Hull (1982).
WHAT EVOLUTIONARY EPISTEMOLOGY IS NOT

Gerhard Vollmer
Justus-Liebig-Universitat
Giessen
1.
CONVERGENCE BETWEEN KONRAD LORENZ AND KARL POPPER?
"Life is a cognltlve process", says Konrad Lorenz, one of

the founders of EE. And "from the amoeba to Einstein, the
growth of knowledge is always the same", claims Karl

Popper, another representative of EE.
Judging from those
and many
similar remarks,
about the same
thing.
In
Lorenz
fact,
and Popper
they
opportunities to stress the convergence
have
of
seem to talk
taken
their
many
views on
the evolution of knowledge (e.g. Popper & Lorenz, 1985;

Lorenz & Kreuzer, 1981).
I will try to show that they not. For, whereas Lorenz
is talking about the evolution of cognitive systems in
general and of our cognitive abilities in particular, Popper

is interested in the evolution of scientific knowledge.
Whereas the evolution of cognitive systems is a problem of

biology and epistemology, the evolution of knowledge is a
problem of history and philosophy of science.
In the
first case, the concept of evolution is quite specific
(essentially Darwinian) and clearly biological, in the
second case, it is quite general and partly metaphorical.
To use the term 'evolutionary
epistemology' for both
approaches is more misleading than helpful.
Before going into any detail, we should face an
obvious objection. Whatever the differences between Lorenz'
and Popper's ideas might be (thus runs the objection), is it
not fine to find similar results, to come to convergent
views, to discover common structures or common laws? Is not

the detection of isomorphisms an essential element of
scientific progress? Is it not enlightening to realize that
a swinging pendulum and an oscillating electric circuit obey
the same law?
In fact, is it not vital to interdisciplines
such as cybernetics, bionics, or synergetics, that they

investigate the common structures in organisms and machines,
or
in
various
complex
consists in establishing
systems?
And
isomorphisms,
if
understanding
have not Lorenz and
203
W Cal!ehaur and R. Pmxren (eds.). EvolutIOnary epistemologv. 203-221
1987 hy D. Reidel PubfishinKCompany
204
G. VOLLMER
Popper contributed extremely to our understanding of several

processes?
Why not give a common name to an
evolutionary
interdiscipline
cultural
that
mediates
evolution?
Why
not
between
call
Names, of course, are arbitrary.

an interdiscipline studying the common
biological
it
EE?
and
Thus, if there were

evolutionary traits
of cognitive abilities and of scientific knowledge, then it

would be perfectly adequate to call it EE.
The trouble is
that such an interdiscipline does not exist.
It might have
started with Toulmin's work on the evolutionary development
of natural science (Toulmin,
1961, 1967,1972). It could
also have been initiated by Campbell's
seminal contribution
to Schilpp's Popper volume, where Campbell explicitly tries

to establish relations between biological evolution and the
evolution of science (Campbell,
1970, 1974a).
It so
happened that both Lorenz and Popper not only expressed
their complete agreement with Campbell's views but that they
accepted and used the term 'EE' as adequate for what they
had
done for years.
Thus,
writes (my translation):

If
In
his
essay
in 'Behind the mirror', Lorenz
'Evolutionary
Epistemology',
Donald T.
Campbell says" ... the natural
selection paradigm of such knowledge increments
can
be
generalized
activities,
such as
science".
contention,
but
only
agree
with this
it as one of the main
tasks of this book to draw the generalizing
comparison, proposed by Campbell, between the
different mechanisms by which different living
systems
acquire
not
to
other
epistemic
learning, thought and
regard
and
store
the
relevant for them." (Lorenz, 1977)
information
And Popper, in his reply to Campbell's article, stresses

"the almost complete agreement, down even to minute details,
between Campbell's views and my own", and writes:
"If we start from a
critical conunon-sense
realism ...
then we shall take man as one of
the animals, and human knowledge as essentially

almost as fallible as animal knowledge ( ... )
The main task of the theory of human

knowledge is to understand it as continuous with
animal knowledge; and to understand also its

discontinuity - if any - from animal knowledge.
205
In all this there is, I believe, complete

agreement between Campbell and myself." (Popper
in Schilpp, 1974, p.1059-61).
2. EVOLUTIONARY EPISTEMOLOGY AND
EVOLUTIONARY PHILOSOPHY OF SCIENCE: COMMON TRAITS
This consent of both Lorenz and Popper to Campbell's
approach may lead to the impression that, in talking about
EE, both are talking about the same thing.
They are not.
Lorenz,
being
a
biologist
occasionally
turning to
epistemology,
is concerned with the wide spectrum of
cognitive systems, everyday cognition, mesocosmic knowledge,
with perception and experience, with our cognitive apparatus
and its
evolution,
with the processes of cognition.
Popper, on the other hand, being a philosopher occasionally
drawing on biology,
is studying scientific knowledge,
abstract theories, the results of cognitive processes and
the evolution of science.
In order to signalize this
distinction
terminologically,
'evolutionary epistemology'
we
shall
use
the
term
for the first, that is Lorenz',
approach, centered on cognitive processes, and 'evolutionary

philosophy of science' for the second, namely Popper's,
centered on science.
True,
both
parties
like
to
stress
the
analogies
between organic evolution and the evolution of science.
The
trouble is that these analogies, though undoubtedly existent

and interesting, are not as substantial and as far-reaching
as Popper, Toulmin, Campbell and others seem to expect. It
might even be that those thinkers are fully aware of this
fact.
In that case, however, they have been widely
misunderstood.
supporting
They
are
read,
cited
and
criticized as
the idea that the growth of scientific knowledge
is essentially Darwinian.
This is not to say that it was not
legitimate to look
for such analogies.

It is, in fact, one of the most
respectable aims of science to identify common structures in
different systems. Moreover, the cognitive aim of science is
the reduction of redundancies in the description of the real
world.
And the discovery of common structures is, indeed,
part and parcel of such a reduction, hence of science.

Nor do we claim that there are, between organic
evolution and the evolution of science, no analogies at all.
On the contrary, there are quite a few and they are
206
G. VOLLMER
extremely interesting.
In a sense, the search for such
analogies has been quite successful.
There is, indeed, a
substantial area of intersection between the two kinds of EE

dealt with in Lorenz' work and in Popper's. Such common
traits are specified in Table 1.
Both theories
- are theories of knowledge
- center on the growth of knowledge
- observe (and welcome) an overall increase of
information in time
- find substantial continuity between animal
and human knowledge
- stress the evolutionary character of the growth
of knowledge
- try to draw epistemological consequences from this

continuity
- rate problem-finding and problem-solving as

essential to the growth of knowledge
-
refer to
trial and error elimination as a road to
knowledge
- may be subsumed under Campbell's conception of
"variation and selective retention" (but blindness
is another matter)
- point in turn to a evolution of these

evolutionary processes
- stress the tentative (hypothetical) character of

all knowledge, but nevertheless
- use some idea of convergence to truth
- incorporate indeterministic elements
- emphasize the role of creativity
- stress the impredictability ("openness") of the

future, including the future of science
- have methodological consequences, and even
- make suggestions how to increase knowledge

Table
1.
Common
traits
ln
Lorenz'
evolutionary philosophy of science.

we
EE
and
Popper's
In view of such an impressive list of common traits,

are perfectly entitled to pose the heuristic question
whether both theories are not, after all, identical. It is

this question to which we now shall direct our attention.
207
3.
Finding
SIMILARITY AND DIFFERENCE: THEIR ROLE IN SCIENCE

common
structures
in
different
systems
is
an
essential element of science.

But it is not all there is.
The world is complex, and an adequate description of a
It must do
complex world cannot be arbitrarily simple.
justice to the intrinsic complexity of real things. What we
strive for are not Just simple theories, but theories that
are both true and simple.
For cognitive purposes, truth is
even more important than simplicity.
That is, we might be
ready to sacrifice simplicity for truth, but never truth for
simplicity. Elimination of redundancy, though being the aim
of science, must end where redundancy ends. And redundancy
ends with a minimal description, a description which cannot
be shortened or made more economic without losing relevant
information.
There is,
minimal description.
by definition,
no redundancy in a
The insight that a description is minimal, that a

thing does not allow of a simpler description, often amounts
to a scientific discovery (1).
Thus, scientific progress
may be achieved in (at least) two ways opposed to each
other: by the detection that two objects (or structures)
hitherto supposed to be different are in fact identical, and
by the discovery that two seemingly equal things (or
structures) are in fact different.
Examples
of
the
first
kind
of
progress
(identification) are listed in Table 2.
Identifications in science
- morning star is (identical with) evening star
(planet Venus)
- stars are giant spheres of glowing matter
- water is a chemical compound of hydrogen and
oxygen (H 20)
- light is (identical with) a specified section of the
electromagnetic spectrum (Maxwell)
- diamond, graphite and soot are just different
configurations of the same substance, namely
carbon
- combustion is a process of oxidation (Lavoisier)
- heat is (identical with) a kind of energy (Mayer)
- heat in gases is (identical with) mean kinetic
energy of molecules (Bacon, Boltzmann)
G.VOLLMER
208
- mass is (identical with) some form of energy

(E = mc 2 ; Einstein)
- gravitational mass is (identical with) inertial
mass (Newton, Einstein)
- mental states and processes are (identity with)
particular brain states and processes (identical
theory as an answer to the mind-body problem; not
universally accepted).
Table 2. The discovery of unexpected identities as essential
for scientific progress.
But there are also many examples of the second kind
(discrimination).
In
fact,
is
not
'Thou
shalt
distinguish ... '
one of the
most sacred
commands in
scientific methodology?
Are not def.initions, explications,
specifications, distinctions, prominent in the tool-box of
every scientist,
every scientific author, and every science
teacher?
Are not scientific instruments made and
enable more and better distinctions ?
What is more,
resolving power is,
used to
quite generally, a
measure of quality.
This is true for any instrument, be it
a conceptual one like a terminology, a theory or a research
program, or a material one like a telescope, a microscope or
any other instrument of observation.

organismic
abilities
It is equally true for
traits
such as sense
organs, perceptional
and other cogn1t1ve structures.
We might even
roughly determine the evolutionary level of an organism from

its capacity to discriminate between different stimuli. In
both cases
advances.
Thus,
increases
in
resolving
identification
complementary aspects of
and
progress,
power
are
seen
discrimination
both in
as
are
science and in
evolution.
It would be quite one-sided to restrict the
concept of progress to one of them (2). Heuristically, it is
a better strategy to keep the situation symmetric. If with
respect to a special subject the differences have been
stressed long enough it might be worthwhile to look instead
for common traits. And if for a couple of objects analogies
and common traits have been widely discussed,
consider the differences.
it may pay to
209
evolutionary
epistemology
evolutionary
philosophy of science
proponents
Lorenz, Vollmer, Riedl
Popper J Toulmin, Campbell
dealing with
evolution of cognitive
systems and abilities,
cognitive processes
evolution of knowledge
(mainly scientific),
cognitive results
"evolution" under-
organic evolution
cultural evolution
stood as part of
concept of evolution
quite specific
qui te general
correlation to
strong
weak, metaphorical,
organic evolution
relevant time scale
(essential identical)
millions of years
tentative, heuristic
tens of years
pertinent scientifie disciplines
biology (genetics,
theory of evolution,
history of science
and technology
ethology J neuroscience) J
psychology J linguist ies
philosophical
discipline
(in particular)
scope
(cognitive levels)
reference
(Objects of
"selection")
regula t i ve idea
or property
approximated by
relation
ideal (f ic t it ious)
final state
... not reached
due to
epistemology
philosophy of science
("cognitology")
perceptual and experiential (mesocosmie) knowledge
all cognitive systems
(amoeba to men, Martians,
possibly even machines
("theory dynamics")
theoretical (or scientifie knowledge
hypotheses, theories
(let theories die instead
of their adherents)
fitness
truth
adaptation
"convergence" (Bavink)
"ver isimili tudell(Popper)
"partial truth ll (Bunge)
truth provides fitness
nei ther fitness nor

adaptation warrant truth
many optimized, mutually
exclusive, but co-existing
species
ever changing environment,
mutations
one single consistent

all-embracing (true)
theory
complexity of the world
tangled hierarchies,
weak causality, chance
events, non-linearity),
limitations to knowledge
G.VOLLMER
210
never regained
non-Darwinian
(al1-or-none decisions)
(forgotten theory)
may be formulated anew
processes
unconscious
opportunistic
conscious
critical
variations
aimed, not blind
information
aimless, blind
playful
copying errors
by genetic inheritance
to own descendants
progress
is an inevitable, but
evolutionary behavlor essentially

lost information
induced by
transmission of
Darwinian (tolerant, allowing many ecological niches)

(extinguished species)
unintended by-product
of evolutionary processes
systematic
problems
by publication
to all fellow scientists

is intended. at least
hoped for. but may be
missed and cannot be
proved
innovations
rate of change
constraints on
innovative trials
character of
constraints
learning strategy
mind-body problem
quas i -cont inuous

("gradual"), conservative
(saltations are too risky
"evolutionary"
many (few evolutionary "licenses")
mainly his tor ical
(no six-legged mammals,)
evolution is "reorganization without closing the
workshop" (Osche)
saltatory,
sometimes radical
"revolutionary"
few
mainly logical
(no contradictions, e.g.)
but also epistemological:
formulizabili ty in a
finite, recursive, intersubjective, argumentative
languagej projectability
to our physical periphery
Nature (evolution, phylogeny) never learns from
(some) scientists learn

from their and others I
mistakes, only from

successes ("poststabilized harmony")
monism (ideot i ty theory)
at the hard core of evolutionary epistemology;
dualism (or Popper's three
worlds) not compatible with
evolutionary epistemology
mistakes and may try and

succeed to avoid them
weakly relevant to an
evolutionary philosophy
of science
Table 3. Major differences between EE and evolutionary theory of science.
211
With respect to EE, the common traits of organic

evolution and the evolution of knowledge have been brought
to the fore by many authors.
It is about time
to trace
out the differences between them and to explode some
misunderstandings based on their uncritical identification.
4.
DIFFERENCES BETWEEN EVOLUTIONARY EPISTEMOLOGY AND

EVOLUTIONARY PHILOSOPHY OF SCIENCE
We could fill pages by characterizing the two kinds of EE.

This would amount to a compaLative
presentation of both
theories and make up a volume of its own.
Let us
use a
more economic alternative.
Let us
content
ourselves
with
relevant differences. They are collected in Table

(3).
These differences show that there exist, in fact, two
disciplines
called
'evolutionary
epistemology',
not one.
They are, despite much overlap, quite independent of each

other.
EE (in Lorenz' sense) could flourish even if
philosophy of science did not exist at all (whereas it
could not dispense with the theory of organic evolution).
And an evolutionary philosophy of science could exist
without even mentioning a biologically oriented EE.
Stressing,
as we do,
all those differences, does not
amount to the claim that the two kinds of EE

common.
had nothing in
This point was already made in section 2
and need
not be repeated.
Our main point is that there are such
decisive differences and that the term 'EE' being used for
both approaches, may obscure them.
What is more, the
ambiguity
of our
term has
in fact
led
to grave
misunderstandings and,
inevitably, to unnecessary attacks
directed against EE
as
a whole.
The
following sections,
then, are meant to counter these attacks by clarifying
of the real claims of EE.

5.
THE EVOLUTION OF SCIENCE
some
IS IT DARWINIAN?
We may even be tempted to ask a more fundamental question
Is the
reveals
with.
growth of science evolutionary?

a basic difficulty historians are
That
there is
change
in science
This question
quite familiar
(and in history)
seems to be evident. But what is meant by 'evolution'?

Is
it nothing but change, or must such change be of a special
kind? How do evolutionary changes differ from revolutionary
G.VOLLMER
212
ones?
Is evolution characterized by its (low)
pace, by
the gradualness of its steps, by its continuity? Are the
concepts of evolution and of revolution mutually exclusive,
or. do they overlap?
Perhaps, revolution is nothing but
rapid evolution, hence evolution after all?
Are there revolutions in science?
And what is normal
in science : stasis, or evolution, or even revolution? Does
Kuhn's distinction of normal science
(dominated by a
paradigm) as against revolutionary science (characterized by
a paradigm shift) really make sense? (Kuhn, 1962; Toulmin,
1970)
We shall not try to answer these questions here. The
reason is that EE and evolutionary philosophy of science
do not differ in this respect.
Whatever their differences:
both of them view the growth of science as evolutionary
(as was indicated in Table 1).
However, the concept of
evolution used here is quite general.
It is, in fact,
universal: all real systems evolve.
But by no means do
they evolve according to the same laws.
In particular,
they need not obey Darwinian principles.
And this leads
back to our main question: is the evolution of science
Darwinian?
Again,
we
are
confronted
with
an
intricate
terminological problem: when is an evolutionary process
Darwinian, and when is a theory describing it a Darwinian
theory?
Are Darwin's personal preferences relevant? Are
they decisive?
But Darwin of all people believed (herein
following Lamarck)
that acquired
characters could be
inherited! Was even Darwin not a Darwinian after all? Which
principles are constitutive for a Darwinian theory of
organic evolution,
and which are peripheral?
Is the
principle of natural selection sufficient to characterize
Darwinism, or is it just a necessary ingredient amongst
others?
How much may we add to Darwin's theory without
distorting it, and, even more exciting, how much of it may
we sacrifice without destroying it?
And again, we shall not try to answer all these
questions. In particular, we shall not give a complete
characterization
of
Darwinism.
We
shall
rather content
ourselves with exhibiting some of the most prominent traits

of organic evolution, namely replication, heritability of
genetic
information,
variation by mutation, and gene
recombination,
and,
last
not
least,
differential
reproduction due to varying fitness (usually called 'natural
213
WHAT EVOLUTIONARY EPISTEMOWGY IS NOT
selection'),
These
science
despite
traits
are
essential
and
necessary
ingredients of Darwinism: drop one of them and you will have

no evolution at all.
Now, EE, viewing the evolution of cognitive systems
as part and parcel of organic evolution, is indisputably
based on the theory of the latter. Thus, the principles
specified above are intricately woven into the descriptive
and explanatory, explicative and argumentative fabric of EE.
These evolutionary traits are, however,
not found in the
evolution of scientific knowledge.
The
evolution
of
is,
all
similarities,
parallels, not Darwinian.

True, we might generalize
specific
conceptions,
to
these
transmission,
analogies
features
and
to less
variation,
and
selective retention, for instance, or to mere selection.

These concepts may tentatively and even successfully be used
to describe the
evolution of scientific theories.
But are
they
Darwinian?
Being
generalizations of Darwinian
principles, they cannot contradict the latter. But being so
general, they are, at the same time, much too weak to
account for organic evolution.
This has been shown by
several authors
criticizing
Darwinian
models of the
evolution of science (4).
It turns out that the common ground
between a
Darwinian-type theory on the evolution of cognitive systems
(EE in Lorenz' sense) and an evolutionary but non-Darwinian,
philosophy of science (in Popper's sense)
is
quite
limited.
In other words, the greatest common denominator
of Lorenz' and Popper's approaches is, though existing,
forbiddingly small.
Thus, the term 'evolutionary epistemology', as it is
used now,
is ambiguous and even equivocal, standing for two
different enterprises. We might leave it at that, keeping in

mind the ambiguity of our term. We might as well restrict
its meaning to that small area of intersection characterized
by some quite general concepts such as 'transmission',
'variation'
and 'selection'.
This, however, would do
justice neither to Lorenz nor to Popper.
And finally, we
could, for the sake of clarity, choose our terms to
designate just one of those two approaches (but not the
other) .
My favorite choice has been made in 1973 and is worked
into Table
3:
I propose
to call
Lorenz' approach
'evolutionary epistemology' and Popper's otherwise (5). But
this is, of course, a terminological decision, hence partly
G.VOLLMER
214
a matter of taste. Much more important is the fact that the

two approaches are, indeed, distinct and even more distinct
than both their adherents and their critics seem to realize.
Therefore we should always make clear what we are talking
about.
Throughout
the
following
sections,
the
term
'evolutionary epistemology'
is meant
to
signify the
evolutionary
systems.
and
EE
even
Darwinian
approach
in that sense holds that
system and our brain are the products of
to
cognitive
our central nervous

organic evolution,
of adaptive and selective processes, and

tries to
investigate
the
epistemological
and
anthropological
consequences of this thesis.
6.
DOES EVOLUTIONARY EPISTEMOLOGY USE

AN EVOLUTIONARY CRITERION OF TRUTH?
This is not to ask whether our conception of truth evolves

(it does!) or whether our assignments of truth and falsity
are changing in time (they are!).
Here we are rather
concerned with the problem how to define and to recognize
truth.
In principle, it would be possible to use a
pragmatist concept of truth, in particular, to identify
evolutionary success with truth or to use it as a necessary
and sufficient criterion for truth.
But this would be a
dubious step, and to my knowledge no adherent of EE has
consciously taken
it.
EE
rather
subscribes
to a
correspondence theory of truth.
Such a conception cannot be justified on evolutiona:y
grounds.
The final criterion for evolutionary success 1S
fitness, not truth, and, relative to a particular ecological
niche, fitness may be provided by quite limited or even
deceptive cognitive means.

Whether
ideas', its genetically determined
an organism's 'innate
cognitive structures,
are correct, whether they are adequate not only to cope with
reality but to build a (partially isomorphic) internal
reconstruction of the world, is not to be judged from its
mere survival.
There is a much
better criterion for
the adequacy of
mesocosmic knowledge, namely scientific knowledge.

The
latter is not crucially dependent on our biological make-up.
Of course, even our scientific knowledge is never final,

perfect, or absolute.
On that account, our appraisal of
biologically conditioned knowledge cannot be final either.
WHAT EVOLUTIONARY EPISTEMOWGY IS NOT
215
If, however, it makes sense at all to talk about truth in

science, at least as a regulative idea, then we are equally
entitled to talk about the truth of mesocosmic knowledge.
This shows that the term 'evolutionary epistemology'
does not imply that all epistemological problems were
solved, or hoped to have been solved, by recourse to organic
evolution.
This is neither claimed by proponents of EE nor
should it be demanded by its critics.
Does transcendental
philosophy, after all, solve all epistemolog~cal problems
in a
transcendental manner?
Do
phenomenology
or
hermeneutics answer all questions phenomenologically or
hermeneutically? Certainly not.
What distinguishes EE is its reference to the facts
and laws of organic evolution, and this unusual trait fully
justifies the epithet 'evolutionary', even if there
are
many problems left to non-evolutionary considerations.
The internal reconstruction of an outside object or

structure may be correct or incorrect.
It is evident that
its correctness enhances the fitness of an organism. And
under conditions of intra- or interspecific competition,
better knowledge about the outside world gives higher
survival value.
This is true as long as the costs for
better knowledge are not too high.
Under competition,
however (that is, in the 'struggle for life'), cost-benefit
relations cannot be ignored.
Different ecological niches
lead to different
cognitive needs and advantages and
therefore to different cognitive systems.
Thus, EE explains, first, the existence of cognitive
systems (by the survival value of cognition), second, the
multitude of cognitive systems (by reference
to the
diversity
of ecological niches),
third,
the partial
correctness
or adequacy
of
cognitive
structures
correctness
of their internal reconstructions?
(by its
fitness-enhancing effect), and fourth, their partial failure

(by introducing cost-benefit considerations).
If we feel indeed entitled to explain the existence of
correct world views by their survival value, may we not take
the survival of cognitive systems as a
clue to the
Of course,
nobody would consider this kind of argument as being a valid
EE, to repeat it,

success a criterion
does not make survival or evolutionary

for truth.
In fact, it is neither
necessary nor sufficient.

But between a deductively valid
inference and a mere logical fallacy, there is a whole
And contained in this
spectrum of types of arguments.
spectrum is the fact that the truth of a conclusion may,
G.VOLLMER
216
under specified conditions,

support the truth of the
premise.
This kind of 'corroborative' argument is used in
science when
we
trust a
theory
which
has
been severely
tested and has survived the test unscathed. And the same
kind of argument is used by EE when evolutionary success
under competition is used as an indicator for truth.
7.
IS THERE A VICIOUS CIRCLE

IN EVOLUTIONARY EPISTEMOLOGY?
Occasionally, EE is blamed for arguing in a circular manner.

We cannot
Such objections come in quite various forms.
discuss all of them here (Vollmer,
1983). One particular
charge, however, is bound to arise
when adherents or
critics confound EE with an evolutionary philosophy of
science, and this charge should therefore be treated in our
context. So let us try to formulate and refute it.
EE is based on science,
that is on theories about the
physical structure of the world in general, on the theory

of organic evolution in particular.
On the other side,
EE claims to be epistemologically relevant. It informs us
about the reliability of our cognitive
apparatus, it
shows why intuition or visualizability may not be used as
criteria of truth, it explains the fit or match between our
cognitive structures and the outside
world,
it
talks
about
the
possibility,
but
non-demonstrability,
of
objective knowledge, etc.
How can that be? Is not the
theory of knowledge prior to knowledge, epistemology prior
to episteme?
Would it not be a vicious circle to use
scientific knowledge when trying to support or to criticize
science itself?
There
is,
indeed,
some
circularity
in
the
argumentative structure of epistemology.
This is, however,
not a vicious circle, but a virtuous one.
What we find, is,
more precisely, a self-correcting feedback loop between our

knowledge and our theory of knowledge.
We start with
everyday knowledge, then pose and answer the first naive
questions about knowledge, thereby correcting and improving
it; we then turn back to reflect our improved knowledge,
thereby initiating epistemology, and back again to criticize
and to improve science and scientific knowledge, and so on.
In this virtuous circle every side is both influencing, and
dependent on, the other.
217
This
characterization
of
the
relation between
knowledge and theories of knowledge is adequate under
historical
as
well
as
under
systematic
aspects.
Epistemological efforts have developed in a virtuous circle
with
science,
and epistemological
arguments
may be
reconstructed
in
like
manner.
As
Einstein
says:
"Epistemology without contact with science becomes an empty

Science without epistemology is - insofar as it is
scheme.
thinkable at all - primitive and muddled." (Schilpp, 1949)
EE fits this description.
It takes the existence of
knowledge as an empirical fact and tries to solve some
problems concerning human knowledge.
In doing this, it
makes ample use of the theory of evolution applied to
cognitive systems.
Its primary field of application is,
however,
intuition
mesocosmic
knowledge,
and
pre-scientific
that
is
experience,
perception,
household
communication and naive logic.

Scientific knowledge is,
though relevant, not a genuine subject of EE. Therefore, EE
will
neither
support
nor
refute
scientific
theories,
certainly not the theory of evolution (which is presupposed

and used). Taking this restriction seriously, we may answer
another frequently heard objection.
8.
WHAT, IF ANYTHING, DOES EVOLUTIONARY EPISTEMOLOGY

CONTRIBUTE TO PHILOSOPHY OF SCIENCE?
EE claims to uncover the biological, genetic, evolutionary

roots of human knowledge.
If it
lives up to this task,
should it not specify the biological
roots of scientific
theories such as group theory, the theory of relativity,
quantum chromodynamics, cosmology, molecular biology, the
theory of evolution, or even EE?
The answer to this question is a simple no. EE is not
able and, to my knowledge, has never promised to unearth
the biological roots of particular scientific theories.
Of course, nobody would mind finding them if there were
any.
But do they EE is relevant and applies to mesocosmic,
not to scientific knowledge.
It explains the
biological
roots
of
our
topological,
metrical,
temporal,
informational,
inferential, statistical, causal intuitions,
and
not
the
theories of physics
which
explicitly
contradict
those
intuitions.
Inconsistencies
between
intuition and science suggest that such roots
don't even
exist:
If
there are,
indeed, biological roots
for
G.VOLLMER
218
mesocosmic conceptions, how could there be, at the same

time,
roots
for
theories
which
are
in
flagrant
contradiction to them?
And if such roots do not exist, why
ask EE to uncover them?
No doubt, this mistaken request directed at EE arises
from the equivocal and therefore misleading use of the term
'evolutionary epistemology'.
If
we confound questions
concerning the origin and structure,
reliability and
limits of our cognitive apparatus with
questions on the
evolution of scientific knowledge, such misunderstandings
are bound to arise again and again. They could be avoided
by restricting the term 'evolutionary epistemology' to the
first type of problems.
That there are no biological roots for particular
scientific theories does not imply that there are none for
science
in
abstraction
general.
and
Memory
and
generalization,
learning,
creation
curiosity,
and
use
of
concepts, formation of hypotheses, communicative needs, the

use of a descriptive and argumentative language, a critical
attitude, and the urge for intersubjective consent - all

these are, indeed, typically human traits, biologically
rooted
and,
at the same time,
constitutive
for
science.
There is a wide field to be searched into by evolutionary

neuroscience,
evolutionary psychology
and EE.
There
is
still
another
respect
in
which
EE does
effectively contribute to philosophy of science. Although it

cannot positively exhibit the roots of the latest scientific
theories (which may be nonexistent), it may explain some
interesting negative features in the history of science.
Most scientific disciplines have started with hypotheses and
theories which are nowadays considered as false, but which
were quite convincing to their creators and their fellow
thinkers.
These early conceptions were accepted because
they were, their falsity notwithstanding, quite successful
in accounting for a characteristically restricted class of
observations. Aristotle's theory of motion, for example, but
also geocentric astronomy,
the impetus theory of the
pre-Newtonian era, or the theories of four (or more)
elements are quite adequate for the world
of medium
dimensions.
early
As has been found in psychological experiments, such

conceptions correspond much more to our intuitive
grasp of
objects,
events,
processes, frames of reference,
probabilities,
etc.,
than our more recent scientific
theories could ever hope or claim to (Tversky & Kahnemann,
219
1981; Kahnemann & Tversky, 1982; Clement, 1982; Mc Closkey,

1983; Vollmer, 1985). They are, as a rule, not mere
inventions, inspired by fantasy and held dogmatically. They
are rather applicable to the world of medium dimensions,
truly mesocosmic.
This explains why they were built,
accepted and firmly kept, even against disproving evidences.
Thus, EE explains, if not the successes of science,
at least some of its failures. This is reason enough to
stress its relevance to the history and philosophy of
science.
NOTES
1. The minimality of a description can never be proven
definitely, as Gregory Chaitin has shown, using arguments of
Kurt Gadel and Alan Turing (cf. Chaitin, 1975).
2. Thus, if Riedl and Kaspar endow all organisms with
a ratiomorphous cognitive principle called "hypothesis of
identification" and characterized as "the expectation that
dissimilarities may be ignored in similarities, and that
similarities will prove similar even in what has not been
perceived" (see Riedl, 1984b; Kaspar, 1984), then this
assertion
must
be
supplemented
by
a corresponding
"hypothesis
of discrimination",
characterized
as the
expectation that objects known to be dissimilar will prove
dissimilar in even more aspects.
3. The editors have suggested that table 3 could and

should be complemented by a row for Piaget's genetic
epistemology,
investigating
ontogenetic instead
of
phylogenetic and
historical processes.
This would be
possible and heuristically fruitful, and I have used such
surveys
in other
contexts focussing,
for instance, on the
chances of improving our knowledge or the risk

it. Here, however, I am primarily concerned
distinction between the two kinds of EE running
same label and causing much confusion if not
distinguished and kept apart.
of losing
with the
under the
carefully
It is true, however, that Piaget's ambiguous term

"genetic
epistemology"
may
lead
to
a
similar
misunderstanding.
For "genetic" may be interpreted as
deriving
from
either
IIgenesis"
or
"gene". Genetic
epistemology would then be a discipline investigating either

how cognition and knowledge arise or how they are influenced
220
G.VOLLMER
or even determined by genes~ that is genetically in the

biological sense. And the r~se of knowledge has itself
phylogenetic, historical and ontogenetic aspects. So, which
aspect is signified by "genetic epistemology"?
It is
known that,
personally, Jean Piaget was
interested in all of these different perspectives. This is
perfectly documented by the scope and content of his
scientific work, for instance by his book Biology and
knowledge (1971). It is also borne out by the following
remark:
"Our problem may be formulated as follows: By
which means does the human mind pass over from a
state of less satisfying knowledge to a state of
higher knowledge? ( ... )
The most fruitful and most natural field
of investigation would be,
of course, the
reconstruction of human history - the history of
human thinking from prehistoric man onwards. But
unfortunately we don't know very much about the
psychology of Neanderthal man or of Teilhard de
Chardin's Homo siniensis. As this dimension of
biogenesis is
not accessible
to
us,
we shall
have
to
turn,
as
biologists
had,
to
ontogenesis." (Piaget, 1972).
From this, it is evident that Piaget would have welcomed an
evolutionary account of human cognition as aspired by EE.
But it is also evident and, besides,
well-known that
Piaget's work is ontogenetically oriented and
that his
'genetic' epistemology refers to the genesis
of knowledge
in the individual. So much so that Hans "G. Furth, one of
his disciples and interpreters, in the preface to his book
on Piaget, makes the following telling remark:
"I am unable to see how one can feel at home
with his Piaget's model of intelligence unless
one sees intelligence as a prolongation of
organic development. Without a biological basis,
Piaget's formal logical model becomes what to
many
it
unfortunately
appears:
a
cold,
artificial system of ratiocination that has no
relevance to full-blooded real life." (Furth,
1969, p. XVII, preface).
221
Thus, although genetic epistemology and EE have much in

common (which could be the subject of another paper), they
are clearly distinguished and not in danger of being
confounded. They have, after all and fortunately, different
names.
4. A most comprehensive critique of an evolutionary

philosophy of science is given by
Thagard (1980), p. 187,
who convincingly argues that "the similarities between
biological and scientific development are superficial, and
that clear examination of the history of science shows the
need
for
a
non-Darwinian
approach
to
historical
epistemology". And in a note to this remark, he adds (p.
193): "My critique of EE is not concerned with the claim
that human biology may be relevant to epistemology in more
direct
ways,
for
example
in
debates
concerning
innate
ideas ( ... ) Nor do I address the 'genetic epistemology' of

Piaget ( ... )"
Another critique is found in
L.J. Cohen (1973). He
comes to the conclusion
that
Toulmin's "evolutionary model
for the history of intellectual disciplines is not entitled

to be put forward under an aura of Darwinian respectability"
(p.41).
Here again, it is not the evolutionary type of
development which is denied to science (as the title seems
to suggest), but rather the specifically Darwinian character
of its evolution. And again it is not EE in Lorenz' sense at
all which is criticized by Cohen (see also Losee, 1977).
5.
According
to
excellent contributions of
this
terminological
Louis Boon,
and Jean Paul Van Bendegem to

wi th EE "proper".
choice,
the
Karin Knorr Cetina,
this volume are
not dealing
THE PHILOSOPHICAL SIGNIFICANCE

OF AN EVOLUTIONARY EPISTEMOLOGY
Andrew J. Clark
University of Sussex
1. INTRODUCTION
What kind of an Epistemology is an EE?
In what sense, if
any,
can such an epistemology
provide
'a dialectic
resolution to many old controversies' (Campbell, 1974a)? To
answer such questions is to begin the somewhat neglected
task of placing evolutionary epistemology in its proper
philosophical niche. The task is an important one since, as
we shall see, much recent criticism and distrust of the new
epistemology is bred by confusion
over
its intended
significance.
The naturalised approach, I shall argue,
should in no way be promoted as a means of resolving old
controversies.
For
most such controversies
are bracketed
the moment we take the biological organism as the knowing

subject
of
epistemological
concern.
The
legitimate
philosophical interest of the discipline of EE accrues
rather, I claim, to the range of new epistemological
questions
it poses.
Some of these questions
are reviewed
towards the end of the paper.

The discussion is in two
parts.
Part I presents an impressionistic account of the
traditional epistemological
enterprise
and
shows why
evolutionary conjecture is essentially maladapted to such an
environment.
Part II attempts to place the new ideas in a
more suitable,
non-foundationalist
fund of philosophical
the new setting.
niche and
questions which may be
sketches the
addressed in
2. THE TRADITIONAL EPISTEMOLOGICAL ENTERPRISE

Epistemology, in the historical philosophical tradition,
is prompted by the challenge of Cartesian scepticism. It
therefore seeks to explain how knowledge of the external
world is possible in the face of Descartes' conjecture that
the
experiential
data-base
of
our
knowledge might
conceivably be disconnected from the real world in which we
are located.
Such disconnection. as Descartes pointed out,
223
W Callebautand R. PinXlen reds.), Evolutionary Epistemology, 223-231.
A.J.CLARK
224
is already familiar to use from the daily phenomenon of

dreaming (Descartes, ed. Haldane, 1955). The problem of
traditional epistemology is how
to guarantee
that a
similar disconnection does not afflict all
our experiences,
waking and sleeping alike. How, that is, can we know we are
not being deceived by some evil demon or, to adopt a more
modern formulation, how do we know we are not just 'brains
in a vat' (Putnam, 1981) in the laboratory of some alien
scientist?
Such questions go to make up what Barry Stroud
has called the philosophical problem of our knowledge of
the external world (Stroud, 1984). Standard epistemology,
as Richard Rorty has insisted at length (Rorty, 1980), thus
grew up in part as an attempt to resolve this philosophical
problem by formulating a theory of knowledge which secured
human wisdom against the sceptical challenge.
In this
environment
the rationalists
maintained
that reason alone
could disclose the unshakable foundations of knowledge in

the form of propositions whose structure guaranteed their
truth.
A prime example of such a response is Leibniz'
claim that the predicate is included in the subject of
every true proposition.
The empiricists (and later the
positivists) also sought to uncover the hidden foundations
of knowledge, but they focussed on (allegedly) indubitable
sense-experiences.
Hume's
'simple
impression'
and
Berkeley's 'immediate perceptions' could both plausibly be

included as examples of this kind of response (Hamlyn;
1970).
More complex positions such as
Kant's empirical
realism and transcendental idealism also embody arguments
against philosophical scepticism.
Generalising over these
examples we may say that epistemology, in the classic
tradition, is primarily concerned with the justification of
human knowledge against the threat of extreme sceptical
doubt.
Such justification, as Rorty (1980) neatly analyses
it, proceeds as a rule by the isolation of a base class of
logically privileged knowledge (knowledge somehow immune to
sceptical doubt) which gets pressed into service as a
foundation for the grand edifice of human knowledge itself.
It is no wonder,
then, that an EE should be incapable
of satisfying such traditional epistemological desires. For

such an epistemology begins well after the traditional
enterprise is meant to have ended.
It begins by taking the
biological organism, a creature in the world, as its
epistemological subject. But to assume, at the outset, that
we know
ourselves aright
as biological organism,
is, from
THE PHIWSOPHICAL SIGNIF1CANCE
225
the traditional point of view, to assume far too much. For

what is the biological organism if not a part of the
external,
objective reality
knowledge
of
which is
supposedly in
question?
There is nothing virtuous, then,
in any explanatory circles generated by the attempt to use
evolutionary conjecture as a response to the traditional
philosophical problem of our knowledge of the external
world. As Manley Thompson has put it:
"The (scepti<;al) challenge in its full force
calls into question any claim that we have
contact with reality and are therefore able to
represent something besides items in our own
consciousness." (Thompson, 1983).
To represent our epistemic situation as 'simply that of a
human biological organism vis-a-vis its environment' (ibid.
p.35) is thus to beg to question against the philosophical
sceptic.
An EE (1) which explains the possibility of
cognitive fit (the fit of our ideas to the world) by using
the apparatus of biological evolution brackets the sceptical
justificatory demand at the very start. If further proof of
this is necessary we need only pause to reflect that any
sufficiently devious Cartesian demon might well scheme to
have us formulate and believe evolutionary
hypotheses
thereby diverting us from the reality of our plight.
In the light of these observations Quine's recent
assertion (2)
that the new 'naturalised epistemology'
constitutes 'an enlightened persistence ... in the original
epistemological question' can be seen to be dangerously
misleading.
For whatever the new discipline may achieve it
necessarily fails even to address the original philosophical
problem.
Campbell's claim, reported earlier, that EE may
provide a dialectic resolution to
old
controversies is
similarly misleading.
For a dialectical argument, in its
Aristotelian sense, is one which is
designed to produce
conviction in the opponent, even though it falls short of
providing watertight proof.
But,
as we saw, if anyone
is at all persuaded by the arguments of
the traditional
sceptic, no change in his
opinions is
likely to come
about from reflection on the evolutionary picture. I do not
wish to dwell too heavily on Campbell's one unfortunate use
of words, the more so since, in later works,
he shows
himself quite
clearly
aware of
the irrelevance
of
evolutionary
epistemology
to
traditional,
sceptically-motivated enquiries (3).
Nevertheless much
A.J.CLARK
226
contemporary opposition
precisely
encourage.
the
kind
of
to EE can be seen
misconception
to
which
spring from
such remarks
Thus critics such as Richard Rorty and Hilary Putnam

seem to believe that the goal of naturalised epistemology is
indeed to solve the classical sceptically inspired problem.
Rorty objects to evolutionary epistemology on the grounds
that it tries fruitlessly to justify 'our criteria of
successful enquiry' and to show why they are 'the right
criteria, nature's criteria, the criteria which will lead us
to the truth'.
The evolutionary epistemologist, Rorty
believes, is a1m1ng to 'de-transcendentalise epistemology
while nevertheless making it do what we had always hoped it
might' (Rorty,1980
p. 299). And what we always hoped it
might do was to justify our knowledge in some independent,
objective way. Yet the naturalised claims, as Rorty rightly
points out, are themselves as dependent on sUbjective social
justification as any other claims; a causal theory of

representation or an evolutionary account of knowledge are
just further parts of our current, socially agreed theory of
the world and hence unable to provide the objective foothold
on knowledge we (allegedly) hoped they might (Rorty, 1980 p.
295).
Putnam likewise mislocates the goal of the new
epistemology
as being the objective
justification of
knowledge. Thus he objects that any naturalised argument for
cognitive fit must be viciously circular since it must rely
on the very things it is meant to justify e.g. our notions
of causality and of observational data (Putnam, 1983).
To
deflect
the
force of
such criticisms the
evolutionary epistemologist
may
rightly
insist that,
contrary to his opponent's beliefs, he is not in the
business of justifying our knowledge in any absolute
foundationalist sense. Such a response is no doubt correct,
but as it stands inadequate.
For it remains to show what
(non-foundationalist)
business
the
evolutionary
epistemologist is engaged in, and to convince the doubters
that the business is epistemological in at least some
philosophically interesting sense. To these matters we now
turn.
3. PHILOSOPHICAL QUESTIONS
What
kind of
foundationalist
Callebaut
and
epistemological
work remains
once the
project
is
abandoned
or
bracketed?
Pinxten
claim that "EE can
offer a
227
naturalistic theory of justification in the

sense of
non-epistemic and of epistemic yet non-foundationalist
appraisal" and offer as a corollary of this claim the
thought that "The justification of induction pre-supposes
that induction is justified; but this is a non-vicious form
of circularity" (Callebaut and Pinxten, 1984).
Such a claim is reminiscent of Quine's frequent
reassurances that the circularity inevitably involved in the
evolutionary 'justification' of inductively-based knowledge
is not to be feared. It is to be feared, he thinks, only if
we accept that all valid epistemological conjecture must be
geared to the repudiation of a global scepticism. But far
from believing this Quine holds that there can be no
'external
vantage
point'
or 'first philosophy'.
He
therefore holds the traditional enquiry to be misconceived
and focusses on a reformulated 'problem of knowledge',
namely the problem of how our knowledge (granted that we
have
some)
could
have
come
about;
i.e.,
in
Quine's
terminology, how it could be that 'our quality space matches

that of the cosmos' (Quine, 1969).
The
epistemological
question
addressed
by
a
non-foundationalist theory of knowledge of the kind provided
by Quine or by an EE is thus not of the form 'Is knowledge
possible?' Rather, it is an instance of a second type of
epistemological
concern
which
Daniel
Dennett
has
characterised as having the general form of the question
'Given that knowledge is possible, how is it possible?'
(Dennett, 1978). This second type of question is, however,
epistemological only in a somewhat indirect (but nonetheless
important) sense. Some philosophers (particularly
Rorty)
believe that to give up the foundationalist project is to
give up epistemology itself as a philosophical concern.
To
ask
the
second,
hypothetically
formulated,
'epistemological' question is not, they will claim, to ask a
philosophical question at all.
Rather it is to ask a
scientific question about the origins of our knowledge; a
question which, they feel, stands in need only of a
scientific answer.
There is some truth

in such
objections.
Most
philosophers (myself included) do believe that they work in
a domain
which,
if
not discontinuous
with science, is at
least recognisably different to science in its goals and

procedures. The question 'Given that knowledge is possible,
how is it possible?' is seen by them to be a scientific
question
and
hence an improper
subject of (direct)
228
A.J.CLARK
philosophical concern.
For philosophers expect their best
arguments and conjectures to be in some sense independent of
particular experimental results or empirical considerations.
But this is not to say that those very arguments and

conjectures may not take as their (falsifiable) base a
situation (e.g.
the state of the brain as a product of
biological
evolution)
which was
itself
revealed by
scientific discovery.
The discipline of philosophy on this
model may deal in conceptual disputes and conjectures which
arise out of the examination of issues and possibilities
drawn from hard science.
It is for this reason that I
believe
the
resistance of
mainstream
philosophy to
evolutionary
epistemology to
be
misplaced.
For an
evolutionary answer to the second question (concerning how
knowledge is possible), though perhaps not of philosophical
interest in itself, can act as an effective springboard for
genuine epistemological discussion.
It cannot, of course,
act as a springboard for any discussion or resolution of the
issue of (global)
scepticism.
But
it
can suggest
conjectures and questions appropriate to the discussion of a
different range of properly epistemological issues, namely
those relating to what D.W.
Hamlyn (1970, p.6) calls the
question of the scope of knowledge. Questions falling into
this latter category have in the past included the question
'whether whole ranges of forms of knowledge are possible and
if so how?' (e.g. the question of a priori knowledge)
(p.6-7), the question whether 'knowledge of reality is
always mediated' (p.124), the (related) question of whether
perception provides 'direct knowledge of physical objects'
(p.147) and the question, addressed by both Kant and
Wittgenstein, whether 'we might have developed a conceptual
structure different from that which we have developed in
fact' (p.72).
I do not, however, wish to claim that EE provides an
answer even to these questions (though it may do so,
especially with regard to the issue of a priori knowledge).
Instead I wish to suggest that its prime philosophical
interest
lies
in
its
providing
basis
of
scientific
conjecture from which essentially new questions concerning

the scope of knowledge can be raised.
Such questions might
include:
(1) The question how far, if at all, an evolutionary
account of mind can be reconciled with a classical mirror or
correspondence theory of knowledge?
Relevant to this issue
would be the reflection that natural
selection is a
229
satisficing, not an opt1m1sing force (4); that it is geared

to the generation of cognitive systems promoting speedy,
approximately valid evaluations of environmental stimuli.

Evolved cognitive systems, on such a model, are likely to
share in the imperfect, approximate and species-specific

character of evolved solutions in general (Tennant, 1983).
(2) The question of cognitive universals.
If, as
evolutionary epistemologists such as Lorenz (1941) claim,
much
of
our thought depends on
an evolved
framework of a
priori concepts, how far can we expect that framework to be

shared by all rational beings? And how far can we properly
even conceive of cognitive frameworks other than our own?
(3) The question, supposing we accept the imperfection

and species-specificity of human cognitive processes, of
whether we can then uphold any standard of objectivity in
science?
Must we go on to allow, with Rescher (1983),
the possibility of non-standard extra-terrestrial science?
And what then of the claim, essential to any evolutionary
epistemology, that science adequately describes the common
reality to which various beings are seen to be variously
adapted?
Can this be preserved in the
face of an
evolutionarily inspired scientific relativism or

relativism a self-undermining project?
is such a
(4)
The question
of
whether
the evolutionary
epistemologist, by picturing us as knowing the world only
indirectly
and
'presumptively'
by means
of evolved
strategies (see Campbell, 1974a, p. 418), must therefore
recapitulate the much-criticised Kantian divide between
Appearances
and
the
world-in-itself.
Is independent
reality, for the evolutionary theorist, necessarily an
indescribable 'something = X' hidden forever behind the veil
of perception and cognition (O'Hear, 1984)?
Taken together, such questions amount to the threat of
a new scepticism.
Far from filling in for Descartes' God
the problem is that evolutionary epistemology may unveil
Cartesian demons in nature herself.
Demons which block out
direct and unbiased contact with reality and which further
deceive
us
with
species-specific
priori
conceptual
frameworks.
The new scepticism thus uses one part of our
knowledge (viz. evolutionary theory) to question the status
of the rest.
Whether such a scepticism can be consistently
maintained, and if not, how the evolutionary arguments are
to be otherwise accommodated, provide further areas for
useful philosophical
discussion.
questions could be further expanded
Indeed our list of

to include issues in
230
A.]. CLARK
normative epistemology and theories of language and meaning.

An evolutionary approach to
But the point is clear enough.
knowledge functions best,
want
to
say,
as
a breeding
ground for new epistemological problems and not as a source

of solutions to old ones.
These new problems concern the
scope and character of human knowledge rather than its
overall certainty or possibility. But they are still proper
matters for philosophical discussion and it is a virtue, not
a vice, of the new epistemology that it should give rise to
them.
I would close, then, by recommending a kind of gestalt
switch.
Under that switch evolutionary epistemology is
perceived as a source of new questions more than as a fund
of answers to traditional ones.
These new questions are
recognisably epistemological, but relate to the issue of the
scope of human knowledge an not to its safety against the
classical sceptical challenge.

succeed in shaking off the
Thus viewed, EE may at last

disreputable mantle
of a
reactionary attempt to revive foundationalist philosophy.
NOTES
1. Here and throughout the paper I use the term 'EE' to
refer
to
what
Donald
Campbell
calls
'biological
Evolutionary Epistemology' and Gerhard Vollmer calls EE (as
opposed to 'EE'). I therefore use it to refer
to the
account of knowledge consequent upon the identification of
human cognitive systems as, in part at least, the products
of actual biological evolution.
I do not use it to refer
to the metaphorical usage ('EE') of a natural selection
model of
theories.
the
generation
and
retention
of
scientific
2. Quine's remark is found in his (1974, p. 3) and is

reported in Stroud (1970, p. 224). Stroud too argues that
Quine's naturalised approach, whatever else it may be,
cannot resolve the classical philosophical problem.
3. Thus Campbell, in one of his most recent papers,
writes that: "When biological evolution is invoked ... for a
philosophical epistemological argument, there has been no
answer to the sceptics" (Campbell, 1984).
4. The word 'satisficing' was coined by H. Simon to
describe 'methods that look for good
or satisfactory
solutions instead of optimal ones'. Its use in the present
context is meant to signal that the forms of sense and modes
of processing selected will be geared to efficacy rather
231
than detailed veridicality. For efficacy and veridicality

(or, if you like, truth) diverge as soon as the parameter of
cost-efficiency is introduced into the equation. For a
discussion of the notion of satisficing strategies see
Simon, (1969), p. 64.
HOMO SAPIENS, HOMO FABER, HOMO SOCIANS:

TECHNOLOGY AND THE SOCIAL ANIMAL
Linnda R. Caporael
Department of Science and Technology Studies
Rensselaer Polytechnic Institute
1. INTRODUCTION
Modern technology is a pinnacle of human progress, the
perfection of reason mirrored in design.
It maintains that
bit of Enlightenment hubris captured in the term Homo
sapiens.
Why,
then, does it seem easier to design
technology than to make decisions about how to use it? That
technologies can puzzle or disappoint or seem out of
control, is explained - by their opponents - as temporary
yet resoluble oversights of rational solutions, or - by
their supporters
as the failure to demonstrate to all
concerned the reasons that the benefits outweigh the costs.
(In spite of the acronym MAD, nuclear armaments debates are
conducted in this vein).
Alternatively, some of our
technologies may puzzle, disappoint, or seem out of control
because the model of human nature is inappropriate. Despite
economic,
hedonic,
and
nonmative
justifications,
the
technology or its implementation somehow doesn't 'make

sense' or 'feel right'.
This has not always been the case.
Until very
recently, the technology invented by humans was naturally
accommodated to the human physical form.
Its requirements
are obvious - no one would design a hand spade that was four
feet wide or chopsticks that came in trios. Technology
connected body and habitat.
Moreover, the consequences of
technologies with undesirable effects was limited to local
populations
and habitats.
Contemporary technology is qualitatively different and
the
dimensions
of
difference
not
clear-out.
Its
requirements are often
far less
obvious because the
technological accommodation is less to form than it is to
mind (e.g., communication technologies). Its consequences,
desirable and undesirable, can have global effects. But most
important,
I think,
is
how technological innovation
intersects with what it means to be a social species.
233
W. Callebaut and R. Pinxten reds.), Evolutionary Epistemology, 233-244.
L. R. CAPORAEL
234
In this paper, I will suggest we are neither Homo

sapiens, nor even Homo faber, but rather Homo socians, and
that technology has been hostage to a sociality that
represents the wisdom of
past selective environments.
Without understanding that sociality, we cannot know if our
wisdom should apply to the current scene or whether we
should construct (rather than be selected for) a new wisdom.
Campbell (this volume) distinguishes between "biological
evolutionary epistemology" and "evolutionary theories of
science". It is his epistemology of the first kind that is
at present most important for considering technology.
In the following section, I shall put aside the issue
of technology in order to consider in some detail a
biosocial theory of general epistemological significance.
Its central proposition is that among human biological
adaptations is sociality, a class of
species-specific
perceptual
and behavioral
mechanisms
sensitive
and
responsive
to
social
stimuli.
Sociality mechanisms
function in the development and maintenance
of group
membership. Two dualisms are
of direct interest.
The
first is that biological human nature is both competitive
and cooperative.
The second is between two kinds
of
rationality
a constraining, (usually)
subcognitive,
nonverbal
social
rationality
and
a
potentiating,
striving-toward-normative,
symbolic
rationality.
The
biosocial theory provides the
grounds for integrating
human evolution into considerations of technology.
2. INTERPERSONAL SELECTION
Much contemporary discussion of human evolution adopts the
sociobiological framework that has been subject to numerous
criticisms (cf.
Brandon & Burian, 1984; Caplan, 1978;
Sober,
1984).
There are three shortcomings that are
especially serious flaws.
First is the presupposition of
sociality in evolutionary reconstructions to explain (and
this is common in non-sociobiological scenarios as well).
For example, Trivers (1971) posits 'mutual dependence' as a
prerequisite for the evolution of reciprocal altruism.
Second, natural selection does not act on analytic
categories such as
'altruism,' 'competition,' 'aggression,'
or 'cooperation.' These complex constructs are more usefully

conceived in the ecological sense as events, that is
dynamic, multimodal changes over time (McArthur & Baron,
1983).
The mechanisms on which selective pressure might
HOMO SAPIENS, HOMO FABER AND HOMO SOCIANS
235
operate would be those involved in the 'interpretation and

organization of responses to properties of social events.
'Mechanisms'
gene:a!ive
is used here in an very broad sense to include
rules,
prototype
formation,
microbehaviors,
cogn1t1ve constraints, affect, and maturational sequences.

The properties of social events include not only the
stimulus properties of interactants, but also higher-levels
principles of social organization (Campbell, 1974).
Third is the selection of altruism as sociobiology's
central problem (Wilson, 1975).
In its restricted sense
(reproductively self-sacrificial behavior) it usually fails
to be meaningfully mapped to human behavioral altruism.
Symons (1980) characterizes Medea thusly: "( ... ) from the
standpoint of reproductive success, a woman who betrays her
father,
sets
up
her
brother's
murder,
and
murders her
children ought to be regarded as a paragon of altruism" (p.

205). In its expanded sense, any time or energy expended on
behalf of a beneficiary is altruistic because it can be used
by
the donor (Barash,
1977,
pp.
77-78), but this
formulation breeds the much criticized adaptationist program
and has little explanatory power in the face of complex
social behavior.
We need to further examine human sociality in its
biological context.
The 'first family'
collection of
australopithecines, three million years old, suggests group
living is an ancient adaptation.
Their descendants are an
extraordinarily social species.
The central problem in
human evolution is the same as for all species, reproduction
and development of offspring to reproductive age. Human
sociality represents a solution to these problems. The
theoretical and empirical issues are to determine the
mechanisms of this solution, how it contributed to the
reproduction of offspring and their survival and development
to reproductive age, and how these mechanisms combine to
achieve the flexibility characteristic of human behavior.
Kin selection, a favored sociobiological process,
certainly allows for protohuman aggregation: proximity is a
necessary condition for sociality and a substitute mechanism
for kin recognition (Holldobler & Michener, 1980; Holmes &
Sherman, 1983). The impetus to aggregation is likely to be
overdetermined, associated with a collection of interrelated
factors.
Protection from predation is one often mentioned
factor; Kurland and Beckerman (1985) provide an extended
argument for information exchange about potential food
resources in foraging; I suggest (Caporael, 1984) increased
236
L. R. CAPORAEL
demands of offspring associated with the beginning evolution

of the lengthy post-natal development.
The consequence of
such a concatenation of factors is that the group would
eventually come to intervene between the individual and the
selection pressures of the habitat and itself become a
selective environment.
Campbell (1983) calls this crucial
human evolutionary process 'selection by functioning social
unit'; I prefer the term 'interpersonal selection' because
it is briefer and avoids confusion with 'group selection'.
Interpersonal selection means that to the extent that group
living conferred a selective advantage over solitary living,
individuals better adapted to life in groups had greater
reproductive success than those not so adapted.
Interpersonal selection placed a premium
on the
evolution of perceptual, affective and cognitive mechanisms
sensitive and responsive to social stimuli.
These are
'low-level'
constructs, some non-controversial examples
being the capacity for individual facial recognition, the

identification of certain emotional states, and imitative
learning.
Tentatively, they are restricted to mental
processes below the level of language (i.e., we are not
instructed on how to recognize biological movement); they
require for their functioning interface with the environment
and are
thus
'open
programs,'
(e.g.,
learning to
communicate), they may take the form of constraints on
learning, they may be quite complex (e.g., stereotyped play
patterns), they may be brought together to form complexes
amenable to analysis at higher orders, and they may be
brought into service in contexts outside their evolved
functional
context.
(My
research on baby
talk to
institutionalized elderly adults (Caporael, 1981; Caporael,
Lukaszweski & Culbertson, 1983) demonstrates these last two
points ).
The distinctively human complex may be viewed as the
outcome of an evolutionary pathway where developments at one
diachronic point may (a) constrain developments at a future
point,
and (b)
provide
the
foundation
for future
developments not only by direct selection, but by coaptation
and exaptation (Gould & Vrba, 1982). Individuals evolving
mechanisms enabling them to maintain positions closest to
the center of the group would be reproductively favored
relative to marginal individuals. The analogy would be to a
school of fish where a position in the center of the school
reduces the risk of predation. Unlike fish, however, the
characteristics defining this center can shift as differing
237
group strategies become

more
effective
in differing
environments,
Depending of the availability of resources,
the center might be relatively flat and inclusive or steep
and exclusive.
Similarly, the proportion of marginal to
central individuals would also vary with social and habitat
conditions,
in some periods there being a higher proportion
of tolerated marginals and greater diversity among them.

Interpersonal selection for mechanisms favoring individual
success in group-living (e.g., behavior related to prolonged
caregiving and dependency, maintenance of group membership,
and prediction of others' behavior) would be available for
supporting increasingly complex social coordination, the end
result being 'intense sociality' (a psychological concept
not to be confused with Campbell's (1975) sociological
ultrasociality). Barriers to cooperation such as aggression,
competition, or spite are thus not eliminated, but such
behaviors are constrained.
Campbell (1983) proposes that
these behavioral tendencies are accompanied by 'criterion
images,'
mechanisms
integrating
instinctive
and
learned
responses for assessing and evaluating primarily other group

members and secondarily the organism's own behavior.
The task of negotiating one's way through the social
environment is enormous.
With Gould & Vrba (1982) I
hypothesize that what we call human intelligence is an
exaptation of the sheer complexity of the human brain, and
further hazard the guess that this complexity evolved for
'social intelligence'.
An analogy might be our ability to
ride bicycles, which is not an evolved adaptation, but is
made possible by adaptations associated with locomotion.
A full elaboration of selection for human sociality
will include (or be) an ontogenetic theory.
The infant,
biologically adapted to survive as an infant, has as its
most sophisticated adaptations (e.g" crying, imitation, the
perception of biological motion) those allowing it to
interact with
the
social
environment,
This social
environment constitutes the initial buffer between the
infant and the non-social,
material environment.
The
environment,
both social and
material, then expands,
modifies, and ritualizes the functioning of these perceptual
and
behavior
mechanisms.
Ontogenetically,
then,
these
mechanisms support the transformation from the biological

adaptedness of infancy to cultural identification, and these
mechanisms are reciprocally transformed by, and support the
maintenance of, adult membership in a small functional group
sharing customs, rituals, norms, history, and bonds of
L. R. CAPORAEL
238
attachment.
I emphasize small in this discussion because
the predominant part of human evolutionary history has been
written in hunter-gatherer sized groups.
3. TECHNOLOGY IN HUMAN EVOLUTION
What are the effects of technology on human biological
evolution?
For many years, lithic technology (usually
associated with hunting behavior) was believed to be a
'prime
mover'
in
human
evolution.
This
view
has
been
challenged from three quarters: first, research on animal

tool use and the primacy of foraging
and scavenging
techniques used by hominids (Kurland & Beckerman, 1985);
second,
by the
interpretation of paleoanthropological
evidence indicating a decoupling of technological advance
and morphological transitions (Eldredge & Tattersall, 1982);
and third, by analysis suggesting that the notion of a
'prime mover' of any sort may be a mere artifact of the
narrative structure of
evolutionary scenarios (Landau,
1984).
I believe that technology is subsidiary to adaptations
for sociality and thus their role in human biological
evolution (but not cultural evolution) is minor. For most
human
evolutionary
history,
technology
has
been
'body-extensive' and in the same category as animal tool
use. The significance of technology is that it provides the
opportunity for ultrasociality, which cannot be achieved by
selection for intense sociality.
The genetic scope of interpersonal selection is likely

to be limited to the size of a microband, a primary group of
about 20 to 30 (mostly related) individuals. Until about
10,000 years ago this was the typical form of social
organization.
One characteristic of hunter-gatherer social
organization is the seasonal formation
of impermanent
macrobands from permanent microbands.
That is, a macroband
is not merely a scaled-up microband, generating the same
selective pressures and having the same functions. Its
members do not randomly form new groups when the macroband
disperses, nor does the macroband eliminate intergroup
competition
indeed, macrobands often form to conduct
formalized intergroup competition.
In addition to (and based on) evolved mechanisms for
sociality, Homo sapiens culturally constructed and to some
extent controlled behavioral events such as competition,
cooperation, and reciprocity within the group. Between
239
groups (i,e., microbands), sociality mechanisms, events, and

their constructions must be reorganized.
definitions
of the
environment,
and
Norms, cultural
strategies for
interacting among
groups,
are
developed
largely by
exploiting the lability of group membership and redefining
it along different dimensions (cf.
Brewer, 1981; Brewer &
Campbell, 1976; Tajfel, 1981), important functions of myth,
ritual, and moral preachings. For the better part of human
history,
however,
between-group
association has been
temporary, limited not by the cognitive capabilities for
forming larger (albeit often unstable) groups, but by the
lack of technologies to more effectively exploit the habitat
to support a higher density of individuals over extended
time periods.
Given extremely favorable habitat conditions
and the development of technology for more effectively
exploiting the ecology, the flexibility for maintaining and
redefining group membership lent itself to patching together
primary
groups
into
supraordinate
social
systems-
ultrasociali ty.
Humans did not biologically adapt to higher density
units created through technology because there was no (or
insufficient) selective pressure to do so.
Sociality and
the intelligence it gave rise to presented such a tightly
integrated biological organization (i.e., canalization) that
by the time high residential density was viable, suitable
biological mechanisms already existed for reorganizing at a
supraordinate
level.
Fundamentally,
the
'natural
environment' of humans did not change - nor has it changed
yet.
It was maintained by psychological and cultural
factors.
The sociality mechanisms, which constitute the
capacity for culture, 'parse' the stimulus information of
the social environment into small group 'grammar'. For the
practical exigencies of daily living, individuals still
organize themselves along dimensions relevant to small
groups
for example, colleagues, friends, relatives, and
strangers
with various levels of intimacy, strengths of
coalitions, and expectations for reciprocation.
Thus far I have proposed that the role of technology
in human evolution has been to facilitate the functioning of
existing morphological and social adaptations. Technology
does alter the gene pool by allowing forms to reproduce that
otherwise might not (e.g., near-sighted individuals), but
there is
no systematic selection
effect because the
distribution of any single technology across habitats is
variable. Dobzhansky (1962) discusses some cases where
240
L. R. CAPORAEL
technology might select against certain genotypes, for

example among the 1% of South Afrikaner porphyriacs fatally
sensitive to barbiturates.
Such effects, however, are
limited to local populations and have little species-wide
impact. Invoking such selection effects in the context of
the evolution of sociality or intelligence assumes not only
directional effects, but also assumes that if there are
somatic consequences of technology, there are psychological
consequences as well, and the consequences are analogous.
The evidential status for these assumptions are problematic.
In summary, technology, like group living, may act as a
buffer against selection pressures in the habitat, but
unlike group living, it is not itself a selective agent.
4. HUMAN EVOLUTION IN TECHNOLOGY
I have argued that technology has had little effect on human
biological
evolution.
But
technological
design and
implementation is generally conformed to human evolution.
This is most apparent, as I stated in the introductory
paragraphs, with body-extensive technologies. The salience
of technology in our lives often results in a confusion as
to whether or not the locus of an effect is technological or
psychological.
My favorite example of this point is the
advent of computing bringing with it a fear that the
technology would result in a heretofore impossible level of
regimentation and order in human life - a fear epitomized in
Orwell's 1984.
But 1984, and its more powerful and
evocative precursor, Zamiatin's We, written and banned in
Soviet Russia in 1929, are remarkable for what their highly
structured, alienated worlds lacked - the computer. They
are tales not about the regimenting effects of technology,
but of the human ability to conform and adapt, the human
vulnerability to social influence.
We are becoming increasingly aware that technology
brings chaos. (The American Internal Revenue Service with
its new computer is the most recent constituency making this
observation.)
The chaos is a consequence of the conflict
between 'mind-extensive' (1) technology and adaptations for
sociality.
The problem is that we do not know by simple
observation the salient features of the mind the way we know
the salient features of the body.
Thus, we can use
technology to build bureaucracies, but as Campbell (1982)
has argued, distortions in bureaucratic rationality are
241
observed when mechanisms that evolved to support clique

solidarity interests (e.g., mutual monitoring and ingroup
favoritism) thwart higher-order collective interests. Both
technology
and
social systems
are
products of the
flexibility of human behavior.
But other characteristics,
specifically adaptations for sociality that evolved in the
context
of small
functional
groups,
appear
to be
evolutionarily stable and show little evidence of change.
We still have conflict between groups, although our methods
for killing are technologically advanced. Familial systems
still exist even though maintained through the telephone
wires.
Such apparent immutability of some categories of
behavior would be expected if the social context attenuated
natural selection pressures from the physical environment
and itself functioned as a selective agent.
A number of ideas and hypotheses, many of them
complicated by trying to describe continuously reorganIzIng
systems, have been compressed into the previous pages, but I
hope
an example
identification is
may
prove illustrative.
Individual
species-specific characteristic, and
infants have a preference for facial features organized as a
face rather than randomly organized.

This general face
schema is a biological adaptation.
Infants subsequently
develop
the ability to
remember
and
recognize
faces, an
interaction between
biology and experience.
But the
evaluation of and preference for faces having 'handsome'
features is learned in the social group. With individual
experience we may incorporate the group's evaluation and
preference, reject it and influence a shift in the ideal,
remain marginal with respect to the group in our preferenc~
or even find a new group.
We
can
also
use
the
face
in technological
applications, as when Chernoff (1971) faces are generated by
computer
and used
to
present
complex
multivariate data.
These are schematic faces (face-plots), in which the value

of a variable is mapped onto a particular facial feature.
Presumably a face-plot is like any other complex data
representation plot (e.g., ca.tles, stars, and ships) in
that their information content ~s carried in distinctive
feature representation.
But
social psychological research
has demonstrated that facial recognition is better for faces

judged
on
fuzzy
attributes
like
friendliness
or
trustworthiness than for judgments based study of the
physical attributes of the face (Patterson & Baddeley,
1977).
This suggests the possibility that the utility of
L. R. CAPORAEL
242
face-plot technology could be under constraints

of a
biologically based system for coding facial information.
Most mind-extensive technologies are like face-plots. To a
greater or lesser degree they may have some utility and be
accommodated and assimilated into our social groups and
social systems, but we do not understand how they work.
5. SUMMARY AND CONCLUSIONS

It has been fashionable to assert that humans, no longer
living the life of hunters and gatherers on the savannah,
have lost their 'natural environment' through the distorting
influence of technologies, agricultural and onward (e.g.
Symons, 1980). I have proposed that, on the contrary, the
'natural environment' has not changed since the emergence
of Homo sapiens, largely because the small group as the
field for action is maintained by psychological and cultural
factors.
Broadly speaking, innovations made it easier to
exploit the habitat, separate the heads from the bodies of
one's enemies,
and
increase
the
comfort
associated with
obtaining life's necessities and the enjoyment of the

non-necessities - all selective tests of new technologies.
But many
new technologies can alter
the basic
structure of our natural environment.
The transportation
and communications technologies of this century sever the
temporal and spatial contiguity that was once fundamental to
groups.
Some
technologies may not work
under such
conditions.
For example, in discussing the failures of an
international computer-mediated conference, Tombaugh (1984)
concludes, "It may be that electronic mail and computer
conferencing in small groups, where individuals know one
another and have a relationship of trust, will prove to be
one method of developing greater computer communication
among scientists" (p.
142). Other inventions potentially
increase the number of dimensions that can be used to define
a group, the
number of possible interacting groups, the
number of different individuals that can fill roles within
groups, and the possibilities for defining supraordinate
~rou~s.
Still other technologies
stretch beyond our
lmaglnation.
The effects of exploiting the habitat may be
felt on the other side of the globe, we no longer see the
whites
of
our enemies'
eyes,
and
life's necessities and
non-necessities are obtained through a global marketplace.
The
perception
of
243
technology
as out
of control or
having an autonomy of its own may derive from the difficulty

of assimilating it to the cognitive mechanisms selection has
worked out for us or the cultural norms associated with
those mechanisms.
To reiterate, culture is built on the
mechanisms
for sociality -
it may continually
alter their
expression, but it must accommodate their existence. Hence,

our intellectual capacity to create new technologies expands
beyond our capacity to comprehend, abstract, and control

them.
The last decade has seen a flowering of research
demonstrating the limits of human rationality (Kahneman,
slovic & Tversky, 1982, have produced a compendium, although
there is yet no compelling theory),
These limitations
produce a problem for evolutionary theorists of Campbell's
(this volume) first category
biological evolutionary
epistemology -,
believers
in
processes that
to say nothing of the problem they poses to
the fundamental rationality
demonstrate the
of
boundedness
humans.
The
of rationality
are implicitly assigned a place beside such social processes

as ingroup bias, conformity or groupthink: all are viewed as
intruding
on
or
disrupting
the normative
rationality by
which we have developed control over the natural world. A

revealing quote on the jacket of Kahneman, Slovic and
Tversky's book tells us that we now know human ratiocination
does
not
conform
to
normative
theory;
rather,
people
'replace' the laws of chance with heuristics, But in fact,

some people replace heuristics with tbe laws of chance that is if they know the laws of chance, a knowledge usually
gained
after
considerable
study
and
training.
Phylogenetically and ontogenetically, the achievements of
reason are hard won,
come in small increments, and are part
of the cognitive economy of groups.

It may be human
behavior evolved to be 'rational' in a small group social
sense,
and that philosophy, science, mathematics, and
technology represent a continual effort to overcome the
shortcomings of this 'social logic' mistakenly applied to
the physical world and to increasingly complex societies,
If so,
as our domains of action are broadened, the
consequences of cognltlve and affective mechanisms evolved
in an evolutionary past become more important,
A biosocial perspective does not doom us,
however, to
talk about our biological limitations,

On the contrary, I
have argued that human evolutionary history has been one of
exploiting potentials.
The rational
methods discovered by
244
L. R. CAPORAEL
human intelligence give us a prospect for

implementation of technology beginning with
improving our
improving our
understanding of ourselves.
NOTE
1.
Writing might be the
of a mind-extensive
technologies
do
earliest non-controversial example
technology.
limit
Of
course, body-extensive
mind-extensive
technologies
somewhat shape
ideational content.
But
function of mind-extensive technology is not
and
the dominant
to extend the
muscular or sensory capabilities of the user;

it is to
extend cognitive events so that those events can be shared
among individuals. I have been toying with the hypothesis
or at least a collection of observations and hunches
that the earliest co-occurrence of body-
and mind-extensive
technology revolves around rhythm. It seems that only a

slight genetic shift might add rhythmic capabilities to the
tone modulated signalling of primates. The tonal and pitch
variations of baby talk include a rhythmic
component, and
infants are exposed to different rhythms through lullabies
and children's songs. The closest we can reliably come to
shared affect is through rhythm. Rhythm unites the fighting
force or the working force in the fields.
Rhythmic hand
clapping expresses approval. The production
of rhythm
produces the feeling of shared membership in a group, one
that can be extended beyond primary group to large-scale
groups. Rhythm can be produced vocally,
with body parts
by clapping, or by striking objects. The invention of stone
tools may have come about not by a brilliant insight, but
accidentally, by striking two rocks together to produce
rhythmic sound.
Partm:
The Piagetian approach
IS PIAGET'S "GENETIC EPISTEMOLOGY" EVOLUTIONARY?

Christiane Gillieron
Universite de Geneve
1. INTRODUCTION
"From what I have understood, (Piaget) just
describes the process of adaptation, or adaptive
mechanisms,
or
the
development
of child
intelligence.
These concerns are already named
in psychology,
which is more precise than
epistemology
or
knowledge"
(Nivnik,
in:
Silverman, 1980, p.1S).
If this were the case, Piaget's genetic epistemology (GE
henceforth) could be viewed as a kind of philosophical
wisdom developed by a very special philosopher, who valued
science greatly and needed hard facts in order to help
himself philosophize.
But Piaget's system is much wider
than just the extension of a psychological theory. In order
to understand how wide-ranging his project was, we have to
go back to his 1918 novel Recherche, where he expounded for
the first time a model that will appear again and again in
later works.
This model, or 'image of wide scope' (Gruber,
1978) is that of the 'circle of the sciences'.

It is the
key to Piaget's GE as
it
simultaneously
shows the
importance of interdisciplinary reflection, justifies the
'genetic' method and demonstrates the central role of
psychology in a scientific approach to knowledge. Piaget
always believed that epistemology would become a true
science, with certain necessary characteristics:
1. Interdisciplinarity.
Every science must be reflective
and must explain itself.
2.
Genetic and historical approach: Without this coupling
we would have on the one hand purely deductive sciences and
a
on the other purely empirical ones. This would lead to
radical dualism swinging between abstract tautologies and
the concrete world.
3.
Central role of psychology:
Indispensable as the link
required
between biology and mathematics.
In its simplest form, the circle

appears as follows (Piaget, 1947b, p.148):
247
W. Callehaut and R. Pinxten reds.), Evolutionary Epistemology. 247-266.

/987 hy D. Reidel Publishing Company.
of
the sciences
c. GILLIERON
248
Mathematics
Psychology
~ Biology
~YSiCS
.-------
While the sciences have developed historically in a

linear order
(Mathematics ->
Physics ->
Biology->
Psychology), a critical examination of their relationships
reveals a tendency of the series to close up. This closure
is
the consequence of
a double
evolution
in
science, or
rather of an
oscillation:
"Scientific
thought ( ... )
oscillates between two poles; the mind explains physical
reality through mathematics; but physical reality explains
the mind and mathematics through biology" (Piaget, 1929,
p.147). Scientific evolution corresponds, from one point of
view, to a progressive mathematization of reality, and from
another, to self-explanation through biological laws (lato
sensu).
It must be noted that this complementarity goes
along
with an opposition between
two epistemological
attitudes:
IIBetween
the
two
poles
of
mathematics
and
biology, thus symmetrically oriented, physics and psychology

in a similarly complementary way
participate in the
idealistic current dominant
in
mathematics,
as well as in
the realistic one, of which biology is the purest example"

(Piaget, 1947b, p.149). But the fact that these oppositions
among (traditional)
epistemologies
coincide
with the
boundary
of
disciplines
is
understandable.
While
mathematics 'use' deduction, i.e., a mathematical mind, they
focus on the 'subject'; biology, which is empirical, focuses
on the 'object'.
Meanwhile, 'subject' and 'object' are
precisely the residue of a polarization, which must be
studied as a
'real'
process: both a biological and
psychological one.
The schema thus summarizes the whole
project of GE, which presents itself as an answer to the
"eternal problem of the agreement between mathematics and
reality" (Piaget, 1955, p.33)(1).
The special role of psychology in the system of
the
sciences must be viewed with respect to
the problem
of the
mathematization of reality. It is then understandable that

Piaget had at heart to elucidate the nature of the link
between psychology and logic/mathematics, and to convince
the professionals in these disciplines of the 'closure' of
249
the circle.
This is a difficult point if one wants to
escape the criticism of psychologism - which Piaget did not,
according to some epistemologists - as well as that of
logicism.
Piaget's
main argument was eventually
the reality of necessity,
facts.
But
here
i.e.,
we have to
developed in the sixties.
based on
the existence of normative

examine the model as it was
2. THE CIRCLE OF THE SCIENCES, VINTAGE 1967

The most detailed presentation of the circle of sciences by
Piaget (1967c) is far from being exhaustive and clear.
Several relationships
analogically and need
are
still
a careful
the
specific
Gillieron,
in prep.).
classification
of
In addition,
implicit or treated
examination (Borel &
it must be noted that
disciplines has undergone
several modifications, justifying Gruber's (1979) claim that

the circle changed in shape.
Meanwhile, the feature
essential for it to
be a 'circle',
at least topologically,
is the closure of the series due to the links between logic

and psychology.
These are discussed repeatedly, and the
reasons given are always
the
same.
These
invariants are
what I shall emphasize here.

(See Beth & Piaget, 1961;
Piaget, 1950, 1952, 1954, 1955, 1964, 1966, 1967a, 1968,
1970 chap.3, 1973, in addition to the papers previously
quoted) .
Logic and psychology are dependent one upon the other
at two levels: the level of the material domain and that of
the derived epistemology.
At the two intermediary levels,
they are independent.
Four levels are distinguished in each science:
A.
The material domain, i.e., its objects, such as numbers
for mathematics, bodies for physics, actions for psychology;

B.
The
conceptual domain,
i.e.,
the theories, the
acquirements specific to it;
c.
The internal epistemology, i.e., the critical theories

of the conceptual domain;
D. The derived epistemology, i.e., the general epistemology
stemming from the
internal epistemology (relationships
sUbject/object
in
this
science,
as
understood
in
relationship to that of the other sciences).
The
interdisciplinary
order at levels A and D,

C (Fig. 8).
connections
show
circular
and a linear order at levels Band
C. GILLIERON
250
Epistemology of
psychology (often)
borrowed from biology
,.
-------
ipis-t:~orOg; of
biology -:
(ghostly), - - -,'
--
"
Figure 8. Interdisciplinary connections as conceived by

Piaget (1967c). Four main groups of disciplines: (I) Logic
and mathematics, (II) Physical sciences, (III) Biology, (IV)
Sociological sciences.
(see text).
A,
B, C and D represent the domains
251
At level B, logic and psychology are independent:

"This independence of the two disciplines, accepted today by
almost everyone,
implies
that we must
renounce all
'psychologism' in logic and mathematics and all 'logicism'
in psychology, which means in effect that the autonomy of
enquiry in each of these respective fields is guaranteed"
(Beth & Piaget, 1961, p.325). At level C, we may accept the
same linear order, due to the fact that level C consists of
a critique des fondements: generally the same reasons that
lead to a relationship at level B would
lead to a
relationship at level C.
In the case of logic, this would
be especially the case,
since there is no clear-cut
distinction between Band C.
The foundations of logic are
studied with logical methods, and the distinction between
logic and metalogic is internal to the conceptual domain of
logic, i.e., domain lB.
But the main hypotheses concern
levels A and D.
When they ask the epistemological question of the
nature of logical objects,
logicians need a subject,
transcendental or natural.
However, the transcendental
position seems difficult to maintain in the face of modern
metatheories.
The limits of formalization (GBdel's and
Church's theorems,
Kleene's hierarchies of predicates,
Tarski on the representation of truth in a system, etc.) as
well as the multiplicity of formalisms and the new idea of a
hierarchy of
structures suggest instead the
idea of
construction.
Platonic realism is ruled out, since "the
theorems of the limitation of formalisms compel us to bind
essences to a construction that never ends, and we find that
a subject necessarily intervenes, be it transcendental or

human" (Piaget, 1967g, p.575). But apriorism seems no more
appropriate, since it cannot explain the relativity of
logical forms.
It is necessary, then, to relate logical
beings either to physical reality, or to a biological
subject via psychology.
If
logical
beings
were
extra-mental,
logico-mathematical knowledge would be no different from
physical or empirical knowledge: a logical law could be
'discovered' in the same manner as empirical properties are
'abstracted' by observation and/or actions, i.e., they
emerge as a consequence of
the interaction of
the organism
with the outside world.

But in this case, how could one
explain the
character
of necessity
that
goes with
mathematical and logical systems? And how could mathematics
anticipate empirical results?
Logic and mathematics are
C. GILLIERON
252
viewed as norms by those who use them, and thus must be

viewed as normative facts by the 'meta-subject', i.e., the
epistemologist.
At level D, the logician-epistemologist
must then admit that his epistemic subject is dependent upon
a 'real' subject, whose portrait has to be drawn by a
naturalist.
As Ladriere (1982, p.18) summarizes the whole
argument:
"Apriorism
as
well
as
psychologism seem
incapable of explaining the real status of
logic. Apriorism fails because it can explain
neither the historicity of logic ( ... ) nor the
relativity of logical forms, which is itself
moreover a cue to its historicity. Psychologism
fails too
because
it cannot
explain the
normativity that characterizes logical rules and
the necessity tied to the derivations made
according to these rules ( ... ).
The idea of
'reading' (a kind of absolute logic inscribed in
the mind or actualized in a separate field
available to some intellectual intuition) must
be replaced with the idea of construction, and
consequently of operatory genesis".
This can justify the closure of the circle at level D, from
IVD to ID.
By the same token, this allows the closure at
level A: the natural subject needed at level D is the
premises where logic grows.
Since an 'operatory genesis'
always needs a preliminary content, the ultimate object of
formalization must dwell in a psychological or sociological
or
linguistical
subject.
The
remaining
derived
epistemologies of
logic, conventionalism/nominalism and
constructivism, all rely on such extralogical entities.
The fundamental role of the circle of the sciences may
be denied by scientists who wish to restrict their role to
domains Band C:
sciences
are concerned with
concepts and
theories while the 'material domains' need epistemological

assumptions in order to be delimited. As for the 'derived
epistemologies' they fall within the competence of the
philosopher.
To this Objection Piaget replied that the
internal epistemologies (C)
evolve with
the sciences
themselves as they evolve.
Since an epistemology is the
critique of the relationships between concepts (B) and
objects (A), it is necessary to take into account the
relationships between the material domains (A).
This fact
implies a general epistemology (D) (1967c, p.1178).
253
It is therefore not by chance that the epistemologies

paralleled by biological and psychological theories (Piaget,
1947a,
pp.
17-25) correspond closely to the derived
epistemologies of logic as presented in Piaget 1967a (see
Table 1).
Logic (IB and Ie) gives the strongest argument
in favor of constructivism (ID) and suggests the central
hypothesis that at its beginning it must be conceived as
the formalization of the opera tory
structures of the
subject.
The process of reflective abstraction, which is
the key process in formalization, works in psychogenesis as
well.
The fact that elementary structures may be described
in psychogenesis and the fact that they can be formalized
present
another
argument,
this
time
from
GE
(IVD) .
3. "ADAPTATION" AND THE SUCCESS OF MATHEMATIZATION

Piaget reasserts the key position of psychology in the
system of science on several occasions.
This is due,
neither to the
questionable
fact
that psychogenesis
recapitulates history, nor to an exclusive focussing on the
subjective side of science, but to the importance of the
logico-mathematical frames in science. This legitimates the
appeal to psychology in order to solve epistemological
problems.
The genetic
study
of
human intelligence
(domain IVB) allows GE (IVD) to 'choose' from the classical
epistemologies a solution that seems to agree with the
internal epistemologies in every field, as well as with
empirical
data.
"On one hand, psychology d~pends upon all the
other sciences and sees 1n mental life the
resultant of the physico-chemical, biological,
social, linguistic, economical factors etc.,
that are studied by each
of the specific
disciplines
concerned
with
objects
or
surrounding reality.
But on the other hand,
none of these disciplines would be possible
without a logico-mathematical coordination which
indeed expresses the structure of reality, but
which can be grasped only through the actions of
the organism on the objects; and only psychology
allows us to study these activities in their
development" (Piaget, 1966, p.12, our italics).
It is necessary at this point
to prevent
a common
misunderstanding, as expressed for example by Flavell:
C. GILLIERON
254
"The mind builds its knowledge structures by

taking external data . and interpreting them,
transforming them, and reorganizing them. It
therefore does indeed meet the environment in
the process of constructing its knowledge, and
consequently that knowledge is to a degree
'realistic'
or adaptive
for the organism"
(1977, our italics).
The interaction of a subject in the
flesh with its
objective surroundings is
but one manifestation of the
encounter of the organism and the outside world. Meanwhile,
it is not the limited experiences.of individuals confronted
with the real world that can explain the miraculous
match
(2)
between mathematics and reality.
The classical
oppositions
of
natural and pure,
empirical and deductive,
synthetic and analytic sciences still apply. Indeed, since

nearly everybody agrees in considering the outside world as
real, that could explain the adaptive character of physical
and biological knowledge: in these fields the problem of
formalization is reduced to the
problem of knowledge
itself.
But if logico-mathematical objects are the results
of a continuous and indefinitely fecund
construction,
both
their
rigor
(the normative
aspect) and their
applicability are mysterious.
It is true that
for
Piaget logico-mathematical
structures are not tautological but 'express' (as he says)
deep properties of reality. However, this expression does
not lead to or permit a 'reading', which due to the
materiality of the
confrontation would be 'correct'.
Mathematics do not merely copy reality, they are not limited
to some structures or transformations which correspond to
effective processes:
on the contrary, they overflow reality
and
even
anticipate
their
own
applicability.
If
mathematical
structures
'express'
the
object,
the
coordination which leads to mathematical structures must be
a fundamental law of the object.
If we exclude any aprioristic
explanation, this
harmony must result from the psycho-bio-physical nature of
genesis.
Logico-mathematical structures emerge out of a
real process, they are deductive constructions by concrete
subjects, i.e., organisms.
The ties are internalto the
subject:
mathematics
are
the
result
of reflective
abstractions starting from the opera tory structures of the
subject which depend upon the coordination of his actions,
which depend upon the nervous connections, which depend'upon
the organic
regression.
structures
Meanwhile,
and
lilt is
so
this
255
on,
in
a perpetual
very regression that
helps us to reach the biophysical reality, but inside the

subject so to speak and not in his empirical or external
searches" (Piaget, 1967g, p.587).
Consequently, the importance of genetic psychology
does not come from the mere description of the naive
theories of
children at different
ages, which could
hypothetically disclose the constraints that the object
prescribes to the subject. It comes from the fact that only
the empirical study of psychological development may lead to
the
understanding
of the
mechanisms
of reflection.
Knowledge seems to us both objective and deducible, as the
result
of a
double
movement
of
objectivation and
internalization, which psychogenesis offers to dissection.
The developing child allows the psychologist to be present
at her construction of reality
that is to say, its
normalization
by observing the genesis, or at least the
chronology of normative facts.
Considering one of the most famous normative facts
studied by Piaget and his collaborators, the principle of
conservation,
one
must
insist
significance of the findings in
mathematization.
i.e.,
The first
evident,
'objective'
fact
unquestionable,
on
the
respect to
for
epistemological
this problem of
the seven-year-old,
'normal' or, let's say it,
'logical' fact, is that in the transformed ball, the

quantity of playdough is still the same. How does the child
come to 'know' that?
How can one show that "it's a fact"?
None of the classical arguments - given by adults as well is truly convincing.
They are either mere descriptions, or
precisely assume what they are supposed to prove. Reshaping
the playdough just suppresses the problem.
Pointing to the
covarying dimensions only shows that the conserver is
convinced that these covariations compen~ate exactly, which
supposes either calculation with proport10ns (formal stage,
around fourteen years of age) or the axiom of equality. The
identity argument does not apply for quantity: the same
playdough doesn't mean the same amount
of playdough.
Finally, when arguing that 'nothing was added, nothing
substracted', the child sets forth the principle that
'adding'
or 'substracting' may result only from some
specific transformations, such as bringing or taking away a
piece of playdough.
The actual manipulation - reshaping -
C. GILLIERON
256
does
not
belong
to
the
class
transformations.
Meanwhile, it
surface area; how does one decide?
of relevant-for-substance
is
relevant
for,
say,
The conservation of matter thus appears as a typical

example of mathematization.
'Matter' is only indirectly
measurable through some of its properties: weight, volume,
inertia, etc.
Meanwhile, 'matter' is conserved before any
of its properties. This shows that the child does not learn
from some empirical reading that the playdough still is the
same: she is putting forward a postulate. The quantity of
matter truly is substantia, hypostasis, a child of the mind
(Gilli~ron, 1982).
When putting forward this postulate of conservation,
the child acts with the same faith as the physicists who
'calculate'
energies and spins and trajectories: reality is
mathematizable.
Meanwhile,
unlike the physicist, the
logician-child does not question
this mathematization;
structures are autonomized before any physicalization. It is
only progressively that she realizes that this application
must
be
justified
theoretically.
The psychogenetic
progression from the logician to the geometrist and the
physicist mirrors the anticipatory nature of mathematics.
4. GENETIC EPISTEMOLOGY, PSYCHOLOGY AND BIOLOGY
GE,
like
any
other
'derived'
epistemology,
is
interdisciplinary. It has been shown that far from being an
appendix of psychogenetic theories, one of its central
theses comes from the image of the closing circle of
sciences, mainly due to the internal epistemology of logic
and mathematics.
As Henriques puts it: "The very existence
of
mathematical thought
is a
major epistemic fact;
neglecting
it,
any
relevant
epistemology
would be
impossible" (1984, pp.
54-55).
Meanwhile, if the links
between logic and psychology are fundamental
for the
epistemologist, one cannot underestimate the 'organic' link
between biology and psychology.
Analyzing this link is not" as easy as it may seem.
The existence of two factions in psychology, the scientists
('serious'
psychologists)
and
the
humanists ('soft'
mentalists) is a living picture of the fundamental problem
in psychology, as summarized by Greco (1967, p.937): "It is
the misfortune of the psychologist: he never is sure that he
is 'making science'.
If he does, he is never sure that it
is psychology".
To the difficulty of delineating the
257
material domains of psychology and biology (the former being

part of the latter) we must add the poverty of the
corresponding epistemologies. Organicists in psychology as
well as biologists are prone to consider their object as
given, and to spare the discussion on fundamentals. This
trend is remarkable in biology (Buscaglia, 1983; Meyer,
1967), but no less real in psychology, where biological
theories (domain IIIB) are promoted to epistemological
credos.
Piaget's GE is often presented as a 'biological'
epistemology.
Indeed, it is 'biologically' rooted as it
comes from a biologist for whom Reason appears as the
necessary end-product of biological evolution (3).
But GE
is a general epistemology, only indirectly. linked to a
specifically biological (or psychological by the same token)
epistemology (domains IIIC or IVC). In fact,
as noted
above,
these are lacking and biology appears as
an
instrumental
discipline,
whose
constructs
an
'organism'
may
serve the
epistemologist in the same way that psychology provides some

of the facts that help epistemology to become objective. In
addition, the relationships between biology and psychology
are distorted in the schema of Figure 1.
Psychology is
better viewed as a subdomain of biology, since indeed a
general biology must incorporate the field of behavior.
This inclusion is distinct from the current reductionism,
for its aim is to compel biologists to question their
traditional way of defining their object.
"The living organism is a kind of isolate antedating
experience, a natural apparatus already constructed in the
universe ( ... ) Biological structures exist, with no more, no
less evidence than the table or the house" (Buscaglia, 1983,
p.12).
This is the realist view that everybody shares.
But what makes the organism different from tables and
houses?
The living systems are not 'isolates':
"The properly biological system paradoxically
extends beyond the limits
of the organism
stricto sensu.
The behavioral field appears as
a structural net projected by the living being
on its surroundings ( ... ) In fact, the question
is
not
to
relate
and
an
'environment',
but rather to point to the
unequivocal trace in the environment of the
structures that make it integrated into the
organized systems" (Meyer, 1967, pp. 785-786).
C. GILLIERON
258
living
system
organism is at
the same
time a physico-chemical
and an observed subjectivity.
It is the
locus and
the
source of
behavior and mind.
That
is
why
"psychology is first a biology" (Piaget, 1968, p.119), but
a biology not restricted
to the study
of
the material
aspects of the
living beings;
preoccupied,
on
the
contrary,
with the organizing role of subjectivity.
This antireductionist identification of the material
domains of psychology and biology may explain the solitary
position of Piaget with respect to the classical dichotomy
of monism and dualism.
Contrary to the 'normal' view of the
day, Piaget (1963) advocates a kind of dualism in which

causal chains at the physiological level and implicative
chains
at
parallelism
the
conscious
level
are parallel.
Such
is on a par with that of mathematics and
physics. For a similar reason, biology is thought to enrich

physics in the same way as it is enriched by psychology.
The true relationships in the circle of the sciences are not
ones
of
asymmetrical
dependence,
but
of
mutual
determination.
GE,
In order to become the theoretical biology needed by

general biology must consent to approach all its
problems -
from
double point
organization.
of
morphogenesis
to
evolution
from this
view of
physico-chemical and
For
indeed,
at every
level,
psychic
organlc
regulations refer, in their functioning, to those components
of
subjectivity
that
significations
are.
Only
significations may
direct
behaviors,
from
the most
elementary to the most evolved. Biological genesis presents
itself as a succession of more and more complex forms, more

and more
diversified
structures.
Abstract structures
produced by the
formalizing activity of human
minds become
themselves the source and content of new structures, while

at the
bottom of this scale
former
structures are
discovered,
'being'
whose
origin
is
of structures is their
more and
more remote.
structuring"
"The
(Piaget, 1968,
p.120).
The
organism
is
the
paradigm
structure,
consequently, and "if we knew our own organism through and
through it would, on account of its double role of complex
physical object and originator of behavior, give us the key
to a
general
theory of
structures"
(ibid, p.40).
(Unhappily, biology is far from that point ... )
Coming back to the circle of sciences, which the
provisionally distinguishes the four groups of disciplines
(the Popperian worlds 1,2,3, & logic?) we may now view it
259
as the projection on a plane of the spiral that schematizes

the (real) genesis of norms, the genesis partly attended
to through psychological
morphogenesis. The elementary
behaviors coordinate into systems that are already logical,
and which lead to the first autonomization of
a structure,
in the shape of objectivity (the major 'normative fact' of
two-year-olds).
Operatory
structures
themselves,
concrete,
then formal,
result from a differentiation
between form and content.
This becomes possible
through
the exercise of the two forms of abstractions.
"Content
and form are in fact two faces of the same process ( ... )
where at the same time the properties of actions are
attributed to the
objects,
and the system of these
properties is 'reflected' in the necessity of the resultant
relationship" (Borel, 1978, p.157). Beyond the constructive
character of reflection, which allows true novelties, the
functional continuity of development is striking,
since
both movement of objectivation and internalization are at
work in science as well as in psychogenesis (see Piaget,
1967d, sp. pp.1259 sqq.).
5. PSYCHOGENY AND PHYLOGENY
Knowledge, intelligence, adaptation: these concepts
are
closely
connected.
'Intelligence'
allows
for
'knowledge' and knowledge is 'adaptive' for the intelligent
knower
adaptive of course, as far as true knowledge is
'adapted'
to reality.
The
adaptation of scientific
knowledge to reality, that of the (individual) intelligent
being to its environment, and that of the organic forms to
the world,
are
correlative.
Meanwhile,
these
terms are
commonly associated with different domains, and they even

appear as headings in a classification proposed by Piaget on
several
occasions
(e.g.,
1929, 1947a, 1967e).
When
discussing
the
correspondences
between
biological,
psychological
and
epistemological
theories,
Piaget
designates them shorthandedly by the terms of theories of
adaptation, of intelligence, of knowledge.
However, the
classification itself demonstrates the communalities behind
them (Table 1).
To the different brands of sophisticated,
explicit epistemologies, correspond the nalve and implicit
epistemologies underlying
biological
and psycl.ological
theories.
H"
>".
0
H"
"
"....
TERTIUM QUID
(NEO)DARWINISM
OPERATORY
TRIAL/ERROR
CONVENTIONALISM/
PRAGMATISM
CONSTRUCTIVISM
EMPIRICISM
PHENOMENOLOGY
PLATONIC IAN
REALISM
APRIORISM
EPISTEMOGOLY
Theories of
knowledge
".,'"'
H"
.,z
rt
P-
"ro
"
"ro
.,",....
...,
.,...
structures . . . .
Construction ~.
Empirical
data
Ready-made
Interaction
Object
primacy
Subject
primacy
Table 4: Traditional epistemologies may be viewed in parallel with the biological

theories of adaptation and the psychological theories of intelligence. This
classification proposed in 1947a by Piaget coincides with that of the derived
epistemologies of logic discussed in Piaget 1967a (4).
Both
Internal
ASSOCIATIONISM
GESTALT
EMERGENCE
Both
(NEO) LAMARCKISM
"DENKPSYCHOLOGY"
"FACULTIES"
PREFORMISM
CREATIONISM
PSYCHOLOGY
Theories of
intelligence
Internal
External
,.... External
'"<
H"
>".
>".
'"X
Factors
BIOLOGY
Theories of
adaptation
E
"',
Cl
(")
261
Every epistemology hypothetizes the existence of two

entities
and
relates
the
two.
The transcendental
epistemologies may refer to a transcendental subject and to
essences or transcendental objects. Ideal as they are,
'subject'
and 'object'
they are named.
Naturalistic
epistemologies refer, on the other hand, to an epistemic
sUbject behind or beyond 'subjects' as well as to an
'objective' or material reality.
Meanwhile, subjectivity
and objectivity emerge as epistemic categories, in natural
minds, after having emerged as practical categories in
living organisms.
From the objectivizing point of view of
the scientist (of the biologist, in this case), only
sensations exist for the organism.
They are finally
experienced as stimuli bringing information either about the
world
(exteroceptive)
or
about
internal
states
(proprioceptive).
The contact zone between skin and world
and the internal transformations resulting from action
(movements, resistance) give rise to correlations, hence
meanings. The personal history of the organism allows it to
differenciate
'objective'
experience
and 'subjective'
experience.
It
can
be
seen
that
'naturalistic'
epistemologies are natural epistemologies for naturalists,
which has already been emphasized. This fact points to the
necessity of describing psychological (or more largely,
biological) adaptation, at both phylogenetic ('forms') and
ontogenetic levels ('learning'), since objectivity, and
science by the same token, must be adaptive.
If we admit Piaget's classification i.e., paralleling
on the one hand
Lamarckism,
associationism and
empiri~ism,
who favor the 'objective' pole of experience (external

origin of knowledge), on the other hand mutationism, trials
and errors, pragmatism, who favor the 'subjective' pole of
experience (internal origin), all these theories leave us
with a crucial problem which is ... adaptation. As long as
adaptation is explained as the consequence of experience,
i.e., interaction of the 'subject' with the world, the
adaptive character of (formal scientific) knowledge is not.
Selection, at both onto- and phylogenetic levels may help to
understand survival but not mathematical truth.
With ready
made structures, genesis is not understood. GE endeavored
precisely to escape from the aporia.
But historically its
first aim was to fight against empiricism, and this fact
explains the importance given to Kantian categories. Later,
as the theory of innate ideas reappeared, Piaget developed
biological arguments against the corresponding systems.
c. GILLIERON
262
The shift is noticeable in the number of biological

articles he published after the classical structuralist
period
(see
Buscaglia,
1982).
When
answering the
epistemologists Lorenz, Monod, and Chomsky, Piaget argues
qua biologist. To Lorenz, specialist es innate cognition,
Piaget opposes
logico-mathematical
knowledge with its
necessity: "If the a priori evolve like some biological
characteristic, being prior conditions for every kind of
experimental knowledge and fixed in heredity as instincts or
innate conceptual framework, then they must lose, along with
their uniqueness and their universality ( ... ), the very
thing that gave them their chief value, which was their
necessity" (Piaget, 1967h, p.360).
Monod equates the epigenetic construction of organs
and behaviors: Piaget asks him to distinguish between
programs and possibilities. The actualization of actions
implies autoregulations,
and
the
very
mechanism of
autoregulation should be part of every explanation of living
organization,
including evolution (Piaget, 1971a).
(A
detailed analysis
of the way autoregulation
acts at
different levels is presented in Piaget (1977).)
The same arguments are expanded in the Chomsky debate
(Piatelli-Palmarini, 1979) where Piaget clearly opposes
'most
biologists'
except
Paul
Weiss
and Changeux,
considering
"how
lightly
behavioral
and
cognitive
adaptations are explained by selection, as if
selection constituted a kind of organizing power
that could transform any kind of haphazard or
random variations into adapted behavior ( .... ).
Adequacy-adaptation ( ... ) implies a teleonomy in
relation to the environment (and) this teleonomy
must be explained" (1979, p.409).
This brings us to the core of our subject, neo-Darwinism
and genetic epistemology.
It may be seen
that here
psychogenesis becomes a side argument: when looking at
evolution, Piaget develops his own biological theory, which
appears as the tertium quid in Table 1 (5).
Both neo-Lamarckism and neo-Darwinism are open to the
criticism of (naive implicit) empiricism. Mutationism is
paralleled with an explicit epistemology which emphasizes
preliminary structures, but iri practice, biologists act as
if biological findings could serve as a base for a theory of
knowledge (e.g., using molecular biology as an argument for
explaining
novel
behaviors),
Hence
nea-Darwinian
263
epistemologies
(internal
and
derived)
are
self-contradictory!
A true nativist like Fodor, who comes from
'formal' disciplines, may well object that the structures
of knowledge
are innate,
while contents depend upon
individual experience.
What is known necessarily reflects
the potential structures, since everything that is, was
possible.
However, according to Piaget, it remains to be
explained why any specific possible becomes real. The core
of the disagreement between Piaget and nativists rests upon
the notion of 'potentiality'. No structure exists without a
content,
and the
actualization of a
structure comes
from
some previous functioning.

This gives rise to latter
constructions.
In the manner that morphogenesis is an
epigenesis, cognitive structures are determined by their
functioning. This explains why Piaget wants to look at
evolution
by considering
Piaget,
evolution
not only forms,
hut behaviors as
well: behavior contributes to the shaping of evolution, as

does mind.
theory of evolution. is highly
The tertium quid
controversial and raises the contempt of those who judge
neo-Darwinism to be a definitive answer to biological
puzzles.
Its Lamarckian shadings appal specialists as an
inverse function of the dimensions of the objects they
study.
But as some
admit,
after Waddington, being
'post-neo-Darwinists' , Piaget hoped to see the postulated
mechanisms of phenocopy being more precisely described by
molecular biology.
We do not need to detail here the
arguments concerning the phenocopy phenomenon (Piaget 1974,
1976, 1977b).
It suffices to recall that according to
substitutes
internal,
hereditary
constraints for environmental one, so that organisms may be

said to 'copy' the effects of their environment on them,
through behavioral adaptation.
The multiplicity of living
forms is not deducible, but it does not result from some
accumulation of unfortunate accidents, which is Monad's view
(mutations,
by
definition,
are imperfections of the
mechanism of conservation). Such a conception of the genome
seems paradoxical:
"this pre-eminently conservatory organ should

have undergone an unceasIng series of palace
revolutions while being for nothing in its own
development. We prefer to suppose ( ... ) that it
is not
so
limited,
and
that
the main
evolutionary
'strategies'
mechanisms
and
its brilliant
or reorganizing capacities are the
c. GILLIERON
264
same, representing 'unceasing responses' to the

'tensions' of the outside world" (Piaget, 1971,
p.35).
The functional dimension of constructivism - be it Piaget's
or Lamarck's
is essential for linking evolution and
adaptation. As Voneche (1984) reminds us,
"for
selectionism,
adaptation
is,
in
a way,
akin to
an
immunological
process against
external perturbations ( ... ).
It is all the
more clear in the linguistic theory that Chomsky
has been professing for nigh on 30 years. The
child's learning of his or her mother tongue
appears as a progressive pruning of the inner
tree of universal language, which itself comes
from a central kernel which is innate and
existed before all experience. Faced with such a
frozen landscape, we can only wonder how the
freezing occurred ( ... )" (op. cit. p.229).
The structures without content of selectionist theories are
hypostatized and 'frozen', as is 'genetic information' when
one ignores the psychological face of epigenesis.
From the
constructivist
point of
view,
on the contrary, knowledge,
psychogenesis and morphogenesis rely on 'information', that

is, in-formation, forming inside (von Glasersfeld, 1982,
p.621n). Information qua information does not exist before
life (Cellerier, 1980, p.1897), and what is 'formed inside'
is not a copy of the world: it is the residue of some
previous functioning.
"The crucial difference between the realist and
the constructivist ( ... ) is not that the one
projects his cognitive structures beyond the
experiential interface, while the other does
not; the difference is that the realist believes
his constructs to be a replica or reflection of
independently existing structures, while the
constructivist
remains
role
as
experiencer' 5
structures,
i.e.,
aware
of
originator
of
the
all
for the constructivist there
are no structures other than those which the

knower constitutes by his very own activity or
coordination
of
experiential
particles"
(von
Glasersfeld, 1974, p.12).

A biochemical substance such as DNA is a 'message' not only
from the point of view of the observer: it is a meaningful
sequence for the functioning organism.
A particular string
265
raises up the making of a protein, i.e., is used as a

If a
coded instruction for an active doer, an interpreter.
specific sequence, at one point of the epigenesis, acts as
a STOP signal, it is a sign needing a cognitive system for
whom it has become meaningful.
In acknowledging the
arbitrariness of the code
why should this sequence of
nucleotides suscitate the making of that particular form one must accept the description of genesis in cognitive
terms.
Goodwin's (1976, 1978) structural or cognitive
biology
points to the
same conclusion as
does the
epistemology of logic/mathematics.
Genesis is construction
of structures, organic, psychological or formal structures
being the
expression of
the
same
natural process.
Constructive
it is,
since we cannot conceive of an
intelligence which would construct its own capacity of
constructive functioning.
"A thought capable of generating
its own capacity of constructive functioning, not given at
the start, would paradoxically exhibit a second-degree
constructive power, paradoxically as it is supposed to be
not constructive" (Henriques, 1984, pp.61-62).
6. EVOLUTIONARY EPISTEMOLOGY AND

GENETIC EPISTEMOLOGIES
EE relies on the "blind-variation-and-selective-retention
theme
(Campbell,
1982;
Wuketits,
1984c).
Genetic
epistemologies
do
not.
Like
Piaget's,
Baldwin's
prototypical
version
is
truly
non-Darwinian,
and
stresses functional continuity (Freeman-Moir, 1982). The
remarkable fact that both Baldwin and
Piaget proposed
their own tertium quid biological theory, that both studied
the development of the child as a key to genesis,
is a
cue to its main characteristics: constructivism. While
transposing the 'mutation/selection' theme to the realm of
scientific theories, one admits that 'content' is selected
through external pressure, i.e., non-viable
ideas die.
However, genetic epistemologies emphasize the fact that
knowing is not perceiving: it is anticipating,
formalizing,
separating by self-reflection. Apostel (1977) noticed the

absence of psychogenetic data relating the construction of
psychological,
sociological,
epistemological
notions.
The ontogenesis of social, epistemic or psychological grasps
of consciousness is still lacking.
But the project of
linking naIve
and sophisticated thought,
formal and
empirical
disciplines,
ontogenesis and cultural history is
266
C. GILLIIJRON
viable: no external pressure has yet extinguished

this
hope.
Scientific theories are but instruments
in
the
evolution
towards
complexity,
differentiation
and
self-reflection. Explication is externalized implication.
NOTES
1.
"L'union de
la mathematique et
de l'experience, c'est
l'ideal que poursuit la science, mais crest en meme temps le
mystere supreme que la science se doit d'expliquer. La

SC1ence ( ... ) est a elle-meme son propre probleme" (Piaget,
1929, p.1s2).
2. The feeling of a miracle is well shared. Einstein
remarked that "one may say the eternal mystery of the world
is in its comprehensibility" (Dyson, 1979, p.sO. See also
op. cit., pp.101 sqq).
3.
This point cannot be developed here. Meanwhile,
to Boden (1979, pp.119-120) who argues that the existence of
phylogenetically primitive
species should theoretically
embarrass 'a
necessitarian progressivist
like
Piaget', we
might answer that Humans were not necessary, but Reason

should early or later be manifested in its present form.
Had Elephants evolved up to formal logic, their mathematics
should reflect Reality in the very same way that ours does.
4.
The picture is somewhat different when Piaget
proposes a
classification of
'general epistemologies'
(1967d).
The table given pp.1240-1241 is more complex, and
the position of logical empiricism different.
5.
As Goodwin (1982) remarks: "to take on both these
traditions (empiricist reductionism and idealist preformism)
and attempt a new synthesis and clarification over such a
range of disciplines is not a modest endeavour: but then,
Piaget was not a modest man" (p.s28).
THE GENESIS OF ATOMIC PHYSICS

AND THE BIOGRAPHY OF IDEAS
Arthur I. Miller
Departments of Philosophy and History
University of Lowell
Department of Physics
Harvard University
I have been exploring the connection
of creative scientific
thinking with the origins of scientific concepts, and the

ways in which this connection may
provide
a better
understand~ng
of
scientific
progress.
clue
to deeper
development
insights lnto these problems is an observation made by

scientists such as Albert Einstein and Henri Poincare, whose
research led them to probe the process of thinking itself-
namely,
that
pre-scientific
psychologists,
II
sc ientific
thought"
thought
is
(Einstein,
particularly Jean Piaget,
1934).
of
Some
reached a similar
conclusion.
A fundamental thesis of Piaget's genetic
epistemology is that the evolving structures in scientific
thought exhibit parallelisms in structures in the formative
psychological processes in children. Inspired in large part
by Piaget's writings, I have approached these fundamental
problems by using results from case studies in the history
of science as data for theories of cognitive psychology.
Conversely, this method tests the claims of the cognitive
theories themselves concerning the construction of knowledge
and creativity (1).
This essay I shall focuses on the part
of my research in which results of historical analysis of
atomic physics during 1913-1927 are taken as data for
genetic epistemology.
We will find parallels between
the hierarchical
structures that are the varlOUS formulations of atomic
physics
during 1913-1927
and
structures
in genetic
epistemology.
And we will find that decisive scientific
progress is signaled by establishment of the following
structures that Piaget found to be essential
in the
construction of pre-scientific knowledge
namely, the
permanent object (i.e., the Bohr atom), reversibility (i.e.,
acceptance of the wave-particle duality
of light and
267
W. Cal/chalit and R. Pinxten (cds.), Hvo/Ulionary Epllfemo{ogv, 267-281
/987 hy D. Reidel PuhlishinJ.[Company.
268
A. I. MILLER
matter), conservation (i.e., maintenance of properties of

atomic ent1t1es whether they are particles or waves), and
systematization (i.e.,
final mathematical version of atomic
physics with the correct physical interpretation). These

structures are hallmarks of other theories that I have
examined.
This result
leads me to
propose that a
combination of genetic epistemology with other cognitive
psychological theories could form the basis of a new view of
scientific progress that is basically a
stage theory of
the genesis of scientific knowledge.
Input from other
cognitive psychological theories is necessary because there
are shortcomings in any genetic epistemological scenario.
For example,
we will find that
the stages
in the
development of representative activity in atomic physics do
not always square with the sequence of
stages in genetic
epistemology.
Furthermore,
genetic
epistemology
de-emphasizes the role of visual thinking as the logical
component of thought becomes more highly developed, in
contradistinction with historical data.
The genesis
of atomic physics
is an excellent
historical case study to use as a laboratory for cognitive
psychology owing to its extensive documentation with sources
archival, primary and secondary. A fascinating aspect of
this episode in the history of science is that for Niels
Bohr
and Werner Heisenberg,
among others, conceptual
problems were often more critical than considerations of
empirical data.
During 1926-1927 the most critical problem
on the frontier of physics was the nature of physical
reality on the atomic level.
The continuing struggles of
physicists such as Bohr and Heisenberg to understand nature
at a level beyond sense perceptions forced them to grapple
with problems that cut to the very core of how we construct
knowledge in the world in which we live: How thoroughly
connected are visualization and intuition?
How is the
intuition that leads to one sort of visualization replaced
by another?
The ramifications of their replies to problems
such as these have set the course of subnuclear physics and
altered dramatically our understanding of physical reality
in ways that are still not completely understood.
Moreover,
these problems span
the
spectrum of
intellectual
history
from
physics,
mathematics
and
philosophy, into the German educational system and the
German cultural milieu.
Through exploring these pathways
historians of science have achieved a better understanding
of the response of physicists such as Bohr and Heisenberg
269
(i.e., atomic physics) to the stimuli (i.e., data) that were

available to everyone.
But our understanding of the
invention of atomic physics is incomplete without
trying to
unlock the black box in which stimuli are transformed into

responses - that is, to explore creative thinking itself. A
valuable clue for proceeding is to assume that there is a
connection between pre-scientific and scientific thought.
I shall develop my approach as follows: The procedure
is to use genetic epistemology as a mold for the genesis of
atomic physics.
Genetic epistemology is
a
rich and
intricately developed body .of knowledge.
As a first approximation in analyzing the genesis of a complex and abstract
scientific theory I will deal with part of the core of
genetic epistemology
namely, the stages of mental
development
and the assimilation-accommodation process.
We
shall find that the deficiencies in genetic epistemology for
this
case
study cannot
be removed
through more extensive
application of the theory. I shall begin by reviewing

certain notions from genetic epistemology for the analysis
to follow.
Then I shall present extensions of genetic
epistemology that are required for historical research.
Next
I
shall
discuss
in
some
detail
a genetic
epistemological scenario for atomic physics during the
period 1913-1925.
Tben, having set my terminology and
methodology, I will outline the scenario through 1927.
Children construct knowledge by actively incorporating
perceptions into schemes, and then adjusting the schemes to
fit the
situation
that is, by the assimilation-accommodation
process.
Thus,
according
to genetic
epistemology we attempt to come to grips with nature by
means of the assimilation-accommodation
process which leads
to
a
hierarchy of equilibrated
structures that are
increasingly better approximations to physical reality.
Each level is richer than the one below.
Piaget divides the
development of representative
activity into
three
stages:
sensorimotor, egocentric
representative,
and
operational.
At the end of the
sensorimotor period there are fleeting pictures of objects
from
the
world
of
sense
perceptions.
The
egocentric
representative
stage
is
divided
into
two
parts:
preconceptual thought and intuitive thought.
The onset of
preconceptual thought
is the
construction
of object
permanency which signals the start of semiotic functions.
The dependence on imagery during pre conceptual thought
decreases during the transition to intuitive thought. In
270
A. I. MILLER
the concrete operational stage operations possess full

reversibility and are internalized, but thinking is still
somewhat dependent on imagery.
In the formal operational
stage the transition from picture to image to symbol is
completed, and reasoning with images is much de-emphasized.
Needless to say, when Bohr or Heisenberg began their
research
they had
years
before entered
the formal
operational stage of thinking.
Clearly, extensions of
certain notions from genetic epistemology are required for
historical research. I suggest the following ones:
(1) The construction of knowledge by children and
physicists is not exactly the same.
By means of the
assimilation-accommodation process
the
child's schemes
develop the mathematical and physical attributes of the
levels of genetic epistemology that are the child's version
of physical reality.
Physicists deal with schemes that
possess ab initio the proper mathematical attributes to set
them in the operational stage.
(2) A theory's interpretation of new data places the

theory into a lower level of genetic epistemology.
(3)
A second definition
of
assimilation: The
application of a scheme to problems involving empirical
data,
data
from thought
experiments,
or aesthetic-philosophical commitments.
(4) With Piaget I define "structure ( ... ) as the set
of possible states and transformations of
which
the system
that actually pertains is a special case" (1972). 'Here I

propose that the relevant equilibrated structures are the
various
formulations of Bohr's atomic
physics during
1913-1927,
the
permanent
object,
reversibility,
conservation, and systematization.
The states or systems
are the wave and particle modes of matter and light. The
transformations are based on the mathematical notions in use
during the particular period of theory development. I turn

next to the genesis of atomic physics.
For 1913-1923 I will present first a historical sketch
which I will then analyze with genetic epistemology. From
1923
on I will focus on the genetic epistemological
analysis.
In 1911 Ernest Rutherford used classical mechanics to

analyze data from the scattering of alpha-particles off thin
metal foils. Rutherford concluded that the atom was a
minuscule solar system with a positively charged nucleus and
planetary electrons.
He was unable to adjust classical
physics to these data because a classical solar system atom
271
is unstable.
The reason is that the revolving planetary
electrons radiate away energy and eventually spiral into the
nucleus. But most atoms are stable, else we would not be
sitting here!
In 1913 Bohr incorporated atomic stability into a
theory in which by axiom there is a lowest orbit beyond
which a planetary electron cannot fall.
Another axiom is
that only certain orbits are permitted.
While in one of
these allowed orbits, the electron does not radiate any
energy.
Bohr referred to the allowed orbits metaphorically
as "waiting places" (Bohr, 1913). The planetary electron's
downward transition between allowed orbits results in the
emission of light.
Although the
drastically altered
classical mechanics could still be used to calculate allowed
orbits, it could not trace the electron's transition.
Consequently, in transit the electron is unvisualizable; it
disappears and appears like the smile of the Cheshire cat.
We may describe these developments
with genetic
epistemology as follows (see levels 1 and 2 in the figure).
Rutherford assimilated the scheme of classical mechanics to
his 1911 data with no attempt at adjustment. In 1913 Bohr
adjusted or accommodated the disequilibrated version of
atomic physics with its unstable atom to a new equilibrated
structure, that is, the 1913 Bohr theory. Central to the
1913 Bohr theory is the permanence of the solar system
atom.
Consequently,
at this
juncture
Bohr's theory
displays representative act1v1ty analogous to egocentric
representation.
Rutherford's 1911 conception of the atom
(the product of assimilation) is the signified. Bohr's
1913 atom (the product of accommodation) is the signifier.
The signifier or image represents the atom, but neither is
it the atom itself nor is the signifier the image of the
scheme
in
Bohr's atomic theory.
Because, as Bohr
emphasized in 1913, the images from classical mechanics were
imposed on the atomic theory through the perception-laden
interpretation of the mathematical symbols of ordinary
mechanics.
From 1913 through 1923 atomic physics problems were

analyzed with visualizable models of solar system atoms.
Indeed by 1923 the Bohr theory had become a magnificant
edifice, and these models led some very prominent physicists
to wax poetic.
For example, Max Born (1923) wrote:
"A remarkable and alluring result of Bohr's
atomic theory is the demonstration that the atom
is a small planetary system ( ... ) the thought
272
A.!, MILLER
that the laws

reflect
exercises
the
of
the macrOCQsmos in
terrestrial
world
great
indeed its form
magic
is
on
rooted
the small
obviously
mankind's
mind;
in the superstition
(which is as old as the history of thought) that

the destiny of men could be read from the
stars".
In 1923, however, the solar system image of the atom

withered away principally because atoms did not respond to
light as a solar system atom should.
A possible resolution
was in Albert Einstein's light quantum. In Einstein's first
annus mirabilis of 1905 he proposed that to understand
certain processes which involved the interaction of light
and matter, it was helpful to assume that light propagated
through space like a hail of shot or light quanta. Most
major physicists, Bohr included, resisted the notion of
light
quanta
because
it
was
cQunterintultlve.
By
counterintuitive I mean counter to what Bohr and Heisenberg,
among others,
referred
to
as
our
'customary intuition',
Customary intuition was a characteristic style of visual

thinking among many scientists and engineers trained in the
German cultural milieu. The term customary intuition is rich

in Kantian overtones.
Customary intuition is the mental
imagery that is abstracted from phenomena that we have
witnessed in the world of sense perceptions. Customary
intuition is the imagery of classical physics. According to
customary intuition light is a
wave phenomenon.
For
example, certain properties of

light like interference is
often explained by abstracting from properties of water
waves.
Then how can particulate light quanta produce
interference?
How can something be both a wave and a
particle, that is, be both extended and localized? Light
quanta were
to
be avoided
right from the
start in atomic
physics. Bohr agreed.

In order to continue to exclude light quanta from
atomic physics, in 1923 Bohr folded into his theory a
mathematical formalism in which
an atom's constituent
electron reacted to an incident light wave as if the
electron were represented by as many harmonic oscillators as
there
are
atomic
transitions.
In
this
way
the
characteristics of the
scattered
light
wave can be
correlated properly with the incident one.
Although we may
visualize an atomic oscillator to be a billiard-ball like
electron attached to a spring, there are so many possible
transitions that a constituent electron is neither localized
273
nor visualizable. Consequently in the post-1923 Bohr atomic

theory, the electron's image is no longer imposed on the
theory by customary intuition but is given by the theory or
scheme.
According to genetic epistemology, during 1913-1923
the thinking of Bohr and most other physicists focused on
visualizable models and so emphasized accommodation over
assimilation - that is, a phase of thought that Piaget calls
'imitation' (see levels 2 and 3 in the figure). Moreover,
since their thought was centered on light as a wave, their
assimilation
of
data
was
incomplete.
Similarly
accommodation was incomplete owing to its limitation to
waves. These are characteristics of preconceptual thinking.
Decentering begins with the formation of a structure that is
an image not of an object, but of a scheme.
This was the
case with the assimilation of the 1913 theory t.o the
requirement
of avoiding light
quanta.
The resulting
accommodation led to the 1923 oscillator version of Bohr's
theory in which a constituent electron was represented by a
very large number of oscillators. The next portion of
egocentric
representation
"intuitive thinking".
scenario.
is
what
Piaget
Historical data
refers
to
as
yield the following
Late in 1923 data became available that strongly

implied the reality of light quanta.
Bohr's response was a
physics of desperation.
In 1924
he incorporated or
assimilated
these
data into his
atomic theory with
accommodation to a level more abstract than previously (see
levels 3 and 4 in the Figure). In the 1924 Bohr theory the
oscillator representation of a bound electron is referred
to as 'virtual oscillators'.
The virtual oscillators
emitted
waves
that carried
probabilities for atomic
transitions, in addition to the usual quantities transmitted
by radiation.
Bohr included probability with its inherent
discontinuity in
place of light quanta.
A stunning
prediction was the non-conservation of energy and momentum
in the individual atomic processes.
To most everyone's
relief, in early 1925 this prediction was empirically
disconfirmed.
In response to the disconfirmation
accommodation of the 1924 Bohr theory to

mathematical structure based on virtual
emitted only waves of radiation and not
levels 4 and 5 in the figure).
there was
a sophisticated
oscillators that
probability (see
A.I.MILLER
274
*
*
r
r
VERSION PREFERRED BY HEISENBERG

(1928)
~ 2nd QUANTIZATION
QUANTUM MECHANICS (1927)
~
THOUGHT EXPERIMENTS
QUANTUM MECHANICS (1925-1926)
HEISENBERG INVENTS THE NEW ATOMIC PHYSICS - QUANTUM MECHANICS

BOHR (1925)
r
r
r
*
DISCONFIRMATION OF BOHR (1924)
BOHR (1924)
~
DATA SUPPORTING LIGHT QUANTA
BOHR (-1923)
~
AVOID LIGHT QUANTA
BOHR (1913)
~
RUTHERFORD I S DATA
CLASSICAL MECHANICS
Figure 9. The figure depicts the assimilation-accommodation

process in the dynamics of scientific progress in atomic
physics during 1913 through 1928.
The asterisks indicate
object permanence (level 2), systematization, conservation
and reversibility (level 7 for Bohr while Heisenberg
constructed conservation in level 8).
275
Nevertheless many physicists had yet to accept fully

the light quantum's reality, once again owing principally to
its
counterintuitiveness.
Take,
for
example,
the
widely-read pa~er on the interaction of light with atoms
written in sprlng of 1925 by Heisenberg and the eminent
physicist Hendrik Kramers. They mentioned light quanta, but
treated light like a wave.
We may say that the new 1925 scheme for the Bohr
theory exhibits what Piaget calls intuitive thinking because
the 1925 theory's descriptive apparatus is given by the
theory or scheme, and is semireversible because the theory
assimilates data to only light as a wave. Disequilibrium in
language (i.e., mathematical notation) serves even further
to place the Kramers-Heisenberg paper and thus atomic
physics, too, in the intuitive thinking stage. For example,
Kramers and Heisenberg used the notation or language Ve
and Va for the frequency emitted and absorbed during an
atomic transition, instead of notation such as V;k for a
transition between ~tomic states i and k. The improper
notation prevented them from realizing the deeper meaning of
representing the
atom
by
an
ensemble
of harmonic
oscillators.
As Heisenberg recalled, we "practically (had)
the matrix multiplication"(2). Besides the spectre of light
quanta, the 1925 Bohr theory was further disequilibrated by
the recently proposed wave-particle duality of matter for
which data were accruing.
Last ditch attempts by Heisenberg to
apply the
visualizable solar system models
to pressing problems
failed.
More than anyone else Heisenberg became convinced
of the far-reaching advantages of renouncing visualization
of atomic phenomena.
In June 1925 he created the scheme or
structure of a new atomic physics or quantum
mechanics.
At
its basis was the virtual oscillator metaphor of the

unvisualizable electron.
To some extent the nascent moment of Heisenberg's
invention of quantum
mechanics
squares with Piaget's
definition
of
creativity
namely,
the
"creative
imagination,
which is the assimilating activity
of spontaneity" (1962).
That
scheme to data for which the
in a state
is, the assimilation of a

scheme was not originally
meant. Briefly, whereas most everyone was concerned over the
dynamics of the interaction between atoms

and light,
Heisenberg alone realized that current problems were rooted
in not understanding the atomic electron's kinematics. So
Heisenberg assimilated a scheme that at first sight was not
A.I.MILLER
276
direct
response
to
the
interplay
between
atoms
and
radiation to
just this problem.
Among the scheme's
ingredients was the virtual oscillator metaphor of the
unvisualizable electron. Accommodation of this scheme led to

the 'unusual kinematics' that was the new atomic physics.
On the one hand, genetic epistemology helps us to

understand the gradual transformation of knowledge exhibited
in the hierarchy
1913-1925.
of
structures
On the other hand,
in
the
genetic
Bohr
theory from
epistemology often
fails to capture the nascent moment of creative thinking

such as Heisenberg's invention of the new atomic physics.
Some
flexibility
in
structure
'reflective
abstraction',
process where reorganization
formation
is
offered
by
which is the unquantifiable

of material from lower-level
structures can lead to the formation

of higher-level
structures.
But here I found useful Gestalt psychology's
notion of
an
disorganized
For example,
invariant or
conserved
field of knowledge
quantity
entering a
to precipitate creativity.
Heisenberg's realization that the new peculiar
mathematics of quantum mechanics agreed with conservation of

energy. I turn next to sketch the remainder of the genetic
epistemological scenario.
During the brief period mid-1925 through fall of 1927
problem after problem that had resisted treatment in the old
Bohr theory was solved, and several exciting and unexpected
new results emerged. Some of these results were puzzling and
misunderstood because until the fall of 1927 the new atomic

physics lacked unambiguous interpretation of its syntax.
For example, although results of calculations agreed with
empirical
data,
manipulations
neither
the
meaning
nor the newly emerging
of subatomic particles were understood.
of
intermediate
mysterious properties
This situation was
rooted in the total failure of physicists to extend the

mental
imagery of classical physics,
i.e., customary
intuition with its perception-laden language, into the
domain
of the
atom.
For although,
in June 1925,
renunciation of customary intuition in the atomic realm had
been a necessary pre-requisite for Heisenberg's invention of
the quantum mechanics, we read in subsequent key papers by

Heisenberg and Born, among others, the desire for some sort
of visualization of atomic processes.
Physicists were adrift with no anchor to the world of

sense perceptions. But they had nothing else, and continued
reliance on customary intuition was the stumbling block in

Bohr's and Heisenberg's heroic struggles at Bohr's Institute
277
in Copenhagen during fall of 1926 into 1927 to interpret the

syntax of the new atomic physics. As Heisenberg wrote in an
article of November 1926, "the electron and the atom possess
not any degree of direct physical reality as the objects of
daily experience" (1926).
On 23 November 1926 Heisenberg
wrote to Wolfgang Pauli,
"What the words 'wave' and
'corpuscle' mean we know not any more" (Pauli, 1979).
Heisenberg recalled that these struggles left Bohr and
himself in a state of 'despair'
(1975).
Indeed the
psychologist Rudolf Arnheim, in his book Visual Thinking,
writes of the "apprehension" that develops in a scientist
when the change is made to a "model of higher complexity
(when) the timeless stability of concepts, cherished by the
thinker, no longer has its counterpart in the world these
concepts describe".
According to genetic epistemology the assimilation by

Bohr and Heisenberg of data from thought experiments that
investigated paradoxes from the wave-particle duality was
partial because it was centered on one or the other scheme
of matter,
i.e.,
wave or particle,
with no reversibility.
Their accommodation was partial since an image of

particle
was
evoked.
Heisenberg's
a wave or
correspondence
and
published papers show that he persisted in focusing on the

corpuscular mode of matter and the wave nature of light.
By February of 1927 new mathematical methods enabled
Heisenberg to achieve reversibility for the mathematical
schemes of the quantum mechanics with its unvisualizable
particles and a rival version of atomic physics that dealt
with
matter
exclusively
as
waves
that
mechanics,
with its accompanying
imagery.
resisted reversibility concerning the physical
is,
wave
Heisenberg
schemes of
particle and wave; in his opinion the wave mechanics did not
discuss adequately the

physical properties
of matter
behaving like waves.
In this intuitive thinking stage
Heisenberg was led to permit the mathematical formalism of
the
quantum
mechanics
to
give
restrictions
on
perception-based symbols such as x for position and v for

velocity - that is, he realized his famous uncertainty
relations.
Unfortunately historical data are too scarce to fold
into genetic epistemology Bohr's acceptance in late 1926 of
the complete reversibility between the wave and particle
aspects of light and matter with conservation of physical
properties.
Bohr's full reversibility and conservation led
him to the complementarity principle, which is a deeper
A. I. MILLER
278
result than the uncertainty principles. Bohr had realized

that the paradoxes of the quantum theory were rooted in the
wave-particle duality of light and matter.
His genial
response was to propose that at the underlying atomic levels
both horns of the dilemma are, in fact, connected. This
'wavicle', to use a term of Arthur Stanley Eddington (1927),
is open to neither our perceptions nor
instruments.
That is, we can imagine
to
or
our measuring
perceive only
quantities that are either continuous or discontinuous but

not both.
According to complementarity, whereas the wave
particle states
are
mutually
exclusive
in an
and
experimental arrangement, they are complementary because
both are necessary to fully describe an atomic entity. In
other words, depending on how you choose to measure an
atomic entity, then that is what it is. For example, if the
experimental arrangement is set up for
wave phenomena, then
the wave side or mode of the atomic entity is 'observed',

i.e.,
experimentally
measured.
Another
aspect
of
complementarity is that there is an essential difference
between the pictures from customary intultlon of how a
system develops in space and time and the actual course or
behavior of atomic phenomena. According to complementarity
the
space-time
pictures
of
customary
intuition
must be
interpreted only as restricted metaphors.

Thus, the assimilation by Bohr and Heisenberg of the
1925-1926 structure of the quantum mechanics to their
thought experiments resulted in accommodation to the 1927
structure (see levels 6 and 7 in the figure). Bohr achieved
reversibility
and
conservation.
Systematization
was
constructed
because the 1927
structure possesses the
mathematical formulation and physical interpretation that we
use today for non-relativistic quantum
in
1927,
Heisenberg
achieved
mechanics.
only
However,
reversibility
mathematical schemes of the wave and quantum mechanics.
of
The reason is that at first complementarity did not

completely
satisfy
Heisenberg
due
in
part
to
a
conceptual-physical point - namely, how the discreteness of
electric
charge
emerged
from
the
electron's
wave
description.
This problem was solved in 1927-1928 through
the invention of a far-reaching mathematical prescription of
transforming the wave
and
particle
states
of
matter and
light into one another while these two states remain

mutually exclusive.
This procedure became known as 'second
quantization'.
To Heisenberg second quantization expressed
the symmetry between wave and particle descriptions that was
279
inherent
in
Bohr's
complementarity
principle,
and
thus
conservation too.
This leads us to suggest that in 1928
Heisenberg assimilated the 1927 structure of the quantum
mechanics as he understood it to second quantization (see
levels 7 and 8 in the figure). "The resulting accommodation
was to the
1928
structure within which he achieved
conservation.
So in atomic physics conservation transcends the item

in Piaget's theory of mental development. Whereas the child
realizes that a piece of clay maintains its amount of
substance when it is rolled into longer pieces, in the
atomic domain the electron maintains its charge, mass,
momentum, wavelength and spin, whether it is a particle or
wave, and whether these properties are irnageable.
The hierarchy of structures. in the figure invites us
to paraphrase Piaget's famous statement from his classic
book, Biology and Knowledge, that "life is essentially
autoregulation", as "science is essentially autoregulation".
We have identified parallels between structures in atomic
physics
and
structures
in
genetic
epistemology.
Systematization is strikingly illustrated in the thinking of
Heisenberg who continued to seek in
the mathematical
formulations of atomic physics total independence from
It is noteworthy that in Heisenberg's
customary intuition.
case
conservatl0n
followed
systematization.
Another
shortcoming in this genetic epistemological analysis is that
throughout the genesis of atomic physics we find in sources
archival and
primary
the longing of
scientists for
visualization of physical phenomena. In fact, it turned out
that Heisenberg restored imagery to atomic physics, in
contrast to genetic epistemology's description of the formal
operational period.
A hallmark of a fruitful scientific
theory in the 20th century is some sort of mental imagery.
Widening the range of application of the fundarnent~l
hypothesis of genetic epistemology to the genesis of a
complex theory (in this case the new atomic physics) that
emerged from theories that were already highly developed and
abstract necessarily brings up the hotly-debated problem of
going beyond formal operational thinking (3). For example,
although mathematically the formal structure of the special
theory of relativity is the same as that of pre-relativistic
electromagnetism, the meanings of basic quantities such as
space and time are different.
Historical studies have
revealed that these different meanings did not emerge in a
strictly logical manner that can be explained by the INRC
280
A. I. MILLER
group of the formal operational stage. Even more striking is

the case of atomic physics whose
syntax (logic) and
semantics (representation) differ strikingly from those of
lower level pre~quantum theories.
In the genesis of both
theories extra~logical grist for the mill of assimilation
was
required
from
the scientist's environment.
But this
grist does not find its proper place in a theory that is

rooted essentially in logic as is genetic epistemology.
We may widen the scope of genetic epistemology by
proposing that it describes the dynamics of scientific
progress
as driven
by
the
upward
spiral
of the
assimilation~accommodation
process.
The
result
is a
hierarchical series of structures in which equilibrium is
achieved only in part. Is it the case that the brain is an
open system in search of equilibration which the enquiring
mind never fully achieves?
If the level
of formal
operations was final then there would be no scientific
progress to theories whose
logic is richer than
either the
classical logic of Piaget's formal operational level or,

said otherwise, the content of the previous lower~level
theory.
The mystery of creative scientific thinking is how
richer
structures or theories
emerge from less
rich ones.
Further probing of this problem necessarily requires the

sort of interdisciplinary collaboration which conferences
such as this are so important to bring about.
In conclusion, an historical case study combined with

psychological analyses
show the necessity for combining
logical reasoning with mental imagery.
It is my hope that
analyses such as this can shed further light on the profound
nascent moment of creativity.
It is apropos to conclude
with an observation of Jean Piaget (1970b):
"A great deal of work remains to be done in
order to clarify this fundamental process of
intellectual creation, which is found at all
levels of cognition, from those of earliest
childhood to those culminating in the most
remarkable of scientific inventions.
1I
NOTES
1. This essay further explores issues from my recent book,
Imagery in Scientific Thought: Creating 20th~Century Physics
(1984) .
281
2. From an interview with W. Heisenberg on 13 February

1963 for Archive for History of Quantum Physics.
3. E.g., see M.L. Commons, F.A. Richards, and C. Armon
(eds.), (1984); and A.I. Miller (1986).
SENSORIMOTOR EMERGENCE:
PROPOSING A COMPUTATIONAL "SYNTAX"
Claude Lamontagne
School of Psychology
University of Ottawa
1. INTRODUCTION
Within the general context of the Conference's theme as
sketched by Callebaut and Pinxten in their position Paper,
the present paper belongs to this set of papers which can
be said to provide elements for an answer to the question of
the relevance of the Piagetian perspective as a means of
tackling some of the central issues raised by Evolutionary
Epistemology.
In this line of thought, the
paper's
contribution is 'seen' as being twofold. Firstly, the
paper stresses a key limitation in
Piaget's proposed
solution to the problem of 'genesis', or 'development', a
limitation which, although fully acknowledged by Piaget
himself and by a number of 'classical'
as
well as
'neo'-Piagetians (see e.g. Pinard, 1981), still seems to
have gone totally unnoticed by many, namely the failure to
define convincingly the process upon which 'development'
itself
critically rests:
the process of 'abstraction
reflechissante' (from now on referred to as 'reflective
abstraction').
And secondly, the paper proceeds to present
the
first
steps
of an
attempt
to
alleviate the
above-mentioned limitation, an attempt to bring to focus the
notion of 'reflective abstraction' in the specific context

of a theoretical encounter,
'of a computational kind', with
the neurophysiology of sensorimotor knowledge systems. The

body of the paper will of course be largely devoted to this
latter aspect of the contribution, which points to what we
have to offer in terms of original research results; the
former aspect of the contribution, which is limited to
stressing an interpretation of already reported research
results (by Piaget and his collaborators), will be dealt
with
as a matter of
context-setting,
as upcoming and only
sub-section of the current introductory section.

283
W Callebautand R. Pinxlen reds.), Evolutionary Epistemology, 283-3JO.

C. LAMONTAGNE
284
1. "REFLECTIVE ABSTRACTION": PIAGET' S ACHILLES HEEL

The relevance of Piaget's work for Evolutionary Epistemology
is, in our view, bound to lie mainly in its genuinely
"diachronic" aspects, i.e. those aspects which address the
issue of the "unfolding" of the more and more encompassing

knowledge systems which are said to punctuate the course
of cognltlve ontogeny, as opposed to those aspects which
address more specifically the issue of the nature
of each
knowledge system,
or "cognitive state of
equilibrium",
as
such.
Piaget has, no doubt, given quite a bit of attention
to these former, "diachronic", aspects of the problem:
"It is evident that a structural analysis of
this kind needs a complement: a model that
accounts for change. Piaget does not limit
himself to the framework of the structural
analysis
of equilibrium
states;
his main
interest is in the transition from one state of
equilibrium to the next, that is to say in
mechanisms of transition between
the older
structures and the
new ones.
It is this
qualitative
jump that is
central
in his
preoccupations ... " (Inhelder et aI., 1977, p.7)
However, a major problem arises from the fact that it seems
extremely difficult to achieve
clear understanding of
exactly how Piaget followed up on
this
concern for
'mechanisms' through which more global cognitive structures
emerge from more local ones: explicit theoretical constructs
are rather scarce.
One key notion nevertheless stands out
in
the
conceptual
'constellation'
inhabiting
this
problem-space, that of 'reflective abstraction':
"Already in 1950 the author insisted ( ... ) on
the
necessity
to
distinguish
between an
abtraction
referring
to
objects
and
a
'reflective abstraction'. The latter originates
in the actions or operations of the subject and
transfers to a superior level what is gathered
in
an inferior level
of activity: hence,
differentiations generate at the superior level
new and generalizing compositions." (Piaget et
aI., 1977, P.5)
285
SENSORIMOTOR EMERGENCE
"Let us recall ( ... ) that both empirical

and reflective abstractions exist at all levels
of development, starting from the sensori-motor
and organic levels all the way up to the highest
forms of scientific thinking." (ibid., p. 319)
Unfortunately, the otherwise rather appealing intuitive
efforts at accounting for cognitive development through the
'mechanism' of 'reflective abstraction' never led to an
explicit formal model, as acknowledged by Piaget himself:
"It certainly is easy in these instances to
speak of syntheses or new
combinations of
established components. But a notion such as
reflective abstraction can only be valuable if
one substitutes its vague formulas by a detailed
model. We did not realize this as yet ( ... )."
(ibid., p. 113)
ACknowledging this state of affairs leaves one with the
task of replacing 'with a detailed md model', the 'vague
formulas' called upon by Piaget and his collaborators to
account for 'reflective abstraction'.
2. SENSORIMOTOR EMERGENCE
From the very onset of our fascination for the Piagetian
endeavour, it had appeared that the 'evolutionary' principle
underlying the primary
concern for "how more global
cognitive structures emerge from more local level ones" had
not been convincingly brought to bear on the most primitive
(hence most fundamental) cognitive structures of all: the
perceptual and motor
ones. Consequently, we had set out to
tackle this task, and some ten years were devoted to
seeking a satisfactory account of the 'nature' of perceptual
and motor systems, with a clear emphasis, throughout the
first
seven years,
on
perceptual
ones, more specifically
through an almost exclusive concern for the visual system.

Right from the start, our endeavour had departed from
'mainstream'
piagetian
research
on
two
extra
counts:
firstly, we had chosen the computational 'formalism' as

ultimate 'linguistic' reference (i.e. as means of rigorously
expressing our theory), and secondly, we had decided to tie
our theorizing about perception and motor control
to a
serious concern for the neurophysiological substrate; our
computational theorizing therefore took, throughout the
endeavour,
the form
of "neural network
design",
in
most
C. LAMONTAGNE
286
cases brought to the stage of computer simulation,

to
concern
for
neurophysiological
and tied
plausibility.
Significant results obtained over these years were reported

in Lamontagne (1973, 1974, 1975, 1976), Lamontagne & Howe
(1980), and Lamontagne & Beausoleil (1982, 1984).
After these ten years, we felt that we had come to a
sufficiently
articulate expression of
the
nature of
perceptual and motor 'knowledge systems' to turn explicitly
to the question of their 'evolutionary filiation', or
'emergence'.
Some four years have now been devoted to this
question, in the explicit context of the 'conceptual crisis'
triggered by the 'reflective abstraction affair', and the
present paper is the first progress report on the matter.
Throughout the exercise, however, the main concern for
perceptual and motor emergence as such has had to share our

energies with a secondary (albeit certainly not of lesser
fundamental relevance to the issue) concern for bringing
our conception of perceptual and motor knowledge systems to
an ever
more formalized state.
The presentation, in
sub-section 1.2, of those of our results which are more
specifically related to emergence as such will therefore be
preceded by the presentation, in sub-section 1.1, of those
of our results which represent the current state of the very
language which we use for talking about perceptual and motor
knowledge systems in our attempt at tackling the issue of
their emergence.
2.1. A computational expression of the nature of the
neurophysiology of perceptual and motor knowledge systems
The concepts presented in this section are meant to
bear on
perceptual and motor knowledge systems in general, i.e.

they are meant to cover what all such systems share.
However, it might be important to note that throughout the
course of our work on these concepts, one single highly
specific system has constantly served as primary reference,
namely the visual line segment detection system which is
fully described and presented as a theory of the physiology
of the visual cortex by Lamontagne and Beausoleil (1982).
This is due to the fact that this particular perceptual
knowledge system was by far (and still is) the most detailed
and complete one at hand (in terms of neurophysiological
plausibility and computational efficiency).
287
Let- us start by ~oting that, in our view, the

fundamental problem aris1ng from a1ming at 'explaining'
perceptual and motor experience through neurophysiology lies
essentially in the fact that state of the art knowledge in
the neuroanatomy and neurophysiology of perception and motor
control leaves us with no other choice than acknowledging
'sensory receptors' (retinal rods and cones for vision,
cochlear ciliated cells for hearing, primary and secondary
muscular receptors for kinaesthesia, etc.) and 'motor units'
('alpha' neurones and associated muscle fibers) as primary
'substrative' support; the problem arises from the sharp
contrast opposing
the acutely local
nature of these
'organism-environment interface neurophysiological units'
and the highly global nature of the perceptual and motor
'experience' which they so compellingly seem to'support'.
How could one not
be struck,
for instance, by the
considerable relative
globality
of
the psychological
experience of perceiving a familiar face, or of acting upon
an intention of executing a somersault, in comparison with
their so-called 'substrative' neurophysiological expression
at the level of the interfacing with the environment where
neural
events are
taken to
'specify' various light
amplitudes
and
frequencies in
and various
regards the former,
various retinal loci, as

contraction amplitudes and
rates in various muscular loci, as regards the latter?

Therefore, tackling our problem essentially implies
conjecturing (i.e. proposing theories) about how the neural
medium can
manage to set
local
events in critical
relationship with global events.
We shall then look first,
in Section 1.1.1, at how one could talk, in computational
terms,
about the most 'primitive' expression of the nervous
'substrative' support of perceptual and motor experience,

locus of those events which are the most acutely local:
'sensory receptors'
for perception,
and
'motor units' for
motor control. Once this basic terminology is established,

we shall move on, in Section 1.1.2, to the question of how
one could
talk,
still in computational terms, about
'setting'
more local neural events in critical relationship
with more global ones.
288
C. LAMONTAGNE
2.1.1 Talking about the most local neural events:

Informational Interface Modalities (11M's), and lIM-bound
neural event diversity
'Interfacing' an organism and an environment first calls

for energetic mediums.
There can be as m~ny facets to
interfacing as there are types of energet1c mediums to
which the organism can react, and types of energetic mediums
to which the environment can react.
In the context of the
priority which we have placed on 'knowledge systems', or
'information systems', the 'structural' aspects of interface
energetic mediums (i.e. their actual setting within the
continuums of space, time, and quantity) will of course
take precedence over their sheer 'capacity of doing work',
as key characteristic of the interaction. 'Sensory' and
'motor control' organs are then but those 'parts' of the
organism
which
embody
this
'structure-oriented', or
'informational', interface: sensory organs correspond to
those parts of the organism which can be affected by
'environmental'
energetic
structures,
and
the motor
apparatus corresponds to those parts of the organism which
can affect 'environmental' energetic structures.
The ability of 'being affected by' and the ability of
'affecting' the structure of whatever type of environmental
energetic medium seem to call for the very same basic
feature in any organism's 'informational interface layout
structure', that of being primarily and essentially involved
in a process of local, or discrete, specification of the
manifestation of the type of energetic medium considered,
along some or all of its physical dimensions. Sensory
receptors and motor units constitute the embodiment of this
"local,
or
discrete,
specification
strategy",
for
perception and motor control respectively. If one takes for
instance the case of 'scotopic' visual perception, one
finds as specific 'sensory interface' the retinal set of
'rods', whose task it is, in the chosen perspective, to
discretize the luminous energetic medium in three ways: (i)
discretize luminous spatiality into 'positions of light
occurrences',
(ii)
discretize
luminous
temporality
into
'moments of light
occurrences',
and (iii) discretize
luminous quantity into 'amplitudes of light occurrences'. Of
course, these dimensions of the environmental energetic
medium do not need to be specified, in the receptor (or
nervous)
domain,
through
an
identical
physical
dimensionalization: if positions of light in space, in the
case
289
of visual perception,
do translate
into positions of
neural events (through the spatial layout of the receptors,

i.e. the retinal space), light amplitudes seem to translate
into frequencies of nervous events (the higher the amplitude
of light stimulating a retinal receptor, the
higher the
'firing' rate of the receptor).
The only need
is one of
specificity for the 'dimensional interfacing':. each chosen
dimension
of
the environmental energetic
medium concerned
has to be paired in a unique and unambiguous way with some

specific
set of neural
units generating specifically
dimensionalized
neural
events,
along
whatever particular
physical dimension of the 'organismic' neural domain. This

constraint will be called the 'principle of dimensional
interfacing specificity'.
On the basis of the above discussion, let us now try
to specify more precisely the nature of informational
interface structures, and let us try to do it by associating
them with the following three (3) particulars: (1) specific
polarity of informational interfacing (p), (2) specific
environmental energetic medium implied (m), and (3) specific
feature set of informational interfacing pointing to the
specific physical dimensions exploited and their respective
discretization
scales (f).
By
'specific polarity of
informational interfacing' (p) we mean the specification of
the 'directionality' of the interfacing: if it is meant to
make the organism affect the environment, it will be
labelled 'output', and if it is meant to let the organism be
affected by the environment, it will be labelled'input'.
By 'specific environmental energetic medium implied' (m), we
mean the form of environmental 'matter/energy' compound
(photonic,
sonic,
etc.) which is 'mobilized' by the
particular interfacing.
And by 'specific feature set of
some informational interfacing pointing to the specific
physical
dimensions
exploited
and
their
respective
discretization scales' (f), we mean the set of descriptors
which represent the respective physical dimensions of the
chosen 'type of environmental energetic medium' upon which
the discretization process is applied, together with each
dimension's 'discretization scale', in terms of a statement
on its kind and extent.
Characterizing
specific
informational
interface
structures, which we shall refer to, from now on, as
"Informational Interface Modalities" (IIM's), then becomes a
matter of specifying 3-descriptor expressions of the form:
IIM:
{p, m, f}
C. LAMONTAGNE
290
where the {p, m, f} set of descriptors

of three elements respectively extracted
three sets:
-The set of POLARITIES (P)
P : {Pl' P2}
{Input, Output}
-The set of energetic MEDIUMS (M)
M : {ml'
is any combination
from the following
. , m k }
{Photonic, Sonic, Muscular, ... etc.}

-The set of dimensional interfacing FEATURE-SETS (F)
F: {fp ... , f j , ... , f n }
where fj is any set of features characterizing a
unique (dsj-ss j ) pairing
-of any element of the set of DOMAIN-SETS (DS):
DS : { ds 1> ... , ds j, ... , ds}
where
dS j
is
any
combination
of
(possibly
repeated} elements of the set of
CONTINUUMS (C):
C : (c l , c 2, c3)
{Space, Time, Quantity}
-and of any element of the set of SCALE-SETS (SS):
5S: {S51' .. "
sSi'
.. "
55}
where SSj is any combination of (possibly repeated)

elements of the set of SCALES (S):
S : {sl' ... , s j, ... , s}
descriptors
of the
where Sj
is any pair of scale
"standard"
form:
is
a
(standard,
extent),
where
statement
about
the
nature
of
the
unit
of the
metric implied, and where "extent" is a statement about the
number of units implied and about what part of the continuum
concerned they occupy.
For instance:
{p, m, f}: {Input, Photonic, (Positions [i],

Moment [ii], Amounts [iii])}
[iJ "positions" standing for the set of "discrete" values
extracted from the
Space
continuum
through a Scale
characterized by a standard of 20" of arc and an extent
of 180 over two dimensions,
[ii] "Moment" standing for the set of discrete values
extracted from the
Time
continuum
through
a Scale
characterized by a standard of 1ms and an extent of 1
moment and
291
[iii] "Amounts" standing for the set of discrete values

extracted from the Quantity continuum through a Scale
characterized by a standard of
candella per m2 and an
extent of 13
log units located between
10- 6 and 107
candellas per m2 in the "luminance" continuum
would be a typical characterization of a 'visual'

interface modality which happens to be somewhat akin to the
human one.
Now, since 11M-bound neural events are critically

linked to 11M's characteristics through the "specificity of
dimensional interfacing"
principle, IIM terminology is
easily transposed into neural event terminology_ Indeed,
whatever the physical dimensionalization of the organism's
'interfacial' neural domain happens to be, the neural events
taking place within it have to be paired "in a unique and
unambiguous way" with those events which characterize the
11M's discretization of the environmental energetic medium,
as specified by the IIM's (p, m, f) set of descriptors_
Referring to the example provided above (i_e. the 'human
type' visual 11M, where {p, m, f}: {Input, Photonic,
(Positions, Moment, Amounts)}), it should be clear that the
neural events associated with this 11M actually specify
'incoming light', in a variety of 'positions' and 'amounts',
and at a given 'moment'; this means in fact that one can

freely refer to the neural events associated with this 11M
as being of the type {Input, Photonic, (Positions, Moment,
Amounts)} and that, more generally, one can refer to the
neural events associated with any 11M as being of the {p, m,
f} type which characterizes the particular 11M. On this
basis, what can then be said about the potential variety of
11M-bound neural events?
Firstly, it is clear that there will be, within any
organism, as many different types of 11M-bound neural events
as there are different 11M's, i.e.
different
{p, m,
f}'s: we shall call this potential source
of neural
event diversity the "inter-modality variety", stressing,
however,
11M's
that in order to 'qualify'
must
differ in
a minimum of
as different,
only
one
any two
aspect
of
only one descriptor (p, m, or f), leaving ample room for

partial inter-modality 'identity'.
And secondly, it is also clear that there will be,
within any specific 11M, as many different types of
11M-bound neural events as the feature-set characterizing
the particular 11M (the f descriptor of the {p, m, f} set of
descriptors)
allows
for,
the
actual
variety
of
those
292
C.LAMONTAGNE
" intra-modali ty" rIM-bound neural events resting essentially

on the respective extents of the scales making up the
scale~set (ssi )
which, paired with the domain-set (ds i ),
const1tute this feature-set (or f) descriptor. In fact,
there will be exactly as many possible such "intra-modality"
11M-bound neural events as there are discretization units
implied by particular 11M's f descriptors: this corresponds
to the sum of all discretization units allowed by each and
every scale specified by the particular 11M's feature-set
(or f) descriptor. For instance, referring again to our
{Input, Photonic, (Positions, Moment, Amounts)} example,
there are,
in this case, as many different possible
"in tra-moda li ty"
11M-bound neural events as there are
different possible Positions, different possible Moments,
and different possible Amounts, all added up together:
indeed,
any 'unique'
combination of
Position
value, a
Moment value, and an Amount value (combination which one

might call a (p, m, a)) specifies a unique possible neural
event within
the neural domain
associated with this
particular 11M. However, it should be noted (as was done in
the
case
of
inter-modality
diversity)
that
this
intra-modality neural event diversity requires in no way
that in order to qualify as different two neural events be
characterized
as different
in
all
aspects
of all
descriptors: in order to qualify as different, any two
neural events must differ in a mlnlmum of only one
"unit-value" of only one scale of the scale-set (ssi)'
leaving ample room for "partial intra-modality identity".
For instance, referring once more to the {Input, Photonic,
(Positions, Moment, Amounts)} example where intra-modality
diversity
has just
been associated with (p,
m, a)
value-combinations,
a large quantity of intra-modality
different possible neural events
are characterized as
sharing identical a-values (Amount-values) and m-values
(Moment-values) for instance, variety being confined to the
positions domain.
Now these latter concepts of "types of variety and
identity" within pools of inter-modality and intra-modality
IIM-bound neural events bring us to the "door" of our target
issue of talking about "tying more local neural events with
more global ones": let us turn to it.
293
2.1.2. Talking about setting more local neural events

in critical relationship with more global ones: Epistemic
Entities (EE's), Essential Characterization (EC), Primary
Differential
Characterization
(PDC)
and
Secondary
Differential Characterization (SDC)
First of all, let us acknowledge the fact that lIM-bound
neural events,
whatever they are, represent, for the
organism within which they occur, the "knowlege", in its
most primitive or local form, which this organism has of its
relationship to the 'environment'.

We shall acknowledge
this by
calling
them
"Level-O
Epistemic Entities"
(O:EE's), calling "Level-1 Epistemic Entities" (1 :EE's)
'immediately' more global Epistemic Entities, and we shall
define the
"setting
of
O:EE's
in
critical computational
relationship with 1:EE's" as

the exclusive link-up of any
given "type" of subset of 0:EE'5 with single 1:EE'5, a
'type' of subset of O:EE's being defined as specifying

any identifiable criterion of relationship between single
O:EE's and any constant number of other O:EE's.
Now
as
O:EE's
are
specified
through {p,m,f}
descriptors, criteria of relationship IIbetween single 0:EE'5
and any constant number of other O:EE's" can be readily

extracted from the particular values of these descriptors,
whether one is considering one or many 11M's. In the case of
the by now familiar example of the {Input, Photonic,

(Positions,
Moment,
Amounts)}
11M,
such
types
of
(necessarily intra-modality) subsets of O:EE's would be, for
instance,
amount
",
the type "adjacent position /
same moment /
or the type "adjacent position
same
same moment /
adjacent amount", These two examples in fact also illustrate
quite well an extremely important possible consequence of

any "mapping" of O:EE's onto 1:EE's, which is that in most
cases it involves a variety of 1:EE's, which opens the door
to further local-global ties where 1:EE's become "more
local" than potential (more global) 2:EE's, which in turn,
in as much as they also still present

diversity, can assume the role of "more
some "residual"
local" EE's in
possible ties with (more global) 3:EE's, and so on and so

forth. Indeed, the above mentioned "adjacent position / same
moment I same amount" type of subset of O:EE's, given the
ranges of values within which Positions, Moments and Amounts
were defined (by the f descriptor),
opens onto vast
quantities of 1:EE's determined firstly by the fact that for
Moments and Amounts, the mapping does not exploit Level
294
C. LAMONTAGNE
variety and therefore transfers it without any modification

into Level-1 variety (i.e. 1:EE's will be characterized by
as many Moments and Amounts as O:EE's), and secondly by the
fact that for Positions, the mapping creates a new variety
which points to "positions of
adjacent
positions of
simultaneous
light occurrences of identical amplitude", new
variety which will nonetheless be very similar to the Level

Positional variety (i.e. 1:EE's will be characterized by
almost as many "positional adjacency positions" as there are
Positions
characterizing O:EE's).
This
leads to the
recognition of the necessity of generalizing our terminology
about O:EE's to EE's of any level, and of generalizing our
terminology about "setting O:EE's in critical computational
relationship with 1:EE's" to "setting n:EE's in critical
computational relationship with n+1:EE's".
In general, then, we shall say that "more local" EEls
are set in critical computational relationship with "more

global" ones when a neuronal "effective decision procedure"
is specified,
which establishes a
one-to-one exclusive tie
between instances of some given type of subset of n:EE's

(referred to as "the more local ones") on the one hand, and
single n+1:EE's (referred to as "the more global ones") on
the other, a "type" of subset of n:EE's being defined as
an exclusive pointer to any identifiable criterion of
relationship between single n:EE's and any constant number
of other n:EE's. We shall then also say that EE's of any
level will be characterized by a set
of descriptors
{Descriptor1' ... , Descriptor i' . , Descriptor n }, which,
for O:EE's, takes the particular form {p, m, f} (i.e.
{polarity, medium, feature-set}).
Establishing a local-global tie therefore essentially
rests on:
-specifying, for any given set of n:EE's, a 'type' of subset

of n:EE's; and
-linking up the subsets of n:EE's with as many single
elements of a new set of EE's, labelled n+1:EE's.
Finally, putting the emphasis on the nature of the
'more global' EE's (i.e. n+1:EE's) which this procedure
generates,
we shall say that
the procedure actually
'creates'
n+1:EE's through a two-fold CHARACTERIZATION
process:
-firstly, in specifying the "type" of subset of n:EE's
which will be mapped onto n+1:EE's, the procedure actually
determines the very essence of the n+1:EE's as Epistemic
295
entities, it provides their very 'meaning', and we shall

call this aspect of the process "ESSENTIAL CHARACTERIZATION
(EC) of n+1:EE's";
-and secondly, in linking up the various subsets of
n:EE's (of the specified 'type') with specific single
n+1:EE's, the procedure actually determines the extent of
the diversity of n+1:EE's,
along "feature dimensions"
formally
specified
through their {descriptor 1 "
descriptorj, ... , descriptorn} set of descriptors, where any
descriptor is either CREATED (if it bears on some aspect of

the diversity characterizing the 'relationship' between
n:EE's implied by the given 'type' of subset of n:EE's) or
simply CARRIED OVER from n:EE's' set of descriptors (if it
bears on some aspect of the identity characterizing the
"relationship" between n:EE's implied by the given "type" of
subset of n:EE's): we shall call the overall form of this
aspect
of the
characterization
process "DIFFERENTIAL
CHARACTERIZATION"
(DC),
calling
"PRIMARY DIFFERENTIAL
CHARACTERIZATION" (PDC) its more specific expression in the
case of "created descriptors",
and calling "SECONDARY
DIFFERENTIAL CHARACTERIZATION"
(SDC) its more specific
expression in the case of "carried over descriptors".
Applying this new terminology to our usual example

of the {Input, Photonic, (Positions, Moment, Amounts)} 11M,
and the intra-modality "adjacent position / same moment /
same amount" critical "type" of O:EE's subset, 1:EE's will
be said to be
-"ESSENTIALLY CHARACTERIZED" by this very 'type' of
subset of
O:EE's on which they critically rest, and which
happens to point to each and every subset of O:EE's
consisting of anyone O:EE and of all those other O:EE's
which share with this one O:EE the very same Amount value
and the very same Moment value, while they differ from it as
to their Position value, a difference which has to be kept
within the range of possible 'adjacencies', and
-"DIFFERENTIALLY CHARACTERIZED" , "PRIMARILY" by the
'new' variety of 'positions' introduced as a consequence of
the pooling of O:EE's under the 'type of subset' criterion
of positional adjacency, giving rise to an 'adjacency
position' descriptor in the 1:EE's' set of descriptors
(based on, but considerably differing from, the 'position'
descriptor
in O:EE's'
own set of
descriptors), and
"SECONDARILY"
by the former variety of 'amounts' and
'moments',
introduced
required along these
as
consequence
dimensions
by
the
of
'type
the identity
of subset'
296
C.LAMONTAGNE
criterion, giving rise to an 'amount' descriptor and a

'moment' descriptor in the 1:EE's' set of descriptors
(directly carried over from the 'amount' and 'moment'
descriptors in O:EE's' own set of descriptors); overall,
then, this calls for a 1:EE's set of descriptors of the
form: (Adjacency-Position, Moment, Amount).
In short, our effort at expressing in computational
terms the nature of the neurophysiology of perceptual and
motor knowledge systems first led to locate the ultimate
roots
of the
problem
in what came
to
be called
"Informational Interface Modalities", or lIM's, specified in
terms of {p, m, f} sets of descriptors (i.e. a polarity
descriptor,
an
energetic
medium
descriptor,
and a
feature-set
descriptor).
Focus was
then turned onto
11M-bound
neural events,
upon which the {p,
m, f}
specification scheme was applied, leading to the concepts of
inter-modality and intra-modality
VARIETY of 11M-bound
neural events. The neurophysiological task (expressed in
computational terms) of perceptual and motor knowledge
systems having been acknowledged as being essentially one of
setting more local neural events in critical relationship
with more global ones, and 11M-bound neural events having
been recognized as most local neural events, referred to as
"Level-O Epistemic Entities" (or O:EE's), Perception and
Motor Control were then viewed as essentially involved in
processes of
-Essential Characterization, where the instances of
some given "type" of subset of Level-n Epistemic Entities
(or n:EE's) are tied to single Level-n+1 Epistemic Entities
(or n+1:EE's), starting with n=O (i.e. O:EE's), and
-Differential Characterization, where varieties of
Level-n+1 Epistemic Entities (or n+1:EE's) are specified
through a set
of Primary
and Secondary Differential
Characterization
descriptors,
starting
with
Level-1
Epistemic Entities (or 1:EE's).
As a final note before turning to the issue of
emergence,
let
it
be
stressed
that
the
strong
uni-directionality of the "logic"
of local-global tie
setting
is
totally
independent
of 11M's' polarity
descriptors (Input or Output): the actual "flow of nervous
activity" within a neural system has, in principle, nothing
to do with local-global ties as such, and should therefore
be considered a totally separate issue.
297
2.2 A computational expression of the nature of

emergence in neuro-physiologically embodied perceptual and
motor knowledge systems
Given
Section 2. l' s "picture" .of the
of perceptual and motor neural networks,
structural filiation
from Informational
Interface
Modalities
(11M's)
onwards,
an
obvious
genetic filiation
pattern 'emerges': neurophysiological
'evolution' of perceptual and motor neural networks has no
choice whatsoever but to start with
11M's and their
associated 'knowledge units', Level-O Epistemic Entities
(O:EE's), and to proceed step by step, over the highly
ordered succession of EE-Ievels achieved by setting current
level EE's (or 'more local' EE's) in critical relationship
with next-level-up EE's (or 'immediately more global' EE's)
through the
processes
of Essential
and Differential
Characterization.
In the beginning ... then ... there are Informational
Interface
Modalities,
and
their
associated
Level-O
Epistemic
Entities
(O:EE's).
Progressively, organisms
appear which access Level-1, and then Level-2, Level-3,
Level
reaching more and
more encompassing experiences
of the universe through sets of more and more 'global'

EE's, and opening the way for yet other organisms reaching
yet more encompassing experiences of the universe through
yet new sets of EE's, and so on and so forth ... !!
But how is this emergence process to be controlled?
It is one thing to 'establish' that some given perceptual or
motor system implies for instance a structural filiation
spanning some number (however large) of EE-Levels (however
fully specified in terms of their respective Essential and
Differential Characterizations);
it
is
an altogether
different matter to pinpoint a 'genetic control structure'
which could possibly have led
to such
a structural
filiation: for just about any non-trivial set of 11M's, the
number of possible 'hierarchies' of local-global ties is so
large that however generous one is prepared to be in terms
of available space and time, it will almost always seem
necessary to account for given outcomes of the genetic
process
through
specific
genetic
determinisms, this
departure from the "straightforward ways of chance", as
means of 'explanation', posing a problem whose scope is
directly proportional to the scope of the departure itself,
and whose nature has extremely little to do with that of
hierarchies of local-global ties themselves.
298
C. LAMONTAGNE
Having acknowledged the 'necessity' of reducing (or at

least ordering?) the set of possible emergence paths through
the combinatorial field associated with whatever set of
O:EE's, one must then specify the "reduction strategy". The
general principle which we adopted in our 'search for a
"reduction strategy" is that of a greater relevance of those
"n:EE local-global tie relationships" which imply greater
similarity amongst n:EE's. Two types of considerations led
to adopting this 'greater similarity' principle: the first
type points to the fundamental belief that, given the view
that
knowledge acquisition systems are
"identity seekers",
Essential
Characterization
is
but
the
conjectured
irrelevance of some dissimilarity between n:EE's, and that
it seems better to keep the extent of the dissimilarity as
limited as possible; and the second type of considerations
points to the fundamental belief that systems tend to
conform themselves to certain rules of "processing economy",
exploiting for instance, in the case of neurophysiological
systems, cell proximity (and ensuing speed of processing),

and
thereby
favoring
the
particular
physiological
relationships happenning to be weaved into this anatomical
one (of cell proximity): in our case, this happens to favor
"greater similarity
relationships amongst n:EE's"
in that,
to a large extent, EE descriptor sets organize EE diversity

along ordered dimensions, keeping "closer together" EE's
characterized with "most similar"
feature values.
In order
to allow
for efficient visualization
of this latter
argument, it might be worth introducing at this point the
"uniformization principle" which we applied in defining our
working representation
of
11M's neural
domains
and their
ramifications through local-global ties.

Our "neural domain uniformization principle" simply
stated that the physical dimensions of the organismic neural
domains implied in IIM dimensional interfacings of interest
should be confined to the spatial continuum, i.e. that EE
diversity be exclusively embodied in spatial diversity.
This principle
was adopted primarily
for easing the
research process, and the bearing of the otherwise arbitrary
choice which it implies upon the actual issue of emergence
should be understood in the context of Section 1.1.1's
presentation of the "principle of dimensional interfacing
specificity", where it is stressed that neural domain
physical dimensions are in no way "expected" to tap the same
299
continuums
as
those
tapped
by
their environmental
counterparts, and are therefore free to exploit whatever
continuums seem to offer some other type of advantage.
Concretely,
our
spatial uniformization principle
amounts to treating
11M
neural
domains
(and their
local-global ramifications) as multi-dimensional spatial
cell
arrays.
Let us
consider,
for
instance, the
intra-modality diversity implied by an 11M whose {p, m, f}
reads:
{Input, Photonic, (Positions, Moments, Amounts)}
where
"Positions"
stands
for
some
two-dimensional
orthogonal set
of
8x8
discrete
(and quantitatively
equivalent) "slabs" of space, where "Moments" stands for a
set of 2 discrete (and quantitatively equivalent) "slabs" of
time, namely moment(O) (present) and moment(-l) (immediate
past), and where "Amounts" stands for a set of 2 discrete
"slabs"
of luminance, namely amount (0) (darkness) and
amount(l) (light) (See figure 10).
In this overall context, then, "greater similarity"
amongst EE's translates into "greater proximity" of EE's,
and "greatest similarity" amongst EE's translates into EE

"contiguity". We chose, in this first phase of research, to
restrict
emergence
paths
(i.e.
local-global
ties)
exclusively to those falling within the boundaries set by
"contiguity"
relationships,
as
extreme
form
of EE
similarity, and we embodied this constraint in a formal
neural network which, expressed according to the "spatial
uniformization" principle, takes the form illustrated in
Figure 11, where Co stands for any single n:EE supporting
cell, and where (C"
C2 , C 3 , ... , Cn ) stand for Co's
contiguous n:EE supporting cells, n varying according to the
particular geometries of n:EE supporting cells.
The different thresholds introduced in the network
(the 2's and the n) in fact imply local-global ties which,
quite apart from being based exclusively on contiguity
relationships between single n:EE's and other n:EE's, are
also
based on
whatever
single
the co-occurrence (given
n:EE)
whatever
chosen
possible
to ignore,
of
all
dimension(s).
order to have the

simplest
definition of the "contiguity"
relationship,
the
contiguous
occurrence of
n:EE's
along
This measure was adopted in
possible
single
relationship: it
in the definition of
the idiosyncrasies of the
under which n:EE's can be dimensionalized.
general
made it
the "contiguity"
various geometries
300
C. LAMONTAGNE
positions
(SxS)
moments
(x2)
=~~~ii~I-;~%~~~~iI~~~l11:Q
amounts;
(x2) ~
1__
I
I
-1 L~
Figure 10 illustrates the form that the "formal" neural

domain associated with such an 11M would take.
Figure 11.
It
301
was decided to
"contiguity
primitive"
start "testing"
the
in the context of a
above~defined
perceptual and
motor emergence setting limited to two (2) 11M's: one

perceptual 11M and one motor 11M. The perceptual 11M, a
"visual ll one, was characterized as follows:
{Input, Photonic, (Positions, Moments, Amounts))

64x64
where "Positions"
discrete
(and
stands for a two-dimensional set of
quantitatively
equivalent) hexagonal
adjacent "slabs" of space,

where IIMoments" stands for a set of
discrete (and
quantitatively equivalent) "slabs" of time, namely moment(O)

(present) and moment(~1) (immediate past),
and where "Amounts" stands for a set of 2 discrete
(and quantitatively equivalent) "slabs" of luminance, namely
amount(O) (darkness) and amount(1) (light).
Figure
12
illustrates part of the "dimensional
geometry"
of
the
neural
domain
associated
with
such a
perceptual IIM.
Most critical for eventual "contiguity
relationships" is of course the hexagonal geometry of the
"positional space": for anyone non-boundary cell, it allows
for exactly six (6) contiguous cells. Two convergent types
of considerations in fact led to the choice of this
particular geometry for the "positional space": the first
type has to do with the particular ease with which the
hexagonal geometry lends itself to establishing uniform
contiguity relationships, which stands in sharp contrast
with the difficulty which other geometries,
like the
"dominant" orthogonal one, present vis--vis this same issue;
and the second type of considerations has to do with the
extremely widespread occurrence of
the
hexagonal (or
quasi-hexagonal) geometry in "natural constructs" in general
(as in flowers, beehives, etc.), and in retinal gatherings

of light~sensitive cells in particular.
As for the motor
11M,
a very primitive "arm control lt
one implying a single

articulation supporting a single
degree of freedom embodied in an agonist~antagonist pair of
muscles was opted for, and it was characterized as follows:
{Output, Muscular, (Positions, Moments, Amounts))
where
"Positions"
stands for
set
of
discrete (and
quantitatively equivalent)
portions
of
space, namely
position(1) (agonist muscle) and position(2) (antagonist
muscle) ,
where "Moments" stands for a set of
discrete (and
quantitatively
equivalent)
portions
of
time, namely
moment(O) (present) and moment(~1) (immediate past) ,
302
C. LAMONTAGNE
Figure 12.
63
~
1
Momen ts
Amounts
Positions
Figure 13
illustr ates part of the "dimen sional geomet ry" of
the neural domain associa ted with such a motor 11M.
303
Perceptual 11M:
Motor 11M:
{Input, Photonic, (Positions,

Moments, Amounts))
{Output, Muscular, (Positions,

Moments, Amounts))
or
or
{I,P, (P,M,A))
~
yielding a potential intramodality EE-diverslty
of:
~
p ri'I ali
p ri'I Ii
ri'I ii
ri'I ii
p ri'I Ii
p ri'I ii
p ri'I ii
(a)
P
P
P
P
P
P
P
ri'I ii
ri'I ii
ri'I ii
ri'I ii
ri'I ii
ri'I ii
ri'I ii
(b)
Figure 14 shows a listing of the combinations making up each

of the two
combinatorial spaces
associated with the
perceptual and the motor 11M's respectively.
single layer
portion of
Level-O EE's
(O:EE's)
single layer
portion of
Level-1 EE's
(1:EE's)
Figure 15.
C. LAMONTAGNE
304
and
where "Amounts"
quantitatively
stands for a set of
equivalent)
portions
force" continuum.
Now these two 11M's open
well
as
intra-modality
up
64
of
onto
EE diversity.
discrete (and
the
"contraction
inter-modality as
We chose to address
first the issue of intra-modality EE diversity, expressing

each 11M's associated EE diversity in terms of the potential
variations within
feature-sets.
Each
the different sub-sets
11M
was
therefore
of the respective
attributed
an
EE
"combinatorial space", made up of all possible combinations

of possible feature-related variations.
Spelling out the combinatorial space associated with each

11M's EE diversity completed the specification of the
Level-Q
neural domain onto which the "contiguity primitive ll
was to be tested.
Figures 15, 16, and 17 show the
prototypical formal "wiring" patterns obtained by applying
the
"contiguity primitive"
diversity combinations
of
to the first
the perceptual
three (3)
11M
II
p ,m,a''
(i.e. those
combinations where diversity is confined to one single

feature space): (p,m,!i), (p,m,~), and (p,m,a).
In Figure 15, the "contiguity primitive" is applied to
contiguity in the "positions" feature space (p,m,a): a

Level-1 EE supporting cell will "fire" if and only i f the
corresponding Level-O EE supporting cell (in terms of
position) AND its six immediately neighbouring (Level-O EE
supporting) cells have "fired".
In Figures 16 and 17, the "contiguity primitive" is

applied to contiguity in the "Moments" and in the "Amounts"
feature spaces respectively (p,m,a and p,m,a); in both
cases, a 1:EE supporting cell will "fire" (i.e. a 1:EE will
be actualized) if and only if the corresponding O:EE
supporting cell (in terms of moment or amount) AND its
immediately
neighbouring (O:EE
supporting)
cell have
"fired".
It should be noted, as regards the application of
the contiguity
primitive
to
the
"Moments"
feature space
(Figure 16),
that the 1:EE supporting
neural domain
(right-hand side portions of 7a and 7b) does not follow the
dimensionalization strategy adopted from the start, where
the vertical dimension was firstly split in "blocks" of time
(i.e. moments), each of which was then split in "blocks" of
quantity
reversed,
(i.e.
the
amounts):
vertical
in the present case the order is
dimension being
firstly
split in
blocks devoted to quantity, which in turn are split into

blocks devoted to time. This was simply meant to push
temporal diversity into the last (fourth) dimension, in
portion of
O:EE's
portion of
l:EE's
305
portion of
O:EE's
portion of
l:EE's
(b)
(a)
Figure 16,
portion of
O:EE's
portion of
l:EE's
(a)
Figure 17,
portion of
l:EE's
portion of
O:EE's
(b)
306
C. LAMONTAGNE
order to mark the fact that in this ~e:y case it does not
present a "genuine"
diversity, pa1r1ng moment(O) with
mom:n~(-l) appearing as strictly equivalent (given identical
pos1t10n and amount), in terms of outcome, to pairing
moment(-l) with moment(O).
It should also be noted, this time regarding the
application of the contiguity pr1m1t1ve to the "Amounts"
feature-space (Figure 17), that if the particular. wiring
implied represents a formal
possibility, it does not
represent
an actual
possibility. Indeed, the co-occurrence
of EE's which differ only as to their amount-value is

actually impossible, because there simply cannot be two
different amounts of "light" detected at the same moment in
the same position.
This particular application of the
contiguity primitive is therefore "doomed" right from the
start! !
Full-scale w1r1ng of
the
three
(3) particular
local-global ties described
above of course implies that
the given instance of the contiguity primitive (p,m,a or
p,m,~ or
p,Gt,a)
is applied repeatedly
for every "suitable"
O:EE supporting cell ("unsuitability" only arising in the

case of some "boundary" cells for which the critical number
of neighbouring cells cannot be reached):
this in fact
implements the so-called "differential characterization"

aspect
of
the
EE
generation
process,
"essential
characterization"
being
implemented
through
the
"pattern" of every instance of the contiguity primitive.
very
Attempting to visualize the global outcome of such a

"networking" task grew more and more difficult as we tended
to introduce in the "picture" the rest of the perceptual
11M's "p,m,a" diversity combinations, those which call for
double-feature and triple-feature diversity (namely "p,rn,a",
"p,m,a", "p,rii,a", and "p,m,a"). However, it was not until we
realized that "differential characterization" could benefit
inunensely from
second order "repetition"
,i.e. recursion,
that the problem seemed to escape our intuitive grasp,

suggesting that maybe time had come to call upon computer
simulation to help figuring out exactly what could happen
within
our growing emergence paradigm:
obviously
basis!!
become
incapable of
Introducing
doing it
recursion
in
the
on
the
mindts eye had
a stand-alone
"differential
characterization"
process
simply meant
repeating the
repetitive application of the contiguity primitive (onto the
newly created sets of EE supporting cells) until no cells
307
remain onto which the primitive can

be applied: the
principle
was
extremely
simple,
but its networking
implications were not!
Steps were then taken to implement the theoretical
paradigm into an actual computer setting.
The highly
parallel nature and the sheer size of the formal neural
networks to be implemented were requiring vast amounts of
available memory and highest possible processing speed;
although Assembly Language had become over the years our
preferred simulation language, we decided this time' to move
back up to a higher level compiled language, the C language,
mainly for purposes of transportability.
To date, we have been able to implement both the
perceptual and the motor 11M's, together with procedures for
the application of the contiguity primitive, which has also
been implemented. The perceptual 11M is fully operational,
and has been provided with a "visual environment" simply
consisting of the "arm" which the motor IIM drives.
No actual "mutants" have been generated as yet, but we
are strongly anticipating two (2) highly interesting early
patterns of emergence, which seem to fit in quite nicely
with current knowledge concerning the actual neurophysiology
of perception in mammals: the first one points directly to a
neurophysiological primitive operator (the "TDU") which we
called upon (Lamontagne,
1976), a few years ago, as
"universal"
neurophysiological support-structure for motion
detection, while the second one points to a whole set of

now "classical" neurophysiologial structures, the so-called
(concentric and linear) "on-off" visual rerceptive fields
(see Hubel (1963) for a succinct survey of the findings in
question). On the other hand it already appears that if the
contiguity primitive seems to
be "necessary",
if one is to
lIinhibition tl
appears
necessary in
hope
reaching neural structures
presenting functional
similarity with the visual cortex of higher mammals, it does
not seem to be
"sufficient"
for
this
purpose: an
order to rid
arising
from
primitive
the
the
also
to
be
networks of a form of "redundancy"

repetitive
aspects of "differential
characterization", However, the above remarks rest largely

on intuItIon, and some important simulation work remains to
be done before one can hope to turn this intuitive basis
into a formal one.
308
C. LAMONTAGNE
3. CONCLUSION
But what does all this have to do
with "reflective
abstraction"? Are we not stretching the notion far beyond
its "breaking point"? How serious is it to argue that the
combinatorics
of
"EE
local-global
contiguity-based
relationships" prefigures the whole Piagetian "equilibration
process"
in general, including the key-aspect of the
acqulsltlon of conservations,
abstraction" in particular?
the
If,
on the one hand,
and
Piagetian
II
re flective
we do acknowledge the fact that
emergence paradigm described in Section 1 falls short

providing an intuitively obviously relevant "detailed
of
model"
of reflective abstraction in the overall
context of
the equilibration process and within the "main-stream"

Piagetian "conceptual network", on the other hand, we wish
to remind the reader that quite apart from the fact

is
in
its prime infancy,
our research
endeavour
that it
aims at
extending the piagetian attempt at deciphering the mysteries

of cognitive development, or emergence, (i) through a new
formal reference (the computational one), (ii) taking into
account
new
levels
of
biological
embodiment
(the
neurophysiological domain), and (iii) starting at a much
more primitive stage of cognitive processing (that of early
perceptual and motor processing).
We therefore suggest an
attitude based on an active search for relevance along the
proposed new line of research, rather than on an expectation
of a final and complete cannonically Piagetian solution. In
other words: "Don't bite my finger, look where I am
pointing!", as W.S. McCulloch used to say (according to S.
Papert, in McCulloch (1965)).
But where are we pointing, exactly? And how does this
"vanishing point" relate to the key Piagetian notions of
lIequilibration",
"conservation",
and
"reflective
abstraction", which, after all, we did choose as conceptual

springboard?
To the first question, quite clearly, the answer is
that we
are pointing to
a computationally efficient
perceptual
and
motor
emergence
process,
where
sets
of
existing perceptual and motor structures are repeatedly

expanded to include ever more encompassing ones.
To the second question, also quite clearly, the
general answer rests on the fact that the scope of the
Piagetian notions in question seems to have been wished, by
309
Piaget at least, to cover the full range of "knowledge

supporting" biological levels, as evidenced by the Piaget et
al. (1977, p.319) excerpt already quoted in Section 0, and
the following similar but more detailed comment by Pinard:
"One might even go as far as Piaget ... in
claiming that the distinction between empirical
and reflective abstraction already
has its
equivalent or
analogue
at
the
level of
biological, organic functions in contemporary
genetic theories, in that these consider both
exogenous environmental factors (which would
correspond to "empirical
abstractions") and
factors
of
endogenous
reconstructions
or
syntheses within the genome (corresponding to
"reflective
abstractions")."
(Pinard, 1981,
p.36).
Obviously falling within this extremely wide scope, our
model of emergence then qualifies as a relevant instance of
the processes in question, and a particularly detailed one
in
point
of
fact.
In
our
particular
case,
"empirical
abstraction" is carried out through the actual "performance"
of the perceptual and motor neural networks as such, given

particular environmental settings which either support or
do not support (as "exogenous factors") the local-global

ties embodied in the particular networks, while "reflective
abstraction"
is carried out
through
the
actual emergence
process as such, where new local-global ties are implemented

within
particular perceptual
or
motor
networks (as
"endogenous reconstructions or syntheses" factors).
As a final comment we wish to mention that at present,
one of the most promising "anchor points"
of
our endeavour
within the canonical Piagetian conceptual realm surrounding

the notion of "reflective abstraction" seems to be that
related
to
the
concept
of
"conservation".
As
can
be
gathered from Piaget's very broad usage of the term (for

instance in Section 11 of Biologie et Connaissance (1967)),
one would expect conservations to emerge from any developing
cognitive system whatsoever,
however primitive, and it does
seem that the way in which we have conceived our perceptual

and motor emergence paradigm lends itself particularly well
to this idea, every new local-global tie implemented making
in fact way to a particular potential
"conservation", which
the formalism should allow to specify unambiguously.
C. LAMONTAGNE
310
It is therefore with these concerns in mind

are entering the "intensive simulation"
phase
research endeavour.
that we
of our
ACKNOWLEDGEMENTS
The financial
support
of the
National
Science and
Engeneering Research Council of Canada over the past few
years of research,
and
the
powerful, fruitful, and
passionate
conceptual resistance offered
by Jean-Roch
Beausoleil,
Celine Cote, Denis Belisle, Jean Bernabe,
Jean-Pierre Delage, Alain Desrochers, and Pierre Tremblay,
over the past ten years of research,
are gratefully
acknowledged
as determinant factors in
obtaining the
above-reported research results.
EVOLUTIONARY EPISTEMOLOGY, GENETIC EPISTEMOLOGY,

HISTORY AND NEUROLOGY
Leo Apostel
1. INTRODUCTION
The papers I will comment upon
(those by Gillieron,
Lamontagne, and Miller) were brought together because all
three of them are concerned with 'genetic epistemology'. As
the editors of this book have remarked, an EE looks for a
model of the history of science in phylogenesis, while
Piaget himself has mainly examined the analogies between
ontogenesis and the development of concepts, in his genetic
psychology in general and in his main treatise on GE (3
vols.) in particular. Piaget & Garcia (1983) offer the only
fully worked out parallel between the history of science and
genetic epistemology.
One might thus claim that GE does
not belong to EE.
If this were true, the models offered in
the papers I am going to discuss would fall outside the
scope of EE.
They
would simply be
examples of a
naturalistic epistemology according to which the study of
the evaluation and history of science must
itself be a
science. The sciences taken as paradigms are different
however:
deductive neurology in
Lamontagne's case, genetic
psychology (and implicitly embryology) in Gillieron's case

and gestalt psychology added to genetical psychology in
Miller's case.
A clear incompatibility between EE and GE is stated by
Gillieron.
As far as
I can
see,
her
conclusion
is
not
necessary but depends on a very specific definition of

evolution.
It is true (as she states) that ievolution is
to be explained by blind variation, selection and retention,
then the mechanisms that describe the historical and

psychological history of knowledge cannot be analogous to
EE.
However, what happens when one looks at the explicit
statements of EE about the development of
knowledge?
for instance,
introduces the
three
Donald Campbell,
Darwinian mechanisms on a great number of levels.
On each
level, change is random (blind variation, selection and

differential retention), but a change at level n may
introduce regulated changes (not random at all)
in level
n-x! Is this classical neo-Darwinism?
Certainly not.
It
311
W. Callehaut and R. Pinxten reds.), Evolutionary Epistemology, 31/-323.

1987 hy D. Rl'idel PuhlishingCompany.
312
L. APOSTEL
is nea-Darwinism
with something
added
proliferation of vicarious selection levels.
no restrictions are imposed on this
hypotheses remain empty;

evolution can be explained
every
by a
to
As
proliferation,
non-random,
random one
it:
long
the
the
as
EE
teleonomic
on another
level.
In fact, during the conference this book is the
result of, Donald Campbell emphatically insisted that his
work has the nature of a challenge: he calls on others to
discover or refute the existence of his levels and he is
fully aware that the EE he is presenting now is purely
speculative. To a lesser degree Jean Piaget
also deviates
from classical nee-Darwinism.
His deviations
are more
severely sanctioned in biological
circles because his
phenocopy hypothesis seems incompatible
with molecular
biology and comes close to Lamarckism.
The idea that a
species which has, during one or n generations, adapted to
external influences, and then reconstructs from the genome
itself the phenotype it has been using to adapt is abhorrent

to nearly all biologists.
But I have no judgment to present here; I simply want
to emphasize the speCUlative character of the theories
defended by EE on the one side (many-level evolution) and of
GE on the other (phenocopy).
None of both derives the
historical or/and individual development of knowledge from
the neo-Darwinian hypothesis as such. The two add essential
elements to the base from which they start. But having
added speCUlative elements, both EE and GE then claim that
the mechanisms causing speciation
in phylogenesis are
analogous to, or causally explain (two different theses, by
the way) the mechanisms causing the change of human models
of the world (knowledge development).
If this is not an unfair account of the state of the
art,
then the central problem
adequate explanation of the

evolution is not denied by
remains:
What
is
the most
evolutionary process? While

most biologists, theories of
evolution are neither certain nor complete; nea-Darwinism is
either in a period of crisis, or - at least - of qualitative

growth.
The problem of the evolutionary explanation of
knowledge generation belongs to a future time, when it will
be possible to compare many detailed histories of scientific
theories
and
neo-Darwinism.
of
everyday
skills
with
an
enriched
EPISTEMOLOGY, HISTORY AND NEUROLOGY
313
it seems to me that Gillieron's

(even if she is quite correct when
speaking about evolutionary theory as it presents itself in
1987) .
Before going on, I want to make a last remark about EE
and GE.
Gillieron quotes with approval Brian Goodwin's
'constructivism'. I admire the elegance and subtlety of
Goodwin's work in embryology, but it seems to me that he
represents
not
'evolutionary
epistemology'
but
'epistemological evolutionism'.
He explains the evolution
of life by means of information-processing concepts. Here
evolution is not the basis, and history of thought the
explanandum; rather the opposite occurs: knowledge gathering
becomes the base and is used to explain ontogenetic and
phylogenetic development.
Both Piaget and Campbell
if
I am not mistaken - should be read as explaining knowledge
by evolution, and not the reverse.
In fact we would have discovered a second difference
between EE and GE, if Goodwin's work were to be taken as
typically Piagetian biology. Campbell has repeatedly stated
that he is a strong realist. In a recent paper, Giere
(1985) attributes this realistic stance to EE as such. If,
however, constructivism entails (as Gillieron claims) that
the structures generated by the knower do not exist at all
in the external
world,
then, by not being strongly
realistic, GE could not belong to the EE family. I would
To that extent,
verdict is premature
claim, however, that even if we need the power of our action
to construct the structure of our models, this does not

prevent these models to be partially homomorphic to real
systems outside the knower, only dimly exhibited by the
perceptual field.
After looking
at these
apparent oppositions, I
conclude that GE is a special case of EE characterized by
(i)
the importance of
two very specific properties:
embryology in evolution, and (ii) the rich material on
space, time, and number gathered during sixty years of work,
trying to answer the basic question "Why is mathematics,
whose source is so different from physics, as fertile in
physics as it appears to be?" In fact all other schools of
EE refrain to attack this basic problem. And while all these
other schools may use a more orthodox explanation of
evolution, this orthodox theory of evolution has not yet (as
far as I am aware) been applied to explain the many
well-known facts about the development of our basic physical
and mathematical notions. I challenge the other schools of
L.APOSTEL
314
EE to come up with their explanation of the facts. After

all we are animals living in space-time; it should be
possible to discover how we mapped so usefully space-time on
organisms and acquired such rich orientation facilities.
This challenge having been made, let me now return,
for a brief remark, to the importance of embryology in

phylogenesis.
Piaget is not
the
only biologist or
philosopher who was fascinated by it.
William Wimsatt has
repeatedly mentioned it, and Stephen Jay Gould has studied
it thoroughly.
But we should know much
more about
evolutionary embryology (especially of the brain) in order
to attack
with
strong
enough
forces,
the
intuitions of
Baldwin and Piaget.

The relation between EE and GE will not become clear,
however, if one does not consider it in the light of the
problem of the reduction of theories to each other. In
quite a few formulations of EE we read that the history of
knowledge and/or science is, by means of EE, 'reduced' to
the evolution of life.
What is Piaget's position with
reference to such
Gillieron clearly has
a 'reductionism'?
this problem in mind when she insists on the circular order

of the sciences, as defended by Piaget. The simplest circle
of
the sciences can be used as an example, even
if
we know
that more sophisticated versions exist: Psychology explains

mathematics, mathematics explains physics, physics explains
biology
and
biology
explains
psychology.
The relations
between the different sciences,

however, are not relations
of asymmetrical dependence, but of 'mutual determination'.
So arrows pointing in two directions are meant and not

arrows indicating only one preferred orientation.
But more
ought to be said: (i) since all sciences are historically
evolving entities,
their relative places on the circle
magnetism and life
were
may
change (for instance, as long as astrology was taken

seriously, astronomy explained psychology, and as long as
thought to
be
related, magnetism
explained biology or inversely); (ii) The different arrows

on the circle do not have the same meaning (for instance,
one may still
try to causally explain biological facts by
means of physics,
that
physics
will
but no one seriously
be
derivable
from
considers, in 1987,
pure mathematics).
Sometimes the arrows mean 'are used in' and sometimes they
mean 'are explained by';
(iii) The vicinity between
psychology and mathematics (through logic) is by no means

necessary: sociology of mathematics and anthropology of
mathematics are also offering partial explanations of this
315
EPISTEMOLOGY. HISTORY AND NEUROLOGY
science;
(iv) The relation between psychophysiology and
logic 1S similar to no other one: causal chains on the
neurological level correspond to implications on the level
of consciousness. So GE claims (as any theory of entailment
and of causality can show)
that these two relations
(causality and entailment) do not have the structural
properties to solve, by some kind of structural dualism, the
mind-body problem which is lurking at the core of EE. The
relations between EE and the Piagetian circle of the
sciences are
unclear.
To be sure,
if a biological
explanation
for
the
history
of knowledge
and its
self-imposed norms were available, this would allow to close
the circle at one point (the relation human biologyinformation acquisition in psychology and history). But
the circle needs much more than that.
Moreover, if mutual
assimilations and accommodation are typical of the circle,
then biology will have to be changed as much by epistemology
as inversely. Yet,
neither in history nor in behavior can
strong arguments in
favor of the circle
be found.
Inversely, even if EE were more than a program and already a
reality,
nothing would compel us to claim either
a
historical or a suprahistorical truth in favor of the circle
of the sciences.
The fact that EE does not entail the
reality of the circle is not identical to the fact that EE
is
incompatible with it.
For those who fear that EE
amounts to a
reduction of higher to
lower levels of
complexity, the possibility (not the necessity) of such an
organization
is an important argument alleviating their
doubts.
Arthur Miller's contribution is an attempt to prove,
by means of a concrete historical analysis, that the
development of one science during a certain period (quantum
physics from 1913 till 1927) can be explained by mechanisms
due to GE.
If the reader is convinced, after my discussion
of Gillieron, that after all GE is an example of EE, then
this case study becomes also a strong argument in favor of
EE; it would
enable, in its Piagetian version, to throw
light on the history of the most advanced sciences. History
- all naturalistic epistemologists agree - is the laboratory
of epistemology.
The paper refers to Miller's (1984) book.
Before entering the discussion, I want to make a preliminary
remark.
Max
Wertheimer's
Productive
Thinking
contains a
famous chapter X, "Einstein: The thinking that led to the

theory of relativity".
Wertheimer, one of the founders of
gestalt psychology,
uses the concepts and notions it
316
L.APOSTEL
developed to explain Einstein's thought, as the latter

explained it to him. Miller's book depends on two schools of
psychology: GE and gestalt psychology. E.g. Miller uses
Piaget to explain the growth of quantum mechanics, and
gestalt
psychology
to
explain
relativity.
Every
psychologist reading this will declare immediately that this
poses a considerable problem. For Piaget, 'gestalten' obeyed
other laws than operational or formal structures. In fact
he opposed them to one another.
How, then, could certain
parts of scientific development be explained by gestalt
psychology (if this were true it would be a refutation of
GE) and another part of the same development by GE (if this
were true it would refute gestaltism)?
Miller, in his
contribution, uses
GE focusing on 2M only. So I will
restrict my remarks to this small part of his work. From a
more general point of
view,
however, the question is: What
is the relation between gestalt psychology and EE? Certain

passages of Campbell's on perception seem to promise that a
relation could exist.
More work has
to be done if we are
to
come
to
a decision
on
this
point.
However, Miller
himself sees clearly that the intervention of imagery

thinking in scientific research is contrary to the Piagetian
view that holds the formal level to be superior to the
concrete-operational or symbolic or sensori-motor stages.
He realizes that he will not be able to use GE in history of
science without modifications.
The first modification has already been described (the
continuous use of imagery representations).
The second
derives from the fact that the adult scientist has evidently
attained the operational and formal stage (Miller, 1984, p.
285).
Miller does not mention the formal stage in this
context but his meaning clearly- implies it. It will thus be
impossible to apply GE's growth patterns if the discovery of
new facts does not repulse the theoretician to a lower level
(Miller,
1984,
p. 285). Moreover, the resistance of
scientific theories against accommodation will be stronger
than the resistance of children's "theories", because they
have so many connotations and represent the historical
development. Miller's awareness of the distance between GE
and himself shows this:
"I propose the following as a new
definition of assimilation: the application of a scheme to
problems involving
empirical data,
data from thought
experiments, or
aesthetico-philosophical commitments, or
dispositions."
If
the
invariance,
concepts
of
accommodation,
317
assimilation,
reversibility, scheme equilibrium will be used,
Miller will give them new meanings that are adequate in the
new context of history of science, This has bearing on the
relevance of his programs as to
their status in EE.
Obviously, a generalized schemata dynamics will have to be
derived from
Campbell's n-level random variation, if the
facts of Miller are to be considered as relevant to EE.
Here I would like to make some remarks about the
application of Piagetian concepts to facts of the history of
science, and address the question "Would another type of EE,
now extant, have done better with the material at our
disposal?"
In 1913, Niels Bohr,
in order to explain the
stability of an atom conceived as a minor solar system
surrounded by electrons, proposed that only certaln paths
can be followed by electrons, which emit energy in finite,
discrete amounts when they jump from one path to another;
they cannot be 'observed' during their jumps (there, no path
exists) and they can never get closer to the nucleus once
they have reached a basic trajectory.
Miller describes this as follows: The Rutherford atom
The Bohr atom (quantized) creates
was 'in disequilibrium'.
an
equilibrium,
by
'adjusting'
or
'accomodating'
Rutherford's atomic model to the facts of 'stability'
(explained by the quantization). However, this Bohr atom is
still far from stability because, keeping the initial
representation of Rutherford (the atom as a small solar
system), it 'displays representative activity analogous to
egocentric representation'.
I attract the
attention of the reader
to the
psychological terms used.
a) Presumably the Rutherford atom is 'in disequilibrium' because its stability is in contradiction with
classical mechanics. So far so good. If all contradictions
in science entail a disequilibrium, then Miller is correct.
History of science shows, however, that not all patent
contradictions create the immediate impulse to change the
theory. Therefore, I would only accept to a certain extent
the equivalence 'contradiction = disequilibrium'.
b) However, the Bohr atom is, for the very same reason
as the later history shows clearly, also in disequilibrium,
because it
certainly also contradicts classical mechanics:
electrons changing orbits do not have paths but should have
318
L.APOSTEL
them.
To call this an 'equilibration' is already difficult
to defend.
One simply jumps from one 'disequilibrium' to
another.
c) I am utterly unable to understand why the Bohr
'theory' displays 'egocentric representation'. Egocentrism
in psychology is the identification of an object with the
perspective on that object as presented to a particular
subject from a specific vantage point. Nothing in the Bohr
atom shows such 'egocentric representation'.
d) If we follow Miller's story, we repeatedly face the
mixture of a creative attempt to explain the history of
science by schemata dynamics and of the observation that
schemata dynamics cannot explain this same history.
For instance, reaching the pure formal stage is
certainly not the result of the Schrodinger-Heisenberg
stage, since both wanted to continue using visual imagery in
their work (and so with excellent results). Moreover,
Piaget's claim that accommodation and assimilation work with
equal velocity and pregnancy is clearly negated when Miller
tells us that during 1913-1923 "the thinking of Bohr and
other
physicists
emphasized
accommodation over
most
assimilation".
'preconceptual
Indeed, Miller attributes to this generation
thinking',
because
assimilation remain incomplete.
both
accommodation and
I believe the reader should not assimilate without

caution a theory to a 'schema'. A theory is an intellectual
structure,
social in
nature, inter-generationally present.
To be sure, we have the right to generalize the concept of

'schema' to such social and intergenerational structures.
But if we do, we should not expect the new concept to follow
the same dynamics as the original one; thus the explanatory
force of psychological observation with respect to the
global theory of dynamics vanishes.
In the story as told by Miller, 'accommodation' is the
modification of a theory to facts, and 'assimilation' the
explanation of these facts by the modified theory. Nothing
is gained in terms of deepening our understanding of the
story by these enlarged definitions. Quite to the contrary:
the
understanding
of Piagetian theory
is made more
troublesome,
because
high-class
results
of
physics
are
sometimes qualified as 'showing intuitive thinking' or

'preconceptual thought'. I cannot follow the details of
Miller's fascinating narrative here; I only wish to stress
two features that show how little explanatory power GE h.as
if his narrative is true:
319
a) When at a certain moment Kramers and Bohr consider

giving up
the conservation of energy, everybody shudders,
and general satisfaction is shown when the 'hypothesis of
new
consideration'
(!) is factually disproved.
This
entails, however, that the physical community was willing to
contemplate,
in order to solve some empirical difficulties,
the disappearance of the most important invariant, In GE,

when a high-class invariant is reached it becomes an a
priori,
and precisely this becoming an 'a priori'
is what
Piaget wants to understand.

Obviously the same does not
occur in physics; therefore the two developments are of
another type.
b) The persistent desire for intuitive visualization
and the willingness to leave behind the rules of classical
logic and syntax (complementarity and
the introduction of
quantum logics) to move beyond the frontier of Piaget's
classical INRC logic also show that neither the finality of
the formal stage nor the systematic dominance of formal
operations over concrete operational ones - two cornerstones
of Piaget's thought - can be preserved in its application to

the history of science. Miller makes this very clear!
But you can't have your cake and eat it.
If - as
Miller and I believe - complementarity overcomes the limit
of
the
systems
studied
by
GE,
then assimilating
'complementarity'
to 'reversibility'
(for reasons not made
clear) is of no explanatory value.

To conclude: Elsewhere, I (Apostel, 1980) have myself
proposed the idea that when the mastery of formal operations
is achieved, the skillful translation of formal problems
into the languages of earlier stages 1S one of the major
heuristic skills that,

by preventing a 'combinatorial
explosion', distinguishes the creative mind from others. In
this respect, one might say that at least sometimes in GE,

Miller's emphasis on visualizability has been taken into
account.
Intimate collaboration between, on the one hand,
visualization in particular, imagination in general and, on
the other hand, conceptualization,looks promising both inand outside science. Moreover, this fact can be easily
explained by EE: As human organisms we have first learned to
find our way in concrete space and time
in
our physical
and social environment.
That we continue to
make use of
our biological heritage on higher levels of thinking is only

to be expected.
320
L.APOSTEL
Disagreements notwithstanding - the main one being the
necessity of redefining the basic GE concepts for use in the

history of science -, Miller's contribution seems to present
a major challenge to the evolutionary epistemologist. Either

he can so generalize his concepts (while preserving their
laws) that he can explain important parts of the history of
science, or he cannot.
In the first case he becomes a
serious contender within the
philosophy of science; in the
latter he has only introduced a heuristic proposal that
cannot be evaluated yet.
Claude
Lamontagne's
Emerging:
the sensori-motor
structure will be relevant to EE if two conditions are met:
a) GE has introduced the concept of 'reflexive abstraction'.
This difficult concept plays a major role in GE; b) If it is
the case (one should read in depth Piaget's Biology and
Knowledge to find out) that 'reflexive abstraction' has
clear biological ancestors and follows laws derivable from
or at least compatible with evolutionary laws, then any
model of 'reflexive abstraction' (especially if it is
expressed in semi-neurological language) will be relevant to
EE. Lamontagne claims correctly that no clear model of
'reflexive abstraction' can be found in the literature and
proceeds to construct one. My comments on his proposal will
be twofold. I shall sketch its major features and show that
in fact and by the explicit avowal of the author no model of
'reflexive abstraction' appears.
This task achieved, I
shall sketch very briefly a direction
in AI (explicitly
unconnected to EE)
in which a
model for 'reflexive
abstraction' can be found.
Lamontagne considers a layered series of neurons.
The neurons of the first level receive inputs and yield
outputs of different types and having various relevant
features. Any neuron of the first layer is characterized by
a triplet {p, m, f,} p indicating if some x is an input or
an output, m indicating the type to which x belongs and f
the relevant features of the same x.
The second layer
superimposed on the first corresponds to specific relations
between the "p, m, f" triplets of the first layer neurons.
In simpler language neurons of layer one classify x's
according to three dimensions, and neurons of the second
type classify series of neurons of the first layer according
to the relations between the classification products of the
layers of the first type.
According to the present
commentator, this model obviously derives from Hebb's The
Organization of Behavior (1949).
The global system is
EPISTEMOLOGY, HISTORY AND NEUROWGY
321
defined statically by the number of layers of neurons, and

the relations between the lower-layer neurons and the
higher-level ones. Dynamically, the system is characterized
by its laws of development and growth (new layers can be
added, the relations between the layers can be changed).
Claude Lamontagne gets busy
exploring the large
multiplicity of growth and development laws one could
conceive
for these schematized and
ordered
neuron systems
(the brain as a whole certainly does not resemble this

model, but Hubel and Wiesel among others have shown that at
least the visual cortex has such a hierarchical structure).
Lamontagne uses the simplest, completely empirical growth
law (as sound methodology recommends) and considers "if two
i-th level neurons are neighbors and are simultaneously
excited along the same dimensions then (after i or n
occurrences of this type) an i-th level neuron grows that
will be excited when the two first are". Given the large
number of definitions of 'similarity' and of 'neighborhood',
even this law of
extreme simplicity (association by
contiguity and resemblance, combined on sensorial" and motor
characteristics ) has the result that the system can no
longer be studied by pure deduction and must have recourse
to computer simulation. I can appreciate the desire to work
with precision on simple models (that very soon become too
complex to be mastered).
However, in what the author
offers, not even the remotest simulation of the (admittedly
difficult) 'reflexive abstraction', is forthcoming.
If
evolutionary
neurology
were sufficiently
advanced to
suggest the evolutionary tendencies of brains and their
neurons over the ages, then the multiplicity of hypotheses
Lamontagne has to study would be drastically reduced. The
major problem I have with this carefully executed attempt is
that (like Hebb's) it remains completely empiricist and
associationist.
This set of ordered neurons is not an
active system, continuously exploring new combinations in
itself and its environment.
As long as such an 'activistic'
feature is not built in (presumably it could be built in),

the fertility of the approach, for the understanding of the
transspecies evolution of the brain and for the modeling of
'reflexive abstraction' remains remote.
As a second reaction to the fascinating problem the
author wants to resolve, I would like to present the
following hypothesis.
Very roughly, I would describe
L. APOSTEL
322
'reflexive
abstraction'
combinations
organism" .
of
as "pattern recognition applied to
internal
and
external
actions
of
the
There begins to exist, in the AI community, a growing

awareness of the need for simulating introspection (a major
step towards the pattern recognition just described).
Battali (1983)
offers a comparative discussion of
important earlier work by Weybrauch and Doyle.
I would now
like to address to Claude Lamontagne on one hand, and to the
friends of EE on the other a challenge:
1.
Can Lamontagne's neuronal model, by means of suitable
(and still understandable) complexifications,
gain the
power
necessary to execute or
simulate computational
introspection?
2.
Can
the simulations of
biological
evolution, poor as
they may be, be introduced in environments such that there

is a non-zero chance that in reasonable periods, frames
akin to the computational introspection programs could be
I grown I?
By
remarks:
way of
general conclusion,
I offer the following
1.
My first recommendation would be, given the current
multiplicity of theories of evolution, to work on the
epistemological consequences of all
these explanations.
No useful purpose is served, for non-biologists, by choosing
in favor of one of the competitors;
the attempt to tie EE
strictly
to one explanatory schema has until now had the
consequence (see Donald Campbell,
Jean Piaget, Konrad
Lorenz) that either the basis was transformed in order to
make the explanandum more understandable, or that the
explanandum was
transformed in order to make the basis
suff icient.
2.
My second recommendation is to follow Miller's example,
apply various theories of the evolution of knowledge to the
history of science, while remaining fully aware that the
transposition
from one
domain to
another introduces
arbitrary choices which must
be noted first, be defended
next, and finally be overcome by using more than one
evolutionary scheme.
3.
My third recommendation is to focus by all means on the
evolution of the brain-in-environment (ecology and ethology
combined with brain science). As far as I am concerned, AI
simulations will have to be selected at their highest
levels
and
their
evolutionary
theory
should
be
reconstructed.
323
The part of the Symposium I had to analyze was

extremely heterogeneous.
I still hope, however, that the
reader will see
how
many
fruitful interdisciplinary
connections are suggested by the concept of 'EE', if only (a
final caveat) EE also takes into account the evolution of
evolutionary mechanisms and the impossibility of one-sided
reduction.
PartlV:
Extensions and applications
THE EXCHANGE OF GENETIC INFORMATION BETWEEN

ORGANISMS OF DISTINCT ORIGIN CAN PLAY AN IMPORTANT
ROLE IN EVOLUTION
Jef Schell and Dani De Waele

1. INTRODUCTION
Recent developments in biological research have implications
for the classical concept of evolution, and especially for
the notion of mutation. At least in the molecular evolution
of contemporary life forms, mutations indeed do not seem
to represent a biologically relevant phenomenon. Although
we can experimentally demonstrate that it does exist, the
role of mutation in evolution can almost be disregarded.
Evolution as we currently observe it is predominantly the
result of what we call the 'horizontal
transfer
of
information'. This phenomenon consists in the transfer of
genetic information between one type of
organism and
another.
The independently pre-evolved information of
different types of organisms is recombined and mixed again
into new combinations.
It is the same phenomenon that we
experimentally
realize
under
the
name
of 'genetic
engineering': the horizontal transfer of genetic information
with a different evolutionary origin.
However, and maybe
ironically, at the same time that human beings think that
they invented this kind of horizontal transfer, we begin to
discover that this is an important phenomenon in natural
evolution itself.
2. THE MOLECULAR BASIS FOR UNDERSTANDING EVOLUTION
In
order to
for evolution,
explain the importance

we must start with
of horizontal transfer
the basics
of molecular
biology.
In molecular biology life is said to be the
reaction between three sorts of elements.
One element is
the coded information, which takes the chemical form of a
linear molecule and is composed of a number of repeated
signals: the code.
The second element is the cell, or a
number of cells together: the organism, a system in which
327
W Callebaut and R. Pinxten fedJ.), Evolutionary Epistemology, 327-335.

J. SCHELL AND D. DE WAELE
328
life is organized in some isolation from the environment.

The third element is the interaction between the system and
the environment.
The reason why life shows some form of stability, is

that whatever takes place in the cellular system, has to be
represented by the coded information it contains, and is
thus constrained by the framework of possibilities provided
by the chemical and physical properties of the proteins that
are present. Since the coded information is responsible for
the type of proteins that are made in the cellular system,
it is clear that the genetic code plays the major role in
evolution.
Proteins cannot
be changed
information, responsible for their

This is an important limitation.
unless the genetic
assembly,
is changed.
3. SELECTION OF PROTEINS VERSUS CHANGES IN THE DNA

Proteins are made on the basis of a template, i.e. a DNA
molecule. The DNA molecule has two major functions: (il
being stable; (iil containing a program for the synthesis of
the protein.
Proteins determine, directly or indirectly, the life
processes in the cell, so that selection operates on the
proteins.
There is no selection on the level of DNA since
DNA merely represents a stable code.
Yet it is thought that evolution is due to a change in
the genetic code, which is true in the end , but only under
the following conditions. When changes occur in the genetic
code, these changes will be either maintained or not,
depending on whether the living system in which they occur
survives or not. The genetic material itself cannot survive
and be reproduced without its cellular environment (except
for short times and in special laboratory circumstances).
If changes in
the genetic material damage
the cell
functions, .then the nucleic acid disappears.
Here the
expression 'survival of
the fittestf
is,
to
some extent,
still appropriate: the evolution of living systems would not

be possible without reproducibility of the cell types in
which you have the genetic program that might be selected.
In fact this is the only sense in which evolution is
determined.
In the context of the research that we shall
discuss, the proteins are measured in terms of their
efficiency, and it is possible to change one program into
ON GENETIC INFORMATION AND ORGANISMS OF DISTINCT ORIGIN
329
another only if any variation in efficiency is reflected in

the differential
reproduction of the
types of cells
containing the genetic information of interest.
Genetic information can be changed either gradually or
abruptly. This can be done by replacements of genetic
information or by the introduction of from information (i.e.
from another living system) into the present system.
4. THE NOTION OF 'SPECIES'
How can this be related to the classical theories of
biological evolution? Biologists were impressed by the fact
that the living world is not a continuum of forms but, on
the contrary, that there exists an identification into
groupings.
As a consequence, the notion of species was
introduced. One way of explaining this discontinuity in the
tenns of modern biology is that each given form possesses
its own set of genes coding for its own set of proteins. The
genetic information of one system is shielded off from the
genetic information of another system.
This explanation,
which is
the
classical
basis
for
understanding species,
seems now to be very incomplete because the frequency of

interactions within a species is significantly higher than
the frequency of interactions between different species.
The probability of new combinations of genetic information
occurring is greater within than between species.
5. DEMONSTRATION OF GENE TRANSFER IN MICRO-ORGANISMS
How do scientists demonstrate this

intraspecific gene
transfer? For example: we take one species of bacterium,
say a bacterium living in our intestine, Escherichia coli.
We isolate this bacterium as a single clone, i.e. we take a
culture of cells each of which has been derived from one
original cell,
so that we have genetically identical
organisms, - unless mutation has occurred.
The following is
an example providing experimental proof of the possibility

of evolution through mutation is.
Say, we grow a bacterium
that is sensitive to a specific
antibiotic, such as
penicillin.
If our popUlation of bacteria is large enough we will
find within it some bacteria that 'have become' resistant to
the antibiotic. When selective pressure (the antibiotic) is
330
J.SCHELLANDD. DE WAELE
applied,
every
bacterium that was sensitive
to the
antibiotic dies and only the mutant bacteria which have
become resistant to the antibiotic through mutation survive.
This is Darwinism at it simplest - a clear proof that
Darwinism is basically correct. . However, if instead of
applying the selective pressure to the culture in isolation,
we first put its component cells in contact with other
living systems (other bacteria, or plants or whatever there
is), then we do not find that mutation has occurred, but
rather that bacteria have received a whole set of new genes
from the other living organisms.
What has happened? We know that bacteria have evolved
very dynamic genetic systems with a variety of mechanisms
for the transfer of DNA. We also know that the most
efficient mechanism for bringing new DNA into a given cell
operates between cells of almost the same type. We have
demonstrated that several types of bacteria dramatically
acquired e.g.
the ability to detoxify specific antibiotics
in this way, even when we did not intend to treat them with
antibiotics. This indicates that bacteria do not evolve
through mutation, but rather by opening themselves up to
genetic information that originates and has become adapted
in other environments. In the usual way, a genetic program,
thus acquired, that provides a better chance of survival for
the cell, will be maintained.
6. MECHANISMS FOR GENE TRANSFER

Recent studies have examined the
mechanisms by which
bacteria (even those belonging to distinct taxonomic groups)
can rapidly exchange genetic information.
In many bacteria
the genetic information is carried not only by a bacterial
chromosome, but also by a plasmid or circle of DNA. In most
circumstances, the bacterium can live without the genetic
information of this circular DNA.
These plasmids have two
types of properties:
1.
The ability to replicate very rapidly.
The
information supporting the replicating machinery lies within
the plasmid itself.
2. The capacity to transfer from one cell to another.
When a cell carrying such a plasmid meets a cell which does
not, the two cells come together and a copy of the plasmid
is passed from one cell to the other.
The probability of
transfer normally increases with the extent to which the 2
types
of
cells are
taxonomically
related,
331
although some
plasmids have been evolved specifically to overcome transfer

barriers.
Instead of having information for only a few proteins,

the plasmid can have
information
in
(e.g.
for
several
proteins or
if only one of
these functions or
properties has a selective advantage, nevertheless the
complete set of properties will be transferred. One change
functions.
the
Even
environment
the
selective
pressure
of an
antibiotic)
can thus lead to the acquisition in the
population of a whole set of new genes.
When such a new
gene - for which there is no selection - has significance
for the cellular system, this can produce an abrupt change
in the evolution of the species concerned.
The phenomenon of gene transfer via plasmids is very
important for understanding the evolution of microorganisms.
It indicates that their evolution comes about primarily
through horizontal transfer of information, and not through
the gradual accumulation of mutations.
7. CONSERVATION OF INFORMATION VERSUS
INNOVATION OF INFORMATION
evolution we see
a balance between
two opposing
tendencies: (i) a tendency to keep genetic information
constant within a species; (ii) a tendency toward the
dynamic transfer of genetic information from one living
system to another.
One of the systems which both allows the transfer of
genetic information and restricts the transfer to fairly
similar organisms, is sex - one of the pleasures but also
often one of the miseries of our life. Linking the exchange
of genetic information with reproduction
is extremely
efficient.
A new generation is made out of the coming
together of the genetic information
of two different
individuals who have but slightly different genetic programs
(variation on alleles).
Since what is transmitted from one
generation to another is only the information carried on by
the germ lines, there is a tremendous limitation to the
transfer of foreign information.
Only information present
in, or able to enter, the germ lines will be propagated to
the next generation.
In
332
Bacteria,
however, do not
between reproduction and exchange
They can exchange
information
make this correlation

of genetic information.
completely
independently of
the cell division.
What about these organisms where exchange of genetic

information is linked to reproduction (= all organisms which
reproduce by sexual phenomena)?
It was generally thought
that in these organisms, the species is very well prot~cted

against input of information from outside and that therefore
evolution would occur through mutation. Scientists now
experience that mutation is not the only way for evolution.
8. INTERSPECIFIC GENE TRANSFER IN A LABORATORY

Before giving an example,
it is essential to point out that

universal,
that
it
is not
species-specific.
This means that a given piece of DNA has
information for a specific protein, regardless of the DNA's
specific cellular environment.
The same piece of DNA, if
transcribed in species A or in species B, will always give
the same protein. What does change from system to system is
the kind of signals that are used to render the DNA
expressive, but once the DNA becomes active, it produces the
the
genetic
same
protein.
code
is
If the code
were not
universal, then even
exchange of
information
could
have
no evolutionary
significance: the two sets of information concerned would
not be compatible because the codes would be different.
Genetic engineering is based on this universality of
the code. Geneticists are looking for those signals which
make the DNA function and are making new combinations of
different DNA
pieces.
The
combinations that became
popularized
first
were those
directed by industrial
interest.
For example, the case of insulin for patients
suffering from diabetes: the genetic information for the
synthesis of insulin is taken from human cells and the
signals are modified so that the DNA can become active in
bacterial cells.
Since bacterial cells duplicate every 30
minutes the synthesis of human insulin
cells can produce large
amounts of
out of bacterial
insulin.
A new
industrial process has been developed with success.

Again,
this proves that horizontal
transfer is
possible.
Human beings suddenly have become an element in
evolution by creating new combinations of properties from
different organisms,
by using mechanisms, however, that are
333
taking place in nature itself (see below).

What continues
to be important is that new combinations of information will
only be maintained if the new information contributes to the
survival of the cell in which it is present. If it does not
contribute to (nor harm) the survival of the cell, the new
information will be maintained with a very low probability.
It may remain in some kinds of cells, where it is useless,
and may subsequently emerge under conditions which re.nder it
advantageous.
In industry, this advantage is created on purpose:
there, new genetic combinations have a very high rate of
survival.
In the hands of human beings, new combinations
acquire an advantage they never had before.
9. AN EXAMPLE OF INTERSPECIFIC GENE TRANSFER
IN NATURE
How tight are the constraints on the intake of foreign
information in naturally evolving species?
Apart from the
horizontal
transfer done by
humans
under laboratory
conditions,
there
is one
well-known
system whereby
horizontal transfer occurs in nature, and this, among widely
different species.
This natural system of gene transfer is
studied in the Laboratorium voor Genetica (Gent, Belgium).
In this case, highly evolved organisms, namely plants,
acquire sets of information from bacteria.
This phenomenon has been observed while analyzing a

plant disease.
Some plants are susceptible to cancer
resulting from the introduction into the plant of a given
type of bacteria, living in the soil, namely Agrobacterium
tumefaciens.
When these oncogenic soil bacteria infect the
plant they cause the formation of tumors on the plant and,
as a result, the plant grows in a completely abnormal way.
In the bacterial cell, the presence of a plasmid ( a circle
of DNA), in addition to the bacterial chromosome, can be
demonstrated. This plasmid is responsible for the induction
of tumors in the plant: without such a plasmid the bacteria
can no longer cause tumor formation if the plant. In the
course of infection, the bacterial plasmid is transferred to
the chromosome in each of a number of
plant cells. Some
genes, carried on by this plasmid, deregulate the normal
growth of the plant cell, producing an undifferentiated mass
of plant cells forming a tumor.
This phenomenon has
biological significance.
One of the genes which is carried
334
by the plasmid is not responsible for tumor formation, but

rather for the formation of a specific organic product in
the growing plant tumor cell.
This organic product is
essential for the growth of these specific, infecting
bacteria
in
competition
with
other
bacteria: only
Agrobacterium can use this organic compound as a carbon
and/or
nitrogen
energy
source.
Agrobacteria
are
sophisticated parasites: by a gene transfer mechanism they
force the infected plant to make an additional product that
only they can use as a nutrient. In this way the metabolism
of the plan is diverted in favor of producing elements for
the metabolism of the colonizing bacteria.
This
system of horizontal
transfer
of genetic
information can be used to introduce into the plants
information other than that which is tumorigenic. Indeed,
geneticists can eliminate the piece of DNA on the plasmid
that is responsible for tumor induction (while leaving
intact the information for the transfer of the plasmid) and
replace it with totally new information.
In this way, new
information can be inserted into the plant, using the
bacterium as a vector, without causing tumor formation.
Instead of a proliferation of undifferentiated cells, the
bacteria induce a specific reprogramming of the plant cells.
For example, the plant can be induced to make roots at
the site of infection, an event which can also take place in
nature.
Since bacteria exchange plasmids with one another,
the plasmid can also be replaced by, or combined with,
another one: if the new combination represents an advantage
for the cells in which it is present, the new information
will be amplified.
In the case described here, we have bacteria which
somehow required a set of genes that has no function for the
bacterium when it is in isolation.
The plasmid becomes
useful for
the
bacteria only after
they have been
transferred to another organism.
For the infected organism
this has a dramatic effect on the
type of cellular
differentiation.
The important possibility of DNA transfer should now
be quite clear. In principle, there is no limit to the
extent of genetic information that living organisms can
acquire.
The concept of species goes against this rule
without delimiting it.
335
10. SUMMARIZING CONCLUSION

Mutations have traditionally been regarded as the sole motor
of evolution.
The
environment was
supposed
to
pick out
those mutations that are advantageous.

Scientists now find
that the horizontal transfer of large sets of genetic
information might play an important role in evolution. Sets
of information with different evolutionary origins can
recombine and when the new combination serves a particular
role in a given situation, it becomes amplified. Since the
system which copies information is blind, selection only
occurs at the level of the proteins which effect the
differential survival of the cells in which they are
present.
FERMAT'S LAST THEOREM SEEN AS

AN EXERCISE IN EVOLUTIONARY EPISTEMOLOGY*
Jean Paul Van Bendegem
Vrije Universiteit Brussel
1. IS EVOLUTIONARY EPISTEMOLOGY RELEVANT FOR UNDERSTANDING
THE MATHEMATICAL PROCESS AS IT ACTUALLY OCCURS?
Out understanding of
the mathematical process
has been and
still is (rightly so) associated with Lakatos' Proofs and

Refutations. But at first sight, the results of his research
have little or nothing to do with evolutionary epistemology.
The scheme he arrives at, is roughly this:
If you have a conjecture, set out to prove it or

refute it, and then try to find counterexamples both to the
conjecture and to
But
as
the 'suspect'
so
often
interesting heuristic,
conclusions.
For there
lemmas.
happens,
first
sight
is
an
but a bad method for reaching

are two hidden lemmas in Lakatos'
method as well:
(a) the method works only if a proof or refutation has been
found;
(b)
the method assumes that counterexamples are accessible

that it is possible to actually construct one,
(that is,
such as is the case in the Euler
example concerning regular
polyhedra) (1).
Concerning (a), it can be remarked that a large part
of
mathematics
consists of unproven
statements; and
concerning (b), that it is not always the case that
counterexamples are 'accessible'.
This means that in all
cases not covered by
the method of proofs
and refutations,
other methods are needed.

What could these be like?
This
paper is an attempt to partially answer that question.
As
it turns out, on the one hand the method presented here is
favorable to an interpretation in terms of evolutionary
epistemology, on the other hand, we are not forced to reject
Lakatos' method, thus proving that second sight is a good
method for correcting first sight.
The structure of this paper IS quite simple. In

section 2, I wander through the history of a particular
mathematical
problem:
Fermat's
Last
Theorem
(FLT
henceforth). In section 3 I try to derive a general scheme
337
W. Cal/ehaur and R. Pinx(cn (eds.), EvolUilUnary bpistemology, 337-363
/987 hy D. Reidel Publishing Company
338
J. P. VAN BENDEGEM
for the historical growth and development of FLT. Of

course, I do not claim to have a scheme that applies to the
whole of mathematics (that would be unwarranted induction in
this case indeed!), but what it does show is that Lakatos'
scheme needs to be complemented and that evolutionary
epistemology is an interesting tool for that purpose.
2. WHY IS FLT SUCH AN INTERESTING CANDIDATE?

The two main reasons are that it fits the requirements for
our purpose:
(a) there is no proof of FLT (and very likely this will be
the case for years to come)
(b) counterexamples are extremely hard to find.
An additional reason is that in 1987 we can celebrate
the 350th anniversary of FLT.
'For all n > 2, there are no integers
What is FLT?
x,y,z such that x"+ y" = z"'
The story that goes with it, says that Pierre de
Fermat (1608-1665) wrote this statement in the margin of his
copy of the algebra of Diophantus and added to it: "I have a
truly remarkable proof for this statement, but the margin is
too sma1l to contain it".
This proof (if that was what it
was) has never been found among the papers of Fermat.
So far no proof has been found, notwithstanding the
fact that it ranks among the top problems in mathematics,
thus (a) is certainly satisfied. As to (b) there is a very
short and ingenious proof by Grunert (1856), showing that if
there is a counterexample (i.e. for a certain n, there is a
triple x,y,z such that xn + yn = zn holds), the following
relation must be satisfied: min(x,y,z) > n.
The proof runs as follows:
suppose that x < y < z (the integers can always be arranged
in that way), then xn
zn - yn
(z - y) (zn., + zn-2 y + . + yn.,)
> (z - y) nyn.,
(by replacing each z by y in the right-hand factor in the
right-hand side of the above equation, each term becomes

, and since there are n terms, this produces nyn-, and,
yn-1
since z > y, this must be smaller than xn)

> nyn-,
(since z - y is positive and larger than 1)
> nx n - 1
(for y is larger than x)

But from xn > nx n ' it follows that x
>
n.
QED.
FERMAT'S LAST THEOREM AS AN EXERCISE
339
since we know today that FLT holds for all n up to 125000,

this implies that a possible counterexample will involve
numbers at least of the following size:
125002125001.
What we are talking about here, are numbers involving more
than 625000
digits!
Furthermore
the
proof has an
interesting consequence: as more cases of FLT are proved
(i.e.
the set of n for which FLT holds), counterexamples
become larger and larger.
Thus, loosely speaking, the
nearer we come to a proof of the full statement, the less we
can say about possible counterexamples.
Together with the
fact that FLT has such a long history, I believe that the
reader understands why FLT has been chosen.
Before going through the history of the problem, a few
methodological remarks must be made:
(a) As sources for the material itself, I have mainly
used Ribenboim's 13 Lectures on Fermat's Last Theorem and
Edwards's Fermat's Last Theorem.
A Genetic Introduction to
Algebraic Number Theory.
If there were disagreements
between these
two
authors,
additional
sources were
consulted, among which of course Cantor's Vorlesungen uber
die Geschichte der Mathematik.
The presentation of the material is highly
(b)
interpretative.
By this I mean that the different stages I
distinguish in the development of FLT are not historical
facts (if they could ever be that) but constructions that
seem to capture the relevant elements of that development.
In other words, the account presented here is not the
account of FLT, but it does fit the facts.
(c) This paper has a very serious shortcoming (and it
is the author's hope that in the near future this can be
remedied): it focuses entirely upon the mainstream results.
But at several occasions prizes were offered for a solution
of FLT (1816, [Paris], 1850, [Paris], 1908, [Gottingen]).
These prIzes generated a stream of solutions and proofs.
But they are totally neglected in any history of FLT. No
doubt a lot could be learned from these mistaken proofs (the
1908 Prize may still be yours until September 13, 2007!)
because they
are
the
products
in
most
cases of
non-mathematicians who know little or nothing about the
mainstream actIvItIes.
Let us not forget that Fermat
himself would today be considered an amateur since he made a
living as a judge in Toulouse.
340
J.P. VANBENDEGEM
(d) Except for direct, simple proofs such as that of

GrUnert, most results will be presented either without proof
or with a rough outline of the proof.
Fortunately the aim
of this paper does not necessitate a detailed study of the
proofs
involved,
rather what counts,
are the major
ingredients of the proof and these can be presented without
getting involved in highly technical mathematical matters.
2.1. INITIAL PHASE: EXPLORATION BY EXISTING METHODS
In the
initial
phase of FLT,
attention was almost
exclusively concentrated on special cases.
Either one
proved that FLT holds for a small value of n, or one tried
to find a counterexample.
None of these proofs gave any
indication as to the nature of a general proof. Now, this
seems remarkable. What could it help to know that FLT holds
for say n = 3, n = 5 and so on, since what we are asked to
prove is that it holds for all n. The most plausible reason
seems to be that the mathematicians were not trying to prove
FLT, but exploring it. To which must be added the fact that
in the initial phase not all mathematicians were convinced
that FLT was an interesting problem. As late as 1816, Gauss
wrote to Olbers that "but I confess that Fermat's Last
Theorem as an isolated proposition has very little interest
for me, because I could easily lay down a multitude of such
propositions, which one could neither prove nor dispose of"
(2). Hence, why waste time trying to prove it! This is of
enormous importance, I believe, because it shows that
problems - in this case, propositions that need to be proved
are not considered as important solely by the fact that
someone formulated them. A problem must in a sense be
perceived by the
mathematica-l community.
It is very
plausible to claim that problems resisting being easily
solved, are good candidates to attract attention. This
feature on its own does not guarantee the continued interest
of the mathematicians. FLT was considered more important
when it became clear that solving FLT would lead to the
solution of other mathematical problems. As we go along, we
will see some examples.
As far as results are concerned, Fermat himself proved
the case n = 4, Euler proved the case n
3 (1822),
Dirichlet the case n = 5 (1828), Dirichlet the case n = 14
(1832) and Lame the case n
7 (1839). In all these cases
the proof was based on the same method: proof by infinite
descent.
The core of such a proof is the following: assume
341
that a solution exists, say (x, y., z). Then prove that
there must be a second solution, (x', y', z') such that [x]
> [x']'[y] > [y']'[z] > [z']. The argument can now be
repeated with the solution (x', y', z') infinitely often
(hence, infinite descent). But as all these solutions must
be whole numbers, this is impossible. Values cannot become
arbitrarily small within the integers.
But in order to
construct such a proof, it was necessary to introduce
certain algebraic identities, specific for the case in
question.
And exactly that prevented generalization to a
complete proof for all cases.
Note: Only cases in which n is a prime number need to
be considered, as all mathematicians involved with FLT knew.
For suppose that n = m.p, where p is a prime. And suppose
you have proved that x P + yP = zP has no solutions. But
then it follows that x" + y" = z" can have no solutions
either. Suppose it did, then we can rewrite x" + y" = z" as
(xm)P + (ym)p = (zm)p, but then x m, ym, zm is a solution of
x P + Y P = z P, which is impossible.
Therefore in what
follows, only prime numbers will be discussed as exponents
and always indicated by p (or p').
2.2. BREAKTHROUGH PHASE:
VARIATIONS ON EXISTING METHODS
(a) Gauss was the first mathematician to attack the problem
in a different way.
But not altogether that different.
His attention was restricted to the case n = 3, so, in that
sense,
we are still close to the initial phase.
But
Gauss did, was the following: all the solutions known up
what
to
then, used properties of rational numbers, combined with the
method of infinite descent.

In the period in which Gauss
lived, complex numbers became more and more accepted.
In
particular Gauss looked at complex numbers of the form a+
b . .[-3
rational numbers,
These numbers are clearly not
because of the presence of .[-3, but, so he found out, they
behaved very much like them. E.g., you could add them:
(a + b . .[-3) + (c + d . .[-3) = (a + c) + (b + d) . .[-3,
you could mUltiply them:
(a + b .~-3) . (c + d . .[-3) = (a.c - 3b.d) + (a.d +
b.c) . .[-3
and so on.
342
J. P. VAN BENDEGEM
Using these numbers that looked

very much like
rationals and still using the infinite descent method, Gauss
could prove the case n = 3. But the most important thing of
all was that the method could be generalized. Thus there
was hope for a general proof (although it later turned out
that the reasons for hoping so were mistaken).
On the one
hand, it is clear that Gauss' method was not that different
from the existing methodes), but this small shift of
attention was sufficient to
open new
possibilities.
As we
investigation was started.
The basic question was: suppose
will see, this pattern will return over and over again.
(b) Simultaneously with Gauss, a second line
there is a solution,
certain n, xn
solution have.
i.e.
of
a triple (x,y,z) such that for a
What properties does such a

Clearly this approach is a more obvious move
+
yn
zn
away from the initial phase.

The method used here might be
called the counter-example search method.
One assumes that
solution
exists
and tries
to
list
properties
such a
solution is supposed to satisfy. Let me give one example of

such a result: the following theorem has been proved by
Barlow in 1810:
if x,y,z are pairwise relatively prime (i.e. x,y,z
have no common divisor) and x P + yP + zP = 0 and p
2 is
not a divisor of 2 then there exists integers t and t1 such
that: x + y = t P , (x P + yP) / (x + y) = t~, z = -tt 1 .
Such theorems were proved by elementary methods, i.e.
the kind of methods that were available at that moment.
E.g.: it is straightforward that if x P + yP = zP then
zP = (x + y)(xP-1 - xP-2 Y + ... + yP-1)
Hence it follows that (x P + yP) / (x + y) is an
integer. This function of x and y viz. Qp(x,y) = (x P + yP)
/ (x + y) was extensively studied.
Although one might expect that these methods were used
to derive properties such that either a counter-example
could
be
effectively
constructed
or
such
that a
counter-example, if it existed, would have inconsistent
properties (hence proving FLT by reductio ad absurdum), this
was not the case. As it turned out, a French mathematician,
Sophie Germain, succeeded in 1823 to use these methods to
prove the following:
Consider the following subproblems of FLT:

- the first case: add to the conditions of FLT that p*x.y.z.
(i.e. p does not divide x, y or z)
343
the second case: add to the conditions that p I x.y.z and

gcd (x,y,z)
1 (i.e. the greatest conunon divisor of x,y
and z is 1, which is another way of saying that x,y and z
are pairwise relatively prime)
then the theorem says:
if p is an odd prime (p
2) such that 2.p + 1 is a prime
too, then the first case of FLT holds.
This result is of particular importance because it is
the first result - notice that it was proved in 1823, that
is almost two centuries after the formulation of the problem
that could cover an infinite number of prime cases. But
the problem is to find all those primes p such that 2.p + 1
is a prime too.
Unfortunately that problem is even harder
to prove than FLT itself.
But Legendre was able to refine
the results of Germain and instead of 2.p + 1, any prime
such that 4.p + 1, B.p + 1, 10.p + 1, 14.p + 1 or 16.p + 1
is a prime too, would do.
The net result was that FLT was settled for all prime
exponents < 100.
Or stated otherwise, a finite subset of
the first case had been settled. The most important feature
of the work of Sophie Germain is the fact that FLT is no
longer treated as a single problem.
Instead it has become
the union of two (clearly) related problems: the first case
and the second case. Although it is close to a trivial
remark to say that if a problem is very hard to solve, it
may help to split it up in parts.
(This sounds like an
advice Polya was likely to give).
Nevertheless I want to
draw attention to the fact that the way(s) in which the
problem can be split up are not arbitrary.
Given FLT,
several decompositions are possible, such as:
FLT = FLT for all exponents 5 n + FLT for all

exponents> n (where n is an arbitrary integer). This
decomposition
alone
generates an
infinite number of
possibilities
- FLT = FLT such that x has ten digits + FLT such that
x has not ten digits.
But
clearly
such decompositions are
of little
interest.
In short, what I want to say is that, given
certain proof searching methods or counter-example searching
methods, there are in some sense optimal decompositions
relative to these methods.
decompositions
mentioned
nevertheless excluded.
Secondly, even if we include the
above,
FLT is
some
decompositions
are
a statement about integers

Hence, if we want to split up FLT,
and powers of integers.

we will necessarily have to refer to properties
of integers
344
J. P. VAN BENDEGEM
and powers of integers.

It simply doesn't make sense to
split up FLT, say, in FLT such that e 3 = 8.97 and FLT such
that e 3
8.97 (3).
I will return to this problem in
section 3.
2.3. REORIENTATION PHASE:

FURTHER VARIATIONS ON EXISTING METHODS
(a)
The next result is due to Ernst
Eduard Kummer
(1810-1893). Contrary to what one might expect, Kummer's
work continued the line of attack started by Gauss. Kummer
had noticed that complex numbers did behave almost like the
rational numbers, but not in every respect.
Hence he
proposed to look at the subset of complex numbers that do
have all the properties, the ideal complex numbers. The
property that was needed above all others, was the unique
decomposition
property.
Fortunately,
this
can
be
illustrated by a simple example. Take an arbitrary integer,
say 130.
Then (as I expect you might know) this number can
be written in a unique way as a product of powers of prime
130 = 2.5.13. But for complex numbers this
numbers, viz.
nice property does not hold in general as the following
example shews, for 5 is now no longer a prime:
5 = (1 + 2.i).(1 - 2.i).
The problem is that restricting the complex numbers to
the ideal complex numbers generates a proof that does not
work for all exponents.
Hence Kummer proposed a new
decomposition:
the exponent is regular or the exponent is irregular.
The theorem he was then able to prove says:
If p is regular, then FLT holds.
Note that this decomposition does not coincide at all with
the division used in the theorem of Sophie Germain. But,
unfortunately, Kummer faced the same problem as Germain: it
is not easy at all to find out whether there are an infinity
of regular primes. Although there is a notable difference :
in the Germain case, if one would ask whether the theorem
holds for a certain prime p, then it is sufficient to
calculate 2.p + 1 and to check whether this number is prime
or not.
In the Kummer case, to determine whether a prime is
regular, you have to calculate a certain number hp

and to
calculate that number you need Bernouilli numbers. Now,
this is a new feature. For Bernouilli numbers existed
already in the mathematical literature and had already
shown their importance. There is a connection, e.g., with
345
the Riemann zeta-function, an important and crucial concept

in number theory.
Let me add some comment to this rather
cryptic statement.
Formally,
Bernouilli numbers are defined in the
following way:
take the function x/rex
1), write this function as an
infinite series
x/reX - 1)
The coefficients Bn are exactly the Bernouilli numbers
The Riemann zeta-function is the function
v (s) = E
Remarkably
n=l
l/n s , for s real and s
enough,
there
is
an
(4).
>
interesting
connection
between prime numbers and V (s), viz.

V (s)
fl 1/(1 - l/pS), where the infinite product
runs over the set of all primes.
And then the following
connection (5) is made
B2k = (-1 )k-l (2(2k) !/(2T<)2k) V (2k).
Apart from the fact that FLT attracted attention simply
because it resisted to be proved, links of this kind
increased the importance of FLT.
This connection is the
first of a series to follow.
Kummer continued his work now
focusing on the irregular primes and amazingly enough,

methods meet and merge, for using the methods developed by
him, he proved theorems of the following kind: if p divides
the numerator of at most one of the Bernouilli numbers B2,
B4 ,
BR-3' then the first case holds for the prime p.
(Bernouilli numbers are fractions, hence the reference to
the numerator of a Bernouilli number) (6).
(b) Parallel to this development, an enormous amount
of work was done to effectively calculate the Bernouilli
numbers and to check for which primes FLT could be settled.
The methods used here are basically refinements of the
existing methods.
Mathematicians such as Wagstaff, had no
intention to prove new results, but they wanted to render
the existing results fit for calculation.
The success of
this approach is clearly demonstrated by the fact that the
work of Kummer, computationally translated, allowed to
decide FLT for all primes up to 125000 (7). Such results,
as I said in the beginning of this paper, provide us with
J. P. VAN BENDEGEM
346
valuable
information
counterexamples.
The
concerning
the
result of Grunert,
size
of
possible
e.g., belongs in
this area. And so are the many partial results, such as

if x,y,z are such that y = x + k and z = x + 2k, then this
triple can never satisfy FLT.
The methods used in these partial results are often at
the
elementary
level
requiring
only
quite general
mathematical tools (the result above requires only a bit of
modular arithmetic and general algebra) (8). As independent
statements, these results are not important, but they can be
used for calculating purposes: suppose that you are actually
calculating a certain example and you find evidence that the
triple X,y,z forms an arithmetic progression.
Then this
result can be dismissed right away and thus saves expensive
computation time.
At this stage of the history of FLT, the matter has

become quite complex.
FLT itself is now the core of a
network of related problems and several methods are being
used for tackling these related problems and thus indirectly
FLT itself.
What happens is that the concepts and methods
Kummer
introduced,
extending and refining the
approach of
Gauss, start to lead an independent life so to speak. If

one takes a broader look at the mathematical field, this
suggests the idea that methods originate associated with
particular problems and then acquire a certain independence
thus being available for other problems.
A fine example
related to FLT itself, is that Gauss could work out his
solution of the case n=3 only because complex numbers were
available.
A century before such a thing would have been
impossible.
Complex numbers
were
'present'
in the
mathematical field not because of FLT, but because of other
problems in algebra and geometry.
2.4. REORIENTATION PHASE:
STILL FURTHER VARIATIONS ON EXISTING METHODS
Ca) In 1909 Wieferich published the following quite amazing
result :
if the first case fails for p, then p must satisfy 2 P-1 "
(mod p2).
(a " bCmod c) means that a can be written as the sum of b
plus a mUltiple of c, thus a = b + k.c, or otherwise said, a
b = k.c meaning that a - b is divisible by c. The result
of Wieferich states then that 2 P-1 - 1 is divisible by p2).
347
This result is extremely powerful because it does not

involve the integers x,y,z themselves.
Furthermore the
criterion is more fit for calculation than the Bernouilli
approach.
On the other hand, it is of course restricted to
the first case of FLT.
In contrast with the simplicity of
the result, the methods used by Wieferich are a collection
of almost everything that is available: not only Kummer's
work is used, but - and this is a new element - derivatives
and pOlynomials are introduced.
Mirimanoff and Krasner had
already established a large part of
this polynomials
approach, that itself can be traced back to methods Kummer
had introduced to deal with the irregular primes. Once the
proof of Wieferich was available, Mirimanoff proved an
analogous case, viz.: if the first case of FLT fails for p,
then 3 P_l " 1 (mod p2).
Note that these theorems although apparently giving a
property of a counterexample, if it should exist, are for
computational purposes to be read in transpositional form,
i. e. if 2 P-l '" 1 (mod p2) then the first case holds.
(b) This result was immediately accompanied by a
computational explosion and we know now that the first case
holds for all primes < 3 .10 9 It is interesting to remark
that Wieferich's theorem left two primes undecided, viz.
1093 and 3511, but these were settled with Mirimanoff's
theorem.
Various other conditions were derived from Wieferich's

condition combined with the already known computational
methods.
An interesting link was found between primes
satisfying Wieferich's theorem and Mersenne primes. These
are prime numbers of the form 2Q
1, where q is itself a
prime number.
Mersenne primes are important in number
theory and in finite group theory. I want to emphasize once
more that these new computational methods do not render the
existing ones redundant, for, as it turns out, in some
reformulations of Wieferich's criterion, Bernouilli numbers
play again an important role.
2.5. REORIENTATION: VARIATIONAL EXPLOSION
In this last section I will simply list a number of
approaches to FLT, without little or no detail, for, a few
exceptions put aside, the technicality of the methods used
is
impressive (even
for
the trained mathematician).
348
J. P. VAN BENDEGEM
Furthermore, if, originally one or two methods were used,

FLT occurs now in an impressive number
of different
mathematical fields (and even in logic).
(a) Furtwangler continued along the lines of Kummer,
Krasner, Mirimanoff and Wieferich.
He reformulated the
methods used in terms of a structure that is now known as
class field theory. Basically what was achieved here, was a
clearer understanding of the work of Wieferich. Proofs
became much shorter and easier to understand. There were no
new spectacular results.
Nevertheless Vandiver managed to
derive
work.
some
new computational criteria
from Furtwangler's
(b) A new line of investigation was initiated, based

on the following observation.
Consider xn + yn
zn.
Divide both side by zn: (x/z)n + (y/z)n = 1. Call x/z, X
and y/z, Y, then the equation reads: Xn + yn = 1. This
represent a curve in the plane (e ..g. if n = 2, then Xn + yn
1 represents a circle with radius 1).
What FLT says in
these terms is, in fact, that if n > 2, then the curve
passes through no rational points (9).
For if it did, such
a solution would be a solution for FLT, thus refuting it.
The most interesting results were not obtained by looking at
'traditional curves' (i.e. defined over the real numbers),
but by looking at curves in general field structures.
Hence, as long as these methods were not developed to a
certain degree, they were of little interest for FLT. But
once they could be applied, they have resulted in a new
breakthrough.
In 1983, Gerd Faltings proved Mordell's
conjecture.
This conjecture is a claim about the number of
possible rational points of a certain set
of curves
(including the Fermat curve) in a algebraic number field.
Actually the conjecture claims that number must be finite.
The proof implies that given a certain n, then there can
only be a finite number of counterexamples for xn + yn
zn.
I mentioned before the strange feature of FLT, viz. the fact
that as a proof seems within reach, counterexamples are hard
to
find.
Falting's
result increases
this feature's
intensity: counterexamples are of impressive size and rare
as well. Apparently you cannot have both at the same time
(at least not in the case of FLT).
(c) The importance of FLT can be judged by the fact
that in domains totally different from number theory (where
FLT originally was formulated), attention is paid to it and,
interestingly enough, sometimes with a completely different
motivation. Let me list here just two examples:
349
consider the following problem. You

partial differential equation to solve:
() n U + 8 n u = (}nU
b
xn
"yn
have
the following
zn
( a indicates a partial derivative)
(10). The question is

whether there is a function u (x,y,z) such that the
equation above is satisfied and that fulfills the following
conditions:
u has period 1 in each variable, i.e. u(x+1,y,z)
u(x,y,z) and likewise for y and z
- there exists aM> 0 such that, for all x,y,z:
bn
I " x
~(X,y'Z)I~ M'I~(X,y'Z)I~
M,I~(X,y'Z)I~
M
a yn
b zn
take the unit cube, that is, the collection of all

triples (x,y,z) such that 0 ~ x ~ 1, 0 ~ Y ~ 1, 0 ~ z ~ 1.
Then it is required that u vanish on two faces of the unit
cube (i.e.
the value of u (x,y,z) for x,y,z on the face of
the cube must be zero).
Given all this, the following remarkable relation can
be established: if n ~ 2 is even, the following two
statements are equivalent:
- there exist integers a,b,c such that an + b n = en
there exists a solution u (x,y,z)
for the above equation
satisfying the requirements listed.

The importance of relations of this kind is clearly
that in an indirect way,
by looking at the partial
derivatives problem instead of FLT, the whole machinery
developed to treat partial differential equations can now be
used.
The existence of such a relation attracts a whole set
of new methods unfit for the original problem itself.

The second example is of a completely different
nature.
FLT has also attracted the attention of the
logicians.
But what these logicians do, is not to try and
prove that FLT is correct; rather they want to show that the
problem cannot be solved at all.
This requires some
clarification.
Ever since the results of Kurt Godel, Alan
Turing and others, we know that there are problems in
mathematics that are essentially undecidable, that is, there
are statements S (in mathematics) such that there is no

proof for S, and no proof for ~S. Note that the issue is not
that we haven't tried hard enough, it is a matter of
350
J.P.VANBENDEGEM
principle.
If there would be a proof (for S or 'S) then a
contradiction would follow.
There is a beautiful example
that illustrates the problem (11). Consider all finite games
(chess, checkers, ... ). Consider now metagame the rules of
which are very simple: the first player makes a first move
by choosing a finite game. The other player then makes the
first move in the finite game, and the game proceeds until
it ends. Consider the proposition 'Metagame is finite' and
its negation 'metagame is infinite'. Surprisingly, metagame
is neither. Suppose metagame is finite. Then the first
player may select metagame. The other player then selects a
finite game, say, metagame and the first player ... So
clearly metagame is infinite. But if it is, then the first
player cannot choose metagame, but must select a finite game
and thus metagame always ends. Hence it is finite. Thus, we
must conclude that metagame is neither finite nor infinite
in principle. The first statements that were shown to be
undecidable were highly technical and far removed from daily
mathematical practice, but, later on, it was demonstrated
that problems of a more concretenature (whatever that means
for mathematicians) were equivalent to these undecidable
statements.
Hence they were undecidable too.
One such
problem was the search for a general solution to the
Diophantine problem. Roughly speaking, Diophantine problems
involve polynomials with integer
coefficients and the
question is to determine whether solutions in integers exist
or not.
Now, the general Diophantine problem has already
been shown to be undecidable (12). FLT is a special case of
a Diophantine equation. Could it be that FLT is undecidable
too?
This is precisely the question logicians are working
on. If this attempt would prove successful, it would imply,
cynically enough, that all the previous time spent in
searching for a general solution was wasted.
Note that if
it were to turn out that FLT is undecidable, this does not
exclude that special cases can be settled, as we have seen.
This kind of research or
approach
would
have been
impossible, say, fifty years ago (Godel published his famous
paper in 1931) but, today it is part of the spectrum of
mathematical activities that, either you try to prove
something, or to prove that it is wrong, or you try to prove
that it is undecidable.
Tracing back this undecidability
approach, we must return to Godel, whose work was based on
the work of Hilbert.
His project was to
show that
mathematics is complete and consistent. He wanted to do that
because there
were problems
in
the
foundations of
351
mathematics, in particular, problems with the foundations of

analysis (what is a real number, a derivative, etc.) and
with transfinite number theory (the theory of infinite
numbers, initiated by the work of Georg Cantor). In short,
the point is that these new methods were developed in other
areas of mathematics (i.e. not directly related to FLT) and
only then applied to FLT.
I labeled this section 'Variational Explosion' and
that is exactly what it is. The selection of examples above
is only a small set. Various other formulations have been
put forward and as far as I know, Faltings' result is the
latest important result concerning FLT. Before leaving this
section, I want to emphasize once more that the material
selected here, is only a part of the full history of FLT.
The conclusions I will derive from this material in the next
section are therefore rather rough and somewhat vague.
Filling in the details, I take to be a separate problem. The
first aim of this paper was to identify a pattern (if any).
Leaving out some details very often facilitates the process.
In addition, it made it possible for me to present the
history of FLT to a public of non-mathematicians.
3.
NEED IT BE MENTIONED THAT LAKATOS' SCHEME

WILL NOT HELP US?
As said before, there is no proof in the very first place,

counterexamples are hard to find
(if not practically
impossible) and there really seems to be no need to look at
the hidden or suspect lemmas for there aren't any around.
This negative view on the splendid work of Lakatos does not
imply - not in the least as a matter of fact - that I reject
his scheme.
As I will show, this scheme is a special case
of the scheme I will suggest below. But I do reject it as a
candidate for a general scheme.
I first present the ingredients of the general scheme
and then the relations between them.
3.1.
Obviously the first thing we need are the
problems.
But as must be clear from the history of FLT, we
cannot work with problems only. A refinement is required.
For as we have seen, if the original problem resists being
proved, the problem is split up in several subcases that
together constitute the original problem, such as e.g. FLT
seen as the union of the first case and the second case, or
FLT seen as the union of the regular prime case and the
352
J.P.VANBENDEGEM
irregular prime case.

Such a refinement
shall call a
P-complex.
At a given moment as is the case for FLT,
several such P-complexes may exist simultaneously. I shall
call this a set of P-complexes. Are P-complexes arbitrary?
I think not as I already suggested in the previous section.
Consider once again FLT.
This problem deals with integers,
powers of integers, prime numbers, addition and so on~
These 'things' and these operations have certain properties,
such as 'a natural number is either even or odd' of 'the sum
of an even and odd number is an odd number'. This whole set
of 'things', operations and properties I call the make-up of
the P-comp 1ex.
.
Thesis
:Changes in a set of P-complexes consists
basically in a reshuffling (or redistribution) of the
make~up of the different elements of this set.
Example: the two P-complexes (first case, second case)

and (p regular, p irregular) may merge into a new P-complex
(first case & p regular, first case & p irregular, second
case).
FLT supports this thesis.
Besides this argument,
there is a stronger one: no matter how we split up the
original problem, we cannot as such change the make-up of
the problem.
These ingredients must remain the same. Of
course, one might argue that in the long term, these
ingredients themselves change as well.
Natural numbers are
not the same 'things' they were back in the 16th century.
And I agree, but what matters is that the change in the
ingredients is slower compared to the change
in the
P-complex.
Otherwise said, as is the case in FLT, the
make-up can be considered constant relative to the changes
in the P-complex.
Let me use a simple image to clarify the idea:
astronomy tells us that stars have no fixed positions in the
sky.
They slowly move about.
But, if I need a particular
star, say, as a guide at sea, then, given my life-time, I
may very well assume that the star is fixed.
If however we
would be discussing the history of navigation at sea by
reference to the stars back to the Babylonians, then of
course the change in position may play an important role.,
An additional advantage of the notion of make-up is its high
pragmatical value. In order to treat a problem such as FLT,
it is not
necessary to invoke
explicitly the whole
mathematical universe.
353
The ingredients necessary in order to understand the

history of FLT are captured in the make-up. Their specific
formulation depends on the problem itself. Thus, in the
original phase of FLT, the make-up required only well-known
features of natural numbers. There was no need to present
the natural numbers in an axiomatic treatment (such as Peano
arithmetic). Surely, there was no need to analyze the
natural numbers in terms of some more abstract theory (such
as
set
theory).
In a later phase
e.g. in the
undecidability approach - it did become necessary, but in
that case too, the make-up did not involve, say geometrical
or topological properties. Although it cannot be excluded
that, hypothetically, there could be a problem that does
involve the whole mathematical universe, such a problem
would have to be very complex indeed. For most cases
however, we do not run into this difficulty. Hence we do not
share e.g. Nickles' opinion that the full background of a
problem must in the end include everything (13).
3.2. A similar story can now be told for the methods.
Here too we see that new methods arise from old methods by
remixing the ingredients and (possibly) adding some new
ones. The recurrence of the method of infinite descent is a
perfect example. This method can then be mixed with methods
for dealing with complex numbers and this is precisely the
method Gauss used for dealing with the case n=3. There
seems to be no problem to speak about M-camplex, the make-up
of a M-complex and of course of a set of M-complexes.
Although I must admit that the identification problem for
methods may be a lot harder to solve than for problems.
What we are lacking at the present moment - or at least is
not well developed
is a catalogue of methods. The
emphasis on the use of formal logic may have been the cause
of us forgetting that there is more than
the basic
distinction: direct proof and reductio ad absurdum.
If we
take, e.g., the problem of solving a set of two equations in
two unknowns x and y, then there are at least two methods:
substitution (calculate y in function of x using the first
equation and
substitute in the
second
equation) or
determin::t: ::f=t:e
::::t:o:i:a:i~l:
:1 and y =1:
:1 / 1: :1)
354
J. P. VAN BENDEGEM
I consider these methods to be different, although both

proceed by a direct construction of the solution. In terms
of the concepts introduced here, the reason is simply
because these two methods have a different make-up.
I therefore pte sent the second thesis in analogy of
the first one:
Thesis 2: Changes in a set of M-camplexes consist
basically in a reshuffling (or redistribution) of the
make-up of the different elements of the set.
3.3. Those are all the basic ingredients we need. The
question now is, of course, what relations exist between a
set of P-complexes and a set of M-complexes. Theses 3 and 4
summarize these relations:
Thesis 3: A set of P-camplexes remains in existence if
an element of the set establishes a S-link with an element
of a set of M-complexes which thereby remains itself in
existence.
Thesis 4: A set of P-camplexes remains in existence if

an element of the set establishes a C-link with an element
of a set of P'-complexes. The same holds for a set of
M-complexes.
A S-link is quite simple: its means that a method has
been successfully applied to a problem. In other words, the
problem is solved either positively (it is proved) or
negatively (it is refuted).
A C-link is equally simple: it means that if as-link
is established with either P or P' then this S-link is
transferred to the other (although not necessarily in a
complete way, it may suggest as-link).
C stands for
cooperation because if a C-link exists, it guarantees the
remaining in existence of the connected P-complexes if one
of them has succeeded in establishing a S-link with some
M-complex. Several remarks need to be made. First of all,
note that I did not put an 'iff' in the statements.
That
appears clearly to be too strong.
In the case of FLT, we
have seen that e.g.
the P-complex introduced by Sophie
Germain had one S-link which was not reinforced - i.e.
there were no further S-links established within the same
355
P-complex - nevertheless that particular P-complex did not

disappear for Kummer used it in his study of irregular
primes.
Secondly, that C-links exist is evident from such
examples as the connection between FLT and the partial
derivatives problem. One might wonder why I do not consider
such problems to belong to the same P-complex. The reason
is quite simple: the elements in a set of P-complexes share
the property that their make-up is more or less identical.
But a problem having to do with partial derivatives has a
totally different make-up.
Hence it belongs to another set
of P-complexes.
Thirdly, when I was going through the history of FLT,
the most amazing feature to me, was the relative constancy
of the development of FLT. Although over a period of two
centuries, an enormous growth had taken place, still,
locally, this growth was quite gradual and modest. I
therefore felt it necessary that any model should be able to
take account of that strange feature.
Now, the model
presented here does precisely that through the concept of
the make-up of a P-complex. The notion itself of a make-up
does not appear to be artificial, something introduced to
explain that itself requires more explaining. The make-up of
a complex is
between the
a mechanism that produces a 'natural' link

problem
and its environment,
viz.
the
mathematical universe.
In that sense, the concept of a
make-up carries explanatory power.
Finally, the model explains why the classical view of
mathematics does not hold, viz. once a problem is solved,
that is it.
If you have a proof or a refutation of your
theorem, you can start to attack the next problem. But,
strangely enough, mathematicians spend a great deal of their
time, rewriting existing proofs, polishing, cleaning up,
trying to find shorter proofs, etc.
Why are they doing
this?
The answer in terms of the model is straightforward:
it guarantees remalnlng in existence.
Take e.g., FLT. A
great deal of the work of Furtwangler was a rewriting of the
proofs of Wieferich, Krasner and Miromanoff. What does the
model say? Wieferich establishes a S-link between a certain
P-complex and a certain method. If a S-link can now be
established between the same P-complex and another method,
then this increases
the continued
existence of that
P-complex because it now has two S-links.
In short, the
more different proofs, the better! I consider this a clear
356
1. P. VAN BENDEGEM
advantage over classical models (certainly the

cumulative models) that fail to explain this large
the activity of mathematicians.
linear,
part of
3.4. How does all of this relate to evolutionary

epistemology (14)?
Obviously, the various terms introduced - P-complex,
make-up, M-complex, S-link, C-link - lend themselves to an
interpretation in terms of an evolutionary model. In a first
draft of this paper, I equated somewhat naively the triple
<make-up, P-complex, reshuffling> with <genes, organism,
variation>.
However, several other identifications are
possible such as <genes, genotype, recombination>. Although
this shows that an unproblematic and unique interpretation
is not likely, it does show that plausible candidates are
available. As a matter of fact, the elements of the model,
taken separately, relate to elements of various models in
evolutionary epistemology. The careful reader must have
noticed that I was careful to device a vocabulary not
involving terms of evolutionary epistemology as I wanted to
avoid selecting a particular interpretation. If a 'P- or
M-complex' is replaced by 'a network of problems or methods'
then clearly the model relates to the work of e.g. W.
Wimsatt (15).
Problems (or methods) are not isolated
entities, but are part of a network. The connecting lines in
the network would then be the C-links.

The
phrase
'remaining
in
existence'
replaces
'increasing its
survival value'
in
the first draft
mentioned. It is perhaps somewhat odd to associate the
survival of a problem to its being solved or not. However,
it is plausible to assume that an isolated problem remaining
unsolved for a long time, is not likely to hold the
attention of the mathematicians.
C-links are important precisely to prevent problems to
disappear if they resist being solved. Furthermore, if a
problem is solved, this does not imply that the problem
disappears. A solved problem indicates a S-link with some
M-complex, which through some C-link might prove to be
important for an as yet unsolved problem.
Notwithstanding the numerous similarities,
it is
obvious that the model presented here is not well enough
articulated to justify
the
choice
of
a particular
evolutionary model (16). Nothing has been said about the
carriers of the mathematical ideas, viz. the mathematicians
themselves. The story presented here takes place in the
357
realm of mathematical problems' and methods. What the model

fails to capture is e.g. the increase in survival value of a
problem if the same S-link is established in different
spatio-temporal
regions
(i.e. different mathematicians
finding/constructing, the same solution). Such aspects are
totally
neglected
in this formulation
and
must be
investigated further.
3.5. To end this section, how does the model relate to
the model of Imre Lakatos?
His method of proofs and
refutations fits with the model in the following way. The
model does not stipulate that the C-links should be formal
links - in the sense of a stepwise formal deduction - nor
does it say that the problem should be unambiguously
formulated.
This is important for it leaves open the
following possibility: a P-complex is derived from the
original problem, but the P-complex does not cover the
original problem completely. This means that if a S-link is
found between the P-complex and a particular method, this
success will not necessarily transfer to the original
problem.
This, in a nutshell, is what happens in the case
Lakatos studied.
The whole discussion about the Euler
conjecture - the statement that V - E + F = 2, where V is
the number of vertices, E the number of edges and F the
number of faces of a regular polyhedron
is basically the
discussion to what P-complex the S-link applies.
Does this
follow from our four theses?
Yes, it does. Take a set of
P-complexes and suppose that for one element of the complex
a S-link has been established.
The continued existence of
the set of P-complexes is guaranteed if more elements in the
set of P-complexes are S-linked.
Hence it is in accordance
with thesis 3, that once a S-link is found, it should be

extended over the other elements of that set.
As a specific example, let me discuss the first rule
of Lakatos' method: Rule 1. If you have a conjecture, set
out to prove it and to refute it.
Inspect the proof
carefully
to prepare
a
list of
non-trivial lemmas
(proof-analysis);
find
counterexamples
both
to
the
conjecture (global counterexamples) and to the suspect
lemmas (local counterexamples) (17).
What does this rule become in the model presented
here? A possible transcription might read:
Rule 1.
If you have a problem P, try to establish a
S-link (whether negatively or positively).
If a S-link has
been established, reshuffle the M-complex that is involved
358
J. P. VAN BENDEGEM
in the link.
Try to find a P-complex such that the S-link
no
longer
holds
for
the
whole
P-complex (global
counterexample) or try to find a M-complex such that the
S-link no longer holds between that M-complex and the
original problem (local counterexample).
The first part is rather obvious, but the second part
requires some explanation.
If the original problem is
reshuffled and the S-link no longer applies, this can only
be because there is an element in the P-complex that resists
the S-link.
This is precisely what Lakatos calls a global
counterexample.
If on the other hand, the S-link is lost
because the method is reshuffled this can only be because
some element of the M-complex resists the S-link and this is
a local counterexample.
The specific element of the
M-complex that is the cause of the breakdown of the S-link
is what Lakatos calls the suspect lemma. The reshuffling is
precisely the way of making clear what the hidden lemmas
are.
This analysis has a very important consequence and I

formulate it here as a corollary to the four theses:
Corollary.
The establishment of a S-link between a
P-complex and a M-complex does not imply that the complexes
involved are no longer subject to reshuffling.
The other rules can be rewritten ln the same way.

Thus one may conclude that the scheme of Lakatos can be
incorporated in the model presented here. Hence my original
statement that I do not claim that Lakatos's scheme is
wrong, it only needs supplementing.
Otherwise said, the
model is an extension of his scheme.
Which is a different
way of formulating the aim of this paper.
4.
It is obvious that
CONCLUSIONS
the best test for the model
is
to look
for other examples and see if these too fit the model.
After all, that is precisely the procedure followed in this
paper.
I have investigated a particular example - Fermat's
Last Theorem - and used it to criticize and extend Lakatos'

model.
When I presented this paper at the November 1984
conference ln Ghent
on Evolutionary
Epistemology, an
excellent suggestion was made: the four-color theorem (18).
This theorem has the fascinating property that a S-link was
359
established some years ago, but with the strange effect that
some mathematicians accepted it as a S-link and some did
not,
The problem was that the proof involved the use of a
computer (some two hundred thousand quite complicated maps
had
to be colored
effectively)
and that
for some
mathematicians such a proof is not a proof since it is not
open to control by individual mathematicians,
In terms of
the approach outlined here, this means that the nature of a
S-link needs specifying in the very same way that the ways
in which the reshuffling takes place needs to be specified.
NOTES
This paper was completed during a stay at the Center for

Philosophy of Science, University of Pittsburgh. Thanks to
all the colleagues who criticized the ideas presented here,
especially the Center's director, Nicholas Rescher. Also
thanks to Ernest Meulepas and Werner Callebaut for their
thorough analysis of the first draft. The author is.a
reshearch fellow at the Belgian National Science Foundation.
1. Euler's theorem states that for any polyhedron

V - E + F = 2, where V is the number of vertices, E the
number of edges and F the number of faces. A typical
counterexample is
satisfy V removed:
E +
+2,
particular
I
I
polyhedron
.,...----'"
,LL--l :
I':
I
I
'I 'I
I
: i,/}---~-7
I"
I
I'
I'
I'
I'
I'
that
such as a cube with an
~------"
1-------
For this polyhedron, V - E
does not
inner cube
J. P. VAN BENDEGEM
360
2.
See Ribenboim (1979), p.3.
3. e stands here for the base of the exponential

function.
Formally speaking, e can be defined in a number
of ways, such as
e = lim(I+I/n) or e = ~ l/n!.
n=O
->00
It is known that e is an irrational number (a real number,

but not an integer, nor a fraction), hence its (possible)
relation with FLT is clearly minimal. Furthermore, since an
actual (finite) calculation can decide whether e 3 = 8.97 or'
not, this splitting up of FLT does not make any sense at
all.
4.
To give the reader a flavor of the difficulties
involved in working
with Bernouilli numbers,
I will
calculate the first one, B,. To expand the function x/(e x-1)
into
an infinite series,
one uses Taylor's
theorem, which
says that a function f (x) can always be written as

00
n=O
(dnf/dx n I )xn /n!.
This is called a Taylor-development of f(x).

If one
compares this Taylor-development with the formula in the
text, then a simple identification tells us that
Bn = dnf/dxnl , where the right-hand-side is the nth
derivative of f(x) at the point O.
And here the problem
starts.
To calculate 8"
we must calculate the first
derivative of f(x). That is easy enough:
df/dx = (eX - 1 - x.eX)/(e X _1)2.
I assume the reader is still familiar with the rules
for derivatives, such as dex/dx = eX, d(u.v)
du.v + u.dv,
and so on.
But if we now take x = 0, then we find df/dx =
0/0.
You can avoid this indeterminate answer by applying
the following rule: if lim g(x)/h(x) = 0/0,
x->O
then take the derivative

calculate lim (dg/dx)/(dh/dx).
x->O
of both g(x)
and hex) and
If the result if meaningful, then it is equal to the

original limit.
There are certain restrictions on this
rule, but these need not bother us here. Calculating the
derivatives of both numerator and denominator, we find
df/dx = (-x.e X )/(2.(e X - 1).e X ).
361
Unfortunately, for x = 0, the result is still 0/0. So

we apply the rule once again to find:
df/dx
(-eX -x.e x )/(Z.(e X-1).e X + Z.ex.e X). This
result is determined, for we find df/dxlo = (-1)/2 = -1/2.
And this was only the first term!
To give an
impression of the numbers involved, B34 = 2577687858367/6.
5.
In the formula for B2k, (2k)!
stands for the
product of all natural numbers, starting with
and up to
and including 2k. Thus:
(2k)! = 1. 2.3. . ... (2k-2). (2k-1). 2k.
6.
see Note 4.
7. The upper bound of 125000 was obtained as recently

as 1976 by Wagstaff.
It is important to realize that the
arrival of new computation methods did not make the previous
methods superfluous. Most of these techniques were not
abandoned but
refined,
e.g.
by the introduction of
computers.
So even after the introduction of the methods
used by Wieferich, leading to a simpler criterion, the
method of
the Bernouilli numbers was
not considered
out-dated.
8. With the same motivation in mind as in footnote 4,
I want to present a proof so that the reader can judge for
him (her)self whether the proof is indeed to be considered
elementary.
Proof: if x,y,z form an arithmetic progression, they
can be written as x, x+k, x+2k. Hence the equation becomes:
x n + (x+k)n = (x+2k)n
The right-hand-side can be written out as a sum (binomium of
Newton) :
n
xn + (x+k)n = ~ C.xm. (2k)n-m

m=O
This sum contains the special
Hence the sum can be written:
n-l
x .
+ ~ C.xm. (2k)n- m
m=O
Eliminate xn from both sides and observe that on the
right-hand side each term of the remaining sum has a factor
k in it, hence it is divisible by k. But then the left-hand
side must be divisible by k as well.
Thus (x+k)n is
J. P. VAN BENDEGEM
362
divisible by k. But the power of a number is only divisible

if the number itself is, hence x+k is divisible by k, and
that can only be if x itself is divisible by k, hence x =
a.k, where a is a certain number. The original equation, .if
x is replaced by a.k becomes (x n can be eliminated):
an + (a+1)n = (a+2)n
a is either even or odd.
But if a is even then a+1 is odd,
so then sum of an + (a+1)n is odd, but (a+2)n must be even,
hence a cannot be even, so it is odd.
We can then write a
as
2b-1,
reads:
where
b is
a certain number.
The equation now
(2b-1)n + (2b)n = (2b+1)n.

A similar reasoning about divisibility can be made: apply
the binomium to both sides, and the equation can be
rewritten as:
2b.n - 1 + (2b)n = 2b.~ + 1.

n and ~ stand for all the other terms in the binomium (e.g.
if n = 3, then
(2b _1)3 = (2b)3 -3. (2b)2 + 3.2b - 1 =
2b 2b)2 - 3.2b + 3) -1)
or: 2b.n + (2b)n - 2b.~ = 2.
The left-hand side is divisible by 2b since every term is,
thus, the right-hand side must be divisible by 2b as well,
but that can only be if b = 1. The equation is now reduced
to:
1+2n=3 n.
But this is impossible, because
1 +2n=l n +2n< (1 +2)n =3 n forn>1.
To see the truth of the inequality, it is sufficient to
write out (1+2)n to see that besides other terms, 1 and
2 n occur in it as well. Hence the right-hand side contains
more terms (which are all positive). QED
9.
Take as an example, the curve x 3 + y3 = 1. Since
we know that this equation has no solution in rational
numbers (except for the trivial values x = 0, y = 1 and x =
1, Y = 0) this means that the curve does not pass through
00, ,"c;0..1 po;oc.
<h. ~f0110W;",,"C1""
363
The curve separates the p~ane in two regions and

contains nothing but irrational p01nts. Given the fact that
the distribution of rational and irrational numbers in the
plane is very dense, it is quite amazing to find a rather
simple curve that manages to meet only irrational points
(except for the trivial values).
10. A partial derivative can be understood roughly as
the extension of the ordinary derivative to functions with
more than one variable. Thus, the formal definition is:
"f(xl' ... ,xi' ,x n )/ " xi = lim (f(xl' ... ,xi+h,
h->O
.... x") -:- i(x".: .x i ~ ... x~))/h. .

The 1th part1al der1vat1ve 1S noth1ng but the ordinary
derivative holding all other variables constant.
40-41.
11.
This example is presented in Smullyan
12.
See,
Steen (1978).
13.
e.g.,
(1983), p.
the contribution of Martin Davis in
See Nickles (1980).
14. Most comments in this section are the result of a

discussion with Werner Callebaut, to whom I express my
thanks.
15. See e.g. Wimsatt, (1981a).
16.
See
the
position
paper
in
this
volume
for
references and details of these various models.
17.
See Lakatos (1976), p.50.
18. This suggestion was made by Ron Giere. Since then

the list has been steadily growing: Goldbach's conjecture,
the
parallel
postulate
in
Euclidean
Geometry, the
Banach-Tarski paradox, ...
LANGUAGE AND EVOLUTIONARY OR DYNAMIC EPISTEMOLOGY

Fernand Vandamme
1. INTRODUCTION
This
paper
epistemology
touches
on
certain
central
issues of
in relation to development and evolution. We
shall use language and linguistics as heuristic means.

Finally we intend to take into account these heuristic bases
to
bring up
certain suggestions
for
the
development and
construction of an evolutionary or dynamic epistemology.

2. EVOLUTIONARY AND/OR DYNAMIC EPISTEMOLOGY
The term "EE" is ambiguous. "Evolutionary" can be considered
as (i)
a modifier of the
object
specification of
noun epistemology,
the noun.
or as (ii) an
In the latter
case
we
are considering an epistemology, treating the problems

or
the field of evolution. What type of epistemology do we
need in order to understand evolution,
development etc.? In
the former case

we are considering the field of evolution
within
epistemology.
Both
fields
certainly
are
philosophically very important.
We think most authors have in
mind the object
specifying interpretation. It can be argued that (i) is a
special case of (ii). In the introductory paper to this
volume the editors refer to yet two
other interpretations
of EE:
In the first place: EE describes and explains man's
common-sense cognitive abilities and the
emergence of
science in terms of his biological adaptation to his natural
and to his self-created environment. An especially strong
form of such an EE is one which explains the construction,
validation and justification of the claims and methods of
actual scientific knowledge in terms of the operation of a
combination of two mechanisms: (i) blind, i.e. unjustified
variation and (ii) selection (cf. Campbell).
In the second place: EE is based on the idea that all
living systems are knowledge systems and that
all forms
of biological cognition share some essential features.
365
W. Callebaut and R. Pmxfen (edl.), Evolulionar.v Epislem%KY. 365-380.
/987 hy D. Reidel PuhlishinKCompany.
EVANDAMME
366
Both views just quoted suggest a third interpretation

(iii). For it is not so much the epistemology that is the
theory of knowledge, but knowledge which is at the centre of
the attention of EE.
EE is therefore in this third
interpretation a specific
theory
about knowledge.
In this
view EE is neither (ii) an epistemological theory about

evolution in general (the object interpretation), nor a
theory about the evolution of knowledge. In the best case it
can be considered to be a specific subcase of (ii). But this
depends on the degree to which a theory about the evolution
of knowledge is developed in a full-fledged epistemology.
This implies that a stand 1S taken on the classical
problems
of the
evaluation,
the
justification, the
discovery and the application of knowledge.
One can remark that in
this (iii)-interpretation, the
evolution of epistemologies~
in its turn, can be considered
as a special case of the development of knowledge. But even
then it is important to differentiate the object from its

theoretical level. One can study Freud's psychoanalytical
theory, as well as Einstein's relativitytheory, from a

sociological perspective. Such a study, whatever its merits,
does
not justify a reductionism
per se.
The
same
can be
argued here.
It is clear that if we stu~y.knowledge in general as

a changing object, it seems prom1s1ng to use the biological
paradigm to see
(i) if the same mechanisms hold which seem
useful in the explanation of
and selection)
and
biological
systems (variation
(ii) if certain laws for the evolution
of knowledge can be determined.

It is clear that dynamic
epistemology in general is a
useful perspective in this
respect.
Although we are in favour of such an outlook, we would

warn against reductionism.
The
Kitchener
work
~
of
Ekeland,
Thorn,
Prigogine,
Popper,
to mention only some authors - illustrates that
the evolution paradigm itself presupposes an epistemology.

What is more, it introduces many as yet unsolved problems,
or unsolvable problems at least for the time being. since
these problems are central,
the evolution paradigm is
certainly
too weak as a reductionistic base for the whole
of knowledge.
We think the epistemological problems
deterministic
example,
character of
laws
are
more
related
to the
peculiar. For
is it possible to know or describe a deterministic
process beyond a certain degree of complexity?
The
point we
want
to make is
367
that the evolutionary
paradigm applied to knowledge is certainly very interesting

provided one does not reduce the whole of epistemology to
it.
At the minimum
what we certainly must be conscious of
is that "knowledge",
at
least
In
Its
social and
psychological dimensions, is always changing according to a
certain context,
or
a certain function. Whether change is
conceived as development or evolution

will depend on the
empirical data (1) and on our theory about development,
evolution etc. We prefer to call an epistemology which
accepts and studies the changing of knowledge
dynamic
epistemology. Dynamic epistemology is in this
view a
specific
subdomain
epistemology then is
epistemology.
of
epistemology.
viewed as a standpoint
An
evolutionary
within dynamic
2.1 DETERMINISM OR INDETERMINISM

When we look at the dynamic aspect of knowledge, inevitably
we think about laws. Naturally the problem of determinism
versus indeterminism arises.
In his book Le calcul, l'imprevu, Ekeland gives a nice
overview on how the deterministic optimism
of
18th century
science slowly but surely has been modified and weakened.

That the classical deterministic view which accepts the
symmetry of time proved to be untenable is true even for
certain cases of hypothetical determinism. The symmetric
feature of the classical deterministic world
view
is that,
given the laws and the positions of the relevant elements at

a certain moment, it is possible to know the position of the
elements at each moment in the past or the future. Past and
future are symmetric. Both are completely determined. The
works of Poincare, Adors and others have shaken our belief

in the symmetry of past and future.
But things are
worse. Not only symmetry is broken.
Even when the laws and the relevant data on the elements are
known at a certain moment, it is not always possible exactly
to know their value in the future.
Ekeland stresses that the new knowledge gathered is

qualitative
rather
than
quantitative.
Already,
the
probability approach in science is the illustration of such
a qualitative approach; another one is Thorn's catastrophe
theory.
EVANDAMME
368
In fact what Thorn's catastrophe theory does, in

Ekeland's view, is to descrihe the possible developments a
dissipative system can have, i.e. along what ways it can go
from a certain stable equilibrium towards another stable
equilibrium. Such a transition is a catastrophe. Under
certain conditions a certain type of basic
forms of
catastrophe structures are possible.
These
basic forms are
by now world-famous. They are called: "queue d'aronde,

ombilic hyperbolique, ombilic elliptique". etc.
Ekeland adds two remarks.
1. This is a qualitative approach. It doesn't
allow to
predict phenomena, but only to classify them by indicating
the stage they are in, and the series of stages they are
most likely to be included in.
2. Time is eliminated from this theory.
Thorn's approach can be placed in the age-old attempt
of the reduction of time to space and the substitution of
movement by geometry.
I quote Ekeland here:
"That's what the theory of catastrophe does,
unfortunately in
too limited a domain. When
the theory is limited to dissipative systems,
the
simplest of
all
dynamic
systems, a
mathematical coherent model of deterministic
systems can be derived, which can be qualified
as creative:
for the model will
neither repeat
itself (hysteresis), it rather will impose order

on form (morphogenesis). This happens while
eliminating time from the constructions. The
architect locks himself up in his building. Time
stays outside and it is only his statue that
thrones in these wide icy palaces. Time is
chased by
the theory of catastrophes from the
very first
minute,
by
decision
of only
restralnlng of the evolution of dissipative
systems, their state of squilibrium. It is
reducing dynamics to statics; the dynamics of
dissipative systems, although poor, contains
nevertheless a number of
as
is
witnessed
reappears later on,
by
interesting phenomena,
thermodynamics.
as the fourth
Time
dimension of
space-time, where a catastrophe is growing. This
geometric image,
reflection
of an irreversible
and fugitive time, evokes another one, the

ellipse of Kepler. The elementary catastrophes
369
of Thorn, like the ellipses of Kepler, are

endeavors in trying to include time into space
and
seize it by
geometry. Whereas Kepler
constructs with mathematical tools, inherited
from the Greeks, Thorn benefits from modern
topology.
The former uses the 'Traite des
coniques' of Apollonius, the latter the theory
of singularities of functions." (Ekeland, 1984,
p.129. )
The equilibrium of a dissipative system, or the occurence
of a catastrophe,
can easily be compared with
a biological
stage, or with a psychological stage of development. The

link with Piaget is easy. But a lot of problems stay open:
1. How to define, how to determine and how to justify a
stage?
2. How does the transition from one stage to another happen?

Is it completely at random, or are there qualitative
regularities?
3. What about freedom: how is it possible to produce the
impression of freedom or at least partial independence from
the past by the present, when taking a future action?
Is
this only an epistemological indetermination, does some
kind
of
ontological
indetermination
occur
or
differentiation between both practically impossible?
is
3. LAWS AND DEVELOPMENT

When discussing problems of evolution and development,
one necessarily meets the task of differentiating plain
change
in the course
of events and
the notion of
development and evolution. It 1S true all development
implies change.
But the reverse is certainly false: not all
change implies development.

The notion of development implies the notion of change
together with
a notion of progress. Inevitably one has to
specify somewhat more what notion(s) of progress one has to
take into account here.
Is not the notion of progress frequently defined in
terms of a function of the present and past complexity of a
certain structure or in terms of nearness to a teleological
target?
The teleological tradition of interpreting development
has its benefits but also its drawbacks. One of the
hypotheses
we want to make is
that
a teleological
interpretation seems adequate as far as the ontogenetic
EVANDAMME
370
dimension
is
concerned,
but
is
plainly
applied to phylogenetic development,

cognitive and symbolic field.
Before we elaborate this
point of
stage
in,
at
erroneous when
least
in
the
view, we introduce
some differentiations and comments about development by

Kitchener.
Quoting Woodward (1980), Kitchener introduces
three different kinds of explanations on development:
"(1) The behavior B of an organism is explained
by reference to a
particular developmental
the
organism
is
e.g.,
Adolphe is
rebellious 'because he is in "that adolescent

stage." Here we have a type of performance
explanation.
In
cognitive
developmental
psychology
and
linguistics,
however,
the
behavior of a child is often explained by
reference to the
structure,
which
able
task
to perform
and
not
underlying
shows how
a certain
others.
This
cognitive stage
the person
is
kind
is
of cognitive
competence
explanation (one variety of a

how-possibly
explanation) .
(2)
A particular stage 5 is explained by
reference to a stage law by means of which the
system is developing (Woodward's developmental
explanation) .
(3) A stage law (e.g., 51-52-53) is explained
by reference to a more general developmental
principle or mechanism, which itself is not a
stage
law.
For
example,
Piaget's
cognitive
developmental
stage
law
of
individual
development
(sensorimotor
intelligence
concrete operational
intelligence
formal
operational intelligence) is "explained" by a
developmental
principle
of
increasing
equilibrium-equilibration,
and
evolutionary
stage laws are "explained" by reference to
Spencer's
principles
of
increasing
differentiation and integration, by Williston's
law
of
increased
specialization
etc. (Kitchener 1983, p.791).
Here the notion of stage and stage law is
of function,
very
central, as
well as a kind of progress notion as is the case with Piaget

and Spencer.
371
The stage laws are explained by the latter two in

terms
of
equilibium-equilibration
or
increasing
differentiation and integration. Both imply a kind of
progress that is not teleologically defined in terms of a
specific stage, at least not in the phylogenetic dimension
but rather as a kind of structure-dependent characteristic
(i.e. as an internal criterion).
With Kitchener
we also
like
to
introduce
a
differentiation between the so-called stage laws, historical
laws and evolutionary laws:
"According to G. Bergman for example, a
stage law (or developmental law as he calls it)
is different' in logical form from a historical
law. A historical law has the logical form:
(x) {Kx [[S1 (x, tw) . S2(x,ty]]
S3(x,tz)]}
i.e. for all XiS of a certain kind K, if x has
the property Sl at tw and the property S2 at ty,
then x will have the property S3 at tz (where tw
and ty may be at the same temporal interval or
not, but where tz is later in time than both tw
and ty). Evolutionary laws, as discussed by
Popper (1964), Hull (1974 ), and Olding (1978)
are "laws"
history
or
or
generalizations
development
of
concerning
the
particular
individual species or organism and hence are not
truly universal
or nomological." (Kitchener
1983, p.793)
Kitchener himself defines a stage law as:
"(x)
{Kx (S1 (x,tw)
(S2 (x,ty) . S3(x,tz},
i.e., for all x's of a certain kind K, if x is
in a certain stage Sl at time tw then x will be
in stage S2 at time ty (where ty is later than
tw) and x will be in stage S3 at time tz (where
tz is later than ty)." (Kitchener 1983, p.792)
So in
a nutshell,
in stage laws
succession between stages
are formulated. In the case of historical laws, we have a

succession of properties. In the case of evolutionary laws
in the sense of Popper it is a succession of stages or
properties
concerning particular species or
organisms that
is formulated. This means evolutionary laws don't have the

same type of universality the stage laws or historical laws
have.
EVANDAMME
372
Kitchener remarks about stage laws that one has to

take into account a set of practical problems which are very
hard to detail and therefore hamper the use
of the
developmental paradigm to a certain degree. These problems
are the following ones:
"1. Since the duration of each stage is not
specified in a stage law, nor does a stage law
indicate when the stage will appear, a stage law
is not only imprecise and vague but such lack of
clarity
seems
to
throw
its
scientific
credentials into doubt.

2. A stage law does not seem to depend upon any
seems
initial conditions
and thus
to be
unconditionally or categorically true. But these
features,
so it is alleged, make the law
non-scientific.
3. What type of law a stage law is supposed to
be remains unclear. It does not seem to be an
ordinary causal law, for example, since it seems
odd to say that a larvae

stage causes a
butterfly stage
or
that
feudalism causes
capitalism. But if it is not expressing a causal
relation,
what kind of
relation is being
expressed?
4. The logical form of a stage law is not clear.
Does it assert, for example, that one particular
stage is a necessary condition for a later
stage or rather a sufficient condition? Neither
choice seems to be adequate.( ... )
S. Finally, there seems to be an indefinite
number of "exceptions' to any purported stage
law.
There are an indefinite
number
of
environmental disturbances and obstacles that
would prevent development from proceeding (e.g.,
death, an enemy, a blow to the head) and there
are likewise an indefinite number of biological
and
social
conditions
in
whose
absence
development would not continue (brain defect,
genetic
defect,
malnutrition,
impoverished
social environment,
etc). In any of these
cases,
development
would
be
arrested or
abnormal, whereas in the absence of these,
development would proceed normally. But here we
have an asymmetry between normal and abnormal
development: a stage law does not (and cannot)
373
mention the absence of all these "abnormal

conditions" and when development is proceeding
naturally, none of these are in fact required to
be mentioned. If we want to explain a final
stage, for example, we need only mention the
stage law together with the earlier stages which
were present; we do not mention all these other
"necessary conditions".
But when development is
abnormal or retarded, this is explained by

appealing to these special factors. How then,
can we understand this basic
asymmetry of
explanation, a type that has long been abandoned
in the physical sciences?" (Kitchener 1983,
p.794).
We will return to these problems later.
4.LANGUAGE AND LANGUAGE CHANGE

Much attention has been paid to language change. For
instance,
many theories have been forwarded on the origin
of language, but also in modern times this topic has
regained
respectability.
The
chain
of
symposia on
glossogenetics of the last few years illustrates this
(Paris, Montreal, Krakow). But beside the problem of the
genesis of language, the diachronic study of language has
been developed and has been dominant for 150 years in the
science of language. With Martinet however, we have to
remark that not
much of the dynamics of language has been
studied. Rather the comparison between several languages,
and between different stages languages would be in, has been
focused on:
"In
traditional
(prestructural) linguistics
people were mostly occupied
with comparing
genetic related languages and tried to render
account of features of the one, while refering
to certain features of an other, instead of
trying to determine how and why a given language
evolves through the centuries. In other words,
researchers had a tendency of signalizing the
correspondences rather than to explain them."
(Martinet 1962, p.162)
They have looked for the universal, i.e. the common
core between different languages. In a certain sense it can
be argued that one looked for the invariable in the
variable.
374
F.VANDAMME
sense.
it,
The laws one introduced have to be understood in that

The problem why a change happens, or what motivates
was neglected and even rejected as unscientific
in the
classical diachronic study. We think, however, that the

answer to these questions can provide important guidelines
and pointers for a better description of the changes.
It is true dependable descriptions and observations
are needed here. But we all know today how observation is
dependent
on
strategies
for
observation.
Strategies
in
themselves imply theories, hypotheses, perspectives. It is

without doubt that explanations, however tentative, can play
an important role in the determination of perspectives and
hypotheses. We observe a similar fact in the development of
biology. Darwin's theory, an explanation of evolution,
whatever its value, has given an important stimulus towards
more extensive and systematic biological observation.

Martinet (1962) introduces two main factors for the
explanation of language change. They are (i) the necessity

for communication (that is the function of language) and
(ii) the language use: (a language changes by its use). In
his treatment of this second principle,
Martinet focuses on
ziff's law as an illustration. Ziff's law states that in

general the more frequently used signs are shorter. But
Martinet tries to explain this law by making reference to
the Shannon and Wiener notion of information.
Nowadays,
information is
Nevertheless,
we know that the Shannon-Wiener

much too narrow
for natural
certain
interesting
notion of
language.
elaborations
and
applications in natural language of the Shannon information

approach are given by M~rti~et. It is clear that - in the
Martinet approach
1t 1S not only the phonological
structure (form) which is modified by the use of language.
The syntactical,
the semantical,
pragmatical
and even the
discourse dimensions of language are changed. Thus, in

language change it is undoubtable that use is an essential
factor.
Let us confront this
constatation
with the
Lamarck/Darwin controversies.
Lamarck's thesis is that

change of the organs is
caused by their use. The adaptation is then transmitted by
inheritance. Darwin opts for survival of the fittest, that
is the best
adapted for survival and
for producing
offspring. In both cases we have an adaptation of the
species to the environment.
375
Language is a tool for a human being. Certainly, as

with all human tools it is changed by its
use and
transmitted to
offspring in a continuum of use and
therefore of change.
So at
first glance it is the
Lamarckian view which seems best suited for explaining
language change and language evolution, if evolution there
be.
Most, if not all, knowledge can also be considered as
a system of symbols
with specific symbol-manipulation
devices, as
is the case with language. Inversely, we can
consider language as a specialised knowledge system, with a
specific function.
So the generalisation rather easily can be made that
knowledge also is changed
in its form
by its use
(Martinet's second factor) and by its function (Martinet's
first factor).
But if this step is taken then we conjecture that the
Lamarck view on biological evolution gets
new impetus.
For
Jakobson has already

stressed how
the whole genetic
reproduction can be viewed as pure communication.
Consequently,
the Martinet laws on explaining change
in terms of (i) function and (ii)
use are of great
importance even biologically, the more the generality of
existence of symbolic signs is recognised.
For sure the Lamarckian thesis still requires that
there exists a dependency between use and function of
macro-communication (use of organs) and micro-communication
(genetic
communication).
If
this
microand
macro-communication and symbolisation cannot be linked with
each other, then the Darwin thesis can hold out, even if the
Martinet standpoints are true on each level.
One may argue against
Martinet that it is true that
the change can be explained by the use of the signs, but
that we
observe that not all changes survive, only some of
them do. Therefore, Darwinian survival of the fittest (here

a sy~bol) might be said to hold. Or not?
The first of Martinet's factors,
that is the
functionality
,
can be responsible for the selection of
the changes. Functionality, - and therefore selection- , in
the long run is not an individual but a social phenomenon.
I t is the group of language users who accepts or rejects a
certain language change; it is in view of their conscious or
unconscious need that the functionality or the usefullness
of a change is decided. Both a need and its satisfaction
are dependent (i) on the organism, with its individual as
F.VANDAMME
376
well as its social features, and
(ii) on the environment as
it is symbolised, seen, experienced by the organism. Indeed,

an environment only exists for an organism to the degree and
in
the
manner
that
it
is
taken
into
account
in
its
symbolisations and its activities (here seen as a particular
form of symbolisation).
When looking at language change in this way, one
could be tempted to argue that we have a kind of synthesis
between Lamarck and Darwin in the Martinet approach of the
change in symbolic system. The Lamarck aspect is present in
the principle that the change is determined by its use. The
Darwinian aspect is present in the selection by
means of
functionality. But we must be aware that it is social
selection, and a social functionality with an ultimately
social survival!
Language and knowledge are primarily
social, with individual aspects. For language as well as
knowledge is meaningless and purposeless, without this
social dimension.
We are led to a series of interesting problems.: Can
this synthesis been generalised to encompass all living
systems and their changes? Do we have to speak about
language change, or may we speak about language development
and language evolution? These last terms imply some more
specific structures of change. The same certainly is valid
for knowledge. Do we simply have knowledge change or do we
have true knowledge development and knowledge evolution?
5. LANGUAGE AND DEVELOPMENT
Let us just touch on the problem of the development of

language and knowledge. We shall stress the qualitative
difference between ontogenetic and phylogenetic language and
knowledge development. Moreover, we want to say something
about Kitchener's epistemological and language-philosophical
problems related to the stage conception which is so
important in general development theories.
5.1. Progress and the phylogenetic
change
versus
the ontogenetic
Many will argue that in both

phylogenetic and ontogenetic
change, progress can and has to be defined. It is clear
however that although progress is to be expected in the
normal case,
that it
is
it
not
is
nevertheless
"default"
not
feature
automatic (meaning
of
system change).
377
Stagnation and even regress is commonly observed. We are

convinced that progress on the phylogenetic level is not
synonymous and not of the same kind as progress on the
ontogenetic level.
As far as symbolic systems (that is: language systems
or knowledge systems) are concerned, their progress on the
phylogenetic level cannot be defined
in terms
of a
teleological target or cause.
At the phylogenetic level symbolic ~ystems are social
in nature. The stagnation or regression ln their progress,
we believe, has to be defined in terms of increasing social
differentiation
and
integration,
adaptation and accomodation

symbolic system.
of
in
view
the group of
of
extending
users of the
Progress here implies change with a direction. However
this direction is determined by structural features of the

system itself and by the efficiency of its repercussions on
the users of the system.
In fact the users of
and actions are an integral
large.
In
technique
the system with their reactions

part of the symbolic system at
phylogenetic evolution, one can, depending on the
one uses,
differentiate
between
several stages
(see below). Eventually, one can trace one or more paths

through the stages. Historical and evolutionary data (see
the differentiation made between these earlier) about groups
of users of one or more
can be of much use here.
But we
change is
sets of analogous
must be aware that
concerned,
as far
symbolic systems
as the ontogenetic
the present situation or
phylogenetic evolution of the symbolic system is

stage
of a
particular individual
in
an
development
(inside
the
overall
social
stage
of a
the target
ontogentic
system
in
phylogenetic process).
Defining progress at an ontogenetic level is therefore
mainly done by reference to an external target, which is
the level of development of the phylogenetic symbolic
system. It is true that in the endeavor to reach this level
a certain ontogenetic symbolic system may deviate from or
transgress
this external target. Such transgressions are
important tools in and for the further development of the
phylogenetic dimension of the symbolic system. But the
phylogenetic evolution
always remains a prerequisite for
EVANDAMME
378
the ontogenetic evolution, among others for the reason that

a recognition of a higher stage requires a generalisation of
this "progress" over the whole group of actors.
5.2. Linguistics and the stage notion
Many important
problems
concerning
the developmental
explanations are related to the vagueness of the notion of
stage (cfr. Kitchener 1983, p.794). How to characterise a
stage, its duration, its beginning and its end?
In modern linguistic theory the notion of stage
becomes more and more a central concept.
In discourse theory, many authors refer to the stage
approach. Others introduce other terms, with a similar
function: theme, frame, register, situation etc. What is
typical for a stage is that all symbols, all data get a
particular perspective relative to it.
Despite the
linearity
of
the
verbal language
structure, pluridimensionality and hierarchy are introduced
by the stage into the language system. Some elements are
more
important,
others
are
more
peripheral
or,
give
subsidiary information. By means of a stage therefore a

hierarchical structure is introduced. Indeed, the stage
methaphor is very useful here (cfr. Grimes 1975, p.323,
Brown & Yulei 1983, p.134, Clement 1979, p.287). The same
metaphor is central in development theory. But, in this view
Ita stage"
is a construction for the sake of
organising and
introducing a hierarchy between symbols. It is important for

the interpretation of the latter.
It is therefore a
construction which plays a decisive role in determining
action and reaction strategies. This implies that it is
naive to treat a stage as an ontological1y real phenomenon.
Its limits,
in time and in relation
to neigbouring
structures,
its organisation and
hierarchy 15 always
pragmatical:
defined in
terms
of
some explorative,
predictive or therapeutic perspective, for instance. The
actual status of the phylogenetic structure of the symbolic
systems plays an heuristically important
role in the
construction
of the stages,
we believe.
This is also true
for the description of the ontogenetic development of a

certain symbolic system for some actors (which is in fact a
synchronic study). The inverse is also true.
The stages
constructed for
ontogenetical development will
be an
heuristic base for the construction of the stages of the
phylogenetic development.
379
Kitchener in his methodological discussion of the

notion of stage also talked about the so-called asymmetry of
abnormal conditions or exceptions. In the developmental
approach the sequence of stages is determined without
conditionality. When development is abnormal or retarded,
then this is explained by special factors (e.g. brain
defect, genetic defect). In the normal case, no reference is
made to this conditionality.
This
too
stresses the
conventional, constructive nature of a stage.
But not only that.

The stage is at the same time a
social norm.
Therefore it is essential that the possible
deviations
are not taken up
default case is
be focused on,
in
its
beyond deviation.
characterisation. The
If not,
deviation would
with all its social repercussions.
All this does

not preclude that deviation is very
important in the ontogenetic as well as the phylogenetic
development
of
a
symbolic
system.
Ontogenetically,
deviations of symbolic systems are sometimes sanctioned, (in
some cases very strongly,
in other cases rather weakly) and
sometimes not. From a phylogenetic point of view, deviations
from the symbolic system are rather interesting experiments

in a search to
ameliorate the system: under certain
conditions, deviations are taken up and socially integrated.
This integration has impact

stage.
on
the
norm-stage
or default
6. CONCLUSION
In Vlew of this analysis of the development of symbolic
systems, in their ontogenetic as well as phylogenetic
dimension, and considering language and knowledge system as
particular cases of symbolic systems,
we propose the
following hypothesis: knowledge systems phylogenetically
evolve
(or progress in the Spencer interpretation) along
internal phylogenetic criteria towards a higher level of
adaptational
and
accommodational
efficiency.
Ontogenetically, the stage (defined vis-a-vis an external

target), plays a central normative and descriptive role.
In epistemology one has to be aware of the dual and
dialectic object of study: of the individual knowledge
system and the social knowledge system. Therefore, in
developmental explanations, one must avoid being trapped
into a one-sided external teleological interpretation of

development.
Knowledge systems are dynamic,
they are
380
F.VANDAMME
continuously changing. We think the distinction between two

types of change (ontogenesis and phylogenesis)
is an
essential one.
In
this
interpretation
dynamic
epistemology,
including EE as a special case,
is the epistemology of the
future. Indeed, dynamic epistemology is the only one which
can be a base for a true applied epistemology. The latter is
so much needed in Artificial Intelligence, in Intelligent
Information Technology,
but also in politics: General
Politics as well as Science Policy
and in cognitive
education strategies as well as cognitive therapies.
NOTE
(1)
With this reference to empirical data we do not
realistic
ontological
point
of
view.
In
take a
strict
nominalistic approach we differentiate between a theoretical

and an empirical or observational level. Both imply a
symbolic dimension,
but the latter
incorporates
than the former the symbolics of action.
much more
ACKNOWLEDGMENT
The informal discussions and interactions of D. Campbell and
G. Vollmer were very stimulating for the elaboration of this
paper.
THE EVOLUTIONARY EXPLANATION OF BELIEFS
Philippe Van Parijs

Universite Catholique de Louvain
1. INTRODUCTION
In my Evolutionary Explanation in the Social Sciences, I

attempt to answer two questions which I consider to be of
central importance to the philosophy of the social sciences.
Firstly, are functional explanations - the explanation of
social practices,
norms, institutions, etc. by reference to
their (latent) function - legitimate in the social sciences,

just as they are in biology? Secondly, what is the "deep
structure" of the social sciences? In particular, do all
legitimate social-scientific explanations fit the action
pattern, do they all consist in explaining actions and their
aggregate results by reference to the subjective situation
of their authors? The treatment of both these questions
leads to a
systematic exploration
of
what
I call
evolutionary
explanations,
i.e.
explanations
which
presuppose a mechanism of "filtering through the (actual)
consequences"
or, put differently, which rely on the
selection of "blind" variants according to the consequences
associated with them (Van Parijs, 1981, section 20).
In that book, however, I am exclusively concerned with
the explanation of practices - behavior patterns, rules,
institutions, customs, etc. - and have nothing to say about
the explanation of beliefs (ideas, theories, ideologies,
representations, cogn1t1ve systems, etc.). One aim of the
present paper is to fill this gap by focusing precisely on
the evolutionary explanation of belief systems. My main
concern, however, will not be with how this change of focus
affects the answer given in the book to the two questions
mentioned above, but rather with how it leaves the central
distinction around which the book is built. This distinction
consists in a simple dichotomy (to be illustrated
below)
between two types of evolutionary explanations:
(1) NS-evolutionary explanations rely on a mechanism
of natural selection, defined as the selection of the
features
entities
to
be
which
explained
these
through
features
the
selection
characterize;
381
W Callebaut and R. Pinxten reds.), Evolutionary Epistemology, 381-401.
/987 by D. Reidel Publishing Company.
of the
P. VANPARIJS
382
(2) R-evolutionary explanations rely on a mechanism

of reinforcement, defined as the direct selection of the
features to be explained within the entity they characterize
(Van Parijs, 1981, p. 95).
This concern is less parochial and therefore, I
believe,
more
appropriate
to
our
interdisciplinary
gathering. For it amounts to raising the general question of
the nature and classification of evolutionary
mechanisms. I
do not claim, in this paper, to provide a full-fledged

typology of evolutionary mechanisms, but only to show why
the above-mentioned
dichotomy,
which itself tried to
overcome
the
ambiguities
and
difficulties
of more
conventional distinctions, will not do - and to make some
suggestions as to what a more adequate classification should
look like.
For this
purpose,
I
will
first
examine the
NS-evolution~ry
e~planation of beliefs and
point out the
difficulty 1t ra1ses for my original distinction (section
1). Next, I will turn to typical functional explanations of
belief systems - which are clearly not NS-evolutionary - and
spell out what they presuppose (section 2). The functional
explanation of religious beliefs will serve to illustrate
the abstract formulation of these presuppositions (section
3). On this basis, I will argue that the underlying
mechanism could not be one of reinforcement any more than of
natural selection, but rather consists in what could be
called accommodation and can be fleshed out with the help of
various aspects of the social-scientific literature (section
4). This mechanism also enables us to make sense of the
theory of ideology, which typically explains beliefs by
reference to functions they perform for others than those
who entertain them (section 5). In the conclusion, I put
forward a modified classification and mention some remaining
problems.
2. THE NATURAL SELECTION OF BELIEFS
Natural selection consists, by definition, in the selection
of features - for example genetically controlled phenotypic
traits, practices, beliefs - through the selection of the
entities
organisms, groups, societies
which they
characterize. It may be convenient to distinguish four forms
of natural selection in this sense, according to whether
(a)
383
the features which are being selected
transmitted)
biological
characteristics
are (genetically
or
(socially
transmitted) cultural traits or grQups.

(b) the entities on which selection operates are individual
organisms or groups.
The four
cases obtained by
crossing these two
distinctions
correspond roughly
to
the neo-Darwinian
synthesis
(biological/individual),
to so-called "group
selectio~'
(biological/collective), to so-called "cultural
ethology"
(cultural/individual)
and
to
"system
functionalism"
(cultural/collective) (Van Parij s, 1981,
section 22). Is it in principle possible for natural
selection, thus described, to operate on what we usually
call "beliefs" (or "ideologies", lIideas", etc.)?
One reason why this might not be possible is that
beliefs, unlike practices, do not directly produce external
effects on which natural selection could have a grip. A
belief
can only
generate such effects in
conjunction with
some goals, values, rules, etc., i.e. with something which

does not qualify as a belief in the (rather loose) sense in
which I will be using the
the belief
that
having
term in this
paper. For example,
lots of children
guarantees one's
salvation will not generate a tendency to have more children
unless those who hold that belief also value salvation. What
matters to natural selection, however, is differential
consequences, whether joint or not, whether direct or not.
It is enough, for natural selection to operate, that
differences in one's beliefs about what matters to salvation

should affect,
other
things
remaInIng
equal, one's
reproductive success.
It is not required by
any means that
beliefs should generate behavior on their own (1).

Secondly,
one might
want
to
argue
that
evolutionary shaping of
beliefs,
the
unlike that of practices,
necessarily pertains to cultural (i.e. socially controlled)

rather than organic (i.e. genetically controlled) evolution.
For there would be no point in distinguishing beliefs from
practices if it
were not possible to
believing something and acting
differentiate between
on that belief.
Since this
is only possible if cognitive contents can be expressed (and
hence transmitted) independently of the behavioral responses

they contribute to generate, it can be sustained as an
analytic truth that beliefs (not practices) are culturally
transmitted. If this argument is correct (2), two of the
four forms of natural selection distinguished
above
those operating on biological characteristics
become
384
P. VAN PARUS
irrelevant in the case of beliefs. This does

natural selection is altogether irrelevant.
at first sight the promise held out by the
forms is not very great.
Take first, the selection of cultural
the selection of the social groups carrying
functionalism"). In Evolutionary Explanation
not imply that

But at least
remaining two
traits through
them ("system
in the Social
Sciences, I argue that this mechanism does play some role in
the field studied by the social sciences, especially when

the groups considered are firms or sects rather than total
societies. But I conclude that explanations relying on it
play such a marginal role in actual social-scientific
practice that mentioning it could hardly be more than
anecdotal.
The
recent
expansion
of
interest
for
natural-selection
models in industrial
economics, however,
demands that this conclusion be duly qualified, even if most

forward mix elements
of natural
models actually put
selection and of reinforcement (Nelson and Winter, 1982).
Beliefs entertained within a firm or a sect are no less and
no more liable to this form of natural selection than rules
or habits, and what has just been said applies equally to
them (Van Parijs, 1981, section 31).
The natural selection of cultural traits through the
selection of individuals ("cultural ethology"), on the other
hand,
is
dismissed
in the
book
as "intrinsically
problematic" even for those traits which (unlike a pattern
of social stratification or the existence of
courts) can be
carried by individuals. The fundamental problem is not only

that such selection is unavoidably restricted to cultural
traits affecting reproductive success (independently from
the social sanctions enforcing conformism), but mainly that
cultural traits are "usually transmitted, maintained and
enforced at the level of society - school, friends, sorcerer
rather than at the level of a single family" (Van Parijs,
1981,p. 86-87). In their very stimulating recent work,
however, Boyd and Richerson have shown how these two
difficulties could to some extent be
circumvented by
construing cultural ethology in a broader manner which
allows for "multiparental transmission".
One way of putting their central point is as follows.

What matters to the fate of any feature, whether genetic or
cultural, is not, as such, the success of its carrier as a
reproducer, but the success of its carrier as a transmitter.
In the case of a
with reproductive
genetic feature,
success.
But in
this happens to coincide

the case
of a cultural
THE EVOLUTIONARY EXPLANATION OF BEUEFS
385
trait, this coincidence can no longer be taken for granted.

Indeed, to take one extreme example, a belief in the virtue
of celibacy may conceivably maximize success in transmitting
one's cultural traits (namely, if being childless makes one
more available or more efficient for teaching purposes).
Imitation, rather than reproduction, is now the key to the
transmission, and hence the spreading, of a feature. A
generalized cultural ethology can study the dynamics of a
population in which this sort of process is at work, whether
on its own or in conjunction with standard natural selection
(with
biparental
transmission),
biased
choice
of
transmitter, reinforcement, etc., as well as attempt to
determine the optimal mix of these processes in terms of
inclusive fitness as the overarching criterion (Richerson &
Boyd, 1984; Boyd & Richerson, 1985).
What matters, for our present purposes, is that such a
process in
its pure
form
is an evolutionary
mechanism as
defined above - features are being filtered according to

their consequences (for transmissive success)-, but that it
is not a mechanism of natural selection, again as defined
above, since the selection of features does not here operate
through the selective elimination of organisms (3). Note,
moreover,
either,
that
it
is
not
mechanism
of reinforcement
since the selection of features according to their

conse~uences is not supposed to take place within particular
organlsms. The evolutionary shaping of belief systems - and
of other cultural traits - can thus take a form which creeps
in the interstices of the basic distinction between natural
selection and reinforcement and thereby challenges the claim
that these two categories cover
the
whole range of
evolutionary explanations. I will return to this challenge
in the concluding section and now turn to the question
whether beliefs can be "reinforced".
3. FUNCTIONAL EXPLANATION AND
EVOLUTIONARY ATTRACTORS
In order to assess whether reinforcement applies in the
case of beliefs just as it does in the case of practices,
let us now turn to explanations of (socially shared)
beliefs actually put forward by social scientists, first in
order to identify and locate those among them which seem to
call for some sort of evolutionary mechanism.
386
P. VANPARIJS
On a first level, the maintenance and adoption of

(socially shared) beliefs, just as the maintenance and
adoption of (socially shared) practices, are often explained
by
social
scientists
in
terms of inertia
and contact: a
social group believes what it does "because it has always

known it", and it alters its beliefs because of some
"influence".
This
kind
of
explanation
tends
to
become
tautological unless it is supplemented, in the case of

practices, by an analysis of the interplay of social
sanctions, and, in the case of beliefs, by the interplay of
what could be called social authorities. What keeps belief
systems unchanged is above all the insertion of each
individual into a "plausibility structure", i. e. in a
community which lives in a common universe and which defines
reality in the same way (4). A change in beliefs, on the
other hand, is basically governed by the interplay of
conflicting
authorities:
the authority
of tradition,
sometimes vested in particular individuals (chiefs and
sorcerers, priests and schoolteachers) must fight it out
with a conflicting authority stemming either
from an
alternative plausibility structure or from a charismatic
personality. Just as a change in practices is the uncertain
outcome of a conflict
between contrary
sanctions, a change
in beliefs is the uncertain outcome of a conflict between

contrary authorities. (See the social-psychological study of
the factors which affect the effectiveness of "persuasive
communication",
as
reviewed,
Berkowitz and Moyer, 1970).
In this case of practices,
for
example,
by
Simons,
however, it may be further
asked why the "superstructure" of social sanctions protects

the practices it does (rather than others, equally possible
within the given context). Sometimes, the justifications
provided by the agents themselves seem satisfactory enough
(for example, the socially enforced prohibition against
urinating in a fountain is justified by the health hazards
involved). But in other cases, they appear less convincing
(for example, the prohibition of marriage with a parallel
cousin is justified by reference to the fact that children
born from such a marriage would have awful teeth and red
eyes). And this is where social scientists tend to introduce
explanations by latent functions,
which call
for an
evolutionary reconstruction.
In the case of beliefs, similarly, we may want to know
why the "superstructure" of social authorities protects the
beliefs it does. The justification provided by the agents
themselves is nearly always that what they believe
387
is true,
that it corresponds to reality. In some cases again, such

justification
may
be
convincing
enough
for
the
social-scientific inquiry to fizzle out at this point.
Suppose, for example, that it is widely believed that
Belgians are stupid. If the investigator knows that Belgians
are indeed stupid or
to put it in a more agnostic (and
tactful) fashion - if he shares this widely held belief, why
should he bother to find a further explanation for the fact
that this belief is so widely held? The interesting cases,
in the social scientists' eyes, are precisely those in which
the believer's justification does not convince them, whether

because the belief is (to them) demonstrably false or
because the evidence is (to them) blatantly insufficient to
ground a reasonable belief. These are the cases.where social
scientists may come forward with functional explanations and
where,
therefore,
evolutionary mechanisms may need to be
appealed to.
It is a remarkable feature of most social-scientific
theories of belief systems that they tend to define their
specific objects in two complementary ways - or to oscillate
between two such definitions. The beliefs they are concerned
with
are,
on
the one hand,
characterized as "distorted",
and, on the other, as "functional" in some sense (5).The

paradigmatic illustration is the theory of ideology. On the
one hand, an "ideology" is often defined as a belief which
"distorts"
reality (e.g. Althusser, 1965, p. 64, 78;
Goldmann, 1966, p. 110), or in terms of a "discrepancy
between what is believed and what can be (established as)
scientifically correct" (Parsons, 1959, p. 25), or else in
terms of "false consciousness" (e.g. Miller, 1972, p.
442-444). But social scientists often shy away from such an
epistemic definition, which seems to attribute them some
superior position of omniscience (See e.g. Mannheim, 1929,
p. 80,83; Schaff, 1967, p. 50-51; Althusser, 1969, p. 23 and
Lang 1980, p. 124). And they turn instead either to a
definition of ideology as a belief which can be explained by
the social position of those who entertain
it (e.g.
Mannheim, 1929, p.69, 77-80), or as a belief which can be
explained by the social functions it serves (see Elster,
1982, p. 130-137). I t is the latter definition which is
relevant for our present purposes. It can be illustrated,
for example, by Louis Althusser's later writings, where
ideology
is
defined
by
the
dominance
of
the
"practical-social
function"
over the "theoretical" one, or
388
P. VANPARIJS
by the fact that the solutions it offers to the problems it

considers are "produced by practical interests" (Althusser,
1965, p. 138 and Althusser and Balibar, 1968, p. 62-63), or
again "by the concept of ideology elaborated on the ground
of the modern Marxist-Leninist philosophy, which defines
ideology in purely functional terms, regardless of the
cognitive
value
of
the
propositions
qualified
as
ideological" (Lang, 1980, p.124) (6).
A very similar story can be told about the concept of
illusion,
as handled by the psychoanalytic theory of
religion. On the one hand, there are clear connotations of
falsehood,
and Freud
has to be
very careful about
distinguishing illusion and error. On the other hand, he
defines it explicitly by the dominance, in the motivation of
a belief, of the accomplishment of a desire (Freud, 1927, p.
353-354). Similarly, students of prejudice define the latter
either as a belief unwarranted by the evidence, as a
spurious
Qvergeneralization,
as
having
no
more
than
superficial relation with reality (e.g. Adorno and al.,

1950, p. 607,612,618 or Allport, 1954, p. 7-8,12-13) or by
the fulfillment of a specific irrational function for the
bearer of the prejudice (e.g. Ackerman and Jahoda, 1950, p.
4). And analogous hesitations could be detected as regards
the notion of rumor (Morin, 1969, p. 17; Rouquette, 1975, p.
17-18), the Durkheimian concept of prenotion (Bourdieu,
Passeron and Chamboredon, 1968, p. 35-36), Pareto's notion
of derivation (Soudon 1984), Smelser's concept of hysterical
belief (Smelser, 1962, p. 94,100), etc.
What matters, of course, is not the detail of these
semantic issues, but the general picture which they reveal.
The general presumption is that there are purely cognitive
processes
which,
in
the absence of
interference by other
factors, would lead beliefs to take a particular shape (if

relevant evidence is available) or to be spread randomly
over the space of logical possibilities (if it is not). Let
us
call
cognitive
attractors
the
(locally
stable
equilibrium) states - if any - in which belief systems tend
to settle as a result of the operation of such processes
(7).
Interesting questions arise,
in a social scientist's
eyes,
when
belief
these states -
point
of
i.e.
systems
(systematically)
deviate from
when reality is being "distorted". The
epistemic definitions
of "ideology", "illusion",
"prejudice", etc. is that they focus on

such situations
as the interesting ones for social scientists to explain.
However,
such definitions are damagingly narrow
a) because
389
they rule out cases in which beliefs which happen

to be
true, or strongly supported by evidence, are actually held
for reasons wholly independent from their truth or their
empirical support; and b) because they
also
rule out
situations in which there are no cognitive attractors
whatsoever,
and
hence
in
which
the
notions
of
"distortion ll , "falsehood", etc. do not make sense.
The
"ideology",
advantage
of
the
functional
definitions
of
"illusion", "prejudice ll , etc. is precisely that
it covers all those cases and that they point straight away
to the kind of explanation which is being proposed. Of
course, the same kind of explanation could also be expressed
without making use of a functional definition, for example
by stating that some ideologies, illusions, prejudices, etc.
(more neutrally defined) perform certain functions which
account for their existence. But functional definitions make
it possible to phrase the matter more succinctly: simply by

calling a belief an ideology, an illusion, a prejudice,
etc., one claims to explain it in a certain way. In either
case,
the general presumption is that processes are at work
which pull belief systems away from where they would be

(i.e. from whatever shape they would take) if only processes
of "cognitive attraction"
were at work.
To the extent that
the functions referred to in functional definitions are

consequences associated with the beliefs and that these
beliefs are not deliberately chosen in order to bring them
about, these processes must be evolutionary mechanisms as
defined above, i.e. mechanisms of "filtering through the
consequences", and the (locally stable equilibrium) states
in which belief systems tend to settle as a result of their
operation can then be called evolutionary attractors (8). In
the
absence
of
"evolutionary
attraction",
beliefs
will
settle where cognitive attraction pulls them, whereas one

should expect
practices to become
extinct (9). This
constitutes
a significant difference
between
the
case of
attraction, and hence calls for the specification

evolutionary mechanism.
of some
beliefs and that of practices. But in both cases, the very

use of functional explanations presupposes evolutionary
390
P. VANPARIJS
4. THE EVOLUTIONARY EXPLANATION OF

RELIGIOUS BELIEFS
Before facing squarely the question of how this mechanism
should be specified, let us consider in some more detail one
particular instance of the sort of explanation we are here
trying to reconstruct. In a Freudian perspective, religion
is an "illusion",
i.e. the accomplishment of a desire.
Its
essential function, the "function to which it owes most of

its influence", consists in soothing men's anguish
and in
consoling them in adversity. Religion is rooted
in man's
distress,
in his powerlessness,
in his craving
for
protection. In particular, the persistence of the child's
distress throughout life explains why man has so obstinately
stuck to the belief in an almighty Father (see e.g. Freud,
1927, p. 337-354, 1930, p. 421-431, 1933, p. 174-181;
Malinowski, 1944, p. 173-174). In a Marxian perspective, on
the other hand, religion is the expression of people's
material misery, of the alienating situations in which they
live. It consoles them with the promise of (illusory)
happiness after death. In this sense, it is produced by
men's social relations, by their material conditions. In
this sense again (and not
in the sense of
a drug
administered by a malevolent doctor), it can be called "the
people's opium". Religion has managed to survive because it
soothes the pains, because it meets the aspirations of
oppressed people (see, typically, Marx, 1844, p. 378-379,
1859, p. 8-9; Marx and Engels 1846, p. 26-27,38; Donini,
1975).
Between
these
two
perspectives,
there
are obvious
differences. One stresses psychological factors and human

nature. The other stresses social factors and historical
conditions. But in either case, the persistence of religion
as an socially shared (and enforced) belief system is
explained by reference to the function it
serves in
fulfilling the believers' wishes. In either case, it is
taken for granted that religious beliefs - from the belief
in an almighty Father to
the
belief
in posthumous
compensation for present suffering - could not be explained
by cognitive factors alone. As mentioned earlier, this does
not require us to assume that religious beliefs are false,
but only that the "evidence" for them is so scarce that it
would not suffice to account for the remarkable universality
and persistence of such beliefs. In other words: since such
391
beliefs do not correspond to cognitive attractors, the fact

that they are so widely held can only be explained by the
fact that they correspond to evolutionary attractors, i.e.
by the fact that they do better, in terms of peoples'
wishes,
needs, interests, than such "local alternatives" as
disbelief in God or in life after death (10).

The object of evolutionary explanations, of course,
need
not
be
restricted
to
the
universality
or
near-universality of beliefs. Evolutionary explanations can
also explain differences in the beliefs adopted by different
groups. Take, for example, Pierre Bourdieu's analysis of the
religious
beliefs
of
Algerian
proletarians
and
lumpenproletarians. Proletarians have stable jobs and the
guarantee that their basic needs will be satisfied. They are
able to make rational plans, to forecast and to save.
Lumpenproletarians, on the other hand, permanently live in a
state of need and insecurity. They are unable to make plans

and oscillate between fatalism and utopia. Arguably, what is
optimal in terms of the former's needs is the belief in a
Paradise with individual bliss, while millenarianism, a
blind longing for a complete subversion of the world, is
optimal in terms of the latter's needs. This might account
for the differences actually observed (see Bourdieu, 1966,
p. 126-127,136-141).
Just as the position of the evolutionary attractor may
vary from one group to another, it may also shift through
time. To illustrate, take the case of the Melanesian "Cargo
Cults", as characterized by beliefs in the imminent landing
of European
goods or
in
an
impending
exchange
of roles
between Blacks and Whites and change in the color of the

natives' skins. The rapid spread and independent development
of such beliefs in many Melanesian islands throughout the
first half of the twentieth century is commonly attributed
to the frustrations generated by the colonization process.
Life is made intolerable for the native population, which
sees no hope of gradually improving its situation through
institutional means. In this context, millenarian beliefs
become better suited to their needs than their traditional
faith (see esp. Worsley, 1957, p. 229-253; Jarvie, 1964, p.
99-101). In other words, the change in the context induces a
shift in the position of the evolutionary attractor, which
generates in turn, with a time lag, a change in the socially
accepted beliefs. In the case of beliefs just as in the case
of
practices,
evolutionary
explanations
(comparative-static) explanations of change.
can
be
392
P. VANPARIJS
5. THE MECHANISM OF ACCOMMODATION

Thus, the logic of explanation seems to be pretty much the
same as in the case of practices. But what about the
mechanism involved?
Even if the natural selection of
cultural traits plays the role granted to it in section 1
above, it could not provide with microfoundations the sort
of explanation we have been considering in the previous
section.
For
the
implicit maximand,
in
all these
explanations,
the
implicit
criterion
of
optimality
is
something like wish-fulfillment or mental comfort, rather

than effectiveness as a transmitter (whether reproducer or
teacher). Of course, in order to explain why such a
criterion happens to prevail, one may have to appeal,
ultimately,
to natural selection.
But this does not
eliminate the need to spell out the intermediate mechanism
if
one
is
to
make
sense
of
the
various
functional
explanations of belief systems

to be
found
in the
social-scientific literature.
In the case of social practices, this problem was
solved by spelling out the mechanism of reinforcement - the
selection of variants according to their consequences within
particular
organisms
conditioning.
as
generalization
of
operant
This enabled us, in particular, to make sense
of "chances of satisfaction",
instead of "chances of
reproduction",
as the relevant maximand. Why not use
reinforcement to reconstruct functional explanations of

beliefs just as we used it to reconstruct functional
explanations of practices? The key difficulty can be put as
follows. In a way, reinforcement is essentially concerned
with beliefs. It consists in learning, through trial and
error, what the consequences of various possible practices
are
and
in
adjusting
behavior
accordingly.
Thus,
reinforcement could be said to consist in getting one's
beliefs straight.
As has been noted
above, however,
functional explanations of beliefs typically assume that
beliefs have gone astray. Or, to use the terminology
introduced earlier,
whereas reinforcement assumes that
beliefs converge on cognitive attractors, it seems essential
to the functional explanations considered that they should

converge on evolutionary attractors which do not generally
coincide
with cognitive ones.
entertained because
Beliefs are here
they are in themselves
said to be
satisfying, not
because they are accurate and therefore make for satisfying

practices. The mechanism could still be one of selection of
393
variants acc?rding to consequences

(however immediate)
within organ1sms, but it would need to leave out the
consequences beliefs generate by virtue of the practices
rooted in them (11).
In order to get a more precise picture of the
mechanism involved, one may want, first of all, to take up
Freud's
remark
that
the
logic
of
religion
as
wish-fulfillment is very similar to the logic of dreams
(Freud, 1927, p. 338-339). There is indeed a very extensive
psychoanalytic literature on the processes involved in the
shaping of dreams (and other phantasms)
such as the
well-known
operations
of
"displacement",
"condensation",
"transformation into images" and "secondary elaboration"

(Freud, 1899, 1901). However, these processes - which make
up what Freud calls "dream work"
are not supposed to
operate
between real-life material and
what "wishful
dreaming"
makes
of
it,
but
rather
between
the
wish-fulfillment which constitutes the latent content of a
dream and the manifest dream which we remember. And the
distortion which is thus effected is not the product of
wish-fulfillment but of social censorship, as instilled in
every child by early socialization.
Consequently, it may be wiser to turn to more directly
relevant social-psychological literature, such as the theory
of cognitive dissonance and related approaches (going back
to Festinger, 1957; Osgood & Tannenbaum, 1955; Rosenberg,
1960, etc.). These are attempts to study belief formation by
reference to the general postulate that beliefs held by an
individual tend to change so as to minimize "dissonance",
"incongruity",
"imbalance",
etc.
In particular, they show
how negative or positive attitudes towards an object act as

powerful filters in the selection of what an individual will
settle with believing.
Alongside the processes which underpin individual
wishful
thinking,
however,
some
specifically social
processes may also play a significant role in shaping
socially shared beliefs according to their consequences.
These processes are most conveniently studied by looking at
the way in which rumors spread. In addition to the processes
sketched above taking place within each individual the
development of a rumor crucially depends on how likely
individuals are to transmit a message which they have
received,
instead of keeping it to themselves and/or
forgetting it. And this in turn depends on how much prestige
can be gained from repeating the story they have heard (by
394
P. VANPARIJS
confirming opinions they expressed previously, by showing

access to a privileged source of information or by giving
news which will please the listener) and on how much they
want the listeners to know the story (Rouquette, 1975, p.
58-59,88). Moreover, the development of a rumor also depends
on how likely the listeners are to accept what they hear
(from an anti-semitic anecdote to a prophetic proclamation),
instead of dismissing it as slanderous or ludicrous. And
this in turn depends at least partly on the listeners'
interests and wishes (13).
Following a suggestion made by Allport (1954, p.14), I
shall call accommodation
the
complex
mechanism thus
sketched. Clearly, the operation of this mechanism is not so
powerful and ubiquitous as to make all thinking wishful, or
to make credible whatever is desirable. The "principle of
reality" - as embodied reinforcement - imposes constraints
on what can become and remain socially accepted belief under
the pressure of the 'principle of pleasure' - as embodied by
accommodation. When these constraints are fairly loose,
however, as is the case with supernatural beliefs, the
process of accommodation is likely to play a powerful role,
by determining evolutionary attractors on which
religious
systems will tend to converge. Even when the constraints are
tighter, the intensity of feelings may be such as to
'distort' the perception of reality. In such situations,
of
which
racial prejudice is
typical,
evolutionary
attractors are sufficiently powerful
to pull socially
shared beliefs away from whatever cognitive attractor there
may be.
6. THE THEORY OF IDEOLOGY
The mechanism sketched in the previous section enables us
to make sense of a large number of functional explanations
of beliefs. But note that it is only relevant to those
explanations which account for beliefs by reference to their
fulfilling their bearer's wishes or interests. Very often,
however and most typically in the Marxist tradition, beliefs
are being explained, qua ideologies, by reference to the
fact that they serve someone else's interests, typically
those of the ruling class. In their most popular versions,
such explanations are clearly conspiratorial: powerful and
malevolent capitalists (and the like) instill into the
workers'
minds ways of thought which suit their own
interests.
Sophisticated
Marxists, however, vigorously
THE EVOLUTIONARY EXPLANATION OF BEUEFS
395
reject such explanations as being exceedingly simplistic

(see e.g. Althusser, 1965, p. 241-242; 1970, p. 102-105). Is
there any other way of making sense of attempts to explain
some society's
classes?
beliefs
by
the
interests
of
its ruling
First of all, the process of accommodation sketched

above generates a systematic tendency for beliefs to be in
some sense "congruent"
with the social
positions
of their
bearers. This tends to justify the rulers' position in their

own. 7yes, but also to reconcile the ruled with their own
pos1t10n.
Religious beliefs,
for example, may reduce
incongruence by playing down the importance of worldly goods
in the eyes of those who are deprived of them. In so doing,
they
systematically
tend
to
transform
contingent,
historically relative features into absolute, natural ones,
and thus to reproduce the existing social order (See e.g.
Barthes, 1957, p. 229-234, 243-244 and Bourdieu, 1971, p.
312-315). From this perspective, however, the contribution
of beliefs to the reproduction of the social order is not a
consequence which enables us to explain why people who have
no interest in this reproduction entertain the beliefs they
do. It is a systematic by-product of the beliefs to be
explained, not a function by reference to which they can be
explained.
However, there is another way in which we can try to
spell out the underlying mechanism. Accommodation implies
that social groups tend to accept and transmit beliefs which
suit their interests. There is therefore no reason to expect
a bias in favor of the interest of one or another group, at
least providing all positions in the network in which
beliefs spread are equivalent. However, it is presumably one
of the attributes of the ruling class that it is better
placed to propagate whatever belief its members happen to
hold. This lack of symmetry is not only due to the fact that
members of the ruling group have more opportunities to be
heard by members of the other groups (by virtue of having
easier access to the mass media, for example). It is also
largely due to the fact that they tend to enjoy greater
(cognitive) authority, which has the effect of lending more
weight to the messages they transmit (Cohen, 1978, p. 291).
Consequently, at least in a society in which there is no
total breakdown of communication between the various social
groups, there will exist a systematic tendency for the
'ruling ideology' to serve the interest of the ruling class.
Moreover,
in this case, the effect pointed out is not
just
396
P. VANPARIJS
a systematic by-product, but it is an effect by reference

to which
prevailing beliefs
can
be explained (14).
Consequently, this kind of story provides us with a way of
making sense of a non-conspiratorial theory of ideology with
the help of the mechanism of accommodation sketched above.
7. CONCLUSION
To sum up: our exploration of the evolutionary explanation

of beliefs has come up with two major difficulties for my
initial claim that all evolutionary mechanisms, whether in
the biological or in the social realm,
were either cases
of what I called "natural
selection"
(selection
of
features through selection of entities)
or
cases of
what I called "reinforcement" (selection of features within
entities).
The first difficulty can arise in
the case of
practices, but it necessarily arises in the case of beliefs.
When the features to be explained are cultural traits, they
are typically subject to "multiparental
transmission", i.e.
to transmission by others than the biological parents. There

is therefore no reason to expect those features to prevail
which maximize chances of survival or reproductive success only transmissive success.
And one cannot say
features are here being selected

elimination, or even through the
of entities carrying them. This
Darwinian mechanism in the cultural
either
that
through
the selective
selective reproduction,
natural extension of the
realm is no longer an
instance of natural selection as defined.
The second difficulty arises

as
one
tries to
reconstruct the mechanism which is presupposed by the
functional explanations of belief systems to be found in the
social-scientific
literature.
The
mechanism
of
accommodation, which shapes beliefs in such a way that they
fit the wishes of those who hold them, satisfies our formal
definition of reinforcement as an evolutionary mechanism
which selects features within entities. At the same time, it
seems to conflict head on with reinforcement as elaborated
in the case of practices, i.e. as a generalization of
operant conditioning. For accommodation, to put things very
grossly,
keeps pulling beliefs away from an accurate
perception
of reality,
whereas the
reinforcement of
practices
requires organisms (and
other entities) to
register causal links as they are in reality. The formally
397
defined category of reinforcement thus appears to be an

heterogeneous set, only one element of which corresponds to
the intuitive notion of reinforcement.
These
two
difficulties
directly
suggest
a
reformulation of the initial distinction. The first category
of evolutionary mechanisms should now be defined more
broadly in terms of differential transmission rather than
in terms of differential survival. It consists in mechanisms
which select features because of the transmissive efficiency
they confer upon organisms carrying them: reproductive
efficiency in the case
of biological characteristics,
"propagation efficiency" in the case of cultural traits. And
the second category should then consist of
mechanisms
which select features because of the gratifications they
yield to the entities carrying them.
It
covers the
accommodation of beliefs
and
operant conditioning as
particular
cases.
And
the
phenomenon
of "adaptive
preferences",
i.e.
the adjustment of
tastes to the
perception of possibilities (Elster, 1983, ch. 3), may
provide a further example of it.
However, a number of problems are still left unsolved
by this classification. Let me briefly mention four of them:
(1) Where does pure differential survival (without
reproduction) fit in? The selective dying out of societies,
for example,
cannot
be
subsumed
under differential
transmission - or should it?
(2) Where do we put the process by which we seek a
comfortable posltlon while asleep? This is not just a
homeostatic process - a process of "equilibration". It is an
evolutionary mechanism - a process of "equilibration through
the consequences"
-, in exactly the same way as the process
of accommodation
by
which
we
end
up entertaining
"comfortable" beliefs (I discuss this sort of example as a
threat to the exhaustiveness
of the natural-selection
/reinforcement classification in Van Parijs, 1982). But can
we just subsume it under the second category, along with
accommodation, operant conditioning and adaptive preference?
(3) What is the inner structure of the second category
of evolutionary
mechanisms?
Is the
fundamental difference
between accommodation and reinforcement the fact that one is

concerned with beliefs and the other
with practices?
Remember that reinforcement too can be viewed as essentially
concerned with beliefs. Or is the fundamental difference
between the two mechanisms the fact that one involves
immediate, and the other one more remote consequences? Note,
398
P. VANPARIJS
however,
that cooking habits,
for example, could be
reinforced by their immediate consequences for the pleasure
people take in eating, as well as by tbeir more mediate
impact on their health.
(4)
Last but not
least,
what do we make of
"evolutionary
epistemology"
in the Popperian
vein?
Is an
evolutionary mechanism, in the sense of a process of

selection according to causal (not logical) consequences,
involved in it? And, if so, is it a case of differential
transmission
(first
category)
or
of
differential
gratification (second category)?
It would be nice if we could construct a neat typology
in which these various problems could be elegantly solved. I
do not claim to have provided one here. But I do hope that
this paper will enable the discussion to go some way beyond
the (pretty unhelpful and desperately ambiguous) standard
distinction between Darwinian and Lamarckian mechanisms.
ACKNOWLEDGMENTS
(The author is a research fellow at the Belgian National
Science Foundation.) He is grateful to Leo Apostel, Sue
Black, Donald Campbell, Peter Halfpenny, Rom Harre, Karin
Knorr, Henry Plotkin, Gerhard Vollmer and above all to Celia
Heyes and Peter Richerson,

versions of this paper.
for useful
comments
on earlier
NOTES
(1) This may suggest the existence of an analogy between
the pairs beliefs/practices and genotypes/phenotypes.
But
such an analogy would be misleading.
What corresponds to
the genotype/phenotype distinction in the realm of cultural
evolution is rather the distinction between what linguists
call "competence" (an internalized set of rules) and what
they call "performance" (the application of those rules in
actual speech)
or,
more
generally,
between
norms
and
actual behavior. Just as genotypes

can
only generate
consequences and thus lend themselves
to the operation
of evolutionary
forces
by
expressing
themselves in
phenotypes, so norms can only have effects
and
thus
subject themselves to evolutionary mechanisms
through the
behavior they govern. When I talk about
the evolution of
practices, I am referring to such norm-governed behavior,
just as those who talk

characteristics
are
(phenotypic) features.
practices, on the other
"competence"
level,
399
about the evolution of biological

speaking
about
gene-controlled
The distinction between beliefs and
hand, can only be made at the
and only
in
so
far
as
this
level
becomes differentiated into objectives (goals, preferences,

values, etc.) on the one hand, and beliefs (representations,
information, etc.) on the other, instead of just consisting.
in a set of instructions.
(2) Doubts are allowed, if only because I do not want
to go here into the tricky issue of giving the notion of
belief a precise definition.
(3)
Richerson & Boyd (1985) still refer to this case
as "natural selection on cultural

"transmissive
success"
certainly
variation". After all,

depends
on survival
prospects and often also on reproductive success. Moreover,

even if it did not, it would still depend, for example, on
not being eliminated from the pool of teachers. And this
seems enough to warrant using the term "natural selection".
Indeed, some may want to keep using the term for any
mechanism consisting in the unintentional filtering of blind
variants, including whatever has been subsumed above under
the heading of reinforcement. But the important issue, of
course, is not how we are going to call the categories we
find useful to introduce, but rather what categories are
useful to introduce if we want to shed some light on this
confusing area. Equivalent results can be achieved with very
different labels.
(4) See Berger and Luckmann's concept of "plausibility
structure", which - as they suggest themselves (ibid. 206)is in much the same relationship to beliefs as the concept
of "reference group" is to practices (Berger and Luckmann,
1966, p. 154-158).
(5)
A notable
exception
is
the
so-called "strong
progrBJll" in the sociology of knowledge, whose basic claim is
that true and false beliefs should treated

Barnes, 1974; Bloor, 1976).
(6)
reference
labelled
alike (See e.g.
Note that this

functional
definition (by
to class interests) is distinct from what is so
for
example
by Schaff (1967,
p. 50): an
400
P. VANPARIJS
(uninterestingly neutral) definition

belief about the desirable course of
development.
of ideology as any
social or individual
(7) The concept of "attractor" is borrowed from the

literature on catastrophe models (e.g. Zeeman, 1977). I
discuss the potential and limits of its use in the social
sciences in Van Parijs, 1978.
(8) See Van Parijs (1981, section 20). Note that

cognitive attraction can also be conceived as a filtering
process, though not one operating through the (objective)
consequences of the items which are being filtered (which
would turn it into a variety of evolutionary attraction).
Selecting scientific statements by checking their logical
consequences
against
empirical
different from selecting

pleasing consequences.
belief
observations
because
is
quite
it generates
(9)
At least providing it can be assumed that
maintaining them involves a cost. If this is the case, this
cost could be incorporated into the evaluation of the
consequences and the absence of the practice could then be
viewed as the evolutionary attractor. Even so, however,
beliefs would still remain different from practices in so
far as there is nothing like "cognitive attraction" applying
to the latter.
(10) See Runciman (1969, p. 173-174) for an explicit
formulation of this presumption in the context of the
explanation of religious beliefs.
(11) From an (NS-)evolutionary perspective, the very
existence of a mechanism pulling beliefs away from cognitive
attractors
is pretty puzzling.
For
whereas one can
understand that natural selection may have "wanted" to
transform weak cognitive attractors into strong evolutionary
attractors (to use an expression aptly suggested to me by
Peter Richerson), it is harder to see why and how a
mechanism could have come about and maintained itself which
generates a systematic discrepancy between what one would be
right in believing and what one takes pleasure in believing.
The answer is bound to be, roughly, that it is not always in
one's (true) interest to know the truth - or to know that
there is no truth. What I want to do in this paper, however,
401
is argue that such a mechanism is actually presupposed by

social-scientific explanations and outline its nature, not
attempt to explain why it exists.
(12) For analogies between "rumor work" and "dream
work", see, for example, Allport & Postman (1947, p.
134-136) and Rouquette (1975, p. 70-81). On the tendency for
rumors
to
substitute
consonance
Rouquette (1975, p. 58-59).
for
dissonance,
see
(13) As Francis Bacon put it: "for man always believes

more readily that which he prefers". See also Pareto (1916,
section 78) and Bourdieu & Passeron (1970, p. 40-41), who
emphasize the importance of selective reception.
(14) As pointed out by Elster (1982, p. 141-142),

however, one should be careful to distinguish between
serving a group's interest in the sense of promoting its
material interests and in the sense of fulfilling its wishes
(reducing dissonance,
etc.).
The
mechanisms suggested can
justify no more than explanations which

in the latter sense.
take the expression
Part V:
Bibliographies
EVOLUTIONARY EPISTEMOLOGY BIBLIOGRAPHY

Donald T. Campbell, Cecilia M. Heyes,
and Werner G. Callebaut
While this bibliography aspires to completeness, the field is growing
so rapidly, and is published in such a wide variety of outlets, that we
feel sure there are major omissions.

information about,
and reprints of I
(We would appreciate receipt of

such items.) This is a successor to
the
most
complete
previous
bibliography
(Campbell,
D.T., 1974,
Evolutionary Epistemology. In P.A. Schilpp (Ed. L The philosophy of Karl
R. Popper. The library of living
philosophers.
LaSalle, IL: Open
Court Publlshlng
Company.
Volume 14 1,
413-463.),
and of its 125 or so
entries we have repeated 40 which we judge to be of contemporary

relevance and/or to have been unduly
neglected.
These are marked by an
asterisk. Needless to say, we have not seen all of the entries listed. We
have made use of the published and unpublished bibliographies of many
persons, especially F .M. Wuketits and G. Vollmer.
*Ackermann J R.J. (1970). The philosophy of science. New York: Pegasus.

Ackermann, R.J. (1976). The phl.losophy Of Karl Popper. Amherst: University
of Massachusetts Press.
Adams, M.B. (1979). From "gene fund" to "gene pool": On the evolution of
evolutionary language. In W. Coleman & C. Limoges (Eds.), Studies in
history of biology, 3. Baltimore & London: Johns Hopkins Unl.versl.ty
Press, 241 285.
Albert,
H.
(1984) .
Die
Moglichkei t
der
Erkenntnis.
Von
der
transzendentalen Fragestellung zur evolutionaren Erkenntnistheorie.
Address delivered at the Simposio International sabre la Filosofia
de Karl Popper.
Madrid,
6-9 de Noviembre, 1984, Universidad
Complutense. Prof. Manuel Garrido, organizer.
Almeder, R.F. (1973). Science and idealism. Philosophy of Science, 40 (2),
242-254.
Almeder, R. (1975). Fallibilism and the ultimate irreversible opinion.
American Philosophical Quarterly, Monograph Series, 9, 33-54.
Alston. W.P. (1976). Has foundation been refuted? Phl.losophical Studies,
29, 287-305.
Alt,
J. A. (1980). Vom Ende der Utopie in der Erkenntnistheorie:
Popper I s
evolutionare Erkenntnl.stheorl.e
und l.hre
praktJ.Schen
Konsequenzen. Meisenheim: Hal.n, 185 pp. (Esp. 1722, 3238, 93102.)
Anderson.
R.
( 1983).
The epistemological status of a naturalized
epistemology. ~4Juiry, 26(3), 333-344.
Anderson. R.E. (19
. The role of cognitive animal ethology.
Paper
presented
at
the
conference
on
"Integrating
Scientific
Disciplines," Georgia State University, Atlanta, GA, May 3-5, 1984.
Annis. D.B. (1982). Epistemology naturalized. Metaphilosophy, 3(3/4),
201-208.
Apostel,
L.
(1970). Naar een verge lij kende Evolutietheorie. Studie
Philosophica Gandensia, 8, 35-127 (Esp. 92-94.)
Apostel,
L.
(1980).
Construction
and
validation in contemporary
epistemology. In L. Apostel, G. Cellerier,
R.
Garcia et a1. J
Construction and validation of scientific theories: The a roach of
~~9~;~~. epl.Stemology.
Gen ve:
on atl.on Arc l.ves
l.aget,
Apostel, L. (1982). The future of Piagetian logic. Revue Internationale de
Philosophie, 36( 142/143), 567-611.
Apostei,
L. (1987). Evolutionary epistemology, genetic epistemology,
history and neurology. (This volume.)
*Ashby, W.R. (1952). Design for a brain. New York: Wiley.
405
D. T. CAMPBELL ET AL.
406
*Auger, P. (1952). L'Homme microscopique: Essai de monodologie. Paris:

F lammar ion.
*Auger, P. (1962). The methods and limits of scientific knowledge. In On
modern physics. New York: Collier, 93-125.:..
Baldwin.
J.M. (1909). Darwin and the humanities. Baltimore:
Publishing Co. Reissued, New York: AMS Press, 1980, 118 pp.
Review
*Barr, H.J. (1964). The epistemology of causality from the point of view of
evolutionary biology. Philosophy of Science, 31, 286-288.
Bartley, W.W., III. (1976). The phl.losophy of Karl Popper, I: Biology and
evolutionary epistemology. Philosophia, 6(3/4), 463-494.
Bartley, W. W. , III. (1982). Philosophy of biology versus
philosopby of
physics. Fundamenta Scientiae, 3( 1), 55-78.
Bartley, W.W., III. (1982). A Popperian harvest. In P. Levinson (Ed.),

In pursuit of truth: Essays on the philosophy of Karl Popper on the
occaSl.on of hl.s 80th bl.rthday.
Atlantl.c Highlands, NJ: Humanl.tl.es
Press & Sussex: Harvester Press, 249-289.
Bartley,
w. W., III. (1983a). Non-justificationism: Popper versus
Wittgeostein. In P. Weingartner & J. Czermak (Eds.), Epistemology and
philosophy of science. Wien: Holder-Pichler-Tempsky.
Bartley, W.W., III. (19B3b). The challenge of evolutionary epistemology. In
International Cultural Foundation (Ed.).
Absolute values and the
creation of the new world.
New
York:
Internatl.Onal cultural
Foundatl.on Press,
2, 835-880. To be reprinted in G. Radnitzky & W.
W. Bartley (Eds.), Evolutionary epistemology, theory of rationality,
and the sociology of knowledge. LaSalle, IL: Open Court, 1986.
Bartley, W.W . III. (1984). Knowledge is a product not fully known to its
producer. In K. Leube & C. Nishiyama (Eds.) The road to serfdom
after 40 years. Festschrift for F. A. Hayek's 85th bl.rthday. Munchen:
Phl.losophl.a. Reprl.nted ln G.
Radnl.tzky & W.W. Bartley (Eds.),
Evolutionary epistemology, theory of rationality, and the sociology
Bateso~~ k~~Wl(1e;9)~a:~*~e~n~L~a~~:: C~U~!~e!::~y

238 pp.
unity. New York: Dutton,
Baumgartner,- H.M. (1981). fiber die Widerspenstigkeit der Vernunft. sich

aus
Geschichte
erklaren
zu
lassen.
Zur
Kritik
des
Selbstverstandnisses der evolutionaren Erkenntnistheorie.
In H.
Poser (Hg.), Wandel des Vernunftbegriffs. Munchen: Alber.
Baumgartner, H.M. (1982). Erel.gnl.s und Struktur als KatetJ0rien einer
geschichtlichen Betrachtung der Vernunft. In N. Luyten (Hg.), Aufbau
der Wirklichkeit.
Struktur und Ereignis II.
Freiburg & MUnchen:
Alber. 175-224.
Baumgartner, H.M. (1984). Die innere Unmoglichkeit einer evolutioniren
Erklirung der menschlichen Vernunft. In R. Spaemann, P. Koslowski, &
R. Low (Hrsg.). Evolutionstheorie und menschliches Selbstverstindnis.
Weinheim: Verlag Cheml.e, 55 71.
Bechtel, W. (1984). The evolution of the understanding of the
cell: A
study in the dynamics of scientific progress.
Studies
in the
History and Philosophy of Science, 15(4). 309-356.
Bechtel,
W. & Rl.chardson, R. (1983). The evolution of complexity:
evolutionary
perspectives
on
the
mind-body
problem.
Australasian Journal of Philoso h , 61(4), 378-395.
Benesc,.
7 . Der Ursprung des Gel.stes. MUnchen: DTV.
Berkson, W. & Wettersten, J. ( 1982). Lernen BUS dem Irrtum: Die Bedeutung
von Karl Poppers Lerntheorie fur dle
psychologl.e
und
dl.e
PhJ.losophJ.e
der Wl.ssenschaft.
Hamburg:
Hoffmann
und
Campe.
Translatl.on: Learnl.ng from error. La88lle,IL: Open Court, 1984.
Bickhard,
M.H.
(1979). On necessary and specific capabilities
in
evolution and development. Human Development, 22, 217-224.
Bickhard, M.H. (1982). Automata theory, artifl.cial intelligence
and
genetic
epistemology.
Revue
Internationale
de
Philosophie,
36(142/143), 549-566.
407
Bjerring, A.K. & Hooker. C.A. (1979). Progress and process: The nature
of systematic inquiry.
In J.
Biirmark (Ed.). Perspectives in
metascience. Goteborg: Kungl Vetenskaps-och vitterhets-Samhallet . J
57 74.
Blachowitz, J. A. (1971). Systems theory snd evolutionary models of the
development of science. Philosophy of Science J 38 J 178-199.
Blazek, B. (1978). On scope and l~ml.ts of analog1es between evolution aDd
cognition. Proe. Syrnp. Natur. Select, Prague, 543-558.
Blazek, B. (1979). Can epl.stemology as a philosophical discipline develop
into a science? Dialectica, 33, 87-108.
Bloor,
Boden,
D.
(1983).
Macmillan, 175
Wl.ttgenstel.n:
&
A social theory of
206, notes 38, 39, 40, 43.
knowledge. London:
M. (1980). The case for a cognitive biology. Proceedings of the

Aristotelian Society, Supplementary Vol. 54, 25-49.
G.
(1977).
Models for the development of science. In I.
Spiegel-Rosing & D. de Solla Price (Eds.), Science, technology
and
society: A cross-disciplinary perspective. London & Beverly Hl.lls:
Sage, 319 351.
Boon, L. (1983). De list der wetenschap. Variatie en selectie: vooruitgan
zonder rationa ltelt. aarn:
0,
5 pp.
Boon. L. (1987). Van.ation and selection: Scientific progress without
rationality. (This volume.)
*Bonsack, F. (1961). Information, thermodynamique, vie, et pens~e. Paris:
Gauthier-Villars.
Borel, M.J. (1978). Discours de la logique et logique du discours.
Lausanne: L' Age d Homme.
Boulding, K.E. (1978). Ecodynamics: A new theory of societal evolution.
Beverly Hills, CA: Sage.
Boulding, K.E. (1980). Science: Our common heritage. Science, 207(4433),
831-836.
--Boyd, R.N. (1985). Lex orandi est lex credendi. In P M. Church land & C.A.
Hooker (Eds.), Images of science. Chicago: University of Chicago
Press, 3-34 (esp. 23-32).
Boyd, R. & Richerson, P.J. (1985). Culture and the evolutionary process.
Chicago: Chicago University Press.
Bradie, M. (1982). A Darwinian approach to the evolution of science.
Proceeding of the Ohio Philosophical Association
(Annual Meeting),
in Toledo, April 3, 1982, 77-85.
Bradie,
M. (1986). Assessing evolutionary epistemology. Biology and
Philosophy, 1(4), 401-459.
Braitenberg, V. (1984). Vehicles: Experiments in synthetic psychology.
Cambridge, MA & London: MIT Press.
Briskman. L. (1974). Critical study: Toulmin's evolutionary epistemology.
Philosofhicai Quarterly, 24, 160-169.
Brooks, H.1980). Technology, evolution, and purpose. Daedalus, 109,
65-81.
--Brown, J.R. (1982). Rescher's evolutionary epistemology. Philosophia.
15(3), 287-300.
Bunge, M. (1979). The mind-body problem in an evolutionary perspective.
Brain and Mind, October 1979, Excerpta Medica (Clba Foundation Series
69) .
Bunge,
M. (1983). Treatise on basic philosophy I v; Epistemology &
methodology I: EXPlor~n, the world. Dordrecht & Boston: Re~del.
(Esp.6-a, 50-59, 67-70, 03 loa, 233 241, 286-290.)
Burhoe, R.W. (1976). The source of civilization in the natural selection
of coadapted information in genes and culture. Zygon, 11, 263-303.
Burhoe l R.W. (1981). Toward a scientific theology. Belfast: Christian
Journals Limited.
Bohme,
408
Burkhardt,
R.W..
Jr.
(1982).
The
evolutionary
development
of
an
evolutionary ethology. June, 1982, Darwin Centenary

Conference i n
Cambridge, England. Department of History, University of Illinois,
Burks,
Urbana, IL 61801.
A.W. ( 1977). Chance, cause, reason: An inquiry into the nature of
scientific evidence. ChIcago: UnIversIty of ChIcago Press, 694 pp.
Callebaut,
W. ( 1982). Evolutionaire epistemologie: Enkele
kritische
en
constructieve overwegingen. Ghent: Communication & Cognition, 72 pp.

Callebaut. W. (1983). Bi'dra e tot een al emene rationaliteitstheorie op
evolutionaire qronds 8?, met een toepasslng op
e organl.satle van
wetenschappehJke ken01S. Unpubhshed Ph.D. thesIs, Unlverslty of
Ghent, 736 pp.
Callebaut, W. (1984). The scientific enterprise as a distr i buted
problemsolving system. In F. Vandamme (Ed.), Representation, use and
acquisition of knowledge or the manufacture of knowiedge. Namur:
ASSOc~8tl.On Internat l. onaie de Cybernetl.que, 76 84.
Callebaut, W. (1987). Why it makes sense to extend the genotype/phenotype
distinction to cultur-e. La Nuova Cr-itica (in pI"ess).
Callebaut, W. ( 1987). Vel's une eplostemologle biologique? La philosophie
au'ourd'hui - Question dis utees, 34, Universite de Liege.
Callebaut, W.
1
. Wl.ssensc a tdynamik: eiDe Metawissenschaft wiirde
geboren. To appear in H. Stachowiak (Ed.), Pragmatics, 5, Pragmatic
Tendencies in Epistemology. Hamburg: Felix MeloneI'.
Callebaut, W. & pl.nxten, R. (19878). Evolutionary epistemology today:
Converging views from philosophy.
the natural and the social
sciences. (This volume.)
(!
calleb~~itiP~~adfgm p~~~~;~~ . RDO~~~:~~t 9:~=f~n7V~;~j!1~a(~h~;i:~j:;~'y: A

Callebaut. W. & Van Bendegem, J.P. (1982). The
distribution approach
to problem-solv i ng in science: Prospects
for
general systems
methodology. In R. Trappl (Ed . ), Cybernetics and systems research.
Amsterdam & New York: North-Holland, 51 56.
*Campbell, D. T. (1959). Methodological suggestions from a comparative
psychology of knowledge processes. Inquiry. 2, 152-182.
*Campbell. D.T. (1960). Blind variation and selective retention
in
creative thought as in other knowledge processes . Psychological
Review, 67, 380-400.
*Campberr,-D . T. (1973). Ostensive instances and entitativity in language
learning. In N. D. Rizzo (Ed.), Unity thr-ough diversity. New York:
Gordon & Breach.
Campbell, D.T. (1974b). Unjustified variation and selective retention in
scientific discovery.
In F.J.
Ayala & T. Dobzhansky (Eds.).
Studies in the philosophy of biology . London : Macmillan, 139-161.
Campbell, D. T . (1914c).
Downward causation" in hierarchically organized
biological
systems.
In
F. J.
Ayala & T. Dobzhansky (Eds.).
Studies in the hilosophy of biology. London: Macmillan, 179-186.
Campbe ll ,
D.T.
1977a .
DescriptJ.ve
epistemology:
PsychologIcal ,
sociological, and evolutionary. William James Lectures, Harvard
UniverSity, Spring, 1977. ( Unpublished, duplicated copies available.)
Campbell, D.T. (1977b). Discussion comment on "The natural selection model
of concep t ual evolution." Philosophy of Science, 44(3). 502-507.
Campbell. D. T. (1979). A trlobal model of the social system vehicle
carrying scientific knowledge. Knowledge, 2, 181-201.
Campbell, D. T. (1982). The " b lind- v8n8tJ.on - and-selective-retention lt theme.
In J.M. Broughton & D.J. Freeman-Moil' (Eds.)' The
cognitivedeve l o mental s cho l ogy of James Mark Baldwin: Current thear and
research 1D genetlc epistemology. Norwood, NJ: Ablex, 87 7.
409
Campbell, D. T.
(1986a). Science's social system of validity-enhancing
collective belief change and the problems of the social sciences. In
D.W. Fiske & R.A. Shweder (Eds.), Metatheory in the social sciences:
Pluralisms and subjectivities. Chicago: Unl.versl.ty of Chl.cago Press,
108 135.
Campbell, D.T.
(1986b). Science policy from a naturalistic sociological
epistemology. In P.O. Asquith & P. Kitcher (Eds.),
PSA 1984,
2. East Lansing, MI: Philosophy of Science Association.
Campbell, D.T. (1987). Neurological embodiments of belief and the gap in
the fit of phenomena and noumena. In A. Shimony, D. Nails, & R.S.
Cohen (Eds.), Naturalistic epistemology: A symposium of two decades.
Dordrecht & Bas ton: Re.1del.
Campbell, D. T. (1 987b). Selection theory and the sociology of scientific
validity. (This vOlume.)
Caporael, L.R. (1987). Homo sapiens, homo faber, homo socians: Technology
and the social animal. (This volume.)
Carrier, L. S. (1 9aO). Perception and animal belief. Philosophy, 55,
193~209 .
Cellerier,
G. (1984). Genetic epistemology Human Development, Sept.
(Special double issue on ontogenesis and
phylogenes.1s organized
by Archives Jean Piaget.)
Changeux, J.-P., Heidmann, T., & Patte, P. (1984). Learning by selection.
In P. Marler & H.S. Terrace (Eds.), The biology of learning. Berlin &
New York: Springer, 115-133.
Cherniak, C. (1986a). Limits for knowledge. Philosophical Studies, 49,
1-18.
Cherniak, C. (1986b). Minimal Rationality. Cambridge, MA: MIT Press
(Bradford Books).
*Child, A. (1946). On the theory of the categories. Philosophy and
Phenomenological Research, 7, 316-335.
Chomsky, N. (1966). Cartesian linguistics: A chapter in the history of
rationalistic thought. New York & London: Harper & Row.
Chomsky, N. (1968). Language and mind. New York: Harcourt, Brace & World.
Churchland, P.S. (1986). Neuroph11osoph : Toward a unified science of the
mind-brain. Cambridge, MA: MIT Press Bradford Books .
Churchland, P. S. & Church land , P.M. (1983). Stalking the wild epistemic
engine. Nalls, n( 1), 5-18.
Churchland, P~ (1979). Scientific realism and the plasticity of mind.
Cambridge: Cambridge Univers.1 ty Press.
Churchland,
P.M.
(1984).
Matter and consciousness: A contemporary
introduction to the philosophy of mind. Cambridge, MA: MIT Press.
Churchland, P.M. ( 1985). The ontolog~cal status of observables: In praise
of the superempirical virtues.
In P.M.
Churchland & C.A. Hooker
(Eds.), Images of science.
Chicago: University of Chicago Press,
35-47 (esp. 45).
Church land , P.M. & Hooker, C.A. (Eds.). (1985). Images of science: Essays
on realism and empiricism. Chicago: Chicago Univers~ty Press.
Clark, A. J. ( 1982). Evolut~onary epistemology and the dialectics of
meaning. In P. Weingartner & J. Czermack (Eds.), Epistemology and
philosophy of science. Wien: Holder-Pichler-Tempsky.
Clark, A. J. ( 1984). Evolutionary epistemology and ontological realism.
Philosophical Quarterly, 34, 482-490.
Clark, A.J. (1983). Meaning and evolutionary epistemology. Theoria, 49,
23-31.
--Clark,
A.J.
(1985). Natural anti-realism. Unpublished Ph.D. thesis,
University of Stirhng, Scotland.
Clark, A.J. (1986). Evolutionary epistemology and the
scientific method.
Philosophica, 37, 151-162.
Clark, A.J. (1987). The philosophical significance of an evolutionary
epistemology. (This vOlume.)
410
Cohen,
J.L.
(1973).
Is the
progress
of
science
. Journal for Philoso hy of Science I 24( 1 ) I
evolutionary? British
41-61 .
---
J.L.
1974. Professor Hu 1 and the evolution of science.
British Journal for the Philosophy of Science. 25(4). 334-336.
Constant,
E.W.
II.
(1978). On the d~versity and co-evolution of
Cohen,
technological
multiples:
Steam turbines
and
Social Studies of Science, 8, 183-210.

Cooper, D. (1979). The argument from evolution.
Aristotelian Society. 53, 207-223.
Corning,
P.A.
('983).
The synergism hypothesis:
Pelton water wheels.

Proceedings
of
the
A theory of progressive
evolution. London: Blond & Brl.ggs.

Cresswell, M.S. (1979). Can epistemology be naturalized? In R.W. Shahan &
C. Swoyer (Eds. L Essays on the philosophy of W. V. Quine. Norman:
University of Oklaboma Press, 109 118.
Cummins, R. (1975). Functional analysis. Journal of Philosophy, 72,
750-751.
Currie,
G.
(1978). Popper's eVOlutionary epistemology: A critique.
Synthese, 37, 413-431.
D'Agostino, F. (1983). Darwinism and language. In D. Oldroyd & I. Langham
(Eds.), The wider domain of evolutionary thought. Dordrecht & Boston:
Reidel, 159-173.
Darden, L. & Maull, N. (1977). Interfield theories. Philosopby of Science,
44, 43-64.
Delbruck,
M.
(1986).
Mind from matter? An essay on evolutionary
Dell,
e~~F~em&lfo~li:~~a~~
(~~~~).e~~~dnUng
tir.r.:
durch Fluktuation": Hine
evolutionare Epistemo!ogie flir menschliche Systeme. Familiendynamik,
6(2), 104-122.
De Hey. M. (1972). Kennis en evolutie: Vergelijkende psychologie van
cognitieve processen. In L. Apostel et a1. (Eds.). Eenheid van de
kultuur. Meppel: Boom.
De Me~ (1982). Genetic epistemology and integrated science of
science. Revue Internationale de Philoso hie. 36(142-3), 636-679.
Dennett, D.C.
1971. IntentJ.onal systems. Journal of Philosophy. 68(4),
87-106. Dennett, D.C. (1978). Brainstorms: Ph~losoph~cal essays
on mind and psychology. Montgomery, VT: Bradford Books, MIT Press.
253 pp.
Dennett, D.C. (1981a). Three kinds of intentional psychology. In R. Healey
(Ed. L Reduction, time and reality. Cambridge: Cambridge University
Press, 37 61.
Dennett, D.C. (1981b). True believers: The intentional strategy and why it
works. In A.F. Heath (Ed.), Scientific explanation. Oxford: Oxford
University Press.
Dennett, D.C. (1983). Intentional systems in cognitive ethology:
The
"Panglossian paradigm" defended. Behavioral and Brain Sciences, 6,
343-390.
Dennett, D.C. (1984a). Elbow room: The varieties of free will worth
wanting. Cambridge, MA: Bradford Books: MIT Press, 183 pp.
Dennet~.
(1984b). Cognitive wheels: The frame problem in artificial
intelligence. In C. Hookway (Ed.), Minds, machines and evolution.
Cambridge: Cambridge University Press, 129-151.
Dennett, D.C. (1986a). Evolution, error and intentionality. Unpublished
manuscript CCM-86-1. Center for Cognitive Studies, Tufts University.
Medford, MA 02155.
Dennett, D.C. (1986b). The myth of original intentionality. To appear in
R. Viale (Ed.), Proceedings of the Conference, Human Mind-Artificial
Mind, Turin, Italy, Apn.l, 1985. Center for Cogn~t~ve Studies, Tufts
unIVersity, Medford, MA 02155.
EVOLUTIONARY EPISTEMOWGY BIBLIOGRAPHY

Depew,
411
D.J. & Weber, B.H. (1985). Innovation and tradition in evolutionary

theory: An interpretive afterword. In D.J.
Depew & B.H. Weber
CEds.), Evolution at a crossroads: The new biology and the new
philosophy of Bel.ence. cambn.dge, MA: MIT Press, 227 260.
Derksen, A.A. (1980). Rationaliteit en wetenschap. Assen: Van Gorcum,

358pp. (Esp. 162-169).
de Vries. G. (1982a). The collectives of 'normal' and I functionalized I
sciences. Dynamics of Science. 2. 268-281.
de Vries, G.
( 1982b). De ontwikkeling van wetenschappelijke kennis,
sociologisch beschouwd. Kennie en Methode, 6(3), 190-220.
Ditfurth, H. V. (1987). Evolution, Transzendenz und Selbsttranszendenz.
In
R. Riedl & F.M. Wuketits (Eds.), Die evolutionire Erkenntnistheorie.
Berlin: Parey.
Dretske, F. (1971). Perception from an epistemological point of view.
Journal of Philosophy, 68, 584-591.
Dretske, F. ('981). Knowledge and the flow of information. Cambridge, MA:
MIT Press, vii, 145, 243.
Dupreel, L. (1949). Essais pluralistes. Paris: Presses Universitaires de
France (Chapter XII: Vers une theorie probabiliste de la vie et de la
connaissance) .
Durkheim,
E.
(1965). Elementary forms of religious life. Translated by
J.W. Swain. New York: Free Press & LondOn: RaCMl.l1an.
Dvorak,
J.
(1983 ~. Historische Bemerkungen zur 'biologischen' oeler
'evolutionaren Erkenntnistheorie. Conceptus, '7, 57-70.
Edelman, G.M. & Mountcastle, V.B. (197~indful brain. Boston: MIT
Press.
Edelman, G.M. & Reeke, G.N. (1982). Selective networks capable
of
representing
transformations,
limited
generalizations,
and
associative memory. Proceedings of the National Academy of Sciences,
79, 2091-2095.
Egidi, R. (1986). Emergence, reduction, and evolutionary epistemology: a
commentary.
In G.
Radnitzky
&
W.W.
Bartley
III (Eds.),
Evolutionary epistemolo , theory of rationality, and the sociology
o
now e ge. LaSalle,
:
en Court.
l.gen,
. & l.n er, R.
(1975). Das S iel: Naturgesetze steuern den Zufall. Munchen: Piper.
Ellis,
B.
1 7
Ratl.ona
e l.e
systems. Totowa, NJ: Rowman &
Littlefield.
Elster, J. (1979). Ulysses and the sirens. Cambridge, MA: Cambridge
University Press.
Engels, E.-M. (1983). Evolutionare Erkenntnistheorie - ein biologischer
Ausverkauf
der
Philosophie?
Zeitschrift
fur
allgemeine
Wissenschaftstheorie, 14( 1), 138-166.
Engels, E.-H. (1984). Was leistet die evolutionire Erkenntnistheorie?
Eine
Kritik
und
Wiirdigung.
Zeitschrift
fur
allgemeine
Wissenschaftstheorie, 16( 1), 113-146.
Engels,
E. M.
(1985a).
Die evolutionire
Erkenntnistheorie in der
Diskussion. Information Philosophie, 13(1), 56-63, (2), 49-68.
Engels, E.-M. (1985b). Evolutl.Onare Erfahrung und Realismus. Spiel, 4(1),
41-69.
Esposito, J.L. (1975). Remarks toward a general theory of organization.
Journal of General Systems, 2(3), 133-143.
Esposito,
J. L. ( 1980). Evolutionary metaphysics: The development of
Peirce's theory of categorl.es. Athens, OH: Ohl.o Unl.versloty Press.
Evans, R.I. (1975). Konrad Lorenz: The man and his ideas. New York:
Harcourt Brace Jovanovl.ch.
Faber, R.J. (1984). Feedback, selection, and function: A reductionistic
account of goal-orientation. In R.S.
Cohen
&
M.W.
Wartofsky
(Eds')J Methodology, metaphysics and the history of science: In
memory of BenJaDll.n
Felson.
Dordrecht & Boston: Rel.del, 43-135.
(Especl.ally append loX 7: Learning by 'variation and selection',
131-132. )
412
D. T CAMPBELL ET AL
Falk, A.E. (1981). Purpose, feedback, and evolution. Philosophy of Science,

48(2), 198-217,
Feibleman, J. (1977). Adaptive knowing: Epistemology from a realistic
Fetz,
standpoint. The Hague & Boston: N~Jhoff.
~82).
Pour
une ontologie genetique.
Jean
Piaget
et 1a
philosophle moderne. Revue InternatlOnale de Phl!osophle, 36( 142/143)
409 434.
Feyerabend, P.K. (1974), Popper. Objective Knowledge Inquiry, 17, 475-507.
Flohr,
H. (1985). Review of F.M. Wuketits,(ed.), 198~nal
and Biological Structures, 8, 395-399.
Florkin,
M.
(1974).
Concepts of molecular biosemiotics and
evolution. New York: Elsevl.er.
of
Social
of molecular
Fox, R~ The matter of mind. In R. Fox, The red lamp of incest. New
York: Dutton, Chapter 7, 166-192 & 252-257.
Freeman-Moil", D. J. (1982). Evolutionary epistemology. In J. M.
Broughton
&
D. J. Freeman~Moir (Eds.), The cognitive-developmental psychology
of James Mark Baldwin:
Current theory and research ln genetic
epistemology. Norwood, NJ: Ablex, 98 121.
Frehland, E. (Ed.). (1984). Synergetics - from-microscopic
to macroscopic
order. Berlin: Springer.
Frey, -G-,- (1980).
Moglichkeit
und
Bedeutung
einer evolutionaren
Erkenntnistheorie. Zei tschrift fur philosophische Forschung, 34( 4),
1-17 ,
Furth, H.G. (1981). Piaget and knowledge: Theoretical
foundations (2nd
ed.). Chicago: Universlty of Chlcago Press (1rst ed.: 1969).
Gamble, T. J. (1983). The natural selection model of knowledge
generation:
Campbell's dictum and its critics. Cognition and Brain Sciences, 6,
353-363.
Gellner, E. (1979). Pragmatism and the importance of being earnest. In I.C.
Jarvie & J. Agassi (Eds.), Spectacles
and predicaments. Cambridge,
MA: Cambridge University Press.
Georgescu-Roegen, N. (1971). The entropy law and the economic process.
Cambridge, MA! Harvard University Press. (Chapter I: SClenCe! A
brief evolutionary analysis).
Gerhardt, G. (1983). Wider die unbelebrbaren Empiriker. Die Ar~umentation
gegen empirische VerSl0nen der Transzendentalphl1osophle bel H.
Cohen und A. Rl.ehl. Wiirzburg: Konlgshausen & Neumann.
*Ghisehn, M. T. (1969). The triumph of the Darwinian
method. Berkeley:
University of California Press.
*Ghiselin, M.T. (1973). Darwin and evolutionary psychology. Science, 179,
964-968,
--Giere, R.N. (1984). Toward a unified theory of science. In J.T. Cushing,
C.F. Delaney, & G. Gutting (Eds.), Science and reality, Notre Dame:
Universi ty of Notre Dame Press, 5-31.
Giere, R.N. (1985). Philosophy of science naturalized. Philosophy of
Science, 52, 331-356.
Gillieron, C. (1987). Is Piaget I s "genetic epistemologyl! eVOlutionary?
(This vOlume.)
Gillispie, C. C. (1968). Remarks on social selection as a factor in the
progressivism of science. American Scientist, 56, 439-450.
Glassman, R.B. (1977). How can so little braln hold so much knowledge?
Applicability of the principle of natural
selection
to mental
processes. Psychological Record, 2, 393-415.
Goldman, A.I. (1975). Innate knowledge. In S.P. Stich (Ed.),
Innate
Ideas. Berkeley: University of California Press.
Goldma~I. (1979). Varieties of cognitive appraisal. Nous, 13, 23-38.
Goldman, A.I. (1985). The relation between epistemology-and psychology.
Synthese, 64,29-68.
Goldma~
(1986). Epistemology and cognition. Cambridge, MA: Harvard
Universi ty Press.
413
Goldman,
M. (1982). Science and play. In P.D. Asquith & T. Nickles

(Eds.), PSA 1982,1, East Lansing, MI: Philosophy of Science
Association, 406-414.
Goodwin, B.C.
(1982). Genetic epistemology and constructionist biology.
Revue Internationale de Philosophie, 36(142/143), 527-548.
Goudge, T.A. (1973), Pragmatam s contribution to an evolutionary view of
mind. Monist, 57(2), 133-150.
Gould, S.J. D9S0T. The panda's thumb. New York: Norton, 79,
Gratch, G. & Schatz, J.A. (1985). Cognitive development: The relevance of

Piaget I s infancy books. In J.D. Osotsky (Ed.) J Handbook of infant
development, 2nd edition.
Gregory, R. L. ( 1981). Mind in science. New York: cambridge University
Press.
Griesemer J J. (1983). Communication and scientific change: An analysis of
conceptual maps in the macroevolution controversy. Unpublished Ph.D.
dissertation, December 1983,
Universi ty of Chicago, Committee on
Conceptual Foundations of Science.
Griesemer, J.
(1984). Presentation and the status of theories.
In P. D.
Asquith & P. Kitcher (Eds.), PSA 1984,
1, East Lansing, MI:
Philosophy of Science Association~.
Grmek, M.D. (1976). Le role du hasard dans la genese des decouvertes
scientifiques.
Medicina Nei Secoli:
Official bulletin of the
Institute for the History of Medicine at the University of Rome, 2,
Haack,
277-305.
S.
(1975).
Hahn,
"1"7"2=182.
L.E. &
The
relevance of
psychology to epistemology.
Metaphilosophy, 6(2), 161-176.
Habermas,
J.
&
Luhmann, N.
(1971). Theorie der Gesellschaft oder
Sozialtechnologie - Was leistet dl.e Systemforschung? Frankfurt:
Suhrkamp. (Esp. 364 369)
Hahlweg,
K.
(1983). The evolution of science: A systems approach.
University Microfilm. Ph.D. 1983, The University of Western Ontario.
Canada, 44/10-A, 3086.
Hahlweg, K.
(1986). Popper versus Lorenz: An exploration into the nature
of evolutionary epistemology. In: A. Fine & P. Machamer (Eds.), PSA
1986, 1. East Lansing, MI: Philosophy of Science Association,
Schilpp, P.A. (Eds.) (1986). The philosophy of W. V. Quine.
LaSalle: Open Court.
H. (1984). Some introductory remarks on synergetics. In
E.
Frehland (Ed.), Synergetics: From microscopic to macroscopic order.
Berlin: Springer, 1 3.
Hamlyn, D. W.
(1978). Experience and the growth of understanding. London:
Rou tledge & Kegan Paul.
Harre,
R.
(1979).
Social be in : A theory for social
sychology.
Oxford:
Blackwell
Chapter 15:
Social
change:
Theor~es
and
assumpt ions) .
Harre, R.
(1980) KnOWledge. In G.S. Rousseau & R. Porter (Eds.), The
ferment
of
knowledge:
Studies
in
the
historiography OI
e~ghteenth century
science.
New York
&
Cambridge: Camhrl.dge
Unl.vers~ty Press, 11 54.
Harre, R. (1981a). The evolutionary analogy in social explanation. In U.J.
Jensen & R. Harre (Eds.), The philosophy of evolution. Brighton,
Sussex: Harvester Press, 161~175.
Harre, R.
(1981 b). The conditions for applying an evolutionary model to
the development of science: Commentaries on Mittelstrasse
[sic],
Laudan and Newton-Smith.
In U.J.
Jensen &
R. Harre (Eds.). The
philosophy of evolution. Brighton, Sussex: Harvester Press, 290-29~
Hattiangad1, J.N.
(1973). Mind and the origin of language. Philosophy
Forum, 14, 81-98.
Hattiangacrr,
J.N.
(1979).
Meaning,
reference
and
subjunctive
conditionals. American Philosophical Quarterly, 16(3), 197-205.
Haken,
414
Hattiangadi,
J.N.
(1982a).
Knowing
that,
Hattiangadi, J.N. (1982b).

32(2), 38-51.
Essence
versus evolution
Duplicated
manuscript.
Dept.
Downsville J Ontario M3J 1P3.
of
and
how (a
Philosophy,
new empiricism).
York
University,
in language. Word,
*Hawkins J D. (1964). The language of nature: An essay in the

of science. San Franc.lsco: Freeman. 46-52 & 252-254.
philosophy
Hebb, D.O. (1980). Essay on mind. Hillsdale, NJ: Erlbauffi.

Heffner, J. (1981). The causal theory of visual perception:
Its scientific
Herber~ (1981).
Marx,
basis and epistemological implications.

Quarterly, 21(3).
International Philosophical
Evolution as a learning process in
Plaget
and Habermas. Ph.D. Dissertation, University of California, Santa

Cruz, 42/07A, p. 3185 RXD81-29261, University Microfilms.
Heyes, C.M. (1984). Conspecific learning in the Syrian hamster. Doctoral
thesis, University of London.
Heyes, C.M. (1987). Cognisance of consciousness in the study of animal
knowledge. (This volume.)
Hines, S.M. (1979). Evolutionary epistemology and political
knowledge:
An inquiry into the epistemological basis of an emerging paradigm in
political science. In M.T. Falco (Ed.), Through the looking glass:
Epistemology and the conduct of inquiry,
an anthology. waslhngton,
DC: Un.lversity Press of America.
Holland, A. & O'Hear, A. (1984). Symposium: On what makes an epistemology
eVOlutionary. Proceedings of the Aristotelian Society, 58, 177-217.
Holton. G. (1973). Thematl.c Orl.gl.ns of SCl.ent.lfl.c Thought: Kepler to
Einstein. Cambridge, MA: Harvard Un.lvers.lty Press, 392-395.
Holzhe~983a). Evolutionare Erkenntnistheorie und okologische Krise.
Schweizer Monatshefte, 63(3). 215-227.
Holzhey, H. (1983b). Genese und Geltung. Das vernunftkritische Resultat
einer Kontroverse zwischen
biologischer
und
kantianischer
Erkenntnistheorie. Studia Philosophica, 42, 104-123.
Hooker,
C.A.
(1975).
Phl.losophy and
meta-philosophy
of science:
EmpiriCism, Popperianism and realism. Synthese, 32, 206-227.
Hooker, C.A. (1977). Methodology and systematic philosophy. In: R.E. Butts
& J.
Hintikka (Eds.), Basic problems in methodology and linguistic.
Dordrecht, 3-23.
Hooker, C.A. (1978a). An evolutionary naturalist realist doctrine of
perception. In W. Savage (Ed.), Perception and cognition: Views in
the foundations of psychology.
Ml.nneapOhs, HN: Un.lversl.ty of
Mlnnesota Press.
Hooker,
C.A.
(1978b).
Metaphysics,
reference
and
(meta-)theory.
Metaphilosophy, 9(2), 133-149.
Hooker, C.A. (1982). Understanding and control: An essay
on
the
~~~~~~~~al dynamics of human cognition.
Man-Environment Systems, 12,
Hooker, C.A. (1985). Surface dazzle, ghostly depths: An exposition and
critical evaluation of van Fraassen I s
vindication of empiricism
against realism.
In P.M.
Churchland &
C.A.
Hooker (Eds.),
Images of science. Chicago:
University of Chicago Press, 153-196
(esp. 172 180 and 187-188).
Horton, R. (1982). Tradition and modernity revisited. In M. Hollis & S.
Lukes (Eds.). Rationality and relativism. Oxford: Blackwell, 201-260.
HBvelmann,
G.
(1984). Sprachkrl.tlsche Bemerkungen zur evolutionaren
Erkenntnistheorie. Zeitschrift fur allgemeine Wissenschaftstheorie,
15(1 ).
Hull, D.L. (1973). A populational approach to scientific change. (Review
of S. Toulmin, 1972.) Science, 182, 1121-1124.
Hull, D.L. (1974). Are members of biological species similar to each
other? British Journal for the Philosophy of Science, 25(4), 332-334.

Hull,
415
D.L.
(1978). Altruism in science: A sociobiological
model of
cooperative
behaviour among scientists.
Animal Behaviour, 26,
685-697 .
Hull, D.L. (1980). Sociobiology: Another new synthesis. In G.W. Barlow &
J. Silverberg (Eds.), Sociobiology: Beyond nature/nurture? AAAS
Selected Symposium 35. Boulder, CO: Westv~ew Press.
Hull, D.L. (1982). The naked meme. In H.C. Plotkin (Ed.), Learning,
development and culture: Essays in
evolutionary epistemo~w
York: Wiley, 273 327.
Hull, D.L.
(1983). Conceptual evolution and the eye of the octopus.
Proceedings of the 7th International Congress of Logic, Methodology,
and Ph~losophy of Sc~ence. July 1 16, 1983, Salzburg, Austn.a.
Hull l D.L. (1984a). Cladistic theory: Hypotheses that blur and grow. In T.
Duncan &
T. Stuessy (Eds.), Cladistic
perspectives
on
the
reconstruction of evolutionary history. New York: Columbl.a University
Press.
Hull, D.L. (1984b). Darwinism as a historical entity. In D. Kahn (Ed.)'
The Darwinian heritage. Wellington, New Zealand: Nova Pacifica.
Hull, D.L. (1984c). Hl.storical entities and historical narratives. In C.
Hookway (Ed.), Minds, machines, and evolution. Cambridge: Cambridge
University Press.
,
Hull, D.L. (1984d). Lamarck among the Anglos. Introduction to reprinted
edition of: J.B. Lamarck's Zoological Philosophy:
An Exposition
with Regard to the Natural
Hl.story of Anl.mals. Chlcago: Chlcago
University Press.
Hull, D.L. (1985). Interactors versus vehicles. Paper presented at ICSEB,
Brighton, England, July 1985.
Husserl, E. (1970). Philosophie der Arithemetik, (Ed. L.Eley; orig. 1890).
(Esp. 359-371.)
Irrgang, B. (1986). Biologie als erste Philosophie? Uberlegungen zur
Voraussetzungproblematik und zum Theoriestatus einer Evolutionaren
Erkenntnistheorie. Philosophische Rundschau, 33(1/2), 103-121.
Isard, W. & Kaniss, P. (1978). Structure, control and language hierarchies
and world organization. Journal of Peace Science, 3(1), 63-91.
Isard, W. & Liossatos, P. (1979). Spatial dynamics and optimal space-time
development. New York: North Holland.
Izuzquiza, I. ( in press). Epistomologia y filosofias de la ciencia
postpopperianas I:
La ep~stemologl.a evoluc~on~sta.
Ov~edo:
El
Basilisco.
Jantsch, E. (Ed.). (1981). The evolutionary vision. Boulder, CO: Westview.
Jensen, U. & Harre, R. (Eds.) (1981). The Philosophy of evolution.
Brighton: Harvester Press.
Johnson,
D.M.
(1983).
Memory
and
knowledge: The epistemological
significance of biology.
American Philosophical Quarterly, 20,
375-382.
Johnston,
T .D. (1982). Learning and the evolution of developmental
systems. In H.C. Plotkin (Ed.), Learning, development, and culture:
Essays in evolutionary epistemology. New York: Wl.ley, 411 442.
Jones, W.T. & Chiaraviglio, L. (1979). Is science an adaptive system?
Behavioral Science, 24, 325-333.
Kaniss, P. ( 1978). Evolutionary change in hierarchical systems:
A
general
theory.
Regional Science Dissertation
and
Monograph
Series, 9. Ithaca, NY: Program
on Urban and Regional
Stud~es,
Cornell University.
Kanitscheider, B. (1981). Wissenschaftstheorie der Naturwissenschaften.
Berlin & New York: De Gruyter.
Kantorovich, A. (1983). The collective a priori in science. Nature and
System,S, 77-96.
Kantorovich, A. (1984). The collective a priori: A genotype/phenotype model
for the
growth of knowledge.
Institute
for
History
&
Philosophy,
Tel Aviv University.
Paper
presented
at
the
416
Symposium "New Directions in Evolutionary Epistemology, II held at the

150th Annual Meeting of the American Association for the Advancement
of Science, New York City, May 24-29, 1984.
Kary, C.E. (1982). Can Darwinian inheritance be extended from biology to
epistemology? PSA, 1, 356-369.
Kary, C.E. (1983).~n analysis of Darwinian evolutionary theory and a
critique of theories of evolutionary epistemology. Duplicated PhD.
dissertation, Dept. of Philosophy, University of Illinois at Chicago.
Kaspar, R. (1980a). Die Evolution erkenntnisgewinnender
Mechanismen.
Biologie in unserer Zeit, 1, 17-22.
Kaspar, R.
(1980b). Naturgesetz, Kausalitat und Induktion. Ein
Beitrag
zur theoretischen Biologie. Acta Biotheoretica, 29, 129-149.
Kaspar, R. (1984). A short introduction to the biological principles of
evolutionary epistemology. In F.M. Wuketits
(Ed.),
Concepts
and approaches in evolutionary epistemology. Dordrecht & Boston:
Rel.del, 51 67.
Kaspar,
R.
(1987).
Materialismus
Idealismus
und evolutionare
Erkenntnistheorie. Paper presented at symposium in Vienna, April
1986.
In R.
Riedl & F.M. Wuketits (Eds.), Die evolutionare
Erkenntnistheorie. Berlin: Parey.
Kekes, J. ( 1977). Popper in perspective. Metaphilosophy, 8( 1), 36-61.
Kekes, J. (1982). Scepticism and evolutionary epl.stemology. In R. Almeder
(Ed.), Praxis and reason. Washington, DC: University Press of
America.
Khin Zaw, S. (1980). The case for a cognitive biology. Proceedings of the
Aristotelian Society, Supplementary Vol. 54, 51-71.
Kitcher, P. ( 1983). The nature of mathematical knowledge. Oxford: Oxford
University Press.
Kitchener, R.F. (1980). Genetic epistemology, normative epistemology, and
psychologism. Synthese, 45, 257-280.
Kitchener,
R.F.
~
Is psychology relevant
to epistemology?
Duplicated
manuscript,
Dept.
of
Philosophy,
Colorado State
Universi ty.
Ki tchener, R. F. (198Sb). Is genetic epistemology possible? Duplicated
manuscript, Dept. of Philosophy, Colorado State University.
Knorr Cetina, K.D. (1981). The manufacture of knowledge: An essay on the
constructivist and contextual nature of science. Oxford: Pergamon
Press,
Chapter 1 (partlcularly 1.6 1.8). German
translation:
Die Fabrikation von Erkenntnis: Zur Anthropologie der Wissenschaft.
Frankfurt: Suhrkamp, 1984.
Knorr Cetina, K.D. (1987). Evolutionary epistemology and
sociology of
science. (This volume.)
Koehler, H. (1983). Erkenntnistheorie als biologische Anthropologie? In G.
Pfligersdorfer (Rg.),
Blickpunkte philosophischer Anthropologie.
Salzburg: Pustet, 43-63.
Koertge, N. (1975). Popper I s metaphysical research program for the human
sciences. Inquiry, 18, 437-462.
Koertge, N. (198~laining scientific discovery. PSA, 1, 14-28.
Kornblith, H. (Ed.). (1985). Naturalizing epistemology. Cambridge, MA: MIT
Press.
Lakomski,
G.
(1983).
Knowledge
and power:
Donald T. Campbell! s
evolutionary
epistemology
and
program
evaluation
as social
experimentation. Unpublished Ph.D. thesis, University of Illinois at
Urbana-Champaign.
Lamontagne,
C. (1967). Emerging: The sensorimotor conjecture. (This
volume. )
Lehrer, K. (1974). Observation and natural selection. In K. Lehrer,
Knowledge. Oxford: Clarendon Press, 173-174.
417
Leinfellner, W. (1983). Evolution of intelligence. In P. Weingartner & J.

Czermack (Eds.), Epistemology and
philosophy of science. Wien:
Holder- Pichler-Tempsky, 161 167.
Leinfellner J W. ('984). Evolutionary causality) theory of games, and

evolution of intelligence. In F.M. Wuketits (Ed.), Concepts
and
approaches in evolutionary epistemology. Dordrecht & Boston: ReJ.del,
233 276.
Leinfellner, W. (1987). Evolutionare Erkenntnistheorie und Spieltheorie.

In
R.
Riedl
&
F .M.
Wuketits
(Eds.),
Die
evolutionare
Erkenntnistheorie. Berlin: Parey.

S. (1982). 1. The ending of the epistemology of divine
knowledge.
II. The. naturalization of epistemology.
III. The humanization of
epistemology. IV.
The
historization
of
epistemology.
Book
manuscript.
Institute for
His,tory
&
Philosophy of Science,
Marulicev Trg 19/1, 41000 Zagreb, Jugoslavija.
Lelas, S. (1986). Epistemic implications of two biological concepts.
Philosophica, 37, 127-150.
Lem,
S.
(1981 ). Summa technologiae. (German translation of Polish
original.) Frankfurt: Suhrkamp.
Lenat, D.B. (1982). The nature of heuristics. Artificial Intelligence, 19,
189-249.
Levinson, P. (Ed.) (1982a). In pursuit of truth: Essays in honor of Karl
Popper's 80th birthday. New Jersey: Humanl ties Press & Sussex:
Harvester Press.
Levinson, P. (1982b). What technology can teach philosophy. In P. Levinson
(Ed.), In pursuit of truth: Essays in honor of Karl Popper's 80th
birthday. New Jersey: Humanltles Press & Sussex: Harvester Press.
Levins~(1982c). Evolutionary epistemology without limits. Knowledge,
3, 455-502.
---Levinson, P. (1 9S2d) MCLuhan' s contribution in an evolutionary context.
Educational Technology, 22, 39-46.
Levinson,
P. (1983), Evolution and rationality as checks on media
determinism. In S. Thomas (Ed.), Studies in mass communication and
technology. Norwood, NJ: Ablex, 231 237.
Levins~1984).
Technology as the cutting edge of cosmic evolution.
Paper presented at the 150th Annual Meeting
of
the
American
Association for the Advancement
of Science,
New York City, May
24-29,1984.
Levinson, P. (1986a). Information technologies as vehicles of evolution.
In C. Mitcham (Ed. L Research in Philosophy and
Technology, 8.
Greenwich, CT: JAr. Forthcomlng.
German editl.on edl.ted by A.
Huning, Vieweg, 1984.
Levinson, P. (1986b). Mind at large: Knowing in the technological age.
Forthcoming.
Lewis, H.A. (1976). The argument from evolution. Proceedings of the
Aristotelian Society, 53, 207-223.
Lewis, H.A., & Cooper, D. (1976). Symposium: The argument from evolution.
Proceedings of the Aristotelian Society, 53, 207-222.
Lewonbn, R.C. (1982a). Organl.sm and environment. In H.C.
Plotkin (Ed.),
Learning,
development
and
culture:
Essays
in
evolutionary
eplstemology. New York: Wiley, 151 170.
*Limoges, C. ( 1970). La selection naturelle. Paris: Presses Universi taire
de France.
Lipsey, M. W. (1972). Scientific values and scientific knowledge: A test
of an evolutionary model. JSAS Catagog of Selected Documents in
Psychology, 2.
Locker, A. (1981). Evolutionare Erkenntnistheorie - ein
Katarrh der
Vernunft. Paderborner Studien, 3(4), 101-106.
Lelas,
418
Loefgreo, L. (1981). Knowledge of evolution and evolution of

knowledge.
In E. Jantsch (Ed.). The evolutionary vision:
Toward a unifying
arad!
of
h sical.
b~olo l.cal
and SOCl.ocultural evolutl.on.
Bou er J CO:
Westv1e~
Press.
Lorenz, K. (1969). Innate bases of learning. In K.B. Pribram. On the

biology of learning. New York: Harcourt, Brace & World.
Lorenz. K. (1977). Behind the mirror: A search for a natural history of
human knowledge. New York
Lorenz.
K.
(1982).
contemporary
Learning,
e l.stemo 0
&
London: Harcourt Brace Jovanovlch.
Kant's doctrine
biology.
of
Reprinted
the a priori
in
H.C.
in
the
Plotkin
light
of
(Ed.).
develo ment,
and
culture:
Essays
in evolutionar
ew Yar: Wl ey.
j a so 10
. . vans
. an
Intervl.ew r , Konrad Lorenz: The man and his ideas. New York:
Harcourt Brace JovanOvl.Ch, 1975, lS1-217.
Lorenz, K. & Wuketits, F.M. (Eds.) (1983). Die Evolution des Denkens.
Munchen: Piper. 394 pp.
Lorenz, K. (1987). Das Apriori-Problem.
In R. Riedl & F.M. Wuketits
(Eds.), Die evolutionare Erkenntnistheorie. Berlin: Parey.
Law. R. (1984) The metaphysl.cal ll.Dll.ts of evolutionary epistemology. In
F.M.
Wuketits (Ed.),
Concepts and approaches in evolutionary
epistemology. Dordrecht & Boston: Rel.del, 209-230.
Law, R. &. Spaemann, R. (Hg.). (1983). Evolution und Erkenntnis-Tragweite
und Grenzen der evolutionaren Erkenntnistheorie in philosophischer
Absicht. In K.
Lorenz &
F.M.
Wuketits (Hg.). Die Evolution
des Denkens. Munchen: Piper, 330-331.
Lowenhard,
P.
(1981)."
Consciousness:
A biological view. Goteborg
Psychological Reports. 11 (10), 1-88.
Lowenhard, P. (1982). Knowledge, belief and human behaviour. GOteborg
Psychological Reports, 12( 11), 1-71.
Lowenhard,
P.
(1986a) . The mind-body problem: Some neurobiological
reflections. In P. Hoyningen-Huene &
F.M.
Wuketits
(Eds.),
Molecules and or anisms: Problems in reductionism and s stems theor
in 101ogy. Dordrec t & Boston: Re1de
10 preparat10n .
Lowenhard, P. (1986b). Evolutionary epistemology: A commentary. In G.
Radnitzky 5: W.W. Bartley, III (Eds.). Evolutionary epistemology,
theory of rationality and sociology of knowledge. LaSalle, IL: Open
Court.
Luhmann, N. (1975). Soziologische Aufklarung, 2: Aufsitze zur Theorie der
Gesellschaft. Opladen: Westdeutscher Verlag. (Esp. 199-200).
Luhmann, N. (1984). The differentiation of advances in knowledge: The
genesis of science. In N. Stehr & V.
Meja (Eds.)' Society and
knowledge. New Brunswick, NJ: Transaction Books.
Lutsky~5). Relativism and evolutionary objectivity in the
development of scientific knowledge. Dept. of Psychology, Carleton
College, Northfield, MN 55057. Duplicated manuscript presented to
the Fifth Annual Symposium of the Jean Piaget Society, June 13, 1975.
Lilt terfelds, W. (1982). Kants Kausalkategorie - ein stammesgeschichtliches
Aposteriori? Philosophia Naturalis, 19, 104-124.
Liltterfelds,
W.
(Ed. ).
( 1986).
Transzendentale
oder evolutionare
Erkenntnistheorie? Darmstadt: Wissenschaftl1che Buchgeselischaft.
Lycan, w. (1985a). Epl.stemic value. Synthese, 64, 137-164.
Lycan, W. (1985b). Conservatism andtEe'""Qita base. In N. Rescher (Ed.),
Reason and rationalit in natural science.
MacCormac, E. .
. Hume s embo 1e l.mpressions. Southern Journal of
PhilOSO~hY, 18(4), 447-462.
MacCormac, E. . (1983). Religious metaphors: Mediators between
biological
and cultural evolution that generate transcendent meaning. Zygon,
18(1), 45-65.
-MacCormac, E.R. (1985). Metaphor as a knowledge process. Chapter 5 in E.R.
MacCormac, A cognitive theory of metaphor. Cambridge, MA: MIT Press.
EVOLUTIONARY EPISTEMOLOGY BIBUOGRAPHY

Mach,
E.
(1976).
Knowledge and
error:
Sketches
inquiry. Translatl.on from the 5th edltl.on of

by P. McCormack & P. Foulkes.
419
on the psychology of
Erkenntnls und Irrtum,
Machamer, P. (Ed.) (1986). Naturalistic epistemology. Dordrecht & Boston:

Reidel.
Mackie,
J. L. (1983). Logic and knowledge: Selected papers. OXford:
Clarendon.
Marbach. E. (1983). Two directions of epistemology: Husserl and Piaget.

Revue Internationale de Philosophie. 36( 142/143}. 435-469.
Margoll.s, J. (1978). Persons and ml.Dds: The prospects of nonreductive
materialism. Dordrecht & Boston: Rel.del.
Maturana,
In
H.R. (1978). Biology of language: The epistemology
G.A.
of
reality.
Miller & E. Lenneberg (Ede.). Psychology and biology of

and thought. New York: Academic Press.
Maturana, H.R. (1980). Auto,poiesis: Reproduction, heredity and evolution.
In M.
Zeleny (Ed.),
Autopoiesis, dissipative structures, and
spontaneous social disorders. Boulder, CO: Westv1ew.
Maturana.!.. H.R. & Varela, F.J. (1980). Autopoiesie and cognition: The
languag~
Maxwell~al~~att'968j. thSc~!~~~icBO:!~~~d~~~:;I~d
the causal theory of

perception. In I. Lakatos & A. Musgrave (Eds.), Problems in the
philosophy of science. Amsterdam: North Holland, 148-160, 167-177.
Maxwell, G. ( 1970). Structural realism and the meaning of theoretical
terms. In M. Radner & S. Winokur (Eds.), Minnesota Studies in the
f~~~192~hY of Science.
Minneapolis: Univers1ty of M1nnesota Press,
Maxwell, G. (1971). Theories, perception, and structural realism. In R.
Colodny (Ed.), The nature and function
of
scientific theory.
Pittsburgh: Univers1ty of Pl.ttsburgh Press, 3-34.
*Maxwell, G. (1974). Corroboration without demarcation. In P.A. Schilpp
(Ed.), The philosophy of Karl Popper. LaSalle, IL: Open Court, "
292-321.
McGinn, C. (1983). The subjective view. Oxford: Clarendon.
McMullin, E. (1976). The fert1hty of theory and the unit for appraisal
in science. In R.S. Cohen et al. (Eds.), Essays in memory of Imre
Lakatos. Dordrecht & Boston: Reidel, 395-432.
Meyers-,-R-.-( 1985). Naturalizing epistemic terms. Paper presented
at
conference on "Naturalism and Rationality", SUNY, Buffalo, March
28-30, 1985.
Miller, A. I. (1987). The genesis of atomic physics and the biography of
ideas. (This volume.)
Millikan, R.G. (1984). Language, thought, and other biological categories:

New foundations for reall.sm. Cambr1dge, MA: MIT Press.
Mohr,
H.
(1965).
Erkenntn1stheoretische und ethische
Aspekte der
Naturwissenschaften. Mitteilungen des Verbandes Deutschen Biologen,
113, 525-535.
Mohr, H. (1967). Wissenschaft und menschliche Existenz: Vorlesungen
uber
Struktur und Bedeutung der W1ssenschaft. Frel.burg: Rombach. Trans!.
Lectures on structure and s1gnlilcance of science. Berlin: Springer,
Mohr,
1977 .
H. (1981). Biologische Erkenntnis: Ihre Entstehung und Bedeutung.

Stuttgart: Teubner.
Mohr, H. (1983a). 1st das 'Ethos der Wissenschaft' mit der evolutionaren
Erkenntnistheorie zu vereinbaren? In K. Lorenz & F .M. Wuketits
(Hg.). Die Evolution des Denkens. MUnchen: Piper.
Mohr, H. (1983b). Evolut1onare Erkenntnistheorie. Biologie in unserer Zeit,
613, 16-20.
Mohr, H. (1983c). Lasst sich Wissenschaft evolutionistisch begriinden? In
P. Weingartner & J. Czer-mack (Eds.), Epistemology
and
philosophy
of science. Wien: Holder-Pichler-TempsKy.
420
Mohr.
Mohr,
H.
(1983d).
Evolutionare Erkenntnistheorie - ein PHidoyer fur
Forschun s rogramm.
sltzungsberlchte Heldelberg:
Wl.ssenscha ten Mathemat. naturwiss. Kl.), 223-232.
AkademlB
ein
der
H. (1984). The ethics of science: Compatible with the concept of

evolutionary epistemology? In F.M. Wuketits
(Ed.),
Concepts
and
approaches in evolutionary epistemology. Dordrecht & Boston: Reldel,
185-206.
Mohr, H. (1987). Evolutionare Erkenntnistheorie, Ethos und Moral. In R.
Riedl & F .M. Wuketits CEds.), Die evolutionare Erkenntnistheorie.
Berlin: Parey.
Honad,
J.
(1970).- L8 hasard
et
18
n~cessite.
Paris: Seui!.
Muench, R. (1974). Evolutl.onare Strukturmale komplexer sozialer Systeme

am Beispiel des
Wissenschaftssystems.
Kenner
Zeitsehrift fur
80ziolo ie und 80zial sychologie, 26, 681-714.
Munevar, G.
1
. Rad1cal knowled e: A hiloso hieal in uir into the
nature and limits 0 SC.lence. n .lanapo 1S: ae ett.
Munz, P. ( 1984). Ph1losophy and the mirror of Rorty. Philosophy of Social
Science, 14, 195-238.
Munz, ~985). Our knowledge of the growth of knowledge: Popper or
Wittfenstein. London & Boston: Routledge & Kegan Paul.
Munz, P. ( orthcomin). Evolution and equidistance.
Musil, R. (1908/1980). Bel.trag zur Beurte11ung der Lehren Machs und Studien
zur Technik und Psychotechm.k. Rel.nbek: Rowobit.
Musil, R. (19a8/ 1982). On Mach s theories. Munchen: Philosophia.
Naess, A. (1972). The lural1st and ossibilist as ect of the scientific
enterprise. Lon n: A en & Unw1n.
Nagel, T. (1986). The view from nowhere. Oxford University Press, 78-82.
Nelson,
J .0.
(1983).
Do an1Dlais
propositionally
know?
Do they
proportionally believe?
Science Philosophical Quarterly. 20(2),
149-160.
Nelson, R.D. (1984). The evolutionary modeling of scientific change.
Dissertation proposal.
=:i:~~: Ri : R. (1 :82~in~~~. lOS~G. of (1~8~) .D~~d~~~~~t~o~~~to~~e~~id~i' economic
change. Cambridge, MA: Harvard Universl.ty ress Belknap.

Niiniluoto, 1. (1982). The evolution of knowledge. In Y. Haila, 1.
Tuominen & O. Vilhu (Eds.), Evolution: Different aspects, common
problems.
Proceedings
of the
1st Tutk.lJal11tto
Symposl.Um,
Tutk1jaliitto, Helsinki, 176-198.
*Northrop,
F.S.C.
(1931). Science and first principles. New York:
Macmillan.
Nozick, R. (1981). Philosophical explanations. Harvard University Press.
Odling-Smee, F. J. ( 1983). Multl.ple levels in evolution: An approach to
the nature-nurture issue via 'applied epistemology.' In G.C.L. Davey
(Ed.), Animal models of human behavior. New York: Wiley.
Odling-Smee, F.J. & plotkln, H.C. (1984). Evolution: Its levels and its
units. Behavioral and Brain Sciences. 7, 318-320.
Oeser, E. (1976) W1ssenschaft und 1nformation. 3 volumes. Vienna-Munich:
Oldenbourg.
Oeser.
E.
(1982).
Kants Beitrag zur progressiven Begriindung
der
komparativen
Wissensehaftstheorie.
Philosophia
Naturalis,
19.
201-250.
Oeser, E. (1984). The evolution of scientific method. In F.M.
Wuketits
(Ed. ),
Concepts and
approaches in
evolutionary epistemology,
Dordrecht & Boston: Rel.del, 149-183.
Oeser, E. (1986). Wissenschaftsevolution. Berlin: Parey (in preparation).
Oeser, E. (1987). Das Realltatsproblem. In R. Riedl & F.M. Wuketits
(Eds.), Die evolutionare Erkenntnistheorie. Berlin: Parey.
o 'Hear, A. (1984). On what makes an epl.stemology evolutionary. Proceedings
of the Aristotelian Society, 58, 193-217.
Olding, A. (1983). Bl.ology and knowledge. Theoria, 49, 1-22.

Oldroyd,
D.
&
Langham, I. (Eds.) (1983) The wider domain of evolutionary
Oppenh!~~~ghp: D~rd~~~!m~ B~~tO(19~). Re~~~~;.

hypothesis.
Concepts,
421
In
H.
Feigh,
M.
Scriven,
of
science as a
and G. Maxwell
working
(Eds.),
Theories, and the mind-body Problem, Minnesota Studies in

~~36~1.10SOPhY of sC.l.ence,
2, Ml.nneapo!l.s: Unl.versl.ty of Ml.nnesota,
Ott,
J.A., Wagner, G.P. & Wuketits, F.M. (Eds.) (1985). Evolution, Ordnung
und Erkenntnis. Berlin-Hamburg: Parey.
Pagninl., A. ( 1984J. Modelli evoluzionistici e crescita della conoscenza.

Intersezioni,4(1).
Paller, B.T. & Campbell, D.T. (1987). Reconciling Maxwell and Van Fraassen
through consideration of sense-organ evolution, the ostensive basis
of the term observe I,
and optimal justificatory practice in
science. In M.L. Maxwell and w.e. Savage (Eds.). Science. Mind and
Psychology J to appear as two issues of Synthese.
Pera, M. (1984). The justification of scientl.fl.C progress. In G. Anderson
(Ed.)' Rationality in science and politics. Dordrecht & Boston:
Reidel.
Perkinson, H.J. (1982). Education and learning from our mistakes. In P.
Levinson (Ed.), In pursuit of truth: Essays in honor of Karl Popper's
80th birthday. New Jersey: Humanl.tl.es Press & Sussex: Harvester
Press.
Peters., K.E. (1982). Religion and an evolutionary theory of knowledge.
Zygon, 17(4), 385-415.
Petrie, H.G. (1981). The dilemma of enquiry and learning. Chicago & London:
University of Chl.cago Press.
*Piaget, J. (1950). Introduction a I' epistemo!ogie genetique, 3 volumes.
Paris: Presses Unlversltalre de France.
Piaget, J. (1980) Adaptation and intelligence: Organic selection and
pbenocopy. Chicago: Unl.versl.ty of Chl.Cago Press.
Piaget~971).
Biology and knowledge. Chicago: University of Chicago
Press.
Piaget~ (1972). The principles of genetic epistemology. New York: Basic
Books.
Piattelli-Palmarini, M. (Ed.). (1980). Language and learning: The debate
between Jean Piaget & Noam Chomsky. Cambrl.dge: Harvard Onl.versl.ty
Press.
Piattelli-Palmarini,
M.
(1981).
Against
evolutionary arguments in
epistemology. Florence Center for the History
and the Philosophy
of Science, Florence, Italy. Paper delivered at the Symposium "Karl
Popper et la Science d'Aujourd'hui", Cerisy, July 1-11,1981, (1984
revision) .
Plotkin, H.C. (1981). The ~volution of closed and open programmes of
development. In D.R. Garrod & J.D. Feldman (Eds.), Development in
the nervous system. Cambridge: Cambridge University Press, 369 389.
Plotkin, H.C. (Ed.) (1982a). Learning, development and culture: Essays in
evolutionary epistemology. New York: Wl.ley.
Plotkin, H. C. ( 1982b). Evolutionary epistemology and evolutionary theory.
In H. C. Plotkin (Ed.), Learning,
development
and
culture:
Essays in evolutionar epistemolo . New York: Wl.ley, 3 13.
Plotki~lol~~r~al :nd 7:o~ia~V~~~!~~~~rfT~i!s!~~~;~~) and the synthesis of

Plotkin, H.C. (1987b). The eVOlutionary analogy in Skinner's writings. To
appear in S. Modgil & C. Modgil (Eds. L B.F. Skinner: Consensus and
Plotki~~ntH~~~rst19~~!~~to~~ !:~~~~i~~:~;' epistemological
approach to the
evolution of intelligence. To appear in the Journal of Human
Evolution (a special issue on the evolution of intel1l.genceJ, 16(1).
422
Plotkin,
H.C.
(1987d).
Game
theory,
hierarchy
theory and evolutionary
epistemology. Commentary for publication in R. Riedl & F .M. Wuketits

(Eds.), Die evolutionire Erkenntnistheorie. Berlin: Parey.
Plotkin, H.C. & Od11ng-Smee. F.J. (1979). Learning, change and evolution:
An enquiry into the teleonomy of learning. Advances in the Study of
Behaviour, 10, 1-41.
Odling-Smee, F.J. (1981). A multiple-level model of

evolution and its implications for sociobiology. Behavioral and
Brain Sciences, 4, 225-268.
Plotkin, H.C. & Odling-Smee, F.J. (1982). Learning in the context of a
hierarchy of knowledge gaining processes. In H.C. Plotkin (Ed.),
Learnin.
development
and
culture:
Esss s
in
evolutionary
Plotki~ &
POlJ,ack~J.s~~mot1ek3) ~ewF;~~k~e~~o~~io:ic~l7;~escriPtion
to socia-historical
description:
The changing metascientific
discourse. Fundamenta
Scientiae, 4(1), 1-27.
*Poppe~
(1959).
The logic of scientific discovery. London:
Hutchinson.
*Popper, K.R. (1963). Conjectures and refutations. London: Routledge &
Kegan Paul.
*Popper, K.R. (1972). Objective knowledge: An evolutionary approach.
Oxford: Clarendon Press.
Popper, K.R. (1977). Die beiden Grundprobleme der Erkenntnistheorie.
Tiibingen: J.C.B. Mohr (Pau! S1ebeck).
Popper,
K.R. (1978). Natural selection and the emergence of mind.
Dialectica, 32(3/4), 339-355.
Popper, K.R. (1981). The rationality of scientific revolutions. In I.
Hacking (Ed.)' Scientific revolutions. OXford: OXford University
Press.
Popper, K.R. (1986). Evolutionary epistemology. In G. Radnitzky & W.W.
Bartley III (Eds.), Evolutionary e istemolo , theor of rationality,
7 K~~~ (~~~~)~g~J.~f e~~~:!~~I:the~r:ti:~he ~osi~~ono~~;.
nd
evolutionaren
Erkenntnistheorie.
In R.
Riedl & F.M. Wuketits (Eds.), Die
evolutionare Erkenntnistheorie. Berlin: Parey.
Pribram, K.R. (1984). EvolutJ.on and the mind/body problem. Paper presented
at the
annual meeting
of the American
Association for the
Advancement of Science, May 27, 1984, New York, NY.
Prigogine, I. (1987). The meaning of entropy. (This volume.)
*Pringle, W.S. (1951). On the parallel between learning and evolution.
Behaviour, 3, 175-215.
Putnam~1979). The place of facts in a world of values. In D. Huff &
O. Prewett (Eds.), The nature of the physical universe. New York:
Wiley.
Putnam, H.W. (1983). Why reason can't be naturalized. Synthese. 52, 3-23.
Reprinted in H.W. Putnam, Realism and reason. CambrJ.dge: Cambridge
University Press, 229-247.
*Quine, W.V. (1969). Ontological relativity. New York: Columbia University
Press, 126-128.
Quine, W.V. (1973). The roots of reference. LaSalle, IL: Open Court.
Quine, w.v. (1975). The nature of natural knowledge. In S. Guttenplan
(Ed.), Mind and language. OXford: Clarendon.
Radnitzky, G. (1982). Popper as a turning point in the phi1oso~hy of
science: Beyond foundationalism and relativism. In P. Levinson (Ed.),
In pursuit of truth: Essays in honour of Karl Popper's 80th ~irthday.
AtlantJ.c HJ.ghlands, NJ: HumanJ.tl.es Press & BrJ.ghton: - arvester
Press.
Popper
423
Radnitzky J G. (1983). Die Evolution der Erkenntnisfahigkeit, de!=! Wissens

und der
Institutionen.
In
R.
Riedl und F. Kreuzer (Hg.).
Evolution und Menschenbild. Hamburg: Hoffmann und Campe, 82-120.
Radnitzky, G. (1983). The SClence of man: Biological, mental, and cultural
evolution. In V. Cappelletti. B. Luisel!i, G. Radnitzky, & E. Urbani
(Eds.)' Saggi di storia del pensiero scientifico dedicati a Valerio
Tonini. Roma: SOC leta Edl.torlale Jouvence. 369 401.
Radnit~ G.
(1987).
Empirismus-RationalismuB
und
evolutionare
Erkenntnistheorie.
In
R.
Riedl &
F.W. Wuketits (Eds), Die
evolutionare Erkenntnistheorie. Berlin: Parey.
Radnitzky. G. & Bartley. W.W. III. (1986). Evolutionar~ e istemolo gy ,
theory of rationality, and sociology of knowledge. La al e, It: open
Court.
Rechenberg, I. (1973). Evolutionsstrategie: Optimierung technischer Systeme
nach Prinzipien der bJ.ologJ.schen EvolutJ.on. Stuttgart: FrommannHoizhoog.
Rensch,
B.
(1961).
Die
Evolutionsgesetze
der
Organismen
in
naturphilosophischer Sicht. Philosophia Naturalis, 6, 288-362.
Rensch. B. (1968) . Biophilosophie auf erkenntnJ.stheoretischer Grundlage.
Fischer: Stuttgart.
Rescher J N. (1977). Methodological pragmatism. Oxford: Blackwell.
Rescher, N. (1982). Response to Rekes. In R. Almeder (Ed.), Praxis and
reason. Washington, DC: University Press of America.
Richaras;---R. J.
(1977).
The natural selection model of conceptual
evolution. PhilOSOJhY of Science, 44, 494-501.
Richards,
R.J.
(1981. Natural selection and other models in the
historiography of science. In M.B. Brewer & B.E. Collins (Eds.),
Scientific inquiry
and the social
sciences.
San Francisco:
Jossey Bass. 37-76.
Richards. R.C. (1987). Darwin and the emergence of evolutionary theories
RiCher~~n~indp~j~ b&haB~~d: c~~caf~~8~)~ve~:~~~a~f ;~~~~~~O~re:~d
culture.
BioScience. 34(7). 430-434.
Riedl,
R.
(i 978).
Uber
die
Biologie des
Ursachen-DBnkens: Ein
evolutionistiscber, systemtheoretischer Versucb. Mannbeimer Forum,
9-70.
Riedl. R. (1980). Biolo ie der Erkenntnis: Die stammes eschichtlichen
Grundlagen der Brnun t.
WJ.t
t e co a oratJ.on 0
. aspar.
Berlin: Parey.
Riedl,
R.
(1981). Die Folgen des Ursachendenkens. In P. Watzlawick (Ed.),
Die erfundene Wirklichkeit. Munchen: Piper, 67-90.

Riedl, R. (1982). EvoiutJ.on und Erkenntnis. Munchen: Piper.
Riedl, R. (1983). EvolutJ.on und Erkenntnis. Biologische Rundschau, 21,
143-154.
Riedl.
R.
(1984a).
Evolution and evolutionary
knowledge: On the
correspondence between cognitive order and nature.
In
F .M.
Wuketits (Ed.), Concepts and approaches in evolutionary e istemology.
Dordrecht & Boston: ReJ. e , 5- .
Riedl. R. (1984b). Biolo
of knowled e: The evolutionar basis of reason.
(With the col a oratJ.on 0 R. Kaspar; translate
rom t e r German
Edition by Paul Foulkes.) New York: Wiley.
Riedl, R. (1985). Die Spaltung des Weltbildes. Berlin: Parey.
Riedl, R. (1987). AdaptJ.erungsmangel der menschlichen Vernunft. In R.
Riedl & F.M. Wuketits (Eds. L Die evolutionare Erkenntnistheorie.
Berlin: Parey.
Riedl. R. & Kreuzer. F. (Eds.) (1983). Evolution und Menschenbild.
Hamburg: Hoffmann und Campe.
Rorty, R. (1979). The unnaturalness of epistemology. In D.F. Gustafson &
B.L. Tapscott (Eds.)' Body. mind, and method. Dordrecht & Boston:
Reidel, 77-92.
Rorty. R. (1980). Philosophy and the mirror of nature. Oxford: Blackwell.
424
Rosenberg, A. (1981). Variation and artificial selection in the
of science.
Duplicated for ERISS
Conference,
June 1981 J
Philosophy, Syracuse University, 10 pp.

Rosenberg, A. (1987). Evolutionary biology and intentional
The uneasy analo," Behaviorism, 14.
Rosenberg,
J.F.
evolution
Dept. of
psychology:
(1981). One world and our knowledge of it: The eroblem of
ost-Kantl.an pers ect1ve. Dordrecht & Boston: Re1del.

Roth,
18ge on natura 1se epistemology and natural science.
PhilOSOPh, of Science J 50 J 482-493.
.
Rozin, P. (1976. The evolution of intelligence and access to the cognitive
unconscious. In J .M. Sprague & A.N. Epstein
(Eds.) f
pro~ress
in psychobiolOgy and physiological psychology, 6. New York: Aca eml.C
Press.
Rubinstein,
R.A., Laughlin, C.D. & McManus, J. (1984). Science as
cognitive process:
Studies toward an
empirical
hiloso h
of
BCl.ence.
l. a e p l.a: nl.versl.ty 0
ennsy vanl.a ress.
Rumbau~ (1984).
Intelligence of animals and humans: From genes to
genius for reasons of control. Presented at
the
Conference
on
Integrating Scientific Disciplines, Georgia State University, May
3-5, 1984.
Ruse, H. (1977). Karl Popper's philosophy of biology. Philosophy of
Science, 44, 638-61.
Ruse, ~983a). Darwin and philosophy today. In D. Oldroyd and I.
Langham,
(Eds.),
The wider
domain of
evolutionary thought.
Dordrecht & Boston: Rel.del. 133-158.
Ruse, M. (1983b). Sociobiology and philosophy: Is evolutionary epistemology
a viable option? Unpublished manuscript, August, 1983, 78 pp. (Dept.
of History aod Philosophy, University of Guelph, Ontario, Canada N1G
realism in
Ruse,
2W1.
M.
(1985). Evolutionary epistemology: Can sociobiology help?
Synthese.
Ruse, ~85/86). Evolutionary epistemology: Let's be Darwinian not
Spencerian. Newsletter: For those interested in the philosophy of
Karl Popper, 2(3/4), 6-8.
Ruse, M. (1986). Takin Darwin seriousl: A naturalistic a roach to
philosophy.
Ox or :
ac e .
sp.
c
vo utl.onary
Epl.stemology", and ch 5: "Darwinian epistemology").
Russell, C. & Russell, W.M.S. (1962). Raw materials for a definition of
mind. In J.M. Scher (Ed.), Theories of the mind. New York: Free
Press, 535-571.
Russell, J. (1978). The acquisition of knowledge. New York: St. Martin's
Press.
Schafer, w. (1978). Normative Finalisierung. Eine Perspektive. In G. Behrne
et aI., Starnberger Studien, 1: Die gesellschaftliche Orientierung
des wissenscbaftll.cben Fortscbrl. tts. Frankfurt: SUhrkamp, 377-415.
Translated as Normatl.ve
hnall.zation,
in
W.
SchAfer (Ed.),
Finalization in science: The social orientation of
scientific
progress. Dordrecbt & Boston: Re~del, 1983, 207-231.
Schell, J. & De Wae!e, D. (1987). The exchange of genetic informatio,n
between organisms of distinct origin. (This volume.)
Seitelberger, F. (1983). Die evolution zur Erkenntnis: Leistungspotenzen
und Leistungsprodukte des menschlichen Gehirns.
In P.
W~ingartner &
~~ld;;:~~~~~e~~~:~~~kY, El~4:i84~Ogy and philosophy of science. Wien:

Seitelberger, F. (1984l' Neurobiological aspects of intelligence. In F .M.
Wuketits,
(Ed.,
Concepts
and
approaches
in
evolutionary
epistemology. Dordrecht & Boston: Rel.del, 123-147.
Seitelberger, F. (1985). Gehirntatigkeit und Bewuptsein. Proc. of the 9th
International Wittgenstein Symposium. Vienna: Holder-Pl.Chler-Tempsky,
112-114.
425
*Sellars, W.S. (1963). Science, perception and reality. London: Routledge &
Kegan Paul, 90.

*Shimony,
A.
(1970).
Scientific inference.
In R. Co!odny (Ed.),
Pittsburgh Studies in the Philosophy of Science, 4. Pittsburgh:
University of Pl.ttsburgh Press, 79 172.
*Shimony, A. (1971). Perception from an evolutionary point
of view.
Journal of Philosophy, 68, 571-583.
Shimony, A.
(1981).
Integral epistemology. In M.B. Brewer & B.E. Collins
(Eds.),
Scientific inquiry
Jossey-Bass, 98-123.
Shrader,
and the social
sciences. San Francisco:
D.W., Jr. (1980), The evolutionary development of science. Review
Siegel~f ~~taf7~~~r' ~~~~lfi~:~i~;~'
discovery and the naturalizing of

epistemology. Philosophy of Science, 47, 297-321.
Simmel, G. (1982). On a relat~onship between the theory of selection and
epistemology. In H.C. Plotkin (Ed.), Learning,
development,
and
culture: Essa s in evolutionary epistemology. New York: Wiley, 63-71.
Simon,
.A.
57. at10nal cho1ce an t e structure of the environment.
In H.A. Simon, Models of Man. New York: Wiley, 261-273.
"'Simon, H.A.
(1969). The sc~ences of the artificial. Cambridge, MA: MIT
Press, 95.
Simonton, D.K. (1984). Genius, creativity, and leadership: Historiometric
inquiries. Cambridge, MA: Harvard Un~vers~ty Press.
Skages~1978). Taking evolution seriously: Critical comments on D.T.
Campbell's evolutionary epistemology. Monist, 61(4), 611-621.
Skagestad, P. (1981a). Hypothetical realism~.B. Brewer & B.E. Collins
(Eds.), Scientific inquiry and the social sciences. San Francisco:
Jossey-Bass, 77 97.
Skagestad, P. (1981b). The road of inguiry: Charles Peirce's pragmatic
realism. New York: Coiumb~a un~vers~ty Press.
Skagestad, P.
(1986). Peirce's conception of truth: A framework
for
naturalistic epistemology. In A. Shimony,
D. Nails, & R.S. Cohen
(Eds.), Naturalistic
e istemolo y:
A symposium of two decades.
Dordrecht J Ho Ian : Re~ e .
Smith,
C.U.M.
(1983).
Herbert
Spencer's
epigenetic epistemology.
Studies in History and Philosophy of Science, 14(1), 1-22.
Smith, R.A. (1979). Analogical l.ncorpor-at~on: Themes and variations in
evolutionary
eputemology.
Ph.D.,
Ph~losophy,
Un1vers~ty
of
Cal1fornia,
Santa Cruz.
Ann Arbor, MI: University Microfilms
International.
Sober, E. (1976). Psychologism. Journal for- the Theory of Social Behaviour-,
8(2), 165-191.
Sober, E. (1981a). The evolution of rationality. Synthese, 46, 95-120.
Sober,
E.
(t961b).
Revisability,
a prior~ truth, and evolution.
Austr-alasian Journal of Philosophy, 59, 68-85.
Soberon, E. ( 1980). Gez~chtspunten veor een bielogistische epistemologie:
Campbell,
D.T. en Piaget, J. Unpublished "licentiate" thesis.
Univer-sity of Ghent.
Somenzi,
V.
(1981).
Scientific discovery from the
viewpoint
of
evolutionary epistemology. In M.D. Gr-mek, R.S. Cohen, & G. Cimino
(Eds.),
On
scientific discovery:
The
Erice
lectures 1977.
Dordrecht & Boston: Re~dei, 167-177.
Sosa, E. (1983). Nature unmirr-or-ed, epistemology naturalised. Synthese, 55,
49-72.
--Stanzione,
M. (1981).
Introduzione. In D. T. Campbell, Epistemologia
evoluzionistica.
Rome:
Armando Ar-mando,
7-62.
(Rough Enghsh
translatl.On available fr-om D.T. Campbell.)
Stanzione,
M.
(1984).
Epistemologia
evoluzionistica: Confronti
e
critiche. In B. Continenza et a1., Evoluzione e modelli: II concetto
di adattamento nelle teorie dei s~stem~ b~ologJ.cJ., cuIturah e
art~ficiali. Rome: Editori R~unitl., 193 300.
426
Steele,
E.J.
(1981).
Somatic selection and adaptive
evolution:
On the
heritance of acquired characters. Chl.cago: Um.versl.ty of Cambn.dge

Press. 2nd ed . 92-96.
Stegmiiller,
W. (1985). Thesen zur 'evolutionaren Erkenntnistheorie,'
Information Philosophie, 13(3), 26-32.
Stegmiiller,
W. (1984). Evolutionare Erkenntnistheorie, RealislIlus und
Wissenschaftstheorie. In R. Spaemann, P. Koslowski, & R. Low (Hrsg.),
Evolutionstheorie und
menschliches
Selbstverstandnis. Weinheim:
Verlag Cheml.e, 5-34.
*Stemmer, N. (1971). Three problems in induction. Synthese, 23(2/3),
287-308.
Stemmer,
N.
(1975a).
---
relative
notion
of
natural
generalization.
Philosoph of Science I 42, 46-48.

Stemmer, N. (97Sb). The Goodman paradox. Zeitschrift fur allgemeine
Wissenschaftstheorie, 6, 340-354.
Stemmer, N. (1977). A solution to the Hempel paradox. Teorema, 7, 119-128.
Stemmer, N. (1978a). The reliability of inductive inferences and our innate
capacities. Zeitschrift fur allgemeine Wissenschaftstheorie, 9(1),
93-105.
Stemmer,
N. (1978b). A partial solution to the Goodman
paradox.
Philosophical Studies, 34, 177-185.
Stemmer, N. (1919). Pro]ectlble predicates. Synthese, 41, 375-395.
Stemmer,
N.
(1981a).
Generalization
classes
as alternatives for
similarities and some other concepts. Erkenntnis, 16. 73-102.
Stemmer,
N.
(1981b).
The
objective
confl.rmation
of hypotheses.
Canadian Journal of Philosophy, 11, 395-404.
Stemme~39_~53. (1982). A solutl.on to the lottery paradox. Synthese. 51.
Stemmer.
N.
(1983a).
Blackwell.
The philosophy and psychology of knowledge. Oxford:
Stemmer. N. (1983b). The roots of knowled~e. Oxford: Blackwell.

Stent, G. (1972). Prematurl.ty and unl.qeness in scientific discovery.
Scientific American, 227, 84-93.
Stent, G. (1975). Ll.rnl.ts to the scientific understanding of man. Science,
187, 1052-1057.
--Stent. G. (1977). Explicit and implicit semantic content of the genetic
information: 1. Genetic information. In R. Butts & J. Hintikka
(Eds.), Foundational problems in the special sciences. Dordrecht:
Reidel, 131-149.
Stent, G. (1980). Thinking about seeing. A new approach to visual
perception. The Sciences, 20, 6-11.
Stent. G. (1981a). Cerebral hermeneutics. Journal of Social and Biological
Structures, 4, 107-124.
Stent,
G. (1981b). Programmatic phenomena, hermeneutics and complex
networks. In C.M. Steinberg & I. Lefkovits (Eds.), The immune system:
Festschrift in honor of Niels Kaj Jerne on the occaSSl.on of h1s 70th
birthday, 1. Basel: Karger, 6 13.
Stent. ~986). Introduction and overview. In M. Delbruck. Mind from
:~!~kW:l1 A~Cl.:~~~{l.C o~ubi:~!~~~~~~rt
is:istemology
Palo
Alto,
CA:
Stent.
G. (in press). Uniqueness in creation in art and science.

Interdisciplinary Science Reviews.
Stich, S.P. (1985). Could man be an J.rrational animal? Some notes on the
~:iS~~~~~~!iho(E~~)~on:!~~~~1~Pi:~~mOl~~~~13~~br~~~~ ~n~~ Mi~
Press, 249~267.
Stroud, B. (1980). The si nificance of naturalised epistemolo ., Midwest
Studies in P l. asap y,
.
epr1nted l.n
. Korn 1t
(Ed.),
Naturalizing epistemology. Cambridge, MA: MIT Press, 71-89.
427
Stroud, B. (1981). Evolution and tbe necessities of thought. In L.W.

Sumner, J.G. Slater & F. Wilson CEds.) Pragmatism and purpose: Essays
~36~i4j~d to Thomas A.
Goudge. Toronto: Un1VerSl.ty of Toronto Press,
Stroud,
B.
(1984).
Naturalized
epistemology.
The
significance
of
Tennani~il~~Of~ij83!) ~k~~t!~i:~~eN~; !~~~~t~~!~~~ ~~~;~~:~I~~~e;~~o~~:~2!::

32-48.
Tennant,
---
N.
(1983b). Evolutionary epistemology. In P. Weingartner & J.

Czermack CEds.), EPistemO!OH and philosophy of science. Wien:
Holder-Pichler-Tempsky, 168-17 .
Tennessen, H. (1973). Knowledge versus survival. Inquiry, 16, 407-414.
Thagard, P. (1980). Against evolutionary epistemology. PSA, 1, 187-196.

Thorn, R. (1979). La g~nese de l'espace representatIrselon Piaget. In
i'
i~;~~~~t :~ur l:r:~~~nc;h:~rte:o:e ;~~~i~t~o3a~Og. Paris: Centre

Thorn, R. (1987). The epistemology of evolutionary theories. (This volume.)
Thompson, M. (1981). Epistemic priority, analytic truth, and naturalised
epistemology. American Philosophical Quarterl , 18, 1-12.
Tibbetts, P.E. (19
e wel.g te co erence t eory of

rationality: A
critique and an alternative. Philosophy of the Social Sciences, 10,
259-272.
Titze, H. (1983). Evolutionare und/oder transzendentale Erkenntnistheorie.
In
P.
Weingartner
&
J.
Czermak
(Hg. ),
Erkenntnisund
Wissenschaftstheorie. Wien: Holder-Pichler-Tempsky, 198-203.
Topitsch, E. ( 1979). Erkenntnis und Illusion. Grundstrukturen unserer
Weltauffassung. Hamburg: Hoffman & Campe.
*Toulml.o, S. ( 1961). Foresight and understanding: An inquiry into the aims
of science. Blooml.ngton, IN: Indiana Unl.Vers1ty Press.
*Toulm1n, S. (1967). The evolutionary development of natural science.
American Scientist, 55, 456-471.
*Toulml.n, S. (1968). Neuroscience and human understanding. In F. Schmitt
(Ed.), The neurosciences. New York: Rockefeller University Press.
Toulmin, S. (1972). Human understanding: The evolution of collective
understanding. Prl.Deeton, NJ: Prl.nceton Unl.Vers1ty Press.
Toulmin, S. (1981). Evolution, adaptation, and human understanding. In
M.B. Brewer & B.E. Collins (Eds.)' Scientific inquiry and the
social sciences. San Francisco: Jossey-Bass, 18-36.
Toulmin, S. (1985). Conceptual evolution in science. In R.J. Cohen &
M. W. Wartofsky (Eds.), A portrait of 25 years: Boston Colloquium in
the philosophy of science, 1960-1985. Dordrecht & Boston: Rel.dei.
Ulrich,
E.
&
von
Wel.Zacker,
C.
(1984).
Errors
and Fitness:
Fehlerfreundlichkeit. In K. Kornwachs (Ed.), Offenheit, Zeitlichkeit,
Komplexitat. Frankfurt & New York: Campus, 167-201.
Unsold, A. (1981). Evolution kosmischer, biologischer und
geistiger
Strukturen. Stuttgart: Wl.ssenschafthche verlagsgeseils(:h8ft.
Van Bendegem, J.P. (1987). Fermat's last theorem. (This volume.)
Vandamme, F. (1987). Language and evolutionary or dynamic epistemology.
(This vOlume.)
Van Fraassen, B. (1984). On the reality of mathematical entities. In M.
Piattelli-Palmarini (Ed.), Livelli di realta. Milan: Feltrinelli,
90-110.
Van Fraassen, B. (1985). Empiricism in the philosophy of science. In P.M.

Churchland & C.A. Hooker (Eds.), Images
of
science. Chicago:
University of Chicago Press, 245-305 (esp. 282-283 and 300).
Van Parijs. P. (1981). Evolutionary explanation in the social sciences: An
emerging paradigm. Totowa, NJ: Rowman and Ll.ttlefl.eld.
Van Parl.Js, P. (1987). The evolutionary explanation of beliefs. (This
volume. )
Varela, F.J. (1979). Principles of biological autonomy. New York: Elsevier.
Vendler. Z. (1972). Res Cogm.tas. London: Cornell Un~versity Press.
428
Vollmer J
G. (1975). Evolutionare Erkenntnistheorie. Stuttgart: Hirzel (3rd
edition, 1983).
Vollmer, G. (1983). On supposed circularities in an empirically oriented
epistemology.
In
International
Cultural
Foundation
(Ed.),
Absolute values and the creation of the new world,

2. New York:
Internatl.onal Cultural Faundatl.on Press, 783-834. To be reprinted in
G. Radnitzky & W. W. Bartley, III (Eds.), Evolutionary epistemology.
Theory of rationality and the sociology of knowledge. La Salle. IL:
Open Court I 1986.
Vollmer I
by
G. (1984a). Mesocosm and objective knowledge: On problems solved

evolutionary epistemology. In F.M. Wuketits (Ed.), Concepts and
6~p~2~~hes in evolutionary epistemology.
Dordrecht & Boston: Re1del,
Vollmer, G. (1984b). New problems for an old brain - synergetics,

cognition and evolutionary epistemology. In E. Frehland (Ed.),
Syner etics from microsco ic to macrosco ic
order. Berlin:
pr1nger, 5
5.
Vollmer.
G. (1984c). Evolutionare Erkenntnistheorie. Darstellung des
Ansatzes. Information Philosophie, 12(5), 4-23.
Vollmer, G. (1985a). Was k5nnen wir wissen? I: Die Natur der Erkenntnis
Beitra e zur Evolut10naren Erkenntn1stheor1e. Stuttgart: H1rze1.
Vollmer, G.
1 5 . Zur Evolut1onaren Erkenntnistheorie. Ein
kurze
Antwort auf die Kritik. Information Philosophie, 13(5), 40-45.
Vollmer, G. (1985c). Zwischen den Fronten
Warum. sich die Evolutionare
Erkenntnistheorie gegen ihre Anhanger unci gegen ihre Kritiker webren
mu~. C
19, 63-69.
Vollmer, G.
a . Wissenschaft mit Steinzeitgehirnen? Mannheimer Forum.
Vollmer, G. (1986b). Was konnen wir wissen? II: Die Erkenntn1S
der
Natur
Beitrage zur modernen Naturphl.loBophie. Stuttgart: H1rzei.
305 pp.
Vollmer, G. (1987a). What evolutionary epistemology is not. (This volume.)
Vollmer, G. (1987b). Evolutionare Erkenntnistheorie und Wissenschaft. In
R. Riedl & F .M. Wuketits (Eds.), Die evolutionare Erkenntnistheorie.
Berlin: Parey.
von Ditfurth, M. (1987). Evolution, Transzendenz und Selbst-transzendenz.
In
R.
Riedl
&
F.M.
Wuketits
(Eds.),
Die
evolutionare
Erkenntnistheorie. Berlin; Parey.
von Glasersfeld, E. ( 1981a). The concepts of adaptation
and viability
in a radical constructivist theory of knowledge. In 1. Siegel, R.
Golinkoff, & D. Brodzinsky (Eds.), New directions in Piagetian
~~e9~: and their applications in educat1on. ttl.flsdale, NJ: Eribaum,
C198tlt,
von Glasersfeld, E. (1981b). An introduction to radical constructivism.

English translation of a chapter in P. Watzlawick (Ed.), Die
erfundene Wirklichkeit. Munchen: Piper.
von Glaser'sfeld, E. ( 1982). An interpretation of Pia,et' s constructivism.
Revue Internationale de PhilOsofhie, 36(142/143), 612-635.
von Schl.lcher. F. & Tennant, N.1984). Philosophy, evolution & human
nature. London: Routledge & Kegan Paul.
von Weizsacker, E.U. (1983). Contagious knowledge: Contagious success
as
a quality criter'ion for' disciplinary and interdiSCiplinary science.
58th Nobel Symposium. August 15-19, 1983.
von Weizsacker, C.F. (1982). Die Riickseite des Spiegels, gespiegelt.
Der
Garten des Menschlichen:
Beitrage
zur
eschichtlichen
Anthropologl.e. unc en: l.SC er.
von Weizsacker, E. & C. (1984). Fehlerfreundlichkeit. In K. Kornwachs
(Ed.), Offenheit, Zeitlichkeit, Komplexitat. Frankfurt & New York:
Campus. 167-201.
WachtershaUser,
G.
(1986).
On
light and life.
429
In G.
Radnitzky &
W.W.
Bartley,
III
CEds. L
Evolutionary
epistemology,
theory
of
rationality, and the sociology of knowled e. LaSalle, IL: Open Court.
Wagner, G.P.
.
e ogl.ca baS1S 0 eva utionary epistemology. In
F .M. Wuketits (Ed.), Concepts and
a;roaches in evolutionary
Wagner
~Pi~~;~Ol(~81 ~~rd~~~t ~i:os~~~~s:~~~e\lrun!l!~!~
der evolutionaren
Erkenntnistheorie. In K. Lorenz & F.M. Wuketits (Eds.)' Die Evolution
der Denkens. MUnchen; Piper, 199-214.
Wagner,
G.P.
(1987).
Die
logische
Begriindung
der evolutioniren
Erkenntnistheorie.
In R.
Riedl &
F.M. Wuketits CEds.), Die
evolutionare Erkenntnistbeorie. Berlin: Parey.
*Wartofsky, M.W. (1967). Metaphysl.cs as heuristic for science. In R. S.
Cohen & M. W. Wartofsky (Eds.) In memory of Norwood Haneson.
Dordrecht: Reidel, 164-170.
Wartofsky, M. W. (1971). From praxis to logos: Genetic epistemology and
physics.
In
T.
Mischel
(Ed.),
COjnitive
development
and
epistemology. New York & London: Academic ress, 129 147.
Wartofsky,
M. W.
(1979).
Models:
Representation
and
scientific
understanding. Dordrecht & Boston: Rel.de1.
Wartofsky, M.W. (1982). Piaget's genetic epistemology and the
Marxist
theory
of knowledge.
Revue
Internationale
de Philosophie,
36(142/3), 470-507.
*Watanabe, S. (1965). Une explication mathematique du elassement d I objets.
In S. Dockx & P. Bernays (Eds.), Information and prediction in
science. New York: Academic Press. 39-45.
Watt. w.c:--( 1979). Against evolution. Linguistics and Philosophy, 3.
121-137.
Weeder. P. & Kester, D. (1982). Variation and selection,
the construction
of an industrial product: The case of Tenax. Dynamics of Science, 2,
282-310.
Weingartner. P. & Czermak, J. (Eds.). (1983). Epistemology and
Philosophy
of Science. Erkenntnis - und Wissenschaftstheorl.e. Wl.en: HolderPl.chler-Tempsky. Sectl.on 1. 5 (Evolutlonary Epl.Stemology).
wenhar~~te~~atl~~:i)' eu~;~~~~iO~~~d:~i~~em(i~~)~ D!:~~~~~~t :a~::r:~d t~~""

I
creation of the new world, 2. New York: Internatl.onal cultural"

Foundatlon Press. 901-908.
Werner.
H.J.
(1978).
Rezension
von
G.
Vollmer:
Evolutionare
Erkenntnistheorie. Philosophisches Jahrbuch, 85. 201-204.
*Whitrow,
G.J.
(1956).
Why physl.cal space
has three dimensions.
British Journal for the Philoso h of Science, 6, 13-31.
Wilden, A.
1
. System and structure: Essays in communication and
exchange. London: TavlStock.
*Wimsatt, W. C. (1972a). Teleology and the logical structure of
function
statements. Studies in History and Philoso h of Science. 3(1), 1-80.
Wimsatt. W.C. (1 7
. T e maehl.oe l.n the g ost: T e
SCl.ences of the
artificial. Unpublished manuscript, University of Chicago Press, 16
pp.
Wimsatt, W.C. (1974). Complexity and organization. In K.F. Schaffner &

R.S. Cohen (Eds.). PSA 1972. Dordrecht: Reidel, 67-86.
Wimsatt,
W.C. (1976). Reductl.onism. levels of organization and the
mind-body problem. In G.G. Globus, G. Maxwell, & 1. Savodnik (Eds.),
Consciousness and the brain: A scientific and philosophical inquiry.
New York: Plenum, 199-267.
Wimsatt, w.e. (1979). Reduction and reductionism. In P.D. Asquith & H.E.
Kyburg (Eds. ), Current research in the hiloso h of science. East
Lansing. MI: Philosop y 0
Clenee BSOCl.atl.on, 5Wimsatt,
W.C.
(1980a).
Randomness
and
perceived-randomness
in
evolutionary biology. Synthese, 43, 287-329.
430
Wimsatt,
w.e. (1980b). Reductionistic research strategies and their
biases in the units of selection controversy. In T. Nickles (Ed.),
Scientific discovery: Case studies. Boston & Dordrecht: Reidel,
213-259.
Wimsatt, w.e. (1981a). Robustness, reliability, and overdetermination. In

M.B. Brewer & B.E. Collins (Eds.), Scientific inquiry and the social
sciences. San Francisco: Jossey-Bass J 124 163.
.
Wimsat~ (1981b).
Units of selection and the structure
of the
multi-level genome. In P.D. Asquith & R.N. Giere CEde.), PSA 1980, 2,
122-183.
--Wimsatt, w.e. (1985a). Developmental constraints, generative entrenchment,
and
the
innate-acquired
distinction.
Unpublished manuscript,
University of Chicago Press, 28 pp.
Wimsatt, W.C. (1985b). False models as means to truer theories.
Talk
given at the Spring Systematics Symposium on Neutral Models in
Biology, 25 pp.
Wimsatt,
w.e. (1985c). Von Baer's law of development, generative
entrenchment,
and scientific change.
Philosophy
Dept.,
The
~i~~~~!~Y o~f scr~~~:~~'
Chicago,
IL 60637
(under revision for
Wimsatt, W.C. f1986a). Heuristics and the study of human behavior. In D.W.
Fiske
&
R.A.
Shweder (Eds.), Metatheory in social science:
Pluralisms and subjectivities. Chicago: Unl.vers1ty of Ch1cago Press,
293-314.
Wimsatt, w.e. (1986b). Forms of aggregativity. In A.

Donagan, A. N.
Perovich Jr., & M. Wedin (Eds.), Human nature and natural knowledge.
Boston & Dordrecht: Reidel, 259-291.
Winston, M. (1987). Recent work in evolutionary epistemology: Critical
notice of Wuketits, Plotkin, and Brewer & Collins. In preparation for
Philosophy of Science.
Winter, S.G., Jr. (1975). Optimization and evolution in the theory of the
firm. In R.H. Day & T. Groves (Eds.),
Adaptive economic models.
New York: Academic Press, 73-118.
Wuketits, F.M. (1978a). Die Ordnung der Natur und die Natur
der
Ordnung. In W. SchAifer (Ed.), Evoluierende Systeme, 3. Frankfurt:
Kramer, 163- 172.
Wuketits, F.M. (1978b). Wissenschaftstheoretische Probleme der modernen
Biologie. Berlin: Duncker &; Hum610t.
Wuketits, F.M. (1981a). Biologie und Kausalitit. Biologische Ansitze zur
Kausalitiit, Determinatl.on und Frel.hel.t. Berhn: Parey.
Wuketits,
F.M.
(1981b).
EVolutl.onare
Erkenntnistheorie:
Die neue
Herausforderung. In K. Lorenz & F .M. Wuketits (Hg.), Die Evolution
des Denkens. Munchen: Piper, 11-28.
Wuketits,
F.M. (1982). Evolution, Kausalitat und Erkenntnis. In W.
Ziegler (Ed.), Organismus und Anpassung. Frankfurt: Kramer, 109-121.
Wuketits, F.M. (1983a). Evolutl.onaryepl.stemology, objective
knowledge,
and rationality: The evolutionary approach
in man I s search for
himself. In International Cultural Foundation (Ed.). Absolute values
and the creation of the new world, 2. New York: Internatl.onal
Cultural Foundatl.on Press, 881-900.
Wuketits, F .M. (1983b). Evolutionsmodelle in der Erklarung
menschlicher
Denkstrukturen im 19. Jahrhundert. Berichte Wissenschaftsgeschichte,
6, 115-122.
Wuketits, F.M. (1983c). Erkenntnis ohne Illusion. tiber

einige Grundlagen
und Konsequenzen der evolutionaren
Erkenntnistheorie. Conceptus,
17, 44-56.
Wuketits, F.M. (Ed.) (1984a). Concepts and approaches in evolutionary
epistemology: Towards an evolutl.onary theory of knowledge. Dordrec;ht
& Boston: ReideL
Wuketits,
F.M.
(1984b).
Evolution,
Erkenntnis,
Ethik. Darmstadt:
Wissenschaftliche Buchgesellschaft.
431
Wuketits.
and
F.M. (1984c). Evolutionary epistemology: A challenge to science

philosophy. In F.M. Wuketits (Ed.), Concepts and approaches in
Wuketits.
F .M. (1985). Zustand und
. . evolutionary epistemology. Dordrecht & Boston: ReJ-del, 1-29.

Bewu~tsein.
Leben als biophilosophische
Wuketi~rtheF~M. H(~~~~/l ~~~5~an:v~~tr:pe~nd
epistemology: Let I s begin

discussion. A reply to Ruse. Newsletter: For those interested in the
philosophy of Karl Popper, 2(374), 8-9.
Wuketits. F.M. (1986). Evolution as a cognitive process: Towards an

evolutionary epistemology. Biology & Philosophy, 1, 191-206.
Wuketits, F.M. (1987). Evolutionare Urspriinge der Metaphysik. In R. Riedl
&
F.M.
Wuketits (Eds.),
Die
evolutionare Erkenntnistheorie.
Berlin: Paul Parey.
Yilmaz, H. (1962). Color vision and a new approach to general
perception.
Biological Prototypes and Synthetic Systems, 1, 126-141.
*Yilmaz, H. (1967/1968). A theory of speech perception. Bulletin of
Mathematical Biophysics, 1, 29. 1967, 793-825j 2, 30, 1968, 455-479.
*Yilmaz, H. (1973). Percept~on and philosophy of science. In R.S. Cohen &
M. W.
Wartofsky CEds.), Logical and Epistemological Studies in
Contemporary Physics. Dordreeht: Relde!, 1-91.
*Zirkel, C. (1941 ). Natural selection before the origin. Proceedings of the
American Philosophical Society, 84, 71-123.
Ackerman~
N. W. & Jahoda, H. (1950). Antisemitism and emotional disorder.
New York: Harper.
Adorno, T.W., Frenkel-Brunswick, Eo, Levinson, D.J. & Sanford, R.N. (1950) .
.!h!. authoritarian personality. New York: Norton, 1969.
Allen, P.M.. Engelen, G. & Sanglier, H. (1984). New methods for policy
exploration in complex systems, communication to a UNU conference at
Hontpellier. France, under the theme Praxis and Management of Complexity. See also special issue of Environment and Planning,
BI2
(1985), 1, 1-138.
Allport, G.W. (1954). Th! ~ .2! prejudice. New York: Doubleday.
Allport, G.W. & Postman, L. (1947) .TI:!! psychology.2f!:!::!!!!2!.' New York:
Holt.
Alston, W.P. (1976). Two types of foundationalism. Journal.Q! Philosophy,
73, 165-185.
Althusser, L. (1965). Pour Marx. Paris: Kasp~roJ 1968.
Althusser, L. (1969). ~n;-;t' la philosophie. Paris: Haspero.
Althusser, L. (1970). IdeologI; et appareils ideologiques d'Etat. In:
Positions. Paris: Editions Sociales, 1976.
A1thus~ Balibar, E. (1968). !4.!:!!! capital, !, Paris: Haspero.
Apostel, L. (1961). Towards the formal study of models in the non-formal
sciences. In: H. Freudenthal (Ed.), The role of the model in mathematics, natural and social sciences. Dordrecht:Reidel-.-- - - - AposteI:"L. (1977). Discussion~. Inhelder, R. Garcia & J. Voneche
(:~ds.), Epistemologie genetique !!! eqUilibration. Neuchitel & Paris:
Delachaux & Niestle, 61-63.
Aristotle. PhYSiCS. Ed. Ross. Oxford: Clarendon.
Armstrong, D.H. (1981). The !!!E!!:! 2!. !!!!!!! and other essays.
Ithaca:
Cornell University Press.
Arnheim, R. (1971). Visual thinking. Berkeley: University of California
Press.
Arthur, W. (1984). Mechanisms .2! morphological evolution: ~ combined genetic, developmental and ecolosical approach. New York: Wiley.
Babloyantz, A., Salazar, J.H. & Nicolis, C. (1985). Evidence of chaotic
dynamics of brain activity during the sleep cycle. Working paper,
Departement de Chimie Physique II, CP 231, Universite
Libre de
Bruxelles.
Barash, D.P. (1977). Sociobiology and ~. New York: Elsevier.
Barnes. .B. (1974). Scientific knowledge .!!!!! sociological theory. London:
Routledge & Kegan Paul.
Barnes, B. & Edge, D. O!:ds.) (1982). Science.!!! ~ Readings !!! the
sociology 2!.!..!!!!!. Milton Keynes: Open University Press.
Barthes, R. (1957). Mythologies. Paris: Seui!.
Battali. J. (1983). Computional introspection. !! Memo 701, HIT, February.
Bechtel * W. (Ed.) (1986). Integrating scientific diSCiplines. The Hague:
Nijhoff.
Berger, P.L. & Luclemann, T. (1966). The .!.2!!.! construction 2! reality. New
York: Doubleday, 1967.
Bergmann,
G. (1957). Philosophy .2!. !..!!!!!. Madison: University of
Wisconsin.
Beth, E. & Piaget, J. (1961). Epistemologie mathematigue !! psychologie.
Paris: Presses Universitaires de France.
Black, H. (1962). Models!!!!! metaphors. Ithaca, NY: Cornell University
Press.
Black, H.
(1985). Ambiguities of rationality. Paper presented at conference on IINaturalism and Rationality", State University of New York,
Buffalo, March, 1985.
433
434
Bloor, D. (1976). Knowledge and social imagery. London: Routledge & Kegan
Paul.
Boden, M.A. (1979). Piaget. Glasgow: Fontana.
Boden, M.A. (1984). Animal perception from an artifical intelligence viewpoint.
In: C.
Hookway (Ed.), Minds, machines, and evolution.
Cambridge: Cambridge University Pre;;:------Boden, M.A. (1984). Artificial intelligence and biological reductionism. In
M.~W.
Ho & P. Saunders (Eds.), Beyond nea-Darwinism. New York: Academic Press, 317-329.
Bohm, D. (1969). Some remarks on the notion of order. In: C.H. Waddington
(Ed.). Towards ~ theoretical biology, 2. Edinburgh: Edinburgh University Press, 18-60.
Bohr, N. (1924). On the spectrum of hydrogen. Lecture delivered to the
Physical Society of Copenhagen on 20 December 1913. Reprinted in
Bohr. N.. The theory !. spectra ~ atomic constitution. Cambridge.
U.K.: Cambridge University Press, 1-19.
Borel, M.J. & Gillieron, C. La classification des sciences selon Jean
Piaget. (In preparation).
Born, M. (1923). Quantentheorie und St5rungrechnung. Die Naturwissenschaften, 27, 537-550.
Boudon, R. (1984). Le phenomeme ideologique: En marge d'une lecture de
Pareto. L'Annee Sociologique, 34.
Boulding, K.E. (1981). Evolutionary ~. Londen: SAGE.
Bourdieu, P. (1966). Le desenchantement du mende. Paris: Centre de Socielogie Europeenne.--Bourdieu. P. (1971). Genese et structure du champ religieux. Revue Francaise ~ Sociologie, 12, 295-334.
Bourdieu, P., Passeron, J.C. & Chamboredon, J.C. (1968). ~ metier ~
sociologue. Paris: Mouton & Bordas.
Brandon. R.N. (1981). Biological teleology: Questions and explanations.
Studies .!!! the History !ill! Philosophy i. Science. 12. 91-105.
Brandon. R. N. ( 1982) . The levels of selection. In: PSA 1982. 1. East
Lansing: Philosophy of Science Association, 315-32~-Brandon, R.N. (1985). Grene on mechanism and reductionism: More than just a
side issue. In: PSA 1984, 2. 345-353.
Brandon. R.N. & Buria;;:-R~(Eds.) (1984). Genes. organisms. populations:
Controversies ~ the units 2! selection. Cambridge, MA: MIT Press.
Brannigan, G. (1979). The reification of Mendel. Social Studies ! ~,
9, 423-454.
Brannigan.
G.
(1982). The social basis f scientific discoveriel!'
Cambridge: Cambridge University Press.
Brecht. B. (1929-1941/1967). Notizen zur Philosophie. In: Gesammelte ~.
20, Schriften zur Politik und Gesellschaft. Frankfurt: Suhrkamp. 125-
178.---- -
--- -
Brewer, M.B. (1981). Ethnocentrism and its role in interpersonal trust. In

M.B. Brewer & B.E. Collins (Eds.), Scientific inquiry and the social
sciences. San Francisco: Jossey-Bass, 345-360.
Brewer. M.B. & Collins. B.E. (1981). Perspectives on knowing: Six themes
from Donald T. Campbell. In: id. (Eds.). Scientific inquiry and the
social sciences. San Francisco: Jossey-Bass. 1-9.
Brewer, M.B. & Campbell. D.T. (1976). Ethnocentrism and intergroup attitudes: East African evidence. New York: Halsted Press.
Brooks-:--n:R".-&-Wiley, E.O. (1986). Evolution ~ entropy: ~ !. unified
theory !. biology. Chicago: Chicago University Press.
Brown. G. & Yule. G. (1983). ~ analysis. Cambridge University
Press.
Buchel, W. (1982). Teleologie und Negentropie. Zeitschrift fUr allgemeine
Wissenschaftstheorie. 13. 40-47.
BUnge. M. (1978). A model of evolution. Applied Mathematical Modelling. 2.
201-204.
Burghardt. G.M. (1985). Animal awareness: current perceptions and historical perspective. American Psychologist. 40. 905-919.
435
Burian, R.M., (1978). A methodological critique of sociobiology. In A.L.

Caplan (Ed.), The sociobiology debate. New York: Harper and Row, 376395.
Burian, R.M. (1986). On integrating the study of evolution and of development. In: W. Bechtel (Ed.), Integratins scientific disciplines. The
Hague: Nijboff, 209-228.
Buscaglia, M. (1982). Jean Piaget biologiste. Archives !!! Psychologie, 50,
31-39.
Buscaglia, M. (1983). Bio!ogie et verites. In: M. Buscaglia, C. Lalive
d'Epinay, B. Morel, H. Ruegg & J. Voneche (Eds.), 1!! crit~res Q.
vlSrite dans .!! recherche scientifigue. Paris: Kaloine, 11-20.
Cairns, J., Stent, G.S. & Watson, J.D. (eds.). (1966). Phage and !!!!.
origins .Q! molecular biology. (DelbrUck Festschrift.) Cold Spring
Harbor Laboratory of Quantitative Biology.
Callebaut, W. (1988). Post-positivist views of scientific explanation. To
appear in the Proceedings of the 13th Quetelet Seminar, Uni vers i t~
Catholique de Louvain.
Campbell, D. T. (1965). Variation and selective retention in socia-cultural
evolution. In H.R. Barringer, G.!. Blanksten & R.W. Mack (1~ds.),
Social change !!!. developing ~ ! reinterpretation 2! evolutionary
theory. Cambridge, MA: Schenkman.
Campbell, D.T.
(1969). Ethnocentrism of disciplines and the fish-scale
model of omniscience.
In: M. Sherif
& C.W. Sherif (Eds.), Interdisciplinary relationships .!!!. the social sciences. Chicago: Ald~
Campbell, D.T. (1970). Natural selection as a epistemological model. In: R.
Naroll & R. Cohen (l!:ds.), ! handbook 2! method !!! cultural anthropo!Qn. Garden City: Natural History Press, 51-85.
Campbell, D.T. (1972). On the genetics of altruism and the counter-hedonic
components in human culture. ~.2! fu!i!!~, 28, 21-37.
Campbell, D.T. (1975). On the conflicts between biological and social
evolution and between psychology and moral tradition. American ~
chologist, 30, 1103-1126.
Campbell, D.T. (1978). Qualitative knowing in action research. In M.
Brenner, P. Marsh & M. Brenner (Eds.), The social contexts of method.
London: Croom Helm, 184-209.
- - - - - - --Campbell, D.T. (1979). CODll1ents on the sociobiology of ethics and moralizing. Behavioral ~, 24, 37-45.
Campbell, D. T. (1982). Legal and primary-group social controls. Journal of
Social ~ Biological Structures,S, 431-438.
--- Campbell, D.T. (1983). The two distinct routes beyond kin selection to
ultra-sociality: Implications for the humanities and social sciences.
In: D.L. Bridgeman (Ed.). The ~ 2! prosocial development: ~
diSCiplinary ~ and strate ies. New York. Academic Press, 11-41
Campbell. D.T. & Fiske, W. 0959 . Convergent and discriminant validation
by the multitrait-multimethod matrix.
Psychological Bulletin. 56,
81-105.
Cantor, M. (1901-1913).
Vorlesungen tiber Geschichte der Mathematik.
Leipzig: Teubner.
Caplan. A.L. (Ed.) (1978). .Th!. sociobiology ~. New York: Harper and
Row.
Caporael, L.R. (1981). The paralanguage of caregiving: Baby talk to the
institutionalized aged. ~.2! Personality.!!!! Social PsychOlogy,
40, 876-884.
Caporael. L.R. (1984). Prolegomenon 12 !!! evolutionary psychology. Unpublished manuscript.
Caporael. L.R.,
Lukaszweski. M.P. & Culbertson. G.H. (1983). Secondary
baby talk: Judgments by institutionalized elderly and their caregivers. Journal .2! Personality ~ ~ Psychology, 44, 746-754.
Capra, F. (1982). ~ turning point. New York: Simon & Schuster.
436
J .C. (1985). Are animals conscious? Paper presented at conference

on History, Philosophy an Social Studies of Biology. South Bend,
Indiana, June 1985.
Carloye,
Cartwright, N. (1983). How the!!!!!.

sity Press.
Chaitin,
G.J. (1975). Randomness
2!
physics lie. Oxford: Oxford Univer-
and
mathematical
proof.
Scientific
232, 47-52, esp. p. 52.

Chernoff, H. (1971). The use of faces to represent points in K-d1mensional
space graphically. Journal.2! the American Statistical Association,
68, 361-368.
~t
Childe, V.C. (1951). Social evolution. London: Watts.

Chisholm, R.M. (1982). The ~ns 2f knOWing. Minneapolis: University
of Minnesota Press.
Chomsky,
N. (1980). Rules and representations. New York:
sity Press.
Columbia Univer-
Church, R. (1984). Popper's "World 3" and the problem of the printed line.
Australasian Journal .2! Philosophy, 62, 378-391.
Churchland, P.S. (1983). Dennett's instrumentalism: a frog at the bottom of
a mug. Behavioral and Brain Sciences, 6, 358.
Cellerier, G. (1980).
The construction of a theory: A prescriptive endeavour, Cahiers 4!!! londation ~:!!!!!. Plaget, 1, 177-199.
Clark, A.J. (1987). The evolutionary constraints on cognitive models (paper
draft).
Clement, J. (1982). Student's preconceptions in introductory mechanics.
~ Journal .2! PhYSiCS, 50,
66-71.
Cohen, G.A. (1978). ~ Marx's theory .2! history: A defence. Oxford:
Oxford University Press.
CODDDons, H.L., Richards, l.A. & Armon, C. (Eda.) (1984). Beyond formal
operations: Late adolescent development and adult cognitive development. New York: Praeger.
Craik, K. (1943). The!!!E!!:! 2!. explanation. Cambridge: Cambridge University Press.
Crombie, A.C. (1967). The mechanistic hypothesis and the scientific study
of vision: Some optical ideas as a background to the invention of the
microscope. In: S. Bradbury & G.L.E. Turner (Eds.), Historical !!.:.
peets of microscopy. Cambridge: Heffer, 3-112.
Dawkins, R. 0976). The selfish &!!!!!.. Oxford: Oxford University Press.
Dawkins, R. (1976). Hierarchical organization: a candidate principle for
ethology. In: P.P.G. Bateson and R.A. Hinde (Eds.), Growing points !!!.
ethology. Cambridge: Cambridge University Press, 7-54.
Dawkins, R. (1982). The extended phenotype. Oxford: Oxford University
Press.
Denbigh, K. (1981). How subjective is entropy?, Chemistry.!!!!.!:ili.i!!., 17,
168-185.
Deneubourg, J.L.. Pasteels, J.H. & Verhaeghe, J.C. (1985). Probabilistic
behaviour in ants: A strategy of errors? Journal of Theoretical
Biology, lOS, 259-271.
--- Dennett, D.
(1982). How to study consciousness empirically: Or, nothing
comes to mind. Synthese, 53, 159-180.
Depew, D.J. (1986). Nonequilibrium thermodynamics and evolution: A philosophical perspective. Philosophica, 37, 27-57.
Descartes, R. (1948). Discours de la ~thode. Paris: Bader-Dufour. (Orig.
1637. )
Descartes, R. (1955). Meditations on first philosophy. In: E.S. Haldane
(Ed.), Th! philosophical .!2.!:!.!.2!.~. New York: Ross.
Dobzhansky, T. (1962). Mankind evolving: The evolution .2! the human .!.2!.:.
eies. New Haven: Yale University Press.
Donchi;;:--X.
(1981). P300 and classification.
In: R. Galambos & S.A.
Hillyard (Eds.), Electrophysiologieal approaches to human cognitive
processing. Neuroscience Research Program Bulletin, 20, 157-161.
437
Donin!, A. (1975). Storia del Cristianesimo. Milano: Teti.

Dretske, F.
(198~etnformational character of representations. Behavioral and Brain Sciences,S, 376.
DysonJ~9). ~rbing the~. New York: Harper & Row.
Eddington, A.S. (1927). The .!!!E:!!:! Qf the physical world. Ann Arbor. Michigan: University of Michigan Press. (From Eddington1s Gifford Lectures
of January-March.)
Eddington, A.S. (1948). The !!!E!!:! 2t the physical world. New York:
McMillan.
Edwards, H.M. (1979). Fermat's last theorem:
! genetiC introduction !
algebraic
theory. Heidelberg: Springer.
Eigen, M. & Schuster, P. (1977-1978). The hypercycle. Naturwissenschaften,

64, 541-565; 65, 7-41; 65, 341-369.
Einstein, A. (1934). The problem of space, ether and the field in physics.
Reprinted in A. Einstein, Essays .!!! ~. New York: Philosophical
Library, 61-77.
Ekeland. S. (1984). Le calcul, l'imprevu. Paris: Seui!.
Eldredge, N. & Gould. S.J. (1972). Punctuated equilibria: An alternative to
phyletic gradualism. In: T.J .M. Schopf (Ed.), !:!2!!!!! .!!! paleobiology.
San Francisco: Freeman, 82-115.
Eldredge, N. & Cracraft, J. (1980). Phylogenetic patterns and the evolutionary process. New York: Columbia University Press.
Eldredge. N. & Tattersall, I. (1982). The myths !!E h!!!!!!!. evolution. New
York: Columbia University Press.
Eldredge, N.
and Salthe, S.N.
(1985). Hierarchy and evolution. Oxford
Surveys !!!. Evolutionary Biology, I, 184-208.
Elster, J. (1976). A note on hysteresis in the social sciences. Synthese,
33, 371-391.
Elster, J. (1978). Logic!!!!! SOCiety. New York: Wiley.
Elster, J. (1982a). Belief, bias and ideology. In: M. Hollis & S. Lukes
(Eds.), Rationality .!!!2. relativism. Oxford: Blackwell, 123-148.
Elster, J. (1982b). A paradigm for the social sciences? [Review of Van
Parijs. 1981]. Inquiry, 25, 378-385.
Elster, J. (1983a). Sour grapes: Studies .!!! the subversion .2!. rationality.
Cambridge: Cambridge University Press.
Elster. J.
(1983b). Explaining technical change. Cambridge: Cambridge
University Press.
Felson, R.B. & Gmelch, G. (1979). Uncertainty and the use of magic. Cur!!!!! Anthropology, 20. 587 - 589.
Festinger, L. (1957). ! theory .2! cognitive dissonance. Evanston: Row &
Peterson.
Feyerabend, P. (1975). Against ~. London: New Left Books.
Flavell, J.B. (1977). Cognitive development. Englewood Cliffs, N.J.: Prentice Hall.
Freud, S. (1899). nie Traumdeutung. Wien: Deuticke, 1950.
Freud, S. (1901). Ueber den Traum. In: Gesammelte Werke, 2. Frankfurt:
Fischer. 1942, 643-7'667 - -Freud, S. (1915-1917/1969). Vorlesunsen E!! Einfuhrung .!!! die Psychoana.!Y.!!. Frankfurt: Fischer.
Freud, S. (1927). Die Zukunft einer Illusion. In: Gesammelte Werke. 14,
Frankfurt: Fischer, 1942, 323-38-0-.-Freud, S. (1930). nas Unbehagen !!! der Kultur. In: GesaalDBlte Werke, 14,
Frankfurt: Fischer, 1942, 419-506.
Freud, S. (1933). Neue Folge der Vorlesungen .!y! Einfiihrung !!!. die Psychoanalyse. Frankfurt: Fischer, 1967.
Fox, M.S. (1981). An organizational view of distributed systems. IEEE
Transactions ~ Systems. ~ and Cybernetics, SMC-ll, 70-80.
-Galambos. R. & Hillyard, S.A. (1981). Electro-physiological approaches to
human cognitive processing. Neuroscience ~ Program~,
20, 141-164.
438
Garcia, J. & Koelling, R.A.

(1966). Relation of cue to consequence in
avoidance learning. Psychonomic~, 4, 123-124.
Geertz, C. (1983). Local knowledge: Further essays !!!. interpretative !!!:.
thropologvo New York: Basic Books.
Ghiselin, M.T. (1974). A radical solution to the species problem. ~
GhiSeli:a~:~. z(~~~r5: ~;te~!~~;:~' life,
and thinking. Behavioral and Brain

Sciences, 4, 269-313.
-Ghiselin, M. T. (1985). Bioeconomics and the metaphysics of selection. Paper
read at the 3rd International Congress of Systematic and Bvolutionary
Biology.
Giere, R. (1984). Realism in the laboratory. Paper presented at the George
Sarton Centennial, University of Ghent, Belgium, November 1984.
Gillieron, C. (1982). Conservation: Forty-five years later. ~.2.!.
Structural Learning. 7, 167-174.
Ginsberg. H. (1961). Social evolution. In H. Banton (Ed.), ~ and
the study .2.!. SOCiety. Chicago: Quadrangle Books.
Godelier, H. (1978). La part ideelle du r'el. Essai sur 11 id'ologique.
L'Homme, 18. 155-188.
Goldmann. L. (1966). Sciences humaines ~ philosophie. Paris: Gonthier.
Goodwin. B.C. (1976). Analytical physiology .2!. !.!..!! .!!!2. developing .2!:.
ganisms. New York: Academic Press.
Goodwin, B.C. (1978). A cognitive view of biological processes. Journal of
~!!!!! Biological Structures, 1, 117-125.
--- Goodwin. B.C. (1982). Genetic epistemology and constructionist biology.
Revue Internationale .Q!. Philosophie, 142-143, 527-548.
Gould, S.J. (1976). DIArcy Thompson and the science of form. In: H. Grene &
E. Hendelsolm (Eds.), Topics in the philosophy 2! biology. Dordrecht:
Reidel. 66-97.
Gould, S.J. (1982). Darwinism and the expansion of evolutionary theory.
Science, 216. 380- 387 .
Gould, ~(1983). The hardening of the modern synthesis. In: M. Grene
(Ed.). Dimensions of Darwinism. Cambridge, U.K.: Cambridge University
Press, 71-93.
- ---Gould. S.J. & Eldredge, N. (1977). Punctuated equilibria: The tempo and
mode of evolution reconsidered. Paleobiology, 3, 11-151.
Gould, S.J. & Vrba, E.S. (1982). Exaptation - a missing term in the science
of form. Paleobiology. 8, 4-15.
Gray. W. & Rizzo, M.D. (Eds.) (1973). Unity throush diversity (2 vols). New
York: Gordon & Breach.
Gr~co, P.
(1967}. Epistemologie de la psychologie. In: J. Piaget (Ed.).
Logigue !! connaissance scientifigue. Paris, Gallimard, 927-991.
Grene, H. (1969). Behm's metaphysics and biology. In: C.H. Waddington (Ed.)
~ !: Theoretical Biology.
2. Edinburgh:
Edinburgh University
Press, 61-69.
Grene, H. (1974). The understanding .Q! nature. Dordrecht: Reidel.
Grene, M. (1985). Hierarchies and behaviour. Paper raad at the 3rd International Congress of Systematics and Evolutionary Biolo8Y.
Griffin, D.R.
(1976). The Question 21 Animal Awareness. New York: Rockefeller University Press.
Griffin, D.R. (1978). Prospects for a cognitive ethology. Behavioral!!!!!
Brain Sciences, 1, 527-538.
Griffin:-n.~) (1982). Animal mind - human mind: Introduction. New
York: Springer.
- - ---- --Griffin, D.R.
(1984). Animal Thinking. Cambridge, HA: Harvard University
Press.
Grimes, J. (1975). ~ ~ 21~. The Hague: Mouton.
Gruber, H.E. (1978). Darwin's 'tree of nature' and other images of wide
scope. In: J. Wechsler (Ed.), On Aesthetics in Science. Cambridge,
M.LT. Press, 121-140.
- ---
GENERAL BIBUOGRAPHY
439
Gruber, H.E. (1979). The changing shape of Pisget's 'circle of the sciences I . Unpublished paper presented at the International School of
History of Science, Ettore Majorana Centre for Scientific Culture,
Erice (Sicily), August 1979.
Ham.lyn, D.W. (1970). The theory .2!. knowledge. London: MacMillan Press.
Hamad, s.
(1985). Bugs, slugs, computors and consciousness. American
Scientist, 73, 121-125.

Haroutunian. S. (1983). Equilibrium!!! the balance:
gical explanation. New York: Springer.
Hansell,
M.J.
Lepidostoma
(1972).
hirtum.
Case
study
!.
psycholo-
building behaviour of the caddis-fly
Journal.2f the
Zoological Society
larva
(London),
167, 179-192.
Rebb, D.O. (1949). The organization .2! behavior: ! neuropsychological theo!I.. New York: Wiley.
Beidegger, H. (1975).!ll.!. Grundprobleme der Phiinomenologie. Engl. tr. A.
Hofstadter. The !!!!! Problems 2! Phenomenology. Bloomington, IN:
University Press, 1982.
Heider, R. &
Simme!, H. (1944). An experimental study of apparent behaviour. ~ Journal .2! Psychology, 57, 243-259.
Heisenberg, W. (1926). Quantenmechanik. lli:!. Naturwissenschaften, 14, 889-
894.
Heisenberg, W. (1975). Discussion with Heisenberg. In: O. Gingerich (Ed.),
The nature of scientific discovery: A symposium commemorating !h!.
500th anniversary of the birth ! ~ Copernicus. Washington:
Smithsonian Press. 556-573.
Henriques. G. (1984). Le preformisme: Critique de ses fondements et de sa
valeur explicative. Archives!!! Psychologie, 52. 53-68.
Hesse. M. (1970). ~.!.!!2 analogies .!!!~. Notre Dame, IN: University of Notre Dame Press.
Hesse, M. (1974). The ~ 2! scientific inference. Berkeley: University of California Press.
Hesse, M. (1980). Revolutions and reconstructions.!!!!h! philosophY .2!.
science. Brighton, Sussex: Harvester Press.
Heyes, C.M. (in preparation). Psychological epistemology. To be submitted
to !mer ican PsycholoB ist.
Hirshleifer, J. (1977). Economics from a biological viewpoint. Journal.Qf
Law and Economics, 20, 1-52.
Ho.
H':'"='W:-& Saunders. P. (Eds.) (1984). Beyond neo-Darwinism. New York:
Academic Press.
Holldobler, B. & Michener, C.D. (1980). Mechanisms of identification and
discrimination in social hymenoptera. In H. Harkl (Ed.), Evolution of
social behavior: Hypotheses and.!!.!ll. Deerfield Beach,~i;i
Chemie, 35-57.
Holmes, W.G. & Sherman. P.W. (1983). Kin recognition in animals. American
Scientist, 71. 46-55.
--Hube1,~963). The visual cortex of the brain. Scientific American,
November.
Hull. D.L. (1974). Philosophy.Q.f biological ~. Englewood Cliffs, NJ:
Prentice-Hall.
Hull, D.L. (1976). Are species really individuals? Systematic Zoology, 25,
174-191Hull. D.L. (1980). Individuality and selection. Annual Reviews of Ecolon and SystematiCS. 11, 311-323.
- - --- --Hull. D.L. (1981). The herd as means. In: ~ 1980, 2. East Lansing. HI,
Philosophy of Science Association. 73-92.
Hull. D.L. (1983). Exemplars and scientific change. In: PSA 1982, 2. East
Lansing, HI: Philosophy of Science Association, 479-503-.Hull, D.L. (1985). Darwinism as a historical entity: A historiographic
proposal. In: D. Kolm (Ed.), The Darwinian heritage. Princeton. NJ:
Princeton University Press, 773-812.
440
GENERAL BIBUOGRAPHY
Humphrey, N. K. (1978). Nature's psychologists. New~, 78, 900-903.

Humphrey, N. K. (1982). Consciousness: a just-so story. ~~, 95,
474-477
Ingvar, D.H. (1985). 'Memory of the future I: An essay on the temporal
organization of conscious awareness. Human Neurobiology. 4, 127 -136.
Inhelder, B. t Garcia. R. & Vonhche, J. (Eds.)(1977). Episteologie geneti!l!!! !! equilibration. Montreal: Delachaux et Niestle.
Irons, W. (1979). Natural selection, adaptation, and human social behavior.

In: N. Chagnon & W. Irons (Eds. L Evolutionary biology and human
social behavior: ~ anthropological perspective. North Scituate:
Duxbury.
Ittelson, W.R. (1952). The ~ demonstrations !!! perception. Princeton:
Princeton University Press.
Jacob, F. (1977). Evolution and tinkering, Science, 196, 1161-1166.
Jacobs, J. (1986). Teleology and reducti~n biology. Biology and
Philosophy, 1, 389-399.
Jantsch, E. (1980). !.h!. self-organisins!!!l!:!!!!!..:. Scientific and ~

implications !. the emerging paradigm 2! evolution. Oxford: Pergamon.
Jarvie, I.e. (1964). The revolution !!! anthropology, London: Routledge &
Kegan Paul.
Jaynes, J.
(1976). The origins .2!. consciousness !!!. the breakdown .2!. ~
bicameral mind. Boston: Houghton Miffin.
Kahnemann, D. &""TVersky, A. (1982). The psychology of preferences. Scientific American, 246, 136-142.
-Kahnema--;;:--D., Slovic, P. & Tversky, A. (Eds.) (1982). Judgment ~!!!!:.
certainty: Heuristics and biases. New York: Cambridge University
Press.
Kaufmann, S. (1974). Elsasser, generalized complementarity. and finite
classes: A critique of his anti-reductionism. In: K.F. Schaffner &
R.S. Cohen (Eds.), PSA 1972. Dordrecht: Reidel, 57-65.
Keat, R. & Urry, J. Social theory !!! !.!.!!lE.!.. London, Routledge and Kegan,
1975.
Berry, J.K. (1980). ~ ill: Knowledge, affection and

toward animals !!! American SOCiety, U.S. Fish and
Wildlife Service, Washington D.C.
Kimura, M. (1983). The ~ theory !. !!!2.!!.!!.!!!~. Cambridge,
U.K.: Cambridge University Press.
Kitchener, R.F. (1983). Developmental explanations. Review .2!. Metaphysics,
XXXVI, (4).
Kitcher, P. (1985). Vaulting ~ Sociobiology!!!!! the quest !Q!. h!:!!!'!!!!.
nature. Cambridge, KA: HIT Press.
Knorr Cetina, K. (1982). Descriptive epistemology of scientific work. Paper
presented at the Boston Colloquium in Philosophy of Science. Boston,
December 1982.
Kramers, H. & Holst, H. (1923). The atom and the Bohr theory of its structure. Translated by R.B. & R.T. Lindsay. London: Gylendal.
Kroodsma, D.E. & Miller, H.M. (1982). Acoustic communication in birds (3
vols.). New York: Academic Press. - - - -Kuhn, T.S. (1970). The structure of scientific revolutions. 2nd, enlarged
ed. Chicago: UnI;;rsity of Chicago Press.
Kuhn, T. (1978). Black ~ theory and the quantum discontinuity. Oxford.
Kurland, J.A. & Beckerman, S.J. (1985). Optimal foraging and hominid evolution: Labor and reciprocity. ~ Anthropologist, 87, 73-93.
Ladriere. J. (1982). Piaget et la logique. ~!!! Psychologie, 50,
Kellert,
S.R. &
17-29.
Lakatos, I. (1970). Falsification and the methodology of scientific research programmes. In: J. Worrall and G. Currie (Eds.), The method.2.!s:l 2i scientific research programmes. Philosophical Papers, 1,
Cambridge.
441
Lakatos, 1. (1976). Proofs and refutations:
!!!! logic
diSCOV1ry. Cambridge: Cambridge University Press.
.!!!.
mathematical
Lakatos,!. 1978). History of science and its rational recontructions. In:

J. Worrall and G. Currie (Eds.), The methodology B!
scientific
research programs. Philosophical Papers, 1, Cambridge.
Lakatos, I. & Zahar. E. (1978). Why did Copernicus' research programme
supersede Ptolemy's? In: J. Worrall and G. Currie (eds.). The Methodology.2! Scientific Research Programs. Philosophical Papers 1, Cambridge.
Lamontagne, C. (1973). A new experimental paradigm for the investigation of
the secondary system of human visual motion perception. Perception,
2. 167-180.
Lamontagne, C. (1974). Defining some primitives for a computational theory
of visual motion perception. ~ ~ Conference Proceedings.
Brighton, U.K.: Brighton University, 90-10l.
Lamontagne, C. (1975). Steps towards! computational theory f visual
motion detection. Ph.D. Dissertation, Edinburgh, Scotland: University
of Edinburgh.
Lamontagne, C. (1976). Visual motion detection: a computational theory and
some of the psychological data which i t integrates. A.I.S.B. Summer
Conference Proceedings. Edinburgh, Scotland: University of Edinburgh,
175-183.
Lamontagne. C. & Beausoleil, J.-R. (1982). Achieving visual spatiality:
Towards a psychologically relevant, physiologically plausible, and
computationally efficient conjecture. Cognition and Brain Theory,
5(4). 341-363.
Lamontagne. C. & Beausoleil, J. -R. (1984). Vers une representation conceptuelle de la representation preconceptuelle. Communication Information, 6(2-3), 177-20l.
Lamontagne, C. & Howe, J.A.M. (1980). Towards a computational theory of
visual motion perception: Macro-issues. Ghent: Communication and
Cognition.
Landau, H. (1984). Human evolution as narrative. American ~, 72,
262-268.
Lang, W. (1980). Marxism, liberalism and justice. In: E. Kamenka & A.E.S.
Tay (Eds.), ~. London: Arnold, 116-148.
Laudan, L. (1981). ~ and hypothesis. Dordrecht: Reidel.
Laudan, L. (1981b). The pseudo-science of science, Philosophy 2! the Social
Sciences.
LeDoux~ Wilson, D.H. &
Gazzaniga, M.S. (1979). Beyond commisurotomy: Clues to consciousness. In: M.S. Gazzaniga (Ed.), Handbook .Qf
Behavioral Neurobiolo8Y, 2. New York: Plenum Press.
Levine, A. & Sober, E. (1985). What's historical about historical materialism? Journal .2! Philosophy, 82, 304-326.
Levins~ R. & Lewontin, R. (1985). The dialectical biologist. Cambridge, MA:
Harvard University Press.
Lewin, R. (1985). Fish to bacterium gene transfer. Science, 227 (March 1.
1985) .
Lewis, G.N. (1930). The symmetry of time in phYSiCS, ~, 71, 570.
Lewis, L.
(1963). Knowledge, danger, certainty, and the theory of magic.
American Journal .2! Sociology, 69, 7-12.
Levontin, R.C. (1970). The units of selection. Annual Review .Q! Ecology and
Systematics, 1, 1-18.
Levontin, R.C. (1974). The analysis of variance and the analysis of causes.
American Journal of Human Genetics, 26, 400-411.
Levantin, R.C. (1978). Adaptatio~tific ~, 239, 212-230.
Levontin, R.C. (1982b). Biological determinism. In: H.C. Plotkin (Ed.).
Learning. development and culture. New York: Wiley, 151-170.
Levontin, R.C. (1983). Darwin's revolution. !i!!! X2!.! ~ .2! Books, 30,
21-27.
442
S. & Oppenheim, P. (1974). Generalization of cf)mplementarity.

Synthese, 28, 117-139.
Lloyd. E.A. (1986). Evaluation of evidence in group selection debates. In:
PSA 1986. 1. East Lansing. HI: Philosophy of Science Association,
483-493.
Lorenz, K. (1941). Kant I s Lehre vom Apriorischen im Lichte gegenwartiger
Biologie. llillll fUr Deutsche Philosophie. 15, 94-124. Translated in
General Systems. 7 (1962). 23 - 35.
Lorenz, K. (1962). Kant I s doctrine of the a priori in the light of contemporary biology. General systems, 7, 23-35.
Lorenz, K. (1963). Haben Tiere ein subjektives Erleben? Jahrbuch der
Technischen
Hochschule MUnchen.
English translation reprinted in
Studies i!! Animal and Human Behavior. 2. Cambridge. MA: Harvard
University Press.
Lorenz, K. (1973). Die Riickseite des Spiegels. MOOchen: Piper.
Lorenz. K. & Kreuzer, F. (1981). Leben ist Lemen. MUnchen: Piper.
MacKenzie, D.A. & Barnes. B. (1979):-Scientific judgment: The biometryMendelism controversy. In: B. Barnes and S. Shapin (Eds.), The natural order: Historical studies .2! scientific culture. Bever!y"Hills:
Sage.
Malinowski, B. (1944). ~ scientific theory f~. Chapel Hill: University of North Carolina Press. 1973.
Mannheim, K. (1929). Ideologie und Utopie. Frankfurt: Schulte-Bulmke.
Marr, D. (1982). Vision. San Francisco: Freeman.
Martinet, A. (1962). Langue ~ fonction. Dne theorie fonctionnelle .2.!:!
langage. Paris: Gonthier.
Martinez, S. & Tirapegui, E. (1985). Preprint, Facultad de Ciencias, Universitad de Chile, Santiago de Chile.
Maynard Smith, J. (1982). Evolution and the theory .2! games. Cambridge,
U.K.: Cambridge University Press.
Maynard Smith. J. (1984). The evolution of animal intelligence.
In: C.
Hookway (Ed.). Minds, Machines, and Evolution. Cambridge: Cambridge
University Press.
Maynard Smith, J. (1986). The problems .2! biology. Oxford: Oxford University Press.
Mayr, E. (1975). Change of genetic endowment and evolution. In: J.S. Huxley
(Ed.), ~ ~!!. process. London: Allen & Unwin.
Mayr, E. (1976). Evolution and the diversity 2! life. Cambridge, MA:
Belknap Press of Harvard University Press.
Mayl.-, E. (1982). The growth .!. biological thought. Cambridge, MA: Belknap
Press of Harvard University Press.
McArthur, L.Z. & Baron. R.M. (1983). Toward an ecological theory of social
perception. Psychological Review, 90, 215-238.
McClelland. J.L. & Rumelhart, D.E. (1986). Parallel distributed processing:
Explorations i!l the microstructure .2! cognition. (2 vols.) Cambridge,
MA: Bradford Books.
McCloskey, M. (1983). Intuitive physics. Scientific American, 248(4), 114122.
McCulloch, W.S. (1965). Embodiments of mind. Cambridge, MA: MIT Press.
McDougall. K.D. &
McDougall, W. (i93~Insight and foresight in various
animals: Monkey, racoon, rat, and wasp. Journal.QI Comparative ~
chology, 11, 237-273.
McKelvey, B. (1982). Organizational systematics. Berkeley: University of
California Press.
McMullin, E. (1983). Values in science. In: PSA 1982. 2. East Lansing. MI:
Philosophy of Science Association, 3-28.
Merton. R.K. (1973). The sociology.QI~. (N.W. Storer. Ed.). Chicago,
University of Chicago Press.
Merton. R.K. & Gieryn, T.F. (1982). Institutionalized altruism: The case
of the professions. In: R.K. Merton (Ed.), Social research and the
practicing professions. Cambridge, MA: Abt Books.
Lindenberg~
443
Mesarovic. M.D. (1964). Foundations for a General Systems Theory. In: H.D.
Mesarovic. (Ed.), Y!!!!! 2!l General Systems Theory. New York: Wiley J
1-24.
Heyer, F. (1967). IISltuation ~pist~ologique de la biologie."

In: J.
Piaget (Ed.). 108i9ue ~ connaissance scientifigue.
Paris: Galli-
mard, 781-82l.
Miller, A.!. (1984). Imagery!!! scientific thought: Creating 20th-century

physics. Boston: Birkhauser.
Miller, A.I. (1986). Imagery, intuition and strategy in creative scientific
thinking: Albert Einstein's invention of the special theory of relativity. For publication in H.E. Gruber and D. Wallace (Eds.), Case
studies in creativity.
--
Miller-,-D-.-(1972). Ideology and the problem of false consciousness, Political Studies, 20, 432-447.
-Horgan:-c.L~g4). ~ introduction !Q comparative psychology. London:
Walter Scott.
Horin, E. (1969). La rumeur d'Orleans. Paris: Seu!l.
Hulkay, M., Gilb;;t~& Waolgar, S. (1975). Problem areas and research

networks in science. Sociology, 9, 187-203.
Hulkay, G. &; Gilbert, H. (1981). Putting philosophy to work: Karl Popper's
influence on scientific practice. Philosophy.Qf ~ ~ ~,
1l.
Nagel, E.
(1977). Teleology revisited. Journal
2!
Philosophy, 74, 261-301.
Reprinted in his Teleology ~ !!!!! other essays.!!!. ~ Eh!:.

losophy and history .2!!..!.!!!!. New York: Columbia University Press,
1979.
Nagel, T.
(1974). What is it like to be a bat?
Philosophical~, 83,
435-45l.
Nelson, K. (1973). Does the holistic study of behaviour have a future? In:
P.P.G. Bateson and P.H. Klopfer (Eds.) Perspectives .!!! ethology, l.
New York: Plenum, 281-328.
Nelson, R.J. (1984). Naturalizing intentions. Synthese, 61, 173-203.
Nickles, T. (1980). Scientific problems: Three empiricist models. In: PSA
1980. East Lansing, HI: PSA, 3-19.
Nickle;:--T. (1980b). Introductory essay: Scientific discovery and the
future of philosophy of science. In: T. Nickles (Ed.), Scientific
discovery. logic !!U! rationality. Boston-Dordrecht: Reidel.
Nicolls, C. and G. (1984). Is there a climatic attractor? Nature, 311, 529532.
-Nicolis, C. & Prigogine, 1. (1977). Self-organization !!!. nonequilibrium
systems. New York: Wiley.
Oakley, D.A.
(1985). Animal awareness, consciousness and self-image. In:
D.A. Oakley (ed.), Brain and mind. London: Methuen.
Oakley, D.A. &
Gormes, 'L.C':"'" m8~The plurality of consciousness. In:
D.A. Oakley (Ed.), Brain and mind. London: Methuen.
Olding, J. (1978). A defenCe'Ot evolillOnary laws. ~ Journal !2!!!!
Philosophy ~~, 29. 131-143.
Osgood, C.E. & Tannenbaum, P.B. (1955). The principle of congruity in the
prediction of attitude change. PsYchological Review, 62, 42-55.
Pareto, V. (1916). Trattato .Q! sociologia generale. Milano; Comunite, 1964.
Parsons. T. (1959). An approach to the sociology of knowledge. In: !!.!!!!:.
~.2f the Fourth World Congress .2f Sociology. Louvain. 25-49.
Pattee, H.H.
(1973). The physical basis and origin of hierarchical control. In: R.H. Pattee (Ed.). Hierarchy theory. New York: Brazi1ler.
71-108.
Patten, B.C.
(1982). Environs: Relativistic environmental particles for
ecology. American Naturalist. 119, 179-219.
Patterson, X.E. & Baddeley, A.D. (1977). When face recognition fails.
~ .2f Experimental Psychology:
~ Learning!!l!!
Memory. 3.
406-417.
444
Pauli. W. (1979). Wissenschaftlicher Briefwechsel mit Bohr. Einstein.

Heisenberg.!:!.!.!.:..t In: A. Hermann, K. v. Hey;;;;; & V.F. Weisskopf
(Eds.). Berlin: Springer.
Piaget, J. (1918). Recherche. Lausanne: La Concorde.
Piaget, J. (1929). Les deux directions de la pens&e scientifique. ~
des Sciences Physiques !! Naturel1es, 11, 145-162.
Pisget, J. (1947a). !:! psychologie !!! 11 intelligence. Paris: Colin.
Piaget, J. (1947b). Du rapport des sciences avec la philosophie. Synthese,
6, 130-150.
Piaget, J. (1950).
Les deux directions de la pensee scientifique.
In:
Introduction ~ I' 'p18t9010818 2enetigue. 1l. 1! pens'e bio!ogigue .!!.
penses psychologique !! ~ pens e sociologigue.
Paris: P. U. F., 329332.
Piaget. J. (1952).
La logistique axiomatique ou 'pure', la logistique
operatoire ou psychologique et les realites auxquelles elles correspondent. Hethodos, 4(13), 72-85.
Piaget, J. (1954). Inconditionm!s transcendantaux et epistemologie gM'tique. Dialectica, 8, 5-13.
Piaget, J. (1955).
Les Hgnes g~nerales de 1'-'pistemologie gen~tique.
Actes ~ congres international 4! l'Union internationale.!!! philosoI!hl!! ~ ~ (ZUrich, 1954),!. Neuchltel: Editions du Griffon,
26-45.
Piaget. J. (1962). Play, dreams and ~.!!l childhood. Translated by
C. Gatagno and F .M. Hodgson. Ne.w York: Norton.
Piaget, J. (1963). L'explication en psychologie et Ie parallelisms psychophysiologique. In: P. Fraisse & J. Piaget (Eds.), Traite.!!!'!y':'
chologie experimentale. !l !!!.!!2!!!!! methode.
Paris: P.U.F., I
123 I-162.
Piaget, J. (1964).
Classification of disciplines and interdisciplinary
connections. International Social Sciences Journal, 16. 553-570.
Piaget, J. (1966). La psychologie, les ~s interdisciplinaires et Ie
syst~e des sciences. ~.!!!. Psychologie,
20, 1-13.
Piaget, J. (1967a).
Epistemologie de la logique.
In: J. Piaget (Ed.),
Logigue .!U. connaissance scientifigue. Paris: Gallimard, 375-399.
Piaget, J. (1967b). L'interpretation biologique des trois formes de la
connaissance. In: Biologie !! connaissance. Paris: Gallimard, 307-395.
Piaget. J. (l967c). Le systeme et la classification des sciences. In: J.
Piaget (Ed.), Logigue ~ connaissance scientifigue. Paris: Gallimard,
1151-1224.
Piaget. J. (1967d). Les courants de l'epistemologie scientifique contemporaine. In: J.
Piaget (Ed.), L08igue ~ connaissance scientifigue.
Paris: Gallimard, 1225-127!.
Piaget, J. (1967e/f). Les deux problemes principaux de llepistemologle
hiologique.
In: J. Piaget (Ed.), Logigue .!U. connaissance scientifi~.
Paris: Gallimard, 893-923 (e), 1114-1146 (f).
Piaget, J. (1967g). Les problemes principaux de 11 epistemologie des mathematiques. In: J.Piaget (Ed.), Logigue ~ connaissance scientifigue.
Paris: Gallimard, 554-596.
Piaget. J. (1968). Le structuralisme. Paris: P.U.F.
Piaget, J. (1970a). L'epistemologie geDf!tigue. Paris: P.U.F.
Piaget. J. (1970b). Structuralism. Translated by C. Haschler. New York:
Basic Books.
Piaget, J. (1971).
Hasard et dialectique en epistemologie biologique:
Examen critique des theses de Jacques Honod. ~ (Paris). 7~,
29-36.
Piaget. J. (1973). Epistemologie des relations interdisciplinaires. UniInformation (Geneve), 31, 4-8.
Piaget, J. (1974).
Adaptation vitale !! psychologie.!!! l'intelligence:
selection organique !!!. phenocopie. Paris: Hermann.
Piaget. J.
(1976). b!. comportement. ~.!!! 1 'evolution. Paris: Gallimard.
445
Piaget, J. (1977a). L'eplstemo!ogie des regulations. In: A. Lichnerowicz,

F. Perroux & G. Gadoffre (&ds.), L'1dee !k regulation dans las sciences. Paris: Halaine/Doin, I-XIII.
Piaget:--J. (l977b). Phenocopy in biology and the psychological development
of knowledge. In: H.E. Gruber & J.J. Voneche (Eds.), The essential
Piaget. New York: Basic Books, 803-813.

Piaget, J. (1982). Functions
and structures of
Plotkin (Ed.),
145-150.
Learnins,
adaptation.
development!!!!!~.
In:
H.C.
New York: Wiley,
Piaget, J., et al. (1977). Recherches!!!!. l' abstraction reflechissante.

Paris: PUF.
Piaget, J. & Garcia, R. (1983). Psychogen~se ~ ~ ~~.
Paris: Flammarion.
Piatelli-Palmarini, M. (Ed.) (1979).
Th';ories.!!!:! langage, th'ories!!!.
I' apprentissage. Paris: Seuil.
Pickering, A. (1980). The role of interests In high energy physics: The
choice between charm and color. In: K. Knorr, R. Krolm & R. Whitley
(Eds.): The social process of scientific investigation. Sociology 2f
the Sciences Yearbook, 4. Dordrecht: Reidel.
Pinard:-A.~ ~ervation !. conservation. Chicago: University of
Chicago Press.
Pittendrigh, C.S. (1958). Adaptation, natural selection, and behavior. In:
A.
Roe & G. Simpson (Eds.), Behavior and evolution. New Haven: Yale
University Press.
Pollock, J.L.
(1985). My brother, the machine. Unpublished manuscript.
Tucson: University of Arizona.
Pomian, K. (1984). L'ordre ~ temps. Paris: Gallimard.
Popper, K.R. (1960). Th!. poverty 2f historicism. London: Routledge and
Kegan, 2nd ed.
Popper f K. R. (1982). The ~ lUli verse. London: Hutchinson.
Popper, K.R. & Lorenz, K. (1985). !!.!! ~ .!!!.2!!!!!.. Miinchen: Piper,
29-31.
Premack, D. & Woodruff, G. (1978). Does the chimpanzee have a theory of
mind? Behavioral and Brain Sciences, I, 515-526.
Prigogine, 1., George,C.:--U;n~& Rosenfeld, L. (1973). A unified
formulation of dynamics and thermodynamics, Chemica Scripta, 4, 5-32.
Prigogine, 1. & Stengers, 1. (1979). Les Deux Cultures aujourd'hui, b!
~ ~ Francaise, 42-59.
Prigogine, 1. & Stengers. 1. (1984). Order !!! 2f chaos. New York: Wiley.
Provine, W. (1971). Origins!!! theoretical population genetics. Chicago:
University of Chicago Press.
Putnam. H. (1981). Reason, truth and history. Cambridge: Cambridge University Press.
Putnam. H. (1983). ~ and~. Cambridge: Cambridge University
Press.
Quine, W.V. (1951/1963). Two dogmas of empiricism. Philosophical~,
60, 20-43. Reprinted in his From! logical point .2! Y.!.!!!.. New York:
Harper Torchbooks. 1963, 20-46.
Quine, W.V.O. (1969a). Epistemology naturalized. In: Ontological relativity
and other essays. New York: Columbia University Press, 69-90.
Quine, W.V.O. (1969b). Natural kinds. In: W.V. Quine. Ontological .!!l!:.
tivity. New York: Columbia University Press. 114-138.
Quine, W.V.O. (1975). The nature of natural knowledge. In: S. Guttenplan
(Ed.), Mind !ru! language. Oxford: Clarendon.
Rescher, N. (1983). Extra-terrestrial science. Paper presented to the 7th
International Congress of Logic, Methodology and Philosophy of Science. University of Salzburg, July 1983.
Ri benboim, P.
(1979). 11 ~ .2!! Fermat I s last theorem. Heidelberg:
Springer.
446
Richel1e, M.N.
(1986). Variation and selection: The evolutionary analogy
in Skinner's theory.
In: S.
Hodgil and C.
Hodgil (Rds.) !:.!.:..
~ ~
Radinson,
H.
(1968).
64, 70-104.
and controversy.
Brighton: Falmer Press.
Sociologie marxiste et ideologie marxiste. Diogene,
Roitblat, H.L.
(1982). The meaning of representation in animal memory.
Behavioral and Brain Sciences, 5, 353406.
Roitblat, H.L., Bever:--T~ Terrace, B.S.
(Eds.) (1984). ~
Cognition. Hillsdale, N.J: Lawrence Erlbaum.
Roll-Hansen, N. (1983). The death of spontaneous generation and the birth
of the gene: Two case studies of relativism. 2!!! Studies 2f i:.
ence, 13, 481-519.
Romane;:--G.J. (1883). Mental evolution in animals. London: Kegan Paul
Trench.
- - ---- - --Rosenberg, A. (1980). Obstacles to the nomological connection of reasons
and actions. Philosophy.2! Social ~, 10, 79-91.
Rosenberg, A. (1985). The ~ .2! biological ~. Cambridge. U.K.:
Cambridge University Press.
Rosenberg, H. J .A. (1960). An analysis of affective-cognitive consistency.
In: C.L Hovland & H.J. Rosenberg (Eds.), Attitude organization and
change. New Haven: Yale University Press, 15-64.
Rouquette, H.L. (1975). 1!.!!!!!!!!!!!. Paris: P.U.F.
RWlC iman , W.G. (1969). The sociological explanation of 'religious beliefs'.
~ Europeennes ~ Sociologie, 10, 149-191.
Sachsse,
H.
(1979).
KausaliUlt-Gesetzlichkeit-Wahrscheinlichkeit.
Darmstadt.
Sahlins, M. (1976). The ~ and ~ 2! biology. London: Tavistock.
Samuelson, P.A. (19m. Maximum principles in analytical economics. ~
these, 31. 323-344.
Schife;:-1i. (1983). Normative finalization. In: W. Schlfer (Ed.), Finalization !!! ~ .TI!!.!2.i!! orientation .2! scientific progress.
Dordrecht & Boston: Reidel, 207-231.
Schaff, A. (1967). La dlHinition fonctionnelle de l'ideologie. L'Homme ~
la societe.
SchilPP':' ~(Ed.) (1949). Albert ~ philosopher-scientist. La
Salle. IL, Open Court. (third edition, 1969),683-4.
Schilpp, P.A. (Ed.) (1974). Th! philosophy 2!!!!:! Popper. La Salle, IL.
Open Court.
Schrodinger, E. (1961). Meine Weltansicht.
Hamburg-Wien: Paul Zsolnay.
English translation: !:!v !!2ili Y!!!!. Cambridge: Cambridge University
Press. Foreword to the 1960 edition.
Schon, D.A. (1963). Displacement .2! concepts. London: Tavistock.
Seyfarth, R., Cheney. D.L. & Harler, P. (1980). Honkey responses to
three different alarm calls: Evidence of predator classification and
semantic communication. Science, 210, 801-803.
Shallice. T.
(1972). Dual fWlctions of consciousness. Psychological ~
view, 79, 383-393.
Shaper~D. (1984). ~ and the !!!h !.2!: knowledge: Investigations i!2
the philosophY 2!.~. Dordrecht: Reidel.
Shapere, D. (1987). Hethod in the philosophy of science and epistemology:
How to inquire about inquiry and knowledge. In: N. Nersessian (Ed.),
.!!!! processes 91.~. The Hague: Nijhoff (to appear).
Sheldrake. R. (1982). A new science of life. London; Blond and Briggs.
Shweder, R.A. (1986)-:- Divergent r;t'ionalities. In: O. W. Fiske and R.A.
Shweder (Eds.), Hetatheory i!2 social science: P1uralisms and subiec~. Chicago: University of Chicago Press, 163-196.
Silverman. H.J. (Ed.) (1980). Piaget. philosophy and the human~.
Atlantic Highlands, N.J.: Humanities Press.
Silverman, P.S.
(1983). Attributing mind to animals: the role of intuition. Journal g! ~ !!!.2. Biological Structures. 6. 231-247.
447
H.A. (1973). The organization of complex systems. In: H.H. Pattee

(Ed.). Hierarchy Theory. New York: Braziller, 1-27.
Simon, H.A. (1979). Rational decision making in business organizations.
American Economic Review, 69, 493-513.
Simon.~983). Reason in human affairs. Stanford, CA: Stanford University Press.
- - - -- --Simons, H.W . Berkowitz. N.N. & Moyer, R.J. (1970). Similarity, credibility
and attitude change: a review and a theory, Psychological~.
73. 1-16.
Smelser, N.J. (1962). Theory 2! collective behavior. London: Routledge &
Kegan Pau!'
Smullyan, R.M. (1983). 5.000 ~ and other philosophical ~. New
York: St. Martin's Press.
Sober. E. (1981). Holism, individualism. and the W1its of selection. In:
PSA 1980. 2. East Lansing~ MI: Philosophy of Science Association, 93Simon~
W.--
Sober, E.
(1984). The nature !. selection: Evolutionary theory i!! philosophical focus. Cambridge, MA: MIT Press.
Sober, E. (Ed.) (1984). Conceptual issues 1!! evolutionary biology. Cambridge, MA: MIT Press.
Sommerhoff. G. (1950). Analytical biology. Oxford: Oxford University Press.
Sommerhoff, G. (1969). The abstract characteristics of living systems. In:
F .E. Emery (Ed.). Systems Thinking. Baltimore: Penguin.
Sperry, R.W. (1969). A modified concept of consciousness. Psychological
Review, 76, 532-536.
Spiro.~ (1966). Religion: problems of definition and explanation. In:
M. Banton (Ed.). Anthropological approaches 12 the study f religion.
London: Tavistock~ 85-126.
Staddon, J.E.R. (1983). Adaptive ~ and learning. Cambridge: Cambridge University Press.
Steen, L.A.
(Ed.) (1978). Mathematics today: Twelve informal essays. Heidelberg: Springer.
Stroud. B.
(1984). The significance ! philosophical scepticism. Oxford:
Oxford University Press.
Symons, D. (1980). Precis of ~ evolution .2! human sexuality (Author's
response). Behavioral and Brain Sciences, 3, 171-214.
Tajfel, H. (1981). Human groups and social categories: ~ i!! ~
psychology. New York: Cambridge University Press.
Tennant. N. (1984). Intentionality and the evolution of language. In: C.
Hookway (Ed.). Minds, machines, and evolution. Cambridge: Cambridge
University Press.
Terrace, H.S.
(1984). Animal cognition. In: H.L. Roitblat, T.G. Bever &
H.S. Terrace (Eds.). Animal cognition. Hillsdale, N.J.: Lawrence
Erlbaum.
Thoday, J .M.
(1975). Neo-Darwinian "evolution" and biological progress.
Nature, 255, 675-677.
Thorn, ~972). Stabilite structurelle II morphogenese. Paris: Intereditions.
Thom, R.
(1974). Modedes mathematigues Q! ~ morphogenese. Paris: U.G.E.
(lD/1S) .
(1975). Dlun modele de la science une science des modeles. ~
Thom, R.
these, 31, 359-374.
Thom, ~1983a). Paraboles ~ catastrophes. Paris: Flammarion.
(Ital.
orig. 1980).
Thorn. R. (1983b). Animal psychism vs human psychism. In: E. de Grolier
(Ed.), Glossogenetics. Paris: Harwood.
Thom. R. (1983c). Semiotique et theorie des catastrophes, Structures cycliques en semiotique. Groupe de recherches semio-linguistiques de
l'EHESS, 10 rue Monsieur Ie Prince, Paris 75006, V47-48. 38-58.
Thompson. M. (1981). Things in themselves. Proceedings.2! the American
Philosophical Association, 57(1).
448
Thompson, P. (1986). The interaction of theories and the semantic conception of evolutionary theory. Philosophica, 36, 73-86.
Tolman, E.C. (1926). A behavioristic theory of ideas. Psychological ~,
33, 352-369.
Tolman, B.C. (1951). Collected papers .!!! psychology. Berkeley: University
of California Press.
Tombaugh, J. (1984). Evaluation of an international scientific computerbased conference. Journal of Social Issues, 40, 129-144.
Toulmin, S.E. (1970). Do~dt;t~n between normal and revolutionary
science hold water? In: 1. Lakatos & A. Musgrave (Eds. >. Criticism
and the growth 2! knowledge. Cambridge: Cambridge University Press,
39-47.
Toulmin, S.E. & Goodfield, J. (1965). Th! discovery.2! l!!!!!. New York:
Harper & Row.
Trivers, R.L. (1971). The evolution of reciprocal altruism. Quarterly
Review 2i Biology, 46, 35-57.
Tversky, A. & lCahnemann, D. (1981). The framing of decisions and the psychology of choice. ~, 211, 453-8.
Ullman. S. (1979). TI!! interpretation 2i Y.!.!!!!! !!2.1!.2!l. New York.
Vandenbrande, R. (1979). La critique litt'raire: Wl espace interme.diaire.
In: W. Callebaut et al. (Rds.). Theory 2i knowledge ~!.i!!!!. policy.
Ghent: Communication & Cognition, 183-189.
van Fraassen, B.C. (1980). The scientific image. Oxford: Clarendon.
Van Parijs, P. (1978). Theorie des catastrophes et mate.rialisme historique,
Revue Francaise 4! Sociologie, 19, 195-220.
Van Parijs, P. (1982). FWlctionalist Marxism rehabilitated. A comment on
Vollme!~st~~. (~:~)7 ~~ts::!e!v~1!!io::;;5!!istemology.

& J.
In: P. Weingartner
Czermak (Bds.), Epistemology and Philosophy .!?!.~. Vienna:
H51der-Pichler-Tempsky, 185-197.
Voneche, J. (1984). Introduction. In: Fondation Archives Jean Piaget (Ed.),
Phylogeny and ontogeny. Human Development, 27 J 227-232.
von Glasersfeld, E. (1974). Piaget and the radical constructivist epistemology. Paper presented at the Third Southeastern Conference of the
Society for Research on Child Development. Chapel Hill, N.C.
Vrba, B.S. & Eldredge, N. (1984). Individuals, hierarchies, and processes:
Towards a more complete evolutionary theory. Paleobiology, 10, 146171.
Waddington, C.H. (1961). The human evolutionary system. In: M. Banton
(Ed.), Darwinism and the study B!. SOCiety. Chicago: Quadrangle Books,
63-82.
Waddington, C.H. (1969). The practical consequences of metaphysical beliefs
on a biologist's work. In: C.H. Waddington (Ed.). ~! ~
.!!!! Biology. 2. Edinburgh: Edinburgh University Press, 72-81.
Walker, s. (1983). Animal thought. London: Routledge & Kegan Paul.
Watson, J. (1966). Crowing up in the Phage group. In: Cairns et a1. (Eds.).
Watson, J. (1968). The double helix. New York.
Whitley, R. (1978)":"'""' Types of science, organizational strategies and patterns of work in research laboratories in different scientific
fields. Social Science Information, 17, 427-447.
Weber, R. (19~Dialogues with scientists ~ sages: The.!.!!!.h for
unity. London: Routledge & Kegan Paul.
Wertheimer, M. (1982). Productive thinking. 2nd ed. Chicago: Chicago Universi ty Press.
Wicken, J.S. (1984). On the increase in complexity in evolution. In: M.W.
Ho & P. Saunders (Eds.), Beyond neo-Darwinism. New York: Academic
Press, 89-112.
Wicken, J.5. (1985). An organismic critique of molecular Darwinism. Journal
2! Theoretical Biology, 117, 545-561.
Williams, C.C. (1966). Adaptation .!m! !l!E!r!!~. New Jersey:
Princeton University Press.
449
Williams, M. (1973). Falsifiable predictions of evolutionary theory. Philosophy .2! Science~ 40, 518-537.
Williams, H. "(""i"9ii') The importance of prediction testing in evolutionary
biology. Erkanntnis, 17, 291-306.
Wilson. H.O. (975). Sociobiology: The!!!!! synthesis. Cambridge, MA;
Harvard University Press.
Wimsatt, W.C. (1981). Robustness, reliability, and multiple-determination
in science. In: M. Brewer & B. Collins (Eds.), Knowing !!!!! validating
in the social sciences. San Francisco: Jossey-Bass, 124-163.
Wimsatt, W.C:-ri'9ii6b). Developmental constraints, generative entrenchment,

and the innate-acquired distinction. In: W. Bechtel (Ed.), Integratins scientific disciplines. Dordrecht: Nijhoff, 185-207.
-Winter. S.G. Jr. (1975). Optimization and evolution in the theory of the
firm.. In: R.H. Day & T. Groves (Rds.), Adaptive ~~. New
York: Academic Press, 73-118.
Woodfield, A. (1976). Teleology. Cambridge, U.K.: Cambridge University
Press.
Woodward. S. (1980). Developmental explanations. Synthese. 44.
Worsley. P. (1957). The ~ shall sound: ! study Q!. 'cargo cults' !!!.
Melanesia. London: Paladin, 1970.
Wright:-s:-T!934). The method of path coefficients. Annals Q!. Mathematical
Statistics, 5. 161-215.
Wright. S. (1980). Genetic and organismic selection. Evolution. 34, 825843.
Zeeman, E.C. (1977). Catastrophe theory. Reading, KA.: Addison-Wesley.
Zuckerman, H. (1977). Deviant behavior and social control in science. In:
E. Sagarin (Ed.). Deviance and social change. Sage Annual Reviews of
Studies in Deviance, 1. Beverly Bills: Sage.
Index of Names
Ackermann 388
Adorno 388
Allen 72
Allport 388,394,401
Althusser 387,388,395
Alston 142
Apostel 3,9,22,23,40,265,319
Aristotle 52,59
Armon 281
Arms trong 109
Arnheim 277
Arthur 19
Ashby 146,157
134
Beth 249
Black 123,187
Bloor 159,399
Boden 7,16,129,266
Bohm 77,78,79,91,92,93,95
Bohr 267,268,270,271,272,273,
274,275,276,277,278,279,
317,318,319
Boltzmann 65,207
Boon 18,24,34,37,44,221
Borel 259
Born 65,271,276
Boudon 388
Boulding 24,25,26
Bourdieu 388,391,395,401
Boyd 21,25,30,32,41,47,144,
153,154,384,385,399
Bradie 3,4,8,17,23,46,54,105
Brandon 8,9,13,14,18,19,47,52,
234
Branningan 185,186
Brecht 5,6,50
Brewer 31,155,239
Brooks 11,33
Brown 378
Biichel 15
Bunge 11,19
Burghardt 131,132,134
Burian 18,29,47,234
Burks 141
Buscaglia 257,262
Berry
Babloyantz 64
Bacon 160,207
Baddeley 241
Baldwin 146,157,265,314
Balibar 388
Banach 363
Barash 235
Barlow 342
Barnes 54,145,399
Baron 234,401
Barthes 395
Bateson 145,168
Battali 322
Beausoleil 286
Bechtel 19,39
Beckermann 235,238
Benard 61
Berger 399
Bergmann 10,371
Bergson 59,99
Berkeley 150,224
Berkowitz 386
Bernouilli 344,345,347,360,
361
Bertalanffy 7
Cairns 161,170
Callebaut 8,11,16,19,23,26,29
34,37,39,40,46,51
53,54,55,139,226,227
283,363
451
INDEX OF NAMES
452
1,4,5,9,15,17,18,20
21,22,24,25,27,30,31
32,34,35,36,37,38,39
40,41,43,47,48,53,55
75,81,82,85,86,94,
105,106,113,139,143
144,145,146,147,148
149,152,153,155,156
165,179,180,182,183
189,193,200,204,205
206,223,225,229,230
234,235,236,237,239
240,265,311,312,313
316,317,322,365
Cantor 339,351
Caplan 234
Caporael 24,235,236
Capra 7
Carloye 116,131
Carrier 30
Cartwright 14
CelIerier 264
Chait in 219
Chamboredon 388
Changeux 87
Chardin,de 220
Cheney 117
Cherniak 39
Chernoff 241
Childe 144
Chisholm 142
Chomsky 126,262
Church 34,251
Churchland 25,110,119,121,122,
140
Clark 25,35,51
Clausius 60
Clement 219,378
Cohen 143,221,395
Collins 31
Campbell
Commons 281
Copernicus 73,160,165
Corning 182,189,190,191,200
Cracraft 44
Craik 15
Crick 170,172
Crombie 150
Culbertson 236
Dallo 11,19
Danto 38
Darbishire 145
D'Arcy Thompson 44
Darwin 12,15,73,111,189,212,
374,375,376
Davis 363
Dawkins 19,48,49,53
DelbrUck 161,169
Denbigh 65
Deneubourg 72
Dennett 38,107,115,116,118,119
120,121,123,124,126,
128,129,130,132,135,
140,227
Depew 6,11,33
Descartes 6,17,51,160,223,224,
229
De Waele 26,144
Dirichlet 340
Dobzhansky 239
Donchin 113
Donini 390
Donohue 172
Dretske 30,80,128
Duhem 156
Dyson 266
Eames 132
Eddington 65,278
Edge 54
INDEX OF NAMES
Edwards 339
Eigen 7
Einstein 58,59,99,203,208,
267,272,315,366
Ekeland 366,367,368,369
Eldredge 13,44,79,90,191,
238
Ellis 141
Elster 10,36,52,120,135,197,
387,397,401
Engelen 72
Engels,E. 30
Engels,F. 390
Euclid 51
Euler 337,357,359
Faltings 348,351
Felson 112
Fermat 337-363
Festinger 393
Feyerabend 159,160,180,182
Feymnann 101
Fisher 53
Fiske 113
Flavell 253
Flewer 34
Fodor 263
Foucault 180
Fouille 146
Fox 51
Freeman-Mayer 265
Frege 100
Freud 12,73,366,388,390,393
Furth 220
Furtwangler 348,355
Galambos 113
Galileo 51,100
Gamble 147
Garcia 130
Gauss 341,342,344,353
453
Geertz 179
Georgescu-Roegen 30
Germain 342,343,344,354
Ghiselin 35,78,83
Gibbs 68
Giere 3,200,313,363
Gieryn 152
Gilbert 173,193
Gillieron 23,256,311,313,314
315
Ginsberg 144
Glasersfeld,von 264
Gmelch 112
Godel 219,251,349,350
Golbach 363
Goldmann 142,143,387
Goldschmidt 44
Goodfield 10
Goodwin 47,265,266,313
Gould 13,14,44,148,191,236
237,314
Gray 7
Greco 256
Greimas 102
Grene 8,52,78,90
Griesemer
Griffin
17
107,108,109,110,111
112,113,114,115,116
117,123,128,130,131
Grimes 378
Gruber 247,249
GrUnert 238,340
Haldane 83,224
Hamlyn 224,228
Hansell 108
Harnad 114
Haroutunian 23,42
Harre 11,27,150,151
Hebb 320,321
Heidegger 59
454
INDEX OF NAMES
Heider 112,122
Heisenberg 268,270,274,275,
276,277,278,281,
318
Hempel 13
Henderson
45
Henriques 265
Hesse 187,200
Heyes 16,30,47,55,139
Hilbert 350
Hillyard 113
Hishleifer 25
Ho 16,30,44,45,46,48
Holldobler 235
Holmes 235
Holton 103
Hooker 25
Howe 286
Hubel 307
Hull 10,13,15,16,19,37,43,
44,49,50,52,54,82,133
144,152,154,180,182,
183,187,201,371
Hume 99,150,151,224
Humphrey 110,112
lngvar 64,65
Inhelder 284
Irrgang 17
It telson 150
190
Jacobs 15
Jahoda 388
Jakobson 375
Jantsch 6
Jarvie 390,391
Jaynes
112
Jenning
146
Jacob
Jensen
11
Kahnemann 218,219,243
Kant 17,20,21,39,54,224,228
Kary 143
Kaspar 20,219
Kaufman 8
Kellert 134
Kepler 150,160,165,368,369
Kimura 44
Kitchener 366,370,371,372,373
376,378,379
Kitcher 11
Kleene 251
Knorr-Cetina 3,27,34,37,40,44,
185,188,194,195,
221
Koelling 130
Kolmogorov 67
Kornblith 38
Kramers 275,319
Krasner 347,348,355
Kreuzer
203
Kroodsma 130
Kuhn 15,27,148,159,165,169,
180,192,197,212
Kummer 344,345,346,347,348,
355
Kurland 235,238
Ladriere 252
Lakatos 28,160,166,169,337,
338,351,357,358,363
Lamarck 12,264,374,375,376
Lame 340
Lamontagne 286,307,311,320-322
Landau 238
Lang 387,388
Laudan 154,161
Lavoisier
207
Le Doux 132
Legendre 343
455
INDEX OF NAMES
Leibniz 17,224
Lelas 54
Levine 10,11,12
Levins 41
Levinson 35
Lewin 144
Lewis 65,112
Lewontin 12,13,18,19,40,41,
42,43,45,53,81,82,
148
Lindenberg 8
Lloyd 18,49
Locke 150
Lorenz 4,20,21,22,40,47,48,
54,80,84,94,110,140,
203,204,205,206,211,
213,221,229,262,322
Losee 221
Lovejoy 45
Luckman 399
Lukaszweski 239
Lyapunov 67
MacKenzie 145
Malinowski 390
Mandelbrot 63
Mannheim 387
Marler 117
Marr 120
Martinet 373-376
Martinez 70
Marx 390
Maturana
54
Maxwell 207
Maynard-Smith 50,88,90,127
Mayr 11,19,26,46,48,52,191
McArthur 234
McClelland 51
McCloskey 219
McCulloch 308
McDougall III
McKelvey 52
McMullin 28
Mendel 162,168,170,186
Mersenne 347
Merton 152,180
Mesarovic 15
Meyer 257
Michener 235
Miller 130,281,311,315,316
317,318,319,320,322
387
Mirimanoff 347,348,355
Monod 45,262
Mordell 348
Morgan,T.H. 145,170
Morgan,W. 111,123
Morin 388
Moyer 207,386
Mulkay 187,193
Mulkey 173
Milller 170
Nagel 5,14,134,189
Nelson,R.R. 27,38,91,384
Neurath 51
Newton 53,58,208,361
Nickles 53,194,200,353,363
Nicolis 32,64
Nivnik 247
Oakley 109,132
Odling-Smee 21,22,24,32,48,85,
86,94
Oeser 36,37,38
O'Hear 143,229
Olding 371
Oppenheim 8,13
Orwell 240
Osgood 393
Paller
25
456
Palmarini 262
Papert 308
Pareto 388,401
Parsons 198,387
Pascal 150
Passeron 388,401
Pasteels 72
Pattee 91,92
Patten 82
Patterson 241
Pauli 277
Peano 353
Pearson 145
Peirce 6,7
Piaget 5,6,12,20,22-24,
42,44,47,80,82,
142,149,219,220,
247-267,269,270,
273,275,279,280,
283-285,309,312,
313,314,316,318,
319,320,322,369,
370
Piatelli 262
Pickering 180
Pike 200
Pinard 283,309
Pinxten 53,54,226,227,283
Planck 58
Plotkin 3,9,14,17,19,20,21,
22,23,24,30,32,47,
48,54,79,80,85,94,
140
Poincare 67,147,367
Polanyi 27
Pollock 110
Pomian 98
Popper 5,25,33,38,39,40,97,
98,114,163,203,204,
205,206,213,366,371
Postman 401
INDEX OF NAMES
Premack 135
Prigogine 5,6,10,11,17,32,33,
35,43,45,46,58,70,
73,366
Provine 145,168
Ptolemy 160
Putman 38,134,224,226
Quine
4,21,112,136,140,141,156
225,227,230
Rescher
27,28,29,38,140,143,
197,229
Ribenboim 360
Richards,F.D. 281
Richards,R. 34,38,46,144,146
Richardson 34
Richelle 87
Richerson 21,25,30,32,41,47
144,153,154,384,
385,399,400
Riedl 20,48,219
Riemann 345
Rizzo 7
Roitblat 125,128
Roll-Hansen 145
Romanes 108,111
Rorty 224,226,227
Rosenberg 16,121,122,146,393
Rouquette 388,394,401
Rumelhart 51
Runciman 400
Ruse 45
Rutherford 270,271,274,317
Sachsse 15
Sahlins 127
Salazar 64
Salthe 79,90
Samuelson 15,52
Sanglier 72
INDEX OF NAMES
Saunders
5,16,30,44,45,46,
48
Schafer 55
Schaff 41,387,399
Schell 26,144
Schilcher,von 19,35,36
Schilpp 205,217
Schon 187
Schrodinger 57,58,72,318
Schuster 7
Seyfarth 117
Shapere 28
Shallice 114
Shannon 374
Sheldrake 47
Sherman 235
Shimony 17,140
Shweder 156
Siegel 146
Silverman 131,132-247
Simmel 112,123,200
Simon 10,29,34,39,41,42,52
90,93,149,231
Simons 386
Slovic 243
Smelser 388
Smullyan 363
Sober 10,11,12,13,18,19,21
47,49,52,140,141,234
Solla Price,de 174
Sommerhoff 84,120
Sosa 141
Spencer 20,376
Spinoza 51
Staddon 87
Steen 363
Stengers 6,11,32,43,45,
58,73
Stroud 224,230
Symons 235,242
457
Tannenbaum 343
Tarski 364
Tattersall 238
Taylor 360
Tennant 19,35,36,134,229
Terrace 126
Thagard 19,43,46,146,221
Thoday 46
Thorn 6,11,14,17,20,32,33,35,
102,104,366,367,368,369
Thompson 11,225
Thorndike 111
Tirapegui 70
Tolman 111,125
Tombaugh 242
Toulmin 10,11,23,27,28,38,144
181,182,183,192,193,
200,204,205,221
Trivers 234
Turing 219,349
Tversky 218,219,243
Ullman
129
Van Bendegem 51,221

Vanden Brande 17
Vandiver 348
Van Fraassen 25
Van Parijs 11,12,17,52,381,
382,384,397,400
Varela 54
Verhaeghe 72
Vinci,da 150
Vrba 90,236,237
Vollmer 4,5,17,20,23,38,
39,54,105,140,216,
219,230
Voneche 264
Waddington
22,23,42,44,47,
77,78,144,263
INDEX OF NAMES
458
Wagstaff 361
Walker 126,134
Watson 112,170,172
Weaver
10
Weber,B.H. 6
Weber,M. 197
Weber,R. 6
Weldon 145
Wertheimer 315
Wheeler 72
Whitehead 103
Whitley 200
Wicken 7,10,33
Wieferich 346,347,348,355,
361
Wiener 374
Wiley 11,33
Williams 11,19,48,51,83
Williston 370
Wilson 235
Wimsatt 7,10,14,18,19,23,29
30,33,49,52,79,113
314,356,363
Winter 26,27,384
Wittgenstein 228
Woodfield 15
Woodruff 135
Woodward 370
Woolgar 193
Worsley 391
Wright 190,195
Wuketits 20,21,140
Wynne-Edwards 49
Yilmar 140
Yuley 378
Zahar
160
Zamiatin
240
Zeeman 400
Ziff 374
Zuckerman
152

Werner Callebaut - Evolutionary Epistemology

Загружено:

Сведения о документе

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

Werner Callebaut - Evolutionary Epistemology

Загружено:

Авторское право:

Доступные форматы

EVOLUTIONARY EPISTEMOLOGY

D. REIDEL PUBLISHING COMPANY

ACADEMIC PUBLISHERS GROUP

DORDRECHT / BOSTON / l.ANCASTER / rOKYO

Library of Congress Cataloging in Publication Dat.

Published by D. Reidel Publishing Company,

In all other countries, sold and distributed

P.O. Box 322, 3300 AH Dordrecht, Holland.

All Rights Reserved

1987 by D. Reidel Publishing Company, Dordrecht, Holland

PART II: EVOLUTIONARY APPROACHES

PART III: THE PIAGETIAN APPROACH

Is Piaget's "Genetic Epistemology"

The Genesis of Atomic Physics and the

PART IV: EXTENSIONS AND APPLICATIONS

DONALD T. CAMPBELL, CECILIA M. HEYES,

more or less wild and more

interesting speculations into a theoretically sound and empivii

multi- or even interdisciplinary yet scien-

!!!! research program. A number of first-rate scientists had

accepted to join our project; here was our chance to prove

could say, is basically about the metaphysical underpinnings

scientific and other. He argues that convergence is

between the corrigible

criticizes the "erroneous emphasis on continu-

ity" in the intellectual genealogies used by historiographers

of science taking an evolutionary view, as well as the exces-

evolutionary account of mind can be reconciled with a

spondence view of truth (yes according to Campbell, no

two evolutionary epistemology programs mentioned before - the

Knorr Cetina and Boon assess

epistemology as a theory of science from the perspective of

that "A conception of

blind, hierarchically structured selection process no longer

pace Campbell, that "discovery in

science may be better understood as an editorial process

part II, ventures to explain why it is that it seems easier

plausible history of the various ways in which mathematicians

some of his own views

development and evolution. In the final paper Philippe Van

view on evolutionary explanations to the explanation

beliefs. He thus hopes "to go some way beyond the (pretty

items have been marked with an

they are all to be found in the second bibliography.

EVOLUTIONARY EPISTEMOLOGY TODAY:

the nineteenth century,

an imposing array of studies somehow relevant to EE has

solved by the clever application of EE;

an exemplar on which one could model further solutions of

tigation of cognition (broadly

from-equilibrium thermodynamics in particular);

W. Caiiebaul and R. PinXlen (edJ.). Rvo/ulwnary EpistemoloRY, 3-55.

w. CALLEBAUT AND R. PINXTEN

It seems imperative to emphasize at the very outset

EVOLUTIONARY EPISTEMOLOGY TODAY

Next (section 2) we will inquire into the concepts of

approach will be contrasted with Piaget's peculiar brand of

tree," Bertold Brecht declared in his fragmentary Notizen

W. CALLEBAUT AND R. PINXTEN

"life itself is a knowledge process. I recognize a tree

Time and again,

the decline of the "Greek",

or "Newtonian", or "Cartesian" approach to science (2), and

EVOLUTIONARY EPISTEMOLOGY TODAY