Questioning Ninetheenth-Century Assumptions About Knowledge - II

Questioning Nineteenth-Century Assumptions about Knowledge

REDUCTIONISM

Edited by

Richard E. Lee Foreword by Immanuel Wallerstein

Questioning Nineteenth-Century Assumptions about Knowledge II REDUCTIONISM

A Symposium coordinated by Aviv Bergman, Jean-Pierre Dupuy, and Immanuel Wallerstein

FERNAND BRAUDEL CENTER STUDIES IN HISTORICAL SOCIAL SCIENCE

Series Editor: Richard E. Lee The Fernand Braudel Center Studies in Historical Social Science will publish works that address theoretical and empirical questions produced by scholars in or through the Fernand Braudel Center or who share its approach and concerns. It specifically seeks to promote works that contribute to the development of the world-systems perspective engaging a holistic and relational vision of the world—the modern world-system—implicit in historical social science, which at once takes into consideration structures (long-term regularities) and change (history). With the intellectual boundaries within the sciences/ social sciences/humanities structure collapsing in the work scholars actually do, this series will offer a venue for a wide range of research that confronts the dilemmas of producing relevant accounts of historical processes in the context of the rapidly changing structures of both the social and academic world. The series will include monographs, colloquia, and collections of essays organized around specific themes.

VOLUMES IN THIS SERIES:

Questioning Nineteenth-Century Assumptions about Knowledge I: Determinism Richard E. Lee, editor Questioning Nineteenth-Century Assumptions about Knowledge II: Reductionism Richard E. Lee, editor Questioning Nineteenth-Century Assumptions about Knowledge III: Dualism Richard E. Lee, editor

ii

Questioning Nineteenth-Century Assumptions about Knowledge II

REDUCTIONISM

Edited by Richard E. Lee Foreword by Immanuel Wallerstein

FERNAND BRAUDEL CENTER STUDIES IN HISTORICAL SOCIAL SCIENCE

iii

Published by State University of New York Press, Albany ©2010 State University of New York All rights reserved Printed in the United States of America No part of this book may be used or reproduced in any manner whatsoever without written permission. No part of this book may be stored in a retrieval system or transmitted in any form or by any means including electronic, electrostatic, magnetic tape, mechanical, photocopying, recording, or otherwise without the prior permission in writing of the publisher. For information, contact State University of New York Press, Albany, NY www.sunypress.edu Production by Diane Ganeles Marketing by Michael Campochiaro

Library of Congress Cataloging-in-Publication Data Questioning nineteenth-century assumptions about knowledge / edited by Richard E. Lee ; foreword by Immanuel Wallerstein. v. — (The Fernand Braudel Center studies in historical social science series) Includes bibliographical references and index. Contents: v. 1. Determinism ISBN 978-1-4384-3441-4 (hardcover : alk. paper) ISBN 978-1-4384-3440-7 (pbk. : alk. paper) 1. Knowledge, Theory of. I. Lee, Richard E., 1945– BD161.Q47 2010 121—dc22 10 9 8 7 6 5 4 3 2 1

iv

2010004836

CONTENTS

Participants

vii

Illustrations

ix

Foreword Immanuel Wallerstein

xi

Introduction Richard E. Lee

1

SESSION I Reductionism in Social Science Andrew Sayer

5

Discussion

40

SESSION II Emergence and Complex Systems Evan Thompson

57

Discussion

90

SESSION III Reduction and Emergence in Complex Systems Jean Petitot

107

Discussion

160

v

vi

SESSION IV Organizers’ Opening Remarks Immanuel Wallerstein

167

Jean-Pierre Dupuy

171

Aviv Bergman

173

Discussion

176

Index

193

PARTICIPANTS

AV IV BERGMA N —Albert

Einstein College of Medicine, New York, NY

J E AN-PIERRE D UP U Y —Ecole

Polytechnique [GRISE], Paris; and Stanford

University, Stanford, CA J OÃ O CA RA ÇA —Director

of Science, Gulbenkian Foundation, Lisbon, Portugal

PAUL CILLI ERS —Philosophy, PAUL D UMOUCH E L —School

Stellenbosch University, Stellenbosch, South Africa of Core Ethics and Frontier Sciences,

Ritsumeikan University, Kyoto, Japan E RI C GOLES —President,

CONYCIT, Santiago, Chile

N. KATHERI NE HAYLE S —English, J E AN PETITOT —Centre

University of California, Los Angeles, CA

de Recherche en Epistémologie Appliquée, Ecole

Polytechnique, Paris, France ISTVA N REV —History,

Central European University, Budapest, Hungary

AND REW SAYER —Sociology,

Lancaster University, Lancaster, UK

ULLICA SEGERST R ALE —Sociology, EVA N TH OMP SON —Philosophy,

Illinois Institute of Technology, Chicago, IL

University of Toronto, Toronto, Canada

IMMA N UEL WA LL E R ST E IN —Sociology, R ICHA RD E. LEE

Yale University, New Haven, CT

(Scientific Secretary)—Fernand Braudel Center, Binghamton

University, Binghamton, NY

vii

I L LU S T R AT I O N S

3.1

Comparison between drawings of visual hallucinations and mathematical models

114

The receptive profile of a simple orientation cell of V1. 1: schematized structure. 2 and 3: mathematical model. 4: empirical recording

118

3.3

The functional architecture of the area V1 of a cat

121

3.4

Left: The association field of Field, Hayes and Hess experiments. Right: Curved Kanizsa illusory contours

123

3.5

Some examples of eigenmodes in V

125

3.6

The retinotopic conformal map, mapping the retina on V1

126

3.7

Lines in V1 correspond to spiral on the retina

126

3.8

Klüver’s planforms are isomorphic to eigenmodes of the bifurcated solutions of the neural network in the synaptic weights of which the functional architecture of V1 has been encoded

127

3.9

Other examples of eigenmodes

127

3.10

Other examples of eigenmodes

128

3.11

The result of the tournament between 12 strategies, each represented by 100 agents

147

3.2

ix

3.12

Nowak and May’s example of Tit for Tat strategy, displayed spatially

152

3.13

The temporal evolution of the subpopulations (c, c) and (d, d) of figure 3.12

153

For b = 2.1 and a 50%–50% InitConfig, defection d dominates immediately and totally

153

The temporal evolution of the subpopulations (c, c) and (d, d) of figure 3.14

154

For b = 1.85 in the critical interval and a 50%–50% InitConfig, the behavior (d, d) begins to dominate; next (c, c) begins to reconquer ground by expanding from nuclei that resisted the initial extermination, but multi-scale nested clusters of c and d appear and expand in a fractal structure

155

The temporal evolution of the subpopulations (c, c) and (d, d) of figure 3.16

155

Evolution of the system for b = 1.85 (inside the critical interval) and an InitConfig reduced to a single (d, d) in a purely (c, c) population

156

3.14

3.15

3.16

3.17

3.18

x

FOREWORD

T

his volume is one of three in a series devoted to the theme: “Questioning Nineteenth-Century Assumptions about Knowledge.” The project was organized by Jean-Pierre Dupuy (a philosopher of science affiliated with the Centre de Recherche en Epistémologie Appliquée, Paris), Aviv Bergman (an evolutionary biologist who directs the Aviv Bergman Laboratory at the Albert Einstein School of Medicine, New York), and Immanuel Wallerstein (a sociologist, formerly Director of the Fernand Braudel Center at Binghamton University and currently a Senior Research Scholar at Yale University). Its Scientific Secretary was Richard E. Lee, the current Director of the Fernand Braudel Center. The underlying premise of this series of conferences was that, in the last thirty years, scholars in all fields have been raising into question some of the fundamental premises of the modern view of knowledge, as it had been developing for at least five centuries and, in particular, as it was codified in the nineteenth century. It was at that time that a view of knowledge that was determinist, reductionist, and dualist came to predominate the intellectual scene, and found parallel expression in the natural sciences/mathematics, the social sciences, and the humanities/philosophy. This consensus, once very widely shared, was seriously challenged in all three arenas in the last quarter of the twentieth century. The object of the series of conferences was to explore the degree of convergence of the questioning in the three arenas, which has often been clouded by the fact that different terminologies are being used in many cases. The format we used was the following. We sought to have sixteen participants: the three organizers and the scientific secretary; three persons to prepare background papers for that meeting, coming respectively from the natural sciences/mathematics, the social sciences, and the humanities/philosophy; and nine others, three from each of the three arenas, who participated in the debate. The only persons

xi

who were present at all three meetings were the three organizers and the scientific secretary. Each meeting had four sessions of a half-day in length: one each to discuss the background paper in each of the three arenas, and a fourth in which the three organizers led an integrative discussion. We found this formula to be very productive. We are publishing three volumes, one for each conference. Each volume contains the background paper and an edited version of the discussion (a very lively discussion, it should be said) on each of the papers, as well as the opening remarks of the organizers at the last session, followed by discussion. We do not consider these volumes to constitute in any sense a definitive resolution of the intellectual issues. Rather, we offer them as what we believe to be a stimulating and intense debate about the underlying epistemological issues. These volumes have the special feature that they bring together scholars from the three main superdisciplines into which the world university system is currently divided. We thereby hope to contribute to overcoming the false separation of the debates caused by the use of differing terminologies in the three domains. Perhaps in the next thirty years the world intellectual community will find a way to reunify the basic epistemology it uses and to overcome some of the limitations of nineteenth-century views about knowledge. The organizers believe that this would be very useful not only in our intellectual pursuits but in the real world to which our knowledge is supposed to apply. We are grateful to the Gulbenkian Foundation which made these meetings possible by its financial assistance and to João Caraça, its Director of Science, who not only supported the project fully but participated in all three of the conferences. I M M A N U E L WA L L E R S T E I N

xii

INTRODUCTION

F

ollowing the 2004 meeting held at Stanford University, which dealt with “determinism,” the second of the three symposia on “Questioning Nineteenth-Century Assumptions about Knowledge” underwritten by the Gulbenkian Foundation was convened at the Maison Suger, Paris, France, 16–17 December 2005. Its charge was to examine the contemporary debates relating to the status of “reductionism” in the sciences, social sciences, and the humanities. Again, participants were invited from a wide range of disciplines in order to insure the broadest variety of opinions possible. It was, however, assumed that all who would participate, either as authors of background papers or commentators, were indeed interested in the epistemological questions in one way or another and were ready to think about the possible limits of nineteenth-century approaches. As with the Stanford meeting, this assumption was born out by the lively, and sometimes surprising, discussions of the contemporary epistemological horizon, both in terms of the problems and prospects of inherited perspectives and of possible alternatives and what they might entail not only for scholarly agendas, but for decision making in the real world. The organization of the book mirrors the organization of the symposium. Each of the first three chapters comprises the background paper specifically authored for the occasion followed by an open discussion. These three sessions were each chaired by one of the organizers. The discussions were wide ranging, as we had hoped, and did not necessarily focus exclusively on the papers. The fourth session, chaired by the scientific secretary, began with remarks from the organizers—in this case, all three of the organizers offered comments on the meeting as it had developed over the first three sessions—and concluded with a final discussion on the numerous issues that had emerged over the two days of the symposium. The first session opened with the presentation by Andrew Sayer, “Reductionism in Social Science.” In a broad overview, Sayer covers upward and downward

1

reductionism and semiosis and flat ontology, as well as anti-reductionism, stratified ontology and the relationship between such stratification and the organization of the disciplines of knowledge. Writing from a critical realist stance, he effectively parses forms of reductionism that fall into the categories of strong social constructionism and anti-essentialism, social ontology and the apathetic actor, the fact-value family of dualisms, and individualism. The thread of social constructionism wove its way through the entire discussion period. Although it was stated that reductionism was ubiquitous, even a prerequisite of science, and ample time was devoted to levels and the asymmetry between or within them and the alternative term, boundaries, it was the politics of epistemology that eventually animated the discussion—for instance, in terms of the putative opposition of structure and agency, the relationship between agency and the fact-value divide, and the intellectual versus organizational boundaries among the disciplines. The second session began with the presentation by Evan Thompson, “Emergence and Complex Systems.” It starts with a discussion of the classical opposition between reductionism and emergentism and their epistemological and ontological components, these positions presupposing a parts–whole or basic–emergent features dichotomy. Given the difficulty of formulating a single concept of emergence without regard to a particular explanatory context, Thompson chooses to focus on emergence implicit in complex systems theory, including the ideas of “circular causality” and “downward causation,” and holism and determinism, to wind up with an examination of the objections to emergence and downward causation from Jaegwon Kim. The conversation following Thompson’s presentation moved from the problem of regularities and the mechanisms of stability and change in systems, both physical and social, to a discussion of the medium in which emergence takes place, fuelled with examples from cognitive neuroscience and literary studies. This led to exchanges about the production of meaning and the idea of meaning as produced in the plural. The productive crossing of disciplinary boundaries was exemplified in the way that the concepts of non-separability in its differing forms and parts

2

and wholes in physical systems shifted the discussion in the direction of Marx and intentionality. The third session started with the presentation by Jean Petitot, “Reduction and Emergence in Complex Systems.” The “narrow, scientific, empirical, and objective sense” of reductionism that Petitot examines concerns mainly complex systems with a micro level and a macro or emergent level where self-organizing structures emerge. He discusses two examples: one from the cognitive sciences, “Phenomenal Experience and Mathematical Reductionism,” and one from the social sciences, “Hayek and the Complexity Problem.” In the first, Petitot presents research that has as its focus the “explanatory gap” between phenomenal consciousness or mental states, representations, contents, structures, events, and physiological mechanisms or neural activity. Readers in the social sciences and the humanities will want to consider the ramifications of this research for their own disciplines. The second example is primarily concerned with complex methodological individualism, causal reductionism and common-sense rules, seen especially through the lenses of the “prisoner’s dilemma.” The discussion led off with what the “brain” is in brain science and continued with a critique of the prisoner’s dilemma, which emphasized that it left out the reality of differentials in power and differences in interests of the individuals involved. What might be a concept of appropriate description, a call to reverse the removal of subjectivity from our concept of nature, and an exchange on the transcendent versus the metaphysical rounded off the session. Immanuel Wallerstein’s remarks at the beginning of the last session make allusion to many of the points that had been hit upon by discussants during the previous three sessions. He moves on from his concern with the non-communication among scholars in different disciplines to speak from the point of view of an ethnographer and historian of epistemological thought as background for his comments on science and reductionism. For Wallerstein, the name of the game is what is useful and how we can explain the complex realities of the human world. Jean-Pierre Dupuy feels that a kind of consensus was reached by the group, in effect that reductionism is no longer an issue, and follows this with a case study

3

of love. Aviv Bergman references what he thinks is the sparse attention paid to the emergence of complexity as a result of linguistics and communicative processes; his overriding point, however, is that in order to understand what is really happening at higher levels it is often necessary to augment the holistic view with lower level information. The group discussion that wound up the symposium began by concentrating on what it might mean to have to get down to the “nitty-gritty” details in order to really understand large-scale dynamics. Examples from biology and social science were proposed and the question of what the consequences would be for observers or manipulators to be inside or outside of the system under consideration was examined. Whether or not there is or should be a basic epistemological stance that is different in the sciences and humanities, and whether or not such a stance as exemplified by the approach to reductionism is changing, formed the final pole of the conversation. Lastly, we would like to recognize and express our appreciation for the exemplary staff support all phases of this symposium received from Donna DeVoist, with Rebecca Dunlop and Katherine Ensor Pueschel of the Fernand Braudel Center, and Marie-Jo Lecuyer (CREA, Paris), who was responsible for local organization in Paris. Their efficiency and good humor were fundamental to the success of this event. RICHARD E. LEE

4

SESSION I

Reductionism in Social Science Andrew Sayer

INTRODUCTION

B

roadly speaking, “reductionism” is used in philosophy to refer to doctrines according to which one can explain some object by reducing it to a different, usually simpler, level—for example, the meaningful to the physical as in behaviorism, knowledge to sense data as in phenomenalism, the social to the biological as in sociobiology. The reductions are not made merely as a way of simplifying complexity, but of locating what their advocates believe to be the causes or sources of the explananda. Anti-reductionists argue that the explananda are irreducible, that even though they may depend on the things to which reductionists appeal—thought on brain cell activity, for example—they have emergent properties or powers which cannot be reduced to those of their constituents without residue. Anti-reductionists therefore argue for a stratified ontology, in which any higher stratum presupposes lower strata but not vice versa—as in the asymmetric relation of the biological to the physical. The strata usually cited are the physical, the chemical, the biological, and the social, but further strata may be invoked within each of these. The plausibility of the idea that the world is stratified arguably provides a warrant for the existence of different disciplines: the physical, the chemical, the biological, and the social deal with different strata of reality. However, as we shall see, the subdivision of social science into disciplines has a much less clear relation to stratification. In part, the rise of different social sciences seems to correspond to the differentiation of spheres in modernity—politics, law, and economics, for example—rather than

5

to different strata, though some might argue that psychology is an exception.1 A third group argues against both these positions, arguing that all objects and processes are on the same level within a relational field, and that what eventuates are products of interaction rather than emergence. In social science, the term “reductionism” is used largely pejoratively, as in the accusation of “biological reductionism” or “psychologism” used by sociologists against those who seek to explain social behavior in biological or psychological terms. It is often used with reference to stratification and emergence held to exist within the stratum of the social, as in accusations of “vulgar materialism,” where actors’ beliefs are treated wholly as a function of their material circumstances. Some arguments about reductionism center not on reduction as such—though they may claim to do so—but the form and direction of the reduction. For example, some opponents of the reduction of the social to the biological may advocate not a stratified ontology in which the social has irreducible emergent powers from the biological, but in effect that the biological is reducible to the social or cultural. In other words, an upward reduction may be substituted for a downward reduction. Upward reductions have become common with the rise of cultural studies and the prioritization of discourse. Support has also grown in some parts of social science, particularly anthropology, for the third, “flat ontology” position (e.g., Ingold 2000). There are also looser usages of the term, again invariably pejorative, which don’t necessarily reduce higher strata objects or processes to lower strata ones (or vice versa) but simply reduce the explanations of multiply-determined processes to a few elements, ignoring others within the same stratum that are believed to be significant. (These might be called horizontal reductionism). The reduction of capitalism to market exchange might be an example of this. For advocates, such reductions provide a way of simplifying and clarifying what they claim would otherwise be unmanageably complex, and they may invoke the prerogative of scientific abstraction and the ideal of explanatory elegance and parsimony to legitimate this. For critics, they involve misattributions of causality and misinterpretations of the meaning of discourses. Thus, for example, the reduction of capitalism to exchange might be argued to allow the effects of power imbalances in control over key resources to be attributed to free exchange.2

6

This last example indicates that reductionism is not merely an arcane matter internal to scientific and philosophical inquiry and lacking wider interest. Reductionism is common in everyday thought and discourse, and it can take forms which have great political significance. Thus, attributing people’s behavioral characteristics in reductionist fashion to their genes has important implications for how we evaluate them and respond to them. Stances on reductionism underlie whether we respond to behavioral pathologies by administering drugs or providing therapy and changing the social environment. Individualistic explanation is a particularly important form of reductionism: by reducing the social to the individual, it attributes to individuals sole responsibility for their fates, so that, for example, individuals are solely responsible for their class and life chances. This is mirrored by a form of sociological reductionism in which individuals have no influence or responsibility for their actions or character and are merely products of wider forces, intersections in discursive networks, etc. These two opposed reductionisms, albeit in more moderate forms, are fundamental to the political divide between right and left. As is usual with reductionism, it is easier to identify the problems of such positions than ways to resolve and avoid them. In this critical commentary on reductionism I shall include both the more technical and looser senses of reductionism, for both cover matters of considerable political significance, though I shall focus mainly on vertical reductionism. Although I am interested in reductionism primarily as an issue in the social sciences, given the importance of relationships between the social and the biological and the physical, it would of course be reductionist (in a pejorative sense) to ignore these. It is also difficult to say much about reductionism without straying into matters of dualism, but I assume that in this context this should be productive. Any general position on reductionism implies some kind of wider philosophical standpoint regarding ontology and metaphysics, in terms of which more specific arguments about reductionism, explanation, and interpretation are developed. It is therefore necessary to give some background on this standpoint, which derives, in my case, from critical realist philosophy. This opposes reductionism and supports a stratified ontology in which emergent powers figure prominently. The main part of the paper develops this approach, offering explanations of emergence, critiques

7

of reductionism and flat ontology positions, and discussions of the implications of interaction among mechanisms at different strata. The second part discusses some influential forms of reductionism in social science—strong social constructionism and its essentialist “other,” the reduction of actors to causal agents and meaning makers, reductionist approaches to values and reason produced by the fact–value family of dualisms, and finally, reductionist treatments of responsibility, whether individualist or socially determinist. Before launching into this, there is a preliminary matter which must be dealt with—the influence of rivalries between academic disciplines and their imperialistic tendencies in driving reductionism.

PRELIMINARIES: DISCIPLINARY IMPERIALISM

The issue of reductionism, of whether one kind of view of the world can be reduced to (and hence replaced by) another without loss, throws different kinds of knowledge into competition, whether for the same ground or over an appropriate division of territory. Raising the issue invites not only debate but competition among disciplines, and hence is as liable also to invite that most tedious of academic tendencies—disciplinary imperialism (Sayer 2000a). Disciplinary imperialism is itself a form of reductionism, at once both imperialistic and parochial, claiming ever greater scope and vision for a particular discipline while remaining within its restricted point of view. The shackling of individual academic ambition to the fortunes of institutionalized disciplines produces claims from each discipline that it is more fundamental and comprehensive than any other discipline. Given disciplinary imperialism, one is tempted to say that economists would say that everything can be reduced to a matter of choice—just as anthropologists would attempt to say everything is cultural, and sociologists would claim that everything is socially constructed. Particularly between disciplines which are close or overlapping in their objects, one finds not only competition but mutual aversions, such as those of sociology and psychology or economics and sociology. These are evident in sociologists’ fear of being accused by colleagues of “psychologism” (reduction of the

8

social to the psychological), which tends to make them refuse to concede anything to psychology, even where doing so would help their own explanations. These are also evident in economists’ scarcely-veiled contempt for sociology, often buttressed by the curious claim that sociologists deal with the irrational and economists with the rational aspects of behavior (which can hardly escape a corresponding implied inequality in status), and by a methodological imperialism that considers deductive reasoning, preferably in mathematical form, as the only kind of approach that warrants the honorific label “scientific.” Disciplinary imperialism invites members of disciplines (who are more “the disciplined” than “the disciples”) to assess theories, explanations, not according to any general standards of empirical adequacy, rigor, coherence, etc., but according to whether they advance the imperialistic ambitions of their discipline. Of course, they rarely do so deliberately; rather they respond to the positive incentives to do so in terms of their personal reputation and that of their discipline, while the arguments of the natives who are being displaced are unlikely to be understood or taken seriously, since they come from another discipline. Individual academics can advance their careers by showing that what was previously imagined to lie outside their discipline’s territory can in fact be better explained by their own discipline’s tropes and theories. As just one example, public choice theory in economics claims to be able to explain any social behavior, not just that generally seen as economic, in terms of the supposedly rational choice of narrowly self-interested individuals. Sometimes, of course, they may be right; this is not a defense of existing disciplinary boundaries—far from it—but a warning of the dangers of such explorations being conducted in a spirit of disciplinary imperialism rather than postdisciplinary learning. One doesn’t have to fall for a sociological reductionism to acknowledge that the competition of the academic field for status and power so brilliantly analyzed by Pierre Bourdieu in Homo Academicus (Bourdieu 1988) has something to do with the way debates about reductionism and the relation of different kinds of discipline, explanation, and theory develop in the context of a field of competing actors and institutions. As Bourdieu argued, the point of acknowledging such tendencies is

9

not to invite a sociologically imperialist reduction of the structure and content of knowledge to a competition for power among academics, but precisely to identify, and hence to limit, the distortion of our understanding of the world by that form of institutionalized competition (Bourdieu 2004).

A CRITICAL REALIST VIEW OF REDUCTIONISM, S T R AT I F I C AT I O N , A N D E M E RG E N C E

The ontological and metaphysical bases of my argument come from the philosophy of critical realism, pioneered by Roy Bhaskar (1975, 1979), and developed most in relation to reductionism by Margaret Archer (1995, 1996, 2000, 2003). This is a fallibilist, post-positivist philosophy, which regards both causal explanation and interpretive understanding as necessary for social science. It is anti-reductionist, arguing that the world is stratified and characterized by emergence. Thus, the meaningful character of discourse or communicative interaction is emergent from, rather than reducible to, physical behavior, and hence must be understood “at its own level,” through verstehen or interpretative understanding. Critical realism’s distinctive analysis of causation is particularly important for dealing with reductionism, and it is therefore necessary to explain this briefly before proceeding. The production of change (which is all we mean by causation)3 cannot be understood by treating it in the conventional manner as a matter of regular successions of events in which event C is regularly followed by event E. Not only does this Humean or successionist theory of causation simply fail to explain what makes such things happen (indeed it merely formalizes what we probably already know and what needs explaining), but its focus on empirical regularities in events renders qualitative change and novelty unintelligible. Stable, precise, empirical regularities depend on objects being internally stable and set within invariant contexts. Such conditions rarely occur spontaneously and generally have to be worked for through the construction of machines and ordering structures in social life in general and through the construction of experiments in science in particular.4 In

10

view of this, the world must be open, capable of many possibilities, not following a predetermined course. Rather, causation is a result of the causal powers and susceptibilities of things (including relationships), such as the power of people to reproduce, inflict violence, or to converse and reflect. The existence of particular powers (susceptibilities) depends on the structure of the objects bearing those powers, such as the neurophysiological conditions of communicative interaction, and often on wider relations or structures or fields in which they are situated, such as language communities. However, these powers are not necessarily activated or exercised, for whether they are depends on contingently related circumstances, and the causal powers and susceptibilities of those circumstances. They exist even when they are not being activated, as in, for example, the powers of an army during peacetime, or the labor power of unemployed workers. If and when they are activated, the particular consequences that flow from it depend on the mediation, (reinforcement, deflection, transformation or blocking) of the process by these latter contextual, causal powers and susceptibilities. Hence causation cannot be understood as a unilateral process in which some agents or things produce certain kinds of change, regardless of the nature of the things on which they are acting. Thus, the reception of this article depends on not only its content and my circulation of it, but on you the readers’ prior dispositions and understandings. Causation, then, is to be understood in terms of powers, which may exist even when not being exercised. Critical realism therefore rejects “actualism”—which reduces causation to interaction or the exercise of powers, without explaining what those powers are or from which they derive—though this is a common form of reductionism in social science; indeed it is central to approaches as different as positivism and actor network theory.5 The production of change is thus always two-sided, dependent not only on the nature of the causal agent as it were, but also on the things which activate its powers and mediate its operation and effects. It may be helpful here to use the analogy of shaping. When we shape something, what happens depends not only on the tools we use but the properties of the object we are trying to shape. Thus wet snow is easier to mold than dry snow because its internal structure offers

11

more resistance. A one-sided view of this causal process which ignored this would be reductionist in a pejorative sense because it would misattribute what is due to the object being shaped to the shaping agent. Now before going further, I want to deal with a common, disarmingly simple defense of reductionism which appeals to the unavoidability of simplification: the world is complex and since we have always to simplify to understand it, reductionism should be seen less as a problem than a solution. On this view, anti-reductionism might be seen as implying a reactionary opposition to the usual methods of selection, abstraction, and simplification that science, and indeed any kind of study, involves. Thus reductionism is often justified by reference to the methodological ideal of parsimony and elegance in explanation. This ideal is most appealing in mathematics but is less so in substantive disciplines dealing with causal processes. It is often coupled to instrumentalist models of explanation which take it to be reducible to a matter of calculation and prediction, where explanation is simply “postdiction” of events, and is assessed in terms of how much of the variation of y is “explained” (i.e., can be calculated) by a formula incorporating other measurable variables.6 In mainstream economics the criterion of parsimony tends to be invoked as a way of buttressing the reductive accounts produced by its own disciplinary imperialism, so that when the reduction of the determinants of action to self-interested rational choice is questioned in terms of its explanatory adequacy or plausibility and charged with misattribution of causal responsibility, it is defended in terms of its simplicity and elegance, and also the scientific license to abstraction (“we proceed by the power of our abstractions”).7 Such an approach confuses questions of how quantitative variation in one variable is quantitatively related to that in certain others with the causal question of what makes something happen. In other words it mistakes calculation for explanation, or the things of logic for the logic of things. More broadly it commits the “epistemic fallacy” of confusing questions of knowledge with questions of being, as if what happened in the material world depended on how we think about it (Bhaskar 1975). Logic deals with relationships between statements, not things: causation concerns processes and transformative relationships between things (Harré 1970; Sayer 1992).

12

Implicitly, such an approach assumes closed systems lacking novelty and emergence. On this criterion, a prediction or postdiction based on simple curve-extrapolation involving no model of causal mechanisms and how they work or of substances and structures is preferable to an approach which does attempt to provide such models and accounts. The ideal of parsimony is problematic where it is taken to trump the criterion of explanatory adequacy, so that inferior explanations are preferred on the grounds that they are simple and elegant, even though they fail to resolve problems explained by more complex accounts. Its only defensible use in substantive sciences is as a tie-breaker where two different substantive theories appear to have equal explanatory adequacy. This is not to reject abstraction, but rather to distinguish between the development of theory through the use of abstraction to attempt to isolate one-sided aspects of objects of interest, distinguishing them from others, and the explanation of concrete (i.e., many-sided) phenomena. The unavailability of practical experiments in social science in which abstractions can be objectified through the isolation, control, and purification of processes throws a heavy burden onto abstraction and thought experiments. While these facilitate theorizing the nature and properties of objects, when it comes to applying the abstractions to the explanation of concrete, many-sided situations in which there are many structures and mechanisms present, there are dangers of a kind of reductionism which mistakenly attributes to the abstracted elements effects which are actually due to things from which the researcher has abstracted, or ignores significant emergent properties of combinations of the abstracted elements (Sayer 1992). Downward Reductionism, Stratification, and Emergence In downward vertical reductionism, higher order phenomena (for example, speaking) are treated as capable of being explained and/or predicted8 purely in terms of lowerorder phenomena (e.g., electrochemical processes) (Bhaskar 1975).9 Opponents of vertical reductionism either (1) argue that lower-order processes provide a basis of contingently enabling and constraining structures and mechanisms for higher-order

13

or emergent powers, which are irreducible to their preconditions, or (2) that all processes are “on the same level,” and that there is no stratification or emergence, only interaction. As we shall see, some opponents of downward reductionism in effect seek to replace it with upward reductionism. Empirically, the evidence counts against this vertical reductionism, since successful instances of such explanations and predictions have not been forthcoming, though that in itself does not discount them as a future possibility. Note that even if reductionism were successful, so that, for example, we could explain consciousness by reference to neurophysiological mechanisms, it would still have to start from an understanding of consciousness, otherwise we would not know what neurophysiology was explaining, or why certain neurophysiological events were significant (Collier 1994). This would of course leave the reality of consciousness intact and not explain it away. The position I wish to advocate—(1)—does not involve either an ontological reductionism (according to which higher-order phenomena are nothing but complicated aggregates of lower-order phenomena) or an epistemological reductionism (higher-order phenomena can be explained in terms of lower-order), but is compatible with a non-reductionist stratified ontology, according to which the world is stratified, broadly, into physical, chemical, biological, social/cognitive, etc.,10 levels, so that, for example, biological powers are dependent on but irreducible to, or more simply, are “emergent from” the chemical and physical. The phenomenon of emergence suggests that the world is not merely differentiated but stratified; the powers of water exist at a different stratum from that of hydrogen or oxygen. Emergence can be understood as follows: Where objects are externally related and only influence each other in ways which do not alter their respective powers (merely the effects of their exercise), no emergent effects arise from their interaction. Where the interaction of objects produces changes in the structure, powers, and susceptibilities of those objects, it can prompt the development of emergent powers—powers not independently or merely additively possessed by those objects. Where these are discovered, explanation of behavior attributable to them need not involve a regress to the powers of the constituents on which they depend. By contrast, mere aggregates consist of externally-related individuals

14

and hence lack emergent powers. Disaggregation, or reduction, as a step toward explanation of the whole, therefore presents no problem in such cases. In the case of internally-related individuals, such as where individuals relate as landlord and tenant, as specialized producers in a division of labor, or as superior and inferior, so that some of the powers and susceptibilities of the individuals are changed by the relationship, emergent powers arise, such as the development of the institution of rent and the productivity gains of division of labor. For such cases, explanation by downward reduction fails (Sayer 1992; Archer 1995). One of the most important kinds of emergence is that of semiosis—the generation of meaning from the relationships of signifiers, signifieds, referents, and subjects. For the social sciences and humanities, semiosis is not only a means for understanding the world but part of the object of study. In combining words in different ways, in particular contexts, people create meanings which are not decomposable into the contributions of the individual words, because the meaning comes from the play of difference among them in relation to the context and actors (Fairclough, Jessop and Sayer 2002).11 These are not only emergent meanings but emergent powers, for they can (co-)produce novel changes in how users think and respond. Reasons and other meaningful features of discourses can be causes, not in the physical sense, but as things which produce (or prevent) change. The very point of discourse is to produce some kind of effect, whether it be maintaining some situation or transforming it. Thus, for example, through neoliberal discourse, attempts are made to encourage individuals and states to see themselves as market actors, choosing what to buy or sell, rather than as citizens concerned with the public good. We need to understand both what the discourse means and what effects it produces. Social scientists don’t have to choose between causation and erklären on the one hand and meaning and verstehen on the other hand: they need both (Bhaskar 1979). This anti-reductionist view of stratification is to be preferred to the flat ontology view (2) on the grounds that the latter fails to acknowledge the asymmetric relations between strata: “One stratum ontologically presupposes another if it could not exist unless the other existed” (Collier 1994, 131). Culture is not ontologically

15

presupposed by physical phenomena, though of course culture is constitutive of the science of physics;12 e.g., all animals are composed of chemicals, but not all chemicals are parts of animals, all persons are composed of biological materials—but not vice versa (Collier 1994, 107). This asymmetry is also reflected chronologically, where biological organisms emerge at a late stage in the geological record, and social and cognitive processes at a still later stage. The higher strata do not break the laws of the lower strata—for example, social practices do not change gravity (Collier 1994, 46–50, 107–136). Hence a tree does not break mechanical laws but in addition it grows according to its biological powers which are enabled and constrained by, but not reducible to, those mechanical laws (Collier 1994, 119). It is vital to note here that in distinguishing social, biological, and other strata, we are indeed identifying objects or processes by their distinguishing characteristics only, ignoring what they have in common. Thus biological organisms also have chemical and physical characteristics, and as social beings we are also biological, chemical, and physical beings and what we can do depends on all these levels, both through the non-reductive dependence of the social on these successive levels, and through our physical, chemical, and biological processes. It is therefore an error to suppose that social phenomena are governed only by social processes (contra Durkheim’s sociological imperialism),13 biological only by biological processes, etc. Hence, as stratified beings, rather than beings existing only in one stratum, we have mechanical powers and susceptibilities, as well as those specific to mammals or humans, and our freedom may be limited at any of these levels. Vertical downward reductionism tends to be associated with assumptions of a micro-regress to successively simpler and more uniform or atomistic elements. However, even in the physical world, variability seems to “go all the way down,” so that no phenomena are exactly uniform (Dupré 2002). The metaphysical assumption of atomism, implicit in many forms of reductionism, appears to be untenable. Emergence need not necessarily have exclusively microscopic or more simple preconditions, but can arise from locations within wider structures, as in the case of semiosis, where the meaning of particular phrases is emergent from their relations to wider tracts of discourses and to context. However, holist reductionism,

16

which takes this latter possibility as universal, so that everything is relational, a product of its relations to everything else, and hence in which nothing can ever be indifferent to anything else, or has the slightest degree of context independence, produces bizarre implications of a world in which no particular kinds of entities could be identified, and language would be impossible, since there would be no relatively stable and commonly understood terms which could be applied among interlocutors, who of course would always be differently located in and constituted through the web of relations.14 It would seem more plausible to argue that there are both internally-related and externally-related objects, that some objects are sharply distinguished from others (e.g., pebbles, knives) while others are not but are instead part of continua of variation and differentiation, and that objects range from the relatively durable and relatively context-independent through to the ephemeral and entirely context-dependent. (Most social objects appear to lie in the middle of this last range). One-sided, restrictive ontologies, whether of atomism or universal internal relations, are unhelpful and unnecessary. The complex mixes of internally and externally related phenomena that social scientists study present formidable problems of abstraction—of isolating one-sided aspects of objects so that their structure and contributions may more easily be assessed—without doing violence to the situation by dividing the indivisible or conflating the separable (Sayer 1992). Simplification and abstraction are unavoidable; the only questions about them are how they should be done and how abstractions should be recombined in the explanation of the concrete. Interaction or Downward Interventions: a Case for Flat Ontology? Higher-order objects can react reflexively on and change lower-order processes from which they were formed; for example, people can alter biological processes through medicine and agriculture. This implies that the higher order mechanisms are not merely epiphenomena but have causal autonomy. However, in noting that people intervene in and modify physical, chemical, and biological processes, including those composing human bodies themselves, we may be tempted to conclude that

17

this refutes the assumption of a hierarchical set of relations in which the physical provides the conditions which enable and constrain the chemical, which in turn enable and constrain the biological. It might seem that the relations are symmetric and horizontal, and hence simply interactions among things which are all at the same level, implying a “flat ontology” or position 2, noted above. I think this is a mistake, and that this kind of interaction or reaction back on lower-order mechanisms is not intelligible in terms of a flat ontology. Firstly, there is the problem of asymmetry: even though we can intervene in lower strata and alter them, it remains the case that the physical could exist without the biological, but not vice versa, the biological without the cognitive, but not vice versa. We may be able to change our heart rate deliberately merely by thinking about certain things, yet the heart can function even in the absence of our thinking about anything, but not vice versa. The asymmetry and emergence still holds. Secondly, it is vital to remember that we are simultaneously biological, chemical, and physical beings, and that we intervene in the world through mechanisms that we possess at those levels, as well as through those mechanisms which seem to be distinctive to humans (and perhaps a few other species) and at a higher stratum. The latter may help distinguish us from other objects, but they only partly constitute what we are, for those powers presuppose lower-order mechanisms too, such as cardiovascular processes, etc. When we intervene in lower-order processes, as we do through agriculture, engineering or medicine, for example, we do so at their own respective levels: we intervene in agriculture at the physical, chemical, and biological levels, through digging, watering, fertilizing, and weeding, and so on. We do not intervene merely by thinking about agriculture, for discourses have to produce responses in terms of material actions to produce material change. Rather, according to our understandings of the world, we attempt to guide actions toward certain ends by physically, chemically, or biologically activating certain ecological mechanisms. Thirdly, we can change objects and processes only in accordance with their own particular powers and susceptibilities, whatever they are, whether we understand them well or very little. This is another corollary of our argument about the double-sided nature of shaping. This is why interventions in nature often fail, why science and

18

practice in general are so difficult, and why they so often produce results other than those intended. They are not simply products of our wishful thinking. That intractability is not merely an obstacle to intervention but what enables it, for it is these independent powers and susceptibilities (that is, independent of how we regard them) that we mobilize. A simple way of putting this is that we intervene in nature in accordance with its own laws. This might seem unsatisfactory, for sometimes it appears that we change those laws—in other words that we change nature at a very basic level—for example, creating new substances and organisms, breaking relations that were formerly assumed to be necessary. However, when we do this, we again produce a manipulation and transformation which is itself enabled and constrained by the properties of the elements being worked upon. Thus reproductive technologies can change relationships that were once thought to be inescapable, necessary conditions of conception, but they do so by shifting to a lower-stratum and by taking advantage of the powers and susceptibilities of the objects at that level to alter what materializes at a higher level. Again it is material interventions that achieve these material changes. Cultural constructions or discourses on their own are insufficient, even where they are necessary conditions for interventions. The phenomenon of interaction or downward intervention in lower strata creates the possibility of co-evolution in which the development of certain lower order structures and powers is enabled and constrained by the contingent presence or intervention of higher order mechanisms. Thus, our mental development is partly dependent upon the existence of particular kinds of contingent social interactions and conditions which enable the emergence of certain capacities, and in turn these may be susceptible to further development and new emergent powers, again provided certain contingent social conditions obtain. Our development arises from a co-evolution of the social and psychological and biological (Ingold 2000). It is precisely this interaction and co-evolution that makes it difficult to decide whether psychological and social phenomena are on the same or different levels. This idea of a flat ontology in which nature is seen as a product of, or indistinguishable from, culture, has become popular in anthropology.15 There are several

19

ways in which this position tends to be supported. Firstly, advocates may appeal to the fact that our natural environment has been thoroughly transformed by earlier human action, and thus is already cultural, rather than belonging to some extracultural realm. However, as we saw earlier, although this is undoubtedly the case, these transformations and the environments and artifacts they produce are enabled and constrained by the properties of the physical and biological processes involved, and do not break their laws. While they are cultural in the sense that, insofar as the transformations are registered by people, they are culturally construed, informed, and guided, they are nevertheless not reducible to their cultural significance but are dependent on physical powers which could and have existed independently of cultural interpretations. There is no warrant for a (disciplinary imperialist) slippage from the idea that anything can be culturally interpreted to the idea that everything is therefore only cultural. Again, this is the reason that attempts to transform nature often fail to produce the intended consequences, and why labor and science are hard work. Thus we could argue that one of the reasons Western industry has caused so much environmental damage, is the problematic Western, instrumental, cultural construction of nature as a resource. These unwanted effects are a consequence of our inadequate understanding of the independently existing causal powers of the processes we intervene in and attempt to mobilize. Secondly, it is sometimes supported by the argument that “we” (humans) are not outside nature, but part of it. However, this is of course accepted. We act within nature as one of its own forces, as one of its internal differentiations. We can and should be aware that we use the term “nature” in two senses—both as inclusive of humans and counterposed to humans (Soper 1995). Thirdly, it is sometimes reinforced by an epistemic fallacy in which it is wrongly inferred from the fact that the knowledge of the sciences is culturally or socially produced, that its objects are too, and in the same sense. This involves conflating objects with actors’ concepts of them (as in the ambiguous expression of scientists “constituting objects”). In addition, it is sometimes bolstered by a slippage from the idea that observations are theory-laden to the idea that they are theory-determined.16 We do indeed necessarily make observations via avail-

20

able schemata, conceptual frameworks or theories, but that does not preclude at least some expectations being confounded. In order to observe weapons of mass destruction (WMD) in Iraq, the weapons inspectors clearly had to have concepts of WMD and what distinguished them from other weapons. But those discursive constructions did not determine what they found; they still had to go and look and try to answer their empirical questions. What they observed was theory-laden but not theory-determined. Fourthly, it may be defended by arguments to the effect that rejections of this position must claim some kind of Archimedean standpoint, outside existing cultural interpretations, one which actually involves an unacknowledged, ethnocentric assertion of a Western, modernist cultural standpoint. After all, many cultures do not use a culture–nature distinction. However, critical realists agree that we can only understand the world via available discourses and cannot get outside them. But discourses are not immune to empirical challenge, as our WMD example shows, nor are they perfectly coherent and consistent. Critique can always work from within discourses exploiting their inevitable inconsistencies. As regards the relative adequacy of different cultural accounts, this is a matter for substantive argument concerning particular understandings. We have already noted one of the ways in which Western understandings of nature are deficient, and while the understandings of other cultures are helpful in identifying these problems, some of the problems may be registered through Western theory-laden empirical feedback, such as that which identifies global warming. Knowledge is primarily a practical matter of coping with the world, and while not just any kind of understanding can cope equally as well as any other, many different understandings of our relation to our environment may be “practically adequate” in the sense of allowing their holders to live successfully. Cultures are not hermetically sealed from or closed to one another, such that dialogue and critique are impossible, though intercultural exchanges are likely to be difficult, and it is illiberal to ban intercultural critique. (As a westerner I welcome other cultures’ critiques of Western culture.) The alarming environmental and social consequences of Western cultural constructs of nature shows that cultural understandings are fallible and susceptible to critique,

21

both from within and from other cultures. The adequacy of the culture–nature distinction or relation itself is a matter of argument, not cultural faith. Moving on from the culture–nature relation, there are other arguments against emergence. Tim Ingold discusses the phenomena of consciousness and the self, criticizing the idea of these being emergent from preconditions existing purely within the individual, and argues instead that consciousness and the self are not bounded by the body but are relational (Ingold 2000, 2005). Thus, it can be argued that the self depends on social relations, that consciousness depends on interactions with others and with the environment, so that in some sense it is “seamlessly distributed across persons, activities and settings” (Ingold 2005). Ingold takes this to imply that the world is not layered, but relational, so that what any one process is and can do depends on its relations to other processes. I agree that these particular claims about the self and consciousness are plausible,17 but that they presuppose, rather than contradict, ideas of stratification and emergence. I noted earlier that emergence is not restricted to cases where the preconditions of the emergent powers are internal to the objects having those powers (i.e., micropreconditions), but may arise from the objects’ insertion into wider relations, as in the case of semiosis and discourse. I would suggest that the self and consciousness are also examples of this kind of emergence rather than of mere interaction. Moreover, they are products of emergence rather than mere interaction because they involve the same kind of asymmetric relations with biological and other physical conditions discussed earlier.

SOME COMMON KINDS OF REDUCTIONISM IN SOCIAL SCIENCE

There are many further forms of reductionism in social science. The work of Margaret Archer represents a far-reaching critical realist critique of many of them, particularly those relating to structure and agency and individual and society, and I commend her recent books on these subjects to readers (Archer 1995, 1996,

22

2000, 2003). Rather than go over similar ground I shall focus on four types of reductionism, only the first of which is a central target of Archer’s critique. Strong Social Constructionism and Anti-Essentialism Strong or idealist constructionism is a form of upward reduction of being to knowledge or discourse, and construction to construal, so that phenomena, including socially constructed phenomena, are reduced to concepts of or discourses about them.18 Conversely, discourses are seen as perfectly performative, perfectly because they not merely contingently produce what they name, but they do so infallibly, regardless of the intractabilities of the materials—physical, psychic, social, ideational—which they use in their processes of construction. Indeed, since the materials are supposedly also merely socially constructed, they have no properties independent from those which the construal gives them. This implies that collective wishful thinking always works. It confuses knowledge with what it is about, and assumes that construals automatically and successfully become objectified as constructions. The reductionist character of strong social constructionism is evident in the popularity of the hyperbolic metaphors of construction and constitution, as in the claim that the subject is “constituted in discourse.” These terms represent complex processes of co-evolution and fallible shaping as the unilateral production of subjects by discourse out of nothing. On this view physical phenomena are no more than what they are socially constructed as (or rather construed as) by physicists or others, and the social constructions and discourse of scientists and others “go all the way down.” The reasons that science is so difficult and laborious, so prone to failure, are inexplicable according to this view. Ironically, it presents a view of knowledge as incorrigible, since there is nothing outside or independent of knowledge about which it can be mistaken. To borrow and extend Bourdieu’s term, it is the ultimate “scholastic fallacy,” a product of the projection by those whose work focuses heavily on discourse and the endless play of difference within it, of that orientation to the world onto the world at large. If we are to understand the fallibility of knowledge, we have to be realists, that is, we have to acknowledge

23

that it is about something independent of any particular observer’s construal, such that they can be mistaken about it. This is entirely compatible with the view that we can only experience and know the world in terms of particular schemata, or discourses. Contrary to a common misconception, it is the fallibility19 of knowledge which gives us warrant for believing that the world—and I include the social world—is not merely our voluntaristic construction. A further reinforcing element is again imperialist concepts of culture, in which not only is everything we know understood in culturally inflected ways, but everything is presumed therefore to be cultural, as if nothing had any qualities that were independent of their cultural significance. With the help of a little disciplinary imperialism it is tempting for sociologists and anthropologists to conclude that since the sciences and other academic knowledge are themselves different ways of making sense of the world, sociology and anthropology are master disciplines affording their disciples privileged insight into all these other kinds of knowledge, able to see what those they observe cannot, namely that their knowledges are (just) particular cultural constructions. “Blank-slateism”—the treatment of people as blank slates on which anything can be written by cultural or social processes, or as “indeterminate material” (Durkheim)—is an example of the kind of reductionism which reduces shaping to a one-sided unilateral process, as if enculturation or socialization had no psychological or biological conditions. It is buttressed by a fear of a converse reduction of the social to the biological. It obviously completely fails to explain why most nonhuman objects are not similarly susceptible to such profound cultural influence and variation, as opposed to being externally construed and used in different ways according to culture.20 Although it can be modified to avoid this problem, Bourdieu’s concept of habitus—the structured set of dispositions acquired by individuals as a result of being subject to a specific configuration of social relations and influences, particularly in early life, dependent on their particular location within the social field—is an example of this reductionism (Sayer 2005). Not just anything can acquire a habitus, and we must have certain properties to enable us to do so. At best it serves as a provisional place filler for an account of psychosocial processes.

24

Similar problems attend the Foucauldian reductionist concept of subjectification. There must be something about people that makes them susceptible to such influences, though accounting for this would take us beyond sociology into fields such as psychology and would hence be problematic for sociological imperialists. One of the most common kinds of reductionism in social science involves forgetting this second side, thereby producing representations of the production of change as a one-sided unilateral process. However, as we noted regarding co-evolution, the development of susceptibilities to enculturation at any particular time is itself, in part, the contingent product of earlier phases of interaction. One of the reasons that strong versions of social constructionism are popular in sociology and anthropology derives from the fact that they are much concerned with cultural variation, with the remarkable variety of different ways people can make sense of and organize their world and themselves, and as a result become very different kinds of people, though each may imagine that theirs is the definitive understanding of the world. Yet for us to be enculturated in various ways, we must be the kind of beings who are susceptible to and capable of such influence—different from objects such as rocks, which are whatever they are regardless of how different cultures or discourses construe them. The high-order phenomenon of cultural variation is rooted in and emergent from particular kinds of lower-order phenomena (neurophysiological preconditions, social interaction), but the latter do not dictate or predetermine, nor help enable us to predict whether we are enculturated to be atheists, Christians, or Muslims, for example. Strong social constructionism is itself fueled by a fear of a converse form of reductionism—usually described as “essentialism”—in which objects are attributed fixed, natural powers which predetermine what they do. The principle target of anti-essentialism has been naturalized accounts of gender that take it to derive from uniform and immutable biological qualities that unilaterally determine our behavior. The effect of this is to naturalize what is a product of a contingent coevolution of the social, psychological, and biological and the effects of particular, local social conventions. The naturalization of such conventions and suppression of difference and alternatives is obviously oppressive. I agree that gender does

25

not have a fixed essence, though it does not follow from this that nothing has. Anti-essentialists may fear that critical realists’ attribution of any causal powers and susceptibilities to actors amounts to essentialism. However, such powers are not necessarily fixed but themselves depend on the contingent reproduction of their preconditions, some of which may be social. Causal powers may mutate or fail to be reproduced. Nor need they be identical in all cases of a particular class of object; the physical powers of people obviously vary significantly. Secondly, to identify particular powers or essences as defining certain objects is not to reduce those objects to those powers, for they may have accidental properties too, which differentiate them, and which may be crucial in certain contexts. Thus, to ascribe certain powers to humans need not entail denying their scope for variation. Thirdly, the activation of causal powers is not inevitable but dependent on context. Fourthly, again as noted earlier, if and when they are activated, the effects also depend on the context. Thus, attributing causal powers or susceptibilities to people does not entail the determinism and suppression of difference that anti-essentialists fear (Sayer 2000b). It allows and indeed helps us to understand the complex kinds of co-evolution involved in socialization noted earlier. The universalism implied in imputing human powers does not entail uniformity and an underestimation of human variety, indeed the remarkable capability of humans for cultural variation presupposes that they have the particular powers which enable this (Collier 2003). The fact that some other species also seem capable of cultural variation does not contradict this but merely invites us to ask what it is about them that enables them to have this property. Social Ontology and the Apathetic Actor Positivist, behaviorist, and other strictly naturalist philosophies of social science, which reduce action to behavior, and actors to agents with only physical powers, represent one scientistic form of reductionism. Anti-naturalists who respond by ignoring causal powers and who reduce the social to the hermeneutic, and hence reduce the task of social science wholly to the interpretation of meaning, invert this

26

reductionism. This also tends to abstract meaning from practice, from habituation and embodiment, thus producing an overly rationalized and discursive model of action, as noted by Bourdieu (Bourdieu 2000). In practice, as theorized by critical realism,21 we need both causal explanation and interpretative understanding to explain social phenomena, for as we noted earlier, semiosis can itself be causally efficacious; indeed that is its point. We also need to recognize the practical, habitual character of much action, in which, as Bourdieu and Merleau-Ponty have emphasized, embodied know-how figures more prominently than discursive, propositional knowledge. While this avoids the opposing reductionisms of the first two approaches, it might still be termed reductionist insofar as it fails to take explicit account of the fact that we are not merely causal agents and meaning makers, but needy beings, characterized by lack, want, and desire, and capable of flourishing and suffering. These are special powers and susceptibilities in that they are not merely potentials but ones that require and seek solutions or satisfactions. They are purposive, and their satisfaction may be crucial for well-being, or indeed survival in some cases. The neediness is not merely physical but psychological, insofar as we have a need for recognition and sociability. We also have the tendency to form attachments and commitments to others, and to practices, causes, and things, whose maintenance can become important for our well-being, so that our attachment is not merely intellectual, but emotional, such that detachment or loss is experienced as a kind of bereavement or injury. Our relationship to the world is not merely one of causal agent or contemplative interpreter, but rather one of care (in Heideggerian terms), or concern, arising from our vulnerability, dependence, and neediness. It is from this relationship to the world and others that normativity and morality arise. To put the point another way: while the analogy of shaping shows the necessity of theorizing the matter of just what it is that is susceptible to socialization or subjectification, it presents a rather passive view of the person, as still conditioned by “hydraulic” social pressures, even though the results depend on her susceptibility to them. This is a form of reductionism (of agency to structure) in that it ignores firstly the mediation of those influences by our perceptions and reflexivity (Archer

27

2003), and secondly our pro-active orientation to the world, of care and concern, stemming from our vulnerability, neediness, and lack. It produces a model of actors as apathetic, as mere cultural dopes. Thus, it is partly our need for food and shelter, and for the support, company, and recognition of others, etc., which makes us susceptible to the prevailing norms and goals of our culture through which those things are contingently provided.22 While there is some recognition, albeit uneven, of the mediating influence of perception and reflexivity, explicit acknowledgements of our orientation to the world of care are rare. Rather individuals are represented as having more or less autonomy, more or less susceptibility to social influence, as being capable of reflection, and sometimes as embodied and practically engaged with the world, but rarely as needy beings, concerned about their world. Possible reasons for this common form of reductionism are: 1) the scholastic fallacy identified by Bourdieu, in which academics project their cognitive, contemplative relationship to the world onto those they study, ignoring the overwhelmingly practical nature of everyday action (Bourdieu 2000); 2) a related scholastic fallacy (ignored by Bourdieu) in which social scientists project their acquired, positive, “de-normativized” academic orientation to the world onto actors, thereby failing to take seriously lay normativity, which stems from this neediness, either by ignoring it or reducing it to “beliefs” or “values,” which in turn are treated simply as facts about their holders; 3) difficulties in empirically identifying needs and concern in a rigorous fashion; 4) a fear of “essentializing” needs and desire, underestimating the social element in their co-evolution; and 5) a fear that acknowledging this will admit a teleological form of explanation into social science.23 Reductionism and the Fact–Value Family of Dualisms: The Derationalization of Values and the Devaluation of Reason Although it seems to me to be simply dogmatic to reject all dualisms a priori, many particular dualisms turn out to be problematic in practice and can be criticized on a posteriori grounds. The problems often involve a form of reductionism in which a complicated pattern of difference is reduced to two single, simple, purified opposites,

28

with no internal complexity and no intermediate terms. In addition the two sides tend to be defined mutually and negatively in terms of what the other is not. The particular problematic dualisms that I want to comment on here belong to what might be termed the “fact–value family of dualisms” (see Putnam 2002). These dualisms are not merely philosophical constructs, but are institutionalized in the very organization of contemporary scholarship and education, with the attempted expulsion over the last two centuries of values from social science, and the corresponding ghettoization of normative thought in moral and political philosophy. They shape the way social scientists think about both their academic work and the nature of values and reason in the lives of those they study: Fact—value is—ought reason—emotion science—ideology positive—normative objectivity—subjectivity It might seem that this family of dualisms has already been widely deconstructed. However, it has so only in a one-sided manner, leaving a highly damaging form of reductionism intact. In effect, in the postpositivist literature, qualities identified by the terms on the right-hand side have been argued to be found also on the left-hand side, so that subjectivity, values, etc., are to be found in science and the factual. Thus, there are probably now more social scientists who believe that social science cannot be value-free than believe it can, and indeed many who believe it should not try to be. Much of the literature in the philosophy of social science and especially feminist theory over the last few decades has addressed this “seepage” of qualities across the dualisms from right to left. However, what has not been noted is the lack of seepage in the opposite direction, in the form of arguments that some of the qualities of the terms on the left—objectivity, reason, etc.—are to be found on the right too. The deconstructions have therefore been one-sided

29

in that they have not challenged the fundamentally reductionist and de-rationalized conceptualization of values, subjectivity, emotion, and normativity that has been distinctive not only of positivism but modernist thought.24 Thus, many who regard themselves as radical critics of the dualisms continue inadvertently to reinforce them by assuming that those on the right-hand side are beyond the scope of reason, that emotions are reducible to affect or feeling and lack cognitive content, that values are not related to the nature of the world, that valuation is radically different from reason, and that normativity has nothing to do with the way the world is. On the one hand, those who still aim for value-freedom regard values as a kind of irrational or arational contaminant threatening the objectivity of social science, while many of their critics, also confusing objectivity with neutrality, respond by keeping values and throwing out objectivity, and hence inadvertently confirm this de-rationalized conception of values. In sociology the de-rationalization takes the form of either subjectivism (values are merely subjective beliefs, not about something capable of being true) or conventionalism (values are no more than conventions—“what we do ’round here”), as if they were unrelated to flourishing or suffering, or as if the latter themselves were merely subjectively or conventionally determined. (These are assumptions that no one can live: for example, when someone seriously harms us, we do not say “in my subjective view, that is wrong” or “we just don’t do that ’round here” or “don’t you realize that that’s culturally constituted as bad.” Instead, we point to the suffering that has been caused.) These problematic conceptions of values continue to limit social science, causing it to provide a largely alienated understanding of the social world, such that it is hard to understand why anything matters to us. This common form of subjectivist reductionism involves a confusion of three different and independent meanings of “objective” and “subjective” (respectively as): 1) value-free and value-laden; 2) true and untrue claims or claims that seek or do not seek truth; and 3) pertaining to objects and pertaining to subjects. As a result, objectivity (2) is confused with and assumed to require value-neutrality (1) and conversely, values are seen as threatening objectivity (2) when they need not. Also subjective views (3) are assumed to be untrue or unlikely to be true

30

(2), when again they need not (Collier 2003; Sayer 2000b). Truth is also often understood as an all-or-nothing matter, which renders the concept unusable and generates concerns about dogmatism. One of the fronts on which this de-rationalization of values has begun to be challenged concerns emotions. Theorists such as Archer (2000), Helm (2001), Nussbaum (2001), and Oakley (1992) argue that emotions are not reducible to their affective dimension but also have cognitive and desiring or motivating elements. Emotions are evaluative judgments of matters affecting, or imagined to affect our well-being, and hence one can speak of “emotional reason” instead of counterposing emotion to reason. Of course, emotional judgments are fallible—we may be mistakenly worried, for example; but unemotional reason can be fallible too. Infallibility is not a requirement of rationality,25 and fallibility presupposes that there is something—the object of our thought/feeling—about which we can be mistaken (Collier 2003). Emotions are obviously subjective (in sense 3, pertaining to subjects) but they can also be about objects (things about which we feel fearful, proud, ashamed, etc.) and their judgments may or may not be mistaken. If we acknowledge that the subjective and the value-laden are not beyond the scope of reason, and indeed that valuation and reason overlap, then we can begin to counter this form of reductionism. To understand the social world we often need to use “thick ethical concepts,” such as “cruel,” “generous,” “arrogant,” or “oppressive,” whose meaning cannot be rid of normative evaluations without loss of positive descriptive power.26 Such concepts allude to whether people flourish or suffer, and flourishing and suffering are in turn simultaneously positive and normative, not one or the other. At present, however, the incomplete deconstruction of the dualisms means that the de-rationalization of values remains largely unchallenged. One possible inhibition to the deconstruction is the liberal belief that conceptions of the good should be regarded as private matters for individuals, and the fear that challenging this and acknowledging the rational character of values will license dogmatic, authoritarian ideas of “moral truth.” However, like positive claims, evaluative judgments are fallible and contestable and can be treated

31

as matters of public deliberation. It is dogmatism, not discussion of values, that we have to fear (Anderson 2004). The de-rationalization of values is complemented in modernism by the reduction of reason to rationality and often to logic, removing its propositional dimension and its involvement in valuation. Thus, in mainstream economics, rational economic actors’ preferences don’t have to be rational (how could they be, for preferences are seen as de-rationalized values), they just have to be consistently or logically followed and acted upon. In order to counter the de-rationalization of values we need to counter the reduction of reason to logic and reinstate an older concept of reason, still present in concepts of what it is to be a reasonable person, as attentiveness to the nature and particularity of the object or situation being dealt with (Collier 2003).27 Such an understanding includes rather than excludes phronesis or practical wisdom oriented to questions of how to live. The de-rationalization of values involved in this form of reductionism complements the previous kind of reductionism discussed, regarding our neediness and concern-ful orientation to the world, for values are directly or indirectly related to our nature as needy beings capable of flourishing or suffering. To acknowledge someone’s need for something is simultaneously to make a positive (“world-guided”) judgment about them and a normative (“world-guiding”) judgment implying, ceteris paribus,28 that something should be changed. Thus the things which do not fall neatly into either side of the fact–value dualisms are not mere curiosities but precisely the things that matter most to us. The de-rationalization of values and the devaluation of reason therefore imply a profoundly alienated social science. Reductionism and Responsibility, Explanation and Justification As we noted at the outset, one of the most powerful forms of reductionism, particularly in common-sense thought, is individualism and its tendency to treat individuals as solely responsible for their character and fate, radically underestimating the extent to which who we are and what we become depends on social circumstances beyond individual control. Thus, crime is viewed as purely a matter

32

of individual responsibility, and there are no social causes or influences. Similarly, individuals are considered to be responsible or largely responsible for their class position. One of the most valuable functions of social science is in countering this reductionism. In so doing, however, it is tempting to flip over into a converse form of reductionism, social determinism, which considers that individuals have no influence or responsibility. This is an untenable position that generates theory–practice contradictions, for like everyone else, in their daily practice, social determinists hold others personally responsible for their conduct—for example, when they mark students’ essays, or cooperate with and rely on colleagues. Individualism tends to be favored by the right, and social or structural determinism by the left, though of course there are many intermediate mixes. One problem for social determinism in social research is precisely the irreducibility of individual reflexivity to its preconditions, so that it can only be roughly predicted on the basis of the latter. Thus some actors may behave in ways not expected in the light of their social and discursive positioning. The difference that individual responsibility makes then appears as a source of “noise” in the data, obscuring the assessment of social influences. Despite its enormous moral significance, it thus tends to be treated as a nuisance rather than an object of interest in itself.29 On certain occasions, despite their emphasis on social determinants of behavior, radicals have to acknowledge the limitations of social determinism and draw back from it. An example that has lately been much discussed concerns social explanations of terrorism. The Right tends to respond to these by complaining that such explanations do not justify terrorism, as if this meant that they could not stand as explanations. The Left responds by saying that they are not justifying terrorism, merely explaining it (Butler 2004; Williams 2003, 2004). Although I think the distinction is legitimate, it depends on a view of responsibility which rejects thoroughgoing social determinism and allows some room for individual (and group) responsibility. This may be uncomfortable for those who have thoroughly decentered and effectively disempowered the subject, but it is an instructive counter to precisely those kinds of reductionism. Yet while the individualist and social determinist positions are easy to criticize, the problem of how we should allow

33

for both individual responsibility and social determination, if not determinism, is immensely difficult. While it is much discussed in philosophy, it is again simultaneously both a positive and a normative matter, requiring an understanding of how individuals and their social context, including their moral sensibilities and character, co-evolve from birth, so that it is possible to assess what is reasonable to expect of them in the circumstances.30 It represents one of the most pressing challenges, not only for philosophy or indeed for social science, but for how we understand ourselves and others in everyday life.

CONCLUSIONS

To claim that some description or explanation is reductionist in a pejorative sense is to say that it misidentifies its object, either by explaining it at the wrong level, ignoring its emergent powers, for example, the sense of pride or shame by reference to physiological changes associated with them, or by omission of important co-determinants of a multi-causal situation and by attributing the effects of those omitted elements to the remaining elements that are acknowledged. Critical realism offers an anti-reductionist philosophy based on an ontology of stratification and emergence. By contrast, philosophies which assume atomism and regularity theories of causation fail to explain emergence, and hence evolution and novelty. Approaches assuming a flat ontology, in which the physical, the biological, the social are all on the same level, in a relational field, fail to address the asymmetric relations among these processes, noticing only their interaction. Critical realism further allows us to incorporate and combine the insights of both naturalist treatments of causation and hermeneutic or interpretive approaches to social research. Nevertheless, it remains a project of substantive research to identify specific kinds of strata and emergent properties, and particular forms of interaction among mechanisms at different strata, and of co-evolution. I have selected just a few forms of reductionism in social science for discussion, and I have paid more attention to forms of “upward reductionism,” such as

34

strong social constructionism, than the more familiar “downward reductionism” of approaches such as sociobiology, not because they are any more dangerous, but because they represent unsatisfactory responses to the problems of the latter. In social science, fear of a deterministic form of essentialism which naturalizes what is actually a (co)product of particular social and cultural forms has driven many scholars toward strong social constructionism and anti-essentialism. In the process, the baby of emergent causal powers and susceptibilities has been thrown out with the bathwater of determinism or essentialism. I have also discussed further, less widely-recognized forms of reductionism in social science, that relate to, firstly, the neglect of people’s evaluative or care-ful or concern-ful orientation to the world, by virtue of their neediness and susceptibility to both flourishing and suffering; secondly, to the de-rationalization of values; and finally, to the vexed question of responsibility. It should, I hope, be clear that while, in the abstract, reductionism seems a purely philosophical matter, it has great political significance, for it has major implications for what kind of beings we understand ourselves to be, and for what we can hold ourselves responsible, and how we should respond to others. Finally, I hope that it will also be clear that if we are to correct such problematic forms of reductionism, it is vital to put aside the disciplinary imperialism that nurtures reductionisms and instead to adopt a postdisciplinary approach.

NOTES

1. I have argued elsewhere that while the disciplinary divisions of natural science have an ontological rationale, those of social science do not and should be replaced by a postdisciplinary social science (Sayer 2000a). 2. Often it is difficult to decide whether the reduction is horizontal or vertical: the problem of class reductionism, for example, could be viewed as a form of horizontal reduction, ignoring influences such as those of gender and ethnicity, but which perhaps exist on the same level as class.

35

3. I add this point because in actor network theory and some regions of cultural studies it is imagined that we can dispense with any concept of cause in accounting for phenomena. Such a view is untenable in the strong sense, as we have to contradict it as soon as we have to offer any explanation of what produces a particular change, such as a change in what people believe and how they behave. Sometimes the attempt to reject causation seems to be based on the belief that causal explanation is about discovering single original causes or founding moments (which would clearly be reductionist); most changes of interest to the social sciences and humanities have multiple causes, and evolutions, and their explanation does not require an infinite regress. Power in the Foucauldian dispersed sense is only intelligible as the sum of the exercise of the causal powers of all the objects in a particular field (Sayer 2004). 4. This work generally requires attempts to purify substances and processes, which otherwise are normally characterized by variation, albeit to varying degrees. As Dupré argues, contrary to atomism, this variability appears to “go all the way down” (Dupré 2002). 5. In sociology it takes the form of interactionism, in which action or practice is explained without reference to the properties of structures or fields (see Bourdieu’s critique of interactionism in understanding markets (Bourdieu 2005)). This may be coupled with a flat ontology in which structure and agency are conflated and stratification and emergence are denied (Archer 1995). Archer terms this “central conflation,” in contrast to “downward conflation” (treating agency as epiphenomenal and collapsing it into structure) and “upward conflation” (treating structure as reducible to individual agency). 6. It is often also coupled with an instrumentalist rejection of the concept of truth, according to which we hold certain beliefs, theories, etc., not because they are true but because they are useful. This may sound appealing, until one asks whether we believe that six million died in the Holocaust because it’s useful to do so or because we think it’s true. 7. Some critics of mainstream economics argue that its insistence on retaining its restrictive assumptions and abstractions is driven by its fetish of mathematical formulation of solubility; any assumptions which cannot provide the basis of analytically soluble models are rejected out of hand regardless of their explanatory power (Fleetwood 1999; Lawson 2003). 8. Bhaskar (1975) separates predictive forms of reductionism from explanatory forms. I have conflated them as the differences do not bear upon my argument. 9. This possibility is assumed in the comical idea, often advanced in newspapers, that physicists are on the verge of being able “to explain everything.”

36

10. As the reader may suspect, my use here of “social/cognitive, etc.” is an evasion of difficult questions regarding the status of the social, the human, the cognitive, the rational. I shall address this later, but at this point it does not prejudice the current argument. 11. It is important not to go overboard regarding this play of difference and scope for novelty by ignoring the practical character of language and its embedding in mostly routine practice. The relationship between the words, referents and contexts and actors are mostly fairly stable, so that the meaning of particular words has at least some stability and limits as to its sense. It is hardly surprising that when language is disengaged from our practical involvements in the world and used primarily for reflection, it becomes more unstable. 12. Even though physical processes can be altered by human intervention guided by cultural understandings, they can only be altered by physical interaction, and according to their culturally indifferent properties, and they do not become internally culturalized objects like persons, capable of cultural variation. Thus the chemical industry may produce sulphur dioxide, but Russian sulphur dioxide is the same as British sulphur dioxide. 13. As Margaret Archer notes, Durkheim could not adhere consistently to this position in his own substantive work. For example, at occasional points in Suicide, he has to acknowledge psychological preconditions and causes of suicide. 14. This tends to go with a flat ontology, interactionist view (see below). 15. Although it may be described as a “deconstruction” of the nature–culture distinction, it arouses the suspicion of a form of anthropological imperialism, in which the natural is reduced to the cultural. 16. Or more accurately, “concept-laden” and “concept-determined.” 17. They need much qualification, however, regarding just how extensively selfhood is diffused, how the claims are to be reconciled with reflexivity and responsibility (see Archer 2000, 2003). Without such qualification the mind-body may be emptied of all content and filled with “social foam,” as she puts it (2000, 317). 18. A moderate constructionism accepts the reality of social constructions, but distinguishes them from the construals that inform and interpret them, and acknowledges the fallibility of attempts at social construction. 19. “The real world is not a dream-world under my control, a fiction my language has created and can mould to . . . my convenience. Reality is more resistant. At many levels, things are not always as I want them to be, or conceive them to be. Realism has to start with the realization that I, or anyone else, can be wrong. Fallibility is part of the human condition. Any view which denigrates reason, and the possibility of truth, and which even

37

doubts our identity, is in fact saying that there is no way we can be mistaken, either individually or collectively. Each epoch will have its own views, its fictions, and even that view will be a fiction. Even the notion of fiction collapses, since there will be nothing left to contrast it with. If we cannot be right, we cannot be wrong, and if we cannot be wrong, we cannot be right. When everything is linguistically constructed, language itself will collapse. So far from the differentiation between subject and object being the consequence of a concentration on language, language itself depends on it” (Trigg 1998, 159). 20. To explain how other species, e.g., dolphins, chimpanzees, seem to exhibit cultural variation, we would again have to identify what it is about them that makes them capable of or susceptible to this. 21. And by Max Weber, albeit unsatisfactorily. 22. Kathryn Dean uses Freud’s concept of “cathexis” as involving “culture’s ‘occupation’ of the biological by way of the emergent subject’s more or less active response to an ‘invitation’ to attend to a part of the world in a particular manner” (Dean 2003, 19). At the same time agency itself can be promoted or limited by culture. 23. By their very nature, needs, lack, and desire need not be satisfied. The fear of teleology is all the more likely where “the subject” is treated as singular rather than multiple so that it appears that social development might have a predetermined goal or destination, as well as where essentialism is misunderstood as a doctrine which argues that the essences or powers of objects unilaterally and deterministically produce certain effects (Sayer 2000b). 24. It is particularly associated with the rise of markets, which are a “reason-neutral” sphere of social action in which buyers do not have to justify their “preferences” in order to buy: all that is needed is the money. 25. Unless we reduce rationality to logic, though as we shall see, that reduction is part of the problem. 26. Theorists of early modernity, such as Adam Smith, predated this tendency to imagine that the normative must be divorced from the positive and that any evaluative content to descriptions could only weaken their objectivity. 27. “Reason is the capacity to behave consciously in terms of the nature of what is not ourselves. We can express this briefly by saying that reason is the capacity to behave in terms of the nature of the object, that is to say, to behave objectively. Reason is thus our capacity for objectivity” (Macmurray, quoted in Collier 2003, 158). While I think this is valuable, it has to be added that sometimes, and particularly where values are concerned, the object is ourselves.

38

28. There may sometimes be overriding considerations which allow us to acknowledge the need, but say that nevertheless it should not be met (Taylor 1967). 29. Conformity to social influences may also depend on reflexivity and the exercise of a sense of moral responsibility. 30. For a discussion in relation to matters of social justice, see Barry (2005).

REFERENCES

Anderson, E. 2004. “Uses of Value Judgments in Feminist Social Science: A General Argument, with Lessons from a Case Study of Research on Divorce.” Hypatia 19 (1): 1–24. Archer, M. S. 1995. Realist Social Theory. Cambridge: Cambridge University Press. ———. 1996. Culture and Agency. 2nd ed. Cambridge: Cambridge University Press. ———. 2000. Being Human. Cambridge: Cambridge University Press. ———. 2003. Structure, Agency and the Internal Conversation. Cambridge: Cambridge University Press. Barry, B. 2005. Why Social Justice Matters. Cambridge: Polity. Bhaskar, R. 1975. A Realist Theory of Science. Leeds: Leeds Books; 2nd ed. 1978, Brighton UK: Harvester. ———. 1979. The Possibility of Naturalism. Hassocks: Harvester. Bourdieu, P. 1988. Homo Academicus. Cambridge: Polity. ———. 2000. Pascalian Meditations. Cambridge: Polity. ———. 2004. Science of Science and Reflexivity. Cambridge: Polity. ———. 2005. The Social Structures of the Economy. Cambridge: Polity. Butler, J. 2004. Precarious Life. London: Verso. Collier, A. 1994. Critical Realism. London: Verso. ———. 2003. In Defence of Objectivity. London: Routledge. Dean, K. 2003. Capitalism and Citizenship: The Impossible Partnership. London: Routledge. Dupré, J. 2002. Human Nature and the Limits of Science. Oxford: Oxford University Press. Fairclough, N., Jessop, R. and Sayer, A. 2002. “Critical Realism and Semiosis.” Journal of Critical Realism 5 (1): 2–10. Fleetwood, S., ed. 1999. Critical Realism in Economics: Development and Debate. London: Routledge.

39

Harré, R. 1970. The Principles of Scientific Thinking. London: Macmillan. Heidegger, M. 1962. Being and Time. Oxford: Blackwell. Helm, B. W. 2001. Emotional Reason: Deliberation, Motivation and the Nature of Value. Cambridge: Cambridge University Press. Ingold, T. 2000. The Perception of the Environment. London: Routledge. ———. 2005. “Brereton’s Blandishments.” Journal of Critical Realism 4 (1): 112–27. Lawson, T. 2003. Reorienting Economics. London: Routledge. Nussbaum, M. 2001. Upheavals of Thought. Cambridge: Cambridge University Press. Oakley, J. 1992. Morality and the Emotions. London: Routledge. Putnam, H. 2002. The Collapse of the Fact–Value Dichotomy. Cambridge MA: Harvard University Press. Sayer, A. 1992. Method in Social Science. London: Routledge. ———. 2000a. “For Postdisciplinary Studies: Sociology and the Curse of Disciplinary Parochialism/Imperialism.” Pp. 85–9 in For Sociology: Legacies and Prospects, ed. J. Eldridge, J. MacInnes, S. Scott, C. Warhurst, and A. Witz. Durham UK: Sociologypress. ———. 2000b. Realism and Social Science. London: Sage. ———. 2004. “Seeking the Geographies of Power.” Economy and Society 33 (2): 255–70. ———. 2005. The Moral Significance of Class. Cambridge: Cambridge University Press. Soper, K. 1995. What is Nature? Oxford: Blackwell. Taylor, C. 1967. “Neutrality and Political Science.” Pp. 139–70 reprinted in The Philosophy of Social Explanation, ed. A. Ryan, 1973. Oxford: Oxford University Press. Trigg, R. 1998. Rationality and Religion: Does Faith Need Reason? Oxford UK; Malden MA: Blackwell. Williams. G. 2003. “Blame and Responsibility.” Ethical Theory and Moral Practice 6: 427–45. ———. 2004. “Praise and Blame.” Internet Encyclopedia of Philosophy: www.utm.edu/ research/iep.

DISCUSSION

Reductionism has come under criticism in all the disciplines, but often using different language and often pinpointing different issues. IMMA N UEL WA LL E R ST E IN:

40

Part of the problem is simply to see how much coherence there is in the critiques of something called reductionism in the natural sciences, in the humanities, and in the social sciences. I do not think we all agree, that seems to me clear; maybe we will agree more at the end. And I think it would be a bit fruitless to simply define reductionism at the beginning. Certainly, people are using it in slightly different ways. Perhaps we can see whether those slightly different ways are as different as we seem to think at the beginning. Karl Popper, the famous British epistemologist, once said “never start with the finish.” I mean, the definition is the result of the reflection, not the premise. Popper also wrote that there is no science that does not aim at reduction. That is the first proposition. Second proposition: all reductions fail to some extent. Third proposition: some of those failures are interesting and they help science to move forward. So you see, that is the gist of the argument. If we stop trying to reduce A to B then we do not do science anymore. So we must aim at that: It is something on the horizon that we will never reach. And the failures: for instance, molecular biology is a failure because we have reduced biology to physics that is no longer the kind of physics we were accustomed to before. It is a physics in which we have circulation or flow of information, in which we even have meaning because we have translations, etc.

J E AN-PIERRE D UP U Y:

IMMA N UEL WA LLE R ST E IN: I absolutely agree that every time we speak about anything we are being reductionists. Certainly anytime we try to explain anything, we are being reductionists. And if by reductionism we mean that we are omitting certain aspects of things in order to get at what we think is the essential of some matter, I do not think that is the end of the game. It seems to me that the whole point is to come up with a good explanation of complexity.

I think the central question is the one of levels. There are different levels and I want to ask if those levels are really ontologically accessible or whether they are a function of descriptions of the world. We can have different criteria for identifying them and I am going to be pushing slightly toward flat ontology.

PAUL CILLIERS:

41

I would be very interested to hear you elaborate the flat ontology view. Basically the levels are defined in terms of preconditioned existence. Can A exist without B, can B exist without A? What is to be made of the asymmetries we get in answering those questions about conditions of existence?

ANDREW SAYER:

The assumption seems to be that emergence implies levels, and I am not so sure that is basic. I totally agree, however, about the forms of asymmetry. But for me there is asymmetry within the levels, not between them.

PAUL CI LLIERS:

But the primary purpose of the social sciences is to study social relations. We do not have to argue about whether we can have soldiers without armies or armies without soldiers. They presuppose one another. The sole existence of armies and soldiers depends upon social relations. One does not have to choose between wholes and individuals. It is the particular form of internal relations which develop among individuals and what follows from that. How we think about reductionism depends upon how we feel about related issues such as the relationship between prediction and explanation. I think that most social scientists would say they are not symmetrical and that it is not necessary to predict in order to explain, even in non-explanatory predictions. The primary goal of social science is not to predict because one cannot. When one does predict, one usually does so without much in the way of explanation. Good extrapolation is often the best way of prediction in the social sciences, but it is usually pretty inadequate as an explanation. I do not know whether people from outside the social sciences expect social scientists to predict. I do not think many social scientists do. Then there is the relationship between mathematical models and cause in explanation. For me mathematics is an acausal language. It does not, as a language, have any concept of producing, forcing, and enabling change. It is a different kind of language and it is useful, obviously. But I wonder if some people are coming with different ideas about the relationship between mathematics and causal explanation here. We could be talking at cross purposes if that were the case.

AND REW SAYER:

42

I would like to propose that between the level picture and the flat ontology, there is actually another possibility for which I do not have a word but maybe I would call the contoured model rather than the stratified or the flat one, and that would be one where we would have different spatial temporal processes working in different spatial temporal webs of interrelations. They would have different morphologies or morpho-dynamics, to use Jean Petitot’s term, and that if we were to think along those lines we would actually have to do some work to reconceptualize what reductionism means. But it is crucial to realize that neither emergence nor reduction really presuppose either the flat ontology or the stratified one. There can be forms of emergence within contrary dynamics and morphologies where we would not have to conceptualize in terms of a base level, a level on top of that, a level on top of that, and so on. EVA N THOMPSON :

I thought the part of your paper that addressed disciplinary imperialisms was very important and I was very aware of how marked one’s discourse is by the discipline from which one comes, because certain claims made in your paper strike someone from my discipline, literary studies, as almost absurd. But I recognize it; it is not an absurdity in itself, it is the marker of disciplinary location. For example, literary studies is still very marked by Lacanian psychoanalysis, so to say there is inadequate attention to that would seem almost absurd from this disciplinary perspective. But I do think there is an implicit contradiction in the ontological status you give to different levels and your very trenchant remarks on disciplines, because I agree that the levels very much correspond to established disciplines. So it seems a bit incongruous on one hand to criticize disciplinary imperialism while reinscribing in an ontological sense the levels instituted in a discipline. I would like to just add one more comment, and this may also be related to disciplinary markers, but I found almost unrecognizable your description of strong constructivism. It differed so remarkably from my understanding of strong constructivism that it seemed almost a straw man set up in your paper. I think the idea that strong constructivism does not recognize any

K ATH ERIN E H AYLE S:

43

kind of constraints originating from chemistry, biology, and so forth is absolutely wrong. In my understanding of the program, strong constructivism, is, on the contrary, about trying to present a more integrated model. Indeed, I think that I could name any number of people working in science studies who would not agree at all with your description of strong constructivism. Strong social constructionism is what I encounter quite a lot in sociology. I am open to the idea of thinking in terms of different spatio-temporal levels. But I would still want to know whether or how asking questions about preconditions of existence of different kinds of entities that may be at different scales works out. AND REW SAYER:

In biology if we look at the interaction between species and ecosystem, we see that species shape the ecosystem, so they cannot be completely divorced from one another. The evolution of a particular trait is not completely divorced from the effect that particular trait can exert on the external environment. Nature has created some very strong boundaries between different levels. I would like to hear you speak to the issue of, on one hand, the mixing between the levels when we are dealing with interaction of the species with its environment and, on the other hand, the existence of very hard boundaries between the species and the individual.

AV IV BERGMA N:

We agree that the problem with the notion of flat ontology is the assumption that it means a homogeneous flat fit. When we talk about reality as being of one level, we mean it is very structured. The issue is one of boundaries, not levels, and there are such boundaries between levels, boundaries between different structures in the system. We get the entities by structuring a network instead of layering the levels.

PAUL CILLI ERS:

If from a naive evolutionary perspective one decided to stick to the notion of the strong ontological existence of levels, then it becomes impossible to

ISTVA N REV:

44

avoid reductionism. This is why I agree that it would be more fruitful and more interesting to talk about boundaries instead of levels. On the other hand, existing disciplinary boundaries are expressions of our notions of a layered existence. The different disciplines are expressions of the different levels we perceive. The existence of these disciplines reinforces our perception of the different layers and levels. There is a very strong interdependency of our institutionalized existence in science and our perception of the world. It is difficult to get rid of this rigid perception because of our institutionalized existence, and this forces us to use reductionism even when we try to avoid it in a naïve, simple way. I would like to pause to think critically and carefully about whether it is quite so easy to say there are the natural sciences and then there are the social sciences. In fact, we also find in the social sciences something else, for example, climate change. What are the levels and what is the distinction between the natural and the social? There are physics, geology, and chemistry. Then there are all the results of human agents and their industrial productions.

EVA N TH OMP SON :

My personal attitude is to start from traditional concepts. I am not refusing the notion of levels, but the next step is to make them complex. Also, I do not fully agree with the idea that disciplines are characterized by their object belonging to a certain level. J E AN-PIERRE D UP U Y:

I am much in sympathy with what Thompson said a few minutes ago. But fifty years ago people would not have said the same thing. People said, yes, there are sciences here and over there are people who study the social world. They are not quite the same as the scientists, but they are trying to be like them. Then there are those who read novels over there. Thus, there was a disciplinary structure there but it was not written in stone, and it didn’t fall from the stars. Rather it emerged over a long period of time. If a certain structure did not exist at a certain time and then came into existence, it suggests that a question we have not thought about yet is the politics of the epistemology, although

R ICHA RD E. LEE :

45

it has been suggested that there might be a politics associated with reductionism. If we ask social scientists today if their work has predictive power, most would probably say that prediction is a very dangerous game. By the same token, there was a time when social scientists certainly did try to predict. Fifty years ago people asked how much investment would have to be made in Algeria for the GNP to be raised by X percent. People actually tried to do things like that. It is not that social science has never tried to predict. But today people are not willing to say that in order to study the social world one needs to be able to make predictions about it. However, the question associated with the politics of epistemology is: what kind of activity are we engaged in if it cannot be predictive about the social world; what can we say about the future? Most social scientists would think they need to say something about the future in order for their activity to have a real raison d’être. It is a quandary; what can we do if we cannot be predictive? I would like to make two comments: one on the paper, and one on the discussion about the paper. On the paper itself, I absolutely agree that there is no intellectual sense to the so-called disciplines within the social sciences as they exist today. In fact I have written on this theme many times. I would say that your idea of upward reductionism, which I think is fruitful and useful, is however in contradiction with the idea that there is no such thing as a flat ontology. It does have tonalities that suggest a flat ontology if there is an upward reductionism as well as a downward reductionism. In the discussion, people have been uncomfortable with the paper because of the idea of levels, and for levels the discussion has substituted boundaries. But where do boundaries come from? What we mean by a discipline seems to me at least two very different things. One is an intellectual construct, and the other is an organizational construct. The intellectual construct is what justifies calling physics, physics, sociology, sociology or literary studies, literary studies. One can ask the question whether intellectually it’s a defensible category. Organizationally we divide up our work because we all cannot do everything, so we say, I do primarily this and you do primarily that and those are the boundaries. Those boundaries are organizational and they IMMA N UEL WA LL E R ST E IN:

46

are temporary; they keep changing as the world changes, and as our knowledge about the world changes. It is not clear to me that fifty years from now there will be something called physics. There maybe something that merges physics and chemistry and biology all into one thing. I do not know enough about the subject matter to predict this but I can see it as an organizational possibility. Once upon a time there was zoology and botany, but nobody talks about them anymore. They have disappeared completely. Now there is biology. For me intellectually it is all one big blob. Organizationally we have to cut into it in different ways and the organizational ways are the ways that are useful, and what is useful today may not be useful tomorrow. So we temporally designate ourselves as X. Now the problem has a social dimension. It is that these temporary categories get ensconced in organizations, in their departments, in universities, and so on and people are very resistant to changing the organizational boundaries. It is hard work to finally convince botanists that there is no such thing as botany. It was done at one point. And it will be hard work to convince people to change. It is not an intellectual problem, it is a sociological problem. It is a reality that people resist this kind of change. But intellectually I am very dubious that there are clear boundaries. There are useful ways of chopping things up in our ongoing work. So we concentrate on problems. And it may well be that the study of climate may turn into a whole discipline. We may have a group of people called climatologists who would look at what we call now the physics of it and what we call now the sociology of it, and see it all as integrated. They would train other people to do it; they would set up departments to do it with PhD’s and so on. Institutes and national foundation awards would follow. The levels are heuristic devices at best. If we want to call them boundaries, call them boundaries; but obviously that is where I come back to my comment on reductionism and complexity. The end of the game is to take a very complex reality and have reasonably adequate explanations of this complex reality that cut through all the reductions. That is why I am anti-reductionist in a very fundamental sense. To me, reductionism represents the idea that when we have reduced whatever we are reducing to something else, we have gotten to some greater truth and some

47

greater reality. And that seems to me far from the case: It is something on the road to where we want to be. J E AN PETITOT: Perhaps it would be interesting to look at some precise examples regarding reduction and emergence. In natural science, the best is the physical example of phase transition. Of course there is a causal reduction. Causal reduction, higher-level phenomena, micro phenomena, and emerging phenomena can be causally explained from the lower micro level. So this is an example of epistemological reductionism. But at the same time, and this is a key point, emergent structures of collective behavior have some autonomy. There are mathematical theories showing that the properties of collective behaviors are largely independent of the fine-grained structure of the underlying physics. And in that case it is for very precise reasons we have symmetry-breaking phenomena and symmetries. Many different physical substrates can have exactly the same symmetry. So the emergent level can be defined by this theory. They are not epiphenomena; this autonomy is formal and it is impossible to deduce from that downward causation. So causal reduction does not entail ontological reduction.

J OÃ O CA RA ÇA : In this paper, emergence is not treated as the inverse of reduction, because we have this notion of upward and downward reduction. I also include the word “aggregation.” The importance of this notion of complexity lies in the fact that we have tremendous difficulty with the disaggregation of things. Indeed, how do we aggregate or disaggregate? I mean that traditional problems up until the eighteenth century were easily disaggregated. The problems of matter and motion, for instance, or a living being and its environment, these are pairs and we cannot set a boundary and say that the living being finishes here and environment starts there. The notion of complexity arises in the difficulty of setting boundaries, separating something from its whole. We cannot disaggregate one object from the whole. That is to say, we will just have to treat open systems. The notion of reductionism is interesting because it brings a value into it; the explanation is something that has a purpose. Indeed, we cannot explain the army by the soldiers because in fact

48

the soldiers are also the product of that army as a whole. Mathematics is a very important language because it is a language that is able to separate the two levels, the ontological from the epistemological. That is to say it separates very clearly the objects from the rules of the objects, how those objects function. I started out by trying to build dynamic simulation models of complex systems. I became dissatisfied with them because they did not actually explain anything for me. Rather they enabled me to calculate possible outcomes under certain initial conditions and parameters. But if I wanted to understand capital accumulation then I would explain that in terms of pressure of competition on employers; they need to accumulate in order to remain in existence and not go bankrupt and so on. There are quite simple qualitative descriptions of causal mechanisms and that is a far different thing from calculating what their effects are. Just because things are all interacting does not mean we can ignore the fact that there are strong asymmetries in terms of their conditions of existence. Conduction currents, for example, can exist regardless of whether we have a love affair with the automobile.

AND REW SAYER:

Durkheim, for instance, does not talk about the individual on the one hand and the multitude on the other or the individual here and the congregation there; he talks about the orator. And the orator is not an individual, but rather an orator is somebody who is in large part defined by the context in which he stands. The crowd over him is also in large part defined by the context. It is not methodological individualism on the one hand and holism on the other. ISTVA N REV:

U L L I C A S E G E R S T R A L E : Somebody [Richard Lee] brought up the notion of politics of epistemology, which I think is extremely important. Of course we have politics of epistemology; I find it very interesting that when you [Andrew Sayer] talk about reductionism toward the end of your paper, you talk about the need to consider background reason in values and background values in rationality. But I wonder if that is something which is very specific for social scientists, because we are dealing with human beings and we have more political responsibilities?

49

Two things in your paper that we have not talked about that much are the relationship between facts and values and the problem of agency. The relationship between fact and value, I think, is a very tricky issue and it seems that you insist on some notion of agency, otherwise it is impossible to ascribe any form of moral responsibility. If we have this as a sort of homogenous constructionist view, it seems difficult to think of something that can be responsible. But at the same time the notion of the individual seems to me to be one of the most reductionist concepts we have. Because we have experienced ourselves as individuals, it is a very seductive one. In any case I do think it is an interesting problem to ask how we think of agency without introducing an atomistic understanding of the individual. And I think the network models we have been talking about can address that issue. PAUL CILLI ERS:

AND REW SAYER: I do not have a problem with the idea that particular claims are value free. What I do have a problem with is the idea that making those claims, actually stating them as an agent, an actor, is value free; one makes claims in particular social contexts and which can have a very damaging effect. Say some terrorists burst into this room and said, which one of you is Immanuel Wallerstein, we want to kill him, and I said, as a matter of fact he is sitting right next to me. The actual fact is value free but the act of making that statement on that occasion would not be value free. And as regards agency, why do I insist on agency? I think one way of testing any kind of proposition is whether we can live it; it is a sort of Aristotelian idea that if we cannot practice an idea then we should just treat it as mere words. And it is nothing more than mere words. We hold each other responsible for lots of things and social life is unimaginable without doing that. You have held me responsible to produce a paper. You hold me responsible for the views and so on. But my view of individuals is also a social view. We are thoroughly social beings, thoroughly dependent on others, thoroughly vulnerable beings, psychologically in need of recognition and of approval. Our personalities are shaped through our relationships with others and so on. Now, I like Adam Smith’s conception of the power of social being which is often forgotten by neo-

50

classical economists and based upon relations having a plastic fellow feeling or sympathy, as he called it. PAUL CI LLIERS: I would like to put it even a little bit stronger. It seems as if you say that we are individuals that have social needs. I think it is even stronger, in the sense that we are constituted socially. AND REW SAYER:

I am happy with that.

The idea of single human beings is a self-contradiction. We are constituted socially. Therefore we are relationally constituted. We are relations, not something pre-human.

PAUL CILLI ERS:

We started by saying that ultimately to some extent we need to reduce, need reductionism. That it is part of science. That it is something fundamental. And so far the way this talk has evolved has been to give some pretty good examples of failure of reductionism and of systems which are not reductive. A reductive system would be a system in which we travel this hierarchy in only one direction, from top toward the bottom. But now we are saying that this does not happen. Actually we can pinpoint or exhibit objects which show that, right? Then the question seems evident. It is, how do we deal with it? We have on the one hand this regulative ideal of reductionism and on the other hand we have these objects, which at first sight at least and perhaps more than at first sight, we cannot reduce. So how do we deal with this?

PAUL D UMOUCH E L:

IMMA N UEL WA LL E R ST E IN:

You could try to give us an answer.

Well I do not really have an answer but I do have some ideas. Among other things, what has happened in this discussion is that reductionism seems to be centered on explanation. To reduce is to explain in a certain way. And these explanations will refer to different levels; they will not necessarily travel

PAUL D UMOUCH E L:

51

in one direction between these levels, but actually travel in a more complex way. So it seems that we need to distinguish what makes a particular type of explanation a reduction instead of simply being an explanation. Because I think we have a tendency sometimes to believe they are both the same thing and then they are different. But we have a hard time pinpointing what makes that difference. Isn’t the problem, that we are into a dualism of structure and agency or wholes and parts which is not real? The structure is obviously composed of a multitude of agents but the structure also has created those agents at the same time. And to constantly talk as though we were moving from structure to agent, from agent to structure seems to me the ultimate irrealism because they are a whole. They are integrated, but each, to use Jean Petitot’s terminology, with a level of autonomy. So the whole has a level of autonomy, so do the multiple agents or the multiple units, or multiple whatever terminology. But the fact is that these are not separable. That is what Istvan [Rev] was just saying. The orator is not separable from the congregation, the congregation from the orator, because they constitute the situation in which there is a religious ceremony with someone standing up there talking to all the other people present. That is what a religious ceremony is. And the member of the congregation and the priest or orator are both constituted by this social situation—which is a social situation. I want to remind you that we talked about politics and epistemology, which I agree with, but I would rather use a different terminology. We have a debate here about social constructionism between your rejection of hard social constructionism and Katherine Hayles saying she did not recognize the description of social construction. I am not sure what a strict social constructionist is. Are there other kinds of social constructionists? Can one be a weak social constructionist? If so, I want to put myself in the category of a weak social constructionist. It seems to me that the categories we use are socially constructed. And it can only be explained in fact by the overall social situation in which we find ourselves. So all this terminology, from structure and agency or physics and sociology, is the consequence of a historical process which occurred in a certain way and has determined the languages which we are using now. As IMMANUEL WALLERSTEIN:

52

Richard Lee suggested, they are not the same languages that we used fifty years ago and as the very meaning suggests they certainly are not the languages used in the middle of the nineteenth century and if one wants to go back, certainly not the languages used in the seventeenth century or those that were used in China. I remind you that what people thought was fantasy in the seventeenth century and what people think is fantasy today are different things. Newton was serious about many things that we would think are absolutely unscientific today. But they did not think it was unscientific in Newton’s time. The very definition of science is itself a totally evolving reality. And we just have to constantly keep in mind, it seems to me, that our discussions are within the context of the world in which we are living now, and in which we may not be living in the future. And in which certainly we were not living in the past. K ATH ERIN E HAYL E S: Well I would like to return to the topic of the politics of epistemology. I think that the description of levels is very much a political description of the situation. You bring into light how that description is laden with politics, but what if we reverse the politics of that description and say level B is more complex than level A because it has the ability to mobilize the powers of level A to create a cell. So the cell has the power to mobilize the chemicals in order to maintain itself. Humans have the ability to mobilize the power in the cell, in that humans have the ability and technology to mobilize the individual atoms and so forth. So that account would give a very different kind of structure. It would privilege the ability of the higher levels to mobilize the powers of the lower and would result in a kind of inverse ontological politics where the emphasis now is not on what came first, but what is able to mobilize the powers of the others.

I am uneasy with the use of terms such as mobilize the power of chemicals or mobilize the power of methodology, because it implies a sort of political, social, and moral relationship. And I do not think that there is such a relation between the cells and the mechanical processes the cell uses. I know that this is a linguistic convention but we live in this world and because of that I have

ISTVA N REV:

53

very serious problems using such language. Besides that, I would like to refer back to the problem of constructivism. Yes, from a certain perspective we can say that all the entities of our enquiries are socially constructed. But on the other hand, it makes a difference whether we deal with something internal to a boundary or something that is external to a boundary. Let me give you an example. If we deal with individuals in the framework of our social inquiry, then we deal with social actors. In one sense, this can lead to methodological individualism. But if we deal with problems related to medical science and if we try to treat the organism as a social entity, then we might commit the fallacy which the paper calls upward reductionism, because this is internal to the boundary of our investigation. Before I let Prof. Sayer have his say, I would like to intervene about Istvan Rev’s example of medical science and things that are external to the boundaries. Let us take what some people say is a very recent medical finding. They have now decided that there is such a phenomenon as therapeutic dogs. In the case of patients with various kinds of problems, allowing therapy dogs to visit them in the hospital seems to have a positive effect, that people who have these visits from the dogs have a higher cure rate than other people. That seems to me something that is external to the boundaries of medical science as we have known it. Maybe it is now a revision of the boundaries. What it suggests is that there are certain mental states that can be created as a result of the visit of the dog which have an affect on the physiology etc. And the case is being taken seriously by an X number of physicians. Medicine is a very good domain in which to see that boundaries as defined by the physicians at a given time constantly are broached by new findings. That has been the whole history of medicine for the last several hundred, if not several thousand years; we really know so little about medicine that it is easy to broach the boundaries. New things get discovered constantly. So I wonder whether that is realistic to make the distinction.

IMMA N UEL WA LLE R ST E IN:

AV IV BERGMA N : However, there are boundaries in natural phenomena that are unavoidable. This is something that cannot be pushed away.

54

I think there are two important aspects of boundaries that one can keep in mind. The first is to think of boundaries not as something exclusionary but as something enabling. So the boundary makes possible, if enabled. The other aspect is the question of where is the boundary, the place of the boundary? If we think in terms of a network, if we think of a complex system in terms of a connected network, then there are few states that compose the system. Then it becomes very difficult to think of a boundary in spatial terms as something out there. PAUL CILLI ERS:

AV IV BERGMA N : Considering boundaries as an enabling condition, the question is, what is enabled? To my mind what is enabled is the beginning of communications. If the two of us did not have the boundary we would not communicate. What is missing from the discussion here is most of all communication.

My major work is on the modern world-system. I believe systems exist and it is central to my own work, and that system has boundaries. One of the points I make about that system is that the boundaries have been changing over time and that is built into the system. There are reasons why the boundaries have spatially expanded over time, as those boundaries are not fixed. That is very important. I do not know about and I am not confident to talk about genes, but genes did not always exist. They evolved. To have boundaries, we must think in terms of boundaries; they are crucial but they are not fixed. Boundaries evolve over time. In fact, at least in social science, it seems to me the evolution of boundaries is basic, a finding constantly reoccurring at every level of our analyses.

IMMA N UEL WA LL E R ST E IN:

I think one kind of asymmetry has to deal with realism, actually. Because the most primitive concept of realism is the idea of objects and that the world exists in some sense independently of our observation of them. And that is obviously something which might be difficult to reconcile with the social world. Let us think about different cases; take geology, for example. The nature of AND REW SAYER:

55

the earth’s surface and the rocks therein is whatever it is, has been whatever it is, regardless of what our concepts are about it. We may influence geology, we may be able to trigger earthquakes but we trigger them by our own physical mechanisms that we have available to us, not just by having a concept of earthquakes. It requires actual physical intervention to produce some change in geology. So it has this very strong form of independence and we do not get this kind of downward interaction in other than the physical sense. Whereas a social phenomenon like money depends on the concept of money. We must have a concept of money. And so language is constitutive of social reality. But in order to be a moderate social constructionist—I think that I am a weak social constructionist—we think about what construction involves. It succeeds or fails according to whether the constructor takes account of properties of the objects, the materials that are used in a construction process. George Bush is a social constructionist in a sense, but he is not my social construction. My attempt to reconstruct him has completely failed. He is independent from me. I lack the political means to make a difference to him. He is independent. There are those kinds of asymmetries still within the social world. So the metaphor of construction should be taken more seriously because then we can have a more sensible view of what involves a non-idealist view and a non-voluntarist view; it does not just involve the idea that wishful thinking necessarily works.

56

SESSION II

Emergence and Complex Systems1 Evan Thompson

T

he aim of this paper is to ponder a number of issues about emergence that have occupied recent philosophy of mind and philosophy of science. I will focus in particular on the concept of emergence in relation to complex (nonlinear) dynamical systems. In this context, an emergent process is said to be one that belongs to an ensemble or network of elements, arises spontaneously or self-organizes from the locally defined and globally constrained interactions of those elements, and does not belong to any single element. This paper falls into two main parts. In the first part, I explore the concept of emergence in the context of complex systems theory. In the second part, I build on this exploration in order to reply to Jaegwon Kim’s well-known criticisms of the concept of emergence (Kim 1999). Discussions of emergence are structured by the classical opposition between reductionism and emergentism (Silberstein 2002). These positions have epistemological and ontological components. Epistemological reductionism states that the best understanding of a system is to be found at the level of the structure, behavior, and laws of its component parts and their relations; ontological reductionism states that the relations between the parts of the system are all reducible to the intrinsic properties of the most basic parts. Epistemological emergentism states that the best understanding of a system is to be found at the level of the structure, behavior, and laws of the whole system, and ontological emergentism states that a whole is more than the sum of its parts and their intrinsic properties.

57

These formulations presuppose a dichotomous conception of parts versus wholes, or basic features (conceived as intrinsic properties of microlevel particulars) versus emergent features. Thus, if part and whole co-emerge and mutually specify each other (as suggested later in this paper), then such a dichotomous conception will need to be revised. Beyond the classical opposition between reductionism and emergentism tricky issues arise, because there are now so many different senses of “reduction” and “emergence” in the scientific and philosophical literature. Many different kinds of phenomena have been considered emergent, and different epistemological and/or ontological criteria of emergence have been used to classify them as emergent.2 It seems impossible, therefore, to search for any single and neatly analyzable concept of emergence independent of particular explanatory contexts. This is one reason why I choose to focus on the notion of emergence implicit in complex systems theory, and the application of this notion to biological and cognitive phenomena.

I EMERGENCE AND COMPLEX DYNAMICAL SYSTEMS

A Working Framework Let me begin by setting up a framework in the form of the following provisional definition of emergence in the case of complex systems (revised from Thompson and Varela 2001). This definition is meant to capture the main features of emergence that researchers in complex systems theory seem to have in mind when they talk about emergence: Definition: A network, N, of interrelated components exhibits an emergent process, E, with emergent properties, P, if and only if: (1) E is a global process that instantiates P, and arises from the coupling of N’s components and the nonlinear dynamics, D, of their local interactions.

58

(2) E and P have a global-to-local (“downward”) determinative influence on the dynamics D of the components of N. And possibly: (3) E and P are not exhaustively determined by the intrinsic properties of the components of N, that is, they exhibit “relational holism.” Terminology: Emergent Processes and Properties Although the term “emergent property” is widespread, I prefer “emergent process.” Strictly speaking, it does not make sense to say that a property emerges, but only that it comes to be realized, instantiated, or exemplified in a process or entity that emerges in time. Emergence is a temporal process, but properties (whether considered as universals or as linguistic abstractions) are atemporal. For instance, the property of being alive did not emerge when life originated on Earth, but came to be realized as a result of the emergent process of molecular self-production or “autopoiesis” that constitutes living cells (Bitbol and Luisi 2005; Bourgine and Stewart 2004; Maturana and Varela 1980). This example points also to the importance of the global and context-shaping characteristics of emergent processes: The emergent network of autopoiesis constitutes a biological individual (a cell) that brings about changes in the external environment; it also creates a structured context in which new kinds of events can take place, such as protein synthesis and RNA/DNA replication, which cannot occur apart from or outside of the protected intracellular environment. Proposition 1: Nonlinear Dynamics The emergent processes covered by this definition occur in networks whose coupled elements have nonlinear interactions. The distinction between nonlinear and linear interactions provides one way of distinguishing between systems that have emergent processes and systems that do not (Campbell and Bickhard 2002). Linear interactions are additive or proportional. They typically give rise (except in the quantum domain) to systems that are “aggregative” (Wimsatt 1986)—ones whose

59

causal features derive from aggregating the properties of the components. Nonlinear interactions are nonadditive or nonproportional. They give rise (by definition) to systems whose activities cannot be derived aggregatively from the properties of their components. It is therefore tempting to say, borrowing the terminology of the classical British emergentists (Samuel Alexander, C. Lloyd Morgan, and C.D. Broad), that nonlinear processes generate “emergents,” whereas linear processes produce only “resultants.” Nonlinearity results from positive and negative feedback relationships. These relationships bring about patterns of behavior, which can be described as constrained alternatives in the space of all possible global states of the system (as attractors in phase space). To understand such activity patterns we need to model them in the language of collective variables and macroscopic-order parameters, while at the same time showing how these collective variables and order parameters are biophysically realized in natural phenomena (see Kelso 1995). The emergent processes of concern to me here occur in complex systems that need to be seen as autonomous. An autonomous system is a self-defining or self-determining system, by contrast with a system defined from the outside or a heteronomous system. A living cell, a multicellular animal, an ant colony, or a human being defines itself as a coherent unity in its interactions with its environment. An automatic bank machine, on the other hand, is defined and controlled from the outside, in the realm of human design. The paradigm for interaction with a heteronomous system is input/processing/output, in which deviations from desired outputs are seen as system errors. The paradigm for interaction with an autonomous system is a conversation, in which unsatisfactory outcomes are seen as breaches of understanding (Varela 1979, xii). According to the embodied or “enactive” approach in cognitive science, living beings and cognitive agents need to be understood as autonomous systems (Thompson 2007; Varela, Thompson, and Rosch 1991). According to Varela, an autonomous system can be precisely defined as a system that has organizational closure and operational closure (Varela 1979, 55–60). The term “closure” does not mean that the system is materially and energetically closed to the outside world (which of course is impossible). On the contrary,

60

autonomous systems are thermodynamically far-from-equilibrium systems, which incessantly exchange matter and energy with their surroundings. “Organizational closure” refers to the circular and recursive (self-referential) network of relations that defines the system as a unity. “Operational closure” refers to the reentrant and recurrent dynamics that organizational closure elicits.3 An autonomous system dynamically maintains its own organization as the crucial invariant through material and structural change. “Organizational-operational closure” is thus is a “morphodynamical” notion (Deffuant, Fuhs, Monnert, Bourgine, and Varela 1995; Petitot 1992): It describes a certain type of invariant form or pattern of collective self-organization. Because this form is precisely the dynamic pattern of a circular network whose constituent processes operate under closure (every product of the network stays within the network), the morphodynamics in the case of an autonomous system defines not merely a formal identity through time, but a formal self-identity. This type of morphodynamics marks an important difference between emergence in autonomous systems (such as a cell) and other oft-cited examples of emergence (such as convection rolls). In the case of autonomy (organizational-operational closure), what emerges is simultaneously a “self ” or “individual” and a correlative niche (the domain or set of interactions possible for such a system given its organization and concrete structural realization). The exemplar and minimal case of this kind of emergence is the living cell (Varela 1997). Another way to characterize autonomous systems is in relation to the contrast between “decomposable” and “nondecomposable” systems (Bechtel and Richardson 1993; Simon 1969). Decomposable systems have a clear hierarchical organization: Each component or subsystem operates according to its own intrinsic principles independent of the others, making the system strongly modular. In a “nearly decomposable” system, higher levels interact with lower levels through top-down or feedback relations. The nature of these interactions determines the extent to which the system is nearly decomposable: “A system will be nearly decomposable to the extent that the causal interactions within subsystems are more important in determining component properties than are the causal interactions between subsystems” (Bechtel and Richardson 1993, 27). As the components of the system

61

become less governed by intrinsic factors and more by the system’s organization, then the system is “minimally decomposable.” A nondecomposable system is one in which the connectivity and interrelatedness of the components give rise to global processes that subsume the components so they are no longer clearly separable. In such a system, the distinction between preexisting parts and supervening whole becomes problematic. Not only does the whole emerge from the components, but also the components emerge from the whole. An autonomous system is at least minimally decomposable, if not nondecomposable. More precisely, when one adopts an autonomy perspective, one ipso facto characterizes the system as at least minimally decomposable. The reason is that an autonomous system is an organizationally and operationally closed network, and hence it is the connectivity of its constituent processes that determines its operation as a network. Neural assemblies can be used to illustrate these ideas. With only a few exceptions, the brain is organized on the basis of a principle of reciprocity: If area A connects to area B, then there are reciprocal connections from B to A (Varela 1995; Varela, Lachaux, Rodriguez, and Martinerie 2001). The traditional practice in neuroscience is to map these reciprocally interconnected brain areas onto a hierarchy of processing levels from peripheral (lower) to central (higher) areas (measured in terms of synaptic distance from sensory stimulation). The sensory end is taken as the starting point, and perception is described as proceeding through a series of feedforward or bottom-up processing stages. Top-down influences are equated with back-projections and feedback from higher to lower areas. This scheme treats the brain as a nearly decomposable or minimally decomposable system (depending on the extent to which the top-down influences are emphasized). From a dynamicist point of view, however, the picture looks different. First, the dynamicist does not depict the brain as a processing hierarchy that starts at the sensory end. Strictly speaking, brain processes are always ongoing, and do not start or stop anywhere. A better entry point for many purposes of analysis can be found in the brain’s own endogenous activity, as reflected in the organism’s states of preparation, expectation, emotional tone, and attention, which are necessarily

62

active at the same time as the sensory inflow (Varela, Lachaux, Rodriguez, and Martinerie 2001, 230; see also Engel, Fries, and Singer 2001; Freeman 1999a, 1999b, 2000; Lutz, Lachaux, Martinerie and Varela 2002). Such activity arises far from the sensors—from the frontal lobes or limbic system or in the middle of the whole network from temporal and associative cortices. There is considerable evidence for the participation of this kind of endogenous activity even in early stages of sensory perception. Although this sort of activity is usually described as top-down or feedback, “top down” and “bottom up” are heuristic terms for what is in reality a large-scale network that integrates both incoming and endogenous activity. It is precisely at this network level that collective-variable dynamics and order parameters become important for characterizing the large-scale integration of widely distributed neuronal activities. Secondly (and to anticipate the discussion of Proposition 2), this large-scale dynamics can modulate local neuronal activity by entraining or “pulling” the behavior of individual neurons into a particular pattern of global activity. This dynamic systems form of global-to-local influence neither requires nor is equivalent to top-down control in a sequential hierarchy of processing stages (Engel, Fries, and Singer 2001; Thompson and Varela 2001). To look at the brain in this way is to characterize it as at least a minimally decomposable system—a system in which the behavior of the components is determined largely by the system’s organization. Yet it is also important to consider that the brain—in its dynamic operation as a large-scale network—might need to be characterized as a nondecomposable system. No doubt it is useful for certain explanatory purposes to characterize the brain as a (nearly or minimally) decomposable system. Yet problems can arise when one assumes that the explanatory strategies of decomposition and localization (differentiating a system into separable components and assigning responsibility for specific tasks to those components) are adequate to characterize brain activity (Bechtel and Richardson 1993; Uttal 2001). In particular, decomposition and localization are insufficient to characterize the operational closure of the brain as a dynamic neuronal network. In Maturana and Varela’s words:

63

Since, due to its constitution as a network of lateral, parallel, sequential, and recursive interactions, the nervous system closes on itself at all levels, the mutilations that it may suffer generally leave a closed neuronal network with a changed structure. Accordingly, the organization of the nervous system is essentially invariant under mutilations, while its domain of possible states, which depends on its structure, and, hence, on its architecture, is not. Yet, due to its closed organization, whatever is left of the neuronal network after a partial ablation necessarily operates as a different whole with different properties than the original, but not as a system from which some properties have been selectively subtracted . . . There is intrinsically no possibility of operational localization in the nervous system in the sense that no part of it can be deemed responsible for its operation as a closed network, or for the properties which an observer can detect in its operation as a unity. However, since every nervous system has a definite architecture, every localized lesion in it necessarily produces a specific disconnection between its parts and, hence, a specific change in its domain of possible states (1980, 129). This characterization implies that the brain—understood operationally as a dynamic network of processes or “brainweb” (Varela, Lachaux, Rodriguez and Martinerie 2001)—is a nondecomposable system. We still lack a theoretical language for expressing the complex behaviors of such systems in dynamic systems terms (see Le Van Quyen 2003). In the brain case, nondecomposability would mean that the brainweb generates global processes that subsume their components so they are no longer clearly separable as components. At this dynamic level, the distinction between pre-existing parts and supervening whole seems to have no clear application: One might as well say that the components (local neuronal activities) emerge from the whole as much as the whole (dynamic patterns of large-scale integration) emerges from the components. “Nondecomposability” and “decomposability” are heuristic, epistemological categories, not ontological ones. It is not my intention to argue for a metaphysical thesis of ontological holism on the basis of nondecomposability (which is not to

64

say that nondecomposability is merely epistemological in the sense an ontological reductionist would assert). My point is rather to call attention to the nondecomposability perspective in order to correct a strong bias in much of classical and contemporary neuroscience toward a modular (localizationist) view of the brain (see also Uttal 2001). Ultimately it is the interplay between these heuristic categories within and across various explanatory contexts that is important, not one heuristic versus the other. As Le Van Quyen states: Following Simon (1973), it is important to consider this interplay as a “loose vertical coupling,” permitting the distinction between levels, and a “loose horizontal coupling,” allowing the separation between subsystems at each level. While the word “loose” suggests “decomposable,” the word “coupling” implies resistance to decomposition. In my view, the characterization of this “loose coupling” represents one of the essential challenges for future developments (Le Van Quyen 2003, 84). Proposition 2: Global-to-Local Influence Complex systems theorists often appeal to the idea of “circular causality,” by which they mean that global patterns both arise from local interactions, and govern or constrain those interactions. In synergetics, a branch of complex systems theory, a vivid but unappealing metaphor is used to describe this global-to-local influence: The global, collective-variable dynamics is said to influence local behavior by “enslaving” the network elements into a particular dynamic régime (Haken 1983). The term “downward causation” is also often used to describe this sort of global-to-local influence. One of the earliest uses of “downward causation” in a scientific context was in an article by Donald Campbell (1974), called “ ‘Downward Causation’ in Hierarchically Organized Biological Systems.” Campbell’s view was that in hierarchically organized biological systems, downward causation from higher to lower levels of organization occurs in the form of natural selection: “Where natural selection operates through life and death at a higher level of organization,

65

the laws of the higher-level selective system determine in part the distribution of lower-level events and substances . . . all processes at the lower levels of a hierarchy are restrained by and act in conformity to the laws of the higher levels” (1974, 180). This idea that higher-level processes “restrain” lower-level processes so that they “act in conformity” to them corresponds to the idea that global processes (collective-variable dynamics) constrain or govern local interactions. What exactly does “constraint” mean in this context? In complex systems theory, constraints can be understood as relational properties had by the parts in virtue of their being integrated or unified (not aggregated) into a systemic network. “Constraint” is therefore a formal or topological notion (Deacon 2003). The form, configuration, or topology of a system limits or prevents certain possible behaviors that the parts could have on their own, while simultaneously opening up new possibilities for them by virtue of the states the system as a whole can access (Juarrero 1999, 132–33). Let us examine this notion of constraints in more detail, beginning with the difference between “context-free constraints” and “context-sensitive constraints” (Gatlin 1972, as cited by Juarrero 1999, 6–7, 131–40). A context-free constraint is one that is externally imposed and alters the probabilities of the available behavioral alternatives of the system’s components. For instance, a container filled with evenly diffused molecules of gas at room temperature is at thermodynamic equilibrium, but inserting a piston (a context-free constraint) into the container and moving the piston so that the molecules are compressed to one side imposes an orderly arrangement on them. If the pressure on the piston is removed, the system will move back toward equilibrium in accordance with the second law of thermodynamics. A context-sensitive constraint, on the other hand, is one that synchronizes and correlates previously independent parts into a systemic whole. Catalysis is a good example (1999, 139–41). Imagine a primeval soup with several types of molecules (A, B, and C) randomly floating around in it. As a result of externally imposed, context-free constraints (e.g., the weather) there will be more of some molecules in certain areas than in others. Now imagine that A catalyzes the formation of B. This relationship between A and B imposes a context-sensitive constraint on both of them:

66

Once the probability that something will happen depends on and is altered by the presence of something else, the two have become systematically and therefore internally related. As a result of the operations of context-sensitive constraints and the conditional probabilities they impose, A is now part of B’s external structure. Because A is no longer “out there” independent of B, to which it is only externally related, the interdependence has created a larger whole, the AB system. Insofar as it is part of B’s new context or external structure, A has been imported into B (Juarrero 1999, 139). Juarrero calls this kind of context-sensitive constraint a “first-order contextual constraint,” because it operates at the same level of organization as the individual components (A and B). A “second-order contextual constraint” is established when the organization of the whole system emerges as a constraint on the system’s components. Thus in a simplified and idealized autocatalytic network, A catalyzes the formation of B, B catalyzes the formation of C, C of D, and D closes the loop and catalyzes the formation of A. The relationship between each pair of catalyzing and catalyzed molecules is a first-order contextual constraint. Once autocatalytic closure occurs, however, then these first-order relationships become subject to the second-order contextual constraint of the organization as a whole. A still more striking example is autopoiesis. A minimal autopoietic system corresponds not simply to an autocatalytic network, but to an autocatalytic network housed within and interdependently linked to a semipermeable membrane-boundary. Autocatalytic networks do not qualify as autopoietic systems, because they do not self-produce their own topological boundaries.4 Autocatalytic networks either have no boundaries, or their boundaries are set by an externally imposed, context-free constraint (such as the walls of an experimental container). In an autopoietic system, however, the membrane forms part of the second-order contextual constraint of the system’s organization. In this type of system, the entire self-producing organization of membrane-plus-internal-autocatalytic-network operates as a second-order contextual constraint from the whole to the parts. Furthermore, in multicellular organisms, the autopoiesis of the individual cells becomes subject to the higher-order contextual constraints of the multicellular organization (second-order autopoiesis).

67

In other words, the autopoiesis of the individual cells becomes subordinated to the maintenance of the higher-order autopoiesis or autonomy of the multicellular organism (Maturana and Varela 1980, 107–111; 1987, 73–89). We now have in hand a possible answer to the question of what “downward causation” can mean in the context of complex systems theory. According to the line of thought just sketched, downward causation corresponds to the secondorder contextual constraint of a system’s organization restricting or limiting the degrees of freedom of the system’s components. Downward causation corresponds to the influence the relatedness of the system’s components has on the behavior of those components. More precisely, it corresponds to the influence of the system’s topological organization on its constituent processes (Varela 1979; see also Deacon 2003). “Downward” is thus a metaphor for the formal or topological influence of a whole with respect to its parts. It is questionable whether this metaphor is a good one. Although there are clearly empirical differences in scale and logical differences in order between the topology of a system and its constituent processes and elements, the two levels do not move in parallel, with one acting upward and the other acting downward, because the whole system moves at once. John Searle makes this point in a related discussion: “The right way to think of this is not so much ‘top down’ but as system causation. The system, as a system, has causal effects on each element, even though the system is made up of the elements” (Searle 2000, 17). From this perspective, the term “downward causation” is symptomatic of a partial recognition of system causation together with an inability to shift completely to a system-causation perspective. Some philosophers might wonder whether the topological influence of a system on its elements should be considered a causal influence. This issue is inseparable from broader conceptual issues about causation. Philosophical debates about emergence and downward causation tend to be structured not only by a strong ontology of part and whole (or basic features versus emergent features), but also by a strong ontology of “causal powers”—the causal powers of basic features versus the causal powers (or lack thereof ) of emergent features. Rather than get caught

68

up in the philosophical debates about causal-power views of causation, I am content to explicate what dynamic systems theorists seem to have in mind when they talk about whole-to-part influence in complex systems. As we have seen, this influence corresponds to the organizational constraint of a system with respect to its components. Such influence is topological. It is therefore not an external force that acts on something, but an interconnectedness or relatedness among processes. This interrelatedness structures the context and background of local interactions, such that certain kinds of events can take place that otherwise would not occur. Some authors describe the constraint of a system’s organization as a standing or ongoing, “structuring” cause, by contrast with an episodic “triggering” cause (see Dretske 1995 for this distinction), and liken organizational constraints so understood to Aristotle’s formal cause (Emmeche, Køppe and Stjernfelt 2000; Juarrero 1999, 125–28). Proposition 3: Relational Holism According to the ontological thesis of part/whole reductionism (also known as “mereological supervenience”), all the properties of a whole are determined by the intrinsic (non-relational) properties of its most fundamental parts (hence the whole is said to supervene on the intrinsic properties of its parts). According to holism (“mereological emergence”), on the other hand, certain wholes possess emergent features that are not determined by the intrinsic properties of their most basic parts. Such emergent features are irreducibly relational. They are constituted by relations that are not exhaustively determined by or reducible to the intrinsic properties of the elements so related. These holistic relations do not simply influence the parts, but supersede or subsume their independent existence in an irreducibly relational structure.5 One might think that relational holism could be invoked to legitimize the notion of downward causation. Downward causation is supposed to be the determinative influence that the relatedness of the system’s components has on their behavior. If this relatedness were holistic, then this fact would presumably account for the determinative influence of the system as a whole on its parts. The problem with

69

this line of thought is that relational holism implies that the relation is the most basic unit, and therefore that the terms of the relation have no independent (nonrelational) status. Hence the components could not constitute an independent lower level subject to higher-level “downward” influence, as the term “downward causation” cannot help but suggest. In other words, given relational holism, “downward causation” seems a misnomer. The presence of relational holism would thus provide another reason for doubting that the concept of downward causation is appropriate for describing the influence of a whole on its parts. The concept of relational holism was introduced largely in connection with nonseparability or entanglement in quantum mechanics, in which the state of the system is not constituted by the states of its parts, and only the whole system can be said to be in a definite state (Teller 1986; see also Belousek 2003). On the basis of nonseparability, a number of philosophers have argued that quantum mechanical systems are holistic or mereologically emergent.6 Silberstein and McGeever (1999) have proposed that nonseparability is a paradigm of “ontological emergence.” By this they mean (i) that it is a kind of emergence that belongs to a system or whole in itself, as opposed to being an artifact of our theories or models; and (ii) that it violates the metaphysical doctrine of (atomistic) mereological supervenience, which states that every property of the whole is exhaustively determined by the intrinsic properties of its most fundamental parts.7 Silberstein and McGeever’s question with regard to complex dynamic systems is whether they too exhibit ontological emergence, or only epistemological emergence.8 The main reason usually given for thinking that nonlinear dynamic systems are not holistic or ontologically emergent is that they are classical deterministic systems. In a deterministic nonlinear dynamic system, at all times (i) every variable has a definite value, and (ii) any change in the value of any variable is nonstochastic. As Silberstein states: “It is hard to imagine how any system that is ‘deterministic’ in both these ways could exhibit mereological emergence” (2001, 83). Nevertheless, because it is impossible for us to deduce mathematically the behavior or global dynamic properties (attractors) of the system, these behaviors and properties can

70

be described as epistemologically emergent. For instance, chaotic systems are deterministic, but can appear random and are unpredictable in the long run, because they are highly sensitive to initial conditions (small differences in perturbations produce exponentially divergent effects), and we are able to specify the values of their variables only to a finite degree of precision. Stephen Kellert expresses this notion of epistemological emergence when he writes: “chaos theory argues against the universal applicability of the method of micro-reductionism, but not against the philosophical doctrine of reductionism. That doctrine states that all properties of a system are reducible to properties of its parts. Chaos theory gives no examples of ‘holistic’ properties which could serve as counter-examples to such a claim” (1986, 90). More recent discussions may cast doubt on this statement. Frederick Kronz (1998) proposes that the key to chaos in classical systems is the nonseparability of the Hamiltonian (see also Kronz and Tiehen 2002, 332–33). Whereas the Newtonian formulation of classical mechanics takes forces as fundamental, the Hamiltonian formulation takes energies as fundamental. The Hamiltonian of a system corresponds to the total energy of the system (kinetic energy plus potential energy). In classical mechanics, the Hamiltonian describing a linear system is separable into a sum of Hamiltonians, with one element in the sum for each constituent of the system. On the other hand, if there are nonlinear terms in the equations of motion, then the Hamiltonian is nonseparable. Building on Kronz’s discussion, Robert Bishop (2002) proposes that the nonseparability of the Hamiltonian is crucial for understanding the emergent global coherence of a complex system: [T]he properties of integrity, integration and stability exhibited by Bénard cells are global properties and involve the nonlocal relation of all fluid elements to each other. This global behavior differs from holistic entanglement in quantum mechanics in the sense that fluid elements may be distinguished from each other while they are simultaneously identified as members of particular Bénard cells and participate in interaction with fluid elements

71

throughout the system. In this context focusing on the nonseparability of the Hamiltonian may be more appropriate because, in contrast to the quantum case, classical states are always separable even when the Hamiltonian is nonseparable (2002, 7). According to this line of thought, there is a kind of holism proper to complex systems that does not seem compatible with the philosophical doctrine of reductionism. On the other hand, this form of holism (nonseparability of the Hamiltonian) is not incompatible with determinism. What determinism means, however, is a whole other matter (see Atmanspacher and Bishop 2002). I do not intend to enter this thicket here, but two basic points are worth making. First, it is important to distinguish between determinism as a feature of a scientific model and determinism as a metaphysical thesis about nature. According to the metaphysical thesis, all physical properties in nature are definite and determinate, and the evolution of the natural world is fixed uniquely (the complete and instantaneous state of the world fixes its past and future with no alternatives). This thesis hardly follows from the fact that we can construct nonstochastic dynamic system models of certain observable phenomena. Second, it is also important to distinguish between nonlinear dynamic systems as abstract mathematical models and as observable biophysical systems. Any concrete empirical system will involve some degree of randomness in the form of stochastic fluctuations (Kelso 1995). For any given empirical system, the analytical techniques most appropriate for characterizing the system’s behavior depend on the hypotheses one makes about its degree of nonlinearity and its degree of stochasticness (Le Van Quyen 2003; Schreiber 1999). “Metastable” dynamic systems display different mixes of nonlinearity and stochasticness at different spatial and temporal scales (Friston 2000; Kelso 1995; Le Van Quyen 2003). Science has barely begun to chart this vast sea of nonlinearity and stochasticness. Within this context, “deterministic” seems best understood as describing certain nonlinear analysis techniques (ones in which there are no noise terms), not as an ontological characteristic of nature (in a classical observer-transcendent sense).

72

Résumé: Dynamic Co-emergence The discussion so far points toward a conception of emergence as dynamic coemergence. Dynamic co-emergence means that part and whole co-emerge and mutually specify each other. Kronz and Tiehen, in their discussion of emergence and quantum mechanics (with reference also to nonlinear dynamic systems), advocate the same idea, which they call “dynamic emergence”: Emergent wholes have contemporaneous parts, but these parts cannot be characterized independently from their respective wholes. Emergent wholes are produced by an essential, ongoing, interaction of its [sic] parts. These are the central features of the new view sketched here; the nonseparable Hamiltonian constitutes an essential ongoing interaction . . . By adopting [this] view, we can say that it does not make sense to talk about reducing an emergent whole to its parts, since the parts are in some sense constructs of our characterization of the whole (2002, 345).

II KIM’S OBJECTIONS TO EMERGENCE A N D D OW N WA R D C AU S AT I O N

With the foregoing ideas in place, we can now profitably examine some recent and well-known objections to emergence and downward causation, raised by the philosopher Jaegwon Kim (1993, 1998, 1999). Kim argues that “reflexive downward causation”—the causal influence of a whole on its own microconstituents—is incoherent when understood to happen simultaneously or synchronically, and is either otiose or violates the “causal closure of the physical” when understood to happen diachronically (Kim 1999, 28–33). If Kim is right, then emergence and downward causation are merely epistemological in the ontological reductionist’s sense. For this reason, it will be instructive to consider his arguments.

73

Here is Kim’s description of the first case of synchronic reflexive downward causation: At a certain time t, a whole W, has emergent property M, where M emerges from the following configuration of conditions: W has a complete decomposition into parts a1, . . . , an; each ai has property Pi; and relation R holds for the sequence a1, . . . , an. For some aj, W’s having M at t causes aj to have Pj at t (1999, 28). What Kim finds troubling and ultimately incoherent about this case is the synchronic way in which it combines upward determination and downward causation: The upward determination of W’s having M at t as an effect of aj’s having Pj at t is simultaneous with the downward causation of aj’s having Pj at t as an effect of W’s having M at t. Nevertheless, there is no obvious incoherence in this case yet. Simultaneous causation, although counterintuitive to common sense, is not obviously incoherent. For instance, there are theories in physics that postulate instantaneous action at a distance; whether such influences exist, and whether they should qualify as causal, are difficult empirical and conceptual issues. Furthermore, simply to pronounce causal circularity in general “unacceptable,” as Kim does (1999, 28), is too facile, given that the notion of circular causality plays a role in complex systems theory. Whether the circular causality of complex systems theory is best seen as synchronic and/or diachronic, or as simply shorthand for reciprocal causal explanation, is an open and difficult question, which cannot be decided simply by armchair pronouncements. Hence, it comes as no surprise that Kim has to derive the incoherence he discerns by introducing an additional metaphysical principle about causation—the “causalpower actuality principle”—to which he thinks “we tacitly subscribe” (though he does not tell us whether “we” means metaphysicians, scientists, or the ordinary person in the street):

74

For an object x, to exercise, at time t, the causal/determinative powers it has in virtue of having property P, x must already possess P at t. When x is caused to acquire P at t, it does not already possess P at t and is not capable of exercising the causal/determinative powers inherent in P (1999, 29). With this principle applied to the case, synchronic reflexive downward causation is incoherent: W’s having M at t is supposed to be an effect of aj’s having Pj at t, which implies (by the causal-power actuality principle) that aj must already possess Pj at t, but aj’s having Pj at t is supposed to be an effect of W’s having M at t, which implies that aj does not already have Pj at t. The notion of downward causation as global-to-local structuring influence casts doubt on the generality of Kim’s principle. The problem with the principle can be seen in the case of Rayleigh-Bénard convection rolls. The orderly rolling motions (which in a closed container form parallel cylinders called “Bénard cells”) emerge from the local dynamics of the fluid molecules and at the same time constrain the states of motion available to those molecules. As Robert Bishop states: When the cells are established at t, this governing property is established at t; it did not exist prior to t. Likewise, prior to t, the trajectories of fluid elements had the property of accessing various states of motion, a property they lose at t due to the global governing effects of Bénard cells. The causal constraints/modifications on the motion of the fluid elements in this case are synchronic: The emergence of the self-regulating global pattern is simultaneous with the modifications of the accessible states of the system. If it were not synchronic, the pattern would not arise (2002, 9). Contrary to Kim, therefore, synchronic reflexive downward causation seems perfectly coherent, as long as it is not conceptualized as efficient causation (in which cause and effect are external to one another), but rather as global-to-local structuring influence.

75

Let us turn to Kim’s argument against diachronic downward causation. Here matters require still more disentangling. Kim distinguishes between two kinds of downward causation, one in his view ordinary and unproblematic, the other highly problematic, but neither kind fits the kind of downward causation found in complex systems theory. The first kind is simply causation from the properties of a whole to properties of its parts—where Kim leaves the notions of “whole” and “part” unanalyzed. A whole causes one of its microconstituents to change in a certain way, and there is a time delay between cause and effect. Kim interprets Roger Sperry’s (1969) analogy of a rolling wheel constraining the motion of its own molecules as a case of this kind of diachronic downward causation: The structural properties of the wheel constrain the movements of the molecules composing it, so that the molecules are caused to move in a particular way as the wheel rolls downhill. Here are Kim’s other examples and his verdict about this kind of diachronic downward causation: I fall from the ladder and break my arm. I walk to the kitchen for a drink of water and ten seconds later, all my limbs and organs have been displaced from my study to the kitchen. Sperry’s bird flies into the blue yonder, and all of the bird’s cells and molecules, too, have gone yonder. It doesn’t seem to me that these cases present us with any special mysteries rooted in selfreflexivity, or that they show emergent causation to be something special and unique (1999, 30). The conclusion Kim draws is that “emergent downward causation should not simply be identified with causation from properties of the whole to properties of its own parts, that is, reflexive downward causation” (1999, 31). What emergent downward causation requires is not simply whole-to-part influence, but a difference in kind between the macrolevel and microlevel events. In other words, the causal relations between the two levels must implicate different properties—unlike the rolling wheel or the flying bird, in which the property of motion is common to the whole and its constituent parts.

76

Kim seems right about these cases. But there is a shortcoming to his analysis: None of the examples he gives in the passage above contains the kind of globalto-local influence found in complex systems. Nowhere in these examples is there the sort of nonlinear coupling of system components typical of a self-organizing system. This absence is particularly evident in the case of Sperry’s wheel, which is an aggregative system, not a self-organizing one. Nowhere is there any dynamic global pattern formation (as in convection rolls and Bénard cells). Nor is there any self-reflexivity, because there is no complex system with operational closure and second-order contextual constraints (as in autopoiesis). Kim neglects the most important and relevant cases for emergence and downward causation. He draws the right general conclusion, but not for the right reason: It is true that emergent downward causation should not be identified simply with whole-to-part causation, but that is because emergent downward causation (at least in the complex systems context) is meant to describe not any kind of whole-to-part influence, but the specific kind of reflexive global-to-local influence that happens in a system that has dynamic global coherence in and through collective self-organization. Kim’s next step is to examine the sort of downward causation in which the macrolevel and microlevel events have different characteristic properties. His model is mental-to-physical causation, but the form of his argument has nothing to do with the mental and the physical as such, and could just as easily apply to the relationship between biological and physical events. The argument he advances rests on three principles (see Kim 1993, 336–57): 1. The Physical Realization Principle: Every emergent event or property M must be realized by (or determined by, or supervenient on) some physical event or property P (its “emergence base”). 2. The Causal Inheritance Principle: If M is instantiated on a given occasion by being realized by P, then the causal powers of this instance of M are identical with (or a subset of ) the causal powers of P. 3. The Principle of the Causal Closure of the Physical Domain: Any physical event that has a cause at time t has a physical cause at t. Hence, “if we

77

trace the causal ancestry of a physical event, we need never go outside the physical domain” (1993, 280). The critical question Kim then asks is why the physical emergence base of the emergent cannot displace the emergent as a cause of any of its putative effects. Here is the basic form of his argument. Suppose emergent M causes M* (which may or may not be an emergent). M* has its own physical emergence base P*. Therefore, M’s causation of M* presupposes M’s causation of P*. In other words, the only way for M to bring about M* is by bringing about its physical emergence base P* (a case of downward causation). Now, M has its own physical emergence base P. The presence of P is sufficient for the presence of M. (Note that P is supposed not to cause M, but synchronically to realize M.) But M is causally sufficient for P* and thereby M*. It follows that P is causally sufficient for both P* and M*. But then M’s status as a cause is preempted by P, thus making M “otiose and dispensable” as a cause of P* and M*. The upshot of this argument is a dilemma for the emergentist who believes in downward causation: Either downward causation is otiose, because the putative causal power of the emergent is preempted by the causal power of the physical elements on which the emergent is based, or downward causation violates the principle that the physical domain is causally closed. This argument, when applied to the case of emergent downward causation in self-organizing systems, is tantamount to denying that the system’s organization exerts any influence on the system’s components. The thought seems to be that all of what we call the macrolevel and emergent causation really happens at the microlevel (by the Causal Inheritance Principle). Talk of the system’s organization having an emergent causal influence is simply a macrolevel description of the microlevel causal events. Such a description can be epistemologically useful for us, even though ontologically speaking all the real causal action goes on at the microlevel of the emergence base (Kim 1999, 33). On this view, the system’s organization is in fact epiphenomenal—an effect of lower level causal transactions with no significant causal status of its own—although it can still be useful for us to talk about the system as if its organization did play a causal role in bringing about certain events (see Kim 1993, 92–108).

78

The problem with this way of thinking is its refusal to countenance the causal importance of a system’s organization (the relations that define it as a system). It does not acknowledge that the microlevel interactions happen as they do because of the way the local processes are organized, which is a macrolevel, relational characteristic of the system. That the macrolevel organization of a complex system is not simply epiphenomenal is evident in at least two respects. First, it is multiply realizable: The very same organization can obtain even though the constituents that realize it can vary. Indeed, in the case of any dissipative (energetically open) self-organizing system, not only can the particular constituents vary, but they also must vary, because they have to be incessantly renewed as the system dissipates energy into its environment. This kind of material renewal is especially striking in autopoiesis: There is and must be a constant turnover of material constituents in the cell while the autopoietic organization remains invariant (otherwise the cell dies). This aspect of a complex system is part of its robustness or dynamic stability, which means that the organization of the system is resilient to small changes or perturbations at the microlevel, and adaptive to large changes. Secondly, the dynamic stability of a complex system is reflected in the fact that different counterfactual statements are true for the macrolevel organization and for the microlevel processes. The organization is necessary for certain subsequent events, but the particular constituents are not: If the organization had been different and those constituents the same, the events would not have occurred; but if the constituents had been different (within certain limits) and the organization the same, the events would still have occurred (because of the way the constituents were organized).9 Given that different counterfactual statements are true of the microlevel constituents and the macrolevel organization, the two cannot be identified. Kim recognizes some of these points, but he goes through contortions to deal with them. In a footnote at the very end of his article, “Making Sense of Emergence,” he admits that “complex systems obviously bring new causal powers into the world, powers that cannot be identified with causal powers of the more basic simpler systems. Among them are the causal powers of microstructural, or micro-based, properties of a complex system” (1999, 36). By a “microstructural” or “micro-based” property he means a macrolevel property that is completely decomposable into a collection of

79

microconstituents plus the microlevel properties and relations that characterize them (Kim 1998, 84). (Given this characterization, it would seem that Kim simply assumes that the microlevel does not include holistic relations, such as quantum entanglement or nonseparability of the Hamiltonian in nonlinear dynamic systems, but this is left unclear.) In this footnote (and the corresponding lengthier treatment in his book Mind in a Physical World), Kim gives away many of the issues at stake in the debate about emergence without ever acknowledging that he has done so (Campbell and Bickhard 2002). Thus he admits that “macroproperties can, and in general do, have their own causal powers, powers that go beyond the causal powers of their microconstituents” (1998, 85, Kim’s emphasis). Nevertheless, he declares, “these properties are not themselves emergent properties; rather, they form the basal conditions from which further properties emerge (for example . . . consciousness is not itself a microstructural property of an organism, though it may emerge from one)” (1999, 36). The reasons Kim denies emergence to these macroproperties are complicated and problematic. First, his conception of emergent properties is derived from the British emergentists, who held that it is simply an inexplicable brute fact of nature—to be accepted in an “attitude of natural piety,” in Samuel Alexander’s phrase—that a certain organization of lower-level elements gives rise to some emergent quality. According to this way of thinking, an emergent property is not to be identified with the macrolevel organization itself, but with a qualitatively distinct property that supervenes on and has as its emergence base that organization (see Kim 1998, 85–86). There is no compelling reason today to endorse this peculiar conception of emergence. The British emergentists rightly drew attention to the significance of organization and relational structure, but did not have the scientific tools for studying them we now do. It is not classical British emergentism but contemporary science that should guide our thinking about emergence.10 Secondly, Kim apparently accepts part/whole reductionism, hence for him all macrolevel, micro-based properties are decomposable into the intrinsic properties of microlevel entities (although, as I mentioned above, this is left unclear). But part/whole reductionism as an ontological thesis about nature is unacceptable, because (among other things) it presupposes a conceptual framework of pre-existing

80

particulars with intrinsic properties that apparently has no coherent formulation in the language of microphysical theory (which is, after all, supposed to be the ground floor on which everything else rests, according to the reductionist). Macrolevel characteristics can be micro-dependent and micro-involving, without being micro-based in Kim’s sense (analytically decomposable into pre-existing microlevel entities and their intrinsic properties). Thirdly, Kim’s approach to emergence is entirely dominated by the Cartesian mind-body problem and its procrustean framework of the mental versus the physical. Given part/whole reductionism about the physical domain (a basically Cartesian conception of nature), and a very wide and problematic conception of “physical” that includes all of biology and psychology save phenomenal consciousness (which is excluded not as something immaterial, but as something that resists physicalistic reductive analysis), the only candidates left over for emergence are “qualia,” the qualitative or phenomenal properties of conscious experience. By contrast, the approach to emergence I have taken in this paper is oriented not by the Cartesian mind-body problem, but by the Kantian problem of self-organization and its relation to the threefold order of matter, life, and mind.11 Let me close this discussion by making a few comments about the three principles at the base of Kim’s argument. His argument is as strong as these principles, and although they have force against classical British emergentism, each one is dubious when applied to emergence in the context of contemporary science. According to the first principle, every emergent is determined by its physical realization base or emergence base; according to the second, every emergent inherits its causal powers from its emergence base. These principles are simply an expression of physicalistic ontological reductionism. This position claims science for its support, but it is metaphysical in the sense of going beyond anything science itself tells us. I see no good reason to believe in such a thing as an “emergence base,” where this means a configuration of pre-existing microphysical entities with intrinsic properties and causal powers that grounds the macrophysical level. This image of nature as a mereologically ordered hierarchy grounded on a base level of particulars is a metaphysical picture projected onto science, whereas the image

81

science projects is of networks of processes at various spatiotemporal scales, with no base-level particulars that “upwardly” determine everything else (Bickhard and Campbell 2002; Hattiangadi 2005). What about Kim’s third principle, the causal closure of the physical? The precise meaning and status of this principle are not clear. The main problem is that it is far from clear what “physical” means—what it includes and excludes—and it is hard to see how one could go about answering this question short of having a complete and true physics (whatever that means and assuming it even makes sense to suppose such a thing). Suppose at some point in the future physicists felt compelled to include mental properties (qua mental) as fundamental properties of physical theory (Montero 1999, 2001). Given that we cannot accurately predict the future course of physics, we have to at least allow for this possibility. In fact, some physicists and philosophers already believe such inclusion to be necessary to account for both mental and physical phenomena (Shimony 1997). But in that case, the closure of physics would include the mental qua mental (as opposed to the mental qua reduced to the physical). This possibility illustrates that the Cartesian mental/physical distinction has become useless. As for Kim, he embraces a very wide sense of “physical” that includes every classical example of an emergent property from physics, chemistry, biology, and psychology, save qualia. Nowhere in my discussion of emergence and downward causation have I appealed to anything outside the physical domain so broadly conceived. On the other hand, if the principle of the causal closure of the physical is construed more narrowly to mean the causal closure of the microphysical, then the principle is certainly not obviously true, and may even be false or incoherent (Dupré 1993; Hattiangadi 2005). For instance, according to Bohr’s interpretation of quantum mechanics, macrophysical concepts are as indispensable to characterize microphysical phenomena as microphysical phenomena are to explain certain macrophenomena (Hattiangadi 2005). Moreover, the causal closure of physics has traditionally been linked to the ideal of the unity of science via intertheoretic reduction—that biology is in principle reducible to chemistry, and chemistry to physics. But such unity via intertheoretic reduction seems nowhere in sight (Dupré 1993;

82

Garfinkel 1981). As Nancy Cartwright says about the classic case of the supposed intertheoretic reduction of physical chemistry to quantum mechanics: “Notoriously we have nothing like a real reduction of the relevant bits of physical chemistry to physics—whether quantum or classical. Quantum mechanics is important for explaining aspects of chemical phenomena but always quantum concepts are used alongside of sui generis—that is, unreduced—concepts from other fields. They don’t explain the phenomena on their own” (Cartwright 1997, 163). One final remark about Kim’s conception of the “physical” is in order here. Part/whole reductionism goes hand in hand with an atomistic metaphysics of basic physical particulars and their mereological configurations, a metaphysics to which Kim apparently subscribes (1993, 77, 96–97, 337). At the same time, he also apparently believes that nothing in the philosophical dispute about emergence depends on precise general definitions of “physical” (1993, 340). But this seems wrong on both counts. In the context of contemporary science, as we have seen, “nature” does not consist of basic particulars, but fields and processes, and this difference between a process viewpoint and an elementary-particle-version of Cartesian substance metaphysics does make a difference to the philosophical issues about emergence (Campbell and Bickhard 2002; Hattiangadi 2005). On the former view, there is no bottom level of basic particulars with intrinsic properties that upwardly determines everything else. Everything is process all the way “down” and all the way “up,” and processes are irreducibly relational—they exist only in patterns, networks, organizations, configurations, or webs. For the part/whole reductionist, “down” and “up” describe more and less fundamental levels of reality: Higher levels are realized by and determined by lower levels (the “layered model of reality;” see Kim 1993, 337–39). On the process view, “up” and “down” are context-relative terms used to describe phenomena of various scale and complexity. There is no base level of elementary entities to serve as the ultimate “emergence base” on which to ground everything. Phenomena at all scales are not entities or substances, but relatively stable processes, and since processes achieve stability at different levels of complexity, while still interacting with processes at other levels, all are equally real and none has absolute ontological primacy.

83

NOTES

1. The material in this essay is a modified and revised version of material taken from Thompson (2007). 2. See Andersen, Emmeche, Finneman, and Christiansen (2000); Schröder (1998); Silberstein (2001, 2002); Silberstein and McGeever (1999); Van Gulick (2001). 3. “Closure” is used here in its algebraic sense: An operation K exhibits closure in a domain D if every result of its operation yields results within D. Thus the operation of a system has operational closure if the results of its activity remain within the system itself. This notion of operational closure has nothing to do with the idea of a materially closed system—a system that is closed to interactions with what lies beyond it. Autonomous systems are and must be coupled to their surroundings; the issue is the nature of the dynamics that defines them and on the basis of which they interact with what lies beyond them (Varela 1979, 1997; Varela and Bourgine 1991). Varela in his writings does not distinguish between organizational and operational closure as I do here, but uses the two terms more or less interchangeably. My usage follows Rudrauf, Lutz, Cosmelli, Lachaux, and Le Van Quyen (2003, 34). 4. Juarrero treats autopoietic and autocatalytic systems as if they were equivalent, but they are not. 5. Another holistic view of emergence goes further by seeing emergence as a kind of “fusion” (Humphreys 1996, 1997a, 1997b). The difference between fusion and relational holism is that a fusion is not a relational structure, because the previously unrelated elements are “used up” in the fusion. 6. See Belousek (2003), Hawthorne and Silberstein (1995), Healy (1991), Humphreys (1996, 1997a, 1997b), Kronz and Tiehen (2002), and Maudlin (1998). 7. Ontological emergence in the form of relational holism violates atomistic mereological supervenience, but perhaps not every form of mereological supervenience. Atomism states that every property of a whole is logically supervenient on, or logically determined by, the intrinsic properties of the whole’s most fundamental parts. On this view, every relation is a reducible relation. On the other hand, if mereological supervenience were taken more liberally to mean that every property of a whole is logically supervenient on, or logically determined by, the properties of the parts and their relations—where this allows for nonatomistic or irreducible relations—then mereological supervenience would be consistent with relational holism. This differentiation of versions of mereological supervenience is not too

84

important at this stage of the discussion. It will be relevant later, however, when we come to emergence as understood by the classical British emergentists (Samuel Alexander, C. D. Broad, and C. Lloyd Morgan) and its influence on contemporary philosophy of mind (see Kim 1999). The British emergentists apparently believed, according to Jaegwon Kim’s (1999) interpretation, that emergent properties are not identical with macrolevel, organizational properties (whether holistic or not), but are additional distinct properties that supervene on those organizational properties as their emergence bases. The emergence of such distinct, irreducible properties (such as “life” or “consciousness”) given a sufficient macrolevel organization, and their supervenience on that macrolevel, were taken as brute facts of nature, to be accepted in an “attitude of natural piety,” again in Samuel Alexander’s phrase. 8. If one criterion of ontological emergence is that it not be an artifact of our theories or models, and if quantum mechanics is supposed to be the best case of ontological emergence, then it would seem a realist interpretation of quantum mechanics is being assumed. Whether there is a coherent realist interpretation of quantum mechanics is an unresolved matter. Bitbol (2007) presents a critical discussion of the realist premises of philosophical treatments of emergence in quantum mechanics. His de-ontologized conception of emergence is close to my conception of dynamic co-emergence. 9. Therefore it would seem that Kim is mistaken when he writes: “A whole has a certain emergent property, M, at a given time, t, and the fact that this property emerges at t is dependent on its having a certain microconfiguration at t, and this includes a given constituent of it, aj, having Pj at t. That is, unless aj had Pj at t, W could not have had its emergent property M at t ” (1999, 29). But if W is a complex system with structural stability, then it could well have had its emergent property M at t, without its constituent aj having had Pj at t. 10. Cf. Kronz and Tiehen (2002, 332–33): “the British emergentists may have placed too much emphasis on the Newtonian formulation of classical physics in motivating their metaphysical doctrine by regarding forces as fundamental, and then developing their metaphysics by analogy with the way in which forces behave. . . . If they had modeled their metaphysical doctrines on the Hamiltonian formulation of classical mechanics, in which energies (both kinetic and potential) are fundamental, the associated metaphysical doctrine may have had a very different character. Poincaré was aware of chaotic classical models, and that they involved nonseparable Hamiltonians; but it is unclear whether the British emergentists were familiar with them.”

85

11. In a mere three sentences in his “Making Sense of Emergence” Kim confronts the Kantian problematic of self-organization (from Kant’s Critique of Judgment) and casually dismisses it without so much as a mention: “But how is it possible for the whole to causally affect its constituent parts on which its very existence and nature depend? If causation or determination is transitive, doesn’t this ultimately imply a kind of self-causation, or selfdetermination—an apparent absurdity? It seems to me that there is reason to worry about the coherence of the whole idea” (1999, 28).

REFERENCES

Andersen, P. B., Emmeche, C., Finnemann, N. O., and Christiansen, P. V., eds. 2000. Downward Causation: Minds, Bodies and Matter. Aarhus, Denmark: Aarhus University Press. Atmanspacher, H. and Bishop, R., eds. 2002. Between Chance and Choice: Interdisciplinary Perspectives on Determinism. Thorverton, UK: Imprint Academic. Bechtel, W. and Richardson, R. C. 1993. Discovering Complexity: Decomposition and Localization as Strategies in Scientific Research. Princeton NJ; Princeton University Press. Belousek, D. W. 2003. “Non-Separability, Non-Supervenience, and Quantum Ontology.” Philosophy of Science 70: 791–811. Bishop, R. C. 2002. “Downward Causation in Complex Systems.” British Society for the Philosophy of Science Annual Meeting, Glasgow, Scotland, 4–5 July 2002. Bitbol, M. (2007). “Ontology, Matter and Emergence.” Phenomenology and the Cognitive Sciences 6 (3): 293–307. Bitbol, M. and Luisi, P. L. 2005. “Autopoiesis with or without Cognition: Defining Life at its Edge.” Journal of the Royal Society Interface 1: 99–107. Bourgine, P. and Stewart, J. 2004. “Autopoiesis and Cognition.” Artificial Life 20: 327– 45. Campbell, D. 1974. “ ‘Downward Causation’ in Hierarchically Organized Biological Systems.” Pp. 179–86 in Studies in the Philosophy of Biology, ed. F. J. Ayala and T. Dobzhansky. Berkeley & Los Angeles, CA: University of California Press. Campbell, R. J. and Bickhard, M. H. 2002. “Physicalism, Emergence, and Downward Causation.” Unpublished. http://www.lehigh.edu/~mhb0/physicalemergence.pdf.

86

Cartwright, N. 1997. “Why Physics?” Pp. 161–68 in The Large, the Small, and the Human Mind, R. Penrose, A. Shimony, N. Cartwright, and S. Hawking. Cambridge: Cambridge University Press. Deacon, T. W. 2003. “The Hierarchic Logic of Emergence: Untangling the Interdependence of Evolution and Self-Organization.” Pp. 273–308 in Evolution and Learning: The Baldwin Effect Reconsidered, ed. B. Weber and D. Depew. Cambridge MA: MIT Press. Deffuant, G., Fuhs, T., Monnert, E., Bourgine, P. and Varela, F. J. 1995. “Semi-Algebraic Networks: An Attempt to Design Geometric Autopoietic Networks.” Artificial Life 2: 157–77. Dretske, F. 1995. “Mental Events as Structuring Causes of Behavior.” Pp. 121–36 in Mental Causation, ed. J. Heil and A. Mele. Oxford: Clarendon Press. Dupré, J. 1993. The Disorder of Things. Cambridge MA: Harvard University Press. Emmeche, C., Køppe, S. and Stjernfelt, F. 2000. “Levels, Emergence, and Three Versions of Downward Causation.” Pp. 13–34 in Downward Causation: Minds, Bodies and Matter, ed. P. B. Andersen, C. Emmeche, N. O. Finnemann, and P. V. Christiansen. Aarhus, Denmark: Aarhus University Press. Engel, A. K., Fries, P. and Singer, W. 2001. “Dynamic Predictions: Oscillations and Synchrony in Top-Down Processing.” Nature Reviews Neuroscience 2: 704–16. Freeman, W. J. 1999a. How Brains Make up Their Minds. London: Weidenfeld and Nicolson. ———. 1999b. “Consciousness, Intentionality, and Causality.” Journal of Consciousness Studies 6: 143–72. ———. 2000. “Emotion Is Essential to All Intentional Behaviors.” Pp. 209–35 in Emotion, Development, and Self-Organization: Dynamic Systems Approaches to Emotional Development, ed. M. D. Lewis and I. Granic. Cambridge and New York: Cambridge University Press. Friston, K. J. 2000. “The Labile Brain. II. Transients, Complexity, and Selection.” Philosophical Transactions of the Royal Society of London. B. 355: 237–52. Garfinkel, A. 1981. Forms of Explanation. Rethinking the Questions in Social Theory. New Haven CT: Yale University Press. Gatlin, L. 1972. Information and the Living System. New York: Columbia University Press.

87

Haken, H. 1983. Synergtics: An Introduction. Berlin: Springer-Verlag. Hattiangadi, J. 2005. “The Emergence of Minds in Space and Time.” Pp. 79–100 in The Mind as a Scientific Object: Between Brain and Culture, ed. D. M. Johnson and C. Ernelling. New York: Oxford University Press. Hawthorne, J. and Silberstein, M. 1995. “For Whom the Bell Arguments Toll.” Synthese 102: 99–138. Healey, R. A. 1991. “Holism and Nonseparability.” Journal of Philosophy 88: 393–421. Humphreys, P. 1996. “Aspects of Emergence.” Philosophical Topics 24: 53–70. ———. 1997a. “Emergence, Not Supervenience.” Philosophy of Science 64 (Proceedings): S337–45. ———. 1997b. “How Properties Emerge.” Philosophy of Science 64: 1–17. Juarrero, A. 1999. Dynamics in Action: Intentional Behavior as a Complex System. Cambridge MA: MIT Press. Kellert, S. H. 1993. In the Wake of Chaos. Chicago: University of Chicago Press. Kelso, J. A. S. 1995. Dynamic Patterns: The Self-Organization of Brain and Behavior. Cambridge MA: MIT Press. Kim, J. 1993. Supervenience and Mind: Selected Philosophical Essays. Cambridge: Cambridge University Press. ———. 1998. Mind in a Physical World: An Essay on the Mind-Body Problem and Mental Causation. Cambridge MA: MIT Press. ———. 1999. “Making Sense of Emergence.” Philosophical Studies 95: 3–36. Kronz, F. M. 1998. “Nonseparability and Quantum Chaos.” Philosophy of Science 65: 50–75. Kronz, F. M. and Tiehen, J. T. 2002. “Emergence and Quantum Mechanics.” Philosophy of Science 69: 324–47. Le Van Quyen, M. 2003. “Disentangling the Dynamic Core: A Research Program for Neurodynamics at the Large-Scale.” Biological Research 36: 67–88. Lutz, A., Lachaux, J.-P., Martinerie, J. and Varela, F. J. 2002. “Guiding the Study of Brain Dynamics by Using First-Person Data: Synchrony Patterns Correlate with Ongoing Conscious States during a Simple Visual Task.” Proceedings of the National Academy of Sciences USA 99: 1586–91. Maturana, H. R. and Varela, F. J. 1980. Autopoiesis and Cognition: The Realization of the Living. Dordrecht: D. Reidel.

88

Maudlin, T. 1998. “Part and Whole in Quantum Mechanics.” Pp. 46–60 in Interpreting Bodies: Classical and Quantum Objects in Modern Physics, ed. E. Castellani. Princeton NJ: Princeton University Press. Montero, B. 1999. “The Body Problem.” Nous 33: 183–200. ———. 2001. “Post-Physicalism.” Journal of Consciousness Studies 8: 61–80. Petitot, J. 1992. Physique du sens. Paris: Editions du CNRS. Rudrauf, D., Lutz, A., Cosmelli, D., Lachaux, J.-P. and Le Van Quyen, M. 2003. “From Autopoiesis to Neurophenomenology: Francisco Varela’s Exploration of the Biophysics of Being.” Biological Research 36: 27–66. Schröder, J. 1998. “Emergence: Non-Deducibility or Downwards Causation?” Philosophical Quarterly 48: 433–52. Searle, J. R. 2000. “Consciousness, Free Action and the Brain.” Journal of Consciousness Studies 7: 3–22. Shimony, A. 1997. “On Mentality, Quantum Mechanics, and the Actualization of Potentialities.” Pp. 144–60 in The Large, the Small, and the Human Mind, R. Penrose, A. Shimony, N. Cartwright, and S. Hawking. Cambridge: Cambridge University Press. Silberstein, M. 2001. “Converging on Emergence: Consciousnesss, Causation, and Explanation.” Journal of Consciousness Studies 8: 61–98. ———. 2002. “Reduction, Emergence, and Explanation.” Pp. 80–107 in The Blackwell Guide to the Philosophy of Science, ed. P. Machamer and M. Silberstein. Oxford: Basil Blackwell. Silberstein, M. and McGeever, J. 1999. “The Search for Ontological Emergence.” Philosophical Quarterly 49: 182–200. Simon, H. 1969. The Sciences of the Artificial. Cambridge MA: MIT Press. ———. 1973. “The Organization of Complex Systems.” Pp. 1–27 in Hierarchy Theory: The Challenge of Complex Systems, ed. H. H. Pattee. New York: George Braziller. Sperry, R. 1969. “A Modified Concept of Consciousness.” Psychological Review 76: 532–36. Teller, P. 1986. “Relational Holism and Quantum Mechanics.” British Journal for the Philosophy of Science 37: 71–81. Thompson, E. 2007. Mind in Life: Biology, Phenomenology, and the Sciences of Mind. Cambridge MA: Harvard University Press.

89

Thompson, E. and Varela, F. J. 2001. “Radical Embodiment: Neural Dynamics and Consciousness.” Trends in Cognitive Sciences 5: 418–25. Uttal, W. R. 2001. The New Phrenology: The Limits of Localizing Cognitive Processes in the Brain. Cambridge MA: MIT Press. Van Gulick, R. 2001. “Reduction, Emergence and Other Recent Options on the Mind-Body Problem: A Philosophic Overview.” Journal of Consciousness Studies 8: 1–34. Varela, F. J. 1979. Principles of Biological Autonomy. New York: Elsevier North Holland. ———. 1995. “Resonant Cell Assemblies: A New Approach to Cognitive Functions and Neuronal Synchrony.” Biological Research 28: 81–95. ———. 1997. “Patterns of Life: Intertwining Identity and Cognition.” Brain and Cognition 34: 72–87. Varela, F. J. and Bourgine, P., eds. 1991. Toward a Practice of Autonomous Systems: Proceedings of the First European Conference on Artificial Life. Cambridge MA: MIT Press. Varela, F. J., Lachaux, J.-P., Rodriguez, E. and Martinerie, J. 2001. “The Brainweb: Phase Synchronization and Large-Scale Integration.” Nature Reviews Neuroscience 2: 229–39. Varela, F. J., Thompson, E. and Rosch, E. 1991. The Embodied Mind: Cognitive Science and Human Experience. Cambridge MA: MIT Press. Wimsatt, W. C. 1986. “Forms of Aggregativity.” Pp. 259–91 in Human Nature and Human Knowledge, ed. A. Donagan, A. Perovich, and M. Wedin. Dordrecht: D. Reidel.

DISCUSSION

I have a couple of questions. One of the assumptions we make when we deal with dynamic systems is that there is a shared mind between the individuals, namely the messages that I am sending to you are going to be interpreted in similar ways as the messages that you are sending me. So the first question is: from where did this shared mind emerge? The second question, which is even harder for me to understand, is related to holism; that is, how do the different levels in the system recognize the emergence of a higher level? There has to be some sort of machinery by which the system itself can recognize the newly emerged patterns which were not there before. So what are the mechanisms by AV I V B E RG M A N :

90

which this operates? These are the two issues that I did not see you address in this paper. A third point is the dynamics of co-emergence. What I would like to propose is something like autonomy, but in terms of maintenance of the self rather than self-maintenance. The distinction between the maintenance of the self and self-maintenance is the fact that indeed self is an organized system, but by interaction with the environment it not only maintains itself but it also maintains its organization in the same way it was before the interaction. One of the things that I have come to think about the notions of autonomy that emerge out of the work of Varela and Maturana is that they need to be supplemented by some account of what is optimal for the system. There is really nothing in their work that says what would make an organism healthy or diseased, what would make it no longer the same organism, except a minimal notion that the autopoiesis breaks down. There is no notion of any kind of internal self-regulating or self-monitoring systems that ensure a certain kind of robustness and flexibility within the limits of a change while enhancing the organism. The bacterium is doing what it is doing because it is good for it to do that; it is good in the sense that it is optimal for the organism. There is nothing in the notions of self-organization and autonomy in Maturana that addresses that issue. And indeed Maturana refuses those notions. He says those are entirely notions that come from the observer. But I do think that that is not right, although I do not have an account of what specifically in terms of models or mechanisms would be required there. EVA N TH OMP SON :

Let me give an answer to the issue of mechanisms of recognition from my own work and therefore from social science. I believe there is something I call a world-system. And I believe that there are many kinds of units within it, and states are one of the kinds of units. One of the problems that have been plaguing social science the last fifty years is: how do states develop? What do they do in order to develop? Now, I believe that there are certain emergent qualities in the world-system that maintain an unequal distribution of reward and

I M M A N U E L WA L L E R S T E I N :

91

therefore constrain the kinds of things that states could do to improve their situation. Now how do they recognize it? Well, one way they recognize it is reward and punishment. If they take X step, they do not do well and if they take Y step, they do fare well. The problem about recognizing it is that the real rules of the system are not necessarily a rhetoric that people give about the rules of the system, and the rhetoric has to be seen as an intervention itself into the system. But the recognition comes from a trial and error mechanism by which if states do X they do not do better and if they do Y they do do better, because that is somehow conforming to rules that they cannot autonomously set, because the rules result from relational processes at the level of the world-system. PAUL CI LLIERS: I have been thinking about the notion of slowness and I think that it is quite interesting because it is counterintuitive. A complex system needs a certain slowness. But that doesn’t mean it has to be slow all the time. But to know how to be appropriately fast it has to be slow some of the time, most of the time. So the identity of the self is not fixed in any way. AV IV BERGMA N:

I completely agree, except that we need multiple levels in the

dynamical system. I would also move back to this issue of the stability of the self. Despite the malleability of boundaries, it seems to be an ingrained characteristic of different objects that they have boundaries. And besides this slow change there is flexibility. As a result of flexibility there’s a restorative capacity for the objects to restore the boundary of the self, of the object, which prevents abrupt change. Not always, though. But there is a resistance to abrupt changes.

ISTVA N REV:

J E AN-PIERRE D UP U Y: If I remember correctly, Francisco Varela used to make a distinction between organization and structure. By definition organization was an invariant. Actually he used the terms in the opposite way of their usual usage.

92

The structure was what was varied. Different matter, different energy, etc., but by definition, in the case of autonomous or autopoietic system, by definition organization is an invariant. Is that correct? In discussing autonomy I wanted to draw attention to the idea of what differentiates an autonomous system in that precise sense of autonomy that I discuss from other examples of complex macro-emergent behaviors, precisely that there is the organizational closure, and that is different from a connection role or something like that. EVA N TH OMP SON :

I want to emphasize a general epistemological point. In the poiesis discussions concerning reductionism and emergence, we always use in one way or another the concept of physical closure. And there is a great ambiguity in this concept. Physical closure is also indefinitely open at the level of empirical results and mathematical modeling. So the content of this closure is completely unknown. We can say that the concept of physical closure is a regulative idea. But the content is a horizon of science. It is at the end of the scientific evolution; that we know. So we cannot use the concept of physical closure as if it were possible to know, to anticipate what is a physical content. So a sentence such as “physics will never explain that” is without meaning.

J E A N PE T I TOT:

I want to say two things. First I agree completely about the causal closure of the physical. And indeed in my paper I try to say something like that. As phenomenologists know, when we are dealing with perception as lived perception, lived experience, we have a transcendental aspect to consider. And that changes the rules of the games. Local interactions are always a function of the global pattern that happens to be in place at that time. There is a context that is set that makes certain kinds of local interactions possible. The individual neurons are able to do what they do because they have been sucked into a particular dynamic regime.

EVA N TH OMP SON :

93

I would like to turn the conversation for a moment to the idea of chain dynamical hierarchies where we have some emergent property coming out at the global level, but then the pattern that emerges at the global level is captured by the primitives of some other kind of system, which is nevertheless coupled with a lower system. And then that pattern results in some kind of global pattern which can be captured in primitives of yet another system. Nevertheless, all the systems are interactive and media seems to be a key. I believe there needs to be a discussion about the medium in which the emergence is happening and what happens if we couple different kinds of media which constitute a larger system, but a system whose levels now are defined by change in media and the ability to capture the lower level patterns.

K ATH ERIN E H AYLE S:

EVA N THOMPSON : I think that is a great point. I think it is fundamentally important that the case that springs to my mind is cognitive neuroscience. If I understand the relationship between neurophysiological activity and mental processes, where there is the picture of the brain that we get in an MRI where we look at blood flow and what that means, and then the picture we get when we look more directly at the macrophysiological signals, I think this is precisely the case of those different types of systemic patterns involved in different media. That, at least with regard to mental activity, we do not understand very well. We have different windows, correlations that we established, but we do not understand how all that fits together to make us mental beings.

I wonder if this might bear on the question Aviv [Bergman] raised about how the system recognizes an emergence. But it is the re-representation, because the emergent pattern is itself a representation. But then that is re-represented in another media.

K ATH ERIN E HAY LE S:

I would be a little bit cautious about the term representation. It can mean many different things and I am comfortable with some uses, but I would prefer to think of it as how different systems constitute patterns and constitute themselves. And those can then by virtue of their interactions have various kinds

EVA N TH OMP SON :

94

of communicative properties. But the notion of representation, and this is really reflecting my background in cognitive sciences, is a problematic notion because it is just a certain kind of semantic referential relationship. Not that we cannot use the word, I am just saying I would want to know a little bit more about what the word is supposed to mean before resting comfortably in it. But how do we tackle the issue of the emergence of literary meaning? I think that would be a very interesting case study. Would Evan [Thompson]’s scheme or framework apply to that? For instance, would we say that the meaning of a literary text, such as a novel or short story, supervenes on the text itself or does it require the intervention of what some literary theorists call intertextuality? J E AN-PIERRE D UP U Y:

AV IV BERGMA N : To add to your question to the literary people, and to continue on the issue of ambiguity, one of the ways meaning may emerge is by the usage of metaphors. And when we make use of a metaphor we create a new meaning. But at the same time we create ambiguity; but nevertheless this ambiguity first of all is maintained and secondly it is being used in very noble ways in languages, and this is something that would add to the issue of emergence of new meaning. But instead of hindering communication, it actually adds to communication. J OÃ O CA RA ÇA : I am not sure whether by sticking to nonlinearity we are able to capture emergence with the same meaning as we were talking about it this morning. You say correctly that when we have no linearity the Hamiltonian is no longer separable. But in the case of the classical system, the parts of the system are still separable. Now what I ask you is: if the parts are separable, are we really watching an emergent property in the same sense that Jean Petitot was saying this morning?

I think there are complicated issues about what the term part is going to mean in different types of systems and hence what relations among the parts are going to mean. When I was talking about the non-separability of

EVA N THOMPSON :

95

the Hamiltonian, I was really introducing that simply because some physicists and philosophers have pointed to it as a possible case in which there is a relationality that is not atavistic. But in the case of a cell, the non-separability that we see, where proteins turn on genes that synthesize the components of proteins and so on, thus a tangled interaction, is a different kind of non-separability, the network entanglement. In the case of the social context, the examples of non-separability given this morning generally involve intentionality in some sense. That is, they involve human intentional agency and human conceptualization. ISTVA N REV: If I may turn back to your question about the production of meaning, although there is a relational intentionality between the text of the author and the meaning of the text, it is a nonlinear relationship. It is non-tangible. And this non-tangibility, normative material of the text makes possible multiple meanings. There is no direct relationship between the text and the meaning despite the intention of this. There is an emergence of meaning.

I would like to go back to a previous question in relation to what Paul [Cilliers] was talking about and with reference to the kinds of systems, large scale social systems, which interest me: the question is that of time. I think it is the big absent player in this scheme. Open dynamic systems come into existence at some point. That is, they are often products of phase transitions which change the state of an existing system into a new form that is not recognizable as a product of that from which it emerges. And that is certainly a nonlinear kind of process. But then after that, in order to be recognizable over the time-span of its existence, in order to be meaningful to us, such systems have to reproduce themselves. And in terms of large-scale time-span, I tend to think of that as a period in which the nonlinear elements of the system are damped to the extent that the system develops in a linear fashion to reproduce itself as a recognizable entity over time. If I think of a process as being a structure reproduced over time, then that translates for the kinds of systems that I am interested in as a series of structures that get reproduced. And it is not automatic. Those kinds of systems

R ICHA RD E. LEE:

96

are not simply systems that can be defined as systems of communication of relatively dense patterns. But it is systems with particular kinds of relationships that then structure the elements themselves over time. And that is always a contested process. In other words, there is a politics associated with the process, with what kinds of meanings are going to be shared, contested, and eventually win out over time. It all depends on real interactions among real people, particularly groups of peoples. So this idea of an emergent process has a real meaning to me. Not just a theoretical meaning, but the way that the large-scale social systems not only are produced, but then reproduce themselves over time. Jean-Pierre [Dupuy] asked about literary meaning. First of all, I do not think there is anything we can call meaning. There are meanings, plural. One of the most useful contributions of the whole field we call cultural studies is to suggest that every reader reads the text differently and therefore gives different meanings to that text. As an author I can say that I get very upset when some critiques of my work attribute to me meanings that I did not intend; they were not my meanings. Either I am misunderstood reasonably, irrationally, or malignantly or whatever, but that is not my meaning. That is his meaning, using my words. It is shared meanings we all want when we put forward texts; we hope that others will share our meaning. In fact, that is the whole purpose of writing an article or book. We make an argument and hope that the reader out there will say, “Oh my God that’s right, I agree with that.” Of course that rarely happens; only a certain percentage of people ever say that. But we try as authors of texts to get shared meanings. The shared meaning, however, is obviously a consequence of large-scale social processes. Richard [Lee] made reference to the fact that racist concepts of a certain time were largely shared even just fifty or seventy-five years ago, but that they are less shared today or at least people are ashamed of asserting that they share them today. This is a consequence of a certain social evolution of the last fifty years which could be explained, I could explain it in terms way up there of what’s happening in the world-system. Somebody else would try to explain it from things way down here. The Prime Minister of Australia, I M M A N U E L WA L L E R S T E I N :

97

faced with obviously racist happenings during which some White young people attacked people of color, said immediately, “Australia is not a racist society.” And other people said, “yes, but you were the one who raised the whole theme in the last three years.” I do not want to get into Australian politics but it illustrates the fact that he denies that he is a racist. In fact, there are racists today who would deny that they are racists, whereas fifty years ago, many of them would have been proud. So what is the shared meaning? Either the shared meaning is that racism is bad or the shared meaning is that it is bad to admit that one is a racist but one may act on the premise equally. In any case I am not sure there is an answer to your question on what is the literary meaning of a text. There is the intent of the author. Some authors are clearer about that than others. I do not know how many of you have just read the Nobel prize speech of Pinter. It is about “how I write my plays.” And Pinter says, “well I don’t know what I’m going to say. I start with a sort of conversation, a little dialogue and I see the way it develops and by the time I finish it I have kind of figured out what I wanted to say. But I did not know that at the beginning.” I do not know if all playwrights write this way. I kind of doubt it. But it fits with my image of Pinter; in any case the point is that he did not even have a meaning at the outset. Or at least he says he did not. All he had in mind was a bit of conversation that he had heard. He says, “I listen to people and I hear an exchange of conversation. And I say, ‘oh that’s interesting.’ And I sort of try to build on that. And that’s how I start writing a play.” I am sure there are many literary critics who write books and articles about Pinter who know what is being taught. But I do not know that Pinter knows what his meanings are. Or at least he claims he does not know. PAUL CILLI ERS: This is not a direct response to your question but a philosophical comment on the notion of meaning. Meaning is an emergent property of the complex system of science. This is often seen as problematic: as soon as we add the word, complex, it seems that meaning can go anywhere. I think that is based upon a basic misunderstanding of the notion of complex systems. We have a lot of complexities; science is so much influenced by chaos theory that we tend to

98

forget that most complex systems are very robust. Therefore a system of meaning can also be robust without being determinate or fixed. AND REW SAYER: I have a simple point following from Paul [Cilliers]’s, which links back to literary meaning, which I would suggest is more open to endless proliferation than scientific meaning, simply because it less disciplined by practice. Sciences largely do not make impractical interventions in the world. They are successfully disciplined by the practical adequacy and efficacy of their interventions more than our social interventions, although those too are disciplined by the structure of the system. So if I try to use British currency in France I will not get very far. Whereas if I try to interpret a painting I am not subject to those kinds of constraints. There is a pretty simple reason for this: different degrees of ambiguity, proliferation, and meaning, and so on.

You [Andrew Sayer] have been talking about meaning as interpretation. People have been talking about meaning as if it were obvious to us, as if we were seeing emergent properties. Question: do we have to know that the emergent properties emerge? Do we have to notice them? I think that there might be emergent properties that emerge without us noticing and they may actually constrain us. This is obvious, but now that I have said this, why? Some have been talking about meaning as if meaning could be without interpretation—literal meaning and so on. This was a fascinating discussion. But I think there may be other meanings of meaning. I was thinking about systems of meaning, such as nonverbal communication systems of meaning, which also constrain us.

ULLICA SEGERST R ALE :

EVA N TH OMP SON : I agree completely that there could be emergent processes that we would not know of or recognize. I think that that is fundamental and crucial for any workable, useful notion of emergence, particularly in a scientific context. Then there is philosophy. There is an interesting debate between what we could call a rationalist view and an empiricist view. The rationalist view is if we know everything about the most fundamental level, then we should be able to deduce all

99

of the subsequent processes and properties that result from them. For instance, we should be able to deduce liquidity from the hydrogen and oxygen molecules in the boundary. Whereas the empiricist view is that we would have to have the concept of liquidity in advance or we would have to have the experience in advance. This is something Jean Petitot will talk about tomorrow in his paper, and I think it is an interesting question. I am inclined to favor the empiricist view in the sense that we need to acquire certain concepts and only if we have those concepts can we recognize the relevant phenomena. But from that, it does not follow that we know what all the relevant phenomena are in advance; that there are not emergent processes that go unrecognized. With regard to supervenience, it seems to me that supervenience can be invoked as a way of trying to have our physicalist universe, our physicalist cake, and eat our mentality too, our philosophy of mind. The difference between that idea and supervenience of, for example, conflicts of evolutionary theory involving biology where we are talking about fitness, is that there we can give supervenience some scientific content. We might not be able to say everything about what that content is, but we have an idea of the kinds of things we need to investigate and explain in order to fill out the relationship between the notion of fitness and the notion of structural relation. PAUL D UMOUCHE L: We can do that locally. We can say, this is what we have to look at for this organism. We cannot do it in a sense that we can identify some properties of environments, or maybe we can, and some properties of organisms which would be the ones that always enter into fitness. That is where we need supervenience. EVA N THOMPSON : In the case of evolution that thing is historical, so we cannot just specify it independently of the history of the organism and environment. In any case, reduction, in the sense of inter-theoretic reduction, is classically linked to the idea of the unity of science and I see no reason to accept that model of unity in science on the basis of the science as it exists now, a proliferation of many highly specialized and detailed technical forms of knowledge and modeling

100

in theoretic experimentation, but with no unity of that sort in sight. If anything, we have a disunity of science. In that context, notions like emergence seem to be more comfortable and make more sense. We cannot build a computer out of beer cans and strings. So I think it is important to say that embodiment matters. This specific contingent system, the way in which it is put together, what its components are, is important. It is not just a neutral medium that has these emergent properties; embodiment matters. But embodiment matters up to a certain point. This can be seen in biology, where the same functionality can be achieved by different means.

PAUL CILLI ERS:

AV IV BERGMA N: One of the things I am working on is sex determination in flies. We can achieve the same two sexes in flies that are very closely related by completely different means. But we end up with the same functionality. So the basic building blocks of DNA are the same, but if we want to go to a higher level, it may not be the same. With respect to evolution, we would have been completely right if the only thing that evolutionary biologists would worry about were the final outcome. But what we are most interested in, and this is the hardest problem in biology, is to measure selection with respect to a particular unit of selection.

We behave as if we were talking about something completely innocent, as if our topic were related only to epistemology, ontology, history, and philosophy of science and not to ongoing real development in the world. I think that fifteen years ago we would have had completely different discussions, partly because the situation in the real world was different. I just want to introduce the idea that besides epistemological issues, we are talking about ideological issues as well. ISTVAN REV:

J E AN-PIERRE D UP U Y: Andrew [Sayer] mentioned Adam Smith in passing this morning. Many people believe that Adam Smith was a methodological individualist, but he was not. What I have tried to show is that what is happening is a dynamic co-evolution of individuals and the social whole and a circular causality or tangled

101

hierarchy. The individuals are not already there. They are being fashioned by what they bring about in a circular way. Andrew Sayer said that unlike neo-classical economists, Adam Smith said X. But we do have to remember the neo-classical economists today do not read Adam Smith in a single one of their courses. I suspect that if we took a poll, 95 percent of professional economists have never opened Adam Smith. What we have is a utilization of a rhetoric of what Adam Smith said. It is true of Marx, but to a lesser degree. Most people who talk about Marx have at least opened a little bit of Marx.

IMMANUEL WALLERSTEIN:

I want to return to the celebrated sentence—Men make history but they do not know what histories make. I think that it is interesting because in general one considers that emergence adds something to the micro underlying level. The whole is more than its parts. But in that case emergence subtracts. The whole is less than its parts. We lose the intentionality. We have agents which are intentional agents, interactions. One of the main theses of complex methodological individualism is to say that we do not have an emergent intentionality, that emergent social structures are without intentionality and that it is the most catastrophic mistake of holism to attribute intentionality to these emergent structures. It is interesting because in that case emergence regards intentionality negatively rather than positively.

J E AN PETITOT:

ISTVA N REV: I think Marx was saying that the outcome does not correspond completely to the original intention and sometimes people who had the original intention are not able to recognize in the outcome what their intention was.

I am not sure I like the way Marx is being used here. It seems to me what he meant was something very simple and is important to the discussion of parts and wholes. He said that men make their own history but they do not make it as they wish. He said that if they are constrained in a IMMA N UEL WA LLE R ST E IN:

102

field which goes from one to ten, the system constrains them to choose from one to seven, but they cannot choose eight, nine, and ten because the system does not allow it. It seems to me that the system in some sense can change the rules and allow two to eight but that within that, there is a range of possible choice for the units. But the range of possible choice is not infinite and is not self-designated but is a consequence of the structure of the system of which they are a part. And I think that is correct and it does seem to me important in terms of the issue of parts and wholes. When we talk of emergent qualities we are talking of rules that emerge, which limit the range of possible choices by any units that are found within the system. It does not tell the unit, “you must choose number two” but it says, “you can’t choose number nine because that would not function within this system,” even though number nine is theoretically possible. That is part of what we mean by emergent characteristics: it is meaningful constraint on how our units can in fact operate within the system. My point was not exactly that. It was that the interaction between intentional agents is Darwinian and without intentionality. So we can have an intentional interaction that is absolutely not intentional. And it is very important for our conception of history. Many people think that cultural evolution is Darwinian, so without any sort of intentionality. But it is made by intentional agents.

J E AN PETITOT:

I want to respond to that. In the case of either Marx or Smith I think that we have intentionality on the individual level. But we have intentionality at the collective level. In the case of Marx it is obvious because he has this central notion. J E AN-PIERRE D UP U Y:

J E AN PETITOT:

But it is a mistake.

J E AN-PIERRE D UP U Y:

No, not in the case of Marx.

IMMA N UEL WA LL E R ST E IN: There is a problem here of the use of the word intentionality. Intentionality to me means that I want something to happen. That’s

103

my intention. The consequence of a systemic process is not intended by anyone. For Marx, it is History with a capital H, and I am not sure what it is for Smith. But no unit, neither the global unit nor the individual, intended it in the sense that they said to themselves, “I want this to happen.” It happens despite them. That is the whole point. So there is in that sense an end, but it is not intended. We are, at least I am, using the term in the sense that it is used in philosophy today, which is actually borrowed from the scholastic name, which has nothing to do with having an intention. Intentionality means, well . . . J E AN-PIERRE D UP U Y:

IMMA N UEL WA LL E R ST E IN:

J E AN-PIERRE D UP U Y:

Meaning.

Well, yes. Aboutness.

AV IV BERGMA N : From an evolutionary point of view, evolution is not something which is directional. It is a process, a very highly constraining wall that does not allow all possible outcomes.

But I thought that when we have evolving characteristics that did not exist at time t1 but do exist at time t2, it is a consequence of the unit adjusting to reality; so that the giraffe gets a longer neck because that is the only way those giraffes with shorter necks die out. And the ones with longer necks survive because they need to. It is what I understood evolutionary biologists to be telling us, that units somehow respond to realities, or they disappear. Over time they develop new characteristics that make them better able to function in the environment of which they are a part. IMMA N UEL WA LL E R ST E IN:

Yes, but this is naïve. We all share extremely robust elements, which are not the result of selection: five fingers, not the result of selection; two

AV IV BERGMA N :

104

eyes, not the result of selection from this or that. In evolutionary biology, we end up with enormous numbers of side effects that have nothing to do with direction. For example, the sunflower seeds that are beautifully spiraled have nothing to do with genetics. The pattern is due to biomechanical constraints on the flower, so we end up with a beautiful spiral. And these are the underlying rules that come from a much lower level that ended up in a beautiful pattern. But this pattern has nothing to do with selection for pattern. I want to make a general point about the importance of constraint. I would go so far as to say that constraints are the necessary conditions of complexity, because complexity is not just degrees of freedom. If we just have a system of many degrees of freedom, we do not necessarily have interesting behavior. Anything is possible. So the structured complexity which is interesting can only arise if the complexity is constrained.

PAUL CI LLIERS:

I also have a question concerning constraints. I quote you [Evan Thompson], “the context-sensitive constraint is one that synchronizes and correlates previously independent parts into a systemic whole.” What is the exact relation between emergence and context-sensitive constraints? Are they only vaguely associated? Is one more general than the other? Are they quite synonymous? Can you elaborate the relation between emergence and context-sensitive constraints?

J E AN PETITOT:

Context-sensitive constraint, as I articulated there, is an ingredient or element that goes into making something emergent, or to put it another way, it is a way of explicating one thing that is involved in something that is being emergent, in the particular sense of emergence that I am concerned with here—complex systems emergence. The second-order contextual constraint is where the organization, the closed loop in the autocatalytic case, becomes a context or constraint on the relationship among the elements. So that is a further complexification that makes possible a more robust kind of emergence. EVA N T H O M P S O N :

105

J E AN-PIERRE D UP U Y: Your point is highly important, that is, the notion that constraints can be considered to be responsible for traits that are not selected for. I see many illustrations of that, perhaps not in the biological area, but certainly in cultural evolution. For instance, the very limited number of different kinship systems.

106

SESSION III

Reduction and Emergence in Complex Systems Jean Petitot

I INTRODUCTION

C

lassically, a reductionist thesis posits that complex high level phenomena, structures, and processes can be reduced, as far as their scientific explanation is concerned, to underlying lower level phenomena, structures, and processes. The most paradigmatic and best-investigated example is the reduction of macroscopic thermodynamics to microscopic molecular and atomic movements (temperature = mean kinetic energy per degree of freedom, etc.). Let us begin with some conceptual precisions. 1. Reductionism can be a particular scientific thesis concerning a specific scientific theory: it is precisely the case with the reduction of macro thermodynamics to micro statistical mechanics. But it can be also a general metaphysical claim on the ultimate nature of reality, as is the case with different forms of monism. Idealist monism posits the universal reducibility of reality to mind while materialist monism posits the universal causal reducibility of reality to matter and energy. In this paper we will consider only scientific reductionism. 2. Reductionism can concern theories dealing with empirical data and in that case focuses on the problem of causality. But it can also concern purely linguistic and formal theories. For instance, lexical definitions or logical

107

axioms consist in trying to reduce complex contents to a small list of primitive contents. In mathematics, many theories deal with the possibility of eliminating higher order concepts, objects, or axioms in proofs and deflate rich theories to more restrained ones (theorems of elimination): for instance one can prove that quantifiers can be eliminated in algebraic geometry (Tarsky-Seidenberg theorem) or that a proof using the axiom of choice can be transformed into a proof without the axiom of choice, etc. In this paper, we will be concerned only with theories having an empirical content. 3. Scientific object-oriented reductionism can be ontological or methodological. It is ontological when it concerns explanations in terms of primitive objects (atoms, neurons, etc.) and methodological when it concerns deflationist nominalist explanations (Occam’s razor). There are well-known examples of eliminative methodological reductionism that have eliminated pre-scientific speculative concepts and entities such as “vital,” etc. Today, a very important debate in cognitive science has to do with the eliminability of “mental” or “conscious” concepts and their reduction to neural concepts (see e.g., the Dennett/Chalmers controversy). In this paper we will be concerned with “ontological” reductionism. However, as we endorse a transcendentalist perspective, we prefer the expression “objective reductionism.” Indeed, transcendental conceptions introduce a strong opposition between the concepts of ontology and objectivity. They posit that scientific theories cannot concern any independent transcendent reality, but only an objective reality mathematically reconstructed from phenomena. For them, any theory necessarily includes epistemic background structures (called priors). 4. In our narrow, scientific, empirical, and objective sense, reductionism concerns mainly complex systems that possess at least two levels of reality: a micro underlying level where a great number of elementary units are in interaction and a macro-emergent one where macro self-organized structures emerge (Thom 1972). In such a perspective, reductionism is inseparable from converse concepts such as emergence, supervenience, or functionalism. Functionalism means that macrostructures that have a functional

108

role can exist only if they are materially implemented in an underlying material substrate, but are at the same time, as functionally meaningful structures, largely independent of the fine-grained physical properties of the substrate they are implemented in. The paradigmatic example is the opposition software–hardware in computer sciences (see philosophers like Putnam, Fodor, Pylyshyn, etc.), but functionalism also applies in natural sciences, where it is an aspect of emergence. 5. There is general agreement on the fact that in complex systems that have different levels of reality at different scales, there exist collective behaviors ruled by laws that are not the laws of the micro underlying level. It is the case for critical phenomena, percolation, self-organized criticality, reactiondiffusion equations, dissipative structures, turbulence, cellular automata, neural networks, ant colonies, swarms, stock markets, etc. According to one’s conception of laws, one can develop different conceptions of this empirical fact. a. Eliminativism and epiphenomenalism: laws being only empirical regularities lacking any objective (and a fortiori ontological) content (Hume’s empiricist thesis), emerging structures are purely epiphenomenal and can be scientifically eliminated salva veritate. b. Holistic realism (the converse of the previous position): laws being real in the ontological sense, the emerging level possesses an ontological reality and cannot therefore be reduced. c. Causal reductionism and objective emergentism: laws being objective, that is, at the same time empirically grounded and mathematically formalized, the emerging level has no ontological content but is nevertheless much more than a simple empirical regularity. It is causally reducible to complex interactions at the micro underlying level but it shares nevertheless some empirical and theoretical autonomy. We will be concerned here with this third type of reductionism. 6. The main difficulty that has to be tackled in such a perspective is the relation between causal reduction and theoretical autonomy. Mathematics plays the fundamental role here. Indeed, the formal equivalent to causal

109

reduction is mathematical deduction. But deducibility is a syntactic property and doesn’t entail any evident conceptual derivation (it is for that very reason that mathematics constitutes an authentically “synthetic” knowledge even if proofs are “analytical”). Therefore the fact that the structures and properties of the macro level can be mathematically deduced from the micro level doesn’t mean that the representational content of its conceptual description can be reduced to the representational content of the microlevel. A very spectacular example is that found in statistical physics where magnetic critical behaviors can be classified, via the renormalization group, in universal classes independent of the specific fine-grained physical structure of the substrate. The renormalization group is a dynamic method that enables us to define these universal behaviors as attractors of a certain dynamics on the space of Hamiltonians. Near the critical temperature (T = Tc ) the macro variables of the system (magnetization, specific heat, magnetic susceptibility, etc.) follow power laws (T) (where  is called the critical exponent of the variable). Empirical data have shown that there exist universal classes of critical exponents linked by very precise relations. These universal classes depend only on very general abstract dimensional and symmetry properties of the substrate and not on its detailed physical structure. The point is that if you prove mathematically that such a critical phenomenon arises from a symmetry breaking of an order parameter, this doesn’t mean that this macro and abstract symmetry breaking modeled via group theory, has something to do with the concept of a spin of a particle. Causal reduction that parallels mathematical deduction is not a conceptual reduction. Moreover, the universality of critical exponents—that is the existence of invariants—proves that the emerging critical phenomenon under consideration has some measure of autonomy and belongs to an autonomous level of reality. As the physical examples of reductionism are very well known, and because we think that it is always better to discuss precise examples, we will focus in some

110

detail on two examples. The first belongs to cognitive sciences and concerns phenomenology of perception and visual neurosciences. The second belongs to social sciences and concerns Hayek’s catallaxy and evolutionary game theory.

II COGNITIVE SCIENCES

In cognitive sciences, reductionism concerns in particular the reducibility of mental states, representations, contents, structures, events, and processes to neural activity. The most difficult challenge is that of the reduction of consciousness. For materialists such as Dan Dennett, mind is epiphenomenal and without any causal efficiency, and must be reduced to neurobiology. Conversely, for philosophers such as David Chalmers, phenomenal consciousness is irreducible and there will always remain an “explanatory gap.” Let us recall very briefly the classical opposition introduced by Ned Block between access consciousness (object-oriented awareness, intentionality, representational contents) and phenomenal consciousness (first person experience, subjective evidence, qualia). It is phenomenal consciousness that raises the main problem for reductionism and many philosophers consider indeed that it yields refutations of physicalism. As the problem is extremely subtle, we will present a very specific example, which can be treated in detail. 1. Phenomenal Experience and Mathematical Reductionism One of the main difficulties encountered in clarifying the concepts of emergence and reduction in cognitive science is that they entangle two completely different problems: 1. A problem of emergence of macrostructures out of physiological mechanisms processed in an underlying microneural substrate N. 2. A problem of matching or agreement between some objective emergent macrostructures and some subjective phenomenal experience E.1

111

If we seek to correlate directly E with N we will run into great hurdles, since the emergence problem is entangled with the opposition “third person objectivity”—“first person subjectivity.” We will argue that a naturalist emergentist answer to the problem of existence of “Neural Correlates of Consciousness” (NCC) can be philosophically clarified only if the phenomenal emergence N ¡ E is “factorized” (as mathematicians would say) through a macro natural level M according to the following diagram: 3 嘷

Agreement

E= Phenomenal Experience

2 嘷 M= Macro emerging level

Naturalist emergence 1 嘷 N= Micro neural level The factorization disentangles the problem: 1. The “vertical” emergence 1 ¡ 2 belongs to the naturalist domain. The problem it raises is not a problem of naturalization but a problem of deducing 2 from 1. As we saw in the introduction, it is a problem of mathematical reductionism, functionalism, ontological dependency, etc. 2. The “horizontal” agreement 1 £ 2 raises a completely different problem of matching representational contents, of naturalization, etc., but not a problem of emergence.

112

As we will focus on problems of visual perception concerning the phenomenal experience of spatial forms, we will call the macro naturalist level 2 a morphological level. The factorization of 1 ¡ 3 through 2: 1 ¡ 2 ¡ 3 reveals the methodological obstruction to any clarification of the problem: 1. The micro neurophysical level 1 is well known thanks to findings in neurobiology and psychophysics; 2. The phenomenal level 3 is well described, thanks to phenomenological descriptive eidetics; 3. But the macro morphological level 2 is largely unknown: it cannot be conceptually derived directly from 1. It can only be mediately mathematically deduced. Therefore, it depends entirely on deep mathematical achievements. In the absence of 2, arguments concerning a supposed emergence 1 ¡ 3 become metaphysical and scientifically undecidable. But in the presence of 2 they depend hic et nunc on the contingent progress of mathematics. It is the main problem of any anti-reductionist thesis positing that “it will be never possible to reduce this to that.” Physicalism must be a mathematical physicalism and the claim that something cannot be proved makes sense only for internal limitations of mathematics and cannot therefore be relevant for physical modeling. What is called the explanatory gap between 1 (neurophysiology) and 3 (phenomenal experience) depends itself entirely on 2. Indeed, the emergence 1 ¡ 2 is of the very same type as the one that is well known in statistical physics (thermodynamics, phase transitions, and critical phenomena). The macro morphological structures at level 2 are nomologically deducible from the micro interactions at level 1. If 2 is lacking, as there cannot be a direct content matching between 1 and 3 for obvious reasons, there exists an explanatory gap. But the stronger the mathematical deduction 1 ¡ 2, the weaker the explanatory gap. And we see no

113

principled reasons to reject the regulative idea that, at the limit, 2 would become sufficiently strong to fill the gap. We will focus on an example and comment on its foundational issues. It is a very limited but very striking example that concerns some recent works on entoptic vision and visual hallucinations. It can be summarized by the comparison between the morphologies displayed at figure 3.1. These morphologies represent spontaneously hallucinated geometric visual patterns: Left: we have an example of 3 (phenomenological report), since these are drawings of visual hallucinations E experienced by subjects. Right: we have an extraordinary example of mathematical models of morphological macrostructures M, since these are models very recently constructed by mathematicians.

Figure 3.1. Comparison between drawings of visual hallucinations and mathematical models (from Bressloff et al. 2001).

114

I will try to present in not too technical a way the key result that these morphological structures M are mathematically deducible from the encoding of a neural functional architecture in the synaptic weights of a neural net. They provide therefore a striking example of mathematical deduction 2. This non-intuitive and non-trivial result has to be proved and cannot, by any means, be anticipated purely conceptually. 2. Entoptic Vision and Phenomenal Consciousness 2.1. Entoptic Vision and Hallucinations Entoptic vision concerns some geometrical patterns that are perceived after exposures to a violent flickering light, absorptions of substances such as mescaline, LSD, psilocybin, ketamin, some alkaloids (peyote), or near-death experiences. I have no personal experience of the matter, but it is known that subjects spontaneously and vividly see typical forms: tunnels and funnels, spirals, lattices (honeycombs, triangles), cobwebs, etc. These typical forms can operate on the drawings of any type of objects, as we find it in the paintings from Indian Mexican tribes. Such visual imagery is dynamic and the illusory contours usually explode from the center of gaze to the periphery, appearing initially in black and white before bright colors take over, and eventually pulsate and rotate in time as the experience progresses (Frégnac 2003). These illusory forms were already classified a long time ago (1928) by the great neurophysiologist Heinrich Klüver (1897–1979), who provided many clinical reports on them. Klüver was a student of Max Wertheimer and introduced Gestalt psychology in the United States. He called the most typical forms planforms.

115

2.2. The Status of the Example The example is particularly interesting because it yields a case of pure phenomenal consciousness. 1. It is clearly a qualitative and phenomenal experience. 2. It is clearly subjective. 3. It is neither conceptual nor linguistic but is the object of graphical reports. 4. It is not a mode of access consciousness. 5. It is without any intentional, representational, and semantic content. 6. It is not embedded in a complex “narratively” organized stream of consciousness. 7. It is not involved in sensory-motor loops. It is therefore particularly adapted for probing a possible matching between a phenomenal content and a neural dynamics. 2.3. The NIMH Program Before coming to the mathematical deduction 1 ¡ 2, let us mention programs at the National Institute of Mental Health (NIMH) that probe the neuroreceptors with varied substances. In the receptor space each substance shifts the balance of activity of the brain away from the origin, by a vector representing the profile of binding affinities at different receptors. In a brain-centered reference frame, the origin is based on absolute levels of activity at each receptor population. The state of the brain is constantly on the move. We can think of it as a complex dynamical system, in which the trajectory follows high-dimensional orbits, and switches among many “attractors.”

116

In this dynamic reference frame, drugs will create a perturbation along the binding vector, thereby pushing the system into a new attractor. (Ray 2004). 3. Spontaneous Emergence of Geometric Visual Patterns 3.1. Reference Papers The first paper on the subject was the one by Ermentrout and Cowan (1979), “A mathematical theory of visual hallucinations.” Bard Ermentrout is one of the major specialists on synchronization in networks of oscillators. Jack Cowan is a specialist on neural networks (Wilson-Cowan equations). Recently, the subject was completely revisited by Paul Bressloff, Jack Cowan, Martin Golubitsky, Peter Thomas, and Matthew Wiener: “Geometric Visual Hallucinations, Euclidean Symmetry and the Functional Architecture of Striate Cortex” (2001). Paul Bressloff is a specialist on models of vision and Martin Golubitsky one of the major specialists on bifurcations in dynamical systems. See also Bressloff & Cowan’s recent paper. “The Functional Geometry of Local and Horizontal Connections in a Model of V1” (2003). 3.2. The Simple Neurons of V1 The receptive field (RF) of a visual neuron is the domain of the retina to which it is connected through the neural connections of the retino-geniculo-cortical pathways. It is decomposed into ON zones (lighting induces a depolarization and a stimulation of the rest spiking response) and OFF zones (lighting induces a hyper-polarization and an inhibition of the rest spiking response). Sophisticated techniques enable investigators to record the level curves of the receptive profiles (DeAngelis 1995). For the receptive profiles of the simple cells of V1, the level curves can be modeled by third order derivatives of Gaussians similar to Gabor functions (see figure 3.2, next page).

117

Figure 3.2. The receptive profile of a simple orientation cell of V1. 1: schematized structure. 2 and 3: mathematical model. 4: empirical recording (from DeAngelis 1995). Color versions of the figures are at http://www.crea.polytechnique.fr/JeanPetitot/ColorFigures_Petitot.pdf.

Due to the structure of its RP, each simple cell of V1 (which operates as a filter by convolution on the optical signal) detects a preferential orientation, which can be assimilated with its RF-content. We define therefore a first (low) level of abstraction by saying that these neurons detect, at a certain scale, pairs (a, p) of a spatial (retinal) position a and of a local orientation p at a. Pairs (a, p) are called contact elements in differential geometry. 3.3. Neural Nets and Hopfield Equations Now, we must do a little bit of mathematics if we want to understand the argumentation. I hope that readers are not too phobic against mathematics, but it is impossible to discuss reductionism and emergence without looking closely at mathematical models.

118

The authors work in the Cartesian product (called a fibration)  : V = R  S1 ¡ R where the base space R is the plane of the retina and the fiber S1 the space of orientations p. We choose in the total space V modeling V1 coordinates (a, ) with a = retinal position and  = the angle of the orientation p. Let z(a, , t) be the activity of V1 at the neuron (a, ) at time t. If we know the real-valued function z(a, , t), we know the evolution of the activity of V1. According to the physicalist scheme, we look for the PDE (partial differential equation) governing the evolution of z. This equation must express the temporal derivative of z as a function of z. We start with standard Hopfield equations for a discrete neural net. Let ui , i = 1, . . . , N be formal neurons with activity zi(t). If time and space are discrete, standard Hopfield equations are local rules of interaction enabling us to compute the activity zi(t) of the neuron ui from its interactions with other neurons uj and from the input. The interactions are governed by what are called “synaptic weights,” that is, weights of the connections between neural units. We get j=N zi(t + 1) =



wij(zj(t)) + hi(t)

j=1 where  is a nonlinear gain function (with (0) = 0), wij the weight of the connection between the neural units ui and uj , and h the external input. We see that the activity of the unit ui at time t+1 is essentially a nonlinear weighted sum of the activity at time t of the units to which it is connected. To find the continuous limit when the temporal step t evolves from t = (t+1) – t = 1 to an infinitesimal step dt, one can put zi(t+1) = F(. . .) into the form zi(t+1)–zi(t)

/ ((t+1) – t) = –z (t) + F(. . .) i

119

So, at the limit, when time is continuous and space discrete, we get a system of N ordinary differential equations: d zi(t) ——— = – zi(t) + dt

j=N



wij(zj(t)) + hi(t)

j=1

On the other hand, if space is also continuous, the index i becomes a spatial variable v and we get a partial differential equation: ∂z(v,t) ——— = –z(v,t) + ∂t



V

w v | v'  (z(v' ,t))dv' + h(v,t)

w 具v | v' 典 being the weight of the connection between the neuron v and the neuron v'. The authors use the following Hopfield equation: ∂z(a,,t) ——— = ∂t –z(a,,t) + — 



0

R

w a,|a' ,'  (z(a' ,' ,t))da' d' + h(a,,t)

where w 具a,  | a', ' 典 is the weight of the connection between the neuron v = (a, ) and the neuron v' = (a', ' ) ,  a parameter of decay and a parameter of excitability of V1. The parameter is the simplest example of localization of a substance in a receptor space and of the fact that the substance shifts the balance of activity of the brain away from the origin. We want now to deduce some emergent global properties of such a typically complex system. To explain how the morphological macrostructures M (2) can emerge from the dynamics of the neural substrate N = V1 (1), the key point is that the functional architecture of V1 can be encoded in the synaptic weights w具a,  | a', ' 典.

120

3.4. The Functional Architecture of V1 Many experimental results are now available concerning the functional architecture (FA) of V1. They show that V1 is organized through a double structure: 1. A “vertical” structure given by the retino-geniculo-cortical retinotopy. The hypercolumns (discovered by the Nobel Laureates David Hubel and Torsten Wiesel in the early sixties) are neural micromodules which associate retinotopically to each position a of the retina R a full exemplar Pa of the space P = S1 of the orientations p. This means that the fibration structure of V = V1 is neurally implemented. 2. A “horizontal” structure given by the horizontal cortico-cortical connections which connect neurons in different hypercolumns (Das and Gilbert 1995). This double structure gives rise to geometrically well-organized structures called pinwheels. V1 is reticulated by a network of singular points which are the centers of pinwheels. Locally, around these singular points all the orientations are represented by the rays of a “wheel” and the local wheels are glued together in a global structure. The relation between pinwheels rays (colors) and preferred orientations (white strokes) is shown in figure 3.3, which represents the FA of the area V1 of a cat.

Figure 3.3. The functional architecture of the area V1 of a cat (from Shmuel and Grinvald 2000).

121

The fundamental feature of this functional architecture is that cortico-cortical connections connect neurons of the same orientation in different hypercolumns. This means that the system implements what is called in differential geometry a parallel transport and is able to know, for b different from a, if the orientation q at b is the same as the orientation p at a. 1. The retino-geniculo-cortical “vertical” connections give an internal meaning for the relations between (a, p) and (a, q) (different orientations p and q at the same point a). 2. The “horizontal” cortico-cortical connections give an internal meaning for the relations between (a, p) and (b, p) (same orientation p at different points a and b). q p=q p=q a Vertical connectons : a=b p⫽q

p

a

b

Horizontal connections : a⫽b p=q

Moreover cortico-cortical connections connect neurons coding pairs (a, p) and (b, p) such that p is the orientation of the axis ab. As was summarized by one of the main specialists of the field, William Bosking, “the system of long-range horizontal connections can be summarized as preferentially linking neurons with co-oriented, coaxially aligned receptive fields” (Bosking et al. 1997, 2112). This functional architecture of V1 is basic for visual perception. It explains the Gestalt principle of good continuation and the striking capacities of integration of contours and boundaries, even of illusory contours, as in Kanizsa’s well-known experiments (figure 3.4, right) (1979). For instance, it explains the pop-out of a curve of local contact elements (ai, pi) against a background of random elements according to the concept of association field of Field, Hayes & Hess experiments (1993) (see figure 3.4, left).

122

Figure 3.4. Left: The association field of Field, Hayes and Hess experiments (from Field et al. 1993). Right: Curved Kanizsa illusory contours.

3.5. The Encoding of the Functional Architecture into the Synaptic Weights Let us now encode this functional architecture in the weights w具a,  | a', ' 典 of the neural net V modeling V1. We must translate the geometrical properties of FA into formulae concerning w具a,  | a', ' 典. o The local vertical connections inside a single hypercolumn yield a term: w a,|a' ,'  = wver( – ' ) (a – a' ) where is a Dirac function and imposes a = a' ( (a – a' ) = 0 if a – a'  0, (a – a' ) =  if a – a'= 0, and the integral of is 1). This formula means that, in a single hypercolumn (condition a = a' ), the connection between a neuron detecting the orientation  and a neuron detecting the orientation ' is weighted by a function wver( – ' ) depending only upon the difference of phases  – '. o The lateral horizontal connections between different hypercolumns yield a term:

123

w a,|a' ,'  = whor(a – a' ,) ( – ' ) where the factor ( – ' ) imposes  = ' and expresses the fact that the horizontal cortico-cortical connections connect pairs (a, p) and (b, q) with p = q and that their weight is only a function whor(a – a', ) of the vector a – a' and the common orientation  of p and q. o Moreover, the fact that p = q = ab, that is the property of coaxiality, is expressed by the fact that the direction of the vector a – a' is  and that the function whor(a – a', ) depends only on the measure s of a – a' with respect to the unit vector e in the direction  : whor(a – a' ,) = whor(s) (a – a' – se ) We must emphasize the fact that the synaptic weights w are E(2)-invariant under the roto-translation group of motions of the plane E(2) and that the PDE satisfies therefore an E(2)-symmetry (E(2)-equivariance). 3.6. Dynamically Emerging Morphologies and Bifurcations We are now ready to explain mathematically why macro morphologies can spontaneously emerge in such a geometrically structured neural network. We suppose that there exists no external input, i.e., h = 0. For = 0, the rest state z  0 is trivially the state of the network and it is stable (you see nothing). Now, the analysis of the PDE shows that, as the parameter increases, this initial activation state z  0 can become unstable and bifurcate spontaneously for some critical values c of . The increase of models an increasing of the excitability of V1 due to the action of some substances on the nuclei (locus coeruleus, raphé) that produce neurotransmitters (serotonin, noradrenalin). The new stable activation states present spatial patterns generated by an E(2) symmetry breaking. The bifurcations can be analyzed using classical (but sophisticated) methods:

124

1. Linearization of the PDE near the solution z  0 and the critical value c. 2. Spectral analysis of the linearized equation. 3. Computation of its eigenvectors (eigenmodes). Figure 3.5 shows some examples of eigenmodes in V. Here, what is of utmost importance is the fact that the eigenmodes can be mathematically deduced from the Hopfield equations while, at the same time, they share some autonomy because they essentially depend on very abstract symmetry breakings independent of the fine structure of the network. In that sense mathematical deducibility is essentially different from theoretical (conceptual) physical reducibility. As we have already emphasized it, the situation is quite analogous to that found in statistical physics, where critical behaviors can be classified, via the renormalization group, into universal classes independent of the specific fine-grained physical structure of the substrate.

Figure 3.5. Some examples of eigenmodes in V (from Bressloff et al. 2001).

125

Figure 3.6. The retinotopic conformal map, mapping the retina on V1 (available as “vision. ppt” at http://www.slideworld.org/viewslides.aspx/103873).

3.7. The Perception of Emerging Patterns as Virtual Retinal Images The last step is to reconstruct from the eigenmodes in V1 the corresponding virtual retinal images that would be hallucinated. For that, we must take into account the retinotopic conformal map mapping the retina on V1 (see figure 3.6). A good dipole model is for instance Log[(z+0.333)/(z+6.66)]. Lines in V1 correspond to spiral on the retina (see figure 3.7).

Figure 3.7. Lines in V1 correspond to spiral on the retina (Petitot).

126

Figure 3.8. Klüver’s planforms are isomorphic to eigenmodes of the bifurcated solutions of the neural network in the synaptic weights of which the functional architecture of V1 has been encoded (from Bressloff et al. 2001).

If we apply the inverse of the conformal map to the eigenstates of the PDE, we get quite exact models of Klüver’s planforms. Klüver’s planforms are isomorphic to eigenmodes of the bifurcated solutions of the neural network in the synaptic weights of which the functional architecture of V1 has been encoded. Many other eigenmodes can be deduced from this bifurcation scheme. Here are some examples (figures 3.9 and 3.10):

Figure 3.9. Other examples of eigenmodes (from Bressloff et al. 2001).

127

Figure 3.10. Other examples of eigenmodes (from Bressloff et al. 2001).

Annex for mathematicians: The spectral analysis of the PDE is rather technical. o After having linearized the PDE around the trivial solution z  0, we look for solutions of the form: z(a, , t) = etz(a, ). When = 0,  = – and the solution is z(a, , t) = e−tz(a, ). o The solutions are stationary only if  = 0. Otherwise they decay ( < 0) or diverge ( > 0) exponentially. In that case, saturation imposed by the nonlinear gain function stabilizes them. o We get an equation (for the eigenvalues ) of the form: z(a,) = –z(a, ) + '(0)



[ w

d' ( – ')z(a,')—— +   R whor(a – a',)z(a',)da' 0

ver



]

where  is a constant measuring the relative strength of the vertical and horizontal connections.

128

o Using Fourier series of z, wver and whor for the periodic variable  and Fourier transforms for the spatial variable a, and identifying the coefficients of the terms in the two sides of the equation, we get dispersion relations of the form  = –+ F(…) where F is a (complicated) function of the Fourier coefficients. o For = 0,  = – and the state decays exponentially to 0. When increases,  will vanish for a certain F(wFver,n, wFhor(q)) (where wFver,n and wFhor(q) are Fourier coefficients) and a certain critical value c. o The bifurcation activates the corresponding terms in the Fourier series and transforms, which elicitates certain eigenmodes. o The key point is that the symmetries imply strong constraints. It is the most technical part of the deduction. 4. Philosophical Discussion We have now at hand sufficient data for testing some philosophical issues. Let us first recall our initial factorization 1 씮 2 ¡ 3: 3 嘷 E= Phenomenal Experience

Agreement

2 嘷 M= Macro emerging level

Naturalist emergence 1 嘷 N= Micro neural level

129

4.1. The Agreement 2 £ 3 As we test the agreement between the macro-emergent level M (2) and the phenomenal experience E (3), we see that it is extremely strong, very much stronger than a mere correlation. It is in fact the strongest form of content matching since it is, at the limit, an isomorphism. But we have to be very cautious here: contents are not conceptual contents but geometric nonconceptual contents. It is a geometrical isomorphism that obtains between the geometrical content of an experienced intuitive space and the geometrical content of a mathematical formal space. We can therefore make the factorization 1 ¡ 2 ¡ 3 much more precise. 3 嘷 E= Phenomenal Experience

Isomorphism

2 嘷 M= Global emerging geometry

Naturalist emergence 1 嘷 N= Micro neural physics We see that the reductionism-emergentism game is not played between holistic phenomenal experience and local neurophysics but between global phenomenal experience, local physics, and emerging global geometry. The isomorphism is a matching between morphologies that are geometrical contents; it is not problematic

130

as such. What is problematic is the matching of the two types of spaces where these morphological contents are embedded and framed. We meet here what is for me the truly hard problem of consciousness, from Kant’s transcendental aesthetics to Poincaré and Husserl: what is the link between space as a phenomenal “form of intuition” and space as a geometrical construct. Space is a multisensorial common format constitutive of phenomenal consciousness. Spatial representations are nonconceptual and nonpropositional phenomenal forms. Space is the form of the phenomenality itself. It is internal and subjective, rooted in the internal temporality of first person experience, but at the same time it is the universal form of third person external objectivity. Moreover it is not only a phenomenal “form of intuition” but also a geometric entity which can be mathematized. If we accept the identity of these two kinds of spatiality, then the matching 2 £ 3 becomes a very strong nonconceptual type-type identity content matching: “eigenmodes” £ “planforms.” If, on the contrary, we don’t accept this identity, then the matching becomes an isomorphism between morphologies embedded in two different (phenomenal and geometric) ambient spaces (which is already a noteworthy result). 4.2. The Emergence 1 ¡ 2 as Neural Correlate The mathematical model enables explanation by means of a long mathematical deduction the morphological emergent level M from the physical activity of the underlying neural substrate N. In that case, there is no explanatory gap (the problem is rather the phenomenality of space, as we have just seen). We meet here a probing example of nonreductionist physicalism that satisfies three constraints that our colleague Jean-Michel Roy called attributive, ontological, and explanatory. o M (2) belongs to N (1) due to the natural properties of N that are necessary and sufficient for its instantiation.

131

o 1 ¡ 2 is stronger than supervenience but eventually weaker than type-type identity if the two modes of space are not identical. o 2 is causally and mathematically reducible but manifests some measure of autonomy since its geometry is an emergent structure. Moreover, in that case, V1 provides a NCC (neural correlate of consciousness) in Chalmers’ sense for the visual experience E. Our mathematicians have proved that N = V1 (1) is what is called a bridge locus for M (2), and through M, for E (3). A NCC for content is a minimal neural representational system N such that the representation of a content in N is sufficient, under conditions C, for the representation of that content in consciousness (Chalmers 2000, 31). Of course, it is well known that V1 doesn’t correlate well with high-level perceptual structures that possess an intentional and semantic content (perception of objects). Higher areas (e.g., IT, inferior temporal) correlate much better. But it is not a problem here because V1 correlates nevertheless extremely well with the geometric morphologies we want to explain. And it is truly minimal since it implements minimally its functional architecture. Now, the main difficulty is to define correctly what we consider to be the representational content of N = V1. It is here that the key concept of functional architecture becomes really unavoidable. We know the local physical representational content (RC) of the simple individual neurons of V1: receptive field and orientation preference. Alva Noë and Evan Thompson are clearly right when they claim that “RF-content is too thin to sustain a match with perceptual experience” and “[i]t is difficult to see how it [a structural coherence] could be built up . . . out of RF-contents atoms” (Noë-Thompson 2004, 90, 14). But we must also take into account the functional architecture of V1 where “vertical” retinotopic connections are completed with “horizontal” cortico-cortical connections. Through its FA, V1 does acquire an emerging global geometric content according to the following table:

132

Local neural RC 

Receptive fields, orientation preference,

Functional architecture



Retinotopy, horizontal connections

Global geometrical RC



Eigenmodes

o If we accept the deduced eigenmodes as a part of the representational content of the neural system N = V1 as a whole, then Bressloff ’s example show that a NCC does exist. o If we don’t accept that contents are logically transitive (in other words that a content C2 mathematically deducible from a content C1 is a part of C1), then we have to define what type of representational content these geometric emergent contents can be. In any case, the fundamental limit of the critiques raised against the NCC and the MCD (matching content doctrine) is in general to underestimate the crucial role of neural functional architectures. We will call deduced RC (RC2) a structure mathematically deduced from the basic RC (RC1) of a neural system N. According to one’s conception—analytic or not—of mathematics, one will accept or not RC2 as part of the RC of N. But in any case the problem is that RC2 cannot be conceptually inferred (without mathematics) from RC1 and depends hic et nunc on the contingent progress of mathematics. 4.3. The Philosophical Status of the Example To conclude this part, let us specify a bit more the philosophical status of Bressloff ’s example.

133

1. It is neither eliminativist reductionism nor epiphenomenalism: the eigenmodes that model Klüver’s planforms exist as geometrical idealities independently of their fine-grained neurophysiological implementation; their necessary and sufficient condition is the functional architecture. 2. It is not dualism: the correlation N £ M is not a relation between two heterogeneous ontological domains. The emerging morphologies are geometrical idealities and there exists no specific ontology of mathematical idealities (no Platonism). 3. It is not ontological emergentism in the sense that the emerging structures would be not only autonomous but also irreducible. Here, they are at the same time autonomous and deduced. 4. It is not functionalism in the classical sense of a mere relation of implementation of the upper macro level 2 into the lower micro level 1. The structures of 2 emerge from a dynamical bifurcation process, namely a critical phenomenon of symmetry breaking. 5. It is not representationalism that eliminates the nonrepresentational properties of mental contents: space is not a conceptual and a representational content but the general condition of possibility of intentional and semantic contents referring to the external world. 6. It is dynamical emergentism: the global geometry M can be deduced neither from the behavior of isolated individual neurons nor from the collective behavior of a differently organized neural net (but it is multi-realizable as long as the functional architecture obtains). Nevertheless it is law-like and can be nomologically deduced from the functional architecture of the NC system N = V1. The most appropriate philosophical approach seems to be a variant of the doubleaspect theory of Thomas Metzinger, “Scientifically describing N and phenomenally experiencing E are just two different ways of accessing one and the same underlying reality” (2000, 4). “The same underlying reality” is here a neural dynamics. It can

134

give rise to a geometric ideal emerging reality (geometric RC) or to a phenomenally experienced reality (phenomenal RC). If we look carefully at the factorization 1 ¡ 2 ¡ 3 we can conclude: 1. There is no really hard problem for the emergence 1 ¡ 2 of global geometry out of local physics. 2. There is no really hard problem for the content matching 2 £ 3 (morphological isomorphism). 3. But there does exist a hard problem concerning the matching between geometrical space and phenomenal space. But this is not a matching of conceptual contents. It is a matching between pure intuitions and mathematical idealities.

III SOCIAL SCIENCES

In the social sciences, an important aspect of reductionism has to do with methodological individualism, that is, with the reduction of collective properties to interactions between individual ones. Complex methodological individualism advocated by thinkers like Jean-Pierre Dupuy is neither holistic nor eliminativist but emergentist. Friedrich von Hayek paradigmatically represents it. Hayek was one of the first to develop the consequences of the theories of selforganization and spontaneous order in cognitive and social sciences.2 We will present a specific example of selection of social rules and show how it can be modeled in terms of evolutionary game theory. 1. Hayek and the Complexity Problem Hayek always strongly emphasized the specific properties of the complex socioeconomic spontaneous order in modern open societies. It is a sophisticated self-organized

135

order where “laissez faire” produces no anarchy, an order that is cognitively founded and would be impossible to obtain in another way. Its endogenous complexity is irreducible and, according to Hayek, dooms to failure any rationalist constructivism that would claim to create it artificially. We meet here a new type of reductionism, namely the possibility to reduce a natural complex order to the application of a system of rational rules. In a certain sense, Hayek claims that there cannot exist an AI expert system for modern societies and markets. The source of complexity has to be found in the fact that, in an open society, knowledge, competencies, and informations are distributed, scattered over a great number of cognitively limited and interacting agents. The systemic properties of such systems cannot be conceptually controlled. The political control of social and economic orders rests on a methodological error. Many consequences derive from this fundamental fact. 1.

2.

3.

136

First, complexity prohibits at the same time a centralized hierarchical organization and a communal link of reciprocity characteristic of small closed communities. In modern open societies the interactions between agents is no longer ensured by consensus on shared values but by exchange of signals such as prices in a market. The market is a way of circulating information in a multi-agent system whose very complexity makes it opaque to its own agents. In a Hayekian catallaxy, everyone cooperates with everyone else but without any shared ends. The individual aims are incommensurable with each other but mechanisms such as free trade and markets guarantee nevertheless a viable cooperation. Complexity is an evolutionary (“Darwinian”) process resulting from a selection of historico-cultural rules of behavior, practices, and institutions that are impossible to master conceptually. In that sense, political, juridical, and social constructivism appears to be the dark side of the Enlightenment. A third consequence of complexity is that rules that govern social exchanges and communication are abstract and formal. Social self-

organized complex systems are governed by civil rights guaranteed by public laws. The main critique raised by Hayek against political constructivism is that it does not understand what a complex order is and is in fact not “progressive” at all, but “regressive.” 2. Toward a Rational Justification of Hayekian Anti-Constructivism Progressively, these problems have become accessible to scientific inquiry and modeling. In particular, the idea that many common-sense rules have been selected by an evolutionary process and constitute an optimizing collective form of “learning,” seems to be essentially right. We will present an illustration, but let us first recall some aspects of methodological individualism. 2.1. The Paradigms of Social Order The concept of spontaneous order must be put in historical context. It posits that pluralism and individual freedom are not sources of disorder, anarchy, and social struggle but, on the contrary, a factor conducive to higher forms of organization. It stands in sharp contrast with three other paradigms: 1. The paradigm of hierarchical order and absolute power theorized from the Renaissance by Machiavelli (1469–1527), then Bodin (1529–1596) and Hobbes (1588–1679), and put in practice for instance in Spain by Charles V and Philippe II, or in France by Richelieu, Louis XIV, and Napoleon. In reaction to this form of absolutism, many demands arose for tolerance and human rights, from Grotius (1583–1645), Bayle (1647–1706), and Locke (1632–1704) to Kant (1724–1804), Humboldt (1767–1835), and Benjamin Constant (1767–1830). This was the source of many revolutions: in Netherlands, England, America, and France (before the Terror). This was the main origin of modern science techniques, the industrial revolution, and prosperity.

137

2. The revolutionary paradigm of rational constructivist order that rejects open society in the name of great ideals of equality and justice and relies on political planning to create a new humanity. 3. The conservative paradigm of natural order, which also rejects open society, but for an opposite reason: it champions a form of organicist holism and accuses modernity of having “atomized” society (individualism) and destroyed “natural communities,” (family, corporations, churches, etc.) (Nemo 2002). The paradigm of spontaneous order posits a new conception of social order as neither natural (permanent and universal) nor artificial (rationally construed), but pluralist and self-organized, non-hierarchical and polycentric. Evident examples of such orders are language, law, or morals: they are not natural in the strict sense of the term, but neither are they artificial, since nobody has ever made them. As the masters of the Scottish Enlightenment David Hume (1711–1776) and Adam Ferguson (1723–1816) emphasized, they are the results of human actions but not of human intentions. 2.2. Methodological Individualism3 In this context, the central problem of methodological individualism (MI) concerns the reducibility of macrosocial structures to micro-individual interactions of agents. In classical sociology, holistic realism is dominant. According to it, social phenomena must be explained in terms of macrosocial and supra-individual collective entities prior to individual agents and transcending them: states, churches, parties, classes, nations, markets, etc. Holism aims at explaining how such “real” social entities prescribe norms and values to individual subjects. 1.

138

For Saint-Simon (1760–1825, De la physiologie appliquée à l’amélioration des institutions sociales, 1813) and Auguste Comte (1798–1857, Système de politique positive, 1851) holism was a sort of “organicism,” a

2.

3.

“physiological” conception of the social reality opposing “mechanistic atomism” developed by “social physics.” With Durkheim (1859–1917, De la division du travail social, 1893, Règles de la méthode sociologique, 1895; Les formes élémentaires de la vie religieuse, 1912), holism is no longer biologically inspired and becomes a true sociological thesis. Social wholes exist and subsist de re, and determine the actions of empirical individuals. Of course, there exist “horizontal” interactions between individuals but the true social causality is “vertical” and “top down” and flows from social wholes to individual parts. By definition, all variants of socialism and communism are also holistic.

MI considers that holism is a mythology. It rejects any substantial hypostasis of global concepts and develops a modern variant of the fight of nominalism against realist conceptions of universals. For it, as for Occam, social groups are aggregates and not substances. It is called methodological because it concerns explanation and not ontology. For Karl Popper (1944), MI is an unassailable thesis according to which all collective phenomena must be reduced to actions, interactions, goals, hopes, and thoughts of individual subjects, as well as to the traditions they have created and maintained. For Jon Elster (1986), it is the thesis according to which all social phenomena, their structure and change, can be explained using only individuals with their qualities, beliefs, goals, and actions. The founders of MI are well known: 1. 2.

John Locke (1632–1704). Individuals are the basic social entities but they interact in a contractual society protected by the rule of law. Bernard de Mandeville (1670–1733) and his celebrated “The Grumbling Hive: or, Knaves Turn’d Honest” (1705), also known as “The Fable of the Bees; or, Private Vices, Public Benefits” (1714). He triggered a

139

3. 4.

tremendous controversy (for instance with Berkeley) because he introduced a principle of inversion between individual intentions (micro level) and non-intentional emerging social properties (macro level). Individuals are intentionally selfish and governed by their private and local self-interest, but their interactions generate, in a non-intentional way, a global social order propitious to public interest. The Scottish Enlightenment: Hume, Ferguson (see above). Adam Smith (1723–1790) and the “invisible hand” (Theory of Moral Sentiments, 1759; The Wealth of Nations, 1776). The essential feature of the invisible hand is that it drives subjects to collective ends that do not proceed from their intentions.

We see that methodological individualism concerns mechanisms of selforganization, which cannot be rationally computed by agents. Social cohesion, cooperation, and prosperity are non-intentional effects emerging from an aggregation of selfish interests. Many variants of MI proceeded from these early works, some more utilitarian and reductionist (John Stuart Mill, 1806–1873; Léon Walras, 1834–1910; and Vilfredo Pareto, 1848–1923, Traité de sociologie générale, 1916), other more organicist (but not holistic, Herbert Spencer, 1820–1903, The Principles of Sociology, 1882–1898). Complex IM was founded by the Austrian school: Carl Menger, Ludwig von Mises, and Friedrich von Hayek. It rejects of course holist mythology, but it is more emergentist than reductionist and mechanistic. Carl Menger (1840–1921, Grundsätze des Volkwirtschaftlehre, 1871; Untersuchungen über die Methode des Socialwissenschaften, 1883) was the first to make explicit this problem of complexity. He was followed by Hayek who gave the best theoretical clarification of self-organized orders (language, religion, law, money, market, state, etc.) that are not the result of a collective intentional will. The parallel with theories in natural sciences (physics, biology, neurosciences) is striking.

140

1. 2. 3. 4. 5. 6. 7.

Order is a consequence of the coordination, be it competitive or cooperative, of individual agents. Emerging collective structures acquire some autonomy even if they are causally reducible to individual interactions. They are structurally stable if agents respect rules of law. These rules result themselves from a form of cultural evolution. Emerging structures are non-intentional and unpredictable (no rational planning is possible). It is a fundamental error to attribute intentionality to them. That error is one of the main sources of totalitarianism. Cultural evolution is Darwinian (and not Hegelian). 3. Cultural Evolution and Emerging Ethical Maxims

At the cognitive level, be it individual or social, according to Hayek, the origin of the rules governing perception and action, as well as that of conventions and norms, is evolutionary. These patterns result from a cultural selection—a collective learning—which is a competitive / cooperative process having favored the individuals and groups that applied them (Kirman 1998; Nadeau 1998). They are like cultural short cuts, enabling people to behave rapidly and adaptively without having to recapitulate every time all the experiences and beliefs necessary to action. For Hayek, common sense is a library of tacit knowledge routines and practical schemes patterning our experience according to generic default schemes. It is absolutely necessary to act without being overwhelmed by the overflow of irrelevant information coming from the environment. For Hayek (as for Mandeville, Hume, or Ferguson), common-sense norms are not repressive constraints but, on the contrary, cognitive achievements deeply adapted to the contingencies of life. Traditions express an “embodied knowledge” which is “phylogenetic” in the sense of cultural evolution, and it is therefore rational to comply with them “ontogenetically.” In much the same way as in evolutionary biology phylogenetic a posteriori operate as ontogenetic a priori, common-sense rules operate for the subjects as

141

a priori frames. In this sense, we find in Hayek an evolutionary theory of the self-transcendence of behavioral rules. Like linguistic rules, they proceed from symbolic institutions whose origin is neither a rational omniscient intelligence nor a deliberative social contract. We see how Hayek articulates cognitive psychology (the “sensory order”) with the sociology of complex spontaneous orders (“catallaxy”). We know that the very concept of cultural evolution is quite problematic. For Hayek, as for Popper, cultural evolution selects groups and not individuals, subjects having to comply with rules that maximize the collective performances of their group. However, for the subjects themselves, it is impossible to understand in what operational sense these norms are socially fruitful, because they encode a “phylogenetic” historical evolution. That’s why they interpret the norms as duties and values. We must emphasize the originality of this conception: 1.

2.

As individuals cannot understand the pragmatic efficacy of norms, they accept them for deontic reasons. We recognize here a thesis that belongs in Kantian ethic. However, norms being socially useful, we recognize also a utilitarian conception of ethics (Jeremy Bentham, John Stuart Mill). The main difference is that the “computation” of moral maxims and actions is cognitively inaccessible for individuals.

Therefore, according to Hayek, cultural evolution implies that maxims of action can act for individuals as transcendent “categorical” imperatives4 while they are at the same time immanent “hypothetical” (pragmatic) imperatives for cultures.5 For cultures, maxims are caused by the viability of a social order from which individuals gain a lot. As was emphasized by John Gray, Hayekian utilitarianism is indirect and exemplifies the general evolutionary principle (Haeckel’s law) according to which phylogenetic a posteriori operate ontogenetically as a priori (see Barry 1984). Hayek was able to reconcile, from within methodological individualism, reductionism with holism: social entities prescribe norms, rules, and maxims to individuals.

142

It is interesting to highlight how Hayek succeeded in renewing the notion of categorical imperative as a deontological (nonconsequentialist) conception of actions. According to deontological theses, actions must be evaluated in a principled way independently of their consequences, while according to consequentialist theses they must be evaluated on the basis of a computation of the costs and benefits of their consequences. But as that kind of computation is impossible for a finite and limited rational mind, it is performed by cultural evolution. As was emphasized by Jean-Pierre Dupuy, cultural evolution is “utilitarian” but bears on “deontological” maxims that can be interpreted in accordance with a test of “categoricity.” 4. An Example of an Evolutionary Game Let us now give an example, taken from evolutionary game theory, of how social modeling upholds some of Hayek’s theses. Ever since the pioneering work of Robert Axelrod, many models have been dedicated to underlying causal mechanisms of complex adaptive social systems. The simplest and best-known example is that of the Iterated Prisoner’s Dilemma (IPD) (Cohen et al. 1998; Poundstone 1993). 4.1. The Prisoner’s Dilemma Let us first recall the classical Prisoner’s Dilemma. There are two players A and B and each player can choose one of two behaviors (strategies): d = defection and c = cooperation. In order to compute gains and losses (profits and deficits), we use a matrix of payoffs with columns A(c) (A plays c) and A(d), and lines B(c) and B(d). Each entry corresponds therefore to a one-shot game and we introduce the players’ payoffs: the column player A’s in the upper right corner and the line player B’s in the lower left corner. The payoff matrix involves 4 terms: T = (d, c) = Temptation, S = (c, d) = Sucker, R = (c, c) = Reward, P = (d, d) = Punishment.

143

For the game to be interesting (there must exist a “dilemma”), payoffs must satisfy the set of inequalities: T > R > P > S and (T + S)/2 < R. Here is a typical example: A(c) B(c)

R=3

A(d)

Behaviors: d = defection, c = cooperation

T=5 Payoffs: T = (d, c) = 5, S = (c, d) = 0 R = (c, c) = 3, P = (d, d) = 1

B(d)

R=3 S=0

S=0 P=1

T=5

P=1

Conditions: T=5>R=3>P=1>S=0 (T + S)/2 = 5/2 < R = 3

This extremely simple game is not trivial, since it represents a situation where individual rationality is at odds with collective rationality. Indeed: 1.

2.

3.

4.

144

If column player A plays c, then line player B gets R if he plays c and T if he plays d. As T = 5 > R = 3, it’s in the interest of B to play d. Now, if column player A plays d, then line player B gets S if he plays c and P if he plays d. As P = 1 > S = 0, it’s still in the interest of B to play d. Therefore, if B is rational in the individualist sense, he must play d whatever A’s behavior. It is said that strategy d strictly dominates strategy c : d is better than c whatever the other player’s behavior. The same holds for A by symmetry.

5.

6. 7.

The rational outcome of the game is therefore the noncooperative behavior (d, d), which leads to the very bad collective payoff (P = 1, P = 1). But clearly, the cooperative behavior (c, c) would have led to a largely better collective payoff (R = 3, R = 3). So individual rationality selects a poor (lose, lose) strategy (d, d) while a collective rationality would have selected a good (win, win) strategy (c, c).

The dilemma comes from the fact that for the above payoff matrix the double strategy (d, d) is the only Nash equilibrium (NE), that is, the only strategy having the property that players would do worse if they unilaterally change their strategy. One can generalize this basic example in multiple ways, introducing asymmetries, non-strict inequalities, neutral behaviors (a player can refuse to play), multiple players, probabilistic strategies (a player plays c with probability p and d with probability 1 – p, etc.). The main result is that the dilemma is robust. How can we therefore explain the emergence of cooperative collective behaviors through an evolutionary selective process? It is clearly a fundamental problem. Remark: The condition (T + S)/2 < R is introduced to make the payoff quadrangle (c, d) – (c, c) – (d, c) – (d, d) – (c, d) convex. It is important for mixed strategies where A plays c with probability p and B plays c with probability q. The payoff expectation of B is (1 – p)qT+pqR+(1 – p)(1 – q)P+p(1 – q)S which is then inside the quadrangle. 4.2. The Iterated Prisoner’s Dilemma (IPD) The situation changes completely when one iterates the game, because defection can then be punished and cooperation rewarded. We can in that case introduce genuine strategies. We must suppose that the number of moves is indeterminate to

145

avoid backward induction (the possibility of defining a strategy by going backwards from the desired result to the initial move), which has the property of leading us back to the noncooperative behavior (d, d) (double defection). We can test strategies such as G = “good” (sucker) = play always c (unconditional cooperation); M = “meanie” (bad) = play always d (unconditional defection); TFT = Tit for Tat = start with c (initial cooperation), then play what the other player played at the previous move; V = “vindictive” = start with c and play d for the rest of the game as soon as the other player plays d (that is, defection is punished as an irreversible betrayal), etc. One pits these strategies against one another over a great number of plays (for instance 1000) and one compares their scores. The notion of a Nash equilibrium (NE) must be revised since the strategy (Id, Id) that iterates the oneshot NE (d, d) remains a NE. But many other strategies yield the same result as Id when playing against Id, and there exist too many NE’s. Hence the concept of “subgame perfect equilibrium,” which is a NE for every sub-game of the game. Among pools of strategies that are not too complex, one finds that the strategy Tit for Tat (TFT) has a striking superiority; TFT does not win every time, but it always gets a very good score. More generally, computer simulations show that the best strategies are nicely cooperative, rapidly reacting to defections (“retaliatory”), rapidly forgiving, and simple (“clear,” without wiles). Figure 3.11 by J.-P. Delahaye shows the result of the tournament between 12 strategies, each represented by 100 agents. The global population comprises therefore 1200 agents. We see that after a stabilization period (which needs about twenty generations; see the following section), all the noncooperative strategies have been eliminated and the winning strategy is TFT.6 4.3. Evolutionary Games Evolutionary game theory considers polymorphic populations of individuals using different strategies and defines new generations using the scores in a generalized competition: strategies with good scores increase their number of representatives while those with bad scores progressively vanish (Binmore 1994; Hofbauer and

146

Figure 3.11. The result of the tournament between 12 strategies, each represented by 100 agents (from Delahaye and Mathieu 1999).

Sigmund 1988; Samuelson 1997; Weibull 1996). Evolutionary theory is more realist than the classical one based on individual rationality. It substitutes a collective selective scheme to an impossible variational calculus. Moreover, it enables us to understand the dynamics that drive agents towards global equilibria. Let {si} be the set of strategies and {pi} their respective probabilities (i.e., the proportions of the population playing them). We suppose that the size N of the population remains constant. Consider the case where there are only two strategies: c with probability = p and d with probability = 1 – p). It is easy to compute the expectation of gains (utilities Uc(p) and Ud(p)) for each strategy as a function of the parameter p. Recall that T = (d, c), S = (c, d), R = (c, c), and P = (d, d). If an agent plays c, the probability that he will play against another agent playing c is p and he will gain (c, c) = R, while the probability that he will play against another agent playing d is 1 – p and he will gain (c, d) = S. If the agent plays d, the probability that he will play against another agent playing c is p and he will gain (d, c) = T, while the probability that he will play against another agent playing d is 1 – p and he will gain (d, d) = P. We get therefore:

147

Uc(p) = pR+(1 – p)S Ud(p) = pT+(1 – p)P

{

The mean gain of the population is therefore given by the quadratic expression: U(p) = pUc(p) + (1 – p)Ud(p) = p2R + p(1 – p)S + (1 – p)pT + (1 – p)2P that is: U(p) = p2R + p(1 – p)(S + T) + (1 – p)2P The evolution of the probability p is given by the replication dynamic p' = p(Uc(p) – U(p)) 4.4. The Tit For Tat Strategy: From Common Sense to Dynamical Models In these models, agents are considered as “phenotypes” expressing “genotypes” identified with strategies, and “micro” strategies influence “macro” population dynamics. Simulations (which specialists of complex systems call “computational synthesis”) provide extremely interesting results. Axelrod has shown that: 1. 2.

3.

148

Anti-cooperative strategies are eliminated, cooperation wins and becomes stable. TFT dominates, but is fragile with respect to mutations; indeed sucker mutants Ic exhibit exactly the same behavior as TFT in a TFT environment; they can therefore substitute themselves progressively and “silently” for TFT, without any observable effect; but then “bad” mutants Id (meanies) can destabilize, invade, and destroy the system. For a strategy to be collectively stable, it must react to defections (be retaliatory), that is, resist destabilizations by “bad” mutants.

4.

5.

If one introduces complex strategies, many subtle phenomena can occur. For instance, a noncooperative strategy can use another one to eliminate cooperative strategies and eliminate its allies in a second step; social disorders enable some noncooperative strategies to survive and even win the game, etc. Simulations show that there exist sophisticated refinements of TFT, which improve its results slightly in more complex contexts. But we can say that TFT is the most efficacious simple strategy.

Now, it is also an empirical anthropological and cultural fact that since ancient times TFT has been selected by common sense. We meet here a typical model of common sense: 1. 2.

Simulations corroborate an old common-sense rule proceeding from collective political and social knowledge. But at the same time, they enable us to overcome naïve common sense and to develop an experimental framework for virtual (modeled) cultural evolutions.

4.5. Sigmund’s and Nowak’s Generalizations Many authors have studied factors that facilitate cooperation in the IPD when one changes the space of strategies, the interaction process, the adaptive responses, etc. We will say a few words on the introduction of “topological” relations of neighborhood between agents, each agent becoming able to imitate the one of his neighbors that makes the best score. Consider, for instance, extremely simple strategies (i, p, q) where: i = initial probability of cooperation, p = probability of cooperation after a cooperative move by the other player, q = probability of cooperation after a defective move by the other player.

149

We have trivially Ic = (1,1,1), Id = (0,0,0), TFT = (1,1,0), cp = (p, p, p) (always c with probability p). Other interesting strategies are GTFT = T–R R–P (1,1,Min 1 – ——— , ——— ) and Kraines’ “Pavlovian” strategy that resists R–S T–P Id well: play c after R or T, and d after P or S. Sigmund and Nowak have shown that “bad” agents Id (meanies) can win at the beginning of the game. However, TFT agents resist. Once the “good” Ic (suckers) have been decimated, the exploiters can no longer abuse them and cooperative strategies of TFT type emerge. But after this emergence of cooperation, the TFT strategies are themselves overtaken by GTFT. However, the GTFT strategy is fragile and allows for the return of “bad” Id.

(

)

4.6. Spatialized IPD In spatial IPDs, there exists a “topology,” each agent having a few neighbors with whom he interacts. For a neighborhood of 4 neighbors renewed at each move and for an evolution process that consists of imitating the neighbor who has the best score, Axelrod get the following results: for initial probabilities (i, p, q) = (0.5, 0.5, 0.5) randomly distributed, suckers S are eliminated by meanies M, whose strategy dominates. In such a system, cooperation cannot emerge. But if neighbors are fixed, then defective strategies can no longer invade the system. Once again TFT strategies dominate because if two TFT agents appear by mutation and meet, they are immediately imitated and their strategy propagates until it has invaded the system. For instance, a sucker S with three TFT neighbors and a meanie neighbor M generated by a mutation is eliminated by this M at a first stage, and M wins. However, at a second stage the M agents must interact and have only TFT neighbors. Then TFT agents win, and the meanies M convert to TFT. In other words, fluctuations generating M agents are recessive. This mechanism explains the strong stability properties of evolutionarily stable strategies such as TFT that cannot be destabilized by mutating invaders.

150

Let us give Nowak and May’s example of systems defined on a square network with 8 neighbors by the payoff matrix:

A(c) B(c)

B(d)

R=1

R=1 S=0

T=b

A(d)

Behaviors: d = defection, c = cooperation

T=b

S=0 P=0

P=0

Payoffs: T = (d, c) = Temptation, S = (c, d ) = Sucker, R = (c, c) = Reward, P = (d, d) = Punishment Conditions: T=b>R=1>P=0=S=0

b is the parameter of the system. Take, for instance, a random initial configuration with 50% c and 50% d. One compares the scores (the score of each site being the sum of its gain and of the gains of its 8 neighbors) and each site adopts the strategy of its neighbor (including itself ) that gets the best result. The conclusion (Zhen Cao and Hwa 1999) is extremely interesting, and a priori unexpected if one is not familiar with critical phenomena in physics and, more generally, with bifurcation processes. One gets: 1. for b < 1.8, c dominates; 2. for b > 2, d dominates; 3. for b belonging to the interval Bc = [1.8, 2]—called the critical interval—there exists a critical transition c ¡ d, with multi-scale nested clusters of c and d.

151

Figure 3.12. Nowak and May’s example of Tit for Tat strategy, displayed spatially. Moves are coded with colors: c then c = blue; d then d = red; c then d = yellow; d then c = green. For b = 1.5 and an initial configuration “InitConfig” 50%–50%, there is an initial catastrophe (in two steps) leading to an overwhelming domination of “meanies” (red). Then cooperation (c, c) restores and dominates progressively, through the extension of residual scattered nuclei having survived the catastrophic initial phase of decimation.

Figure 3.12 is a Mathematica™ implementation I computed using an algorithm by Richard Gaylord and Kazume Nishidate. We code moves with colors: c then c = blue; d then d = red; c then d = yellow; d then c = green. For b = 1.5 (under the critical interval) and an initial configuration “InitConfig” 50%–50%, we see that there is an initial catastrophe (in two steps) leading to an overwhelming domination of meanies (red). Then cooperation (c, c) (blue) restores and dominates progressively, through the extension of residual scattered nuclei having survived the catastrophic initial phase of decimation. Domination of cooperation is, by the way, noncomplete and leaves fracture lines of oscillating noncooperation (d, d). If we represent the temporal evolution of the subpopulations (c, c) and (d, d) we see very distinctly the initial decimation followed by a reconquest presenting small oscillating fluctuations (figure 3.13).

152

Figure 3.13. The temporal evolution of the subpopulations (c, c) and (d, d) of figure 3.12.

For b = 2.1 (above the critical interval) and a 50%–50% InitConfig, we see that d dominates immediately and totally (red totalitarianism, figure 3.14).

Figure 3.14. For b = 2.1 and a 50%–50% InitConfig, defection d dominates immediately and totally.

153

The curves of evolution are evident (figure 3.15, no comment):

Figure 3.15. The temporal evolution of the subpopulations (c, c) and (d, d) of figure 3.14.

But if the value of the parameter b = 1.85 belongs to the critical interval and if we take a 50%–50% InitConfig, we see that (d, d) begins to dominate, next that (c, c) begins to reconquer ground by expanding from nuclei that have resisted the initial extermination, but that, contrary to the first example b = 1.5, multi-scale nested clusters of c and d appear and expand: inside a blue island expanding in a red sea appears an expanding red lagoon, inside which emerges a smaller blue island, etc. (figure 3.16). This critical dynamics is very apparent on the evolution curves where the curves (c, c) and (d, d) present, besides small oscillating fluctuations, large-scale oscillations (figure 3.17). The InitConfig plays, of course, a major role in the evolution of the system. Figure 3.18 (next page) shows an example for b = 1.85 (inside the critical interval) and an InitConfig reduced to a single (d, d) in a purely (c, c) population. We see that this system lacks reactivity, so the noncooperative state proliferates and progressively invades the system. 154

Figure 3.16. For b = 1.85 in the critical interval and a 50%–50% InitConfig, the behavior (d, d) begins to dominate; next (c, c) begins to reconquer ground by expanding from nuclei that resisted the initial extermination, but multi-scale nested clusters of c and d appear and expand in a fractal structure.

Figure 3.17. The temporal evolution of the subpopulations (c, c) and (d, d) of figure 3.16. 155

Figure 3.18. Evolution of the system for b = 1.85 (inside the critical interval) and an InitConfig reduced to a single (d, d) in a purely (c, c) population.

IV CONCLUSION

We have seen in simple examples of evolutionary games how cooperation can spontaneously emerge as collective behavior in a population of individuals. We have retrieved very well-known common-sense rules such as the evolutionarily stable Tit for Tat strategy, and we have modeled the evolutionary stability, explaining how such a rule can have been selected by cultural evolution. There exist today an extremely impressive number of such computational syntheses that provide us with an experimental method for what we could call “Hayekian science,” that is, science based on methodological individualism and

156

causal reductionism, which investigates the natural selection of behavior rules, and explains how collective structures can emerge through self-organizing dynamics.

NOTES

1. See for instance Noë’s and Thompson’s target paper (2004) against Chalmers’ work. 2. See Hayek’s references in the bibliography. 3. See Laurent (1994). 4. For Kant, a normative judgement is “categorical” when it is independent of any end. Categorical prescriptions are purely “procedural.” 5. For Kant, a normative judgement is “hypothetical” when it is conditioned by an end and prescribes means to achieve the end (consequentialism). 6. “Donnant-donnant” = TFT; “Majorité-mou” = play as the majority and play c if there is an ambiguity 50%-50%; “rancunière” = “vindictive” = start with c but always play d as soon as the other player plays d; “gentille” = “good” = always play c; “méfiante” = “suspicious” = TFT with d as starting move; “méchante” = “meanie” = always play d.

REFERENCES

Barry, Norman. 1984. Hayek’s ‘Serfdom’ Revisited: Essays by Economists, Philosophers and Political Scientists on ‘The Road to Serfdom’ After 40 Years. London: Institute of Economic Affairs; Albuquerque NM: Transatlantic Arts. Binmore, K. G. 1994. Playing Fair: Game Theory and the Social Contract. Cambridge MA: MIT Press. Bosking, W., Zhang, Y., Schofield, B. and Fitzpatrick, D. 1997. “Orientation Selectivity and the Arrangement of Horizontal Connections in Tree Shrew Striate Cortex.” Journal of Neuroscience 17 (6): 2112–27. Bressloff, P. and Cowan, J. 2003. “The Functional Geometry of Local and Horizontal Connections in a Model of V1.” Pp. 221–36 in Neurogeometry and Visual Perception, ed. J. Petitot and J. Lorenceau. Journal of Physiology (Paris) 97 (2–3). Bressloff, P., Cowan, J., Golubitsky, M., Thomas, P. and Wiener, M. 2001. “Geometric Visual Hallucinations, Euclidean Symmetry and the Functional Architecture of Striate Cortex.” Philosophical Transactions of the Royal Society of London. B. 356: 299–330.

157

Cao, Z. and Hwa, R. 1999. “Phase Transition in Evolutionary Games.” International Journal of Modern Physics A 14 (10): 1551–59. Chalmers, D. 2000. “What is a Neural Correlate of Consciousness?” Pp. 17–39 in Neural Correlates of Consciousness: Empirical and Conceptual Questions, ed. T. Metzinger. Cambridge MA: MIT Press. Cohen, M., Riolo, R. and Axelrod, R. 1998. “The Emergence of Social Organization in the Prisoner’s Dilemma: How Context-Preservation and Other Factors Promote Cooperation.” Santa Fe Institute Working Paper 99–01–002. Das, A., and Gilbert, C. D. 1995. “Long Range Horizontal Connections and Their Role in Cortical Reorganization Revealed by Optical Recording of Cat Primary Visual Cortex.” Nature 375: 780–84. DeAngelis, G. C, Ohzawa, I. and Freeman, R. D. 1995. “Receptive Field Dynamics in the Central Visual Pathways.” Trends in Neuroscience 18: 451–58. Delahaye, J.-P. and Mathieu, P. 1999. “Des surprises dans le monde de la coopération.” Les Mathématiques sociales, Dossier Pour la Science 58–66. Dupuy, C. and Torre, A. 1999. “The Morphogenesis of Spatialized Cooperation Relations.” European Journal of Economic and Social Systems 13 (1): 59–70. Dupuy, J.-P. 1992. Le sacrifice et l’envie. Paris: Calmann-Lévy. ———. 1999. “Rationalité et irrationalité des choix individuels.” Les Mathématiques sociales, Dossier Pour la Science 68–73. Elster, J. 1986. An Introduction to Karl Marx. Cambridge: Cambridge University Press. Ermentrout, B. and Cowan, J. 1979. “A Mathematical Theory of Visual Hallucinations.” Biological Cybernetics 34: 137–50. Field, D. J., Hayes, A. and Hess, R. F. 1993. “Contour Integration by the Human Visual System: Evidence for a Local ‘Association Field.’ ” Vision Research 33 (2): 173–93. Frégnac, Y. 2003. “Introduction” in Neurogeometry and Visual Perception, ed. J. Petitot and J. Lorenceau. Journal of Physiology (Paris) 97 (2–3). Hayek, F. 1952. The Sensory Order: An Inquiry into the Foundations of Theoretical Psychology. Chicago: University of Chicago Press. ———. 1982. Law, Legislation and Liberty. London: Routledge & Kegan Paul. ———. 1988. The Fatal Conceit. London and New York: Routledge. Hofbauer, J., and Sigmund, K. 1988. The Theory of Evolution and Dynamical Systems. Cambridge: Cambridge University Press.

158

Kanizsa, G. 1980. Grammatica del Vedere: saggi su percezione e gestalt. Bologna: Il Mulino. English ed. 1979. Organization in Vision: Essays on Gestalt Perception. New York: Praeger. Kirman, A. 1998. “La pensée évolutionniste dans la théorie économique néoclassique.” Philosophiques XXV (2): 219–37. Laurent, A. 1994. L’individualisme méthodologique, Que sais-je? n° 2906. Paris: Presses Universitaires de France. Metzinger, T., ed. 2000. Neural Correlates of Consciousness: Empirical and Conceptual Questions. Cambridge MA: MIT Press. Nadeau, R. 1998. “L’évolutionnisme économique de Friedrich Hayek.” Philosophiques XXV (2): 257–79. Nemo, P. 2002. Histoire des idées politiques aux Temps modernes et contemporains. Paris: Presses Universitaires de France. Noë, A. and Thompson, E. 2004. “Are there Neural Correlates of Consciousness?” Journal of Consciousness Studies 11: 1. Petitot, J. 1999. “Morphological Eidetics for Phenomenology of Perception.” Pp. 330–71 in Naturalizing Phenomenology: Issues in Contemporary Phenomenology and Cognitive Science, ed. J. Petitot, F. J. Varela, J.-M. Roy and B. Pachoud. Stanford CA: Stanford University Press. ———. 2000. “Vers des Lumières hayekiennes : de la critique du rationalisme constructiviste à un nouveau rationalisme critique.” Friedrich Hayek et la philosophie économique. Après le Colloque de Cerisy, ed. A. Leroux and R. Nadeau. Revue de Philosophie économique 2: 9–46. ———. 2003. “The Neurogeometry of Pinwheels as a Sub-Riemannian Contact Structure.” Pp. 265–309 in Neurogeometry and Visual Perception, ed. J. Petitot and J. Lorenceau. Journal of Physiology (Paris) 97 (2–3). Popper, K. 1944. The Poverty of Historicism. London: Economica. Poundstone, W. 1993. The Prisoner’s Dilemma. Oxford: Oxford University Press. Ray, T. 2004. “The Chemical Architecture of the Human Mind: Probing Receptor Space with Psychedelics” in Toward a Science of Consciousness VI, conference, Tucson, April 7–11, 2004. http://www.conferencerecording.com/newevents/tsc24.htm. Samuelson, L. 1997. Evolutionary Games and Equilibrium Selection. Cambridge MA: MIT Press.

159

Shmuel, A. and Grinvald, A. 2000. “Coexistence of Linear Zones and Pinwheels within Orientation Maps in Cat Visual Cortex.” Proceedings of the National Academy of Sciences USA 97 (10): 5568–73. Spencer, H. 1882–1898. The Principles of Sociology. London: Williams and Norgate. Thom, R. 1972. Stabilité structurelle et morphogénèse. New York: Benjamin; Paris: Édiscience. Weibull, J. 1996. Evolutionary Game Theory. Cambridge MA: MIT Press.

DISCUSSION

J E AN-PIERRE D UP U Y:

You [Jean Petitot] said physics is mathematical modeling

of natural phenomena. J E AN PETITOT:

No, of physical pattern.

J E AN-PIERRE D UP U Y: No, you seemed to say that the reduction is not physical because it is mathematical. J E AN PETITOT: No, it is not the reduction which is mathematical. We have the Hopfield equation and this is a sort of physicalist reduction. But mathematical deduction is upward. J E AN-PIERRE D UP U Y:

Ah, it is an upward reduction.

An upward emegence! So it is not an eliminativist reductionism. It is not an epiphenomenalism, it is not a dualism, it is not an ontological emergentism. It is a causal reductionism and a formal emergentism, a formal geometrical emergentism. It is not a functionalism. It is not a representationism. So it is something like dynamical and geometrical emergentism, a very good example of that. In the second part of the paper, I discuss complexity in the perspective of complex methodological individualism as in the style of Hayek. First I will give some brief references to the history of the idea and after I will give an example of how we can construct some mathematical models of common-sense rules. We have

J E AN PETITOT:

160

common-sense rules that can be selected by a cultural evolutionary process, which constitutes a tacit knowledge, an embodied knowledge, practical skills, etc. Let us take the very well-known example of the spatialized iterated prisoner’s dilemma, as in the works of Axelrod, Sigmund, Nowak, May, and many others. We have here, in the 3-D plot, populations playing with some strategies. We have a temporal evolution, and along the vertical axis we have the efficacy of the strategy. You see that the defective strategies lose very rapidly and that Tit for Tat wins. It wins because it is a retaliatory cooperation. The general rule is “we cooperate,” but if there is a defection we respond by defection. So in this model, agents are considered as phenotypes expressing genotypes (strategies), and we look at the population dynamic in a sort of Darwinian way. The agents playing with good strategies reproduce themselves, and so it is the bad strategies that are eliminated. Simulations show that anti-cooperative strategies are eliminated and that cooperation wins and becomes stable. But, and it is a very interesting point, the Tit for Tat strategy is fragile with respect to mutations. Why? Because in a Tit for Tat society, purely cooperative, nonretaliatory mutants exhibit exactly the same behavior as Tit for Tat, since we have only cooperation. All Tit for Tat can be replaced by pure cooperators. But in that case, a bad mutation and the occurrence of one defector completely destabilizes the system. So we do not have structural stability. We have general theorems that show the necessity of a strategy to react to defections, to be retaliatory, as a condition for being collectively stable. Let me talk about the prisoner’s dilemma. I have never been impressed with the prisoner’s dilemma example because it has no relation whatsoever with what happens in the real world. It is the essence of reductionism. We have a model here in which we have various actors. And we make two simplifying assumptions which are the essence of unreality. One is that the actors are all somehow equal, they all make choices, and the ability to make choices is exactly equal in the situation. Secondly, we assume that they all are interested in maximizing gain in this particular situation. But in the real world some people may have religious values that tell them to be cooperative no matter what the consequences are. They are not interested in maximizing their situation IMMA N UEL WA LLE R ST E IN:

161

in this, but simply want to retain the validity of the importance to themselves of this religious value. Then we have actors who are more powerful than other actors and they may use pressures, such as propaganda or arm twisting, on other actors to have them act in ways that are not in their own maximal interest. So that from the real-world complex situation, we have left out most of the reality, and then we have this different pattern, which over time leads in the ways that you have suggested. But that is what reduction is. It eliminates the reality of the world, a reality that is much more complex in making actual decisions than in a prisoner’s dilemma. There may be prisoner’s dilemmas in the real world but people in prisoner’s dilemmas do not act in the ways you suggest; well, some of them may, perhaps. But this is the typical model that a large percentage of economists will use to analyze how people act in the market, how people act in political economy, but it is not reality. It leaves out the most important elements of how people act. If you leave out all these important elements then they act in this way, which is true. But it does not help us to explain what actually happened in real situations. That is exactly what we call reductionism. The prisoner’s dilemma model is a reductionist model that takes us far from reality. Perhaps, it is the main epistemological point. It may seem that models are crude local approximations of small parts of the real world. But in fact they can be deductively enlarged and reveal very deep relations between different parts of the real world. The alternative to models are concepts. But it is an illusion to believe that concepts are better than models. With concepts, we can grasp some aspects of the world but we cannot go inside its complexity. To go inside we need mathematics and mathematics “derealize” the real world.

J E AN PETITOT:

I do not agree with Immanuel [Wallerstein] for two reasons. First, he confuses reduction with abstraction. Secondly, it is impossible to do science without abstraction. J E AN-PIERRE D UP U Y:

I think this discussion reveals the meaning, or at least the use of reductionism, which was present but which was not explicit. What is an PAUL DUMOUCHEL:

162

appropriate description? In other words, we say this is reductive when we say this description of the phenomenon is not appropriate. The question is: what are good criteria of what is an appropriate description? We all have different intuitions about this. We have some disciplines with scientific requirements. My personal opinion is that we should be very ecumenical as to what is an appropriate description. I think what we want is to have a description that allows us to be able to recover the type of basic empirical information we have. This is what is so striking in the first example in Jean [Petitot]’s paper. We have a description that is the phenomenological description of the consciousness which appears as fairly acceptable to everybody. Nobody questions the value of that description. Then using a completely abstract model like the Hopfield equations, which does not look like the brain in any way whatsoever, we make this turn and in the end we are really happy and we think that our choice of level of description must have been good. I personally believe that our sole criterion of what is a good choice of level of description should be whether in the end we retrieve something that is interesting. And it is true in social sciences because of the prevalence of everyday language and the idea that we order ourselves in the world using this everyday language. As soon as the descriptions of phenomena in social sciences become a little abstract, become a little idealized in relationship to that everyday language, we have a spontaneous tendency to say that this is reductive at the level of the description and this is not satisfying. I think this is the claim which is involved in the second example, that in the end we will use this very abstract, not very interesting description and we will retrieve something of the equivalent of a rule that is evident. Now this is the question: biologist Peter Hamstogen, in a book called Genetic and Cultural Evolution of Cooperation, says that actually recipropal altruism (Tit for Tat is recipropal altruism) just about does not exist in this world. In other words, since the work of Hamilton and Rivers 30 years ago, biologists have been looking everywhere to find good examples of recipropal altruism, but they cannot find them. Axelrod knew very well that cooperation does not exist in nature nor in society because the rationality of agents plays against cooperation. If you want

JEAN PETITOT:

163

cooperation to emerge you must enforce appropriated strategies. Cooperation is not a natural behavior but you can try to formulate general conditions for the emergence of cooperation. The model I presented is not a model of natural behavior. It is a normative model describing the difficulty of the problem, which is not the same thing. It is not the model of a sort of untypical data. It is quite a normative model. I would like to say something about the popular use of description. What we generally think of as appropriate descriptions also seem appropriate to traditional kinds of analytic outcomes; different descriptions have different consequences. For instance, I make a description of a certain phenomenon, then I do the social science. That is, I apply the mathematics; I do the operation and suddenly I have a proposal, a law, that says we will reduce welfare payments and then people who are on welfare will get jobs. That was a description of a certain phenomenon, which then had a certain kind of outcome. It seems to me a lot of our discussion about reductionism is really about reductionism versus holism. In social science, when we talk about what is the opposite of the reductive description, we mean the type of relational description where we do not simply describe the object of analysis according to its characteristics, but according to the ways it is constituted or constructed within a certain totality. That then leads me to a whole different way of conceiving of the way that I am going to talk about my object and the way I want to conceive of my own role in social science, and in the possible outcomes—at least in the social sciences. And I am a little wary of the idea that there is any appropriate description or that we can talk about appropriate descriptions without specifying for whom.

R ICHA RD E. LEE:

EVA N TH OMP SON : On the question of successful reduction or a failed reduction, the fact that we should not accept a straightforward identity between phenomenological space and geometrical space suggests what is a transcendental problem. It is a problem about phenomenality, about subjectivity, it is a well-recognized problem in philosophy from Kant but especially from Husserl, and I think it points to the need within the science of the mind itself for a cognitive science that is irreducibly

164

phenomenological. That has an irreducible, irreplaceable phenomenological level of analysis that brings out and thematizes these transcendental questions, not in the way it can relate them to questions about brain activity, but to link them productively to questions about the brain and the organism. That is a different kind of science because being reductionistic, going back to the point about Galileo, the whole Galilean project, as Husserl analyzes it, is in a crisis because it arose from the removal of subjectivity from the conceptualization of nature. And now when we come back to this line of analysis, we find we are returning to subjectivity and we cannot remove it. We need to learn to think about what that kind of science is trying to be. I propose it is indeed something revolutionary. AV IV BERGMA N : Do you think this is a result of miscommunication, an inability to understand one another in language?

No, I do not think it is a result of miscommunication. I think it is a result of pursuing a certain kind of analysis that is premised on the removal of mind, of subjectivity, which is reductionism. But follow me out: the consequences are that from nature conceived as exterior objective we come back to the subject and we find we cannot remove subjectivity anymore.

EVA N THOMPSON :

AVIV BERGMAN: Yes,

but this is precisely the issue that was raised here when we were talking about the representation to the V1. The fact that we have such a beautiful analogy between the external world and V1 is not coincidence. It is a result of the exposure of building a different mental state. The language which we process is not shaped by an artificial or social organization but rather by an imposition of nature upon our machinery. What we are talking about here is of leaving mathematics as well; we have generated “an artificial language.” It may work for some things. It may not work for other things. But when we are talking about consciousness we might not be able to talk about it using these artificial languages; we might need a different mathematics, we might need a different representation. But my point may be that we may not be able to ever get there because we cannot communicate

165

using the same machinery that we have, which is the result of the languages by which we process our interactions, our experiences with the world. EVA N THOMPSON : If I could say just one thing. There is also the deep question about the relationship between mathematics and the mind, which the paper does not touch upon. It does not discuss the emerging morphologies that are geometrical ideologies or the fact that there is no specific ontology of mathematical ideology. But if mathematics is not this independent ideal realm, what is it? Well, it is some highly invariant transcendental aspect of the mind. So in pursuit of our mathematics, which is how we understand nature, we are led back again to the conceptualizing, modeling mind. The mind then is irreducible; it is un-go-behindable. And the recognition of that, of which you know arguably we have already seen one case, I’m talking about the quantum domain. But now we are seeing it in the case of trying to understand our own subjectivity as biological organisms, and we cannot get behind that. So that is a kind of irreducibility that is really quite radical.

I just want to make sure what Evan [Thompson] means with this phenomenological cognitive science. We talk about the irreducibility there. Is that an acknowledgement of the metaphysics that we cannot escape? Because it seems that if we reproduce the subject in that way, we are back to the idea of the transcendental ego, which is the only point of reference that is irreducible and therefore metaphysical. That sounds like a sound phenomenological position to take. But to what extent does that lead to any practical cognitive science? How does that transcendental metaphysical position translate into something we can use?

PAUL CILLI ERS:

For me, there is not an entailment from the transcendental to the metaphysical in the sense that I think you mean. That is, it does not go back to a transcendental ego, conceived in a strong metaphysical way, some kind of disembodied watching spectator subjectivity. It does not mean that to me at all. The transcendental means the conditions of possibility by which any investigatable world is disclosed or shows up or is manifest, and that always refers us back to the situated inquiring subjectivity of the scientist or the mathematician.

EVA N THOMPSON :

166

SESSION IV

Organizers’ Opening Remarks Immanuel Wallerstein

I

absolutely agree that the various divisions within the social sciences, the socalled disciplines, no longer have an intellectual meaning and that intellectually we should think of them as part of a single unified domain that I call historical social science. Nonetheless, there are certain techniques that are associated with the historical disciplines and I shall pretend first to be an ethnographer, and then pretend to be a traditional historian. As an ethnographer, that is as somebody coming from the outside to a strange tribe and trying to figure out what is going on, I notice two things. First, in reading scholarly papers, I usually start by reading the bibliography. I find bibliographies are very revealing. Reading the bibliographies of these three papers, I notice that not a single item is cited on all three bibliographies. There are some authors who are replicated, but the items are different. This seems to me to validate the premise of this whole set of conferences, that in the three superdomains of social sciences, humanities, and the natural sciences, people read different things, use different languages; and even though they may be talking about the same thing, they do not read what the others say. The second thing I notice as an ethnographer is the tonalities of the three papers in regard to the subject matter. The first paper by Andrew Sayer on reductionism is rather negative about reductionism. In fact in the beginning half, he says that we all reject it these days in social science. Of course he shows that we verbally reject it, but in fact many of us still practice it. But nonetheless there is an intellectual rejection of this concept. I think this is largely shared by people in literary studies.

167

Jean Petitot’s paper is an attempt to salvage reductionism. He says that there have been simplified versions of reductionism in the sciences and they are not very good, but here is a much more sophisticated version of reductionism that is kind of legitimate. And Evan Thompson plays what I think of as the traditional role of the philosopher, which is to mediate among multiple forms of knowledge. He says basically that it is a non-problem because it is a dualism that is not real; that we are all reductionists and emergentists at the same time; it is co-evolution, etc. When I think back just fifty years to when I was a graduate student, those were not the tonalities, certainly not in a degree program in sociology at Columbia University, and we did not say that reductionism is bad. We basically were taught that reductionism is very good, but we had not yet figured out as sociologists how to do it as well as the physicists, but we would, we would. Which is of course what Prof. Petitot is telling us: someday we social scientists will finally figure out how to do it right. Now if I can change my hat to that of a historian of epistemological thought, it seems to me that there is an important shift. If we go back to the middle of the eighteenth century, the concept of the two cultures was invented in the so-called divorce between science and philosophy (by philosophy, what is really meant is the humanities). The split becomes institutionalized in the nineteenth century. The story of the relationship of science on the one hand and the humanities on the other and social sciences in the middle is the growing social and therefore intellectual dominance of science. Science triumphs, reaching an absolute pinnacle in the years after the Second World War when everybody says science is right. The rest of us are wrong or at least different. And the scientists certainly felt they were right. The social scientists were split between those who stuck with the humanities and were empathetic and hermeneutic and those who were very scientistic—the economists being at the head of the list of the super-scientistic. Then there begins to be a social reaction in the late 1960’s. I can explain it in terms of the shifting realities of the world-system, but I am not going to spend energy on that now. I observe it. There is the growth of complexity studies within the sciences that challenges Newtonian principles in ways that everybody here knows. Cultural

168

studies develop in the humanities that challenge the traditional dominant ideas. And social scientists, who drag their feet, came after and they picked up on this. Challenges came from across the superdisciplines. And that is of course a premise of these conferences. All these challenges are occurring in all these fields and in slightly different languages. As I said yesterday, I am a strong believer in the social construction of these entities, these boundaries, these disciplines, these wholes and parts. First of the all, the great superdisciplines were created. I like to cite the fact that in the late eighteenth century in the University of Konigsberg, Emmanuel Kant was a professor—I see a smile already—and Emmanuel Kant taught courses which we today would define as astronomy, poetry, international relations, political theory. And he did not think he was being multi-disciplinary: He thought he was teaching knowledge. There was no distinction in his mind among these fields. But nobody could do that today. Or if someone would do it, we would call him a renaissance person, a really unusual type who is transgressing boundaries. I end my attempt to be an ethnographer and a historian. Now I want to talk about science and reductionism. At the very beginning of the workshop, one of our guests said we should define reductionism, but everybody said, “let’s not do that, we’ll spend endless energies doing that.” But of course people have been doing it throughout the session and defining it in various ways. So let me say what I mean by reductionism. Reductionism is a simplification of reality that claims to explain reality by getting down to the basic units, whatever the basic units are in a given entity. And the more we get down to the basic units, the more we can explain reality well. Emergentism said, that’s not right since it misses enormous numbers of all those things that are created only at the level of the whole and that affect everything else. Therefore reductionism does not make any sense. But I am really on Evan Thompson’s side; it is a false dichotomy, a false dualism. But it is also a doctrine, a rhetoric and we have heard it repeated here quite regularly. I want to say I think reality exists. This is Paul Cilliers’s point. There is something real out there. There is no point in any of our activities unless we are saying something about what is real. It is not all in our brains. I am not a Berkeleyan philosopher. I do not think I invent the world. The world is there. My job is to try to figure

169

out what is most useful. Paul Dumouchel said what we want to do is find out something that is interesting. I like that. That is to say, what is interesting is what is useful, although useful to whom and to what is a big question. The physical scientists have long been saying that they should get lots of money because they do things that are useful, such as invent light bulbs, invent computers, invent all sorts of machines that are useful to humanity. The biologists say they develop new medicines. Everybody is claiming to be useful, and as a social scientist I want to claim to be useful, too. I want to figure out how the world might be made better. To me, science is trying to figure out how the world works in order to apply some moral judgments and push it in directions that are, from my point of view, better or worse. So the reality exists. Now, it is obviously true that any time we talk about anything, we are simplifying it. If I say this is a table, I am simplifying things. The table is an abstract concept, so of course we are simplifying. There is nothing wrong with simplifying. What is wrong is staying there in some sense and thinking “Ah ha, we got there because we simplified.” Obviously I think what we do is simplify here and simplify there and there and there. Then we get a very complex intermixture of simplifications that comes closer to the real world in some sense. It is an addition of these simplifications. Now I want to take issue with the idea of the distinction between concepts and mathematical models. I think all concepts are quantitative. The concept I use all the time, core and periphery, is quantitative—perhaps in a simplified form—but when I use a concept like that I have quantified. I have nothing against quantification per se. In fact, the better my quantification is, the more plausible what I am saying about something is. However, there is the old joke about the drunk who is found by a policeman underneath a street lamp. The policeman says, “What are you doing?” The drunk says, “I’m looking for my lost keys.” And the policeman says, “Have you lost them here?” And the drunk says, “No, but the light is better here.” There is the whole issue, and I think it is not a minor issue, of what I would call premature mathematization. Indeed, it seems to me to be the major problem in social science. The mainstream economists have gone mad with premature mathematization. It is so far from a description of reality that it is irrelevant. They can play with those figures as long as they want, but they are not telling us very much about the real

170

world. And evidence shows that all their predictions turn out to be wrong. They have not come near to biologists who invent occasionally a better medicine. So a premature mathematization is a bane. And I agree with João Caraça that mathematics has to be a function of what is to be explained and how much real data we have. We cannot look for the keys under the light. We have to look for them where we think they are. And where we think they are may be very difficult to measure. I have done a lot of studies in the last twenty or thirty years in which the measurements are undoubtedly very crude. But from my point of view, they are the best measurements that I can get at this stage of collective knowledge about the problem at hand, and setting up a nice equation would not get me anywhere. In fact, it would push me in the wrong direction. So it seems to me that what we are saying about reductionism is linked to the issue of mathematical models, that the mathematical models assume too much real knowledge, and that one starts with the real problem, tries to get some simplifications and cross them over, and finally get somewhere. That may take us a long time and may make us invent a new mathematics. But it seems a plausible way to go. The answer is not to imitate what was useful for Newtonian-Einsteinian physics, when a major portion of physicists today are questioning the usefulness of those models. And in any case, let me say the end of the game is certainly not simplification: The end of the game is complexity. The end of the game of science is to get to a point where we can maximally explain complex phenomena. I think that is true of natural science just as much as social science. But social science is the most complex of the complex realities in the world. It deals with the most different variables and it will be the longest before we get anywhere near the kind of quantification that makes some people comfortable.

Jean-Pierre Dupuy Yesterday it was suggested that we should try this afternoon to define a kind of consensus among us. To put the question, is there any form of consensus? And I think there is one. It is that the issue of reductionism is no longer an issue. It

171

is already passing. The ways we think in several domains from the social sciences and hard sciences make the question “is this reductionist or it is not reductionist” not true or false, but meaningless. We are beyond the issue of reductionism. When I searched for the origin of the notion of reductionism in the sense we are using it in this conference, I found in the Britannica that the date was 1948. First I was amazed. I thought the origin was much earlier. And then I asked myself, what was the context of that invention? I think I have the answer: It is really a chance event. It is a conference that took place in 1948 at Cal Tech called the Hixon Symposium, where some of the most important thinkers of the twentieth century were gathered. Several of them were cognitive scientists, and although at the time the phrase did not fit, cyberneticians. One of them was one of the most important geniuses of the twentieth century, John von Neumann. We owe him so many things, good and bad, from game theory to automata theory to the H-bomb. It was in the framework of this Hixon Symposium that he put forward what is known today as the conjecture on complexity. The conjecture started with a recursive definition of complexity. He said that to be complex is to be capable of complexification. A complex being is not only complex, it is capable of generating something even more complex than itself. That is how, by the way, he resolved the quasi-theological paradox expressed in the intention to construct an automata. In other words, to be the cause of a being, by definition, that is its own cause. If the automata is a complex being I may be the one who constructs it, but the behavior of the automata eludes it because it is more complex than its structure. So Neumann’s conjecture has had a considerable ontological scope. For example, it renders non-contradictory the following two propositions. First proposition: Physical chemical mechanisms are capable of producing life. Second proposition: Life is infinitely more complex than the physical chemical mechanisms that have generated it. Those two propositions are not incompatible, thanks to Neumann’s conjecture on complexity. So there are algorithms that are capable of generating an epiphenomenon more complex than the structure that generated it. So generation is a one-way path and we cannot go the other direction. And my conjecture is that that is how in 1948 the idea of reductionism entered the philosophical debate.

172

Consider in analogous fashion these two propositions. First: It is human beings who set in action their society. Second proposition: Society transcends them because it is infinitely more complex than they. These two propositions are not incompatible, thanks to Neumann’s conjecture on complexity. They have been known for a long time as being not incompatible. I am referring here to the philosophical tradition which starts with the so-called Scottish Enlightenment and goes to Hayek and contemporary thinkers, a tradition that consists in considering the social to be a complex automaton with a spontaneous order that no will wanted, no consciousness conceived. And we see here the possibility of conceiving the irreducibility of society to individuals without making the former society into a substance or subjectivity. That is, we cannot have at the same time the non-holistic ontology of the social and the non-reductionist ontology. Self-transcendence is another phrase to designate the same idea. I think I have been able to, but I suppose I am hearing a resonance with Andrew Sayer. I have been able to locate this pattern at the root of the thought of Adam Smith, but also, strangely enough, in such philosophical systems as general equilibrium theory, which is generally supposed to be the incarnation of reductions. It is not a matter of assimilating the contents of these theories to each other but simply of identifying in them a single form, the form of self-transcendence. One may judge by the following example, which is a failed attempt at thinking through self-transcendence, Rousseau’s political system. At the end of his life, Rousseau defined a political problem as follows (which he compared to the squaring of the circle and at a time when it was still possible to square the circle, before we knew it is impossible to do). In his own terms, the political problem consists in putting the law above men even though it is men who made the law and they know it. But to Rousseau that remains a contradiction, unfortunately.

Aviv Bergman There are many things that have been said about reductionism in this conference and one that strikes me the most and that I wish we could have elaborated on is

173

the emergence of complexity as a result of linguistics and communicative processes. We touched very little on it in the philosophical part of the conversation and we lack it in the others. I am not going to talk about it now, but I would like to propose to continue the way Jean-Pierre Dupuy started his conversation, by saying that reductionism is in many ways passé. Most scientists do not really care about the modes of knowledge that are out there, may it be reductionism, dualism, or any other mode of knowledge; we simply practice science. However, part of the practice is very strongly influenced by the social context in which we live, and the science of complexity has penetrated a traditionally reductionist science like biology. What biologists are now hoping for, and this was illustrated very nicely by the presentation from the mathematicians in the group, is the fact that we strive to be able to observe some general characteristics of a particular complex system, out of which we will be able to extract some function or property of that system. And one characteristic that people have given attention to is architecture, may it be of the brain, like neural networks, may it be a gene network, may it be a social network, may it be social organization, and the way elements within the organization are communicating. The hope is that just by looking at the architecture, the topology, the interactions of the elements, we will be able to say something intelligent about yet another observation, like robustness. How is it that an organization falls apart when the central component of it is going to be eliminated? But the representation that we are dealing with, such as genetic or mathematical, is a very important part of how we analyze the system, or how we achieve this hope in terms of looking at global things. Another way of looking at the same system is to create a Boolean representation. One interesting characteristic that has received attention lately, and this has been observed not only in the sciences but also in the social sciences, is that most if not all biological as well as artificially constructed networks show something in common. I am talking about, for example, traffic. I am also talking about internet connections, etc. What they all share is that they have very few nodes that are connected to many, many things. On the one hand, can we not employ a very simple reductionist view by studying every single particular node in the system, manipulate it, and observe the effect it has? On the other hand, can

174

we not say something general about this, because as I said, it has been observed in many networks, biological, traffic, internet, manmade, etc? I do not want to go into greater mathematical detail about what it is, but there is already a very famous and celebrated picture of the way by which proteins are interacting. This is the entire representation of yeast, for example. So the biologists thought that we would predict that if we were to knock out the functionality of some element, for example, it would have a major effect, this by looking at the global representation of the system. This is an emergent property which is the result of the way the genetic interactions are created. What we have done for example, was to take each gene in the list and knock it out and see what effect it has on the survival of the yeast. And we see that indeed there is a very nice correlation, and we can say, “yes, we can look at a global property out of which we will say something intelligent about its effects on the characteristics.” We can see that indeed there is a correlation; however, the explanatory power of this correlation is extremely low. There is no explanatory power to this very nice correlation. So just by looking at the property of the characteristics, we cannot say anything intelligent about what one can predict. We say this is a problem of the data, of having very little data to play around with. So what we can do, and this is something that we are very much engaged with, we are generating completely random interactions. And we ask precisely the same question: Can I, by knocking something out, predict something about its functionality? So what we have here is a system by which one has complete knowledge both globally as well as locally, as well as dynamical property and we see that indeed there is a very nice picture. However, if we were to look at the same correlation, it does not give us anything. There is a very weak relationship between the number of connections and how much effect it has. However, if we were to study the details of the system, that is, back to the lower level, we start to see that we have a very strong correlation and very strong explanatory power. Thus, we cannot say something intelligent about complex phenomena just by looking at the higher level. We have to go back to the reductionist view to augment the holistic view of how many nodes are connected, what is the architecture of the system, what is the structure, what is the configuration—that is, the very

175

detailed, reductionist information of the system. So this is just to show that indeed we are seeing the same thing not only in simulated systems, but also in the same experiment that we did with the yeast, where we are now looking at the effect of our correlative power if we were to augment this information with very detailed reductionist information about the effect of each gene in the network, not only in terms of the global picture that this particular system has, but also in terms of the details that this particular gene is involved with. So the take-home lesson from this particular exercise as far as this conference is concerned, is that as scientists we hope to be able to say something intelligent just by looking at the global because this is very appealing. However, as a practitioner of science, I think that we cannot avoid going down to the nitty-gritty details because we are not there yet. I do not think that we will ever be able to eliminate the need to go in both directions.

DISCUSSION

J E AN-PIERRE D UP U Y: It has been estimated that 80% of sexual relationships in a country like Brazil or the United States involve 20% of the population of lovers. Let us consider that a sexual relationship has two partners and let us consider that pair as a network. Two nodes are connected if one makes up to the other and in general it is an asymmetrical relationship. The statistics mean that very few people have lots of sexual relationships and a huge number of people have very few relationships. This is, of course, extremely relevant when we think of a preventive policy against AIDS because then we have to determine where the hubs are. And maybe the ethics is to concentrate one’s attention on the hubs. To conclude, it has been determined, unfortunately for political correctness, that a huge proportion of those hubs are homosexuals, which renders true the prejudice that homosexuality was the major cause of AIDS. So as far as I know, U.S. policy makers are a bit embarrassed, because they want to be efficient and thus they have to concentrate their efforts on homosexuals.

176

But what we have tried to study here is precisely this very attractive and seductive gestalt that you describe, namely that by addressing something that is a general observation about the network, namely that a hub might have a major effect on the characteristic, may not be true. This is precisely the point. And you have to go into the nitty-gritty details. You have to be reductionistic in order to completely eliminate the effect.

AV IV BERGMA N:

What type of information in this context would be the nitty-gritty information? PAUL D UMOUCH E L:

AV IV BERGMA N: For instance, what kind of information would be beside the location in the gene network or beside the location in the network of interaction; what type of messages are going to be transmitted; what is the extent of the intensity of the messages that are going to be transmitted, etc.?

Take for example the transportation system, which is more transparent to us. It would seem at first sight that if Chicago or St. Louis were to blow up, it would disrupt the American air traffic a lot more than if a little city were destroyed. But according to what you say, we should not believe that to be as evident as it seems to be, right? Therefore the question is what should we take into account?

PAUL D UMOUCH E L:

AV IV BERGMA N: I do not know. An expert would have to give us a reductionist view of transportation.

Of course we have to go down into the nitty-gritty to explain. But let me take for example some things I work on. Looking at the world-system as a whole I have a hypothesis concerning one of the major problems of the last fifty years, which is how countries improve their economic standing in the world-economy. Every state is developing; every state wants to expand per capita and other things, and obviously some states do better than others. If we IMMA N UEL WA LLE R ST E IN:

177

start at the level of the world-system, I can argue a thesis that has to do with the structure of the whole, which is that not every country can develop. For some countries to develop and go up, others have to go down, because the percentage of the whole remains constant. Then take a real example: in 1960 the general consensus of analysts was that Ghana was doing better than South Korea. In 20002005 South Korea is doing so much better than Ghana that we cannot begin to imagine it. In order to explain why Ghana did so poorly and South Korea did so well, I have to get into the nitty-gritty, but I also know that there is nothing that Ghana could have done that could have duplicated South Korea’s growth, because the system does not have room for both of them to develop simultaneously. This is what I mean by features of the whole that are ignored by most people who analyze economic development. The analysts consider what Ghana, South Korea, Brazil, or Botswana are doing, and say that one state does it right, and another does it wrong. That is utter nonsense. It is perfectly true that up to a very limited point the actual policy of the states has some effect on the outcome. But there are so many other factors that have much more effect on the reason that South Korea did well. For example, South Korea was geopolitically important for the United States in the period 1950-1990. And Ghana was totally irrelevant geopolitically. Therefore, money flowed into Korea; various things were permitted, and so on, of which Korea could take advantage. Whatever Ghana or Brazil wanted to do would not have matched what Korea wanted to do. This had nothing to do with state policy. It had nothing to do with analysis at the level of the units. It had to do with the operation of the structure at the level of the whole and how that had an impact on the differential ability of some units to do X and other units not to do X. So in order to provide a convincing explanation, I would have to look very carefully at what Ghana was doing, what South Korea was doing, and what other countries were doing in relation to that. And what the World Bank was doing in relation to them, and so forth. And the books that I would write would spend an enormous amount of time on what we are calling the nitty-gritty. But my analytic tools are not dependent on the nitty-gritty. My analytic tools have to do with a whole set of features that are largely out of the control of the units.

178

AV IV BERGMA N: I am completely in agreement with what you said; however, our goal in this particular exercise was to challenge precisely this idea of having a tool, making a prediction, and the effect on what is going on in the system. What we have shown is that this is not sufficient. The tool that we have is not sufficient.

Tool used by whom? Is it a tool used by units or a tool used by the outside manipulator?

IMMA N UEL WA LLE R ST E IN:

AV IV BERGMA N: It is the tool used by the outside manipulator, not by the unit. But the tool is used in order to manipulate the units. And the hope was that as an outsider, an observer would say something intelligent about what one might predict about the outcome.

Yes, but you see you are a biologist dealing as a outsider with a reality which is enclosed and you can let the human beings have the ability to intervene in that in one way or another. I am dealing with a worldsystem with nobody outside there except God. There is the world-system and there are its units. The manipulations that go on are manipulations at the level of all the units in the system and in my analysis I am saying that there is a limited power of the units to manipulate the system to their advantage, because of the constraints of how the whole is operating. IMMA N UEL WA LL E R ST E IN:

When I say we as external individuals manipulate the system, I have in mind something that is naturally possible—namely, we are not manipulating something that in nature will never be observed. For example, knocking out a gene is not an artificial element. It is not something that does not occur naturally: manipulating a gene, changing its expression, changing its functionality . . .

AV IV BERGMA N:

IMMA N UEL WA LL E R ST E IN:

AV IV BERGMA N:

May happen spontaneously or . . .

It happens spontaneously under stressful conditions.

179

IMMA N UEL WA LL E R ST E IN:

Artificially . . .

Artificially, I try to mimic what happened naturally. Then we ask, if it had happened in nature, can we say something intelligent, with some predictive power about the outcome of the system?

AV IV BERGMA N:

Once again, this is one of the differences between social science and biology and physics: in the social sciences the power to do artificial experiments is quite limited.

IMMA N UEL WA LL E R ST E IN:

ISTVA N REV: I would like to respond quickly to the remark that we have not talked enough about humanities. One of the reasons why we cannot even say interesting things about reductionism in the humanities is that this is probably a completely different issue than in the sciences. For example, in history the most important and successful school in the past two decades has been microhistory. I think these historians managed to do something that cannot be characterized as less interesting than what Braudel did about the Mediterranean. One of the important results of writing history in the past two and a half decades was the marriage between history and anthropology. Historical anthropologists who try to use the repertoire of anthropology in history writing have enriched historical analysis in an unbelievably complex way. On the other hand, we know the story of cliometrics, which was an important tool in economic history in the 1960’s and 1970’s; we also know that psychohistory was unable to produce the results that had been hoped for. This is not an issue of moving from one discipline to another which necessarily leads to reduction. Although I might say that it might be possible to conclude, on the basis of what is happening in the humanities, that the question of reductionism versus complexity is a wrong issue, or at least it is a different issue than it had been.

One of five different laws passed by the French Parliament said that French schools have to teach that colonialism was not all bad, that it may even have had positive aspects. Another law said that the extermination of

J E AN-PIERRE D UP U Y:

180

the Armenians by the Turks was genocide. So French historians are saying, “This is our job to say those things, it is not a job for the parliament.” There is a French historian who last year wrote a book on the history of slavery. He is being sued because he wrote that most people who know this topic say that not only the West practiced slavery but well before the West the Arabs and the Greeks did. He is being sued by a particular lobby of people from Martinique and Guadeloupe, islands which are no longer French colonies; they are part and parcel of the French nation. Let there be no confusion here: These people are suing this historian because they consider that the slavery of the Negroes by the West is the only slavery to be considered. There is a political discussion going on in France now about who has the right to write history. In that discussion the words reductive and reductionist come up again and again. So it is absolutely essential. IMMA N UEL WA LL E R ST E IN: I think that when parliament passes such laws it is making a political rhetorical statement. And I am not sure it is illegitimate for parliaments to make rhetorical statements. Parliaments do have an actual governmental duty to decide what should be taught in schools. And it is the role of the specialists who also want to guard their turf to say that they should make these decisions rather than the politicians; historically it is a compromise. There has been a great debate in India under the BJP government; they revised the Indian high school history textbooks to give a bigger role to Hinduism. The Indian specialists wrote a statement to the effect that it is most inappropriate for parliament to do this and it is wrong history and so forth. Of course there is no right history, is there? The fact is that nobody ultimately has the right to be the exclusive controller of rhetorical truth. Neither the state nor the politicians nor indeed the specialists, but specialists are regularly proved wrong by successive specialists, so that is a political problem of how to balance the right of the state as an expression of the collectivity versus the right of intellectuals to pursue the truth as they see it.

To come back to this issue of the differences between the humanities and the other disciplines, it seems to me that natural science is involved

A N D R EW S AY E R :

181

in explanations of what is and what has happened. The humanities are thinking more creatively about new social forms, new kinds of people, and such. Artistic genres, simple rules which generate complexity, generate emergent forms. Artistic genres actually are about novel experience, novel kinds of sensibility, new modes of social being. And I just want to make a final comment about ethics and social science—and Foucault, of whom I am not a fan—that makes a distinction between ethics of authenticity and ethics of creativity. Ethics of authenticity are judgments that are tied to what we take human beings to be, whereas ethics of creativity are about creating new kinds of human beings. There are dangers with both. I think some creativity in being disregards susceptibility to suffering. And the ethics of authenticity can be deeply conservative in essentializing and naturalizing contingent social forms and preventing development towards more enabling ones. But we need both. The creativity comes out of what we are. The possibility being generated by what we are generates what we could become. Social science is mostly about what we are and have been, but it is also about what we could be, what our potentials are. It is about thinking creatively. Predisciplinary theorists of the Scottish Enlightenment were very interested in not only what we are but what we could be. I want to return to the opposition between simplicity and complexity in mathematics. Indeed, mathematics simplifies reality and eventually oversimplifies the reality; but simplicity is only a part of the story. I call it the problem of generativity. Equations, algorithms, programs are simple, but the solutions can be incredibly complex. And we cannot read the diversity and the complexity of the solutions looking at the equation. We can say that an equation is an oversimplified conceptualization of reality. But it is in general impossible to anticipate the diversity and the complexity of solutions. The simplicity is a forced simplicity and in that overt simplicity we have a covert complexity. Very simple non-numeric equations can be unpredictable, chaotic, and sensitive to initial conditions in their solutions. We can generate complexity from simplicity. It is exactly the same for a different set of equations, those of general dynamical systems. The key is that although we iterate elementary operations, simple elementary opera-

J E A N PE T I TOT:

182

tions, we can have incredibly complex behaviors. We also have what is called an infinitesimal generator, which is an equation itself. So we have the diversity, the empirical complexity, and a simple equation, but inside we have the complexity of the solutions due to iteration. I want to add a second very short point, for Prof. Wallerstein. Your remarks reminded me of the history of Ecole des Hautes Etudes en Sciences Sociales (EHESS), because at the beginning Braudel was very enthusiastic concerning a sort of Galilean/Newtonian idea for social science. We need mathematics, we need models, etc. It was an optimistic phase of social science. Then about ten years ago there appeared a dossier in Le Monde directed against the Galilean ideal of social science, which stated that the scientific idea in social science has completely failed and social science must find a new way, a hermeneutic way, and use narrative, etc. Today there is again a new invention and the first priority in the future is a link between natural science and social science created by the science of complexity, complex systems, etc. I am not so sure that we want the social sciences and the natural sciences to become one. I am for difference, and for keeping the differences apart. It is not that they should not talk to each other, but the idea of reducing them to the same thing seems to be a meaningless effort. I want to argue for the social sciences, for the humanities, for the arts in a certain way, by making an irrational argument from complexity that goes like this: We do not have access to the world in an unmediated way. That means we cannot understand the complex world in its complexity. What we do in order to understand is make models. I also like to use models in a wider sense of just formal models: in that sense America is a model, separate rules is a model, holism is a model. They reduce the complexity so that we can understand. But we know there is a reduction of something that is complex, which means that the bits we leave out in the reduction are nonlinear and therefore we cannot predict the size of an eventual error. This means the models are always per definition problematic. That is not bad; it is not that we can avoid it. The point is that there is no model for making models. There is no position from which we can do it. We choose certain models and in that choice

PAUL CI LLIERS:

183

we cannot escape the moment of component. And in that moment of component, this choice of ethics comes before rationality. Ethics is the first philosophy here. And that domain is the domain of the humanities, of the art or creativity of responsibility. Since it comes before rationality, we cannot do science without first being in a domain that is normally that of the arts and humanities. The paper by Prof. Sayer started by saying that, rhetorically, nobody in the social sciences is a reductionist anymore, but the rest of his paper demonstrated that a lot of people are, in practice. And it is a bit parallel to the rhetoric about racism. A hundred years ago people were very happy to say that they were racists or to put forward theses which we would think of today as racist. Today nobody is racist by self-definition. But in fact it is not hard to demonstrate that most of the people who deny that they are racists are in fact racists. So the only question is, why has the rhetoric that racism is not legitimate changed? What has changed in the social situation? I think it is similar for reductionism. I think reductionists are alive and well and matter a great deal, but for various reasons nobody is willing to admit they are a reductionist, because not being a reductionist is considered a good thing. Everybody says, I am not a reductionist because I think X, Y, and Z. It seems to me that it is not an issue of generativity. I come back to what I said before—that concepts are models and if models that are quantitative are generative over the long run by iterating, then rather simple concepts are simply simple models and are as generative as more complex models which are called mathematical equations. So that is not the issue. And the issue is not that we look at empirical reality in its detail. I have spent a lot of my own energy looking at empirical reality in its detail. Again, I try to explain the historical development of the modern world-system and I cannot do it without looking at what happened in 1623 in France, for instance. I have to in order to make any kind of sense; but that is not being reductionist. Being reductionist, again, is to assume that units within a system are the only thing that matters in terms of generating the rules. To take a simple example: if I have a 1000 units, I tabulate them and discover that the rule is X, Y, and Z. This leaves out the fact that there

IMMA N UEL WA LL E R ST E IN:

184

are rules in the system which constrain what the units do. And it is constantly reductionist when people try to generate the rules by looking at the behavior of units in the system. This, it seems to me, goes on in the social sciences certainly. I think it goes on in the natural sciences, I think it goes on in the humanities. I do not think it is an irrelevant concept in literary studies or history, if you want to put in the humanities. I put history in social sciences, but I do not want to have that argument. I think reductionism occurs in many areas and I think it is nefarious in its impact on the kind of scientific work that we all do. If we talk of bringing together the social sciences and the natural sciences, of course that has been a project of various people for a hundred and fifty years. It has been constantly under discussion. The general form that it has taken has been for the natural scientist to preach to the social scientist that they are insufficiently like the actual sciences and that they should copy the behavior of natural scientists in various ways. There is another form which is much less widely used but Ilya Prigogine came close to saying it, which is that the natural sciences should merge with the social sciences by copying the social sciences. They should import into the natural sciences the techniques of the social sciences. It is another road to merger; as a percentage, it is a smaller road than the other. From my point of view, epistemologically the most important task before us in the next twenty-five to fifty years is overcoming the concept of the two cultures. In the basic argument of the two cultures there are two epistemologies. There are two different ways to know reality. One way is by experiment, the scientific method, and another way is more intuitive or hermeneutic. It also then separates the search for the true and good and beautiful as separate activities. This was not historically the case. Before ca.1750, almost nobody would have argued that there are two separate epistemologies. Certainly Aristotle would not have argued that. He wrote on everything, but he did not have different epistemological methods when he wrote on different subjects. And so should not everybody else. The origin of the two cultures is one question, but the nefariousness of the idea that we have two separate epistemologies is the exception and to me it is a problem. It does not mean that if we all agree that there is some

185

kind of single epistemology by which we approach reality that everybody will be doing the same thing. Obviously some people will still be working on novels and other people will still be working on genes, and the third set of people will still be working on what we now call social scientific problems. Everybody cannot do everything: That is not the issue. The issue is how we know what we know, and whether there are two different modes of knowing how we know. Reductionism is central to the point of view within the natural sciences and those social sciences that attach themselves to the natural sciences, which says that our way of knowing is radically different from the way of knowing of someone who works on literature or history. They know things somehow in a different fashion. We really know it, we observe the truth: they do something else. And that reductionism, that tonality, that insistence, for example the insistence that only by creating serious mathematical models can we arrive at understanding complexity, seems to me plain wrong. That is the heart of the issue of reductionism. It is not an irrelevant issue even in this room, where people presumably are at least ready to ask questions about reductionism. I am not sure there will be a consensus among us but science is fundamentally not doing experiments. It is the building of models. In this sense it is true of natural science. It is true or maybe should be true of the social sciences. It could even be true of the humanities and philosophy. But I am summarizing what some people here said. There was a radical break in the philosophy of model building. It occurred at some time in the history of ideas. For me the year was really 1948. What was exactly the nature of this revolution? In the words of mathematic modeling, there are models that have the two-fold property of being apparently simple, but actually they are complex in the laymen’s sense. That is, they are capable of generating outcomes more complex than they are. So let us suppose that the only way, or the best way, or the simplest way, to model a given reality is to put forward the model that has this property. It looks like a reduction, but it is not a reduction in the sense that Jean Petitot explains so well, because the model itself is unpredictable.

J E AN-PIERRE D U P U Y:

186

I think what happened about fifteen to twenty years ago has some consequences that relate to what we are talking about now. To put it in a very simplified, not reductionist, way, one can characterize the twentieth century as a deep continuous attempt at social intervention on the macro scale, which you call in your paper social constructivism. What is meant here is social planning, the conviction that there are predictable social outcomes of human action. This is what failed so spectacularly in the twentieth century. Around 1989, it was the end of an era. In fact, I would say this era ended before that—it ended at the moment when it became untenable to continue with the continuous social intervention on the macro scale. And somehow this coincided with the emergence of new technological opportunities, new levels of scientific and social inquiry when the focus turned to nanophenomena, information flow, genetic manipulation, that is, intervention on the micro level instead of the macro level. Emergentism has something to do with this and is related to the dependence or interdependence among all of these elements. Nowadays something is offered as a response to intervention; this comes in the form of emergentism. There are self-organizing networks, decentralized networks that are outcomes of human action. It is impossible, tragic, and dangerous to try to centralize them, organize them, plant them, and construct them, but it is happening. There is a movement in the world of information theory, information technology, and also in biology. Instead of revolutions, we have emergence and a new social possibility for the world.

ISTVA N REV:

From my point of view as someone who works in the humanities, the descriptions of the humanities offered so far seem to me particularly impoverished relative to my understanding of the humanities. As far as I am concerned, the humanities constitute the area of human knowledge whose central concern is with meaning. I do not mean meaning in the sense of “what does this equation mean?” I mean meaning in relation to other human beings. In fact, from the point of view of the humanities, we could almost define a human as a meaning-seeker. And so the humanities are profoundly concerned with relationality because it is in relationality that the meaning inheres, if we are talking about K ATH ERIN E HAY LE S:

187

what something means for human beings. The techniques of the humanities, in trying to get at this relationality and how this relationality can be established, what its boundaries are, what its significance is, in my view have much to do with contemporary ideas of complexity. That is because they are also concerned with meaning as something that emerges out of local interactions. Central issues in the humanities are also central issues in how we think about complexity. So I think the humanities have traditionally been concerned with complexity and now there is the possibility for forging a vocabulary between the social sciences and sciences and the humanities that will relate some of the traditional concerns of the humanities to issues of complexity in general. I would like to say a word about the history of ideas. Almost everything that anybody says today has been said over the last 2000 years by somebody else. That is elementary. And it does not prove anything to say this idea was already said by Adam Smith, Albertus Magnus, or Sun-Tzu. The issue is not whether the idea was ever broached by anyone, but whether the idea gets significant acceptance at a particular social moment. What we need to trace is not who said it first, but at what point did it become socially important. I think reductionism, whether or not the word was used, became quite socially important in the nineteenth century. Feudalism was a term that was only invented during the French Revolution, but the feudal era existed long before that. We do not need the term to have the reality. Complexity study is interesting not because it just started in the 1970’s, but because it can be traced back to all sorts of important figures who said essentially the same thing but were not taken seriously when they said it. And all of a sudden there appears what I call a knowledge movement. A knowledge movement occurs when a significant number of people with a significant level of prestige argue a particular thesis. So complexity studies, for example, only exist, in my view, since the 1970’s. It is a sociology of knowledge question: why are these ideas picked up in the 1970’s but not the 1870’s or the 1770’s?

IMMA N UEL WA LLE R ST E IN:

One of the attractive things about the idea of emergence is that it helps us appreciate that the world is open to many possibilities, not just one.

AND REW SAYER:

188

Society could develop in many different ways. But actually I do not think Hayek does justice to that at all. He calls one of his books The Fatal Conceit, and the fatal conceit is constructivism. But he is guilty of the conceit himself. He passes off the development of capitalism as the product of rules that have arisen spontaneously, knowing they were proposed by dominant groups to further their own interests. It is one particular way of organizing society, but it is just the one that they were successful in imposing. Markets themselves depend upon particular selective forms of planning, and constructivism cannot have a market system without a high-level state system. There cannot be a market without taxes because the state has to be funded. Capitalism is not just market; it is also large units of capital that act in a very constructivist manner both in their return to their own operations and their intents to dominate government. Notice how domination is effaced in the Hayekian discourse, which is profoundly fatalistic and profoundly apologetic. I would like to try out my own little piece of potted history on you for a moment. Two hundred years ago the structures of knowing underwent a crisis and in large part the questions were quite practical, due to something very important that happened in France. The French Revolution suggested to some social critics that the better world was located in the past; they called themselves conservatives. And other social critics tended to think that the better world was in the future and it would be the outcome of revolution; more or less they were the radicals. But what was interesting about this is that each group made an argument based on a set of values. But the world could not go forward according to one set of values without leaving the other set of values behind; they were incompatible. Now this is very much the story of the first half of the nineteenth century until Huxley came along. Certainly he may not have been the first, but he articulated it very well in the 1880’s: In thinking about what we need to do in terms of political policy to move the world forward, we need to be more like people who call themselves scientists. Now why did he say that? He said that because of what the scientists do. They do not take into consideration whether one is a conservative or a socialist or what kind of bias one has because of race or nationality. They are not interested in that. In fact they purposefully leave that

R ICHA RD E. LEE:

189

out. Now why don’t we do something called social science or political science in this realm of human reality where we leave out all of that? We think about the world more like scientists think about their objects of study. And indeed a lot people ask, why don’t we do that. Thus we have this legacy of the bias of value neutrality in the social sciences. And if we leave out values in the social world, I cannot think of anything more reductionist than that. But let us skip forward from the birth of the modern social sciences in the middle of the nineteenth century to the period that we’re talking about, which I think begins in the 1950’s with structuralism in the humanities; perhaps it doesn’t start then, but it begins to be extremely important socially. What seems so interesting about these times is that two things are happening almost contemporaneously. We have complexity studies in the sciences that absolutely state that very simple systems with very few degrees of freedom can give rise to extremely complex and unpredictable behavior. At the same time in the humanities people are saying that the world is a relational world, that human systems are relational systems, that they are systems of difference. And each of those in its own way is an attack on the idea of reductionism. So it is not just in the sciences, it is not just in the humanities, but it is happening across the spectrum of modes of knowledge formation. That, I think, says that there is a real crisis in the entirety of the way that we think about and study the world. I think Immanuel [Wallerstein] was responding to this idea of the sciences and social sciences converging or merging. I do not think that the idea is a matter of merging, but of creating bodies of knowledge that are non-contradictory. In other words, there are not people over here who are creating bodies of knowledge that are value based, while there are other people over there who are creating bodies of knowledge that supposedly enumerate or discover facts about a world from which values are excluded. I think that’s the message of complexity studies: That the world is a creative one, natural systems are created and create. This is the idea of bifurcation at critical points. But the same thing is happening in the humanities. That seems to be something that we should not be blind to; we should not say it is just something happening in the humanities or just something happening in the social sciences.

190

J E AN-PIERRE D UP U Y: I want to respond to Andrew Sayer’s point. I fully agree that Hayek at the end failed in his programs. And actually one of his major concepts means to abandon oneself almost, in his words, to the dark forces of social processes.

When we started this conversation this afternoon you [JeanPierre Dupuy] said that there is consensus that reductionism is passé; it is over.

PAUL D UMOUCH E L:

J E AN-PIERRE D UP U Y:

I said if there is a consensus.

Yes, I think the answer is that there is not. And let me put this a little ironically perhaps; I think that Immanuel Wallerstein believes that reductionism is irreducible. And I heard that even though people do not talk about reductionism anymore, it is there. Just like racism. They were not talking about it, but it really is there. Now this may be true. But there is also the problem of whether or not reductionism is the same thing when we do not talk about it. And if we are taking into account that this is a complex system, then maybe it is not the same thing. PAUL D UMOUCHE L:

I absolutely agree. It is not the same thing. Go back to racism. Whether people are openly racist or hide it reflects a different social atmosphere. But we are not here to discuss racism. We are here to discuss reductionism and it is an interesting phenomenon that people are denying that they are reductionist. That does not mean that reductionism is passé, even if reductionism takes a somewhat different form at this point. And for me it still poses a problem. It is not an irreducible problem. I did not suggest that there could never be a world in which people were not being reductionist. In fact I am quite optimistic about the opposite and real possibility that over the next twenty-five to fifty years we will arrive at new epistemological consensuses that will exclude reductionist thinking. But as of 2005, it is not excluded that the reductionists are alive and well all over the place. And since I think it is a negative phenomenon,

IMMA N UEL WA LLE R ST E IN:

191

I think we should try to see the forms that it takes and what it keeps us from seeing and doing in an intelligent way. To erase it rhetorically does not solve the problem. To insist that we are not being reductionist, when in fact in one form or another we are still being reductionist, is not helpful intellectually and ethically. I also agree with Paul Cilliers that the ethical component is irreducible, that is to say, that scientific knowledge always includes an ethical component; that one of the problems of nineteenth-century science and social science was indeed the denial, even though again we can explain why they did it and what function it played and so forth. I do not have time to do that nor do I think it is relevant at this point. The point is, can we arrive at an epistemological consensus that excludes reductionism as I define it. We have not arrived there as of 2005 and it is a dangerous illusion to think that we have arrived at the elimination of reductionism as a real ongoing problem in the world of knowledge, not in sciences or social sciences, but in the world of knowledge generally. There is no assumption that we would arrive at a consensus. At least I never believed that. Had we arrived at a consensus, I would think that we had assembled a group that was too close in views at the beginning.

192

INDEX

Abstraction, 12, 13; objectification of, 13; problems of, 17; unavoidability of, 17 Actions: costs/benefits of consequences of, 143; evaluation of, 143; practical character of, 27; rationalized model, 27 Actors: as agents with only physical powers, 26–28; apathetic, 26–28 Actualism: reduction of causation to interaction in, 11 Aesthetics: transcendental, 131 Agency: conflated with structure, 36n5; dualism of, 52; epiphenomenal, 36n5; human, 96; insistence on, 51; intentional, 96; promotion/limitation by culture, 38n22; structure and, 2, 27 Aggregation, 48 Alexander, Samuel, 60, 80, 84n7 Altruism, reciprocal, 163 Ambiguity, 95 Anthropology, 6, 19, 24 Anti-constructivism, 137–141 Anti-essentialism, 2, 23–26, 35 Anti-naturalism, 26 Anti-reductionism, 2, 5, 12, 113 Archer, Margaret, 10, 22, 23, 31, 37n13 Aristotle, 69, 185 Association field, 122, 123fig Asymmetry, 42; forms of, 42; intervention and, 18; realism and, 55; relationships and, 22; in stratification, 15, 16

Atomism, 16, 34, 36n4, 51, 83, 84n7; mechanistic, 139 Attachments, 27 Autonomy, 91, 93; causal, 17; empirical, 109; in higher order mechanisms, 17; theoretical, 109 Autopoiesis, 59, 67, 68, 79, 84n4, 91 Axelrod, Robert, 143, 162, 163 Bahavioralism, 5 Bayle, Pierre, 137 Behavior: action reduced to, 26; alternatives, 66; collective, 48, 156; complex, 64, 183; critical, 110; emergent structures of, 48; explanations of, 6; of neurons, 63; pathological, 7; physical, 10; renormalization group and, 110; social, 6, 9; social determinants of, 33 Being, stratified, 16 Beliefs: as function of material circumstances, 6; subjective, 30 Bénard cells, 75, 77 Bentham, Jeremy, 142 Bereavement, 27 Bergman, Aviv, 4, 44, 54, 55, 90, 92, 94, 95, 101, 104, 165, 173–176, 177, 179, 180 Berkeley, George, 140 Bhaskar, Roy, 10, 12, 13, 36n8

193

Bishop, Robert, 71, 75 “Blank-slateism,” 24 Block, Ned, 111 Bodin, Jean, 137 Bonaparte, Napoleon, 137 Bosking, William, 122 Boundaries, 2; broaching, 54; changes in, 47, 55; clear, 47; complexity and, 48; constructionism and, 54; disciplinary, 9, 45; enabling, 55; evolution of, 55; as ingrained characteristic, 92; levels and, 44, 45; malleability of, 92; in natural phenomena, 54, 55; in nature, 44; organizational, 2, 46, 47; origin of, 46; place of, 55; revisions of, 54; of self, 92; semipermeable membrane, 67; setting, 48; spatial terms, 55; topological, 67; transgressing, 169; unfixed, 55 Bourdieu, Pierre, 9, 23, 24, 27, 28, 36n5 Brain, 3; as dynamic neuronal network, 63, 64; endogenous activity in, 62, 63; hierarchy of processing levels in, 62; localization and, 63, 65; modular view of, 65; as nondecomposable system, 63, 64; organization of, 62; principle of reciprocity and, 62 Braudel, Fernand, 180, 183 Bressloff, Paul, 117, 133 Broad, C.D., 60, 84n7 Campbell, Donald, 65 Capitalism, 6 Caraça, João, 48, 95, 171 Care, 27, 28 Cartwright, Nancy, 83 Catallaxy, 142 Catalysis, 66, 67, 84n4

194

Cathexis, 38n22 Causal Inheritance Principle, 77, 81, 82 Causality: attempts to reject, 36n3; Causal Inheritance Principle in, 77, 81, 82; circular, 2, 74, 101; critical realism and, 10–13; dependence on context, 26; downward, 2, 68, 73–83; failure to reproduce, 26; mechanisms of, 13, 49; mental-to-physical, 77; misattributions of, 6; not unilatereal process, 11; Physical Realization Principle in, 77, 81, 82; powers existing even though not exercised in, 11; Principle of the Causal Closure of the Physical Domain in, 77; as production of change, 10; reduced to interaction, 11; successionist theory of, 10; synchronic reflexive downware, 74, 75; transformative relationships and, 12 Chalmers, David, 111 Change: in boundaries, 47; environmental, 59; in knowledge, 47; linguistic, 53; material, 19; in natural environment, 20; production of, 10, 11, 12, 15, 25; qualitative, 110; resistance to, 92; of rules, 103; slow, 92; stability and, 110; in susceptibilities of individuals, 15; by taking advantage of lower stratum properties, 19; through interaction, 14; unintelligibility of, 110 Charles V (King of Spain), 137 Cilliers, Paul, 41, 42, 44, 50, 51, 55, 92, 96, 98, 99, 101, 105, 169, 183 Cliometrics, 180 Closure: causal, 93; physical, 93 Co-emergence, 73; dyamics of, 91 Co-evolution, 19, 23, 25, 168; complex, 26; in socialization, 26

Commitments, 27 Common sense, 3, 141, 149, 160, 161 Communication: of dense patterns, 97; emergence of complexity and, 4; emergence of new meaning and, 95; enabling, 55; non-verbal, 99; rules of, 136 Communism, 139 Complexity, 171; abstract, formal rules and, 136, 137; boundaries and, 48; conjecture on, 172; constraints and, 105; emergence of, 4, 174; endogenous, 136; as evolutionary process, 136; explanation of, 41; notion of, 48; prohibition of centralized hierarchical organization and, 136; of reality, 47; simplicity and, 182–183; source of, 136; structured, 105; studies, 168 Comte, Auguste, 138 Conceptualization: human, 96 Consciousness, 165; access, 111; emergence and, 22; gap between and neural activity, 3; interaction with others and, 22; neural correlates of, 112; phenomenal, 3, 111, 116, 131, 163; reality of, 14; reduction of, 111; and self, 22; stratification and, 22; understanding, 14 Constant, Benjamin, 137 Constraints: complexity and, 105; in complex systems, 65–69; context-free, 66; context-sensitive, 66, 67, 105; first/second order contextual, 67, 68; importance of, 105; on interaction, 66; repressive, 141; on states, 92 Constructionism, 51; boundaries and, 54; essentialist “other” in, 8; moderate,

37n18; political, 136, 137; problems of, 54; rational, 136, 138; social, 2, 8, 23–26, 35, 54, 56, 136, 187; strong, 43, 44 Conventionalism, 30 Cooperation, 140, 145, 146, 149, 150, 156, 163, 164; emergence of, 164; retaliatory, 161 Cowan, Jack, 117 Cultural: constructions, 24; discourse, 19; evolution, 103, 141–143, 161; interpretations, 20, 21; norms, 28; selection, 141; significance, 20, 24; studies, 6, 168–169; variation, 25, 26, 38n20 Culture: imperialist concepts of, 24 Darwin, Charles, 103, 141 Dean, Kathryn, 38n22 Deducibility: mathematical, 110, 125 Dennett, Dan, 111 Dennett/Chalmers controversy, 108 Description: reductive, 164; relational, 164 Determinism, 2; as feature of scientific models, 72; holism and, 72; irreducibility of individual reflexivity and, 33; limitations of, 33; as metaphysical thesis on nature, 72; social, 8, 33 Development: from co-evolution of social, psychological, biological and, 19; economic, 178; mental, 19; of states, 91, 92, 177; of susceptibilities to enculturation, 25 Disaggregation, 15 Disciplines: advancement of ambitions of, 9; boundaries in, 9, 45; competition

195

Disciplines (continued) among, 8; defining, 46; discourse of, 43; historical, 167; imperialism in, 8–10, 24, 43; intellectual construction, 46; of knowledge, 2; lack of intellectual meaning in, 167; master, 24; mutual aversions in, 8; in natural science, 35n1; non-communication amongst, 3; organizational construction, 46; in social science, 46; stratification and, 5 Discourse: coherence in, 21; consistency in, 21; cultural, 19; disciplinary, 43; empirical challenges to, 21; meaningful character of, 10, 15; misinterpretation of meaning of, 6; neoliberal, 15; out of nothing, 23; performative, 23; point of, 15; prioritization of, 6; production of effect by, 15; reductionism in, 7; of scientists, 23; subject constituted in, 23 Dogmatism, 31, 32 Dualism, 2, 7, 134, 160, 168; of agency, 52; deconstruction of, 29; fact-value family of, 2, 8, 28–32; false, 169; problems in, 28, 29; “seepage” of qualities in, 29; of structure, 52 Dumouchel, Paul, 51, 100, 162, 163, 170, 177, 191 Dupuy, Jean-Pierre, 3, 41, 45, 92, 95, 97, 101, 102, 103, 104, 106, 135, 160, 162, 171–173, 174, 176, 180, 186, 191 Durkheim, Emile, 16, 24, 37n13, 49, 139 Ecosystem: shaped by species, 44 Eigenmodes, 125, 125fig, 126, 127fig, 128fig, 131, 133, 134 Eliminativism, 109

196

Elster, Jon, 139 Embodiment, 101 Emergence, 2, 13–17, 108, 109; arguments against, 22; British view, 80, 84n7, 85n10; Cartesian mind-body problem and, 81; of communicative interaction, 10; of complexity, 4, 174; in complex systems, 2, 3, 57–83, 107– 157; consciousness and, 22; in contrary dynamics, 43; of cooperation, 164; critical realist view, 10–13; criticisms of, 57, 73–83; defining, 14, 58, 59; denial of, 36n5, 80; dynamic, 73; epistemological, 57, 70, 71; explanations of, 7, 8; as fusion, 84n5; of geometric visual patterns, 117–129; holistic, 84n5; intentionality and, 102; levels in, 42; of meaning, 95, 96; medium of happening, 94; mereological, 69; micropreconditions and, 22; naturalist, 112; as neural correlate, 131–133; of new meaning, 95; nonlinear dynamics in, 59–65; ontological, 34, 57, 70, 84n7, 85n8; phenomenal, 112; physical closure and, 93; recognition by systems, 94; reductionism and, 48; relationship to context-sensitive constraint, 105; of selforganizing structures, 3; semiosis and, 15; stratification and, 14; subtraction by, 102; as temporal process, 59; upward, 160; vertical, 112 Emergentism, 169, 187; dynamical, 134; formal, 160; objective, 109; ontological, 160; reductionism and, 2 Emotion: cognitive content in, 30, 31; reduction to affect, 30

Empiricism, 99, 100, 109 Energy: kinetic, 71; potential, 71; total, 71 Enlightment, Scottish, 138, 140, 173 Environment: changes in, 59; evolution and, 44; interaction with, 91 Epiphenomenalism, 109, 134, 160 Epistemology: politics of, 2, 45, 46, 49, 53 Ermentrout, Bard, 117 Essentialism, 25; attribution of causal powers in, 26; deterministic forms, 35; misunderstandings of, 38n23 Ethics: of authenticity, 182; of creativity, 182; utilitarian conception of, 142 Evolution, 100; of boundaries, 55; cultural, 103, 141–143, 161; environment and, 44; as process, 104; scientific, 93; of traits, 44; units of selection in, 101 Existence: institutionalized, 45; layered, 45; preconditioned, 42 Experience: lived, 93; near-death, 115; phenomenal, 3, 111–115, 130; of spatial forms, 113 Fantasy, 53 Feminist theory, 29 Ferguson, Adam, 138, 140, 141 Flexibility, 92 Foucault, Michel, 25, 182 Freud, Sigmund, 38n22 Functionalism, 108, 109, 112, 134, 160; in natural science, 109 Game theory, 145–155; Prisoner’s Dilemma, 143–145; Prisoner’s Dilemma,

Iterated, 145–155, 161; tit-for-tat strategy in, 148, 149, 150, 152, 156 Gender, 35n2; naturalized accounts of, 25 Golubitsky, Martin, 117 Gray, John, 142 Grotius, Hugo, 137 Habitus, 24 Haeckel’s law, 142 Hallucinations: entoptic vision and, 115; retinal images and, 126–129; visual, 114, 114fig Hamstogen, Peter, 163 Hayek, Freidrich von, 135–137, 140, 141, 142, 143, 160, 173 Hayles, Katherine, 43–44, 52, 53, 94, 187 Hierarchy: chain dynamical, 94; order of, 137; organization of, 61; of processing levels in brain, 62; relational, 18; in systems, 65; tangled, 101–102 History, making, 102 Hobbes, Thomas, 137 Holism, 2, 84n5, 90, 142, 164; as mereological emergence, 69; as mythology, 139, 140; organicist, 138; quantum mechanics and, 70; relational, 59, 69–73, 84n5, 84n7 Hopfield equation, 118–120, 160, 163 Humboldt, Alexander, 137 Hume, David, 109, 138, 140, 141 Husserl, Edmund, 131, 164, 165 Identity, 37n19; self, 61, 92; type-type, 131 Imperialism: cultural, 24; disciplinary, 8– 10, 20, 24, 43; reduction of knowledge content and, 10; sociological, 16, 25

197

Individualism, 2, 138; complex methodological, 3, 135; conservatism and, 33; methodological, 49, 54, 101, 135, 137–141, 142, 160; responsibility for fate and, 7; responsibility in, 32–34 Ingold, Tim, 22 Integration: of boundaries/contours, 122 Intentionality, 96, 102; collective, 103; defining, 103, 104; individual, 103; interaction and, 103; meaning and, 104 Interaction, 6; among mechanisms of different strata, 8; causal, 61; changeproducing, 14; communicative, 10; complex, 109; consciousness and, 22; constraints on, 66; critical realism and, 11; downward, 56; environmental, 60, 91; flat ontology and, 17–22; in/between subsystems, 61; between individual properties, 135; input/processing/output, 60; intentionality and, 103; local, 65, 69, 93; nonlinear/linear, 59–65; social, 19, 25; susceptibilities to enculturation as product of, 25; tanled, 96 Interactionism, 36n5 Intertextuality, 95 Intervention: adequacy of, 99; asymmetry and, 18; co-evolution and, 19; downward, 19; failures of, 18, 19; flat ontology and, 17–22; in lower-order processes, 18; material, 19; in nature, 18, 19; obstacles to, 19; social, 99, 187; through mechanisms distinctive to humans, 18; unintended consequences of, 19 Isomorphism, 130

198

Kant, Emmanuel, 81, 131, 137, 157n4, 157n5, 164, 169 Kellert, Stephen, 71 Kim, Jaegwon, 2, 57, 73–83, 84n7, 86n11 Klüver, Heinrich, 115, 127, 134 Knowledge: change in, 47; collective, 171; coping with world through, 21; disciplines of, 2; discursive, 27; embodied, 27, 141, 161; fallibility of, 23, 24; as incorrigible, 23; multiple forms of, 168; practical, 21; privileged insight into, 24; production of, 20; propositional, 27; questions of being and, 12; routines, 141; sociologically imperialist reduction of content of, 10; tacit, 161; teaching, 169; technical, 100, 101 Kronz, Frederick, 71 Labor: division of, 15 Laissez faire, 136 Language: acausal, 42; ambiguity and, 95; artificial, 165; change in, 53; of collective variables, 60; communities, 11; for describing complex behaviors, 64; determinism of, 52; everyday, 163; mathematics as, 49; social reality and, 56; stability of, 37n11 Learning: postdisciplinary, 9 Lee, Richard E., 4, 45–46, 49, 53, 96, 97, 164, 189–190 Locke, John, 137, 139 Logic: defining, 12; reason reduced to, 32 Louis XIV (King of France), 137 Machiavelli, Nicolo, 137 Magnus, Albertus, 188

Mandeville, Bernard de, 139–140, 141 Marx, Karl, 3, 102, 103, 104 Materialism, 6 Mathematical models, 114, 114fig Mathematics: as acausal language, 42; internal limitations of, 113; as language that separates, 49; synthetic knowledge and, 110 Maxims, 142 Meaning: abstracted from practice, 27; contested, 97; differing, 97; of discourse, 15; emergence of, 15, 95, 96; as emergent property of complex systems, 98; generation of, 15; intentionality and, 104; interpretation of, 26, 99; literary, 95, 97, 98, 99; making, 8, 27; metaphors and, 95; misunderstood, 97; non-intentional, 97; as produced in plural, 2, 97; production of, 2, 96; proliferation of, 99; scientific, 99; shared, 97, 98; of texts, 96 Mechanics: classical, 71; Hamiltonian formulation, 71; Newtonian formulation, 71; quantum, 70, 82, 83, 85n8 Media, 94 Medicine, 54 Menger, Carl, 140 Mereological supervenience, 69, 84n7 Merleau-Ponty, Maurice, 27 Metaphor, 95 Metzinger, Thomas, 134 Mill, John Stuart, 140, 142 Mises, Ludwig von, 140 Modernity: destruction of natural communities by, 138; differentiation of spheres in, 5–6

Monism: idealist, 107; materialist, 107 Morality, 27; responsibility and, 51 Morgan, C. Lloyd, 60, 84n7 Natural selection, 65 Nature: boundaries in, 44; conceptualization of, 165; cultural constructs of, 21–22; definite, determinate physical properties in, 72; determinism and, 72; fixed evolution of, 72; inclusive/exclusive of humans, 20; intervention in, 18, 19; as product of culture, 19, 20; of reality, 107; subjectivity in concepts of, 3; transformations in, 20 Needs, 27, 28; empirical identification of, 28; essentialization of, 28 Neumann, John von, 172, 173 Neuroscience: cognitive, 94 Newton, Isaac, 53 Noë, Alva, 132 Nominalism, 139 Nonlinearity, 59–65; positive/negative feedback relationships in, 60 Non-separability, 2 Normativity, 30 Norms: duties and, 142; pragmatic efficacy of, 142; susceptibility to, 28; values and, 142 Objectivity, 29; confused with neutrality, 30; external, 131 Objects: constituting, 20 Observation: theory-laden/theorydetermined, 20–21 Occam’s razor, 108, 139

199

Ontology: of emergence, 34; flat, 2, 6, 8, 15, 17–22, 36n5, 41, 42, 43, 44, 46; non-reductionist stratified, 14; one-sided, 17; restrictive, 17; social, 2, 26–28; stratified, 2, 5, 7, 34, 43 Oppression, 25 Order: as consequence of coordination of individual agents, 141; natural, 138; rational constructivist, 138; sensory, 142; social, 138, 142; spontaneous, 138, 173 Organization: autopoietic, 79; boundaries and, 46, 47; in disciplines, 46, 47; of disciplines of knowledge, 2; hierarchical, 61; higher forms of, 137; as invariant, 92, 93; of nervous system, 64; self, 91, 140; system, 78, 79; temporal categories in, 47 Pareto, Vilfredo, 140 Parsimony, 12, 13 Perception: defining, 62; interdependency with institutionalized existence, 45; lived, 93; mediating influence of, 28; of objects, 132; phenomenology of, 111; processing stages in, 62; sensory, 62, 63; visual, 113, 122 Petitot, Jean, 3, 43, 48, 93, 100, 102, 103, 107–157, 160, 162, 163, 168, 182 Phenomena: boundaries in, 54, 55; collective, 139; critical, 109; critical emerging, 110; emerging, 48; higherlevel, 48; micro, 48; objective reality reconstructed from, 108; reduced to actions, 139; reduction of, 107; relevant,

200

100; social, 56, 138, 139; symmetrybreaking, 48 Phenomenalism, 5 Philippe II (King of Spain), 137 Phronesis, 32 Physicalism, 113; nonreductionist, 131 Physical Realization Principle, 77, 81, 82 Physics: local, 130; social, 139; statistical, 110, 125 Pinter, Harold, 98 Pinwheels, 121 Planforms, 115, 127, 127fig, 131 Platonism, 134 Pluralism, 137 Poiesis, 93 Poincaré, Henri, 131 Popper, Karl, 41, 139, 142 Positivism, 11, 30 Power: absolute, 137; academic, 9; causal, 11, 26, 68; competition for, 9, 10; differentials, 3; emergent, 14, 15; exercise of, 11; imbalances, 6; laws, 110; mechanical, 16; mobilization of, 53; predictive, 46; of social being, 51 Prigogine, Ilya, 185 Principle: of causal closure of the physical domain, 77, 81, 82; causal inheritance, 77, 81, 82; causal-power actuality, 74; physical realization, 77, 81, 82 Principle of the Causal Closure of the Physical Domain, 77, 81, 82 Prisoner’s Dilemma, 3, 143–145 Prisoner’s Dilemma, Iterated, 145–146, 161

Process(es): alteration of, 37n12; biological, 16; change in, 18; competitive, 141; contested, 97; cooperative, 141; cultural, 24; Darwinian, 136; distinguishing characteristics of, 16; emergent, 59; evolutionary, 104, 136; global-tolocal, 65–69; identification of, 16; intervention in, 17–22; mental, 94; nonlinear dynamics of, 59–65; politics in, 97; reproduction of structure over time, 96; social, 24, 97; temporal, 59 Production: of change, 10, 11, 12, 15, 25; of division of labor, 15; of knowledge, 20; of meaning, 2, 96; of subjects, 23; of subjects by discourse out of nothing, 23 Psychologism, 8–9 Psychology, 6 Racism, 97, 98, 184 Rationalism, 99 Rationality: background values in, 49; collective, 145; fallibility and, 31; individual, 144, 145; reason reduced to, 32 Rayleigh-Bénard convection rolls, 75 Realism, 2; asymmetry and, 55; holistic, 109, 138; primitive concepts of, 55 Realism, critical, 10–13; analysis of causation in, 10; causal powers and, 26; interpretative understanding of social phenomena and, 27, 28; rejection of actualism by, 11; understanding world through available discourse, 21

Reality: adjusting to, 104; complex, 47, 171; of consciousness, 14; different strata of, 5; emergent, 135; empirical, 184; existence of, 169, 170; levels of, 108; nature of, 107; objective, 108; phenomenally experienced, 135; reducibility of, 107; resistance of, 37n19; simplifications of, 169; social, 56, 139; structure of, 44; transcendent, 108; ways of knowing, 185 Reason, 29; defining, 38n27; devaluation of, 28–32; emotional, 31; nature of, 29; overlap with valuation, 31; reduced to rationality, 32 Reasoning, deductive, 9 Recognition, 91, 92 Reductionism: anti-essentialism and, 23–26; atomism and, 16; avoiding, 45; basic features v. emergent features, 57, 58; biological, 6; causal, 3, 48, 109, 160; class, 35n2; common forms, 22–34; of complex systems, 3, 107–157, 108; conceptual, 110; consciousness and, 111; critical realist view, 10–13; criticisms of, 40–41; defense of, 12; defining, 41, 107, 108, 169; disciplinary imperialism and, 8–10; downward, 1, 2, 6, 13–17, 35, 46, 48; eliminative, 108, 134, 160; emergentism and, 2, 57; epistemological, 14, 48, 57, 73; essentialism and, 25; in everyday thought and discourse, 7; explanationcentered, 51, 52; explanation of higher order phenomena, 13; explanations of multiply-determined processes

201

Reductionism (continued) and, 6; explanatory forms of, 36n8; fact-value dualism and, 2, 8, 28–32; failures of, 41, 51; form and direction of reduction in, 6; holist, 16, 17; horizontal, 6, 35n2; individualism and, 32–34; individualistic explanation and, 7; inseparability from converse concepts, 108; instrumentalist models of explanation in, 12; justification of, 12, 32–34; limits of nineteenth century approaches to, 1; linguistic and formal theories and, 107, 108; mathematical, 3, 111–115, 171; methodological, 108; micro, 71; need for, 51; objective, 108; ontological, 14, 57, 73, 108; oppositional, 7; parts v. whole in, 57, 58, 69–73, 80, 81, 83; as passé issue, 3; as pejorative, 6, 7, 12, 34; philosophical stance on ontology/metaphysics, 7; physical closure and, 93; physical examples, 110, 111; physicalist, 160; politics associated with, 7, 45–46; predictive forms of, 36n8; reconceptualizing meaning of, 43; regulative ideal of, 51; responsibility and, 32–34; salvaging, 168; scientific object-oriented, 108; seen less as problem than solution, 12; simplified versions of, 12, 168; social constructionism and, 23–26; in social science, 5–35, 7; status of, 1; subjectivist, 25, 30, 31; theories with empirical data and, 107; ubiquitousness of, 2; upward, 1, 2, 6, 14, 34, 46, 48, 54; vertical, 7, 13–17, 35n2

202

Reflexivity: mediating influence of, 28 Relationship(s): agency/fact-value divide, 2, 8; asymmetric, 22; causal reduction/theoretical autonomy, 109; change-producing, 15; cognitive, 28; consciousness/self, 22; culture/nature, 21–22; emergentism /reductionism, 2; to environment, 21; feedback, 60, 61; hierarchical, 18; intellectual/ organizational disciplinary boundaries, 2; mathematical models/causes, 42; moral, 53; nonlinear, 96; semantic referential, 95; social, 22, 53; stratified ontology/ organization of disciplines, 2; structural, 100; susceptibility of, 11; symmetric, 18; transformative, 12; values/facts, 51 Representation, 94, 95, 134, 160; spatial, 131 Responsibility: individual, 32–34; moral, 39n29, 51; reductionist treatment of, 8 Retinal images, 126–129 Rev, Istvan, 44, 49, 53, 92, 96, 101, 102, 180, 187 Richelieu, Armand Jean du Plessis de (Cardinal), 137 Rousseau, Jean Jacques, 173 Roy, Jean-Michel, 131 Saint-Simon, Claude Henri, 138 Sayer, Andrew, 1, 2, 5–35, 42, 44, 49, 51, 55, 99, 101, 102, 167, 173, 181, 188 Science: defining, 53; influence of chaos theory on, 98, 99; intellectual dominance of, 168; philosophy of, 57; split from philosophy, 168; unity of, 100

Science, cognitive, 3, 111–135; enactive approach to, 60; entoptic vision in, 115–117; mathematical reductionism in, 111–115; phenomenal consciousness in, 115–117; phenomenal experience in, 111–115; spontaneous emergence of geometric visual patterns in, 117– 129 Science, natural, 45, 48; disciplinary divisions, 35n1 Science, social, 1, 2, 45; alienation in, 32; causal explanation in, 10; complexity problem in, 135–137; cultural evolution and, 141–143; disciplinary divisions in, 35n1; emerging ethical maxims and, 141–143; evolutionary games and, 143–155; historical, 167; interpretive understanding in, 10; methodological individualism in, 138–141; purpose of, 42; rational justification of anti-constructivism in, 137–138; reductionism in, 5–35; subdivisions in, 5; unavailability of practical experiments in, 13; values in, 29 Searle, John, 68 Segerstrale, Ullica, 49, 99 Self: boundary of, 92; consciousness and, 22; determination, 60; identity, 61, 92; maintenance of, 91; organization, 61, 79, 91, 140; stability of, 92; transcendence, 173 Semiosis, 2, 15, 16, 22, 27; understanding of the world and, 15 Shaping, 18, 19; fallible, 23; reduced to unilateral process, 24 Slowness, 92

Smith, Adam, 38n26, 51, 101, 102, 103, 104, 173, 188 Social: behavior, 6, 9; being, 16, 51; circumstances, 32; cohesion, 140; construction, 2, 8, 23–26, 37n18, 54, 56, 187; constructivism, 136; contexts, 96; determinism, 33; entities, 138; environments, 7; exchanges, 136; groups, 139; influences, 24, 28; interaction, 19, 25; intervention, 99, 187; life, 51; ontology, 2, 26–28; order, 137–138, 138, 142; phenomena, 16, 27, 28, 56, 138, 139; planning, 187; pressure, 27; process, 24, 97; reality, 56, 139; reduction to hermeneutic, 26; reduction to the biological, 24; relations, 22, 24, 42, 53; rules, 135; structures, 102 Socialism, 139 Socialization: complex co-evolution in, 26; conditions of, 24; susceptibility to, 27 Space: as geometrical construct, 131, 164; as phenomenal form of intuition, 131, 164 Spencer, Herbert, 140 Sperry’s wheel, 77 Stability: change and, 110 Stratification, 13–17; anti-reductionist view of, 15; asymmetric relations in, 15, 16; consciousness and, 22; critical realist view, 10–13; denial of, 36n5; disciplinary, 5; emergence and, 14; flat ontological view of, 15; higher/lower, 16; ontology and, 2, 5, 7; ontology of, 34; of world, 10, 14

203

Structure: agency and, 2, 27; autonomous, 141; conflated with agency, 36n5; dualism of, 52; emerging collective, 141; macro self-organized, 108; multitude of agents in, 52; non-intentional, 141; ordering, 110; of power-bearing objects, 11; of reality, 44; as reducible to individual agency, 36n5; reproduction over time, 96; self-organizing, 3; social, 102; system, 103; varied, 92, 93; of the whole, 178 Subjectification, 25; susceptibility to, 27 Subjectivity, 30, 164; conceptualization of nature and, 3, 165; first person, 112; removal of, 165; spectator, 166; understanding, 166 Sun-Tzu, 188 Superdisciplines, 169 Supervenience, 100, 108 Symmetry, 48 Systems: aggregative, 59, 77; autocatalytic, 84n4; autonomous, 60, 61, 62, 84n3, 93; autopoietic, 67; causal effects on elements of, 68; changing states of, 2, 96; chaotic, 71; closed, 13; closure, 84n3; collective self-organization in, 61; decomposable/nondecomposable, 61, 62; dynamic, 96; emergence recognition by, 94; equilibrium, 61; errors, 60; failures of, 51; heteronomous, 60; hierarchical biological, 65; interactive, 94; of meaning, 99; metastable, 72; morphodynamical notions in, 61; nervous, 64; nonlinear, 57–83; open, 48, 96; operations, 84n3; organizational/operational closure in,

204

60, 61; organization of, 61; physical, 3; primitives in, 94; quantum mechanical, 70; relations among parts in, 57, 95, 96; reproduction of, 96; rule changing in, 103; self-defining, 60; self-determining, 60; self-monitoring, 91; self-organizing, 77, 79; selfregulating, 91; shared mind in, 90, 91; social, 96; stability in, 2; structure, 57; understanding, 57; as unity, 61 Systems, complex, 2; circular causality and, 65; as classical deterministic, 70; collective-variable dynamics in, 66; component behavior in, 68; downward causation in, 65, 68, 73–83; dynamic stability of, 79; emergence and, 2, 3, 57–83, 107–157; emergent global properties of, 71, 120; global-to-local influence, 65–69; holism and, 69–73; levels of reality in, 109; microstructural properties of, 79, 80; misunderstanding of notion of, 98, 99; randomness in, 72; reductionism in, 107–157; relational holism and, 69–73; robust, 99; slowness in, 92 Tarsky-Seidenberg theorem, 108 Terrorism, 33 Theory: actor network, 11, 36n3; assessment of, 9; of causation, 10; chaos, 71, 98, 99; complex systems, 57–83; conflicts of, 100; development of, 13; double-aspect, 134; economic, 9; of elimination, 108; equilibrium, 173; evolutionary, 100; evolutionary game, 135, 143–155; feminist, 29; of higher

order concepts, 108; priors in, 108; public choice, 9; regularity, 34; of selforganization, 135; self-transcendenceof behavioral rules, 142; of spontaneous order, 135 Thomas, Peter, 117 Thompson, Evan, 2, 43, 45, 57–83, 91, 93, 94, 95, 99, 100, 101, 105, 132, 164, 165, 166, 168, 169 Thought: collective wishful, 23; modernist, 30; positivist, 30 Totalitarianism, 141 Traditions, 141 Transformation: culturally construed, 20; of natural environment, 19, 20 Truth: denigration of possibility of, 37n19; instrumentalist rejection of concept of, 36n6; moral, 31 Universals: realist conceptions of, 139

Utilitarianism, 142 Values: as conventions, 30; derationalization of, 28–32; as irrational contaminant, 30; nature of, 29; norms and, 142; prescribing, 138; in rationality, 49; relationship to facts, 51; relation to nature of the world, 30; in social science, 29; as threat to objectivity, 30 Wallerstein, Immanuel, 3, 40–41, 46, 51, 52, 54, 55, 91, 92, 97, 102, 103, 104, 161, 167–171, 177, 179, 180, 181, 184, 188, 191 Walras, Léon, 140 Weber, Max, 38n21 Wertheimer, Max, 115 Wiener, Matthew, 117 Wilson-Cowan equations, 117 World-system, 91, 92

205

SOCIOLOGY / PHILOSOPHY

D

uring the last few decades, the fundamental premises of the modern view of knowledge have been increasingly called into question. Questioning Nineteenth-Century Assumptions about Knowledge, II: Reductionism provides an in-depth look at the debates surrounding the status of “reductionism” in the sciences, social sciences, and the humanities in detailed and wide-ranging discussions among experts from across the disciplines. Whether or not there is or should be a basic epistemological stance that is different in the sciences and humanities, and whether or not such a stance as exemplified by the approach to reductionism is changing, has enormous consequences for all aspects of knowledge production. Featured are an overview and subsequent discussion of this pervasive concept in the social sciences that parses reductionism into the categories of strong social constructionism and antiessentialism, social ontology and the apathetic actor, dualisms, and individualism. Also of interest in chapters and follow-up discussions are the relations between essentialism and emergentism in complex systems theory. RICHARD E. LEE is Professor of Sociology and Director of the Fernand Braudel Center at Binghamton University, State University of New York. He is the author of Life and Times of Cultural Studies: The Politics and Transformation of the Structures of Knowledgee and the coeditor (with Immanuel Wallerstein) of Overcoming the Two Cultures: Science versus the Humanities in the Modern World-System.

FERNAND BRAUDEL CENTER STUDIES IN HISTORICAL SOCIAL SCIENCE

4 5"5 &
6 / * 7 & 3 4 * 5 :
0 '
/ & 8
: 0 3 ,
1 3 & 4 4 ; ; ; 7 9 2 = 4 6 ) 7 7  ) ( 9