Comments on Donald T. Campbell's

Home , Cecilia Heyes, Francis Heylighen, Neuroepistemology, William James Lectures

Contents

W. Callebaut/R. Riedl 2 Preface

Donald T. Campbell 5 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity

Linnda R. Caporael 39 Vehicles of Knowledge: Artifacts and Social Groups

W. D. Christensen/C. A. Hooker 44 Selection Theory, Organization and the Development of Knowledge

Andy J. Clark 49 Evolutionary Epistemology and the Scientific Method

Ed Constant 55 Comments on Donald T. Campbell’s “From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity”

Steve Fuller 58 Campbell’s Failed Cultural Materialism

Francis Heylighen 63 Objective, Subjective, and Intersubjective Selectors of Knowledge

Aharon Kantorovich 68 The Rationality of Innovation and the Scientific Community as a Carrier of Knowledge

Elias L. Khalil 71 Can Artificial Selection Remedy the Failing of Natural Section with Regard to Scientific Validation?

Kyung-Man Kim 75 D. T. Campbell’s Social Epistemology of Science

Marc De Mey 81 Vision as Paradigm: From VTE to Cognitive Science

Erhard Oeser 85 The Two-stage Model of Evolutionary Epistemology

Henry Plotkin 89 Knowledge and Adapted Biological Structures

Massimo Stanzione 92 Campbell, Hayek and Kautsky on Societal Evolution

Franz M. Wuketits 98 Four (or Five?) Types of Evolutionary Epistemology

Evolution and Cognition: ISSN: 0938-2623 Published by: Konrad Lorenz Institut für Evolutions- und Kognitionsforschung, Adolf-Lorenz-Gasse 2, A-3422 Alten- Impressum berg/Donau. Tel.: 0043-2242-32390; Fax: 0043-2242-323904; e-mail: [email protected]; World Wide Web: http://www.kla.univie.ac.at/ Chairman: Ru- pert Riedl Managing Editor: Manfred Wimmer Layout: Alexander Riegler Aim and Scope: “Evolution and Cognition” is an interdisciplinary forum devoted to all aspects of research on cognition in animals and humans. The major emphasis of the journal is on evolutionary approaches to cognition, reflecting the fact that the cognitive capacities of organisms result from biological evolution. Empirical and theoretical work from both fields, evolutionary and cognitive science, is accepted, but particular attention is paid to interdisciplinary perspectives on the mutual relationship between evolutionary and cognitive processes. Submissions dealing with the significance of cognitive research for the theories of biological and sociocultural evolution are also welcome. “Evolution and Cognition” publishes both original papers and review articles. Period of Publication: Semi-annual Price: Annuals subscription rate (2 issues): ATS 500; DEM 70, US$ 50; SFr 60; GBP 25. Annual subscriptions are assumed to be continued automatically unless subscription orders are cancelled by written information. Single issue price: ATS 300; DEM 43; US$ 30; SFr 36; GBP 15 Publishing House: WUV-Universitätsverlag/Vienna University Press, Berggasse 5, A-1090 Wien, Tel.: 0043/1/3105356- 0, Fax: 0043/1/3197050 Bank: Erste österreichische Spar-Casse, Acct.No. 073-08191 (Bank Code 20111) Advertising: Vienna University Press, Berggasse 5, A-1090 Wien. Supported by Cultural Office of the City of Vienna and the Austrian Federal Ministry of Science, Research and Culture.

Werner Callebaut/Rupert Riedl

Preface

his issue of Evolution and Cognition is dedicated relativism, or, as he called it, “‘cult-solipsism’, por- T entirely to the memory of Donald Thomas traying sciences as self-deceiving social systems in- CAMPBELL (1916–1996), one of the most remarkable capable of distinguishing truth from tribal myth” scientists and humanists of his generation. (CAMP- (CAMPBELL 1979; cf. KIM, this issue). ERISS convinced BELL never pursued scientific aims for their own many STS scholars that a dialogue with ‘the enemy’ sake, but was always motivated by deeper humanis- remained possible and could even be profitable (cf. tic concerns. See, e.g., his Presidential Address to the RESTIVO in press). CAMPBELL’s descriptive epistemology, American Psychological Association [CAMPBELL as developed in his William JAMES Lectures at Har- 1975].) CAMPBELL ended his scholarly career as Uni- vard University in 1977, relies on physiological, psy- versity Professor at Lehigh University, Bethlehem, chological, and sociological approaches to deal with Pennsylvania, after having taught at the University epistemological issues, for it includes the (fallible) of Chicago (1950–53), Northwestern University theory of how the processes studied by these ap- (1953–79), and Syracuse University (1979–82). proaches “could produce truth or useful approxima- A social psychologist by training (he was a student tions to it”. Isn’t it somewhat ironic or paradoxical of Edward TOLMAN and Egon BRUNSWIK at Berkeley), that CAMPBELL-the-naturalistic-psychologist in- CAMPBELL’s uniquely creative blending of method- creasingly stood for a respectful conversation with ological rigor and daring theoretical speculation the epistemological tradition (including its skepti- about the individual and collective human mind (cf. cist offshoot, which he deeply admired) at a time De MEY, this volume) was to make him highly influ- when many a philosopher (including such influen- ential in many other fields as well. Many psycholo- tial evolutionary epistemologists as Ronald GIERE or gists and social scientists know him best for his David HULL), following the QUINEAN suggestion that contributions to methodology, which include mea- psychology and sociology should replace epistemol- surement issues (e.g., convergent and discriminant ogy, surrendered to “nihilistic naturalism” (Susan validation by the multitrait-multimethod matrix), HAACK), a position which CAMPBELL himself more experimental design (most notably, the method of graciously described as “epistemologically vacu- quasi-experimentation [COOK/CAMPBELL 1979]), and ous”? evaluation research. CAMPBELL for all his modesty (CAMPBELL 1981) Biologists and philosophers tend to regard him as could do all this, and more, because he was a true the major American evolutionary epistemologist of encyclopedist in Otto NEURATH’s sense, one of a our century. Scholars in the booming but divided handful of genuine generalists of the post-World new field of STS or Science and Technology Studies War II generation who transformed cybernetics into remember him most as the grand old man who sym- General Systems Theory (cf. HEYLIGHEN, this vol- pathized with all of the warring factions: philoso- ume). Contrary to a common perception, CAMP- phers speaking in the name of rationality and BELL’s EE is not neo-DARWINIAN in the dominant, correspondence truth, relativistic and constructivist reductionist sense of the orthodoxy promoted by, sociologists doing their thing, cognitive scientists say, George C. WILLIAMS or DAWKINSIAN sociobiol- painfully turning naturalistic, verstehende histori- ogy: Not only did he consider a hierarchy of levels on ans…His personal contribution to this ongoing de- which DARWINIAN mechanisms operate (which al- bate, which in his inimitable manner he dubbed lowed him to highlight the phenomenon of “vicar- ERISS, for “Epistemologically Relevant Internalist ious selection”); he also and increasingly took into Sociology of Science” (cf. his later “sociology of sci- account internal (“structural”) selection factors and entific validity”), was an attempt to reidentify the mechanisms. “something special about science, which gives it The concept of a nested hierarchy of vicarious selec- some greater legitimate claim to objectivity than tion processes, which was in part inspired by systems- other social systems” in the light of post-KUHNIAN theoretical considerations (cf. CAMPBELL 1973), may

Evolution and Cognition ❘ 2 ❘ 1997, Vol. 3, No. 1

Preface

well be CAMPBELL’s single most important contribu- learning select behaviors on behalf of natural se- tion to EE and evolutionary theory more generally lection. (RICHERSON/BOYD in press) (see, e.g., CONSTANT, DE MEY, and KIM, this issue). CAMPBELL’s insistence on the “blindness” of varia- RICHERSON/BOYD (in press) pinpoint its meaning and tion (and retention, but that is a different matter) relevance against the background of the inadequacy has generated quite some misunderstanding. Most of DARWIN’s theory of inheritance in The Descent of importantly perhaps, is that this has prevented peo- Man (1871). As a consistent naturalist, philosophi- ple in the Artificial Intelligence community from cally speaking (see, e.g., CAMPBELL 1988), DARWIN appreciating the useful complementarity of CAMP- took great care to suggest continuity between humans BELL’s views and those of that other great contempo- and other animals, which he secured by attaching rary polymath, Herbert SIMON, on levels ontology, great importance to imitation and other forms of the adaptive (“satisficing”) behavior, and reduction of inheritance of acquired variation: complexity in general (THAGARD 1988 is a case in His theory allowed him to account for the essen- point). Yet as CONSTANT (this issue) points out, the tial similarity of all living humans, while account- point CAMPBELL wanted to drive home may be for- ing for the vast diversity in human behavior, by mulated as a simple reductio ad absurdum. attributing the underlying similarities to conser- The U.S. and Austrian varieties of Evolutionary vative traits and by attributing variation between Epistemology (EE) turn out, then, to share some of human groups mostly to labile traits strongly in- their intellectual ancestry, for BRUNSWIK (who fluenced by inherited habits. DARWIN’s distinc- coined the term “ratiomorphic apparatus”) influ- tion between more conservative and more labile enced Konrad Lorenz and Karl POPPER as well, and a traits did the same work for him that the modern major feature of the ‘Vienna–Altenberg’ conception gene-culture distinction does for us. (RICHERSON/ of evolution is its systems-theoretical orientation (cf. BOYD in press) LUHMANN 1997, 1, p431). In a way it is sad that the One problem with DARWIN’s view is that while rapprochement which the publication of CAMPBELL’s human culture operates as a ‘use inheritance’ sys- paper, “From Evolutionary Epistemology Via Selec- tem, such a (‘LAMARCKIAN’) system is rudimentary or tion Theory to a Sociology of Scientific Validity” in lacking in animals without culture. DARWIN thus left this issue consolidates is a posthumous one, but such a major gap in evolutionary theory that contempo- is the vagary of biological and intellectual existence. rary workers, and CAMPBELL in particular, tried to A first draft of the comprehensive paper Evolution fill. RICHERSON and BOYD argue that the idea of vicar- and Cognition is honored to publish here was prepared ious selection is crucial in this respect: by Don CAMPBELL for an invited talk at one of the sem- How could it be that animals could acquire adap- inars on “The Evolution of Knowledge and Invention” tive variations that anticipate what natural selec- of John ZIMAN’s Epistemology Group, on 24 May tion would favor? DARWIN certainly believed that 1995. He was so kind to prepare a second version for humans and animals could acquire adaptive vari- this journal early in 1996. The additional time this ations, but gave no clue as to how this neat trick most conscientious and generous of men had kindly could itself evolve. CAMPBELL noted that if varia- required to complete work on the paper was not tions are acquired other than randomly, it must granted him by Life as he suddenly died on 6 May be because organisms have the capacity to use 1996. We are grateful to Barbara FRANKEL, his widow some sort of rules to modify behaviors or struc- and a sociologist herself, and to his favorite student, tures. Indeed, plants and animals have many such Celia HEYES, a psychologist, who kindly accepted to capacities, the most familiar of which is ordinary edit the manuscript so as to give it a shape they and trial and error learning, itself an we hope CAMPBELL would example of blind variation and have found suitable. More selective retention in CAMP- Authors’ address detailed information BELL’s view. Natural selection about the editorial process, has arranged sensation of re- Werner Callebaut, Department of Philoso- and a summary of the pa- ward and punishment so that phy, Maastricht University, P.O. Box 616, per—which CAMPBELL learning normally favors be- 6200 MD Maastricht, the Netherlands. characterized as “a person- haviors that are useful to sur- Email: callebaut@philosophy.unimaas.nl alized retrospective history Rupert Riedl, Konrad Lorenz–Institute for vive and reproduce. CAMPBELL of ideas”—are provided in Evolution and Cognition Research, A-3422 termed such processes “vicari- HEYES’ and FRANKEL’s Edi- Altenberg/Donau, Austria. ous selection”; the rules of tors’ Introduction.

Evolution and Cognition ❘ 3 ❘ 1997, Vol. 3, No. 1 Werner Callebaut/Rupert Riedl

A number of CAMPBELL’s younger colleagues and ments speak for themselves; there is no need to friends have accepted to write brief contributions summarize them here. Although several of these com- highlighting one or another aspect of CAMPBELL’s EE ments are quite critical of one or another aspect of sensu lato, as developed in this ultimate paper of his. CAMPBELL’s stance, they all reflect a deep sympathy for Their disciplinary backgrounds are as diverse as social the man and his work. We like to think of this not as anthropology (Linnda CAPORAEL), cognitive science a closed project, but as ongoing work. CAMPBELL liked (Andy CLARK, Marc DE M EY), complexity theory (Fran- to challenge us, and many of the fascinating ideas in cis HEYLIGHEN), economics (Elias KHALIL), philosophy his rich paper could not be dealt with in sufficient of science (W. D. CHRISTENSEN and Cliff HOOKER, Aha- depth at this occasion. But there will be other occa- ron KANTOROVICH, Erhard OESER, Massimo STANZIONE, sions: special conferences, books, journal issues, etc., Franz WUKETITS), sociology and social policy (Steve devoted to his oeuvre, which will continue to inspire FULLER), history of technology (Ed CONSTANT), and ex- many of us. perimental psychology (Henry PLOTKIN). These state- Don CAMPBELL was such a kind and brilliant man.

References Social Science: Selected Papers. Overman, E. S. (ed). Uni- versity of Chicago Press, Chicago. Campbell, D. T. (1973) Ostensive instances and entitativity Cook, T. D./Campbell, D. T. (1979) Quasi-Experimentation: in language learning. In: Gray, W./Rizzo, N. D. (eds) Unity Design and Analysis for Field Settings. Rand McNally, Chi- Through Diversity, Vol. 2. Gordon and Breach, New York, cago. pp. 1043–1057. Darwin, C. (1871) The Descent of Man and Selection in Rela- Campbell, D. T. (1975) On the conflicts between biological tion to Sex. John Murray, London. and social evolution and between psychology and moral Luhmann, N. (1997) Die Gesellschaft der Gesellschaft (2 tradition. American Psychologist 30, pp. 1103–1136. Vols.) Suhrkamp, Frankfurt am Main. Campbell, D. T. (1979) For vigorously teaching the unique Restivo, S. (in press) In the wake of the winnower: Donald T. norms of science: An advocacy based on a tribal model of Campbell and the sociology of objectivity. Philosophica scientific communities. In: Callebaut, W./De Mey, M./ 58. Pinxten, R./Vandamme, F. (eds) Theory of Knowledge and Richerson, P. J./Boyd, R. (in press) Built for speed, not com- Science Policy, vol. 1. Communication and Cognition, fort: Darwinian theory and human culture. Philosophica Ghent, pp. 50–69. 58. Campbell, D. T. (1981) Perspective on a scholarly career. In Thagard, P. (1988) Computational Philosophy of Science. Brewer, M.B./Collins B.E. (eds) Scientific Inquiry and the Bradford Books, MIT Press, Cambridge, Mass./London. Social Sciences: A Volume in Honor of Donald T. Camp- Williams, G. C. (1992) Natural Selection: Domains, Levels, bell. Jossey-Bass, San Francisco, pp. 454–501. Reprinted in and Challenges. Oxford University Press, New York/Ox- Campbell (1988). ford. Campbell, D. T. (1988) Methodology and Epistemology for

Evolution and Cognition ❘ 4 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell

From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity

Edited by Cecilia Heyes and Barbara Frankel

Editors’ introduction ‘knowledge’ should be reserved for the products of vicarious selection processes, such as perception, trial-and- This paper originated in a set of background readings error learning, and scientific enquiry, that “short-cut se- assembled by Donald T. CAMPBELL (DTC) for a talk at a lection by the life and death of genetic variants”. meeting entitled “The Epistemology Group: The Evolu- Sections 3–6 examine the implications of general se- tion of Knowledge and Invention”. The meeting was lection theory for the status of scientific knowledge and organized by John ZIMAN, and took place on 24 May justification. Throughout, DTC emphasizes that the re- 1995, at The Royal Society of Arts, London. The editors alization that science proceeds through selection proof Evolution and Cognition encouraged DTC to develop cesses does not provide justification for scientific beliefs the manuscript for publication, and in his notes on the or theories, and, in this sense, general selection theory is version completed before his death, DTC described it as “epistemologically irrelevant”. He underlines this scepti- “in a halfway state of editing”, and as “a personalized cal point in section 3 by noting that religious, as well as retrospective history of ideas”. He also wrote that “The scientific, beliefs can be selection-based, and in section 4 main goal of the paper is to provide a naturalistic episte- by drawing attention to the many selection processes in mology for science. It turns out that this will have to also science that are antagonistic to “competence of refer- be a sociology of scientific validity. Fragmentary sugges- ence”, i.e. correspondence truth of theories. Primary tions along this line take up much of the space.” DTC among these are the processes required to sustain science pursued this goal first by identifying the type of evolu- as a social system, as a “vehicle” or embodiment of tionary epistemology in which he was engaged as ‘gen- knowledge. In sections 5 and 6, DTC continues to “side eral selection theory’ (section 1). Rather than focussing with the sceptics” by stressing the ubiquity and inelim- on the biological evolutionary origins of categories of per- inability of the problem of induction, and the oversimpli- ception and thought, exploring detailed analogies fication entailed by the use of language. between biological evolution and scientific change, or In the spirit of DTC’s hypothetical realism, there is a examining the role of innate perceptual mechanisms in switch of perspectives between sections 6 and 7. Having language acquisition, general selection theory asserts insisted in the first half of the text that general selection that variation–and–selective–retention processes operate theory cannot justify scientific beliefs, and therefore that at a number of hierarchically organized loci or domains realism about those beliefs is necessarily hypothetical, or in nature, and that all improvements in ‘fit’ between sys- a matter or faith, DTC offers in the latter part of the text tems and their environments are attributable to the oper- a selectionist rationale for greater trust in scientific than ation of these selection processes. in religious beliefs (section 7), and argues that public Section 2 provides further preliminary clarification. commitment to the norms of science, even if it is hypo- Therein DTC announces that he now believes it counter- critical, functions to increase competence of reference productive to describe adaptive organic form, resulting (section 8). Thus, he shows that, if one assumes what from genetic selection, as ‘knowledge’. Instead, the term cannot be demonstrated (using selection theory or by any

Evolution and Cognition ❘ 5 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell other means), that science is (sometimes) successful in 1. Four types of ‘Evolutionary describing nature, selection theory can explain how this Epistemology’ success could be achieved, and provide the basis for recommendations about how science should be organized to Its length notwithstanding, the present essay cov- maintain or enhance the (presumed) validity of the be- ers but a small portion of all that goes on under the liefs it generates. In other words, those used in the title of ‘evolutionary epistemology’ rubric. (For bibliogra- the paper: one type of evolutionary epistemology, selec- phies and surveys, see CALLEBAUT 1993, CAMPBELL & tion theory, yields a descriptive epistemology for science, CZIKO 1990, CAMPBELL, HEYES/CALLEBAUT 1987.) and, more specifically, a sociology of scientific validity. In editing this manuscript, we have sought merely to 1.1 Evolutionary origins of Kant’s a priori clarify the import of the original. To maintain the mo- categories mentum of the text, we have inserted passages which try to impersonate DTC’s style, rather than interrupting in Most completely in CAMPBELL (1974b), I offer docu- our own editorial voices. However, all modifications and mentation on the many independent discoveries additions are printed in italics (as used here), and are of the notion that the a priori categories of percep- therefore clearly identifiable. tion and intuition are the products of biological The first editor (HEYES) performed the exacting task of evolution. The insight goes back to DARWIN’s note- cleaning up a scanner-generated computer disk, since the books, Herbert SPENCER, William JAMES, and a hun- original disk could not be found after DTC’s death. She dred others. I am continually finding new then made a very able first pass at smoothing out rough predecessors previously missed, such as Hans transitions, clarifying sections that presumed too much VAIHINGER (1911), but modern attention to it has familiarity with DTC’s ideas on the part of the reader, been primarily stimulated by the writings of Kon- and generally improving without abbreviating a very long rad LORENZ (1941, 1951, 1973). I first became aware manuscript. It was in that form that the paper was sent of LORENZ’s essays through WHYTE (1951) and BER- to the other authors whose papers are included in this TALANFFY (1955). I initiated and edited the widely special issue of Evolution and Cognition. reprinted English translation of LORENZ’s 1951 The second editor (FRANKEL) took a rather less reverent essay on KANT’s categories, first published in 1962. approach, and thus allowed herself the prerogative of The recent vigorous developments in Austria and overriding, to a degree, DTC’s long-standing habit of can- Germany are primarily of this type (e.g., RIEDL nibalizing his own earlier work freely, thereby saving 1982, 1984, VOLLMER 1975, 1985, 1986, ENGELS himself the labor of rephrasing things he had said to his 1989) and this is the only type of evolutionary own satisfaction before. In his lifetime, Don CAMPBELL’s epistemology endorsed by such naturalistic episte- lengthy self-quotes were a standing joke between hus- mologists as QUINE and SHIMONY. In a widely used band and wife, and now that he can no longer defend typology, BRADIE (1986) designates this as EEM himself she has taken shameless editorial advantage. (Evolutionary Epistemology of Mechanisms). It is Some very long quotes have been therefore abbreviated neglected in this essay. (using ellipses in the body of the text) and/or partially reduced to briefer summaries. The other authors in this 1.2 Analogies between biological evolution and volume will recognize that the paper has been shortened the ‘evolution’ of scientiﬁc theories since they read it, however, every effort has been made to preserve all parts of the argument in the original. This type is central to the present essay. POPPER Equally important, the changes are not so extensive as devotes a paragraph or so to it (1959 and presum- to make Don CAMPBELL’s final publication anything ably 1935) and it is central to the more recent works other than his own work, in his own inimitable voice— of TOULMIN (1967, 1972, 1981), HULL (1988a,b) and or anything less than the persuasive document, the ser- RICHARDS (1987). BRADIE (1986) called it EET (Evolu- mon to the unconverted that he intended it to be. DTC’s tionary Epistemology of Theories). lifelong dedication to science and his faith in the good that it ultimately produces were central themes of his 1.3 Shared innate perceptual reiﬁcation of middle- career. Although he realized that all human beings and sized objects all human enterprises are imperfect, his own achievements stand as testimony to the intellectual and moral This type accents a link between 1.1 and 1.2. It integrity that can be aspired to by members of the scien- focuses on the phenomenon that makes language tific community in which he so delighted. and the bulk of culture possible. It is because

Evolution and Cognition ❘ 6 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity infants and their parents share unjustified but use- 2. My 1960 model and a proposed ful perceptual reification of external objects that a revision shared language can be transmitted through ostensions that are always to some degree equivocal (see 2.1 The 1960 dogma 6 below, CAMPBELL/PALLER 1989, and CAMPBELL 1973). Also, in the ideology of the scientific revolu- The following is rearranged from an early essay on what tion, ‘proof’ calls for ‘demonstration’ using the has come to be called “evolutionary epistemology”—or, same level of ostensions usable in language learn- more aptly, “naturalistic epistemology”—based upon a ing. selection-and-retention model of knowledge processes. Between a modern experimental physicist and 1.4 General selection theory some virus-type ancestor there has been a tremen- dous gain of knowledge about the environment Most of those who have elaborated on 1.2 above … This extended usage of ‘knowledge’ is a part of have employed too close an analogy between sci- an effort to put ‘the problem of knowledge’ into a ence and biological evolution, carrying over many behavioristic framework which takes full cogni- details from the latter that are inappropriate (see zance of man’s status as a biological product of an section 3 below). This essay approaches an episte- evolutionary development from a highly limited mology for science from the perspective of a much background, with no “direct” dispensations of more general ‘selection theory’ in which biological knowledge being added at any point in the family evolution is just one nested cluster of exemplars. tree. The bibliographical citation of the several This biological cluster is, of course, to be mined for sources converging on this approach to the prob- useful insights and analogies, but is not to be taken lem of knowledge, and the discussion of its rela- as a compulsory model to be followed in every tion to traditional philosophical issues and to the detail. Among the exemplars of general selection strategy of science are presented elsewhere (CAMP- theory is the JERNE–BURNET theory of acquired BELL, l959). [See also CAMPBELL 1974b.] Suffice it to immunity. say here that the position limits one to ‘an episte- According to this theory, the presence of a toxin stim- mology of the other one.’ The ‘primitives’ of ulates a proliferation of potential antigens, some of knowledge can not be sought in ‘raw feels’ or in which by chance immobilize the toxin molecules ‘phenomenal givens’, or in any ‘incorrigible’ ele- and are triggered into mass production. This meta- ments. While man’s conscious knowledge pro- phorical base has stimulated GAZZANIGA’s (1992) be- cesses are recognized as more complex and subtle lated independent invention of evolutionary than those of lower organisms, they are not taken epistemology, among others. Trial-and-error learn- as more fundamental or primitive. In this perspec- ing (of a blind animal for purists like me – see CAMP- tive, any process providing a stored program for BELL 1956) was Karl POPPER’s first metaphorical base. organismic adaptation in external environments THORNDIKE (1898), ASHBY (1952), and PRINGLE (1951) is included as a knowledge process, and any gain were among those who have called attention to the in the adequacy of such a program is regarded as common selection theory model shared by trial- a gain in knowledge. If the reader prefers, he can and-error learning and natural selection. Early evo- understand the paper adequately by regarding the lutionary epistemologists such as SIMMEL (1895) and term ‘knowledge’ as metaphorical when applied BALDWIN (1909) have called this ‘selection theory’, to the lower levels in the developmental hierar- and I should have done so too, instead of using ‘evo- chy. But since the problem of knowledge has re- lutionary epistemology’. That ‘selection theory’ sisted any generally accepted solution when would have been a better term is shown by the titles defined in terms of the conscious contents of the of several of my historical bibliographic lists in the philosopher himself, little seems lost and possibly appendices to CAMPBELL (1974b): “Appendix I: something gained by thus extending the range of Trial–Error and Natural–Selection Models for Cre- processes considered. ative Thought”, “Appendix II: Natural Selection as In bulk, [the knowledge gained between the vi- a Model for the Evolution of Science”, and “Appen- rus-type ancestor and the physicist] has represent- dix III: On the Ubiquity of Multiple Independent ed cumulated inductive achievements, stage by Invention.” CZIKO (1995) provides an extensive sur- stage expansions of knowledge beyond what vey of the many exemplars of a general selection could have been deductively derived from what theory. had been previously known. It has represented

Evolution and Cognition ❘ 7 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell

repeated ‘breakouts’ from the limits of available contact with the traditions of philosophical episte- wisdom, for if such expansions had represented mology. only wise anticipations, they would have been ex- Instead, I now wish to identify ‘knowledge’ with ploiting full or partial knowledge already point 2.1.2 of The Basic Selectionist Dogma, i.e., achieved. Instead, real gains must have been the with those ‘vicarious’ processes which short-cut se- products of explorations going beyond the limits lection by the life and death of genetic variants. Vi- of foresight or prescience, and in this sense blind. sual perception is the most important of these. Two In the instances of such real gains, the successful levels of creative thought are others, as are linguistic explorations were in origin as blind as those transfer of belief, and the improvement of belief in which failed. The difference between the success- dialogue. ful and unsuccessful was due to the nature of the This revision still includes as knowledge processes environment encountered, representing discov- some strange items from a philosopher’s point of ered wisdom about that environment. view: Blind trial-and-error exploration (discovery The general model for such inductive gains is and/or learning on the part of a blind person) is in- that underlying both trial-and-error problem cluded. The resulting ‘knowledge’ is so similar to solving and natural selection in evolution, the that provided by vision that it would be remarkable analogy between which has been noted by several if no philosopher had attended to it. (I await help persons (e.g., ASHBY 1952, BALDWIN 1900, 1909, from historians of epistemology.) Echolocation by PRINGLE l95l). Three conditions are necessary: a blind humans—dull sense that it is—is included, as mechanism for introducing variation, a consis- well as exploration of local space with a cane. tent selection process, and a mechanism for pre- Echolocation by radar and sonar are also included, serving and reproducing the selected variations. seen as mechanical protheses for humans. Visual perception by non-humans, and echolocation by The Basic Selectionist Dogma bats and cave birds (NAGEL 1974) also count as knowledge processes under my revised view. (Un- doubtedly, the reflections of supersonic squeaks are 2.1.1. A blind-variation-and-selective-retention displayed on a brain map previously evolved for vi- process is fundamental to all inductive achieve- sion in protobat ancestors.) ments, to all genuine increases in knowledge, to all increases in fit of system to environment. 3. The focal contrast: Analogies 2.1.2. The many processes which shortcut a more between biological evolution and the full blind-variation– and-selective-retention pro- evolution of scientific belief do not help cess are in themselves inductive achievements, ‘justify’ scientific ‘knowledge’ containing wisdom about the environment The many efforts to model the ‘evolution’ of scien- achieved originally by blind variation and selec- tific knowledge on the principles of biological evo- tive retention. lution (e.g., TOULMIN 1967, 1972, RICHARDS 1981, 1987, HULL 1982, 1983, 1988a, 1988b) I now judge 2.1.3. In addition, such shortcut processes contain to be epistemologically irrelevant. MAYNARD-SMITH in their own operation a blind-variation-and-se- (1988) has made this point in his conspicuous lective-retention process at some level, substitut- review of HULL’s (1988b) famous book. Referring to ing for overt locomotor exploration or the life- cooperation among scientists, he says and– death winnowing of organic evolution. [HULL’s] explanation for such cooperation is that [Rearranged from CAMPBELL 1960, pp380–381] the replicators (genes) in the cells are identical... in different members of the group and will be 2.2 A 1995 modification transmitted to future generations only insofar as the group as a whole... is successful. Now an anal- In the above (and in CAMPBELL 1959 and 1974b), ogous argument might explain the loyal coopera- adaptive organic form is treated as ‘knowledge’ (e.g., tion of members of a tightly knit research group, above, “Any process providing a stored program for but would equally well explain the cooperation of organismic adaptation in external environments is the members of a religious sect or of a group included as a knowledge process.”) This broad inclu- bound together by a common political or artistic sion I now reject. It is a needless obstacle in making program. [MAYNARD-SMITH 1988, p1182]

Evolution and Cognition ❘ 8 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity

HULL (1988b) does provide many important insights would not have given us untrustworthy eyes.” But as to how the social system of science leads opportu- even if they noted that the social system of science nistic and egotistical scientists to report their find- requires great (albeit selective) trust of fellow sci- ings honestly (and why falsifying data is a much entists, neither the old providentialist nor the more major sin than plagiarism). But these contribu- evolutionary epistemologist would find it plausi- tions to a future ‘Sociology of Scientific Validity’ do ble to argue that “(God) (biological Natural Selec- “not“ come from his analogies to biological evolution) would not have given us untrustworthy tion. Independently, and with more privacy (CAMP- fellow scientists.” If we can in fact often validly BELL 1988b), I made the same point as did MAYNARD- trust fellow scientists, this is because of culturally SMITH. My own insight came from the fact that the evolved and fragile social systems, not because of variation-selection-reproduction-speciation-geneal- innate honesty and objectivity. [CAMPBELL 1987b, ogy model applied equally well to the Appalachian 151–152] Bible-belt free churches in which my grandfather, uncle, and cousins participated. TOULMIN (1981) 4. Plausible co-selection of belief by inadvertently made the same mistake in an evolutionary epistemology advocacy which used the spe- referent ciation of the Romance languages as an exemplar. In no way is Portuguese adapted to Portugal. Romanian 4.1 Justiﬁcation of visual percepts would serve just as well (if expanded with a dozen or so loan-words for use in cork gathering). As I now see it, the beliefs about normal middle- Here are more limited ways of making the same sized objects and events which vision produces are point, i.e., that evolutionary models of science are ‘justified’ by two separate plausibilities. epistemologically irrelevant. … the stance of the modern biological evolutionary epistemologist 4.1.1 Eyes are adaptive. If the theory of biological can be epitomized: “Natural Selection would not evolution is approximately correct, then eyes are have left us with eyes that regularly mislead us.” adaptive in the normal ecology within which they Thus, reference to natural selection can be used to evolved. Eyes that produced dangerously misleading ‘justify’ visually supported beliefs in the formula beliefs would have been weeded out. (After all, as ‘Knowledge is justified true belief,’ in the weak- SIMMEL 1895, pointed out long ago, in evolutionary ened interpretation of ‘justification’ used by all epistemology, ‘true’ and ‘useful’ become con- modern epistemologists except skeptics. founded.) This makes contact with Alvin GOLD- This general program of evolutionary episte- MAN’s, e.g., 1986, “reliability” theory of justification. mology (or of providentialism) can only with great difficulty, if at all, be extended to the social 4.1.2 Co-selection by the objects of belief. It must processes producing scientific belief. I will give also be plausible that the objects and events of the two brief epitomes of the problem. In the evolu- current belief-independent environment have been tionary epistemology program, any ‘validity;’ or co-selectors. (This is a version of Alvin GOLDMAN’s usefully competent reference, is attributed to the 1967, “causal theory of belief”. GOLDMAN 1986, pro- selection processes which weed out, rather than vides an integration of reliability and causal theory.) to the competence of the generation processes The belief in the selective reflection of the emitted producing the variations. We know of so many radar beam makes it plausible that an airplane is over selection processes in the generation, publication, there (although a compact flock of birds might gen- teaching, and believing of scientific truth claims erate a similar reflection). Detailed pattern in the re- that are irrelevant or inimical to improving the flection increases the plausibility (CAMPBELL 1966), competent reference of beliefs that it becomes but pattern that can be plausibly explained by the hard to argue for a dominant role for ‘Nature Her- characteristics of the radar machinery (rather than self’ in the selecting. This is in contrast to the case the presumed referent) reduces the experienced va- we can make for Her role in the biological evolu- lidity (CAMPBELL 1992 ms). tion of the eye and brain. A reflexive use of biological evolutionary theo- 4.2 Only co-selected ry provides a complementary perspective. Both CARTESIAN and evolutionary providentialists Here I wish to make the point that, even when the refer- could plausibly say “(God) (Natural Selection) ents of beliefs play a part in their selection, beliefs are

Evolution and Cognition ❘ 9 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell also selected by processes that could be expected to de- 4.2.2 Vehicular requirements for embodied tract from their validity. This is a point I wish to be- knowledge–general. A different tack on “only co- labor from several perspectives. In 4.2.1 I use a non- selected” comes from my 1987 participation in a sym- epistemological example illustrating the way that or- posium on naturalistic epistemology: ganic form is selected, not only by the environment in Even though naturalistic epistemology as a move- which it must function, but by history or ‘historicity’, ment announces the relevance of the anatomy the forms from which it descends. In 4.2.2, I argue that, and evolutionary biology of eye and brain, most far from being confined to biological evolution, co-se- of its discussion, including much of my own lection is an inevitable feature of embodied knowledge, (1959, 1966, 1974a, 1974b), employs philosophi- all knowledge recognized by a naturalist. The embodi- cal concepts and vocabulary. In contrast, the ment, the physical vehicle of knowledge – whether it present essay attempts to keep to a language of consists of stone fragments, nervous tissue, or, in the physical substances, placing ‘knowing’ in a frame- case of science, a social system (4.2.3)—is itself a co- work of material things and systems, of physical selector. Bear with me or scan rapidly. objects and processes. Pure epistemology may often deal quite profitably with disembodied, unre- 4.2.1 Co-selection by interests and history. It is alized, and abstract belief and knowledge. Not so, plausible that beliefs ‘fit’ the environment because however, for the descriptive epistemology I at- that environment has selected them. But that envi- tempt. Instead, the knowledge it studies will be ronmental selection, leading to ‘competence of ref- physically embodied in some substance, some ve- erence’ (a.k.a. validity, truth), will have been only hicle or carrier. This vehicle will have its own one of “many” selective forces, one of many ‘co-se- physical nature and limitations. lectors’. ‘Competence of reference’ selection will Let us make this more vivid by considering a have been only one of several kinds of selection, al- mosaic mural done in stone fragments and pictur- though in biological and cultural evolution there ing a street scene, as an example of embodied may have been selection to maximize such selec- knowledge of the street, buildings, and persons tion, and to render minimal (or as stable back- depicted. The size of the stones, the thickness and ground for contrast) the most dominant of the color of the cement, the range of natural colors other co-selectors. available, the restriction to a two-dimensional Let me offer a non-epistemological example. surface, the required rigidity, etc., all contribute The asymptotic true and complete laws of hydro- to the substantialized belief or knowledge that is dynamics have been co-selectors of the shape of carried, all become a part of the picture, reducing the killer whale and the shark (the killer whale and its validity from any ideal of perfection, were such bear are closer relatives), and are ‘in part’ responsi- a conceptualisation feasible. The end product, ble for their similarity in shape, muscles, and fins. knowledge, at its realized best, is some compro- Other co-selectors making for similarity in shape mise of vehicular characteristics and of referent have included their shared interests in predatory attributes. Where validity is our goal we of course speed, and the prior adaptations which each was minimize the vehicular contribution as much as modifying (historicity). The shape and muscula- possible, as by using smaller and smaller pieces of ture of the squid has also been co-selected by the stone, and cements that are thinner and more true laws of hydrodynamics, but co-selected from transparent. But we can never completely elimi- different prior adaptations (historicity) and per- nate vehicular restriction and bias for embodied haps purposes. The mathematical models and con- knowledge. This also holds true for retinas made textual language of the physics of hydrodynamics of rods and cones, for nerve cells, brains, memory have also been selected by these same true laws. processes, visual perceptions, innate reflexes, The latter have been co-selected by the traditions stimulus-response associations, thought and cog- of paper and pencil mathematics, two-dimen- nitive structure. sional graphics, and historicity. In no sense is Without having done the logical analyses that mathematical hydrodynamics a pure ‘representa- might make them compelling, I have leaped to tion’ independent of the products of other co-se- some general principles that will guide my explo- lectors. (Nor is it independent of selection by use, rations: the vehicular substance that carries but it may have achieved greater multi-use possi- knowledge is unavoidably alien to the referents of bilities.) knowledge-–it is a different substance with different structural characteristics. Complete flexibility

Evolution and Cognition ❘ 10 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity in depiction, reflection, transmission, or record- the mathematical formulae of modern physics, as ing, is precluded by the structural requirements of prime exemplars of embodied knowing. I recog- the vehicle. If the vehicle is completely flexible it nize that most epistemologists, descriptive or oth- lacks the rigidity to hold together the picture it erwise, will profitably stay within these bounds. carries. These vehicular-structure requirements But I also feel that it may be useful for some of us produce not only restrictions on fineness of detail, to try placing these prime exemplars in radically but also bias and limitations of aspect. Keeping different conceptual frameworks. I have started the vehicle intact becomes a requirement in rival- such an exploration in adding the strange exam- ry with the requirement of validly mapping the ples of mosaic murals and plaster-of-paris cast- referent. ings. [CAMPBELL 1987a, pp167–169] This alien, limited, biasedness I extend to less obviously physical vehicles of knowledge, such as 4.2.3 Vehicular requirements in the social system spoken, written, and remembered language, logi- of science . In a partially overlapping presentation, cal symbol systems, and mathematical notations. I extend this attention to vehicular maintenance to Their rigid structures of terms and syntax are ve- the social vehicles of scientific knowledge: hicular requirements distorting the referents to The requirement of vehicle maintenance becomes some degree. This analysis can also be extended a structural requirement operating as a selective to the self-perpetuating social systems that are the factor in the winnowing of knowledge representa- vehicles for scientific knowledge (CAMPBELL tions. First, a biological example accepting the 1979a). The social glue that holds such groups to- common metaphor of the gene as a code embod- gether has structure-maintenance requirements ying adaptive ‘knowledge’. Consider a gene bom- that limit and bias the portrait of the world such barded by cosmic rays that disrupt and rearrange social groups sustain. its prior structure. In order for the resulting mate- Descriptive epistemology will need eventually rial to compete as a mutant gene which might im- a physical theory of optimal vehicles. Think of prove the fit of organism to environment it must plaster-of-paris casts, clay, magnetic tapes, photo- first meet the structural selective requirement of sensitive chemicals, and fixing processes in pho- being a gene at all, of being a stable alternate form tography: do these always involve a two-phase of DNA molecule capable of duplication. The great process, one phase of maximal flexibility … and a bulk of the disruptions produce rearrangements second phase of rigidity. Think of how we choose that fail to meet this structural requirement, being stone and wood for realistic sculpture: is it re- incomplete or imbalanced. There are also other quired that the physical structure of a good vehi- intraorganismic selection levels involved that cle be fine-grained? Do nervous tissue and genetic could be separated out with profit, but which I will codes conform to such principles insofar as they lump for now with the structural. Thus, the stable differ from other bodily tissues? How are these DNA molecule must be one around which a mes- physical requirements for stable record related to senger RNA can form, with this RNA capable in revising, expanding, and improving embodied turn of serving as a template around which a stable knowledge? Under what conditions are partial re- molecular alternative among the proteins can as- visions possible? Is total substitution of a different semble. It also must be a DNA molecule that at portrait generally a more mechanically feasible times escapes the inhibitory influences that inac- procedure than retouching it? tivate most genes most of the time. The proteins A similar applied physics of structures is needed produced must form nonlethal composites with for detection and transmission systems. Fritz the preponderance of the proteins other genes HEIDER in his ‘Ding und Medium’ (1926, 1959) was have produced. If, after all of this internal, struc- thinking about such issues. A ‘transparent’ medi- tural selection an adult, fertile phenotype is pro- um seems to be one that contributes least of its duced, this phenotype is then subject to an own structure to the knowledge it transmits. But external natural selection. Of all of these many it must have some structure to transmit other pat- selective systems, only this last can involve an im- terns at all … provement in the fit of the organism to the envi- Let us pause for a moment in this physicaliza- ronment, an increase in the ‘knowledge’ which tion. Like most traditional and modern epistemol- the genome carries in the external world. ogists, I, too, regard conscious experience, visual Similarly, before a scientific community can be perception, memory of past events, language, and a self– perpetuating social vehicle for ever-improv-

Evolution and Cognition ❘ 11 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell ing a set of beliefs about the physical world, it the interference with the validity of scientific be- must first meet the social-structural requirements liefs coming from the necessary tribe–mainte- of being a self-perpetuating social system. The re- nance vehicular requirements? [CAMPBELL 1979a quirements of achieving this ‘tribal’ continuity pp184–186 and CAMPBELL 1988a pp492–493] come first, even if they compete and interfere with the cognitive task of increasing the validity of the 4.3 Conclusion (Intermediate) image of the physical world carried by the ‘tribe’. A scientific community must recruit new mem- Analogies between theory change in science and bers and reward old members well enough so that biological evolution do “not” help justify scientific young recruits will be attracted to a lifelong com- beliefs. Instead: mitment to the field and will justify the drudgery ‘A selectionist model for a scientific belief ‘justi- and the painful initiation rites. Journals must be fies’ such a belief to the extent that it is plausible published, purchased, and read. Members must that ‘the way the world is’ has participated as one of remain loyal to the group and not ‘defect’ to other the systematic selectors of that belief from among tribes. Jobs must be found for loyal followers. So- the historically developed rival beliefs.’ Spelling this cial facilitators are needed to keep the group to- out will lead to rather orthodox conclusions: exper- gether and must be rewarded for this role, even if imentation is important as, too, are competitions in this means giving them scientific honors not the prediction of natural (e.g., astronomical) events. earned by their contributions. The requirements The ideology of the 17th century scientific revolu- of leadership for coordination and continuity tion held out as an ideal a social-construction sys- may produce leaders whose decision-making tem that would plausibly increase the role of power is used to protect their own social positions selection of scientific beliefs by their presumed ref- and their own scientific beliefs against internal erents. challenge from young rivals. The deeply ingrained In what follows I reexamine ‘knowledge’ and ‘jus- social custom of building ingroup loyalty by mo- tification’, then introduce the grounds for ‘compe- bilizing hostility and disgust toward outgroups tence of reference’ in ordinary language, and finally may be employed as a convenience (and perhaps come back to give this conclusion in more detail. even occasionally as a necessity) in maintaining group cohesion and continuity. Without meeting 5. An evolutionary perspective on what these social-structural requirements, there can be no scientific community to serve as the vessel car- we can expect for ‘knowledge’ and rying scientific knowledge. ‘justiﬁcation’ These social-structural requirements make it appropriate to ‘accuse’ a scientific community of 5.1 ‘Knowledge’ is more indirectly (but more being tribelike, that is, of having some basic simi- precisely) selected than are biological structures larities to other self-perpetuating social belief (and superstition) maintenance systems. This ‘accusa- While some naturalistic epistemologists (e.g., tion’ will be appropriate to single schools within KORNBLITH 1985, 1–13) see appeals to evolution as a a scientific discipline, to whole disciplines, and to mode of answering the skeptics, I do not and have coalitions of disciplines such as the physical sci- never done so. In terms of traditional epistemol- ences. ogy, I have from the first (CAMPBELL 1959) sided Calling attention to the functional required- with the skeptics. The shift from my 1960 dogma, ness of these shared tribal features may help to announced above in section 2.2, adds to my make them seem more compatible with a respect emphasis on the presumptiveness and indirection for science as a social system remarkably effective of the ‘foundations’ of knowledge. Vicarious selec- in its achievement of valid shared belief. This task tors, such as vision, employ presumptive vicars for is incomplete, however, until a sociologized ver- ‘the environment’, not the environment ‘directly’. sion of the “demarcation” problem of POPPER And, of course, the environment is not very pre- (1959) and other philosophers of science is ad- cisely represented even in the natural selection of dressed: How does the social system of science dif- genes and, through them, of proteins and protein fer from that of other self-perpetuating belief adjacencies. Selection at this level is a highly sto- systems, and, more particularly, what, if any, are chastic process based upon slight probabilistic the social system features relevant to minimizing advantages. The environment ‘represented’ is

Evolution and Cognition ❘ 12 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity always a past one, with only a slight advantage 5.3 ‘Knowledge is undefeated justiﬁed true belief’ being given to the most recent periods. Very strong are the implicit ‘assumptions’ concerning the regu- This latest amendment of the Anglo-American larity of nature, the representativeness of past sam- orthodox definition of knowledge (LEHRER 1989, ples, and the competence of accumulated models. POLLACK 1974) acknowledges that ‘justification’ is The strongest regularities will be crudely mapped always potentially defeasible, never complete. first, with contingent modifications occurring in ‘Knowledge’ is ‘not yet defeated’ belief. HARMAN’s later adaptations (see CAMPBELL 1988b, 467–471, (1965) account of ‘justification’ in terms of ‘infer- 1987a, pp182–185). Take for example ence to the best explanation’ acknowledges this … the New England fruit tree: It has achieved its map- same incompleteness and, moreover, makes ‘justifi- ping of the seasons not by an inductive procedure in cation’ a comparative process, in which 5.1 and 4 which records of thousands of years have been aver- are not denied, but hopefully ‘held constant’ by aged, but rather by retaining the remnants of many being shared by the competing beliefs. This occurs tried-out seasonal rules. These no doubt began with in a sort of “pragmatic eliminative induction” single-contingency rules, perhaps based upon temper- (DUNN 1995), in which elimination of rival expla- ature alone, later superimposing contingencies based nations is only ‘plausible’, historically dated, and on the amount of daylight, etc. Within each of the never complete. contingencies, the homeostat setting or reference signal undergoes continuous editing as late frosts and 5.4 The ubiquity of ‘inductive incompleteness.’ missed warm springs affect differential survival. Both early and late in this sequence, the fruit tree’s map is It is convenient to make this overlapping point by more simple than the seasons themselves. In certain quoting from a 1993 essay (cited below) The point is years, dramatic evidence of misfitting occurs, as when that all induction is incomplete in that it disregards the a whole season’s crop is lost to a late frost, but the existence of a multitude of interpretations or hypothe- species is well-advised to overlook most such anoma- ses that are, like the chosen one, consistent with the lies and counterevidence, to avoid being overrespon- data. The passage that follows provides examples of sive to a misfitting year or to overfitting a specific this inductive incompleteness or ‘underdetermination’ locale. [CAMPBELL 1988b, p468] in science and in visual perception, and argues that Vicarious selectors such as trial-and-error learning, while it is ubiquitous and inescapable, it does not ren- and especially vision, while much more presump- der useless labor to eliminate some plausible rival tive than natural selection of genes, are nonetheless hypotheses in any given domain. My own agenda for more competent for the immediate environment of the past thirty-five years has been to relate the phi- behavior. While the presumptions that make them losophers’ epistemological problems to evolution- work are historically as deep as any aspect of bodily ary theory, and to that more abstract model of form, the ‘inventions/discoveries’ by which they discovery and adaption shared by trial-and-error work include aspects sensitive to a much narrower learning, natural selection, cultural evolution, ‘specious present’, and their ‘selection ratio’ (were acquired immunity, radar, sonar, echolocation, we to borrow such a concept from the statistical and vision: ‘selection theory’ for short. Pursuant to theory of evolution, population genetics) is much that agenda, I would like to relay graphically in fig- higher, more precise. ure 1 what I take to be the consensus position of modern epistemologists and philosophers of sci- 5.2 A perspective-free, context-free and interest- ence. It is a perspective that provides philosophical free embodiment of knowledge is impossible warrant (were any needed) for the symmetrical, relativist, social constructivist, sociology of scientific This is the conclusion of section 4 above. The con- beliefs. text of co-selectors will be there. These will include One of the ‘scandals of induction’ can be ex- the vehicular co-selectors for sure, but also there pressed by noting that science makes use of an in- will be far less relevant historical co-selectors. ‘Dia- valid logical argument, making the error of the lectic indexical historicity’ connotes the point that ‘undistributed middle term’, or of ‘affirming the the language of new theories and the design of new consequent’. But while invalid, the argument is not experiments involve contrasts with predecessors, necessarily useless. The logical argument of science and are uninterpretable without such awareness. has this form:

Evolution and Cognition ❘ 13 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell

Figure 1: NEWTON’s gravitational theory as an ‘incomplete in- Figure 2: Visual perception as ‘incomplete induction’ (with duction’. apologies to DESCARTES).

If NEWTON’s theory “A” is true, then it should be agram like Figure 1 could be drawn, with a fringe observed that the tides have period “B”, the path of area of plausible rival interpretations: Thus, the Mars shape “C“, the trajectory of a cannonball form reference to ‘QUINE–DUHEM cop-outs notwith- “D“. Observation confirms “B”, “C”, and “D” (as standing’ in the second term of the syllogism judged by the scientific consensus of the day, QUINE- above. DUHEM cop-outs notwithstanding). Therefore NEW- The quasi-EULER diagram is also useful in presenting TON’s theory “A” is ‘true’. DESCARTES’s skepticism about sense perception, as in We can see the fallacy of this argument by view- Figure 2. DESCARTES was perhaps the best neurophysi- ing it as a quasi– EULER diagram, as in figure 1. The ologist of vision and physicist of vision of his day. In invalidity comes from the existence of the cross- this role he took a reflexively realistic stance in taking hatched area, that is, other possible explanations the machinery of perception to be made up of real for “B”, “C”, and “D’s” being observed. But the objects and events in the world, comparable to the syllogism is not totally useless. If observations in- ordinary objects of perception. He studied the phys- consistent with “B”, “C”, and “D” are agreed on ics of light rays, their propagation through pinholes by the consensus of participating scientists, these and lenses (both of glass and from ox eyes). He pos- impugn the truth of NEWTON’s theory “A”. The ited a subsequent message transmission through argument is thus relevant to a winnowing pro- nerves to the brain (hydraulically, by fluids in neural cess, in which predictions and social consensuses tubes). All of these mechanical links increased his on observations serve to weed out the more inad- awareness of the possibility of malfunction, of equate theories. Furthermore, if the predictions pseudotransmissions initiated at an intermediate seem confirmed by the consensus of current ex- link rather than by the ‘perceived object’ itself, and of perimentalists, the theory remains one of the pos- intrusions into this mechanical sequence from tan- sibly true explanations. gential causal chains. All this increased the plausibil- All inductive achievements are ‘incomplete ity of the skeptic’s argument from illusion. Trust in inductions’ (CAMPBELL l990b), with an incom- perceptions produced by such a mechanism required pleteness such as is graphically illustrated in Fig- faith in powers or processes that would keep the vul- ure 1. It is now generally recognized that this nerable causal chain insulated and free of defect. incompleteness is equally so for the so called Lacking selection theory, DESCARTES chose God. ‘facts’ that test or ‘falsify’ theories. Any ‘well-es- Those modern evolutionary epistemologists who tablished’ scientific fact which falsifies a theory invoke biological evolution (e.g., QUINE 1969, is a socially negotiated consensus for which a di- RESCHER 1977, GOLDMAN 1986, and the many others

Evolution and Cognition ❘ 14 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity

cited in CAMPBELL 1974b and CZIKO/CAMPBELL 1990) use dear old Mother Natural Selection to support a parallel trust in vision, albeit a more qualified trust, not providing incorrigibility. But many selectionists at the level of the evolution of the visual system tend toward a complacent foundationalism with regard to the momentary operation of vision. DESCARTES got to his skepticism about vision from what he took to be the illusory vividness of his own dreams, from an up-to-date knowledge of the physics, anatomy, and physiology of vision, and from a pathological need for certainty. But his analysis has been a part of the great tradition of perceptual skepticism back to the pre-SOCRATICS. PLATO’s parable of the cave (bk. 7 of ‘The Repub- lic’) has that theme. The “strange prisoners” are “like ourselves.” “They see only … shadows. To them, the truth would be literally nothing but the shadows of the images.” In this allegory, “the prison house is the world of sight.” Note how com- Figure 3: Verbally transmitted belief as ‘incomplete induc- patible this is with our modern physics and tion’. physiology of perception, in which the brain reifies objects from patterns of light indirectly and which was triggered by shadow form or photo superficially reflected from them. contours. From the epistemology exemplified by Figures As social animals, we acquire confident beliefs 1 and 2, all knowing can be epitomized as guess- through the reports of others. The layers of equivo- ing what is casting the shadows –– the shadows on cality are then more numerous, as shown in figure 3. our retina or the shadows on our laboratory As the process is diagrammed here, I may end up con- meters. fidently believing that, in the next room, out of my [CAMPBELL 1993, pp90–93] sight, the cat is on the mat. This belief is compatible In the article from which this section is extracted with the cat’s really being on the mat (the inner clear, (CAMPBELL 1993) two pages were devoted to an circle). But the reporter who supposedly is in a posi- ambiguous silhouette which when seen alone looks tion to see the cat in the next room may have halluci- like a gunman, but in the second figure is displayed nated, and the cat was not really on the mat. The as the shadow of a woman tennis player. These fig- inner two circles are as figure 2 (DESCARTES). The ures were reporter speaks out, “The cat is on the mat.” Now this … an advertiser’s illustration of the equivocality may be due to the cat’s being on the mat, and being of shadows. But the more fine–grained detail of a perceived as on the mat, or it may be due to idiosyn- photograph (or of a ‘direct’ perception), differs cratic semantics for “cat”, “mat”, and “on”, or it may from the silhouette shadow only in degree, not in be due to the reporter’s wanting me to believe that fundamental epistemology. Psychoepistemologi- the cat is on the mat whether or not it actually is. cally, the ‘guesses’ of direct perception are uncon- For example, the reporter may be my child, who sciously automated, and the conscious sides with the cat’s preference for sitting on the experiential ‘givens’ are of external objects as sofa and does not want me to go in and discipline though directly, unmediatedly, known. But this the cat. This source of equivocality is one peculiar did not mislead PLATO or DESCARTES, and it should to social vertebrates, not shared by the social in- not mislead us as epistemologists. [N]ote that any sects. For us social vertebrates, in our public truth transient belief that the shadow caster was a dark claims, there are always two motives: (1) to report alley gunman, or the belief that the photo was of validly to the best of our own knowledge and (2) a tennis player, is only co-selected by the shapes to influence the decisions we expect to be made of shadow and photo: Essential also to their for- on the basis of our report in a direction deemed mation are the culture and experiences providing favorable to us. These two motives are often in the repertoire of possibilities, one or another of conflict. The second is often the stronger, partic-

Evolution and Cognition ❘ 15 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell

ularly if our very lives and livelihoods depend on viously published articles as attempting “to refute those decisions. Since 99% of the beliefs of a sci- the MENDELIAN theory by all costs”. This dramatic entist are solely dependent on the observations of reversal is the result of a social persuasion process others, this makes social control of the validity of in which DARBISHIRE’s own data played a major reporting central to an epistemology of science (as role. Hull 1978 has noted) and to ordinary knowing of DARBISHIRE’s behavior conformed to the norms social animals (even if not for a nonlinguistic sol- of science as traditionally viewed. But neither itary perceiver). DARBISHIRE’s nor PEARSON’s nor WELDON’s behav- The crosshatched areas of figures 1, 2, and [3] iors are, however, explained by identifying their can never be entirely eliminated. Beliefs, and the examples of, or violations of, ‘scientific rational- best of current scientific theories, will always be ity’. Their behavior, instead, was the product of underdetermined, underjustified. This ubiqui- social system, social locus, and individual person- tous ‘inductive incompleteness’ (CAMPBELL ality. From a future, more thoroughly developed 1990[b]) leaves ample room for the influence of sociology of scientific validity, one could gener- social and personal interests seemingly tangen- ate recommendations for optimal individual be- tial to scientific inquiry. The research achieve- havior and optimal institutional norms for the ments of the symmetrical, social constructivist, goal of optimizing the validity of the consensus relativist programs in the sociology of scientific beliefs of a focal group of scientists. These recom- knowledge amply document such influences. mendations, we may anticipate, will have much [CAMPBELL 1993, pp93–96] in common with the ideology of the early scientific revolution, and with what scientists refer to 5.5 ‘Rational inference’ and the coherence as ‘the scientific method’. Conformity to such strategy of belief revision norms may be collectively ‘rational’ for a scientific community, but is not explained by so desig- Those making efforts to refute the claims of relativ- nating it. The term ‘rational’ at its best refers to ist, constructivist sociologists of scientific knowl- ideal norms, not causes of behavior. But even as edge sometimes argue that ‘scientific rationality’ norm, the meaning of ‘rational’ is in flux. It can adequately explains the adoption of new scientific no longer be identified with ‘logical’. Indeed, it is beliefs in response to new evidence, without recourse now generally recognized that where science re- to sociological explanations. Such an argument has sults in belief in a theory’s truth, it does so by way been made for the conversion to MENDELISM of DARBISH- of an invalid, but pragmatically useful, syllogism IRE. He was the major student of Weldon, who, with … [See above, the first section of 5.4 and Figure 1.] Karl PEARSON, militantly defended continuous variation Microeconomics is based upon defining ratio- biometric theory against William BATESON’s MEN- nality in terms of an individual person’s rational DELISM. With this, I vigorously disagree. And, more gen- optimization of his or her own utilities (not a erally, I believe that philosophers’ use of rationality as a group’s optimality), and this model is colonizing cause of belief change will not stand up under scrutiny. substantial segments of sociology. This model KIM (1994) examines the historic episode of belief may or may not be appropriate. For a sociology change from biometry to MENDELISM, circa 1910, using of scientific validity, the goals of a collective, not a sociology of scientific validity to rebut an earlier exter- an individual, need optimizing. But even were nalist ‘sociology of scientific knowledge’ focussing on the individualized rationality adequate for our pur- same episode. In my introduction to KIM’s book, I provide poses, it would not help us much. In the case of a critique of the use of rationality as a sufficient expla- Lincoln STEFFENS’s predecessor graduate student nation for belief change in science. This ‘rationality’ [CAMPBELL 1994, p xiii], falsifying his data was cannot be understood as a context-free deductive system rational behavior, rewarded by career success. producing entailed conclusions. Thus we find that: Given the parallels, DARBISHIRE’s behavior might DARBISHIRE’s bias in his initial report bears a socio- seem irrational. But to make that computation, logical similarity to episodes of fraud, but, as KIM we have to know for him the negative utility or (1994) dramatically reports, under BATESON’s pri- pain of dishonesty, and many other personal util- vate cross-examination of DARBISHIRE’s data in ities we can only speculate about. correspondence, and under CASTLE’s published To employ a model of individual rationality, criticism of his analyses, he confessed his errors one needs to also know DARBISHIRE’s information in print, even to the point of describing his pre- base. Rational actors never have complete infor-

Evolution and Cognition ❘ 16 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity

mation, particularly about the future. DARBISH- world and science. All of these presumptive axioms IRE’s behavior becomes selfishly rational if he are incomplete inductions. believed that there was a ‘truth’ to the matter, For examples, look back at Figures 1 and 2 of sec- that the consensus of fellow scientists would tion 5.4 above. The crosshatched areas of each con- soon converge on that truth, that his own data tain infinities of potential rival hypotheses. ‘Proof’ indicated that MENDELISM was correct, and that it consists of eliminating only those few plausible al- was best for his career if he joined that future ternative explanations which our community has consensus as early as possible. His quasi-factual made explicit. ‘Absurdly implausible’ rivals are not estimate of the likelihood of being caught fudg- considered. (Note that even in his small and tidy ing the data, and the degree of humiliation it domain, EUCLID had to employ the illogical, non- would entail, might also be involved. These fac- entailing ‘reductio ad absurdum’.) tual beliefs are surely social system products. Second, contrary to the EUCLIDEAN ideal, the corre- Probably the negative utilities of lying and spondence rules between posited things in the world shame, the degree to which one values one’s hon- (objects, actions, events) and their logical or alge- or among peers, how much one values HULL’s braic vicars are imperfect, with an imperfection that (1988) “conceptual inclusive fitness”, and so is context-dependent, and which may differ in vari- forth, are also social system products to a substan- ous locations in the deductive sequence. For descrip- tial degree. tive purposes, even the law of contradiction may not Not all conversions from one partisan position hold. The logical ‘ps’ and ‘qs’, and the algebraic ‘xs’ to another are to be interpreted as symptoms of and ‘ys’, are pure and mono-attributional. The real- the second position’s superior validity. In many world referents of these are invariably loci in n-di- cases of conversion, a sociological analysis may mensional space, multi-attribute syndromes. Using find that social power, within the scientific com- ‘ps’ or ‘xs’ as their vicars in a logical or mathematical munity or external to it, provides the most plau- deduction is a very approximate affair, and, as a pre- sible explanation. In the case of DARBISHIRE’s dicted experimental outcome, ‘p’ may not be quite conversion, however, it seems overwhelmingly the same as it was earlier in the deductive network. more plausible that it was his own data and a social system which, among other things, made 5.6 Natural kinds and concepts possible his being cross-examined on his data and gave him the freedom to change sides without My third proposed compromise with the EUCLIDEAN ideal loss of job or career. [CAMPBELL 1994, pp xv-xvi] of rational deduction involves rejection of the idea that The goal and the approximative practice of ‘ratio- natural kinds have essences and are defined by necessary nal inference’ should be retained, but the concep- and sufficient conditions. The following passage (from tualisation of what ‘rational inference’ is, or could CAMPBELL 1988b, pp457–460, a transcript of the Will- be, must be made more realistic. In other words, iam JAMES Lectures at Harvard University, 1977) makes our conceptualisation of rational inference must this point by reflecting on the processes of language be, sadly enough, relativised and contextualised. At learning. least three compromises with the EUCLIDEAN ideal There is another aspect of those samples of the real of rational deduction must be made if the deduc- world heavily utilized in the ostensive teaching of tions are to be relevant to the validity of descriptive the initial vocabulary that enhances entitativity, beliefs about a belief-independent world. cognizability and -talkaboutableness. This is the First, it must be recognized that there are many more preponderant emptiness of the n– dimensional at- axioms, and they are much less secure, than the EUCLID- tribute space and the resulting discreteness of EAN ideal assumes. Under that ideal there are ‘a few ‘natural kinds’. Imagine a space of possibilities indubitable axioms’, and many true deductions with a dimension for each possible descriptive from them. But history has shown these axioms variable that might be used to describe cats, dogs, (some at least) ‘are indeed dubitable, and none are squirrels, robins, ducks, geese, fir trees, oak trees, of proven truth’. For rational inference in science, dandelions, grass, stones, clods, rivers, and other we need thousands of ‘axioms’, i.e., unproven pre- natural kinds. The attribute dimensions could in- sumptions we tentatively trust. Rational inference clude height, length, breadth, weight, redness, is possible only within a community of discourse greenness, moisture content, carbon content, that shares most of the same ‘presumptive axioms’, fuzziness, furriness, angularity, dendricity, com- unproven but trusted beliefs about the nature of the pactness, location and mobility in latitude, longi-

Evolution and Cognition ❘ 17 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell

these subsets need overlap with those used by a dictionary maker, yet all three may be effectively D diagnosing the same natural kind. A natural kind E once identified and named is rich with attribute A characteristics yet to be discovered. Thinking in

B terms of definitions and essences, there seems to be rich redundancy of usable essence-sets or potential definitions, most yet to be discovered. On the other hand, the view can move one, and it moves me, to abandon the concept of definition Dimension One for natural kind words, and to recognize that they are learned as ostensively as are people’s names or proper nouns in general. Rather than ‘defini- C tions’, we have workable short lists of diagnostic symptoms substituting for larger syndromes of such symptoms, mostly yet undiscovered. When we introduce a new proposed hypothetical natu- Dimension Two ral-kind entity in the course of intellectual devel-

Figure 4: Some ‘natural kinds’. opment, we are at the same time asserting the existence of a syndrome we have only very partially mapped. tude, altitude, etc. If we plot the location of One other feature of this n-attribute space mod- individuals of one natural kind in such an n- el further undermines the use of essentialist defi- space, they cluster tightly together. In contrast, nitions. There are so many attributes, and the the space between any two natural kinds is vast. spacings are so far apart, that any one of the usual Let me make my imagery clearer by locating a few attributes can be missing, can have a zero value, natural kinds in a 2-space [see Figure 4]. Even and the individual still will be closer in the n- though there are only two dimensions, this draw- space to its natural kind than to any other kind. ing illustrates the point that most of the space is Thus a black swan is still a swan if feather color is empty. Even though kinds A and B overlap on all that is out of line with the usual symptom set, dimension 1 and dimension 2, in the 2-space prototype, or stereotype. Thus even before DAR- there is empty space between them. The overlap WIN, LINNAEAS had classified snakes as quadru- between kinds D and E is not so resolved, but if we peds, that is with the reptiles, for which four- were to add dimension 3, you would find that footedness was an essential characteristic. For kind D hangs close to the blackboard, while E is anatomists and physiologists, they shared so out here away from it. By the time there are five many attributes that even though four-footedness to ten dimensions most natural kinds are separat- was missing, snakes still obviously belonged. Thus ed by wide spaces. Stephen GOULD (1980) has the plucked duck ready for roasting, lacking feath- pointed out that even within the compact at- ers, ability to fly, webbed feet, long flat bill, fertil- tribute space of the three dimensions generating ity in breeding with other ducks, and any other of mollusk shells from flat clams to helically coiled the usually used symptoms of duckhood, may be snails, the great bulk of the attribute space is emp- still much closer to ducks in the n-dimensional ty and species tend not to overlap. Adding more attribute space than to any other natural kind … attributes generally increases the empty space. KUHN’s (1974, pp472–82 and 500–513) impor- There are, of course, hundreds of usable at- tant discussion of learning natural-kind terms is tribute dimensions. It is characteristic of natural both partially similar and has partially inspired kinds that any two kinds differ on innumerable this exposition. Note particularly his insistence dimensions. (Is this a counterpart of the empti- on the importance of empty inter-type space in ness of the space?) Numerous small subsets of at- the world of objects (p475) if similarity perception tributes have equivalent practical effectiveness as is to operate and his insistence that learning to distinguishing features. The attributes a child uses name ducks, geese, and swans inseparably also in- to recognize a cat may have no overlap at all with volves learning something new about the nature those used by a paleontologist, and neither of of the world. In his discussion with SUPPES, SUPPES,

Evolution and Cognition ❘ 18 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity

and SHAPERE (KUHN 1974, 500–513) he would, I 1898). There is a competitive conversation between believe, have been aided by a more radical rejec- two cartographers. The Englishman brags: “We tion of ostension as ‘defining’ (in contrast to puz- now have mapped all of England one mile to the zle-setting), also by a rejection of the existence of inch”. The German replies, “That is nothing. We ‘definitional’ (substitution-permitting) relations have a map of Germany one inch to the inch, but between observation terms and theoretical terms, the farmers won’t let us unroll it”. The goal of com- and by an explicit introduction of a quasi-osten- pleteness for knowledge is profoundly misleading. sive process of puzzle-setting (but not ‘defining’) ‘Knowledge’ is implemented by superficial reflec- in all learning of theoretical concepts in science, tions. as will be described below. As I understand him, he and I are in agreement on these points. [CAMP- 5.8 Maps are better epitomes of knowledge than BELL 1988b, pp457–460] are propositions The ‘deductions’ in scientific reasoning are gross ‘ceteris paribus’ shorthands, and the ‘other things’ They obviously employ interest-relevant selective assumed to be ‘equal’ in the deduction include the simplification while still retaining the possibilities of effects of many as yet undiscovered laws. validity and error (TOULMIN 1972, Chapter IV, GIERE These three modifications of the EUCLIDEAN ideal 1995). ‘do not at all rule out useful rational inference dialogue’ among a community of scholars. This dia- 5.9 ‘Competence of reference’ is primary: logue will employ the strategy of raising rival ‘Representation’, where approximated, is in the hypotheses and evaluating their plausibility. service of competence of reference, and is always This plausibility evaluation of rival hypotheses is also very incomplete the ‘coherence’ strategy of belief revision. LEHRER (1974) working within the ‘knowledge is justified As a naturalist, I take the history of animal adapta- true belief’ tradition, contrasts the coherentism tions (innate and/or learned) and nervous systems as which he advocates as a mode of justification with relevant to ‘knowing’. The great evolutionary divide the perceptual foundationalism advocated by Ch- (independently occurring in insects, crustaceans, isholm (and many others). POLLACK, HARMAN, GOLD- mollusks, and vertebrates) occurs when sense recep- MAN, and QUINE (see CAMPBELL 1990ms) and many tors evolve into distance receptors in object-seeking others are coherentists in this sense. This Anglo- locomotor animals. American coherentism must be distinguished from Before that divide, knowledge represents ‘if-then’ the continental ‘coherence definition’ of the mean- rules, where the ‘ifs’ are single sense-organ cell activa- ing of truth. Coherentism is compatible with a cor- tions, and the ‘thens’ are specific muscle contractions. respondence meaning of the word ‘truth’. LEHRER While it is misleading to use the word ‘reference’ for himself (1989, p132) says: “Knowledge arises when knowledge at this level, I will do so as a temporary there is the appropriate sort of match between all of stopgap. ‘Competence of reference’ at this level is what a person believes and external reality.” But nei- achieved when the activation produces a response ther LEHRER nor the militant correspondence theo- that is on the average adaptive. Through evolution, the rists of truth, such as QUINE and POPPER, have ever ‘if-then’ rules at this primitive level are greatly elabo- suggested that correspondence is available as a truth rated into contingent rules such as ‘if then but only test for specific beliefs. Indeed their advocacy of the if’. These reduce the equivocality (the inductive in- QUINE-DUHEM problem (or its equivalent) makes it completeness) of the ‘belief’, but of course do not clear that they reject this possibility. eliminate it. On our side of this great evolutionary divide, these 5.7. Knowledge is always oversimpliﬁed, interest- complex contingency rules merge into the fallible re- relevant and superﬁcial identification of patterns. Consider the activation of a single retinal cell: the number of events and objects A complete videotape with sound-track of one’s in the world which might plausibly have activated it past would be a useless form of memory. It would is very large. But for a pattern of multiple retinal cell take longer to search it than it did to experience it. activations, this equivocality is greatly reduced By way of my philosopher friend, Mark BROWN, (though still technically infinite). Note too, that it is comes this simplified version of a parable from now ‘the pattern’, not the same specific retinal cells, Lewis CARROLL (presumably to be found in CARROLL on each re-cognition. Competent reference (justified

Evolution and Cognition ❘ 19 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell belief) still leaves unresolved the Gettier problem (our tive sentences and the thus associated English ones long-time friend just might have an identical twin we seem to match up in respect of appropriate occa- never knew about). Inductive incompleteness is ine- sions of use, the linguist feels confirmed in these liminable. hypotheses…” (QUINE 1969, p33). On our side of the evolutionary divide, it is not only the These are precious passages (cf. also CAMPBELL ‘stimulus’, but also the action or ‘response’ that has 1990). They deny foundational status both to os- changed. It is no longer the twitch of a specific mus- tension and to simple conditioning. There is a fun- cle. Rather it is a distal response, an act or achieve- damental equivocality in the assumed shared ment, with substitutable muscle movements reference that results. (The present authors would extemporized to reach a subgoal that is itself a percep- argue that this partakes of the equivocality that tual pattern. “Pattern-matching is essential in distal plagues all induction, all relationships between the- knowing” (CAMPBELL 1966). Once again, a map rather ory and data.) This situation holds for children than a proposition best epitomizes knowledge. learning to speak (and protohumans inventing a language for the first time) fully as much as for the 6. Competence of shared reference in radically uninformed translator. But usable translation is in fact achieved; … chil- language dren do learn to use words in effectively sharing We who have distance receptors have unjustifiably reference … And insofar as valid beliefs are verbally reified stable external space containing discrete transmitted, ostension (equivocal though it is) has objects. Although it is unjustifiable, we reidentify been absolutely essential in the language learning with considerable competence ‘the same’ objects and process. events on successive occasions. As a biologist and psy- The improbable and tedious process thus de- chologist, I posit that these unjustified reifications of scribed is speeded up and made nearly error-free by objects occur very similarly from individual to indi- the shared innate and learned tendencies to reify vidual within the same species. It is this person-to- middle-sized physical objects and boundable acts. person similarity of reifications (not their validity per Without hesitation or awareness of alternatives, the se) that make language possible. The following child and translator guess that “gavagai” means excerpts from CAMPBELL and PALLER address these issues “rabbit,” rather than rabbit-aspect, rabbit-part, rab- [Extending QUINE’s model of radical translation, we pro- bit-moment, transient sense data, direction-of- mote] an antifoundationalist emphasis upon osten- pointing, etc. The perceptual reification of inde- sion to explain shared reference. Such ostension pendent objects and events, described in the previ- cannot be definitional because ostension is un- ous section, will have been naturally selected for the avoidably equivocal, as both QUINE and WITTGEN- usefulness available when stable discreteness, ma- STEIN (1953) have emphasized. Nonetheless, nipulability, and reoccurrence are typical, thus ostension provides a selective restraint, whereby making possible approximately adaptive learning the referent has some likelihood of participating in about them. It is around such pervasively shared the selection of the language learner’s guesses. In reifications that the foundations for usefully shared subsequent ostensive instances, and the learner’s linguistic reference can be built. use of the term, the mentor or learner may recog- The nature of the referents—their “entitativity” nize that the learner’s hypothesis as to the word’s (CAMPBELL 1973)—has already participated in se- meaning is wrong, and new guesses as to word lecting the perceptual reifications of the language meaning can be generated. The occurrence of ‘ra- learners (and their pre-linguistic ancestors tional errors’ on the part of children’s language use equipped with image-forming distance receptors). are symptomatic of this process (e.g., CAMPBELL The nature of these referents (their “ostensionabil- 1988a, pp462–464). From QUINE: “Such is the quan- ity”) also operates as a strong selective restraint on dary over ‘gavagai,’ where one ‘gavagai’ leaves off word meanings that can become socially shared in and another begins.... The only difference is how the first-level ostensive vocabulary: … you slice it. And how you slice it is what ostension The role of entitativity in guiding the boundaries or simple conditioning, however persistently re- of conceptualisations and words can be seen peated, cannot teach” (QUINE 1969, pp31–32). In- through examples of the kinds of designations stead, whatever shared reference is in practice that do not become words. Take words about achieved is a result of “hypotheses of translation – fragments of trees, for example: A word for a tree- what I call analytic hypotheses... Insofar as the na- fragment including leaf may or may not be

Evolution and Cognition ❘ 20 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity

present in a given language. If it is present, it will mutual ‘demonstration’ in the ideology of the scien- divide leaf from tree at that point where leaves tific revolution are of the same class that is employ- typically separate from tree-limbs. There won’t able in teaching language to an infant.) be a word for the extreme three centimeters of This model of language learning presumes that pre- leaf, nor a word for the whole leaf … plus the linguistic reifications of objects precedes language adjacent three centimeters of branch … [T]he learning, and is indeed shared by dogs, pigeons, apes, language will follow PLATO’s advice and ‘cut na- and pre-linguistic infants. There is a rival model, the ture at her joints’… SAPIR, WHORF, CASSIRER, BORGES, and COLLINS (1985) Language evolves in a speech community. model, in which language comes first and determines We can imagine there being a continual muta- the reification of perceptual entities. From this model, tion of new conceptualisations and namings. interpersonal sharing of competent reference seems Only a few of these become part of the common to me to be impossible (CAMPBELL 1989). coinage. The selection pressure of the learner’s guesses reduces word meaning discriminanda 7. Why pre-scientiﬁc cultural evolution identifiable by others in the language community. Words unattached to dependable discrim- did not produce competence of inanda are lost from the start. Words utilizing reference to presumed invisible causes of subtle discriminanda where adjacent striking visible effects discriminanda go undesignated are rapidly vul- The short answer is that the pre-scientific, or reli- garized so that through the multiple confusions gious, domain of believing had been co-opted for of common usage, the meaning drifts to the social solidarity purposes. Sociobiologists posit for us striking discriminanda. vertebrates (who have not eliminated genetic compe- Thus, through providing the referents uti- tition among the cooperators as the social insects lized in the editing of trial meanings, and have) selfish and nepotistic tendencies to optimize through providing the basis for the conspicuous personal inclusive fitness by being ‘free-riders’, by and popular hypotheses as to word meanings, cheating on the social contract, by avoiding self-sac- the common objects of the ‘real’, non-linguistic rificial altruism. Cultural inventions such as super- world to some extent edit any language. For the natural sin-recorders and after-death rewards and basic vocabulary that makes language learnable, punishments have curbed these selfish tendencies by language is limited to the talkable-about. A com- means of belief-systems involving invisible causes of mon preverbal evolutionary background visible effects, preempting this domain of belief. strongly biases human beings in the direction of When they are adaptive, these prescientific beliefs may finding ‘stable’ ‘entities’ highly talkable-about. have some metaphorical competence-of-reference, (CAMPBELL 1973, pp1050–51). but science posits much more specific and nearly lit- The shared perceptual reifications thus become eral invisible causes. To do so, it has had to declare socially ‘foundational’ in achieving linguistic independence from traditional religious belief. The transfer of valid beliefs from person to person. Even ideology of science is incompatible with traditional at the later non-ostensive stages of language learn- ideology even in its Euro-Mediterranean homeland ing – when new words are learned by describing as well as in other areas of the world. their meaning in terms of other words already Elsewhere I have put forward a lengthy argument con- known, the term ‘definition’ is not quite right. In- cerning the role of certain religious beliefs in the social stead, verbally produced exemplars are used, and orders of ancient civilizations (CAMPBELL 1991, pp97– once again the learner has to guess at the meaning, 99, 102–112). In it I suggest that, at least in the case of the guesses being disciplined but not entailed by archaic city-states, supernatural or “transcendent” reli- the exemplars. This is particularly clear for new gious beliefs functioned to promote intragroup coopera- terms in scientific theories. In the process, the de- tion by encouraging group members to believe that gree of shared reference may be severely reduced, cooperative behavior would yield personal gain after and also the selective role of the real nonlinguistic death. Groups holding such beliefs were favored by group- referents. (Expanded coherence networks may cure level cultural selection. Referring to BOYD/RICHERSON this degeneration to some degree.) [CAMPBELL/ (1985), I argue that this process is likely to be rare and PALLER 1989, pp238–240] fragile compared with both individual-level cultural se- (In future presentations I wish to give more stress to lection, and cultural transmission in which selection the proposition that the ostensionable objects of plays no significant part.

Evolution and Cognition ❘ 21 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell

7.1 Some puzzles in the sociology of the beliefs of more superstitious, more credulous, than their pre- archaic civilizations decessor cultures. Were these supernatural cosmologies … merely perceived as myth and poetry? I Ancient ‘high’ civilizations such as those in Egypt, Sumer, judge that they were believed … in a manner com- and Babylon, along the Yangtze and Yellow Rivers of parable to today’s beliefs in magnetism, gravity, China, along the Ganges and Indus Rivers of the Indian electromagnetic waves, atoms, genes, etc., that is, subcontinent, in ancient Peru along desert rivers and in as invisible but physically real sources of observ- the highlands, and in pre-Columbian Mexico show cer- able physical effects. tain interesting uniformities. All were organized around (6) Ubiquitous in these religious cosmologies full-time division of labor, with priests, rulers, skilled were rewarding and punishing heavens, hells, and craftsmen and later soldiers—specialists who did not feed reincarnations. These uniformly extended individ- themselves but had to be fed by the labor of others. Storage ual hedonic calculations beyond one’s own biolog- of food, usually in the form of cultivated grain, was ical lifetime … always present. Urban residence was characteristic for at (7) Also ubiquitous were wasteful royal funer- least some portion of the population. als, containing provisions for a royal afterlife. The It is on the moral orders of these city-states that I commonsense, materialistic, calorie-counting, focus. I find in them several puzzles that seem to economic optimizing of modern sociobiology me to be solved by BOYD and RICHERSON’s “con- (fused in anthropology with optimal foraging formist, frequency-dependent cultural transmis- strategy) has no tools to explain such wasteful- sion” (1985). All but one are puzzles of uniformity. ness. Fully useful horses, soldiers, wives, weapons, (1) All of these protocivilizations were accompa- jewels, and money were interred … Were this a nied by political centralization, coordination, lead- culturally isolated occurrence, no functional ex- ership, and hierarchical downward– command planation would be called for … The economic structures headed by a single person. All were well- and biological wastefulness is undeniable … [and] organized tyrannies or despotisms. clearly a selective force continually selecting (2) Although independently socially evolved, all against such customs. Their ubiquity requires … of these archaic city-states ended up with a very [an] overriding functionality, which I posit lies in similar set of moralizing preachments. All their affirmation of the reality of the afterlife. The preached the value of duty to the political organi- functionality of extending individuals’ hedonic zation and its customs. All preached the duty of calculi beyond their biological lives probably self-sacrificial military heroism in defense of the needs no arguing … It contributes to the survival of state. All preached within-group honesty. All the social group as an entity, and perhaps also to preached against self-interested deviations from the combined biological inclusive fitness of the duty (covetousness, jealousy, etc.). members (although not to any single individual’s (3) All supported their moralizing preachments inclusive fitness). with a supernatural cosmology that provided authority and sanctions for these preachments. (Why 7.2 Anadaptive cultural transmission and two types were not the force of custom plus interpersonal re- of adaptive cultural evolution inforcements sufficient without such cosmologies?) We need as a background an anadaptive model of (4) The details of these supernatural cosmologies cultural evolution. This should probably be more were extremely heterogeneous, differing widely like a model of nonadaptive genetic drift than a from city-state to city-state. (This is the puzzle of HARDY-WEINBERG equilibrium in which character diversity. All others in this list are puzzles of uni- frequency remains constant in successive genera- formity.) This … fact argues in favor of … their in- tions (Beatty 1987a, 1987b). Adjacent generations dependent invention … in a contiguous lineage are more similar than in (5) Compared to the supernatural beliefs of their noncontiguous ones … due to cultural borrowing acephalous predecessor societies, the pantheons from the previous generation. Across generations and cosmologies of the archaic city-states were … cultures change in a meandering way that should more incredible (as judged from a modern secular not necessarily be interpreted as ‘adaptive’ to a sys- viewpoint) rather than less so. While we can recog- tematic selective environment. Against this back- nize in these archaic city-states a general cultural ground, we can distinguish two forms of possible advance toward modern civilization, they were cultural adaptation.

Evolution and Cognition ❘ 22 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity

7.2.1 Individual level cultural adaptation. The posit group-level adaptiveness in cultural evolu- first is exemplified by the cultural evolution of tion. This is much more problematic, and for sev- tools, weapons, and knowledge of … materials … eral reasons, to be specified below. Change is due to the fact that individuals can gener- Let me illustrate from some classic small- ate variations on the culturally received form, and group experiments, initiated by Alex BAVELAS … can confirm their efficacy (satisficing, not opti- (see GUETZKOW 1961). Sets of six persons were mizing) … provided with communication links of contrast- The basic recipe for evolutionary adaptations is ing form: circle, hue-and-spokes, and fully con- haphazard variation, selection, and blindly loyal nected. Each member was given a few playing transmission. All of the fit is achieved by selection. cards, and the group was to assemble the single The variations show no foresight … they are ‘hap- best poker hand from the total of their cards. hazard,’ ‘blind.’ The only requirement is heteroge- The spokes pattern was clearly superior to the neity … [S]urviving variations (the genuinely fully linked and the circular pattern. This held adaptive plus the anadaptive and maladaptive true even where the hub, or communication variations not yet weeded out) are … reproduced clearinghouse position, was occupied by the with blind loyalty, both the maladaptive and the least competent person. When fully connected adaptive, although selection reduces the frequen- groups played repeated rounds, there was spon- cy of the maladaptive … taneous disuse of some links, resulting in a For both types of cultural evolution, there is an spokes pattern. This organizational pattern is an analogue to this blindly loyal retention. Uniquely attribute of the group (unattributable to individ- flaked spear points remained constant for tens of uals in isolation) and with a group-level adap- thousands of years, testifying to the strength of tiveness in this experimental ecology. cultural orthodoxy … Cultural evolution has as its For the central theory … we must posit such a raw material of variations not only chance devia- group-level selection not only for moral norms, tions from the inherited orthodoxy, but also the but also for religious-political ideologies. Adap- products of vicarious blind-variation-and-selective cultural evolution at this stage is much more tive-retention processes at the individual level, problematic than for ‘individual’ cultural evolu- such as vision (CAMPBELL 1956, 1974b) and cre- tion for many reasons: (1) There are, on the group ative thought (CAMPBELL 1960). These vicarious level, fewer ‘units’ and fewer ‘degrees of freedom’ processes are not of entailed validity but depend (proportionally to the size of the group). The ba- upon the imperfect validity of their presumptions. sic statistical theory of adaptive evolution re- These ‘intelligent’ sources of variation are indeed quires large numbers of quasi-independent units, often adaptive for the wrong reasons. and shared, consistent, selection pressures. (2) I judge that the adaptiveness of cultural evolu- The time units of trait exhibition and selection tion at this ‘individual’ level is undeniable … are longer and fewer. (3) Complex, multiattribute All adaptive processes require powerful reten- ‘objects’ of selection for cultural complexes make tion mechanisms for the cumulation of already it much less likely that a specific attribute be se- achieved adaptions, as a base upon which fringe lected. In contrast, the selective pressures on the variations are explored. Blind cultural conformity form of a spearhead are much more focused. (4) is individually adaptive for this type of cultural For those beliefs and organizational forms that evolution, increasing individual biological inclu- are beneficial for the group as a whole, but costly sive fitness. for individual inclusive fitness (producing self- For the theory that follows, we must posit that sacrificial altruistic behavior), there is individual- the individually adaptive products are so valuable level selection pressure operating against the that a general tendency toward blind conformity adaptive group selection. There are no doubt oth- has a net individual inclusive fitness advantage … er obstacles. I should doubt that cultural evolu- Readers should be warned that this is one of the tion at the group attribute level had taken place most vulnerable parts of the theory. were it not for the great obstacle to ultrasociality which I judge genetic competition among the co- 7.2.2 Group-level cultural adaptions. For our operators to be, and were it not for the seven cen- theory of archaic moral orders, we also need to tral puzzles of archaic city– states.

Evolution and Cognition ❘ 23 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell

7.3 The Boyd and Richerson model: nonfunctional, or afunctional, explanation of in- Intragroup homogeneity tracultural uniformities which the BOYD/RICHER- SON (1985, esp. chap. 7) model provides. This new Of the many important features of BOYD and RICH- functionalism does, however, still retain the con- ERSON’s great ‘Culture and the Evolutionary Pro- cept of ‘latent’ functions (functions not obvious cess’ (1985), I will make use of only one: to those who practice and transmit the custom, or conformist frequency-dependent nonlinear rationalized by them in other ways) even though (multiple parenting) transmission (‘conformist it was the concept of latent function that so re- transmission’ for short). Like their major prede- laxed the self-critical discipline of the old func- cessors (e.g., GINSBERG 1944, WADDINGTON 1960, tionalists, making it possible for them to treat reviewed by CAMPBELL 1965), BOYD and RICHERSON every feature of … [any society as functional. Now note that cultural evolution makes use of cross- with BOYD and RICHERSON’s help, functional theo- lineage borrowing (they call it “multiple parent- rists are forced to distinguish between ‘accidental’ ing”) in sharp contrast with biological evolution cultural uniformities and ‘selected’, or functional, (save for a few isolated exceptions). Under condi- ones. This distinction requires that a plausible the- tions of ecological diversity and migration, they ory of selection at that functional level be provid- find that it would be optimal for the learners to ed. The functional level upon which this essay adopt the majority (or plurality) position of the focuses is that of the coordinated social group. mentors (i.e., the ‘conformist’ version of frequency– dependent cultural transmission)… 7.3.1 Parentheses on reciprocal altruism. At this Add to conformist transmission the condition point, I interrupt my presentation of the BOYD and of stable small groups semi-isolated from each RICHERSON model for an important aside. These other. In a dozen generations, these groups will be ‘neutral’ homogeneities within groups, in the con- moved to internal homogeneity on all traits … In text of sharp differences between nearby groups, different groups the chance pluralities will be in almost certainly have a function whatever the spe- different directions, in a cultural analogue of ge- cific content of the homogeneity, and even if this netic drift. function was not involved in the selection for the Several things can be noted about this outcome. difference. TRIVERS (1971) in one of sociobiology’s Cultural unity on a trait need not be interpreted most important papers has presented the concepts as a product of adaptive selection. Cultural differ- of ‘reciprocal altruism’ and ‘moralistic aggression’. ences between nearby tribes need not be interpret- Reciprocal altrusim is also the key to AXELROD’s in- ed as adaptations to different ecologies. This is a fluential book (1984) on the evolution of coopera- great emancipation for the believer in cultural tion. For TRIVERS and AXELROD, the tendency to evolution. Previously (e.g., in my 1965 model) my form reciprocally altruistic cliques … is explicable anthropology friends would challenge me. “In our in terms of purely individual considerations. The people, twins are put to death at birth. In the reciprocal altruist pairs or cliques are precarious, neighboring people, twins are given special treat- and vulnerable to selfish defection. For them to ment and reared for shaman roles. Both live in the emerge requires long-lived individuals, who are same mosquito-ridden yam culture. Are you going likely to encounter the same specific others again to claim that this can be explained as different and again, and who have the capacity to identify adaptations?” (Nancy LEIS and Philip LEIS, person- … the specific others. Given these conditions, an al communication.) Cultural evolutionists have innate readiness to form such cliques could been at least as much burdened by excess adapta- emerge. TRIVERS posits that under such conditions tionism as the sociobiologists criticized by GOULD there would also evolve … ‘moralistic aggression’ and LEWONTIN (e.g., 1984). Indeed, such excesses against partners who violated reciprocity … … have been the major reason for the rejection of It has been pointed out (CAMPBELL 1979b, pp42– … functionalism in sociology and anthropology. 43, BREWER 1981) that a culturally-inherited mem- The new functionalism which I advocate at- bership in such a reciprocal altruist pact would re- tempts to avoid this excess adaptationism by reduce the risks involved in negotiating a new one. quiring for each functionality … a plausible It would be in the biological inclusive fitness inter- selection process at the organizational level of the ests of … parents to force such culturally-inherited function (CAMPBELL 1974c, l990a). This new re- membership upon their offspring. All group uni- strained functionalism is greatly helped by the formities on trait-specifically neutral features

Evolution and Cognition ❘ 24 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity would be useful signs of co-membership in such a learned … [H]omogeneities on neutral traits be- reciprocal altruistic pact. Easily perceivable homo- come a symptom of kinship. Thus conformist so- geneities in dialect, dress, rituals, and scarification cial transmission, and the in-group uniformity it would be particularly useful … Moralistic aggres- produces, receives individual-selection support sion becomes death-to-traitors in this functional from kin selection also. explanation of the roots of tribal ethnocentrism. If we turn the phrase from ‘reciprocal altruism’ 7.4 Cultural group selection in cultural evolution to ‘clique selfishness’, we note that the internally altruistic groups are exploiting unorganized per- BOYD/RICHERSON (1985) point out that it is this in- sons, or organized out-groups. Here is an area in ternal-group homogeneity and intergroup vari- need of clarification. Some presentations of recip- ability which set the stage for group selection … [if] rocal altruism read as though it would be to each the traits … provide a group-level advantage. This person’s inclusive fitness advantage if all humanity is a central concept for the Type 2 cultural evolu- were in a single reciprocal altruist pact, and that its tion of group attributes, ideologies, organizational only problem would be that of preventing anony- traditions, etc. mous free– riders. OLSON’s pioneering study (1968) It is important to emphasize that this is an orga- provides formal models and cites experimental nized (or at least face-to-face) social group (rather studies showing that small groups are much more than some nominal group, type, species, etc.). It is likely to achieve mutually altruistic cooperative re- also important to emphasize that this is a selection lationships. But this does not provide a rationale of culturally transmitted attributes, not biological. for … anti-out-group polarization which is so ubiq- (For biological evolution, this paper—at least tenta- uitous in human sociality. The concept of ‘clique tively—accepts the dogma of individual selection’s selfishness’, emphasizing the exploitation of out- dominance.) Groups (social organizations) can ‘die’ groups, comes closer. Each ingroup can plausibly when their biological individuals join other groups, or accuse the other group of clique selfishness and use are converted to other ideologies and organizations. … this accusation to mobilize their own in-group sol- The selective process could be pure emulation by idarity. From this point of view, the accidental in- unsuccessful groups of the successful. Or it could be group homogeneities produced by conformant the forcible imposition of the victor’s culture upon cultural transmission play a role comparable to the vanquished. Biological extinction of weak that of the unique nest and hive odors of ants and groups … [or] fertility of successful ones, could also bees. They provide signals as to who is to be admit- further the selective reproduction of ideologies, but ted and who excluded… … are not essential. The ‘group selection’ posited is This discussion of reciprocal altruism has been a selection of culturally transmitted beliefs, social- presented as a diversion from the BOYD and RICH- organizational structures, religious ideologies. It is ERSON theory. But it may be an essential addition. not a ‘group selection’ of genes. If cultural group selection produces group func- Where selection occurs at several organizational tional, self-sacrificial altruism, as we shall argue it levels, the levels operate in part as competing orga- does, then this produces an individual selection nizations. E.g., ‘[s]elfish DNA’, reproducing itself pressure against it, which would tend to eliminate without regard for whole animal functionality, is the conformant cultural transmission tendencies in rivalry with whole animal optimization … So, which produced it. The math modeling and com- too, individual biological person and social group puter simulations which BOYD/RICHERSON (1985) are—to some degree—in competition. DAWKINS report have not yet covered this feature. The plau- (1976) made famous the conception of “the selfish sibility of their model would be strengthened by gene” (not referring to selfish DNA). In my judg- the explicit addition of individual-selectionist sup- ment, he confused the unit of retention (the gene) ports. The social inheritance of membership in a with the unit of selection, and it is only the units reciprocal altruist clique is one of them … of selection … that can have purposes, including Kin selection is the other individual-selection selfishness. Vis-a-vis individual interests, we need route to quasi-altruism. The reciprocal-altruist to keep in mind a ‘selfish group’ concept and rec- cliques are most advantageous when they are com- ognize that effective selection at that level is selec- posed of close relatives and when individuals are tion for organizational and institutional self- less closely related to members of other nearby perpetuation, at the expense of the individual if cliques. But status as a relative is predominately need be (and within limits).

Evolution and Cognition ❘ 25 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell

7.4.1 Systematic selection pressures in the group their support of these norms. It seems that there selection of ideologies. John BOWKER (1973) … ar- are many specific cosmologies, origin myths, and gues that, if God existed as a part of the environ- pantheons that will support the moral norms … ment during the course of human evolution, then explaining the great heterogeneity of such beliefs. the human mind would be selectively attuned to If we use universality as a symptom of recurrent that reality (as it may well be to quasi-EUCLIDEAN selection pressures on content, then there seems geometry and quasi-NEWTONIAN mechanics). I to have been … survival value in the belief in su- want to accept the general mode of the argument, prahuman invisible authority, gods, or a God … but disagree if BOWKER sees it as justifying the spe- Such beings, or one Supreme Being, are to be taken cifically Christian origin myth, theology, and as real, as the invisible but real causes of visible claims for revelation. physical effects, comparable to our beliefs in in- Considering the dozen independent evolu- visible causes such as gravity, magnetism, wind, tions among archaic human city-states, and the and sunshine (i.e., ‘natural’ rather than ‘supernat- dozen times among the social insects that the syn- ural). The BOYD and RICHERSON theory of adaptive drome has emerged of stored nonspoiling food- conformist transmission requires this credulity, as stuffs, full-time division of labor (including social do the group-level effects. roles that are well fed but gather no food), and Explaining the ubiquity of invisible, transcen- professional soldiers in a ubiquitous role, I have dent authority is of course much more complex argued (CAMPBELL 1965, 1974c, 1983) for the ex- than the above paragraph explains … SWANSON’s istence of common “laws of sociology” as part of brief, provocative ‘The Birth of the Gods’ (1960) the ecological niche of all twenty-four cases, in- is, in general, supportive of the latent-functional- sect and human. ism of this paper. But it offers a nonfunctional It would be nice to be able to derive such laws explanation for the ubiquity of the hierarchies of from general principles, and then find them con- gods, and of one Supreme God. These pantheons, firmed in the observations. But even in biology, he argues, are metaphors for cultural-evolution- discovery of the ecological niche often follows the ary truths at the organizational level for which discovery of the puzzling animal or plant form. there exists no ‘literal’ language. The local human Such confounding of theory and evidence is at political organization is used as a source of meta- least as great a problem in the present arena. How- phor. The functional ubiquity lies at the political ever, conceptually, one might develop, on sys- level … The ubiquity of high and highest gods tems-analysis grounds, a model for human social may be thus explained, without arguing the func- behavior to optimize individual inclusive fitness tionality of the theology per se. in a central range of human environments … If Biological evolution has, presumably, selected these analyses are appropriately general, then our erogenous sense organs, our hedonistic sweets symptoms of universal conflicts should appear in and bitters, pleasures and pains, in such a way as all archaic city-states. We propose that the ubiq- to increase genetic inclusive fitness … It has no uitous features appearing in all archaic moraliz- doubt also selected for … long-term rational he- ings are the symptoms of this conflict. Note that donic calculation, which weighs future rewards sociobiology presents a model of vertebrate social and punishments against present temptations. If behavior optimizing individual inclusive fitness. cultural evolution … can lead credulous believers to Note also that the recurrent image of sinful, temp- extend this hedonic calculus to include rewards tation-ridden human nature in worldwide moral and punishments in an afterlife (heaven, reincar- systems is in remarkable agreement with the so- nation), this supports obedience … even in the ciobiologist’s picture. face of death, and … sacrifice of pleasures even in With regard to the shared moralizings of archa- the absence of observers and sanction systems. ic states, it seems to me plausible that any confor- What I am arguing is functional augmentation, mant transmission event that ended up not necessary requisite … that would tend to lead containing part of the universal moral norm pack- to more effective collective action. Hence, where age would have some systematic tendency to be the BOYD and RICHERSON belief-homogenization selected, however slight, and that the ubiquitous processes have produced such beliefs, the groups common set of moral norms is in general what is holding them may have functioned more effec- under selection pressure. Ideologies will be select- tively, their ideologies more imitated by other ed not for their own content, but incidental to groups, etc.

Evolution and Cognition ❘ 26 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity

Wasteful royal funerals may not be quite as 8. Ideology of the social system of ubiquitous in archaic city-states as I have claimed. scientiﬁc validity But they are certainly too frequent and too independent to be explained by accidental belief–ho- In this section I am going to concede, to constructionist mogenization and nonfunctional diffusion … sociologists of science and others, that science has a They would be more likely selected if they had several great deal in common with social systems that generate latent advantages, so I need not seek a singular and perpetuate superstitious beliefs, and much more explanation … [T]wo functions … are plausibly than its explicit, anti-traditional norms would imply. related to overcoming the social-organizational However, I argue that these norms—including commit- problems created by the biological human nature ment to ‘facts that speak for themselves’ —should con- produced by … genetic competition among coop- tinue to be preached, however hypocritically, because erators. they provide the opportunity for ‘the way the world is’ to The explanatory principles central to this essay co-select scientific beliefs via experiments and demon- seem useful only for the archaic city-states … strations, and thereby to enhance their competence of [A]lthough burials showing belief in an afterlife reference. This competence of reference, or validity, is nec- and in ghosts and spirits … no doubt already existed essarily hypothetical for the many reasons given in section in the simpler egalitarian predecessor societies … 5. I am saying that if we assume, without justification, other selective advantages must be found to ex- that scientific beliefs have greater validity than those of plain these. But these precursors may have provid- other social systems, then that difference can be explained ed useful seeds for exaptation into city-state by scientists’ partial adherence to the norms that origi- ideologies where they were selected by different nated in the ideology of the scientific revolution. functions. The discussion that follows is borrowed from CAMPBELL These elaborately wasteful royal funerals usual- (1986a pp108–135 and 1988a pp513–522 and 1979a ly had details testifying to the ruler’s belief in an pp192–198 and 1988a pp498–503. See also CAMPBELL afterlife. They presumably … also increased its 1986b.) credibility among the local population. Thus my first functional explanation is dependent upon 8.1 The anti-tribal norms of science the more obvious functionality of belief in after- death rewards and punishments. [In 4.3.2 above, and in a non– excerpted case study The second possible function seems unrelated (CAMPBELL 1979a and 1988a pp493–497), I have to the first, but not therefore incompatible. Cov- testified] to my conviction that scientific commu- etous envy is biologically natural but undermines nities must meet the tribal requirements, the social the division of labor, as is evidenced by the ubiq- structural requirements, of group cohesion and uity of anti-envy moral preachments in division- perpetuation. Many of the younger descriptive of-labor societies. Envy is exacerbated by the un- epistemologists, sociologists, and historians of sci- just share of collective products which those oc- ence, supported by anthropological readings, have cupying ‘communication clearinghouse’ roles are gone beyond this to the further conclusion that able to achieve for themselves and their offspring science is not different than other social supersti- … Conceptualizing rulers as divine, as a different tion preservation systems. Instead, I want to assert order of being, and ceremonializing this differ- differences as well as shared features: Among belief- ence in wasteful royal funerals may help reduce preserving mutual admiration societies, all of such envy. which share this common human tribalism, sci- I recognize these explanations to be weak. ence has different specific values, myths, rituals, What I will persist in seeking are functional ex- and commandments. These differences are related planations. These seem to be most likely to be to what I presume to be science’s superiority in im- found in social-organizational functions, rather proving the validity of the model of the physical than in individual– person functions … or in sur- world which it carries. plus-disposal functions. However, a seasonal need My ontological nihilist friends do not deny that to keep a large labor force organized and occu- some of science’s norms are different, but instead pied when its directly functional agricultural ac- assert that these different norms are hypocritically tivities were not possible may provide one preached—since, after all, the community of sci- function for pyramid building. [CAMPBELL 1991, ence does not live up to them, but instead behaves pp97–99, 102–112] like other tribes. In addition, such critics point out

Evolution and Cognition ❘ 27 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell that one cannot prove that the tribe of science is In all social communities, narcissistic people better than other tribes at sustaining a valid model with competitive egocentric pride are a problem. of the physical world, so that I am working on a Cooperative people who defer to the majority, pseudo-problem. In deference to the presumptive who get along and go along with others, and who nature of descriptive epistemology, I will confess hold the team together, get preferential treat- that both the problems I work on and the solutions ment even if they are less competent. This is true I offer are presumptive. of scientific communities too, contrary to scien- Focusing (as I feel we critical realists should) on tific norms that encourage vigorous internal crit- the high quality epistemological-relativist chal- icism even if feelings are hurt … Yet, scientific lenges accompanying this ontological nihilism, communities no doubt differ somewhat from the minimum we can do is generate a presumptive other belief tribes in the rewards given compe- model of social knowing which could produce in- tent arrogance. creased validity in beliefs about nature if the world No cult, sect, or other belief community can were as we assume it to be … isolate itself from the larger society. Science is The charge of hypocrisy on the part of scientists influenced by the external social system in many is a charge that I concede in advance if it asserts only ways counter to optimizing scientific truth. a descriptive inconsistency between what scientists Thus, the status systems of the larger society, do and the values they inculcate. But I do vigorous- based on political and economic power and so- ly reject an implied conclusion that therefore sci- cial class, contaminate the internal status system ence should stop inculcating these values and of science. Given equal ability, it helps a young instead openly acknowledge its conformity to the scientist … to be well-connected in the extrasci- tribal prejudices shared with the other superstition entific real world … to have good manners, conven- maintenance systems. Instead, I regard these spe- tional social views, and to come from a high-prestige cial scientific norms as precious, and would sooner university. All such contamination violates im- recommend their inculcation with increased vigor portant norms of science which hold that the than recommend their abandonment. contribution to scientific truth should be the All self-perpetuating belief communities are tra- only determinant of status within science. dition-ridden, viewing current events through the Should this norm be given up as hypocritical? Or spectacles of their pasts … But whereas most belief has it in fact some effect, making science less sub- communities locate truth in a long-past revelation ject to this contamination than it would other- or … locate the ideals of life in some past heroic wise be? … period, … science’s norms go explicitly counter to These values of science I want to keep alive and this, idealizing truth as lying in the future and de- available for use in the arguments that are made crying tradition as a burden and source of error … in the course of institutional decision– making. Do these antitraditional norms … not offer some These values will, I believe, occasionally make a advantage, however slight, to innovators … and difference-– a difference in favor of truth. Ex- make the sciences less tradition-ridden than other poses demonstrating that science violates these tribal groups? values can go two ways: In shocked disapproval, The organizational requirements of group conti- we can try to advocate them more effectively. As nuity and career attractiveness give administrators a sociologist of science, I approve of this naive and leaders power … beyond what their declining moralistic reaction and am thus sympathetic to competence and increasing rigidity merit. Such the institution-preserving motives lying behind geriarchical and authoritarian biases scientific com- the outraged reactions KUHN and FEYERABEND munities share with all other tribes. Thus off-the- have evoked … The exposes can also have the record advice which young recruits to a thriving opposite effect, in a call to give up the hypocrisy scientific laboratory receive indeed will usually be by ceasing to affirm these values. This I do vigor- much like that received by an army recruit: “You’ll ously oppose. find that if you want to get ahead in this lab, you’d But I do not want to exaggerate the effect of better go along with the old man’s ideas. He just preached norms. Institutional arrangements doesn’t know how to take suggestions or criticism.” that provide selfish incentives for norm– sup- Yet, military communities and churches have ex- porting behavior are more powerful. Honesty, plicit ideological support for this practice, while sci- for example, is an important norm for science as ence’s ideology explicitly decries it… for all other self-perpetuating social groups. But

Evolution and Cognition ❘ 28 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity the exceptional honesty of experimental physi- 8.2 Experiments as divination rituals cal scientists where science is concerned is probably not due to their superior indoctrination for In a final extension of the tribal model to science, honesty (though the sciences may recruit per- I would like to argue that certain of its oracle rit- sons who have an exceptional desire for an occu- uals and magical divination ceremonies could pation in which they can be honest). Rather, it is contribute to the validity of scientific beliefs, due to science’s exceptional punishment of dis- even if these rituals were adhered to superstitious- honesty and to the possibility of … exposure … ly and rationalized on inconsistent grounds. My which competitive replication of crucial experi- imagery for this comes from papers on traditional ments provides … [R]epeated failure of others to divination ceremonies by MOORE (1957) and AUB- be able to replicate a given experiment is cause for ERT (1959). Caribou hunters roast a shoulder fear, shame, and anxiety … Fields lacking the blade on the fire and use the cracks resulting to possibility or practice of competitive replication choose the direction the hunting party should thus lack an important social system feature sup- take. The ceremony … has many features de- porting honesty … signed to prevent human hunches from influenc- This brings us to the important scientific belief ing the outcome, thus providing an in ‘facts’, ‘the hard facts’, ‘facts that speak for uncontaminated channel through which the su- themselves’. This is such a pervasive normative pernatural powers can speak if they will … These belief-complex in the practice of science that I and similar divination rituals were used when vis- feel it must have great positive social system val- iting the well-known hunting sites had … yielded ue, contributing to the objectivity and validity of no game and justified tedious explorations into … science. Yet, descriptive epistemological analy- regions that would otherwise have gone unex- ses have been particularly debunking of this val- plored. They had the further social role of blam- ue. It is not only that scientists often fail to ‘face ing no one group member for the frustrations of up to the facts’, as this norm says they should, such exploration … They also … rendered the but, perhaps more important, that the hard fac- hunters’ behavior unpredictable … a [subtle] stra- tuality of the facts disappears on closes examina- tegic advantage explicated in the theory of games tion … Laboratory facts are only facts for those of VON NEUMAN/MORGENSTERN (1944). None of who share presumptions and background as- these adaptive wisdoms is explicit in the beliefs sumptions … Disconfirming meter readings are that accompany the divination ceremonies. In- regularly explained away as equipment failure or stead, there is a quite incompatible rationale of mistaken auxiliary assumptions. NEWTON, MEN- supernatural beings that are potentially helpful DEL, and DALTON are said to have doctored their but perverse and undependably placatable. The data to unacceptable degrees in order to make the wisdom of the custom is hidden in its manifest evidence for their theories more dramatically justification. persuasive. In their cases the theories were right In contrast to these, there is a second type of and replications approximately confirmed them, traditional ritual oracle so designed as to provide even if not again so elegantly. But such overedit- supernatural authority for the human wisdom of ing processes must often go wrong and create shaman or priest. In ancient Egypt some of the pseudo-discoveries. hidden voice tubes and mechanisms for getting In spite of the theory-ladeness and noisiness statues to move indicate a priestly sophistication of unedited experimental evidence, it does pro- about the deception they practised. Perhaps the vide a major source of discipline in science. oracle at Delphi was also managed this way. But Thus, though in some sense literally untrue, the for many more, such as those described by MOORE ideology of ‘stubborn facts that speak for them- and AUBERT, the procedures are, on the contrary, selves, independently of any scientist’s whim’ designed to keep the shaman’s wisdom from de- seems to me an extremely important norm to termining the answer, and are performed by de- preserve, and one that has a functional truth. vout shamans sincerely dedicated to providing a (Though I have not taken time to explicate this, channel through which the supernatural can evolutionary epistemology leads to giving up speak instead of oneself. (Many more are the notion of literal truth while still holding mixed…) onto the goal of truth [see CAMPBELL 1974a, and It would be characteristic of the exciting and 1975b p1120]). provoking new sociology and history of science,

Evolution and Cognition ❘ 29 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell

which takes as a duty the working hypothesis that cial tendency to split up into like-minded groups science is no truer than other forms of tribal mag- on specific scientific beliefs, but at the same time it ic, to interpret the scientist’s laboratory experi- requires a like-mindedness on the social norms of ment as just another divination ritual (BARNES the shared inquiry. Sociologically, this is a difficult 1974). As perspective-expanding exercise, I be- ideology to put into practice. MERTON (1973) has lieve this would be worth exploring in consider- described the requirement as ‘organized skepti- able detail. But I already know enough to insist cism’ … yet social settings in which organized skep- that the experiment is a ritual of the first type, ticism can be approximated are rare and unstable. meticulously designed to put questions to ‘Nature Nonetheless, it may be regarded as a viable socio- Herself’ in such a way that neither questioners nor logical thesis about a system of belief change that their colleagues nor their superiors can affect the might improve beliefs about the physical world answer. The supplicants set up the altar, pray rev- (including the not-directly-observable physical erently for the outcome they want, but do not world) were such to exist … control the outcome … A narrow window has In terms of my model of variation, selection, and been provided through which ‘Nature’ can speak, retention science puts greatest emphasis on the first two free from the scientist’s control. to the neglect of the third … (e.g., CAMPBELL 1974a). The brilliant historians and theorists of science To so stress variation and selection and neglect re- of recent years have convinced me that the galva- tention in the official ideology would be adaptive nometer reading is not at all the ‘solid fact that only if, at a particular historical period, retention speaks for itself’ we once imagined it to be. In- were grossly overemphasized in the general cultur- stead, it turns out to be highly equivocal, inter- al ideology and practice. At the time of the scien- pretable only at the cost of many unprovable and tific revolution, retention had gotten entirely out revisable assumptions. Yet, the laboratory scien- of hand insofar as beliefs about unobservable phys- tist’s phenomenology is not altogether wrong: ical processes and competence in negotiating with these stubborn laboratory facts are not speaking the invisible physical world were concerned. An in the experimenter’s own voice. Within the de- antitraditional counteremphasis was adaptive at grees of freedom the apparatus allows, they are that time … With such plausible apologies for cer- out of the control of one’s own hopes and wishes. tain aspects of early science that in the seventeenth [CAMPBELL 1979a pp192–198 and 1988a pp498– century did not need underscoring, I believe we 503] should seriously consider the ideology of the scientific revolution as a useful, albeit contingent, thesis 8.3 Disputatious communities of ‘truth’ seekers in an epistemologically relevant sociology of science. The title of this section denotes one sociological fea- From my perspective, the ideology and norms of ture of scientific belief exchanges. I use it to intro- science are not clearly distinguished from ‘scientif- duce my version of the ideology of the scientific ic method’. Scientific method is also to be seen as revolution. However, I have yet to integrate it with a product of cultural-evolutionary process on the the history of the scientific revolution or with HAB- part of a bounded belief-transmitting subsociety of ERMAS’s concept of an ideal speech community, with many generations. With FEYERABEND (1975), I which it probably has considerable communality would agree that new criteria of method are devel- (HABERMAS 1970a, 1970b, MCCARTHY 1973). oped as new choices provide new arguments. Like The ideology of science was and is explicitly anti- religious commandments, the ‘rules’ may be mu- authoritarian, anti-traditional, anti-revelational tually incompatible in the sense that if any one and individualistic. Truth is yet to be revealed. Old were to be followed with complete loyalty, it would beliefs are to be doubted until they have been re- interfere with compliance with the others. Each is confirmed by the methods of the new science. Per- … interpretable only against a background of prior suasion is to be limited to egalitarian means, and current norms and practices. While historical- potentially accessible to all … The community of ly both methods and ideology have fed on concrete scientists is to stay together in focused disputation, successes, it is convenient to regard the ideology attending to each other’s arguments and illustra- and practice of cooperative truth-seeking as com- tions, mutually monitoring and keeping each oth- ing first and method as a rationalized summary of er honest until some working consensus emerges successful usage in the community. This is more … [T]he ideology explicitly rejects the normal so- obviously so for the hermeneutic methods, but I

Evolution and Cognition ❘ 30 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity

believe it also holds for MILL’s canons of cause and structural and ideological changes in the social sci- FISHER’s analysis of variance. [CAMPBELL 1986a ence community, in publication practices, reward pp108–135 and 1988a pp513–522] systems, funding priorities, and the like. [CAMP- The discussion that follows stresses the importance of BELL 1986 p121 and 1988, pp515–516] ‘demonstrations’ in science. It should be expanded to emphasize that scientific ‘demonstrations’ are to 8.6 Replicability of fact involve objects and events that one can see with ones own eyes (and touch, hear, smell, or taste), A crucial part of the egalitarian, antiauthoritarian which are of the same order as those ostensionables ideology of the seventeenth-century ‘new science’ employed in teaching an infant language, i.e., basic was the ideal that each member of the scientific to interpersonal shared competence of reference, as community could replicate a demonstration for discussed in section 6 above. himself. … Thus, alchemy’s ideology of secrecy was an anathema to scientific exchange … and COL- 8.4 Visual demonstration and assent to facts. LINS’ sociological studies showing the absence of replication in current physics (1975, 1981a, It is my belief that some version of the fact/theory dis- 1981b) are to be taken very seriously … [T] early tinction is essential to sociology of science (cf. STEG- study of electricity will show hundreds of Leyden MÜLLER 1976, on KUHN and the theory-ladenness of jars, Voltaic piles, and static electricity wheels gen- facts). Here ‘facts’ are understood as shared, visually- erating sparks in hundreds of labs … few of these supported beliefs introduced as demonstrations in a per- experiments were published, but all figured important- suasive process. The terms ‘demonstration’ and ‘experi- ly in the social persuasion process. A healthy comment’ have much the same referent in early physics, munity of truth seekers can flourish where such chemistry and biology. The early persuasive role of ‘dem- replication is possible. It becomes precarious onstrations’ for both lay and scientific audiences was, I where it is not. assert, more important than ‘experiments’ as a social Replications can be attempted, but too fre- grounding of scientific belief, even though current exper- quently fail, in the most exciting fringes of experimental science seldom relies upon them. imental social psychology (a referential– ecology ‘Facts’ were originally theoretical inferences sup- problem, at least in part). Perhaps as a result, social ported by processes built into the nervous system psychology has the custom (atypical of successful by both natural selection and learning—ontolog- science) of trusting a single dramatic study in go- ical assumptions built into neural information– ing on to the next experiment without explicit or processing channels. At a more mature stage, facts implicit replication. The effort and cost of replica- may be micro– theories no longer controversial tions within a social system that regards them as within the scientific community. [CAMPBELL 1986, unpublishable and of low prestige contribute to p121 and 1988 p515] their absence. The lack of replications … in social psychology means that the discipline lacks the social 8.5 Referential ecology control that exists in those sciences in which replication is feasible and regularly succeeds. Successful science rests upon assent to agreed-upon facts In general, the absence of the norms and prac- (many implicit) as a background for demonstrating new tices of replication … are major problems for the facts. In the social sciences, difficulty in achieving agree- social sciences. From the standpoint of an episte- ment upon facts is certainly a major source of its failure mologically relevant sociology of science, this ab- to achieve genuinely scientific status. sence makes it theoretically predictable that the Much of this is a referential-ecology predicament social disciplines will make little progress. Can that is unavoidable, since it is intrinsic to social planned changes in science policy … change the science topics. Some of the problem, however, is a situation? [CAMPBELL 1986a pp121–123 and 1988 larger societal ecology-of-support issue. Were so- p516] cial scientists to limit their work to topics on which Hermeneutic approaches to history and other fields in factual assent could be readily achieved, it might the humanities create disputatious communities quar- be that society would not support their research, reling over the meanings of specific facts (including some nor students attend lectures limited to their find- that are theory-laden, i.e., resting on culturally-shared ings, because of their banality … But some of the conceptions of human nature). The scrupulous mutual fact-assent problems might be alleviated through monitoring in such communities often generates a mis-

Evolution and Cognition ❘ 31 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell trust of theory as leading to disregard of facts. (Thus Whether or not physics can ever quite achieve SPENGLER, TEGGART and TOYNBEE in history, and such this, the biological and social sciences are much early anthropologists as FRAZER, TYLOR, and WESTER- less likely to be able to do so and if they did would MARK have been used as cautionary examples of the evils be shifting to a different set of scientific puzzles of theorizing). than they have at present undertaken (CRONBACH 1986). This difference in referential ecology right- 8.7 The ecology of explanations and anticipations ly motivates social science methodologists and of facts metatheorists to be wary of borrowing uncritically a theory of science based solely on the physical It is our ontological predicament that the events sciences. The methods of the humanities appro- and stabilities we come to know lie at the intersec- priately become attractive alternatives to be con- tion of innumerable forces, restraints, and causal sidered. But it is intrinsic to understanding … that processes, most of them unmapped at any given the principles invoked are to some extent tran- stage. This is true both of the biological evolution stemporal and transcontextual, however condi- of sensing and predictive machinery and of cul- tionally hedged. Complete situational and ture or science. The survival value of perception historical uniqueness eliminates not only theory and memory lies in those ecologies in which the but also any grounds for ‘understanding’ or highest order interactions of all of the variables shared ‘meaning’. are not significant, in which ceteris are approxi- Insofar as the disputatious communities of mately paribus. Similarly, the growth of science scholars dispute about theory, they are likely to … has required not only the accumulation of facts, enter into arguments in which mutual persuasion but also the achievement of successful approxi- becomes possible only where there exist degrees mative theory relating facts to facts. This is most of freedom sufficient to make possible cross-vali- possible in those ecologies where powerful, over- dation … Such degrees of freedom can come only simplified ceteris paribus laws can be invented … from attempting generalization across instances, to sustain the group’s feeling of progress … persons, provinces, times, or the like. Of course, Success in this regard must often be a matter of we do not want to reject the ‘one-shot case study’ the referential ecology. Take, for example, the at- (as in CAMPBELL/STANLEY 1963) … [I]nstead we titude-change research epitomized by dissonance want to join my later recognition [CAMPBELL theory in experimental social psychology. In 1975a] of the degrees of freedom available in a most respects, the participants acted correctly in case study that come from ability to check multi- terms of my tentative sociology of successful sci- ple implications of a theory in that setting. Yet ence. The generally recognized collective fatigue until we have successful cases of mutual persua- and search for other models (documented by sion converging upon an agreed-upon theory GERGEN 1982) was in my judgment due to refer- achieved by such methods, we should continue to ential ecology. Major effects were replicated far regard the problem as serious. beyond chance, but unevenly and with no emer- This issue is not at all specific to the humanistic gence of dependable laws as to why and when. methodologies. It exists in extreme degree for Some critics I respect believe that the waning in- quantitative economics … which focuses on na- terest was due to faintheartedness, boredom, and tional economies, where only short runs of thirty careerist topic changes and that we now have for or forty ‘comparable’ years are available, rather dissonance theory dependable interaction laws than on the economics of, for example, neighbor- quite sufficient to support a major scientific edi- hood laundries, where degrees of freedom for test- fice. But these loyalists do not agree on which ing hypotheses on not-yet-used samples abound. dependable laws result. [CAMPBELL 1986, pp123– [CAMPBELL 1986, pp124–125 and 1988a pp517– 124 and 1988 p517] 518]

8.8 Degrees of freedom and historical/contextual 8.9 Eliminating rival hypotheses through uniqueness discretionary ramiﬁcation-extinction

Classical physics and chemistry sought laws which were The social crux of science, so it seems to me, is universal and timeless—indeed, time-reversible—so that the ability to render rival hypotheses implausi- provenance and history became irrelevant. ble. This focus, exemplified in the quasi-experi-

Evolution and Cognition ❘ 32 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity

mental tradition (CAMPBELL/STANLEY 1963, the profession. The most esoteric of poets or the COOK/CAMPBELL 1979), seems to me now more most abstract of theologians is far more con- central than experimental isolation or experi- cerned than the scientist with lay approbation of mental controls … all of which are devices for ren- his creative work, though he may be even less dering implausible the rival hypotheses that … concerned with approbation in general. That dif- disputative colleagues have effectively raised. ference proves consequential. Just because he is Randomized assignment to treatments does not working only for an audience of colleagues, an prove the hypothesis under test, nor disprove ri- audience that shares his own values and beliefs, val hypotheses, but instead renders many rival the scientist can take a single set of standards for hypotheses improbable … The narrowing of ex- granted. He need not worry about what some perimental comparisons by specialized control other group or school will think and can there- groups clearly illustrates the role of the current fore dispose of one problem and get on to the contents of the disputatious dialogue … Con- next more quickly than those who work for a trasting examples here are the abandonment of more heterodox group. Even more important, sham-operation controls to rule out surgical shock as the insulation of the scientific community from the cause of changes in behavior following experi- society permits the individual scientist to con- mental lobotomies, versus retention of the double- centrate his attention upon problems that he has blind strategy for drug trials. good reason to believe he will be able to solve. The QUINE-DUHEM equivocality of any experi- Unlike the engineer, and many doctors, and most mental result is a very real problem for any com- theologians, the scientist need not choose prob- munity of scholars. It can only be resolved by lems because they urgently need solution and discretionary judgments of plausibility. None- without regard for the tools available to solve theless, scientific communities often achieve them. In this respect, also, the contrast between working consensus, often against the interests of natural scientists and many social scientists the established and powerful. The central mode proves instructive. ‘The latter often tend, as the of argument involved is closer to the hermeneu- former almost never do, to defend their choice of tic methods than to some idealizations of scien- a research problem’—e.g., the effects of racial dis- tific certainty. The strategy of trusting most of crimination or the causes of the business cycle’ the fabric of corrigible benefits while you chal- chiefly in terms of the social importance of lenge and revise a few (the 1 to 99 doubt/trust ra- achieving a solution’. Which group would one tio) is central … and ramification extinction of then expect to solve problems at a more rapid rival hypotheses is ubiquitous … It was thus (as rate? (KUHN 1970 p 164, emphasis added) MOYER 1979 has so well described) that the Brit- Here I argue that the dependency of scientists on support ish community of astronomers and physicists from the larger society makes it probable “that science changed between 1915 and 1925 from over- works best on beliefs about which powerful eco- whelming faith in NEWTONIAN gravitational the- nomic, political, and religious authorities are indif- ory to complete acceptance of general relativity ferent (RAVETZ 1971) … However … visual … Something similar is described by CLAUSNER/ demonstrations vary greatly in clarity and persua- SHIMONY (1978) for ten years of testing of BELL’s siveness … and if they are convincing enough, dem- theorem. Each particular experiment was flawed, onstrations can even overcome political but through ramification extinction of the alter- relevance.” (CAMPBELL 1986a, p127 and 1988a, native explanations, these flaws permitted even pp519–520) the hidden– variable theorists … to be for the For example, 17th century Chinese emperors replaced most part convinced… [CAMPBELL 1986a pp125– their well-entrenched, politically important court as- 126 and 1988a pp518–519] tronomers with socially powerless Italian astronomers who could successfully predict lunar and solar eclipses 8.10 Insulation of the social system of science from (SIVIN 1980, pp25–26). “But the combination of per- that of the larger society ceptually unclear demonstrations with highly important political beliefs, such as is found in the Thomas KUHN says of the physical sciences that: applied social sciences, is … unlikely to produce be- [T]here are no other professional communities in lief change in the direction of increased compe- which individual creative work is so exclusively tence of reference.” (CAMPBELL 1986a, p127 and addressed to and evaluated by other members of 1988a, p520)

Evolution and Cognition ❘ 33 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell

8.11 Critical mass and success experiences communication networks seem better realized in anthropological linguistics and in … archaeology.) Here I suggest that there are certain sociological require- Contrast the ethnomethodology movement ments for maintenance of communities of truth-seekers. within sociology with the behavior-modification Most important are (1) a critical mass and (2) the appear- movement within psychology. Both are proud, ance of progress (collective success experiences). self-conscious deviations from the mainstream of Fad phenomena in the natural sciences’ choice of prob- their disciplines. Both have social-solidarity needs lems characteristically generate enthusiasm and in- that press for the inhibition of internal divisive- tense informal communications. In turn, these … ness, and hence for the inhibition of mutual criti- supply the critical mass, mutual monitoring, cross– cism, in order to shore up the intramovement validation and sometimes sustained perceptions of morale against the neglect or attacks of the domi- progress. However, [w]ithout perceived break- nant paradigm. The behavior modifiers withdraw through into further problem areas, interest dwin- to their own journals and within them pursue vig- dles and experimental energy becomes available orous internal disputation. The ethnomethodolo- for new perceptions of hot problems and promis- gists, on the contrary, produce isolated illustrations ing techniques. Certainly there are many areas of of their method and theory but, owing to their lack the social sciences that lack critical mass at the mu- of numbers and embattled status, never disagree tual monitoring level … Sociology-of– science with each other about matters of fact. Insofar as studies might well ask scholars in various fields they disagree about matters of theory, they tend about a specific publication “If you are wrong toward further sectarianism and reduced commu- about this, who will notice? Who will try to check nication rather than mutual monitoring. The same by replicating? Who will publish (or formally pub- can be said of … those who identify their method licize) their disagreement? Who will let you know and theory with Verstehen, hermeneutics, critical- privately about a successful or unsuccessful replica- emancipatory theory, dialectical materialism, phe- tion or other data that support or weaken your po- nomenology, and symbolic interactionism (except sition?” These studies should focus both on the for its atypical ‘labeling theory’). Mutual monitor- level of fact and on the level of theory … Without ing fails, not only within these movements but … having such studies available, let me nonetheless also in their roles vis-a-vis their parent disciplines. hazard some opinions. These are all movements of great actual or potential There exist mutually monitoring communities value for mainstream social science as penetrating in religious hermeneutics for such issues as who criticisms and suggestions for revision. But this ef- borrowed from whom in the New Testament gos- fect can only be achieved if both the radical critics pels and the proper translation of crucial verses in and the mainstream scholars remain within a com- the Old and New Testaments. There have in the mon communication network and listen seriously past and may still exist such communities in HOM- to each other. [CAMPBELL 1986a, pp127–129 and ERIC scholarship. It might thus be reasonable for 1988, pp520–521] practitioners to claim that cumulative progress had been made. 8.12 Observations on belief selection by In anthropological ethnography, no such com- sociologists of scientiﬁc knowledge munities exist, LEWIS (1951), BENNETT (1946), HOLMES (1957), FREEMAN (1983), FIRTH (1983), and I end this paper with yet another plea to the social con- BRADY (1983) notwithstanding. Instead, one seeks structionists in sociology of science. I do so as part of my a region as yet unstudied on long-term effort to persuade one’s special topic and, once them to entertain the possibility, successfully published, may Editors’ address however remote it might appear jealously try to prevent oth- to them, that the real world Celia M. Heyes, Dept. of Psychology, Univer- ers from allegedly needless might play a role in selecting the sity College of London, Gower Street, Lon- replication of one’s work … beliefs we come to have about it. dond WC1E6BT, United Kingdom. [T]he genuine collective in- If so, it follows that scientific Email: [email protected] terest in describing all van- discoveries may not be made up Barbara Frankel, 681 Taylor Street Bethle- ishing cultures before they hem, PA 18015-3169, USA. entirely out of the whole cloth. disappear provides justifica- Email: [email protected] Once more I quote what I have tion. (Mutually monitoring said elsewhere:

Evolution and Cognition ❘ 34 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity

Let us return now to the sociologists of scientific Reflexively, we should of course use interest knowledge (SSK). Their case studies of the social theory (BARNES 1977, 1983) to critique such a data construction of scientific consensuses report on set. The scientists being reported on shared an ide- the proposal and abandonment of many hypoth- ology probably leading them to exaggerate the eses about process and instrumentation. They of- role of the referent in belief selection, even in their fer microprocess studies of belief selection apparently unguarded gossip and shoptalk. How- appropriate to selectionist accounts. Thinking of ever, the SSK authors of these works might have LATOUR/WOOLGAR (1979), KNORR-CETINA (1981), had an opposite bias, in favor of the dramatic and and PICKERING (1984), for example, we probably more publishable message of ‘social construction have a hundred or so instances. These could be out of whole cloth’. It is conceivable to me that tentatively classified as to the type of selection they did not start out with this bias, that it was for involved. Some of these episodes will be classified them a discovery in the research process. (If we as purely social: An idea is not followed up be- add EDGE/MULKAY 1976, that initial bias was prob- cause it would offend the laboratory head, or be- ably lacking.) Interview testimony on this point cause it would give comfort to a rival research would abet our coherence-based discretionary group, or because of lack of funding. judgment. However, since they were all pressed by Other ideas are reported as being tried out and the need to write brief, vivid, and publishable found not to ‘work’. In such episodes it is possible books and articles, in which only one-tenth of that the way the world is participates in belief se- their notes could be used, a postresearch selection lection, even though there are vast negotiable re- bias in the ‘constructed out of whole cloth’ direc- sources to … deploy to settle why it did not work, tion is possible. Access to the full field notes might that is, QUINE–DUHEM cop-outs. More borderline provide a less biasedly selected set of episodes. On cases are those in which an idea is rejected because the other hand, we might expect an opposite bias of reasons why it will not work, or because of ru- on the part of science ethnographers such as HULL mors that a trusted researcher is known to have (1988b) and GALISON (1987). While I have not tried it and failed. If those reasons and rumors done the systematic rereading and coding, I am themselves have been coselected by the way the sure that these properly venerated texts (LATOUR/ world is, then (still more indirectly) coselection by WOOLGAR 1979, KNORR–CETINA 1981, PICKERING referent may have been involved. LATOUR’s (1987) 1984) make it plausible that the purported refer- chapter “Laboratories” also provides several ex- ents of belief are participating to some extent in amples of beliefs being abandoned by the resis- belief selection. [CAMPBELL 1993, pp103–104] tance encountered in laboratory practice.

References chance in evolutionary biology. In: Kruger, L./Gigerenzer, G./Morgan, M. S. (eds) The probabilistic revolution, Vol- Ashby, W. R. (1952) Design for a brain. New York: Wiley. ume 2, Ideas in the science. Cambridge, MA: MIT Press. Aubert, V. (1959) Chance in social affairs. Inquiry, 2, 1–24. Bennett, J. W. (1946) The interpretation of pueblo culture. Axelrod, R. (1984) The evolution of cooperation. New York: South-western Journal of Anthropology, 2, 361–74. Basic. Bertalanffy, L. v. (1955) An essay on the relativity of catego- Baldwin, J. M. (1900) Mental development in the child and ries. Philosophy of Science, 22, 243–263. the race. New York: Macmillan. Bowker, J. W. (1973) The sense of God: Sociological anthro- Baldwin, J. M. (1909) Darwin and the humanities. Baltimore: pological, and psychological approaches to the origin of Review Publishing Co.; London: Allen & Unwin, 1910. the sense of God. Oxford: Oxford University Press. Barnes, B. (1974) Scientific knowledge and sociological theo- Boyd, R./Richerson, P. J. (1985) Culture and the evolutionary ry. London: Routledge & Kegan Paul. process. Chicago: University of Chicago Press. Barnes, B. (1977) Interests and the growth of knowledge. Bradie, M. (1986) Assessing evolutionary epistemology. Biol- London: Routledge & Kegan Paul. ogy and Philosophy, 77–85. Barnes, B. (1983) On the conventional character of knowl- Brady, I. (Ed.) (1983) Speaking in the name of the real: Free- edge and cognition. In: Knorr-Cetina, K. D./Mulkay, M. man and Mead on Samoa. Contributions by A. B. Weiner, (eds) Science observed. Beverly Hills, CA: Sage Publica- T. Schwartz, L. Holmes, and B. Shore. American Anthro- tions, 19–51. pologist, 85, 908–47. Beatty, J. (1987a) Natural selection and the null hypothesis. Brewer, M. B. (1981) Ethnocentrism and its role in interper- In: Dupre, J. (Ed.), The latest on the best: Essays in evolu- sonal trust. In: Brewer, M. B./Collins, B. E. (eds) Scientific tion and optimality. Cambridge, MA: MIT Press. inquiry and the social sciences. San Francisco, CA: Jossey- Beatty, J. (1987b) Dobzhansky and drift: Facts, values, and Bass.

Evolution and Cognition ❘ 35 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell

Callebaut, W. (1993) Taking the naturalistic turn, or how real ence Association, 14–26. philosophy of science is done. Chicago, London: Univ. of Campbell, D. T. (1987a) Neurological embodiments of belief Chicago Press. and the gap in the fit of phenomena to noumena. In: Shi- Campbell, D. T. (1956) Perception as substitute trial and er- mony, A./Nails, D. (eds) Naturalistic epistemology: A sym- ror. Psychological Review, 63, 330–42. posium of two decades. Dordrecht: D. Reidel. Campbell, D. T. (1959) Methodological suggestions from a Campbell, D. T. (1987b) Selection theory and the sociology comparative psychology of knowledge processes. Inquiry, of scientific validity. In: Callebaut, W. G./Pinxten, R. (eds) 2, 152–82. Evolutionary epistemology: A multiparadigm program. Campbell, D. T. (1960) Blind variation and selective reten- Dordrecht: D. Reidel, 139–58. tion in creative thought as in other knowledge processes. Campbell, D. T. (1988a) (Overman, E.S., Ed.) Methodology Psychological Review, 67, 380–400. and epistemology for social science: Selected papers. Chi- Campbell, D. T. (1965) Variation and selective retention in cago, Il: University of Chicago Press socio-cultural evolution. In: Barringer, H. R./Blanksten, G. Campbell, D. T. (1988b) Descriptive epistemology: Psycho- I./Mack, R. W. (eds) Social chance in developing areas: A logical, sociolog– ical, and evolutionary. In: D. T. Camp- reinterpretation of evolutionary theory. Cambridge, MA: bell (Overman, E. S., Ed.), Methodology and epistemology Schenkman. for social science: Selected papers. Chicago, IL: University Campbell, D. T. (1966) Pattern matching as an essential in of Chicago Press, 435–86. distal knowing. In: Hammond, K. R. (Ed.), The psychology Campbell, D. T. (1988c) A general ‘selection theory’ as imple- of Egon Brunswik. New York: Holt, Rinehart, E Winston, mented in biological evolution and in social belief-trans- 81–106. mission-with– modification in sciences. Biology and Campbell, D. T. (1973) Ostensive instances and entitativity Philosophy, 3, 171–77. in language learning. In: Gray, W./Rizzo, N. D. (eds) Unity Campbell, D. T. (1989) Models of language learning and their through diversity, Volume 2. New York: Gordon & Breach, implications for social constructionist analyses of scientif- 1043–57. ic belief. In: Fuller, S. L./DeMey, M./Shinn, T./Woolgar, S. Campbell, D. T. (1974a) Unjustified variation and selective (eds) The cognitive turn. Boston, MA: Kluwer Academic, retention in scientific discovery. In: Ayala, F. J./Dobzhan- 153–58. sky, T. (eds) Studies in the philosophy of biology. London: Campbell, D. T. (1990 ms.) Exegesis on 15 famous paragraphs Macmillan, 139–61. from Quine. (For Quine’s visit to Lehigh University, Octo- Campbell, D. T. (1974b) Evolutionary epistemology. In: ber 15–18, 1990.) Duplicated manuscript, 19 single–spaced Schlipp, P. A. (Ed.), The philosophy of Karl Popper. LaSalle, pages. IL: Open Court, 413–63. Campbell, D. T. (1990a) Levels of organisation, downward Campbell, D. T. (1974c) ‘Downward causation’ in hierarchi- causation, and the selection-theory approach to evolu- cally organized biological systems. In: Ayala, F./Dobzhan- tionary epistemology. In: Tobach, E. O./Greenbert, G. sky, T. (eds) Studies in the philosophy of biology. London: (eds) Scientific methodology in the study of mind: Evolu- Macmillan. tionary epistemology. Hillsdale, NJ: Lawrence Erlbaum. Campbell, D. T. (1975a) ‘Degrees of freedom’ and the case Campbell, D. T. (1990b) Epistemological roles for selection study. Comparative Political Studies, 3, 178–93. theory. In: Rescher, N. (Ed.), Evolution, cognition realism. Campbell, D. T. (1975b) On the conflicts between biological Lanham, MD: University Press of America, 1–19. and social evolution and between psychology and moral Campbell, D. T. (1991) A naturalistic theory of archaic moral tradition. American Psychologist, 30, 1103–1126. orders. Zygon, 26, 91–114. Campbell, D. T. (1979a) A tribal model of the social system Campbell, D. T. (1992) Distinguishing between pattern in vehicle carrying scientific knowledge. Knowledge: Cre- perception due to the knowing mechanisms and pattern ation, diffusion, utilization, 2, 181–201. Reprinted in plausibly attributable to the referent. (Unpublished manu- Campbell 1988a. script.) Campbell, D. T. (1979b) Comments on the sociobiology of Campbell, D. T. (1993) Plausible coselection of belief by refer- ethics and moralising. Behavioral Science, 24, 37–45. ent: All the objectivity that is possible. Perspectives on Sci- Campbell, D. T. (1982) Experiments as arguments. Knowl- ence, 1, 88–108. edge: Creation, diffusion, utilization, 3, 327–37 Campbell, D. T. (1994) Toward a sociology of scientific valid- Campbell, D. T. (1983) The two distinct routes beyond kin se- ity. In: K. M. Kim, Explaining scientific consensus. New lection to ultra-sociality: Implications for the humanities York: Guilford Press, ix– xviii. and social sciences. In: Bridgeman, D. L. (Ed.), The nature Campbell, D. T./Cziko, G. A. (1990) Comprehensive evolu- of pro-social development: Interdisciplinary theories and tionary epistemology bibliography. Journal of Social and strategies. New York: Academic Press, 11–41. Biological Structures, 13(1), 41–82. Campbell, D. T. (1984) Can we be scientific in applied social Campbell, D. T./Heyes, C. M./Callebaut, W. G. (1987) Evolu- science. In: Conner, R./Altman, D. G./Jackson, C. (eds) tionary epistemology bibliography. In: Callebaut, W./ Evaluation studies review annual, Volume 9. Beverly Hills, Pinxten, R. (eds) Evolutionary epistemology: A multipara- CA: Sage Publications. digm proqram. Dordrecht: D. Reidel, 405–31. Campbell, D. T. (1986a) Science’s social system of validity- Campbell, D. T./Paller, B. T. (1989) Extending evolutionary enhancing collective belief change and the problems of epistemology to ‘justifying’ scientific beliefs (A sociologi- the social sciences. In: Fiske, D. W./Shweder, R. A. (eds) cal rapprochement with a fallibilist perceptual foundation- Metatheory in social science: Pluralisms and subjectivities. alism) In: K. Hahlweg & C. A. Hooker (Eds) Issues in Chicago, IL: University of Chicago Press, 108–35. Reprint- evolutionary epistemology (pp. 231–257) Albany: State ed in Campbell 1988a. University of New York Press. Campbell, D. T. (1986b) Science policy from a naturalistic so- Campbell, D. T./Stanley, J. C. (1963/66) Experimental and ciological epistemology. In: Kitcher, P./Asquith, P. D. (eds) quasi–experimental designs for research. Chicago, IL: Rand PSA 1984, Volume 2. East Lansing, MI: Philosophy of Sci- McNally.

Evolution and Cognition ❘ 36 ❘ 1997, Vol. 3, No. 1 From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity

Carroll, L. (1898) Sylvie and Bruno concluded. London: Mac- interpersonal behavior. New York: Holt, Rinehart & Win- millan. ston. Clausner, J. I./Shimony, A. (1978) Bell’s theorem: Experimen- Habermas, J. (1970a) On systematically distorted communi- tal tests and implications. Reports on Progress in Physics, cation. Inquiry, 13, 205–18. 41, 1881–1927. Habermas, J. (1970b) Toward a theory of communicative Collins, H. (1975) The seven sexes: A study in the sociology of competence. Inquiry 13, 360–75. a phenomenon, or the replication of experiments in phys- Harman, G. (1965) The inference to the best explanation. ics. Sociology, 9, 205–24. Philosophical Review, 74, 88–95. Collins, H. (1981a) Stages in the empirical programme of rela- Heider, F. (1926) Ding und medium Symposium, 1, 109–57. tivism. Social Studies of Science, 11, 3–10. (Translated as Thing and medium in Klein, G. S. (Ed.), Psy- Collins, H. (1981b) Son of seven sexes: The social destruction chological issues. New York: International Universities of a physical phenomenon. Social Studies of Science, 11, Press, 1959, 1–34.) 33–62. Holmes, L. D. (1957) The restudy of Manu’an culture: A prob- Collins, H. (1985) Changing order: Replication and induction lem in methodology. Ph.D. dissertation, Northwestern in scientific practice. Beverly Hills, CA: Sage Publications. University. Cook, T. D./Campbell, D. T. (1979) Quasi-experimentation: Hull, D. L. (1978) Altruism in science: A sociobiological mod- Design and analysis for field settings. Chicago, IL: Rand el of cooperative behaviour among scientists. Animal Be- McNally. haviour, 26, 685–97. Cronbach, L. J. (1986) Social inquiry by and for earthlings. In: Hull, D. L. (1982) The naked meme. In: Plotkin, H. (Ed.), Fiske, D. W./Shweder, R. H. (eds) Metatheory in social sci- Learning development and culture: Essays in evolutionary ences. Chicago: University of Chicago Press, 83–107. epistemology. Chichester and New York: Wiley & Sons, Cziko, G. A. (1995) Without miracles. Cambridge, MA: MIT 273–327. Press. Hull, D. L. (1983) Conceptual evolution and the eye of the oc- Cziko, G. A./Campbell, D. T. (1990) Comprehensive evolu- topus. Proceedings of the 7th International Congress of tionary epistemology bibliography. Journal of Social and Logic, Methodology, and Philosophy of Science, July 1–16, Biological Structure, 13(1), 41–82. 1983, Salzburg, Austria. Dawkins, R. (1976) The selfish gene. New York: Oxford Uni- Hull, D. L. (1988a) A mechanism and its metaphysics: An evo- versity Press. lutionary account of the social and conceptual develop- Dunn, W. (1995) Discovering and testing rival hypothesis ment of science. Biology and Philosophy, 3, 123–56. with pragmatic eliminative induction. Working paper. Hull, D. L. (1988b) Science as process. Chicago, IL: University University of Pittsburgh, Graduate School of Public and In- of Chicago Press. ternational Affairs. Kim, K. M. (1994) Explaining scientific consensus. New York: Edge, D./Mulkay, M. (1976) Astronomy transformed: The Guilford Press. emergence of radio astronomy in Britain. New York: Knorr-Cetina, K. D. (1981) The manufacture of knowledge. Wiley. Oxford: Pergamon. Engels, E.-M. (1989) Erkenntnis als Anpassung? Eine Studie Kornblith, H. (1985) Naturalizing epistemology. Cambridge zur evolutionaren Erkenntnistheorie. Frankfurt am Main: MA: MIT Press. Suhrkamp Verlag. Kuhn, T. S. (1970) The structure of scientific revolutions, sec- Feyerabend, P. K. (1975) Against method. London: NLÂ ond edition. Chicago, IL: University of Chicago Press. Press. Kuhn, T. S. (1974) Second thoughts on paradigms. In: Suppe, Firth, R. (1983) Review of Derek Freeman’s ‘Margaret Mead F. (Ed.), The structure of scientific theories. Urbana, IL: and Samoa’. RAIN, 57, 11–12. University of Illinois Press, 459–517. Freeman, D. (1983) Margaret Mead and Samoa. Cambridge, Latour, B. (1987) Science in action. Cambridge, MA: Harvard MA: Harvard University Press. University Press. Galison, P. (1987) How experiments end. Chicago, IL: Uni- Latour, B./Woolgar, S. (1979) Laboratory life: The social con- versity of Chicago Press. struction of scientific facts. Beverly Hills, CA: Sage. Gazzaniga, M. S. (1992) Nature’s mind. New York: Basic Lehrer, K. (1974) Knowledge. Oxford: Clarendon. Books. Lehrer, K. (1989) Knowledge reconsidered. In: Clay, M. /Lehr- Gergen, K. J. (1982) Toward transformation in social knowl- er, K. (eds) Knowledge and skepticism. Boulder, CO: West- edge. New York: Springer-Verlag. view, 131–54. Giere, R. N. (1995) Viewing science. In: Burian, R./Hull, D./ Lewis, O. (1951) Life in a Mexican village: Tepoztlan restud- Forbes, M. (eds) PSA 1994, Volume 2. East Lansing, MI: The ied. Urbana, IL: University of Illinois Press. Philosophy of Science Association. Lorenz, K. A. (1941) Kant’s Lehre vom apriorischen im Lichte Ginsberg, M. (1944) Moral Progress. Glasgow: Jackson. gegenwartiger Biologie. Blätter Fur Deutsche Philosophie, Goldman, A. I. (1967) A causal theory of knowing. Journal of 15, 94–125. (Translated as Kant’s doctrine of the a priori in Philosophy, 64, 357–72. the light of contemporary biology. In: Bertalanffy, L. v./ Goldman, A. I. (1986) Epistemology and cognition. Cam- Rapoport, A. (eds) General Systems: Yearbook of the Soci- bridge: Harvard University Press. ety for General Systems Research. Vol. VII, New York: Soci- Gould, S. J. (1980) The evolutionary biology of constraint. ety for General Systems Research, 1962, 23–35, translation Daedalus, 109(2), 39–53. reprinted in: Evans, R. I. (Ed.) Konrad Lorenz: The man and Gould, S. J./Lewontin, R. (1984) The spandrels of San Marco his ideas. New York: Harcourt Brace, Iovanovich, 1975, pp. and the panglossian paradigm: A critique of the adapta- 181–217 and in: Plotkin, H. C. (Ed.) Learning, develop- tionist program. In: Sober, E. (Ed.), Conceptual issues in ment, and culture: Essays in evolutionary epistemology. evolutionary biology. Cambridge, MA: MIT Press. New York: Wiley, 1982, pp. 121–143). Guetzkow, H. (1961) Organizational leadership in task-ori- Lorenz, K. (1951) The rule of Gestalt perception in animal ented groups. In: Bass, B./Petrullo, L. (eds) Leadership and and human behavior. In: Whyte, L. L. (Ed.) Aspects of

Evolution and Cognition ❘ 37 ❘ 1997, Vol. 3, No. 1 Donald T. Campbell

form. New York: Pellegrim and Cudahy. phie, 1 (1), 34–45. (Translated as: On the relationship be- Lorenz, K. (1973) Die Ruckseite des Spiegels. Munich: Piper- tween the theory of selection and epistemology. In: verlag. (Translated as Behind the mirror. New York: Har- Plotkin, H. C. (Ed.), Learning, development and culture: court Brace Iovanovich, 1973). Essays in evolutionary epistemology. New York: Wiley, Maynard-Smith, J. (1988) Mechanisms of advance. Science, 1982, 63–71. 242(25), 1182–83. Sivin, N. (1980) Science in China’s past. In: Orleans, L. A. McCarthy, T. (1973) A theory of communicative competence. (Ed.), Science in contemporary China. Stanford, CA: Stan- Philosophy of the Social Sciences, 3, 135–56. ford University Press, 1–29. Merton, R. K. (Storer, N. W., Ed.) (1973) The sociology of sci- Stegmüller, W. (1976) The structure and dynamics of theo- ence. Chicago, IL: University of Chicago Press. ries. New York: Springer-Verlag. Moore, O. K. (1957) Divination, a new perspective. American Swanson, G. E. (1960) The birth of the Gods. Ann Arbor, MI: Anthropologist, 59, 72. (eds) On the path of Albert Ein- University of Michigan Press. stein. New York: Plenum Press. Thayer, M. L. (1989) Ideas of death and afterlife in pre-Bud- Nagel, T. (1974) What is it like to be a bat? Philosophical Re- dist China. M. A. thesis, Department of Asian Studies, Set- view, LXXXIII, 43 5–450. on Hall University. Olson, M. (1968) The logic of collective action. New York: Thorndike, E. L. (1898) Animal intelligence: An experimental Schocken. study of the associative processes in animals. Psychological Pickering, A. (1984) Constructing quarks: A sociological his- Review Monograph Supplements, 2 (4, whole no. 8). tory of particle physics. Edinburgh: Edinburgh University Toulmin, S. (1967) The evolutionary development of natural Press. science. American Scientist, 55, 456–471. Pollock, J. (1974) Knowledge and justification. Princeton, NJ: Toulmin, S. (1972) Human understanding: The evolution of Princeton University Press. collective understanding , Volume 1. Princeton, NJ: Princ- Popper, R. R. (1935) Logik der Forschung. Vienna: Julius eton University Press. Springer. Toulmin, S. (1981) Evolution, adaptation, and human under- Popper, K. R. (1959) The logic of scientific discovery. New standing. In: Brewer, M. B./Collins, B. E. (eds), Scientific York: Basic. inquiry and the social sciences. San Francisco, CA: Jossey- Pringle, J. W. S. (1951) On the parallel between learning and Bass, 18–36. evolution. Behaviour, 3, 175–215. Trivers, R. L. (1971) The evolution of reciprocal altruism. Quine, W. V. (1969) Ontological relativity. New York: Co- Quarterly Review of Biology, 46(4), 35–57. lumbia University Press. Vollmer, G. (1975) Evolutionare Erkenntnistheorie. Stuttgart: Ravetz, J. R. (1971) Scientific knowledge and its social prob- Hirzel. (3rd ed., 1983) lems. Oxford: Clarendon Press. Vollmer, G. (1985) Was konnen wir wissen? Vol. 1: Die Natur Rescher, N. (1977) Methodological pragmatism. Oxford: der Erkenntnis—Beiträge zur evolutionaren Erkenntnis- Blackwell. theorie. Stuttgart: Hirzel. Richards, R. J. (1981) Natural selection and other models in Vollmer, G. (1986) Was konnen wir wissen? Vol. 2: Die the historiography of science. In: Brewer, M. B./Collins, B. Erkenntnis der Natur—Beiträge zur modernen Natur- E. (eds) Scientific inquiry and the social sciences. San Fran- philosophie. Stuttgart: Hirzel. cisco, CA: Jossey–Bass, 37–76. Von Neumann, J./Morgenstern, O. (1944) Theory of games Richards, R. J. (1987) Darwin and the emergence of evolu- and economic behavior. Princeton: Princeton University tionary theories of mind and behavior. Chicago, IL: Uni- Press. versity of Chicago Press. Waddington, C. H. (1960) The ethical animal. London: Allen Riedl, R. (1982) Evolution und Erkenntnis. Munich: Piper. & Unwin. Riedl, R. (1984) Biology of knowledge: The evolutionary basis Vaihinger, H. (1911) Die Philosophie des Als-Ob. Berlin: Re- of reason. (Trans., P. Foulkes) New York: Wiley. (Original uther und Reichard. work published as: Biologie der Erkenntnis: Die stammes- Wilson, D. B. (1974) Kelvin’s scientific realism: The theologi- geschichtlichen Grundlagen der Vernunft. Berlin: Parey, cal context. The Philosophical Journal (Transactions of the 1980). Royal Philosophical Society of Glasgow), 11, 41–60. Simmel, G. (1895) Uber eine Beziehung der Selectionslehre Wittgenstein, L. (1953) Philosophical investigations. New zur Erkenntnistheorie. Archiv fur systematische Philoso- York: Macmillan.

Evolution and Cognition ❘ 38 ❘ 1997, Vol. 3, No. 1 Linnda R. Caporael

Vehicles of Knowledge: Artifacts and Social Groups

Prolegomena Abstract “unnatural” (yet necessary) demands of culture ALLEBAUT When Werner C CAMPBELL identified artifacts and human social orga- for cooperation. He be- asked me to write a nization as “vehicles of knowledge.” This comment lieved that human nature response to Don CAMP- further develops these two concepts. Artifacts may could be distinct from BELL’s last paper, I represent transformations of nature, making it acces- culture, and that this na- accepted without hesita- sible to the human mind. The social organization of ture, fundamentally self- tion. Don and his stu- science is an effective vehicle of knowledge because it interested, was “in the dent, Marilynn BREWER, bears close parallels to “primitive” social organiza- genes”. A well-known slo- were major forces shap- tion. What differentiates science from other forms of gan of CAMPBELL’s was ing my interest in evolu- belief-maintaining organizations is the rejection of “cooperation between ge- tion when I was a supernatural cause. netic competitors”. Cul- graduate student. Then, ture held selfish human evolution was as popular in psychology as original nature in check through moral preaching, the exer- sin. CAMPBELL (1975) was the only person who cise of power, and the appeal to costs and benefits could talk about both—at the same time! To my in a future afterlife. CAMPBELL rejected “biological” surprise, writing this paper has been impossibly dif- group selection, demanding that any adequate the- ficult, and not just because of Don’s unexpected ory would have to provide a genetic explanation for passing. I was reminded of when, some years ago, self-sacrificial heroism. Yet, one could hardly argue he asked me why I admired his work when he was a that CAMPBELL favored a simple-minded genecen- sociobiologist. I said, lamely, that he wasn't like trism. other sociobiologists. (Not long ago, Barbara Many of CAMPBELL’s insights about evolutionary FRANKEL told me that he described himself as a “left- processes, as well as his empirical work, lead to a wing sociobiologist.”) But the fact of the matter different perspective, one that most influenced my was that CAMPBELL, “militantly ambivalent” as he own views about selection and design. CAMPBELL could be, worked in at least two, to my mind, con- was one of the first to discuss “downward causa- tradictory, evolutionary modes. One was reduction” (CAMPBELL 1974) in hierarchically organized tionistic and recalled the early days of systems. He recognized the group level of organiza- sociobiology; the other was systems oriented and tion as a selector on individuals (CAMPBELL 1982). contributed to an expanded evolutionary perspec- He knew well that humans were social animals with tive. This writing was arduous because CAMPBELL’s strong proclivities to forming groups. He acknowl- writings on human nature were drawn primarily edged individual and group selection could go in from his sociobiological mode. Yet his influence on the same direction. He recognized multiple levels of my own efforts at a naturalistic epistemology were social organization and the potential for synergy shaped by his views on evolutionary processes. and conflict between levels (HEYLIGHEN/CAMPBELL CAMPBELL had created two distinct standpoints for 1996). In 1965, CAMPBELL rejected self-interest as the a psychologically informed epistemology, but he engine for psychological explanation, argued for an stood on one, while I stood on the other. “innate ambivalence” between social and selfish CAMPBELL explictly advocated nature-nurture du- motives, and hypothesized advantages of cognitive alism. He saw a profound conflict between a biolog- efficiency for groups. Yet, just a few years later, he ically-based, selfish human nature and the recanted this vision of human psychology because

Evolution and Cognition ❘ 39 ❘ 1997, Vol. 3, No. 1 Linnda R. Caporael

it implied biological group selection (CAMPBELL Artifactual Vehicles 1972). Although I shared CAMPBELL’s ambition for an CAMPBELL’s prototypical example for an artifactual embodied, material and descriptive epistemology vehicle was that of a mosaic mural of a street scene. and have started down paths he cleared, there have The thickness of mortar, the size of the glass pieces, been significant detours. My comments recall the available colors and the restriction to two CAMPBELL’s earlier vision of psychology, reject the dimensions limited the validity of the picture repre- dualism that he later defended, and are sympathetic sented. A photograph of the street scene would still to group selection (CAPORAEL 1995; CAPORAEL in be limited (although not as much as the mosaic), press.). The differences between CAMPBELL’s empha- but it would have higher validity. The street scene ses and my own inform the background of this com- could be carved in ice or painted in icing on a cake. ment on the physical and social vehicles of These media would be both lower in validity and knowledge. These are subjects where CAMPBELL em- too flexible for most knowledge purposes: the ice phasizes antagonism between the medium and melts and the cake is eaten. CAMPBELL’s examples of knowledge, and where I will suggest more focus on physical vehicles, such as mosaics and plaster-of- coordination. paris casts suggest he conceived of structural vehicles, and of validity, as copies of knowledge. How- Vehicles of Knowledge ever, his inclusion of maps suggests a somewhat different story. As the joke about the farmers who In his target paper, CAMPBELL described two kinds will not allow the cartographers to unroll their 1:1 of vehicles, artifactual and social. According to map over the fields indicates, a “true” copy of CAMPBELL, the “vehicular substance that carries nature is not necessarily useful at all. knowledge is unavoidably alien to the referents of In many cases, representing knowledge in an arti- knowledge.” Any map, for example, is unavoidably fact demands a radical transformation—not a alien to the shoreline it represents. Similarly, the copy—of the knowledge. In these “cognitive arti- tribal structure of science is unavoidably alien—of facts,” the transformation reduces the computa- a different substance—to the knowledge it repre- tional complexity of problems. Knowledge is re- sents. Consequently, the representation of knowl- represented in an alien medium in order to change the edge is limited and biased by the structure of the cognitive tasks that must be undertaken. As an ex- representational medium, which may be artifacts ample, consider finding the square root of the prod- or social organizations. If a vehicle of knowledge is uct of 54 and 23. For most people, paper-and-pencil too flexible, it would be unable to hold together must be used to keep track of the intermediate cal- the picture it contains. At the same time, keeping culations. The calculations can be considerably sim- the vehicle intact conflicts with its function to rep- plified if they are re-represented in an old-fashioned resent knowledge. Ultimately there is a tension or slide rule. One lines up two scales, shifts the hairline conflict between the referents of knowledge and indicator, and reads out the number. The paper-and- the media of its representation. While not denying pencil calculation has been transformed into a per- that conflict exists, I suspect that it is the coordina- ceptual judgment task. The slide rule does not re- tion between parts that enables vehicles of knowl- quire that the user even know how to take the square edge—artifactual and social—to be selected for the root of a number. All the user needs to know is when functions they perform. “square root” is relevant and what scale to use when In the case of artifacts, I will suggest that they the square root is needed. Given the user knows the transform nature to make it more usable. A similar scale, the possibilities for errors are greatly reduced claim could be made for social vehicles, but my by using a slide rule compared to solving the prob- comment on that subject will be directed to ex- lem on the back of an envelope or even with a calcu- panding on CAMPBELL’s tribal model of science. In lator. a highly speculative mood, I suggest that the tribal The slide rule is not perspective-free, context-free, model of science bears a resemblance to a tribal or interest-free. It is the embodiment of historical model of knowledge among hunter-gatherers. I contingencies, trial-and-error learning, and tinker- conclude by suggesting that the demarcation being, which has accumulated over a period of time tween science and other belief systems might be greater than that of the lives of the individuals who simply the result of rejecting anthropomorphic contributed to its invention and development. An modes of explanation. abacus can be used to do the same task, but it, too, is

Evolution and Cognition ❘ 40 ❘ 1997, Vol. 3, No. 1 Vehicles of Knowledge: Artifacts and Social Groups the result of diverse historical and cultural contin- any naturalistic epistemology. From the Western gencies. “Same task”, however, needs to be qualified. perspective, Micronesian sailing is based on invalid From a computational perspective, the task is the knowledge, a conception of the heavens revolving same: whether one uses the different artifacts of a around the earth. Yet it worked for thousands of slide rule or abacus, the result of our computation years for the purposes of Micronesians. In Microne- will be the same (barring error). However, from a sian and Western navigation, techniques are se- naturalistic perspective, the calculating tasks are not lected by the rigorous test of making a landfall. the same. Abacus and slide-rule demand dramati- Today Micronesian sailors use Western navigation cally different background assumptions, motoric be- methods, not because of their greater validity, but havior and perceptual judgments. Inheriting the because the Western sailing tools greatly extend the cultural framework enables the use of the device. available ranges for sailing. More knowledge can be We may also wish to designate a category of “na- stored and distributed more widely when it is medi- ture’s artifacts.” Here there is neither copying nor ated by physical vehicles (e.g., charts and computa- direct transformation of materials. The artifact is tional devices) than by oral traditions and face-to- jointly a product of culture and nature, a cultural- face interaction alone. cognitive system. For example, Micronesians, using traditional methods of navigation, routinely em- Social Vehicles barked on voyages that took them out of sight of land for several days (HUTCHINS 1995). Yet the navi- CAMPBELL is well-known for his characterization of gator was able to accurately indicate his point of de- the “tribal structure of science.” By tribal structure, parture, destination, and islands over the horizon CAMPBELL alludes to the practices, beliefs, ideologies and out of sight. (This following description of Mi- and in-group loyalties that make science a self-per- cronesian navigation is highly oversimplified, but petuating community. The community itself, in his serves to make the point.) The navigator uses a sideral view, is the embodiment or “vehicle” for scientific compass, based on the night sky. The compass is knowledge. Like physical vehicles, however, social based on how humans experience the relationship vehicles are also limited and biased. They are between earth and the heavens. Relative to each unavoidably alien to the knowledge they attempt to other, stars are fixed in their positions. But as the represent and transfer between generations. Many earth turns, the constellations of stars appear to of the processes required to sustain science as a move in a path across the sky, rising at the eastern social system, that is, as a vehicle or embodiment of horizon and setting at the western horizon. A star knowledge, are antagonistic to competence of refer- path is a set of constellations that follow each other ence. I suggest that it may be the case that the pro- in a fixed arc across the sky. Position can be deter- cesses required to sustain science are also necessary mined by noticing where one is relative to the place to whatever competence of reference exists. on the horizon where a constellation is rising (or Humans are obligately interdependent, unable to setting). The navigator imagines a reference island, survive and reproduce reliably in the absence of a below the horizon and beneath a specific star. In the group. This interdependency, as CAMPBELL (1983) Micronesian conceptual framework, the canoe is recognized, would result in a feedforward process se- still, and the invisible island moves from a place in lecting for psychological mechanisms that would front of the canoe to one behind the canoe as the enhance group solidarity. For CAMPBELL, the advan- navigator approaches his destination. By observing tages of solidarity were primarily in the realm of the location of the star above the invisible island, the predator protection and group conflict. I part com- navigator is able to fix his location and others, as well pany with CAMPBELL at this point. In his view, groups as determine how long the trip will take. could be described as a homogeneous aggregate of This “artifact” is different from the slide rule. The innately selfish individuals with self-preserving so- stars do not have the accumulated knowledge of hu- cial sentiments. In my view, groups have come to mans. They are not transformed to make the knowl- serve as an interface between individuals and the edge available. In one sense, the knowledge is the habitat (CAPORAEL/DAWES/ORBELL/VAN DE KRAGT stars themselves, their patterned regularity. In an- 1989) other sense, the knowledge is the cultural traditions CAMPBELL’s description of a “tribal” structure to that make the stars usable, just as cultural traditions science, rather than, say, a market structure or bu- make abacus and slide rule usable. The sideral com- reaucratic structure, was especially insightful. There pass also puts into high relief a central problem for are parallels between the typical organization of no-

Evolution and Cognition ❘ 41 ❘ 1997, Vol. 3, No. 1 Linnda R. Caporael

madic hunter-gatherer groups (JARVENPA/BRUMBACH detect an animal equipped with natural camouflage 1988) and of scientists (HULL 1988). Both have a and distinguish predator from prey. Gatherers must nested hierarchy composed of work groups (research distinguish toxic from nourishing plants. Scientists groups), bands (primary face-to-face groups), and in the lab must detect the signal in their data from seasonal macrobands (tribes). In both instances—in the noise. science and among hunter-gatherers—group-size Primary social groups function as “staging com- numbers at these levels are fairly constant; groups of munities” (JARVENPA 1993), which serve as “general about 3-5 individuals (hunters or lab research processing and maintenance centers” for resources group), 30-50 individuals (bands, conceptual demes, and information retrieved from the smaller dis- or primary social groups), and 300-500 individuals persed groups; they are also loci of shared group (tribes of hunter-gatherers (BIRDSELL 1972) and of sci- identity. In hunter-gatherer workgroups, food from entists (HULL 1988)). the hunt is often brought back to the domestic base In some sense, both hunter-gatherers and scien- and shared in the band. Among scientists, research tists work in the (hypothetically) “real world.” results and their interpretation are brought back to Hunter-gatherers must find resources for survival; the primary social group in the form of workshops scientists search for explanations of phenomena in and small intimate conferences. the world. Scientists and hunter-gatherers have a va- For both scientists and hunter-gatherers, seasonal riety of techniques directed, with considerable ef- macroband meetings (or yearly conventions) are im- fort, to reducing the unpredictability and portant for the exchange of myths, gossip, and in- uncertainty of their experience. In both instances, formation about more distant areas and groups. isolated individuals have reduced viability. Individ- Macrobands meetings are also arenas for competi- uals must be part of a subgroup with sufficient “crit- tive games as well as the affirmation of common ical mass” to persist. Isolated hunter-gatherers are worldviews, the maintenance of languages (hunter- vulnerable to many dangers; isolated scientists are gatherers) and specialized professional terms (scien- not very productive. Similar to hunter-gatherers tists), and the exchange of people (mates, new PhDs, groups, which reproduce by fission, scientists do not or disgruntled members). usually begin a new research group as social isolates; they must bring some of the old group members The Demarcation Problem with them or attract new recruits. Similarities of function also exist in the subgroup The parallels between hunter-gatherers and scien- configuration structure of scientists and hunter- tists seem to suggest no genuine demarcation gatherers. Learning to fashion specialized tools, in between science and other belief systems. CAMPBELL, stone or other materials, or to use specialized labora- of course, believed it was impossible to justify any tory techniques or “primitive” food preparation differentiation, so he recommended a course of techniques is a hands-on, situated activity (like golf believing (without justification) that science pro- or tennis). Such learning requires feedback from ma- duced more valid knowledge than other belief sys- terials, equipment, and an experienced mentor if tems. If we make that concession, then science’s novices are to develop finely-tuned sensory-motor higher validity can be explained by the anti-tribal microcoordination. (Remember, too, that the equip- norms of science. I am rather skeptical about this ment has knowledge embedded in it.) Frequently this last point. Many areas of inquiry have anti-tribal learning takes place in dyadic interactions, not un- norms (cf. BARZUN/GRAF, 1992) and the case could be like the parent-child interaction through which chil- made that many of science’s notable achievements dren learn shared reference in language by ostension. may require tribal norms (GRIFFITH/MULLINS, 1972). For both lab groups and small hunter-gather “Tribally normative” background assumptions are groups ambiguity is a major attribute of the task en- essential for stabilizing the interpretation of data. vironment. Small groups of 4 or 5 people distribute Nor am I convinced that the late appearance of sci- cognition; they share tasks ence can be explained, as such as perception, classifica- CAMPBELL says, by the co-opt- Author’s address tion, inference, memory, and ing of belief systems by reli- contextually-cued responses gion for solidarity purposes. Linnda R. Caporael, Rensselaer Polytechnic when they interpret data or in- Many factors contributed Institute, Troy, NY 12180, USA. teract with uncertain features Email: [email protected]. to the ideological shift in the in the habitat. Hunters must 17th century origins of mod-

Evolution and Cognition ❘ 42 ❘ 1997, Vol. 3, No. 1 Vehicles of Knowledge: Artifacts and Social Groups ern science. A relatively small one from the modern- would not be so for other communities, say, the ist perspective may have had quite substantial world of the American Kennel Club, where beliefs results. An important element, which we almost are organized around aesthetic considerations. Self- take for granted, was the exclusion of anthropomor- organization is likely to be another factor, which is phic appeals in explanations. If humans are social crucial for natural phenomena to operate as a co- creatures, with minds geared to interacting with selector, and is characteristic of most scientific and other, unpredictable social creatures, it would not hunter-gatherer communities. State organization of be surprising if we “defaulted” to anthropomorphic science can result in episodes such as Lysenkoism, or explanations (CAPORAEL 1986; CAPORAEL /HEYES simply lower rates of discovery. 1996). Even modern humans casually attribute hu- It would be a mistake, I think, to continue to man characteristics to artifacts and nature. The preach the anti-tribal norms of science knowing modern automobile owner coaxes the car to start. them to be hypocritical. Hypocritical preaching in- The computer user begs her machine to produce the vites scientists to cloak themselves in ignorance (and desired output instead of pages of zeros. If humans arrogance). For the less arrogant, the extent to which typically default to anthropomorphic explanations the traditional norms of science are compromised by when they have no alternative account for a phe- culture and context comes as a rude, possibly disil- nomena, then excluding anthropomorphism (gods, lusioning shock to the young men and women who ghostly beings, elves, etc.) from the domain of pos- go into science. It would be far better to develop new sible explanation is a truly surprising, novel, and norms and expectations more suited to a post-mod- radical concept that forces the search for a material ern world on the edge of environmental and social explanation. crisis. Such norms would recommend humility in If we make the concession CAMPBELL urges, there the face of what science does not know; an appreci- are probably many factors entering into an explana- ation for the cultural and social context of scientific tion for the higher validity of scientific knowledge. work; a critical stance within a scientific community Central to these, as CAMPBELL points out, is that phe- and a generosity of spirit across the boundaries of nomena in the world are co-selectors of beliefs. This knowledge communities—norms, in fact, embodied is the case for scientists and hunter-gatherers, but in the work and life of Donald T. CAMPBELL.

References Caporael, L. R. (1995) Sociality: Coordinating bodies, minds and groups. Psycoloquy 6(1), http://www.cogsci.soton. Barzun, J./Graf, H. F. (1992) The modern researcher (5th Ed.). ac.uk/cgi/psyc/newpsy?6.01, ftp: //princeton.edu/pub/ Harcourt Brace Jovanovich, New York. harnad/psycoloquy/1995.volume.6/psycholoquy.95.6. Birdsell, J. B. (1972) An introduction to the new physical an- 01.group-selection.1.caporael thropology. Rand McNally, Chicago. Caporael, L. R. (in press.) The evolution of truly social cogni- Campbell, D. T. (1965) Ethnocentric and other altruistic motion: The core configuration model. Personality and Social tives. In: Levine, D. (ed), Nebraska symmposium on moti- Psychology Review. vation, 1965. University of Nebraska Press, Lincoln, NE. Caporael, L. R./Dawes, R. M./Orbell, J. M./van de Kragt, A. J. Campbell, D. T. (1972) On the genetics of altruism and the C. (1989) Selfishness examined: cooperation in the ab- counterhedonic components in human culture. Journal of sence of egoistic incentives. Behavioral and Brain Sciences Social Issues 28: 21–37. 12: 683–739. Campbell, D. T. (1974) “Downward causation” in hierarchi- Caporael, L. R/Heyes, C. M. (1996) Why anthropomorphize? cally organized biological systems. In: Ayala, F./Dobzhan- Folk psychology and other stories. In: R. W. Mitchell/N. sky, T. (eds), Studies in the philosophy of biology. Thompson/L. Miles (eds.), Anthropomorphism, anecdotes Macmillan, London, pp.179–186. and animals. SUNY Press, Albany, NY. Campbell, D. T. (1975) On the conflicts between biological Griffith, B. C./Mullins, N. C. (1972) Coherent social groups and social evolution and between psychology and moral in scientific change. Science 177: 959-964. tradition. American Psychologist 30: 1103–1126. Heylighen, F./Campbell, D. T. (1996) Selection of organiza- Campbell, D. T. (1982) Legal and primary-group social con- tion at the social level: obstacles and facilitator of measys- trols. Journal of Social and Biological Structures 5: 431– tem transitions. World Futures: the Journal of General 438. Evolution 45: 181–212. Campbell, D. T. (1983) The two distinct routes beyond kin se- Hull, D. L. (1988) Science as a process. University of Chicago lection to ultrasociality: Implications for the humanities Press, Chicago. and social sciences. In: Bridgeman, D. L. (ed), The nature of Hutchins, E. (1995) Cognition in the wild. MIT Press, Cam- prosocial development . New York: Academic Press. bridge, MA. Caporael, L. R. (1986) Anthropomorphism and mechano- Jarvenpa, R./Brumbach, H. (1988) Socio-spatial organization morphism: Two faces of the human machine. Computers and decision-making processes: Observations from the in Human Behavior 2: 215–234. Chipewyan. American Anthropologist 90: 598–618.

Evolution and Cognition ❘ 43 ❘ 1997, Vol. 3, No. 1 W. D. Christensen/C. A. Hooker

Selection Theory, Organization and the Development of Knowledge

A Selective Critique of D. T. Campbell’s Evolutionary Epistemology1

I. Campbell’s theory condition C iff: (a) I is cor- Abstract of knowledge related with C because of a selective history, and (b) Donald CAMPBELL has long advocated a naturalist Donald CAMPBELL devel- epistemology based on a general selection theory, with S treats I as a vicariant for ops a naturalist theory of a major qualification (made in the present paper) re- C. epistemology based on a stricting the scope of knowledge to vicarious adaptive This last qualification general selection theory processes. But being a vicariant is problematic be- is important because it which posits three major cause it involves an unexplained epistemic relation. rules out many problem- components to selection- We argue that this relation is to be explicated organi- atic cases and potential ist processes (section 1): zationally in terms of the regulation of behavior and counterexamples. For ex- the production of vari- internal state by the vicariant, but that CAMPBELL ‘s ample, contemporary vi- ants (V), selection across selectionist account can give no satisfactory account ruses have apparently these variants (S), and the of it because it is opaque to organization. We argue been no less structured by retention of those vari- the need for a general theoretic framework for under- VSR processes than has ants which survive the standing organized systems, and the need to under- the scientist, but they are selection process (R). The stand epistemic capacity in terms of systems with no more epistemic than major intuition underly- high-order globally constrained regulatory organiza- their ancient ancestors, ing CAMPBELL’s theory, tion. possessing no organized which he describes as the vicariant structure. This is “1960 dogma” (section 2) is that knowledge in sharp contrast with the lineage of the scientist, emerges through, and only through, VSR adapta- which is marked by a general tendency towards in- tion to the environment. In this respect the creasing epistemic sophistication realized through epistemic sophistication of a scientist compared organized neural structure supporting complexly with his virus-type ancestor simply reflects the nested conditional response structures whose genet- cumulative inductive achievements of millennia of ically inheritable component arose, we shall suppose DARWINIAN evolution (blind-VSR). with CAMPBELL, through VSR processes. Thus we CAMPBELL combines this selectionism with a the- have VSR processes for the refinement and condi- ory of vicariance: there are intrasystemic processes tionalizing of the products of earlier VSR processes, which function to shortcut evolution by vicariously i. e., second and higher order VSR processes. (It is at anticipating characteristic environmental condi- least plausible that the enhanced anticipatory capac- tions (so either avoiding or encouraging them); the ity these processes conferred created for the lineage knowledge of these processes is derived from the VSR a selective gradient favoring refined and deepened processes which produced them, they themselves op- vicariance processes.) So both the contemporary vi- erate according to VSR principles, and the scope of rus and the scientist have experienced first-order knowledge is restricted to them (section 2.2). This VSR structuring, but only the scientist has experi- leaves us with the following account of knowledge. enced second- and higher-ordered VSR structuring, State I of system S is knowledge of environmental and it is this high-order VSR structuring which is

Evolution and Cognition ❘ 44 ❘ 1997, Vol. 3, No. 1 Selection Theory, Organization and the Development of Knowledge

constitutive of the scientist’s massively enhanced local CO2 gradient. CAMPBELL’s theory is somewhat epistemic capacity. Since a VSR process is common vague concerning the epistemic relations involved to both cases, VSR structuring in itself is at best a in vicariance, because it almost entirely neglects the necessary feature of epistemic systems, and the virus contextual regulatory aspects which must be under- case makes it look a relatively uninformative require- stood to properly account for vicariance.3 (2) Selec- ment. tion theory is fundamentally unable to address The distinction between a system feature which is regulatory issues. Despite widespread attempts to merely adaptive and a feature which is a vicariant the contrary, it is simply the wrong kind of theory. plays a critical role in CAMPBELL’s theory, but it is itself Consequently, whatever the role of VSR processes somewhat problematic inasmuch as being a vicari- may be in knowledge development, contra CAMP- ant involves an unexplained epistemic relation. That BELL’s central thesis they do not represent a sufficient S uses I as a vicariant for C means roughly that S treats condition for its occurrence (cf. also HOOKER 1995). I as an indicator of C; however, what the indication We now turn to elaborating this point. relation comes to is unclear. It cannot simply mean that I is correlated with C because any system feature II. Why selection theory is unsuited as a which is an adaptation carries some degree of mutual information with the environment (though see foundation for natural epistemology CHRISTENSEN/COLLIER/HOOKER 1997 section V.2), and The basic problem with selection theory as a theory consequently would meet the criterion irrespective of natural epistemology is that it is radically impov- of its epistemic significance. The least question beg- erished as an explanatory model of organized sys- ging, and from our perspective most interesting, in- tems, but understanding organization is central to terpretation of the indication relation we can make epistemology. Selection theory washes out or is that I indicates C for S if I plays a role in regulating glosses over almost all of the actual dynamics of the S’s processes so that they are appropriate to C. This system being modeled, being focused just on fre- means not only that I has been selected for correla- quencies of outcomes. This is not a problem if the tion with C, but that I is embedded within a more primary objective is just to understand the dynam- general regulatory context in which S has a teleolog- ics of population statistics. However, we contend ical-control relationship with C.2 that many of the dynamical features which selec- However, although a regulatory interpretation of tion theory washes out play important roles in vicariance saves CAMPBELL’s VSR epistemology from understanding the epistemicly significant features being question begging, his relative neglect of it of complex organized systems. raises several further problems for him. (1) CAMPBELL Standard evolutionary theory takes as its primary repeatedly emphasizes that ‘competence of refer- unit the population, and measures changes in gene ence’ is the hallmark of a vicariant state qua knowl- frequency within the population over time. Much of edge, where he seems to understand this condition the usefulness of selectionist models derives from in a purely correspondence manner. However, com- being able to choose conditions such that the adap- petent process regulation involves much more than tiveness, or mutual information, of a particular gene competence of reference in this sense, which is use- is determined by its relative change in frequency as less in itself unless the system can make appropriate compared with other genes within the population. use of the vicariant. The extent to which I is a vicar- This occurs when the population is treated as a de- iant of C for S depends as much on the systemic composable, near-to-equilibrium system. These lin- context of I—in particular on the way C interacts earizing conditions are often only implicit in the with S and S’s goals—as it does on I’s referential com- models, but only under such conditions can out- petence. Indeed, treating competence of reference as come distributions be treated as independent of dy- the primary epistemic relation is misleading: to namical path from their starting values (e.g. as in the function as a regulatory vicariant for C, I must mod- HARDY-WEINBERG law). Moreover, these assumptions ulate S’s processes in ways that are appropriate to S’s are central to the explanatory capacity of the theory: interaction with C, but I needs only limited compe- If the population is not decomposable into quasi- tence of reference to do this, it simply needs to con- independent systems (organisms, genes), because it trol certain aspects of S’s interaction with C. The displays holistic dynamical constraints, then rela- mosquito flying up the CO2 gradient does not need tive frequency measures may not be well defined. reference to it even as an extended stream, it only There are many systems of this kind, and plausibly needs to control its flight direction by response to organisms and social systems—the epistemic sys-

Evolution and Cognition ❘ 45 ❘ 1997, Vol. 3, No. 1 W. D. Christensen/C. A. Hooker tems in question—are amongst them (see below). theory which explicitly treats dynamical and orga- Further, even if the system is decomposable or nizational factors. nearly so but the population is not near-to-equilib- If we are to successfully model knowledge devel- rium with its environment, then relative change in opment as an adaptive process—the root intuition gene frequency cannot be assumed to be a good mea- behind CAMPBELL’s evolutionary epistemology—we sure of adaptiveness because, beside correlation with must distinguish amongst different kinds of adap- the environment, path-dependency effects (e.g. tive processes. As CAMPBELL is all too aware, features founder effects in migration) will also play a role in of social systems (such as religious ideas) may be determining frequency distribution. Note that we adaptive without being epistemically reliable. On are not claiming that there cannot be nonlinear pop- the other hand epistemic reliability does seem to be ulation-genetic models in these latter cases, just that a strong feature (if not the only one) of the adaptive- in these cases relative frequency distribution will not ness of scientific ideas. Clearly there must be some directly measure adaptiveness. difference between the adaptive processes of the re- However, just noting the limitations in the appli- spective systems, but as we have seen, although se- cability of selectionist models is only part of show- lectionist models may in certain circumstances be ing that they represent an unsuitable foundation for sensitive to differences in adaptiveness, they are not natural epistemology—more important are the rea- sensitive to the underlying reasons for these differ- sons for these limitations. Although a selection ences. Consequently selection theory cannot distin- model compares the relative frequency changes of guish between adaptive processes which are stable traits over time, such a model provides no distinctively epistemic and processes which are insight into the dynamical processes by which the adaptive for other reasons. A theory of vicariant pro- frequencies evolve. Moreover, by only comparing cesses may well be an important step towards under- intra-populational differences, the contribution to standing distinctively epistemic adaptive processes, adaptiveness of the shared organization of the mem- but it cannot be developed from within selection bers of the population drops out of the picture—in theory. effect, it is suppressed by the high level abstracted structure of selectionist models. However, it is pre- III. Organization and the generation of cisely such dynamical and organizational information which we need if we are to be able to make knowledge distinctions between various sorts of adaptive pro- These problems result in CAMPBELL’s chronic inabil- cesses, e.g. between gene-based change of fixed first- ity to discriminate in a principled way between sci- order traits to improve environmental fit and high- ence and other cultural systems, and in his order modification of adaptable traits in organism- misguided account of the constraints associated learned fit. with embodiedness, the so-called ‘vehicular’ and The problem cannot be rectified simply by ex- ‘co-selection’ constraints on competence of refer- tending selection theory to intra-organismic or so- ence. To deal with the latter issue first, organized sys- cial processes. Although any system with stable tems characteristically face global constraints, and traits can have a black-box frequency counting these constraints may function both in an inhibi- model imposed on it and may sometimes be useful, tory fashion, by ruling out certain otherwise avail- the basic explanatory gap remains—there is no ac- able complexions of the system, and in a count of process organization. Organisms and social functionally amplifying fashion by making available systems are marked by hierarchical, holistic organi- certain capacities which the system would be other- zation where such modularity as occurs (cells, or- wise unable to achieve. We refer to the latter as gans, institutions) is heavily constrained by global enabling constraints (see HERFEL 1997). For example, functional organization. These kinds of systems are the cell membrane is a global enabling constraint not dynamically decompos- with respect to intra-cellular able because: (a) their compo- biochemical organization nents are highly Authors’ address because it provides a quasi- interdependent, and (b) their isolated environment within W. D. Christensen and C. A. Hooker, Dept. dynamics tends to amplify which critical parameters of Philosophy, University of Newcastle Cal- this holistic organization. such as ionic concentrations, laghan, NSW, Australia 2308. Consequently they can only enzyme activity etc., may be Email: [email protected] be adequately modeled by a maintained and organized so

Evolution and Cognition ❘ 46 ❘ 1997, Vol. 3, No. 1 Selection Theory, Organization and the Development of Knowledge as to perform useful work. Without the cell mem- receptors are highly organized, showing both mu- brane the energy gradients and physical organiza- tual inhibition and activation. In other words, the tion necessary for work would relax, and complex behavior of individual retinal neurons is heavily biochemical functionality would be impossible. constrained by the local and global organization of CAMPBELL, however, models all such constraints as the retina, and it is precisely these constraints (to- limitations on epistemic capacity. For example, he gether with those of the visual cortex) which enable characterizes ‘vehicular’ constraints in terms of reso- the general pattern recognition capacity of the pri- lution and coloration limitations on reference mate visual system.4 The CNS also has more VSR capacity (section 4.2.2). There is no hint that some functional structuring than the IS, but this is para- constraints may play an enabling function. sitic upon its greater organizational depth, and it is Similarly, CAMPBELL struggles to distinguish sci- the increased control depth in particular which is ence from other types of cultural systems, resulting the critical enabling constraint which permits object in the recognition that science like all cultural sys- reification. tems faces ‘tribal’ constraints, together with the pe- The organization of science parallels the features culiar recommendation that science should of the CNS in many respects. The theory-ladenness hypocritically ignore these constraints in its rhetoric of facts, far from being a competing constraint to (section 8.1). The problem derives from the fact that objectivity, is evidence of the extremely high level VSR structuring and embodiment constraints are of control depth in science. A single experiment in common to both scientific and non-scientific cul- physics, e.g., may involve appeal to all or most of tural systems—the differences will lie in organiza- physics in its design and data interpretation, includ- tional features of science which play an enabling ing the theory under test (HOOKER 1975). In this re- role in knowledge development. spect, theory ladenness is an enabling constraint for For example, both the immune system (IS) and science which ought to be embraced (with due cau- central nervous system (CNS) have VSR functional tion) rather than suppressed. characteristics (as well as origins), but they have very different organizational—and epistemic—charac- Conclusion teristics. If the IS is properly characterized as epistemic (and we think that in some respects it is), As a very rough approximation we believe that the it is a very low-order epistemic system whose ‘object characterization of the scientist in terms of high- reification’ capacity is virtually nonexistent. The order VSR structuring is on the right track. How- CNS, on the other hand, is an ‘object reifier’ par ex- ever, CAMPBELL’s account of vicariance—the confer- cellence (though it is also much more than this), but rer of epistemic content to the theory—is not because it has ‘more VSR’ than the IS. Funda- fundamentally deficient because vicariance is an mentally, the CNS is capable of object reification organizational property which selection theory per because it has far greater control depth than the IS. se cannot illuminate. If we are to understand If we consider the organization of a primate retina, epistemic systems, and the role of VSR processes to take just one instance, we may observe that there within these systems, we need a general theoretic are approximately 100 million photoreceptors framework for understanding organized systems. In which detect light signals which synapse onto a particular, we need to understand systems with mere 1 million axons in the optic nerve (CHURCH- high-order globally constrained organization LAND/SEJNOWSKI 1992, p148). Connections amongst which has embedded parallel competitive features.

Notes 2 BICKHARD (1993) presents an analysis of indication (and representation) which is of this general type (see also BICK- 1 All unnamed section references are to CAMPBELL’s paper in HARD/TERVEEN (1995)). The details need not concern us, the this volume. The conceptual background of this paper is important point for the present argument is the introduc- indebted to the complex systems ideas developed over a tion of a regulatory context to analyze an epistemic rela- number of years by the highly interactive research group at tion. Part of the significance of this is a shift in focus from Newcastle (Australia) led by C. A. HOOKER and including J. the upstream causes of a signal to the downstream modu- D. COLLIER, W. E. HERFEL and W. D. CHRISTENSEN. Our ideas latory effects of a signal. For a detailed theory of the on the metaphysics of natural epistemology have also epistemics-relevant metaphysics of complex organized sys- greatly benefited from discussions with M. H. BICKHARD. tems see CHRISTENSEN/COLLIER/HOOKER (in preparation),

Evolution and Cognition ❘ 47 ❘ 1997, Vol. 3, No. 1 W. D. Christensen/C. A. Hooker

COLLIER/HOOKER (1997), and for a regulatory theory of fun- 4 CAMPBELL does refer to the role of mutual interaction in damental epistemology see CHRISTENSEN/HOOKER (in prep- equivocation reduction (section 6), but with respect to VSR aration), and HOOKER (1995). epistemology this stands as an ad hoc insight without sys- 3 CAMPBELL does discuss some contextual issues under the tematic basis. banner of ‘co-selection’ which we shall discuss below, but this does not address the regulatory context of vicariance.

References in preparation. Churchland, P. S./Sejnowski, T. (1992) The computational Bickhard, M. H. (1993) “Representational content in humans brain, Cambridge, Mass.: MIT Press. and machines,” Journal of Experimental and Theoretical Collier, J. D./Hooker, C. A. (1996) ‘Complex organised dy- Artificial Intelligence, 5, 285–333. namical systems’, submitted. Bickhard, M. H./Terveen, L. (1995) Foundational issues in ar- Herfel, W. E. (1997) ‘How social constraints enable scientific tificial intelligence and cognitive science. Impasse and so- practice’. In: Herfel, W.E./Hooker, C.A. (eds.) Beyond rul- lution. Amsterdam: Elsevier Scientific. ing reason. Toward non-formal reason. In preparation. Christensen, W. D./Collier, J. D./Hooker, C. A. (1997) ‘Au- Hooker, C. A. (1975) ‘On Global Theories’, Philosophy of Sci- tonomy, adaptiveness and anticipation: towards founda- ence, 42, 152–79; reprinted in Hooker, C.A. A realistic the- tions for life and intelligence in complex, adaptive, ory of science, Albany: SUNY Press 1987. selforganising systems’, in preparation. Hooker, C. A. (1995) Reason, regulation and realism, Albany: Christensen, W. D./Hooker, C. A. (1997) ‘Very simple minds: SUNY Press. Towards systematic foundations for intelligent systems’,

Evolution and Cognition ❘ 48 ❘ 1997, Vol. 3, No. 1 Andy J. Clark

Evolutionary Epistemology and the Scientific Method*

1. An evolutionary epi- Abstract By imposing the selec- stemologist1 extends to tive paradigm onto the mind and knowledge that What is the proper attitude of the evolutionary episte- matter of basic cognitive account of adaptive mologist towards science? Should he regard science as orientation the evolu- nature and purpose usu- disclosing (or aiming to disclose) information con- tionary epistemologist ally reserved for the cerning the way the world is in itself, independently of can explain, at a stroke, explanation of gross the species-specific needs, bias and cognitive orienta- both startling achieve- physical characteristics; tion of the human life-form? Or should he conceive it ments and apparently characteristics such as as intrinsically limited and indelibly marked with the perverse inaptitudes. The sharp teeth, long necks or stamp of his own humanity? Either way there is a location of food sources is acute hearing. This exten- problem. If he adopts the first, objectivist, interpreta- understandably high on sion is hardly unwar- tion he faces the charge of hypocrisy; why does he not the honey-bee’s ‘list of ranted. Obviously, extend the results of his conjectures concerning cogni- priorities’ so we can make neither sharp teeth nor tion in other species to the enquiring animal, man? To good (albeit post facto) sharp ears will contribute make that extension, and to regard our scientific sense of the evolutionary to the proliferation of the knowledge as biased and limited in ways analogous development of the amaz- genes concerned if they to those attributed to the lower animals, is, however, ingly intricate dance rou- are not combined with an to breed a deeper discomfort. For if he adopts a spe- tines and mutual appropriate control sys- cies-specific, non-objectivist account of scientific interpretative capacities tem (call it ‘mind’) trans- knowledge then the status of the evolutionary conjec- used to communicate in- lating input into survival- ture itself is brought into question. For by what right formation concerning the does the evolutionary theorist then quantify over all en-hancing action. Sight distance and direction at evolved life-forms in formulating his general picture of prey or sound of preda- which food is to be found. of the relation between cognition and reality? tor must be suitably pro- Bee-dance, we conjecture, cessed or interpreted to is choreographed by the result in use of teeth or use of feet accordingly. selective process itself. Attention to the details of this Nature must be red not just in tooth and claw, but in process helps to explain not only what various ani- instinct and desire also if the teeth and claws are to mals can do, but also what they cannot do, cogni- be put to good use. tively speaking. The watershrew, for example, is On the plausible assumption that basic cognitive distinguished in the literature (LORENZ 1941/1962 orientation or instinct is as adoptively strategic as trans. p32) mainly in virtue of its incapacity to find gross bodily form the evolutionary epistemologist a shortcut (literally) to save its life. For having once brings his selective paradigm to bear on the issue of laid down a route to B from A via C, it can never the relationship between an animal’s environment progress to a direct route A → B even if the trip to C (as we recognize it) and its knowledge of that envi- involves a long, looping detour. This cognitive ‘de- ronment (as expressed in its observable behaviour). ficiency’ is explicable too (in a way more precisely He is concerned to account for the form and content detailed below) once we consider that it lives and of that knowledge in terms of the process of random reproduces perfectly successfully in default of any mutation, recombination and differential survival capacity to sustain the complex internal representa- to reproduce; a process known familiarly as natural tions of its, environment necessary to determine a selection. short-cut.

Evolution and Cognition ❘ 49 ❘ 1997, Vol. 3, No. 1 Andy J. Clark

A theoretical model adequate to the explanation of being knows it to be. One way to put the present of both cognitive achievements and inaptitudes is question is to ask whether science can intelligibly available to the evolutionary epistemologist. For the aspire to transcend this latter basic epistemological process of natural selection is characterised by a per- dualism and leave behind the random, species-bi- vasive dual aspect. On one hand, there is the pressure ased and cost conscious character of the process to survive; this brings the species into mental and which made the brains which do science. To suppose physical contact, over evolutionary time, with the it cannot is -to call into question the objective valid- environment in which it is competing. On the other ity of the evolutionary model (itself a branch of the hand, there are the limitations inherent in the rather biological sciences) itself and hence to intimate that minimal goal (viz. survival and reproduction) as- the naturalised angle on knowledge is a self-under- cribed to the selective process and responsible for the mining one. To suppose it can is to invite the accu- ‘contact with reality’ which it can support. For a pro- sation of ignoring our own epistemic situation as cess geared solely to survival may be expected to human beings3, for whence the phylogenetic discon- yield limbs and cognitive strategies alike which are tinuity between the knowledge of the lower animals geared to the special needs of a given being in a given and that attained by man? Is there safe water be- niche. Further slack with any notion of absolute ver- tween the Scylla of cognitive imperialism and the idicality enters with the observation that the whole Charybdis of cognitive relativism? And if there is, process is blind where by this is meant that the op- can it be consistently occupied by an evolutionary tions among which selection takes place are random epistemologist? To reach a decision we must take a mutants; beings whose particular mutated nature closer look at the scientific method itself. stands in no causal relation to the nature of the en- 2. On any plausible view of the scientific method vironment in which they are to be ‘tested’. And, fi- the conduct of science involves the performance of nally, the selective process is to be deemed sensitive some range of cognitive operations upon some to the non-optimising demands of cost-efficiency. If choice of data. The cognitive operations may include a neat approximation is both effective and econom- some kind of ranking of competing explanatory hy- ical it will be selected for against a more detailed but potheses in terms of the delicate balance between energy-intensive rival.3 Considerations of cost-effi- simplicity and comprehensiveness and utility (SOBER ciency, species-bias and random generation may calls this the trade-off between simplicity and fruit- thus explain the various inaptitudes of lower ani- fulness). And the data may be in the form of direct mals in the same theoretical context which explains observational reports or it may be more or less im- their successes. pregnated with theory depending perhaps on the The selective model thus briefly sketched, we may extent to which previously accepted hypotheses are now ask after the appropriate attitude of the evolu- assumed in the construction of the evidence upon tionary epistemologist towards his own (human) which some current claim is to be based. But no mat- conception of reality. Is he simply to extend the acter how intricate the web of intervening theory it count of basic cognitive capacities (instinct, degree will remain at root true to say that science takes ob- and nature of internal representation of the environ- servational reports as inputs, generates explanatory ment, input-action transformation strategies) in laws and models as outputs, and decides amongst lower animals to man himself? Or is science sup- competing laws and models by employing consider- posed somehow to be exempt from any repercus- ations of simplicity and fruitfulness. The explana- sions of the constraints on basic cognitive content tory laws and models which get accepted are discussed in 1 above? therefore subject to two sources of constraint. The In his dialings with the knowledge of other species first source lies with the observed phenomena them- the evolutionary theorist stands committed to what selves; a theory must be true to the facts. The second Donald CAMPBELL has called ‘an organism-environ- source lies with the structure of human (and perhaps ment dualism’ (CAMPBELL 1974, p449). This dualism all) rationality; a good theory should be simple, is both ontological and epistemological. It is onto- beautiful, comprehensive, suggestive and so forth. logical insofar as the world must be conceived as It would be natural to think that if some species- physically independent of mind; it is a mind-pro- based epistemological infection were to afflict sci- ducing, not a mind-produced, system. And it is epis- ence, the site of the infection would be with this temological insofar as it involves a dualism of second source of constraint. SOBER, indeed, has sug- knowledge and reality; how the universe is (in itself, gested that it might be unwarranted to believe that as it were) may always transcend how a given type any cognising being must share the kind of human

Evolution and Cognition ❘ 50 ❘ 1997, Vol. 3, No. 1 Evolutionary Epistemology and the Scientific Method rationality evinced by reference to the ‘parochial indirect route). In being faithful to the phenomena feature(s) of our own adaptive machinery’ (SOBER it maintains the original tie established by the selec- 1981, p117). And this could well include the kind tive process between the phenomena as known by a of heuristic constraints mentioned above. Against being.And the real world in which the being must this it may be held that some features (such as the live. Yet by dealing only with the phenomena which desire for simple hypotheses) may naturally result happen to be experienced by human beings it inher- from demands of informational economy derivable its also the species-specific interests and random ca- from the broad evolutionary bias towards cost-effi- prices of fate which combined to render accessible cient and prompt processing of data. This option those particular aspects of reality in that particular too is signposted by SOBER. I think, however, that it way. According to which of these two aspects of the is a mistake to see the scientific issue as essentially phenomena are stressed we get a more or less realistic bound up with our attitude to the heuristics at all. picture of the activity of science. For the prime site of empistemological infection 3. Just how much realism does the evolutionary must lie, I shall now argue, with the range and na- account require if it is not to collapse under its own ture of our access to phenomena and hence with weight? To get some idea we may consider a typical the first source of constraint on scientific theories. evolutionary claim. The claim is that Even if we are obectivists about the heuristics (tak- The hydrodynamics of sea-water, plus the ecolog- ing them as essential to any rational approach) this ical value of locomotion, have independently will not be sufficient to insulate science from the shaped fish, whale and walrus in a quite similar shock waves of the evolutionary account of our sen- fashion … But the jet-propelled squid reflects the sitivity to observational data. And if we hold the same Hydrodynamic principles in a quite differ- heuristics to be contingent, biased and unprivi- ent … shape. (CAMPBELL 1974, p447) leged too, then so much the worse for a traditional For such claims to be intelligible the evolutionary scientific realism. theorist must claim some right to employ our scien- The observation that one of the two major con- tific account of the hydrodynamics of sea-water as straints on scientific theory-building is to keep faith descriptive of the common reality to which both with the phenomena (to save the phenomena, as fish and squid are adapted. In some sense then the DUHEM puts it4) ought to be enough to transmit some world revealed by science must be justifiably taken of the basic evolutionary infection of bias and limi- to describe the mindindependent environment in tation to the body of scientific knowledge itself. For which adaption has occurred. to admit that science aims to explain and systema- Is such a role for science compatible with the evo- tise the phenomena is to tie the possible content of lutionary epistemologist’s account of cognitive lim- science to the range and nature of the phenomena itation and bias, supposing that account to be accessible to the particular biological organism des- extended to include our own sensitivity to phenom- ignated ‘man’. It is at just this point that any thor- ena? I think it is, and one way to show how this is so oughgoing scientific realism which would see is to focus on the idea of science as modelling an extra- science as penetrating to the unique noumenal roots experiential reality.5 of nature must founder against the evolutionary The notion of a model seems a particularly apt one rocks. For what is accessible to man (the bare obser- for the evolutionary epistemologist to employ. For vational data to which all theoretical constructions there is no implication that a model is a perfect rep- must answer) is determined by the very same contin- lica of what it models. Rather, we conveive a model gent, species biased and limited modes of sensory as bringing out particular features of some realworld access and basic processing to which the evolution- entity, perhaps to the exclusion of other features. ary scenario of section one unequivocally applies. And just what features are stressed will depend (a) on Even the instrumental augmentation of human sen- what information the modeller has at his disposal sory capacities must answer to some checks in gross and (b) on the particular needs and interests which observational accuracy or we would have no cause the model is designed to serve. These two features to accept such augmentation as in any way veridical. correspond satisfactorily to the contingency of the Science, for all its sophistication, thus looks unable range of realworld phenomena to which man has to transcend completely the humanity of its obser- direct observational access and to the particular kind vational base. of interest which man’s needs and the nature of the Science, thus conceived, partakes of the dual as- human brain allow him to have regarding the acces- pect of all evolved cognitive modes (albeit by a more sible realm.

Evolution and Cognition ❘ 51 ❘ 1997, Vol. 3, No. 1 Andy J. Clark

We may now clarify the nature of the proposed nitive orientation we happen to possess. Thus, to linkage between our scientific models and extra-ex- give a simple example, a program written in Cobol periential reality by introducing a special relation of would not prove a useful arrangement of informa- tolerance. Thus we may call a basic cognitive strategy tion for a computer which could only process com- or sensory modality (let P stand for this disjunction) mands coded in Basic. tolerated by the adaptive environment iff The strong conclusion to draw from the picture of P affords a means of classifying, predicting or re- science as aiming at tolerated models would be that acting to things and events which, when applied even at the ideal limit of human enquiry there might by beings of a given biological constitution in a be a plethora of available models all of which are given niche makes for successful (= survival en- observationally and Heuristically adequate (such a hancing) action in the world. conclusion is endorsed by PUTNAM 1983, pp1–25). And we may call a scientific theory P’ maximally tol- For our purposes, however, something weaker will erated by extra-experiential reality iff do. We may conclude simply that scientific enquiry P’ affords a means of conceiving of things and is still not the only possible ‘correct’ representation events which, when applied by beings of a given of reality even if relative to our cognitive constraints biological constitution, enables them to account and observational accessthere are no visible alterna- for successfully (= explain and perhaps predict) all tives. In other words, given the natural possibility of the phenomena accessible to a being so constitut- alternative life-styles, needs, capacities and cogni- ed. tive structures it makes no sense to identify our ideal P’ is then to be conceived as an ideal scientific scientific model of reality with the ultimate nature model in the sense of model outlined above. Such a of the world-in-itself. A model is still just a model, it model is then related to the real world it models by is not the ore true description worshipped by the virtue of the relation between the phenomena it metaphysical realist. Just because we do not regard explains and the world, such links being constituted our models as unique or necessary, however, does by the original tolerance relation between P and the not mean we may not regard them as valid represen- environment. The justification for calling P’ a tations, in the light of our interests and structure, of model of the real world thus rests squarely on the the available information. It is this combination of evolutionary justification for taking the phenom- cosmic contingency and limited objective validity ena which are modelled to be appropriate (if partial which allows the evolutionary theorist his scientific and biased) representations of the world they cope account of the common adaptive enviro nment with. while admitting the cognitive bias and limitations A true scientific theory, we may now say, would implied for man by the adaptive account itself. One be one that is maximally tolerated by the reality ac- interesting consequence of this analysis is that we cessible to man.6 And there will be an infinite. gra- must accept the possibility of alien epistemologists dation of tolerances between the minimal (perhaps even alien evolutionary epistemologists) (accounting for only a small number of phenomena) working successfully with a different model of the and the maximal (accounting for all the phenom- ‘common reality’ to our own! Such epistemologists ena). No maximally tolerated theory has yet been may even diagnose man’s models as a natural and found, and perhaps none ever will be. But the crucial explicable outcome of our own biological nature as point is this; even if one were found, still the reflex- it appears to their science. We, of course, might do ivity of the formulation of the tolerance relation (its the same for them! Each scientific model would relativisation to human and contingent capacities) therefore be sufficiently powerful to embrace the would rob it of any claim to be the one unique meta- working of the other. The questionas to which physical truth fated to be agreed by all rational be- model is the correct one would never be raised. ings. 4. The question finally arises whether the spectre The intelligible goal of science, we may now say, of the world-in-itself, apparently attendant upon the is not the description of the world-in-itself but the epistemological dualism diagnosed in 1 above, has production of more and more highly tolerated mod- been successfully exercised or merely relocated? For els of the world we find around us. And a model is, to adopt the quasi-realistic notion of science as aim- ultimately, nothing more or less than a useful ar- ing to produce tolerated models is to invite the phi- rangement of information. Just what arrangements losopher’s retort ‘models of what?’. Two courses are of information we find useful will depend on our open to the evolutionary epistemologist here. He human needs and capacities and the particular cog- may allow that all such models are models of the one

Evolution and Cognition ❘ 52 ❘ 1997, Vol. 3, No. 1 Evolutionary Epistemology and the Scientific Method

(alas indecribable) objective, mind-independent re- stances described by the statement actually obtain, ality to which all beings are variously adapted. Or he we may still make sense of the minimal evolutionary may dig in his heels and refuse to countenance any claim viz. that various models may succeed in coping conception of reality save that of whatever is said to with reality (we may observe the survival and exist by some successful model (be it a human or achievements of beings employing such models). non-human one). So either we ghive up the very idea But what further evidence could there be to warrant of the world-in-itself (as RORTY and DAVIDSON urge us us in assenting not just to a plethora of models but to do7) and replace it with the notion of multiple to there being one, ultimate, unknowable way the valid species-specific descriptions whose objects are world actually is beyond how it appears in the vari- determined by the descriptions themselves, or we ous models we, or any other sentient being, might retain the idea of the world-in-it self as a bare nou- construct? menal something = X which somehow supervenes In choosing therefore to give up the notion of the (or maybe transcends) the totality of possible de- world-in-itself the evolutionary epistemologist must scriptions of it. Whichever we choose, the divorce of simultaneously ‘resist RORTY’s alternative descrip- science from the description of noumenal reality is tion of reality as whatever human beings can agree ratified. Of the two oprtions suggested, I find myself at a given time exists. (See RORTY 1972, pp661–663). attracted to the more austere alternative of dropping For human beings, we have seen, can recognise the the notion of the world-in-itself entirely. The dual- bias and contingency of their own descriptions of ism of organism and environment would then re- reality from a position within biological science. To main as a part of the theoretical model of biological simply identify ‘the world’ with the world of man is, science, which model itself would be regarded as we may be sure, mere anthropomorphic conceit. The non-unique and cosmically unprivileged. Bot there alternative, recommended in this paper, is to em- would be no need to assume, in addition to this, that brace the difficulty of admitting multiple valid de- all the possible models of reality themselves stand on scriptions and to assert that to be is to be perspec- one side of a dualism of models and the world-in- tivally. itself. Aside from the general thought that the idea Finally, let us observe that the denial of any privi- of the worl-in-itself can now be seen as theoretically leged status to the model of human science renders spurious to the evolutionary account (which re- our whole account of the tolerance relation itself quires only the acceptance of an organism-environ- harmlessly self-referential. For our theoretical models met dualism within a given explanatory model are ultimately justified by keeping faith with observ- which takes science to provide the necessary account able phenomena. These observable phenomena are, of the environment) there are two reasons which tell on the theoretical model of evolutionary theory, ac- in favour of abandoning the notion. The first is the cessed and characterised by sensory capacities and ba- recent and influental polemic launched by Hilary sic forms of processing which have stood the test of PUTNAM (see notes 3 and 6) against the notion of survival. They are hence assumed to constitute a spe- there being one true (if unknown) description of cies-valid arrangement of information concerning how the world is. Such a beliefe, PUTNAM argues, can the external world. Theory is thus justified by theory be shown to be false on model-theoretic grounds in a cosy epistemological circle of the kind sometimes alons. The second reason has to do with the intelli- described as ‘virtuously’.9 A direct consequence of this gibility of the very idea of the world-in-itself. For is that our belief in the relation of tolerance is itself such a world looks to be necessarily indescribable justified only as a tolerated belief. It is thus an accept- (description implying point of view, cognitive bias able representation, for beings of our knowledge and and so forth). But to claim that something about constitution, of the relation of sense and thought to which we can necessarily say nothing exists may be an external reality. But we may not elevate the scien- to claim nothing which we can properly grasp at all. tific model which employs the idea of tolerance to the For any such claim looks dis- level of a unique or metaphys- tinctly dubious in the light of ically privileged representa- DUMMET’s recent investiga- Author’s address tion of the relation of thought tions into meaning. If we ac- and sense to the world. The Clark Andy, School of Cognitive &Comput- cept, with DUMMET8 , that evolutionary epistemologist ing Sciences, University of Sussex Falmer, meaning attaches to state- dare not claim to possess the Brighton BN1 9QH, United Kingdom. ments in virtue of our capacity one true account of the rela- Email: [email protected] to recognise when the circum- tion between mind and the

Evolution and Cognition ❘ 53 ❘ 1997, Vol. 3, No. 1 Andy J. Clark material realm. The best he can do is to say that it is Acknowledgements an account, acceptable to us, and one which avoids the metaphysical excesses of a traditional scientific Thanks to A. BRENNAN for suggesting the relevance realism. As Clive JAMES once observed: of the notion of a model to an evolutionary view of “There are limits to the altitude that can be science. achieved by hauling on one’s own bootstraps” (JAMES 1981, p35).

Notes 6 This corresonds with PUTNAM’s idea of truth as the ideal end-point of the series of warrantedly assertible claims con- * Reprinted by courtesy of Philosophica. cerning the nature of reality which human beings could in 1 Examples of work in Evolutionary Epistemology would be principle come to make. See PUTNAM (1981). CAMPBELL (1974), LORENZ (1941). Or TENNANT (1983). 7 See RORTY (1972 and 1980). Also DAVIDSON (1974). 2 These constraints are signposted by both TENNANT (1983) 8 See especially DUMMETT (1978). and CAMPBELL (1974) and also by SOBER (1981). 9 The terminology is, I think, due to Rescher. A virtuous circle 3 The very same accusation is made by H. PUTNAM against his is one which provides an improvement in understanding earlier metaphysically realist self in the introduction to in spite of any element of self-reference involved. Thus, in PUTNAM (1983). the present case, we learn, by the application of our under- 4 See DUHEM (1974). standing, something of the reasons why we might trust our 5 This kind of account of science is most fully developed in understanding to reveal something of the world in which VAN FRAASSEN (1980). we evolved.

References losophie 15. Translated in: Bertalanffy, L. von/Rapoport, A. (eds) General Systems, Ann Arbor 1962. Campbell, D. T. (1974) Evolutionary epistemology. In: Putnam, H. (1981) Reason, Truth and History, C.U.P., 3 4. Schlipp, P. A. (Ed.), The philosophy of Karl Popper. LaSalle, Putnam, H. (1983) Realism and Reason. C.U.P., vii, xi IL: Open Court, 413–63. Rorty, R. (1972) The world well lost. (WWL) J. PhiL vol. LXIY, Davidson, D. (1974) On the very idea of a conceptual scheme. no. lg Oct.1972 In: Proceedings of the American Philosophical Associa- Rorty, R. (1980) Philosophy and the Mirror of Nature. Ox- tion, vol. 47, pp. 5–20. ford: Blackwell. Duhem, P. (1974) The Aim and Structure of Scientific Theory. Sober, E. (1981) The Evolution of Rationality’. Synthese, vol. New York: Atheneum. 46, no. 1. Dummett, M. (1978) The philosophical basis of intuitonistic Tennant, N. (1983) A defence of Evolutionary Epistemology. logic. In: Truth and Other Enigmas. London: Duckworth. Theoria vol. IL 1983, part 1. James, C. (1981) Unreliable Memoirs. London: Picador. van Fraassen, B. (1980) The Scientific Image. Oxford: Claren- Lorenz, K. (1941/1962) Kant’s Lehre vom Apriorischen im Li- don Press. chte gegenwärtiger Biologie. In: Blätter für Deutsche Phi-

Evolution and Cognition ❘ 54 ❘ 1997, Vol. 3, No. 1 Ed Constant

Comments on Donald T. Campbell’s “From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity”

irst, let me say that I Abstract of “rationality” it might F am deeply pleased contradict is one claiming that the editor‘s of Evolu- In this comment I discuss CAMPBELL’s insistence on that valid knowledge can tion and Cognition have the ‘blind’ character of variation in all evolutionary only obtain from absolute chosen to publish this variation-selective retention processes, which has of- deductive completeness, penultimate expression ten been misunderstood, and the similarities and dif- a view that no naturalized of Donald CAMPBELL‘s ferences between communities of scientists and other epistemology dare hold. oeuvre. It is a pleasure believing communities. The key to CAMPBELL’s profoundly compromised true position is the idea of by the realization that this is the last of his remark- a nested hierarchy of variation-selective retention able work we are likely to see. processes. The structure of this hierarchy permits vi- In my comments, I’d like to focus on two aspects carious exploration and selection, that is, “processes of this synthetic statement of the issues and themata which shortcut a more full blind-variation-and-se- that have concerned CAMPBELL over the last forty lective-retention process,” and “are themselves in- years. One aspect, the importance in all evolutionary ductive achievements, containing wisdom about the variation-selective retention processes of some form environment achieved originally by blind variation of “blind” variation, is from near the beginning of and selective retention.” (ibid., p8) Such vicarious his work in evolutionary epistemology. The other, exploration and selection can be executed by many the similarities and differences between communi- means, have many expressions: (fallible) vision it- ties of scientists as knowers, and other believing self, our beliefs (proofs) about Euclidean space, Bool- communities, especially archaic religious communi- ean logic, the theory of the conservation of energy, ties, is a relatively recent development. Both, oddly, taste and smell perceptions, HAMILTONIAN Quarteri- are somewhat muted in this presentation. ons (PICKERING 1995). All serve us as means to vicar- Probably no other single issue so plagued, or, in ious exploration and vicarious selection: they permit my view, generated so much misunderstanding of, a nested hierarchy of conscious variation and vicari- CAMPBELL’s evolutionary epistemology as his use of ous selection processes that approximate “rational” “blind” variation as the source of the raw material deduction, intentional means-ends analysis, and on which selective retention (or, later, a selection rigorous evaluation. process) could operate. The long list of synonyms All that CAMPBELL insists upon is that for genuine and circumlocutions CAMPBELL tried—“blind,” “ran- novelty—that which has not been produced by dom,” “unforesighted,” “going beyond the limits of prior blind variation and selective retention—to en- foresight or prescience” (this volume, p8)—bears ter such processes, some form of blind variation witness to the difficulty of making this notion clear. must occur at some, usually very low, level in the Critics, unfairly I believe, and for reasons I still can- hierarchy. His proof of this contention is a simple not fathom, set the idea of blindness or randomness reductio ad absurdum: If not blind or random varia- in the generation of variants in sharp juxtaposition tion, then either some form of foresightedness, to “rationality” and “intentionality,” and therefore which only can result by extrapolation or deduction labeled CAMPBELL’s model “irrational.” But as used in from previously achieved knowledge, or supernatu- CAMPBELL’s theory of a “nested hierarchy of blind ral revelation. variation-selective retention processes,” “blindness” I suspect part of the difficulty in explicating the entails no such juxtaposition: the only conception character of these blind variation processes was

Evolution and Cognition ❘ 55 ❘ 1997, Vol. 3, No. 1 Ed Constant

CAMPBELL’s reliance on introspective accounts of cre- focus on the social character of believing collectivi- ativity, most notably Henri POINCARÉ’s (1921). ties rather than on their beliefs. POINCARÉ, and also Jacques HADAMARD (1945)— Such an alternative emphasis reveals much stron- whose theory CAMPBELL saw as simply a rehash of ger parallels, to the point of identity, not only be- POINCARÉ’s—, both postulated, via introspection, tween archaic religions and science, but also among some unconscious, random association or combina- religious communities, science, and all other quasi- tion of “ideas,” which were then subjected to some altruistic social collectivities. The basic structure for equally unconscious preliminary selection, perhaps all such collectivities (although he does not make on “aesthetic” criteria. Only after this unconscious such a broad claim) is laid out by P. Steven SANGREN random variation-selective retention process might (1988, 1991) in his studies of the MA TSU cult in Tai- “thoughts” or concepts be presented for conscious wan and adjacent mainland Fukien province. Ac- evaluation. Such introspective accounts are quite cording to SANGREN’s description, MA TSU, a naturally suspect, for CAMPBELL’s evolutionary epis- presumably historical figure who saved Taiwan from temology as for the self-reported verbal “protocols” invasion in the ninth century, serves as an “alien- of people’s problem-solving activities that are so of- ated representation” of group solidarity and efficacy: ten cited as justification for artificial intelligence the ancient victory is attributed to the intervention simulations. of the goddess, not to the altruistic cooperation of In a sense, CAMPBELL’s necessary reliance on intro- her devotees. Thus the power and solidarity of the spective accounts parallels DARWIN’s plight in simply present-day community of believers is alienated to having to postulate some mechanism for the inher- some remote, and, not coincidentally, incorruptible itance of advantageous mutations without knowing realm. precisely what that mechanism might be, which of In temple rituals, but especially in pilgrimages, course, later, reconstructed MENDELIAN genetics af- devotees learn the ritual production of selves in ap- forded (BRANNIGAN 1979). It may well be that con- propriate communal idiom, thereby socially produc- nectionist models of cognition, or neuron nets, with ing themselves, and socially reproducing the their strong probabilistic foundation, will provide community itself and its alienated representation. CAMPBELL’s MENDELIANISM—or they might not. What Only specific, temporal reifications of the deity’s will matters, however, is that above this minimal level in or beneficence are subject to real-time human inter- the nested hierarchy of variation-selective retention pretation, or misinterpretation or misuse, not her processes lie strata upon strata of conscious variation transcendental existence or power. This social robust- and vicarious selection processes, the only rationality ness in the face of human frailty and mendacity is, of a naturalized epistemology can have: as CAMPBELL course, the answer to CAMPBELL’s parenthetical query argues, all “rationality” is itself an inductive achieve- about archaic religions, “Why were not the force of ment, forever corrigible and hypothetical. custom plus interpersonal reinforcements sufficient My other comments relate to the other end of the without such cosmologies?” (ibid., p22) More strik- scale, as it were, both temporally and dimensionally: ingly, a little reflection clearly reveals that what spe- from the neuronal to the communal. CAMPBELL long cific forms these alienated representations might take argued that all the objectivity to be found in science is completely irrelevant to their latent social func- was to be found in its communal practices, not in the tion: as CAMPBELL puts it, “The details of these super- cognition, morality, or behavior of individual scien- natural cosmologies were extremely heterogeneous … tists. In his extended discussion of the parallels and (This is the puzzle of diversity…).” (ibid., p22) non-parallels between archaic forms of belief (reli- This form-independence of alienated representa- gion), which he, correctly I believe, sees as serving tions suggests that the form does not necessarily the latent function of socially producing coopera- even have to be supernatural. Any belief system that tion among genetic competitors, and the social sys- provides an alienated representation capable of sus- tem of science, CAMPBELL focuses mostly on taining communal solidarity, that performs the la- comparative belief systems, tent function of inducing their propagation (from BOYD quasi-altruistic cooperation and RICHERSON, density-de- Author’s address among community adher- pendent replication), and ents, will do, including ideol- Ed Constant, Dept. of History, Carnegie their epistemology. An alter- ogies: “the market,” MARXIST- Mellon University, Pittsburgh, PA 1521- native, surely latent in CAMP- LENINIST dialectical material- 3890. BELL’s own discussion, is to ism, or “the scientific

Evolution and Cognition ❘ 56 ❘ 1997, Vol. 3, No. 1 Comments on Donald T. Campbell’s “From Evolutionary Epistemology Via Selection Theory to a Sociology of Scientific Validity” method.” The ideology of science, its exaltation of religious ceremony, is always contestable. Novel rit- method, and its normative structure, all historical, uals—new science, new experimental methods— contingent products of the scientific and social rev- quickly set off disputes about fidelity to the ideal: olutions of the seventeenth century, albeit extended witness the many versions of MENDEL, or the ugly in later centuries far beyond the bounds of proper fights among morphologists, pheneticists, and cla- English “gentlemen,” (SHAPIN 1988, 1994) are iden- dists about who was “really” the proper DARWINIAN tical in structure and function to those in any other (HULL 1988). If one were skeptical about the real ex- social collectivity. istence of some spatiotemporally universal, immu- CAMPBELL, in his characterization of “experiments table, invariant scientific method, as distinct from as divination rituals,” (ibid., p29) properly recog- its ritual reifications in scientific practice (as one nizes the ritual nature of “producing” science. What might be skeptical about the supernatural existence he doesn’t articulate quite as clearly is that these rit- of MA TSU or any other god), then quarrels about uals produce both science—reified bits of knowl- what constitutes “good science,” what specific be- edge—and scientists. On the model sketched above, haviors and phenomena are to be acknowledged by scientists, in their professional rituals—running ex- the scientific community as expressing the veridical periments, writing reports and papers, refereeing scientific method, are both to be expected and abso- others’ papers, reading papers at professional society lutely critical to the survival of the enterprise—and meetings, serving on advisory panels—learn as to a sociology of scientific validity. though pilgrims to produce themselves in commu- I think it is the great virtue of Donald CAMPBELL’s nal idiom. In so doing, they also reproduce the com- work, that all the while acknowledging the quintes- munity itself and its alienated representation: the sentially social character of science, he also hewed awesome power of “the scientific method.” And it is, hard and fast to the argument that the spatiotempo- of course, that very deity, that method, that, in SAN- rally particular rituals of science do include validity- GREN’s happy phrase, “mutually authenticates” the enhancing practices through which the world as it bits of science, and the scientists, so produced. is can at least co-select our beliefs about it: it is his Because the alienated representation of scientific always presumptive, corrigible and hypothetical, method is conjured anew in each scientific ritual, “fallibilist realism.” the fidelity of the conjuring, like the orthodoxy of

References gy. In: Schlipp, P. A. (ed) The Library of Living Philoso- phers, Volume XIV, Book I: The Philosophy of Karl Popper, Brannigan, A. (1979) The Reification of Mendel. Social Stud- Open Court, La Salle, IL, pp. 413–463, section 3.6 Mne- ies of Science 9: 423–454. monically supported thought, pp. 427–431. Hadamard, J. (1945) An Essay on the Psychology of Inven- Sangren, P. S. (1988) History and the Rhetoric of Legitimacy: tion in the Mathematical Field, Princeton University Press, The Ma Tsu Cult of Taiwan. Comparative Studies of Soci- Princeton, NJ. ety and History 30:674–697. Hull, D. L. (1988) Science as a Process: An Evolutionary Ac- Sangren, P. S. (1991) Dialectics of Alienation: Individuals count of the Social and Conceptual Development of Sci- and Collectivities in Chinese Religion. Man (N.S.) 26: 68– ence. University of Chicago Press, Chicago. 69. Pickering, A. (1995) The Mangle of Practice: Time, Agency, Shapin, S. (1988) The House of Experiment in Seventeenth- and Science. University of Chicago Press, Chicago, IL. pp. Century England. Isis 79: 373–404. 113–156. Shapin, S. (1994) A Social History of Truth: Civility and Sci- Poincaré, H. (1921) Mathematical Creation. In The Founda- ence in Seventeenth Century England. University of Chi- tions of Science. Science Press, New York, pp. 383-94; cited cago Press, Chicago. in Campbell, Donald T. (1974) Evolutionary Epistemolo-

Evolution and Cognition ❘ 57 ❘ 1997, Vol. 3, No. 1 Steve Fuller

Campbell’s Failed Cultural Materialism

he longer that Abstract for CAMPBELL is not T Donald CAMPBELL merely a synonym for developed his evolution- ‘decision’ but refers liter- CAMPBELL’s evolutionary epistemology had the poten- ary epistemology, the tial to treat the ‘selection environment’ for science as ally to background con- more he was inclined to bearing more than a metaphorical resemblance to the straints on the efficacy of regard his reliance on material environments as it features in biology.His decisions of which the DARWINIAN natural selec- concern for knowledge ‘vehicles’ and ‘co-selectors’ relevant agents may be tion theory as a sugges- pointed in that direction. However, CAMPBELL failed only dimly, if at all, tive metaphor rather to follow through on his materialist promise because aware. than a commitment to he could not envisage that science may become worse I shall argue that CAMP- biology as the metatheo- as it comes to dominate its material environment. I BELL fell far short of his retic framework for trace this to a prejudice he inherited from SPENCER. materialist promise. Nev- explaining the emer- Here guidance could have been sought in economic ertheless, CAMPBELL‘s la- gence and growth of sci- history’s grappling with the concept of ‘development’, tent ecologism should be entific knowledge. especially in a world where some develop at the ex- commended for enabling CAMPBELL (1997) is quite pense of others. I trace some of the implications of this us to see more clearly the explicit on this point. ‘path not taken’, while granting that CAMPBELL’s evo- strengths and weaknesses However, I would urge lutionary epistemology was much more on the right of his evolutionary episte- that CAMPBELL let himself track than, say, KUHN’s more famous version. mology. A good point of off the hook too easily contrast here is with the and that in fact he implicitly accepted enough of somewhat different appeals to evolutionary theory the materialistic assumptions of contemporary biol- that have been made by Thomas KUHN and Niklas ogy to be seen as treating the ‘evolution’ side of evo- LUHMANN, both of whom have been associated with lutionary epistemology at least as literally as, say, ‘autopoietic’ or ‘self-organizational’ models of scien- Marvin HARRIS (1979) does in his pursuit of a ‘cul- tific change. Central to both KUHN’s and LUHMANN’s tural materialist’ paradigm in anthropology. This conceptions is an irreversible insulation, or ‘auton- point is easily obscured if we immediately confer omization’, of science from the larger social environ- metaphorical status on appeals to evolution as soon ment at a certain point in its history (say, with the as the ‘selected’ entities or properties are not founding of the Royal Society)—not merely in the straightforwardly related to entities or properties sense of science being protected from ambient social recognized by current genetic theory (hence, the pressures but more importantly in terms of science’s vigorous but inconclusive efforts by Richard DAWK- development being defined exclusively in the scien- INS and Daniel DENNETT to promote ‘memes’ as tists’ own terms. In effect, autonomization enables genetically respectable units of cultural selection). science to simulate the frictionless medium of However, this requirement is needlessly strict. Like thought that philosophers since PLATO have consid- HARRIS, CAMPBELL accepted a literal form of what ered ideal to the pursuit of knowledge. Conse- might be called ‘ecologism’, namely, that selection quently, neither KUHN nor LUHMANN have much to occurs on the material—or, in the case of knowl- say about the vehicular requirements of scientific edge, what CAMPBELL tended to call ‘vehicular’— knowledge or even those aspects of society that have characteristics rather than ideational ones, which in been, in CAMPBELL’s terms, ‘co-selected’ with scien- turn reflects the fact that, in the final analysis, the tific theories. Rather, one is presented with a seem- environment provides the non-negotiable context ingly irreversible, if strictly non-teleological, story of to which prospective inhabitors must conform or functional differentiation of the scientific enter- else simply not survive. In this respect, ‘selection’ prise, as inquirers encounter impasses in the day-to-

Evolution and Cognition ❘ 58 ❘ 1997, Vol. 3, No. 1 Campbell’s Failed Cultural Materialism day business of puzzle-solving that force them to di- able trajectory that encompasses the entire world. vide their efforts in order to clarify the overall direc- This is, of course, compatible with a considerable tion of their inquiries. Here evolution becomes little degree of variation across local environments, which more than a metaphor in which ‘selection’ results can in turn be explained by the differential impact from the expectations of inquirers being con- of earlier events of common ancestry. CAMPBELL’s founded by their experiences, which in turn cause evolutionary epistemology is most naturally read in them to take a fateful decision. this fashion, but his own rather SPENCERIAN cultural An important sign of evolution’s metaphorical prejudices prevented his model from serving as an status in these autopoietic models is that it is never exemplar of this ‘third way’. Before revealing the na- clear whether the process in question is supposed to ture of these prejudices, let us first turn to the field be unique or repeatable. For example, is KUHN’s (1970) of economic history, especially its long-standing theory of scientific change supposed to model the concern with modeling ‘uneven development’ (see stages through which the history of science as a sin- e.g., BLOMSTROM/HETTNE 1984), which indeed does gular global phenomenon passes, or rather the offer guidance on the third way. phases through which specific sciences pass in any The concern in question is an outgrowth of LE- of a variety of times and places? Were KUHN to have NIN’s theory of imperialism, according to which the intended the former, his model would have approx- logic of capital expansion enables the first industri- imated a literal application of evolution to episte- alizing nations to force latecomers into a position of mology. But in that case, he would have been forced subservience, either as mere providers of raw mate- to confront how the initial scientization of certain rials or as consumers of goods produced by the early fields at certain times and places—specifically, ex- industrializers. In this way, a ‘core-periphery’ rela- perimental physics in 17th century Western Eu- tionship is perpetuated—that is, until the Commu- rope—set constraints on later developments, even in nist Revolution occurs. Whatever its own remote fields, times and places. In other words, how shortcomings, LENIN’s theory managed to eliminate does the prior existence of certain forms of inquiry the ‘metaphysical residue’ in Marxist political econ- constitute the environment against which subse- omy, namely, the idea that the essence of economic quent forms of inquiry are selected? Clearly, KUHN development was first manifested in Western Eu- has not been read in this way, but rather as having rope’s transition from feudalism to capitalism and advanced a multiply repeatable, perhaps even uni- would subsequently be reproduced by the rest of the versalizable, model of scientific change that is just as world, largely without change. (The most systematic relevant to, say, sociological inquiry in the 1960s as scholarly elaboration of LENIN‘s perspective—albeit physical inquiry in the 1660s. It is ironic, given without LENIN’s apocalyptic outcome—is world-sys- KUHN‘s reputation for having ‘historicized’ the his- tems theory: see e.g. WALLERSTEIN 1991). Even LENIN‘s tory and philosophy of science, that his model ‘bourgeois’ critics, while questioning the inevitabil- should be applied in such a mindlessly ahistorical ity of imperialism, nevertheless conceded that one fashion. However, precisely because CAMPBELL built needs to consider the world-historic environment in the relevant material constraints into his evolu- that defines what often turns out to be a narrow band tionary epistemology, his model is unlikely to suffer of possibilities within which development can occur from the same degree of popularity. Indeed, it is in a given time and place. For example, while late amazing that CAMPBELL never seriously pursued the developers have often industrialized more quickly, implications of his materialism. Had he done so, we they have typically to supply local substitutes for the would have seen much greater discussion of the de- institutions of the originators (GERSCHENKRON 1962). mographic, economic and technological dimen- KUHN’s theory of scientific change—at least as in- sions of the scientific enterprise. terpreted by his admirers—retains just the sort of Today it is commonly assumed that because sci- metaphysical residue that LENIN endeavored to ence does not follow the same path in all times and purge from MARX’s theory. Not surprisingly, KUHN places, it must therefore follow different, unrelated has been eagerly embraced by the most famous re- paths that can be explained only by citing local fac- cent metaphysician of economic development, tors. This inference, characteristic of the ‘ethno- Francis FUKUYAMA (1992, pp80–81, 352–353), who graphic’ or ‘postmodern’ turn in science studies, regards the inviolate ‘logic’ of scientific inquiry as constitutes a false dilemma. It overlooks that the described by KUHN to be the motor of economic evolution of scientific knowledge may be exactly like progress that will eventually lift all nations to the biological evolution in being a single and unrepeat- standard of living currently enjoyed by the United

Evolution and Cognition ❘ 59 ❘ 1997, Vol. 3, No. 1 Steve Fuller

States. One need not be a LENINIST to find evolution- the scientific enterprise so that more of its partici- arily relevant reasons for doubting FUKUYAMA’s prog- pants would have a better chance of having their nosis. The average American already consumes six contributions recognized. On the contrary, PRICE times more energy than the average person. Barring stressed ‘economies of scale’ as improving the over- a breakthrough in the design of artificial environ- all quality of science. In other words, an ever widen- ments, that alone precludes the US from providing ing field of competitors would keep all scientists on the basis for an ecologically sustainable political their toes, which would in turn ensure their best per- economy. Interestingly, the same applies to Ameri- formances, leading to an increase in the overall per- can science as a model for world science—and not centage of quality publications—not to mention an just at the level of metaphor. However, compared increase in the absolute number of unrecognized sci- with the idea of limits to economic growth (ARNDT entists ‘wasted’ by the system. MERTON went so far as 1978), limits to scientific growth has been rarely to argue that if indeed genuine contributions to touched upon (RESCHER 1984). However, a telling knowledge are likely to go unrecognized, disadvan- precedent was indirectly set by the man who first taged scientists should try to get advantaged scien- drew the Little Science/Big Science distinction, tists to promote the findings as their own. Derek de Solla PRICE. PRICE and MERTON clearly envisaged a ‘selection- In his tireless search for ‘science indicators’, PRICE ist’ sociology of science, albeit a crude one that let (1978, p87) found that the economic indicator most environmental vicissitudes do most of the work that closely correlated with scientific productivity (pa- rationalist philosophers of science would assign to pers per scientist) was electrical energy consumption explicitly formulated epistemic aims and norma- (kilowatt-hours per capita), which is in turn related tively acceptable constraints on their pursuit. The to economic productivity (GNP per capita) by a closest that PRICE got to specifying the ‘ends of sci- power of 3/2. In a nutshell, the US consumes 25–35% ence’ was in terms of the indefinite accumulation of of the world’s energy resources and produces a com- highly cited scientific articles. This metric of scien- parable portion of its scientific papers. (See OLSEN tific progress is about as primitive as measurements 1992 for recent confirmation.) The finding high- of wealth prior to Adam Smith, when it was com- lights the extent to which science itself consumes monly thought that a nation increased its wealth resources, thereby reversing the economists’ image simply by accumulating more bullion in its treasury of science as exclusively a ‘factor of production’, a than (and often at the expense of) its neighbors. perpetual motion machine. By drawing attention to Smith proposed that the range of living standards consumption patterns, PRICE unwittingly brought and the organization of labor be taken as alternative into the focus the question of waste in the scientific indicators of a nation’s overall wealth. We have yet enterprise. Consider Michael POLANYI’s famous defi- to take the ‘SMITHIAN Turn’ in evolutionary episte- nition of free inquiry as the ability to waste resources mology, whereby the progressiveness of science with impunity in the name of truth. These resources would be judged more by the quality of its people include not only money, paper, machines and en- than its papers. Taking the turn would require merg- ergy but also people. Here I am alluding to what Rob- ing the ‘selection environment’ for science with that ert MERTON (1973, pp439–459) has euphemistically of the surrounding political economy. Consider the called science’s ‘principle of cumulative advantage’ case of science’s ‘invisible’ tendency toward elitism, (also known as the ‘Matthew Effect’), which, like noted above. Behind it may be the fact that as re- capitalism’s ‘invisible hand’, is presented as a situa- search becomes more ‘advanced’, in the sense of spe- tion that appears cruel to individuals but eventuates cialized, it becomes more expensive in almost every in long term systemic benefits. Accordingly, in sci- respect and hence subject to various problems of ence, those who are rewarded earlier (say, in terms scarcity (of skills, equipment, etc.), which are then of elite training, early posts and publications) are sociologically resolved by concentrating resources in rewarded more and longer, with the difference be- a few researchers and institutions. Yet, when this tween the initially advantaged and disadvantaged relatively unproblematic observation is set against increasing rapidly over time, such that two-thirds of the backdrop of science’s contributions to labor-sav- scientists give up on publishing altogether after their ing devices and other forms of technical efficiency, first article. However, Social DARWINISTS of science it would seem that science helps economize in the policy that they were, neither MERTON nor PRICE ever rest of society in order to make room for more lavish suggested that this tendency should discourage the expenditure on its own activities, as measured by not recruitment of new scientists, thereby downsizing only the number of researchers but also the time the

Evolution and Cognition ❘ 60 ❘ 1997, Vol. 3, No. 1 Campbell’s Failed Cultural Materialism general population must invest in acquiring creden- what a scientist knows about a given field may be tials for any line of work. Certainly, this is one (un- based almost entirely on whom she trusts for reliable charitable) way of interpreting the growing presence information. Again seen in strictly economic terms, of ‘knowledge-intensive’ activities in the economy this puts us in the realm of brokered exchanges, at large (STEHR 1994). It may also capture Moritz whereby potential buyers no longer know first-hand SCHLICK’s (1974, pp94–101) brand of evolutionary what is for sale and hence must rely on third parties. naturalism, whereby the advance of civilization is A situation of this sort arises in the ‘informal econo- marked by science gradually reversing its role from mies’ of the former Soviet Bloc countries, which being a means to other ends in life to an end pursued have experienced the breakdown of socialism within itself, and ultimately the only end worthy of un- out yet installing market mechanisms that reliably conditional pursuit. transmit PRICE information. Do we really want to call From an economic standpoint, to regard science such a state-of-affairs ‘functional’? as an ‘unconditional pursuit’ is to forgo any scrutiny Earlier I alluded to ‘SPENCERIAN prejudices’ that of its return on investment. Many would argue that prevented CAMPBELL from following through on the this is as it should be, since science’s returns are typ- materialism implied by the significance given to ‘ve- ically ‘intrinsic’, ‘deferred’ or ‘diffuse’ in ways that hicles’ and ‘co-selection’ in his evolutionary episte- defy normal accounting procedures. However, the mology. These prejudices appear most clearly in very use of the enquoted terms concedes the elusive- section 7 of CAMPBELL (1997), which portrays reli- ness of the boundaries that define science’s selection gious authority as a principle of social solidarity that environment. But there is more for the economist to enables a robust form of ‘adaptive cultural evolution’ worry about here. Consider the expression ‘func- that is nevertheless distinct from one that encour- tional differentiation’, which is supposed to struc- ages science. Like Herbert SPENCER, CAMPBELL held ture the ever expanding scientific enterprise. When that the coercively induced need to conform with the metaphorical grounds shift from embryology to tradition was the ultimate obstacle to genuine scien- political economy, we are asked to imagine an in- tific and social progress. This belief informed not creasingly specialized division of labor in science. At only CAMPBELL’s reading of history but also his de- first glance, this suggests that the pursuit of inquiry sign of experiments, such that he could treat the is becoming more efficient, as each scientist does problem of intergenerational knowledge transmis- what only she can do and then collaborates with her sion as a species of small group conformity studies fellows to get the entire job done. Adam Smith (JACOBS/CAMPBELL 1961). Moreover, also like SPENCER would be pleased. And while such an image may well (though less reflectively), CAMPBELL seemed to be- capture the microdynamics of the research team, it lieve that scientific and economic rationality were has virtually no bearing on the sense of ‘division of co-selected, so that the growth in wealth that is char- labor’ that operates at the systemic level of scientific acteristic of economically rational regimes—that is, activity. At this level the issue turns more on entitle- capitalist ones—becomes the ‘natural’ environment ment than efficiency, namely, the settlement of ju- for sustaining the growth of science. (The giveaway risdictional disputes between competing inquirers, here is CAMPBELL’s en passant remarks about the which then becomes the basis for relations of mutual ‘wasteful funerals’ in religiously based regimes.) deference and trust. In some evolutionary models, Thus, in neither SPENCER nor CAMPBELL do we find the importance of resolving these disputes is made ecologically inspired concerns about whether sci- explicit, as when a KUHNIAN paradigm subdivides to ence could become ‘too expensive’ or that we could conclude a period of ‘crisis’; in other models, such as have ‘too much’ knowledge or, for that matter, CAMPBELL’s, its importance remains implicit in, say, whether our investments in scientific inquiry might the ‘fishscale model of omniscience’ or his preoccu- become ‘unproductive’ or their returns ‘maldistrib- pation with ‘competence of reference’ as a leading uted’. If one can never be too rich, neither can one aim of science. This image of ever be too smart: In both sci- science as an abstract sort of ence and capitalism, more is real estate presupposes that Author’s address always better. However, a tra- over time the access that any jectory of endless accumula- Steve Fuller, Professor of Sociology & Social individual scientist (or scien- tion and growth does not sit Policy, University of Durham, Durham DH1 tific discipline) has to reality well with classical images of 3JT, United Kingdom becomes more circumscribed inquiry having a clear aim Email: [email protected] and mediated, so that most of (e.g. ‘Truth’), whose achieve-

Evolution and Cognition ❘ 61 ❘ 1997, Vol. 3, No. 1 Steve Fuller

ment would presumably bring science to an end chaeologist Joseph TAINTER (1988) argued that these (FULLER 1997). CAMPBELL resolves this tension the societies eventually became too complex for their usual way by gesturing toward the ‘asymptotic ap- own good, as the cost of governance exceeded the proximation’ of the Truth, which neatly embeds the benefit that accrued to either the governors or the idea of endless inquiry in a framework that still ap- governed. Collapse came in the form of either a sim- pears, in some sense, goal-directed. plification of the administrative apparatus or a frag- Because of these SPENCERIAN prejudices, CAMPBELL mentation of the regime itself. If our highly could never envisage an evolutionary epistemology ‘functionally differentiated’ science system is like one that features the decline or corruption of science as a such ancient civilization that let its material develop- long-term consequence of its having been co-selected ment go unchecked, then we might expect epistemic with the capitalist ethic. However, the pressures on collapse to come from, on the one hand, the elimina- governments today to divest its massive funding tion of disciplines and specialties (say, through the commitments to science suggest that we may not be reorganization of the university) and, on the other, so far from a world that would force this awareness the privatization of knowledge in the form of intellec- upon us. CAMPBELL would perhaps recognize its pre- tual property. Perhaps future strains of evolutionary cedent—and recoil in horror. In a fascinating essay on epistemology will incorporate these very material the collapse of the great ancient civilizations, the ar- changes in the selection environment for science.

References Kuhn, T. (1970) The Structure of Scientific Revolutions. 2nd edition. (Orig. 1962) University of Chicago Press: Chicago. Arndt, H. (1978) The Rise and Fall of Economic Growth. Uni- Merton, R. (1973) The Sociology of Science. University of versity of Chicago Press: Chicago. Chicago Press: Chicago. Blomstrom, M./Hettne, B.(1984) Development Theory in Olsen, M. (1992) The energy consumption turnaround and Transition. Zed Books: London. socioeconomic well-being in industrial societies in the Campbell, D. (1997) From evolutionary epistemology via se- 1980s. In: L. Freeman (ed.) Advances in Human Ecology, lection theory to a sociology of scientific validity. This vol- vol. 1. JAI Press: Greenwich CT, pp. 197–234. ume. Price, D. de S. (1978) Toward a model for science indicators. Fukuyama, F. (1992) The End of History and the Last Man. In: Elkana, Y. et al. (eds), Toward a Metric of Science. Free Press: New York. Wiley-Interscience: New York, pp. 69–96. Fuller, S. (1997) Science. Open University Press: Milton Key- Rescher, N. (1984) The Limits of Science. University of Cali- nes UK. fornia Press: Berkeley. Gerschenkron, A. (1962) The Relative Advantage of Back- Schlick, M. (1974) The General Theory of Knowledge. (Orig. wardness. Harvard University Press: Cambridge MA. 1925). Springer-Verlag: Berlin. Harris M. (1979) Cultural Materialism. Random House, New Stehr, N. (1994) Knowledge Societies. Sage: London. York. Tainter, J. (1988) The Collapse of Complex Societies. Cam- Jacobs, R./Campbell, D. (1961) The perpetuation of an arbi- bridge University Press: Cambridge UK. trary tradition through several generations of a laboratory Wallerstein, I. (1991) Unthinking Social Science: The Limits microculture. Journal of Abnormal and Social Psychology of 19th Century Paradigms. Polity: Cambridge UK. 62: 649–658.

Evolution and Cognition ❘ 62 ❘ 1997, Vol. 3, No. 1 Francis Heylighen

Objective, Subjective, and Intersubjective Selectors of Knowledge

Introduction Abstract ogy is part of the larger evolutionary-cybernetic It is with great pleasure It is argued that the acceptance of knowledge in a philosophy which, to- that I use this opportu- community depends on several, approximately inde- gether with others, I am nity to comment on pendent selection “criteria”. The objective criteria are trying to develop in the Donald T. CAMPBELL’s last distinctiveness, invariance and controllability, the Principia Cybernetica paper (1997). I came into subjective ones are individual utility, coherence, sim- Project (JOSLYN/HEY- contact with Don’s work plicity and novelty, and the intersubjective ones are LIGHEN/TURCHIN 1993). at the beginning of my publicity, expressivity, formality, collective utility, Compared with a purely research career, in 1984, conformity and authority. Science demarcates itself evolutionary approach, a during a conference on from other forms of knowledge by explicitly control- cybernetic epistemology evolutionary epistemol- ling for the objective criteria. puts more emphasis on ogy at the University of the structure of cognitive Ghent. Since then, his writings have been a con- systems, on the processes by which they are constant source of inspiration. After we met, in 1990, structed, on the control they provide over the envi- we started to regularly exchange publications. Each ronment, and on the communication of knowledge. time I received a bunch of his papers, I began read- Such a cybernetic analysis, for example, allows the ing them with much pleasure, because I knew that I reinterpretation of CAMPBELL’s (1974) “nested hier- would find every paragraph teeming with deep archy of vicarious selectors” as the result of a series insights and surprising observations. We finally of metasystem transitions, producing subsequent con- decided to collaborate, producing an ambitious trol levels (HEYLIGHEN 1995). review paper about the evolution of social systems The ideas of cybernetics inspired much of CAMP- (HEYLIGHEN/CAMPBELL 1995). We were planning to BELL’s work, as illustrated by his recurring references write more papers together, but that has been made to the work of ASHBY, his long time support for the impossible by his untimely death in 1996. I see the perceptual control approach of POWERS (1973), and present paper as an opportunity to somehow con- his enthusiastic endorsement of the Principia Cyber- tinue my collaboration with Don, adding my netica Project. However, I guess it was a lack of ex- insights to his in a collective publication. pertise in the mathematical and computational As may have become obvious, there is virtually no models of cybernetics which kept him from using disagreement between my philosophical position the “cybernetics” label more explicitly. and the one of Donald CAMPBELL. The differences in approach have more to do with theoretical back- Different Co-selectors of knowledge ground and methods than with aims or convictions. While CAMPBELL (1974) called his philosophy of A large part of CAMPBELL’s (1997) paper, which pro- knowledge “evolutionary epistemology”, I would vides the focus of this memorial issue, is devoted to a characterize mine as “evolutionary-cybernetic episte- discussion of the different selectors which together mology” (HEYLIGHEN 1993). “Cybernetic” refers here determine the evolution of knowledge. The main to the broad domain of cybernetics and general sys- thrust is that the referent, i.e. the external object tems theory (ASHBY 1956; VON BERTALANFFY 1968), which the knowledge is supposed to represent, only and its transdisciplinary study of organization, com- plays a relatively small part in the selection of a par- munication, control and modeling. This epistemol- ticular idea or belief. In spite of its ideology, scientific

Evolution and Cognition ❘ 63 ❘ 1997, Vol. 3, No. 1 Francis Heylighen knowledge too is the product of multifarious selec- its particular niche within the larger shared environ- tive forces, most of which have little to do with ment. However, within the shark niche, some shark objective representation of the referent. Of the other designs will be fitter than others. In both organisms co-selectors, CAMPBELL pays special attention to the and beliefs, “being fitter than” is a partial order, not vehicle through which the knowledge is expressed an absolute one (HEYLIGHEN 1997). Such a philoso- and to the need to maintain the community which phy synthesizes the relativism of philosophers who carries the knowledge. In addition to these primarily emphasize the “incommensurability” of theories, social selectors, he discusses the selectors of interests with the more traditional belief in the objectivity of and historicity, which function on the individual scientific progress. level. Whereas CAMPBELL analyses these selectors struc- Objective Selection Criteria turally, that is, by the specific object or component responsible for the selection, I will here try to classify Since, as CAMPBELL (1997) reminds us, we have no them functionally, that is, by the role they play in the direct access to the “Ding an Sich”, we can only use evolution of knowledge. The class of selectors that indirect means to determine whether a belief corre- promote the same type of characteristics can be said sponds to an objective reality. Like the constructiv- to determine a selection criterion. Implicit in CAMP- ist cyberneticians VON FOERSTER (1981), and BELL’s examples, we can find three superclasses: ob- MATURANA/VARELA (1987) note, in the nervous sys- jective criteria (selection for fit to the outside object), tem there is no fundamental distinction between a subjective criteria (selection for assimilation by the perception and a hallucination: both are merely individual subject) and intersubjective criteria (se- patterns of neural activation. However, subjectively lection for sharing between subjects). These super- most people have no difficulty distinguishing classes can be divided into more fine-grained dreams or fantasies from perceptions. subclasses. The resulting classification will allow us To find out whether a perception is real, you to highlight the differences between scientifically should determine whether it is caused by an external derived knowledge, which supposedly privileges the referent, or by an internal mechanism (e.g.,, imagi- objective criteria, and other types of knowledge and nation, or malfunctioning of the perceptual appara- belief, where subjective and intersubjective factors tus). According to attribution theory (KELLEY 1967), play a larger role. people attribute causes of perceived effects to those However, as CAMPBELL emphasizes, it is impossible phenomena that covary with the effects. External to really separate the different selectors. All the dif- phenomena will covary with their external causes, ferent types of selectors will affect the evolution of but not with changes that only affect internal, sub- knowledge, scientific or other. Therefore, the overall jective variables. This leads to the following criteria probability for a belief to be selected will be a kind of for judging objectivity or “reality”: weighted sum of the degrees to which it fulfils each 1. Invariance: phenomena should not disappear of the criteria. For example, an idea that scores high when the way of perception is changed. The larger on the objective criteria and low on the subjective the domain over which it remains invariant, the ones, is less likely to survive selection than an idea more “real” it will be (cf. BONSACK 1977). KELLEY that scores high on both counts. In this view, no (1967) proposes the following more specific types of single criterion can guarantee selection, or provide invariance: justification for a belief. We can only use the simple a. invariance over modalities: if the same phenom- heuristic that the more criteria an idea satisfies, and enon is perceived through different senses (e.g., the higher the degree of satisfaction, the “fitter” it is, sight and touch), points of view, or means of obser- and the more likely to win the competition with ri- vation, it is more likely to objectively exist. val beliefs (HEYLIGHEN 1993). b. invariance over time: a perception that appears In such a view, there is in general no single “best” or disappears suddenly is unlikely to be caused by a idea. An idea may score high on certain criteria, stable referent. while another idea scores high on other criteria. c. invariance over persons: a perception on which Such ideas are in general incomparable. The one is different observers agree is more likely to be real than likely to win the competition in certain contexts, but one that is only perceived by a single individual. to lose in others. This is similar to the natural selec- 2. Distinctiveness: different referents produce differ- tion of organisms: sharks are not more or less fit than ent perceptions (KELLEY 1967; cf. CAMPBELL 1992). A seaweed or than shrimps. Each species is adapted to perception that remains the same when the attention

Evolution and Cognition ❘ 64 ❘ 1997, Vol. 3, No. 1 Objective, Subjective, and Intersubjective Selectors of Knowledge is directed elsewhere is likely to be produced by the ready knows. For example, the model of a beryllium perceptual system itself (e.g., a particle of dust in the atom may seem simple for a physicist well-versed in eye). Moreover, “real” perceptions tend to be charac- atomic models, but hopelessly complex for a layman. terized by richness in contrast and detail (imagined or More generally, the ease with which a cognitive dream perceptions typically are coarse-grained and system assimilates new ideas depends on the support fuzzy) and to exhibit “Gestalt qualities”, such as reg- they get from ideas assimilated earlier (this is an ex- ularity, closure and simplicity, thus proposing a dis- ample of CAMPBELL’s (1997) “historicity”). This re- tinct, coherent pattern, rather than an unstructured quirement for ideas to “fit in” the existing cognitive collection of impressions (STADLER/KRUSE 1990). system may be called coherence (THAGARD 1989). Co- CAMPBELL (1966 1997) too notes that detailed pattern herence encompasses connection and consistency. increases the plausibility of percepts. Since learning is based on strengthening associa- Extending this logic of covariation, I would like to tions, ideas that do not connect to existing knowl- add the criterion of controllability: a phenomenon edge simply cannot be assimilated. The preference that reacts differentially to the different actions per- for consistency follows from the fact that a fit indi- formed on it, is more likely to be real than one that vidual must be able to make clear-cut decisions. Mu- changes randomly or not at all. This criterion under- tually contradictory rules (“cognitive dissonance”) lies the method of preparation-detection, which will create a situation of confusion or hesitation, characterizes scientific experimentation. Controlla- which is likely to diminish the chances for survival. bility, however, is to some degree dependent on the Complementary to the conservatism promoted observing subject: although I am not able to influ- by the coherence criterion is the criterion of novelty. ence the trajectory of a far-away plane, its pilot is. New, unusual or unexpected ideas or perceptions This leads us to the subjective criteria. tend to attract the attention, and thus arouse the cognitive energy which will facilitate their assimila- Subjective selection criteria tion. This is another adaptation, which helps organisms to cope with unusual situations. It shows itself For beliefs to be accepted and retained by an indi- in the exploratory behavior of animals. The corre- vidual, it is not sufficient that they correspond to sponding human emotion is curiosity. distinct, invariant and controllable phenomena. A relativistic quantum field model of the beryllium Intersubjective selection criteria atom may fulfil all objective criteria to be valid knowledge, yet very few people would ever assimi- Most of the beliefs a subject has were not individu- late, remember or pass on such knowledge. There- ally constructed, but taken over from others. This fore, from a selectionist point of view, the model is process of diffusion plays an essential part in the rather unsuccessful. selection of ideas. Only ideas that are transmitted Most obvious among the subjective selection cri- frequently are likely to be assimilated frequently. teria is individual utility. People will only do the effort Each time an idea is communicated, it replicates, i.e. to learn and retain an idea that can help them to is copied into another cognitive system. Thus, ideas reach their goals. From a long-term evolutionary can be modeled as replicators similar to genes: perspective, such goals and values derive from inclu- memes (cf. HEYLIGHEN 1992). The conversion of an sive fitness. Organisms that assimilate knowledge individual to a new belief is in a way similar to an which increases their fitness are more likely to sur- infection, i.e. the passing on of a “cognitive virus”. vive and pass on that knowledge to their offspring. The first criterion which will determine how often Assimilating useless knowledge, on the other hand, an idea is transmitted is the amount of propaganda only burdens the subject. or publicity, that is, the effort the subject carrying the Indeed, the capacity of a cognitive system is lim- idea invests in making it known to others. That mo- ited. Therefore, knowledge should be easy to learn. tivation largely depends on the other criteria: you The most straightforward determinant of learning will be more inclined to spread an idea if it is simple, ease is simplicity: the more complex an idea (i.e. the useful, novel, etc. However, some beliefs include more components and more connections between their own motivation. This is most visible in reli- components it has, see HEYLIGHEN 1997), the higher gions, cults, fashions and ideologies, which often in- the burden on the cognitive system. Simplicity is clude explicit rules that believers should go and listed as a subjective criterion, because it is relative to spread the word. This may be explained by “selfish the concepts and associations which the subject al- meme” selection (HEYLIGHEN 1992): selection at the

Evolution and Cognition ❘ 65 ❘ 1997, Vol. 3, No. 1 Francis Heylighen level of the meme, which benefits the spread of the interacting group. This selective pressure may be idea, but which is useless or even dangerous for the called conformity. individual carrying the idea. Such ideas can be com- Complementary to this homogenizing influence, pared to cognitive parasites, which “hitch a ride” on we find the diversifying effect of the division of la- a cognitive system without caring for the well-being bor. Because of their limited cognitive capacity, in- of that system. dividuals within a complex society tend to specialize All memes, “selfish” or not, need a communica- in a particular domain. As illustrated by GAINES’s tion medium in order to be transmitted. Ideas that (1994) computer simulation, this process of cogni- are easy to express in a particular language or me- tive differentiation is driven by a positive feedback dium will be propagated more easily. This is the cri- mechanism: individuals who were successful in solv- terion of expressivity. It depends on the medium: ing a particular type of problem will get more of some ideas are easier to formulate in one language these problems delegated to them, and thus develop than in another. Thus, like CAMPBELL (1997) notes, a growing expertise or authority in that domain. The the medium will co-select the idea. For example, it is backing of a recognized expert will contribute to the difficult to imagine the evolution of physical theo- acceptance of a particular idea. This is the criterion ries without the mathematical language in which of authority. they are formulated. The integration on the level of norms and codes The expression of an idea in an intersubjective code fostered by conformity and the differentiation on or language does not yet guarantee its accurate trans- the level of expertise fostered by authority together mission. All expressions are to some degree indexical: produce a complexification (cf. HEYLIGHEN 1997) of their meaning depends on the context. Different peo- the social system. The process is similar to the meta- ple are likely to interpret them differently, thus assim- system transition which produced the differentiated ilating an idea different from the one that was organs and tissues in a multicellular organism (HEY- expressed. However, some expressions are formulated LIGHEN/ CAMPBELL 1995; HEYLIGHEN 1995). in less context-dependent way. The resulting lack of equivocation may be called formality. The more for- Scientiﬁc Validity and the Demarcation mally an idea is expressed, the better it will survive repeated transmissions. For example, ideas are more Problem likely to be communicated accurately through logic The selection criteria we discussed are summarized and mathematics than through poetry or painting. in table 1. When we look at the evolution of scien- The group equivalent of usefulness may be called collective utility. Some forms of knowledge benefit Objective Criteria Invariance the collective, while being useless for an isolated in- Distinctiveness dividual. Languages, traffic regulations, technical Controllability standards and moral codes are examples of cognitive Subjective Criteria Individual Utility entities that have value only for intersubjective pur- Simplicity poses. Such collective ideas will be selected at the Coherence group level: groups having such beliefs will be more fit than groups lacking them. This how the supernat- Novelty ural cosmologies characterizing archaic civilizations Intersubjective Publicity discussed by CAMPBELL (1997) have been selected. Criteria Expressivity However, as CAMPBELL emphasizes, such group se- Formality lection often runs counter to the more powerful and Collective Utility direct force of individual selection. Therefore, he Conformity proposes a mechanism that suppresses individually selfish deviations from these collective beliefs: con- Authority formist transmission. As illustrated by the mathe- Table 1: summary of the proposed selection criteria matical model of BOYD/RICHERSON (1985), all other things being equal, it seems evolutionarily optimal tific knowledge, it is clear that all these criteria play for subjects to adopt the majority or plurality belief a role in the selection of ideas. The objective criteria rather than a minority idea. Thus, already popular obviously underlie the experimental method: new ideas tend to become even more popular, leading to concepts are operationalized by specifying the an eventual homogeneity of belief within a closely observations that will distinguish their referents, by

Evolution and Cognition ❘ 66 ❘ 1997, Vol. 3, No. 1 Objective, Subjective, and Intersubjective Selectors of Knowledge subjecting them to controlled ity.) The objective criteria Author’s address experiments, and by trying to have been built into the scien- find results which are maxi- tific method. They have be- Francis Heylighen, PO, Free University of mally independent of place, come part of knowledge itself, Brussels, Pleinlaan 2, B-1050 Brussels, Bel- time, observer or means of gium. rather than an outside force to observation (as CAMPBELL E-mail: [email protected] which knowledge is sub- (1997) notes, the latter is http://pespmc1.vub.ac.be/HEYL.html jected. The scientific method, often difficult in the social in CAMPBELL’s (1974) termi- sciences). Moreover, subjec- nology, is a vicarious selector, tive interpretation is minimized by formalization an interiorization of external selectors. of the theories and concepts. However, from other- This vicarious selector functions at a higher hier- wise equivalent ideas, scientists will still tend to archical level than the knowledge it produces. Be- prefer those that may bring fame and fortune, that cause other forms of knowledge are not selected at are simple, coherent with what they already know, this higher level, they will evolve in a less efficient and novel. In addition, the social system of science way, and are therefore likely to be of lower quality. will prefer ideas that have vocal advocates, are strik- The difference between scientific and other knowl- ingly expressed, benefit the community, and are edge is not an absolute one, between objective and supported by the majority, or by authoritative subjective, or between justified and unjustified, but experts. one of degree, between the products of a systematic In what way, then, can science be demarcated process of improvement, and those of a slow, hap- from other knowledge producing systems, such as hazard process of trial-and-error, where neither trial religion, fashion or tradition? The difference is that nor error are consciously controlled. science explicitly promotes the objective criteria. (To a smaller degree, as CAMPBELL (1997) notes, science Acknowledgment also tries to neutralize the criteria that are likely to detract from objectivity, such as authority which is During this research the author was supported as a not backed by expertise, conformity for confor- Senior Research Associate by the FWO (Fund for Sci- mity’s sake, and—at least in the pure sciences—util- entific Research, Flanders).

References Heylighen F. (1993) Selection Criteria for the Evolution of Knowledge. In: Proc. 13th Int. Congress on Cybernetics Ashby, W. R. (1956) Introduction to Cybernetics. Methuen, (Association Internat. de Cybernétique, Namur), pp. 524– London. 528. Bonsack, F. (1977) Invariance as a Criterion of Reality. Dialec- Heylighen, F. (1995) (Meta)systems as Constraints on Varia- tica 31 (3–4):313–331. tion, World Futures 45:59–85. Boyd, R./Richerson, P. J. (1985) Culture and the Evolutionary Heylighen, F. (1997) The Growth of Structural and Function- Process. Chicago University Press. al Complexity during Evolution. In: F. Heylighen (ed) The Campbell, D. T. (1966) Pattern Matching as an Essential in Evolution of Complexity. Kluwer, Dordrecht. [in press] Distal Knowing. In: K.R. Hammond (ed) The Psychology of Joslyn, C./Heylighen F./Turchin V. (1993) Synopsys of the Egon Brunswick. Holt, Rhinehart and Winston, New Principia Cybernetica Project. In: Proc. 13th Int. Congress York), pp. 81–106. on Cybernetics. Association Internationale de Cyberné- Campbell, D. T. (1974) Evolutionary Epistemology. In: tique, Namur, pp. 509–513. http://pespmc1.vub.ac.be/ Schilpp P.A. (ed) The Philosophy of Karl Popper. Open Kelley, H. H. (1967) Attribution Theory in Social Psychology, Court Publish., La Salle, Ill., pp. 413–463. Nebraska Symposium on Motivation 15:192–238. Campbell, D. T. (1992) Distinguishing between Pattern in Maturana, H. R./Varela, F. (1987) The Tree of Knowledge. Perception due to the Knowing Mechanism and Pattern Shambhala, Boston. Plausibly Attributable to the Referent. (unpublished manu- Powers, W. T. (1973) Behavior: the Control of Perception. Al- script). dine, Chicago. Campbell, D. T. (1997) From Evolutionary Epistemology via Stadler, M./Kruse, P. ( 1990) Theory of Gestalt and Self-orga- Selection Theory to a Sociology of Scientific Validity. Evo- nization. In: Heylighen F./Rosseel E./Demeyere F. (ed) Self- lution and Cognition (this issue). Steering and Cognition in Complex Systems. Gordon and Gaines, B. R. (1994) The Collective Stance in Modeling Exper- Breach, New York, pp. 142–169. tise in Individuals and Organizations. International Jour- Thagard, P. (1989) Explanatory Coherence. Behavioral and nal of Expert Systems 71:22–51. Brain Sciences 12:435–467. Heylighen, F./Campbell D.T. (1995) Selection of Organiza- von Bertalanffy, L.(1968) General System Theory. Braziller, tion at the Social Level. World Futures 45: 181–212. New York. Heylighen, F. (1992) ‘Selfish’ Memes and the Evolution of von Foerster, H. (1981) Observing Systems. Intersystems, Cooperation, Journal of Ideas, 2 (4), pp 77–84. Seaside CA.

Evolution and Cognition ❘ 67 ❘ 1997, Vol. 3, No. 1 Aharon Kantorovich

The Rationality of Innovation and the Scientific Community as a Carrier of Knowledge

1. The Balance Abstract the scientific community Between Selection interacts with ‘Nature Her- and Generation: It is suggested that epistemological considerations self’ via reliable means, Socio-Evolutionary may not be restricted to the process of selection, as such as experiment and Mechanisms for CAMPBELL maintains. Psychological and social mech- logic. So the criteria for Generating New anisms of generating new variations in science are publication, for example, Variations in Science shown to yield some conclusions about rational ways reflect predominantly the of creating scientific ideas and theories. Further con- scientists’ broad (‘univer- CAMPBELL adopts the view clusions are drawn from the function of the scientific sal’) interest of exposing that the social dimension community as a carrier of knowledge. Departure from truth rather than local or is essential to science. Sci- tradition is made possible by unintentionality and by sectarian interests or per- entific beliefs are beliefs exploiting existing ideas for solving new problems. sonal tastes. produced by a social sys- The norms of the scientific tribe should encourage this Selection is according tem. If they have greater patterns of behavior. to CAMPBELL the main epis- validity than those pro- temological tool in sci- duced by other systems, this is due to the norms of ence, since “validity is attributed to the selection the scientific community. Thus, science has different process” rather than to generation. Indeed, a message specific values, myths, rituals and commandments. preached by Karl Popper and adopted by CAMPBELL is The question is therefore what are these norms and that in science, as well as in DARWINIAN evolution, what is their epistemological significance. generation or variation is ‘blind’, namely does not In CAMPBELL’s text we can find the following char- have any methodological implications. However, acterizations of science as a social system: science is generation is too important to be left to mere chance. one of the “superstition preservation systems” and Generation should yield a limited number of plausi- the scientific community is “among belief-preserving ble variants from which to select, otherwise selection mutual admiration society”. However, the scientific would be a futile game and progress will be impossi- community differs from other tradition-ridden ble. Scientific rationality thus depends on both varia- groups by recognizing that “tradition is a source of tion and selection. error”. If there is a way of eliminating error, it would CAMPBELL diminishes the role of variation in line guide scientists in departing from tradition. This is with the spirit of logical empiricism which expels dis- the point where Natural Selection enters into the pic- covery from the realm of rationality. As the distinc- ture: it supplies the model for creativity. tion between the context of justification (which is The three main ingredients CAMPBELL borrows considered to be governed by logic and reason) and from the DARWINIAN evolutionary dogma are: varia- the context of discovery (thought to be beyond the tion, selection and retention. But he typically concen- reach of reason) has been blurred we have to deal with trates on selection. He confronts “selection by the contribution of both contexts to the rational con- nature” with “selection by the scientific community”. duct of science. Scientists are not engaged with criti- This is portrayed almost as a dichotomy: “It becomes cism for its own sake; criticism goes hand in hand hard to argue for a dominant role for ‘Nature Herself’ with generating or discovering new ideas or theories. in the selecting.” But how does the scientific commu- Often the overthrow of an existing theory occurs only nity select? Or what are the criteria for “the genera- after a better theory is found. Furthermore, there is a tion, publication, teaching and believing of scientific widely accepted view that a replacement of a theory truth claims”? It seems that in natural science (but not by a new one is the only viable way of ‘refuting’ the in the humanities and perhaps also in social science) theory (LAKATOS 1970). Thus, discovery cannot be

Evolution and Cognition ❘ 68 ❘ 1997, Vol. 3, No. 1 The Rationality of Innovation and the Scientific Community as a Carrier of Knowledge treated separately from selection and if “validity is this way their creativity will be enhanced. And more attributed to the selection process”, it should be at- generally, they should be alert so that if in the course tributed to discovery as well. of their efforts to solve one problem they come up Moreover, as we shall see, there are ways of gener- against an idea which gives a clue for solving another ating new ideas which enhance the chances for these problem, they should seize this opportunity for solv- ideas to become discoveries, regardless of questions ing the other problem instead of insisting on solving about their validity. For example, there are certain only the original one. Or, if scientists are not success- ways of combining ideas, or ways of making associa- ful in solving a given problem they may turn to an- tions between different ideas, which lead more often other. And after acquired the skill of using a given to discoveries. Hence, generation is not entirely be- (experimental or theoretical) tool, they might look for yond the scope of reason. It was suggested (KANTOR- other fields where it can be exploited—this is the basis OVICH/NE’EMAN 1989) that within the framework of for the phenomenon of intellectual migration. evolutionary epistemology ‘blind’ discovery be inter- The scientific community has at its disposal an- preted not just as random variation but as serendipi- other source of creativity: the interpsychic process of tous discovery, specified in the following way. When epistemic cooperation. Here the combination of ideas scientists generate, or look for, ideas they are not en- takes place in the social arena. To the extent that sci- gaged with sheer gambling. They are supposed to ex- ence is essentially a social phenomenon (see for ex- ploit intrapsychic and interpsychic mechanisms ample ZIMAN 1968), this process perhaps (SIMONTON 1988) which yield involuntary mental and distinguishes science from all other human activities. social processes respectively. These processes can be A set of social norms and codes of behavior would characterized as ‘blind’ since eventually scientists guide scientists in exploiting this extra dimension of solve via these mechanisms different problems than creativity. Epistemological cooperation fosters unin- those they intended to solve originally. The intrapsy- tentionality and serendipity. Thus an idea or theory chic creative mechanisms are not exclusive to scien- which were generated by one scientist (or group of tific thinking. They appear also in ordinary creative scientists) in order to solve a given problem may be thinking. SIMONTON describes such a creative process employed by other members of the community in as generated by chance association or combination of order to solve an entirely different problem. Again, we mental elements (such as ideas, formulas or images). interpret this process as another kind of ‘blind’ varia- The process may end up with the emergence of a sta- tion since the original discoverer of the idea was not ble configuration. The latter is the product of creative aware of the problem which is eventually solved by thought, i.e. a discovery or an invention. This process his colleagues or predecessors; the initiator of the pro- can be represented as a semi-blind variation and se- cess was blind to the problem solved. The coupling lection, where the principle of selection is related to between both kinds of variation—the intra- and the stability. inter-psychic—is what makes science such a powerful Thus, rationality in the domain of discovery would novelty-generating evolutionary system. require that scientists follow some norms of conduct The following codes of behavior would guide sci- in order to take advantage of the creative mecha- entists in exploiting the interpsychic mechanisms of nisms. These norms can be derived from what we discovery. Scientists have to use the communication know about the nature of these mechanisms. Ratio- network of the scientific community; to attend con- nality in this domain would therefore reflect our es- ferences, to read papers published in the field and to tablished theoretical knowledge about these submit papers for publication in scientific journals, phenomena which take place in our minds. For exam- and so on. And there are further recommendations to ple, the above theory of forming chance configura- the scientific community: epistemic cooperation tions would imply the following rules for fostering or should be encouraged by establishing appropriate cultivating creativity. Scientists should be engaged si- norms of conduct through the scientific education multaneously in several problems; they should not be system, the award system and other institutions. one-track-minded. Thus, in their efforts to solve a Thus, according to this view, it is not true that va- given problem, they may gen- lidity or rationality is attrib- erate or encounter an idea or Author’s address uted only to the selection mental element which was process. The mechanisms of missing in the process of form- Aharon Kantrovich, 10 Drezner St., Tel Aviv generating new variations in ing a stable configuration for 69497, Israel. science yield some conclusions solving a different problem. In about the rationality of creat-

Evolution and Cognition ❘ 69 ❘ 1997, Vol. 3, No. 1 Aharon Kantorovich ing scientific ideas and theories so that the context of stract knowledge detached from its material carrier. discovery or generation does contribute to the ratio- The following parallelism exists between the two nality of science. Rationality here is not full-proof; it kinds of knowledge. There are restrictions imposed on does not necessarily lead to true ideas or theories. the way one material system (an organ) represents However, it complies with the nature of our knowl- another. The restrictions are determined by the na- edge-acquisition system. And in an atmosphere of ture or structure of the representing system. In a sim- post-positivism and naturalism we cannot ask for ilar fashion, the restrictions imposed on disembodied more than this. knowledge are determined by the set of categories or concepts available to the representing system. 2. Carriers of Knowledge A reptile’s belly, the horse’s hoof and the human foot are organs representing different facets of the CAMPBELL distinguishes between embodied (endoso- earth’s surface. The rhinencephalon of lower mam- matic) and disembodied (exosomatic) knowledge. mals, the neocortex of humans and the scientific Referring to embodied knowledge he says: “the community do not embody representations of phe- vehicular substance that carries knowledge is nomena, rather they are different ways or methods of unavoidably alien to the referents of knowledge—it coping with reality: they guide the different species in is a different substance with different structural char- controlling their internal and external environment, acteristics”. Thus, the horse’s hoof embodies knowl- constructing representations (such as theories) and edge or information about the earth’s surface. We explaining the phenomena. The brain and the scien- have here one material system (the carrier of knowl- tific community are both producers and carriers of edge) which has been adapted to another (the refer- knowledge. CAMPBELL discusses the function of the ent of knowledge) through the evolutionary process. scientific community as a vehicle of knowledge. As a Light will be shed upon the relation between the vehicle of knowledge it should have some rigidity “to carriers and the referents of knowledge if we treat the hold together the picture it carries”. Rigidity means first as hypotheses (theories) and the latter as the facts being faithful to tradition. The opposite alternative is to be explained or accounted for. An organ, such as that the vehicle is completely flexible. A transparent the horse’s hoof, embodies a hypothesis about a por- carrier of knowledge “contributes least of its own to tion of the environment, i.e., about the earth’s sur- the knowledge it transmits”. This means that the so- face. The fact that the vehicular substance here is alien cial structure of a scientific community, or its ‘tribal’ to the substance which roads are made of corresponds nature, which contributes to the rigidity of this vehi- to the methodological requirement that an explana- cle of knowledge is what holds together the scientific tory hypothesis should be expressed in theoretical vo- world picture. And CAMPBELL raises the important cabulary which is alien to the observational question: what makes a tribe of scientists capable of vocabulary. For example, the story of the kinetic the- effectively generating “valid” scientific knowledge, ory of gases is told in terms of particles in motion, what demarcates it from other “superstition preserva- whereas the observational story is told in terms of tion systems” or what makes it superior to every other volumes, temperatures and pressures. “belief-preserving mutual admiration society”? To continue the analogy, the relation between an CAMPBELL maintains that the selectionist model pro- organ and a species corresponds to the relation be- vides the answer. But this is only a partial answer. It tween a theory about a particular kind of phenome- seems that the model of unintentional or serendipi- non and a general theory explaining a wide range of tous innovation completes the answer. Departure phenomena. from tradition is made possible by unintentionality Disembodied or abstract knowledge consists of and by exploiting existing ideas for solving new prob- ideas, beliefs and theories. This is the kind of knowl- lems. The norms of the scientific tribe should encour- edge traditional epistemology is engaged with; ab- age this patterns of behavior.

References (ed) Criticism and the growth of knowledge. Cambridge University Press, Cambridge, pp. 91–196. Kantorovich, A./Ne’eman, Y. (1989) Serendipity as a source Simonton, D. K. (1988) Scientific genius: A psychology of sci- of evolutionary progress in science. Studies in History and ence. Cambridge University Press, Cambridge. Philosophy of Science 20: 505–529. Ziman, J. (1968) Public knowledge. Cambridge University Lakatos, I. (1970) Falsification and the methodology of sci- Press, Cambridge. entific research programmes. In: Lakatos, I./Musgrave, A.

Evolution and Cognition ❘ 70 ❘ 1997, Vol. 3, No. 1 Elias L. Khalil

Can Artificial Selection Remedy the Failing of Natural Section with Regard to Scientific Validation?

his commentary is Abstract digms, claim superiority T concerned only with with respect to the one aspect of Donald Given the difference between natural and artificial se- approximation of truth. CAMPBELL’s (1997) multi- lection, CAMPBELL finds artificial selection the more Paradigms would be faceted essay. Namely, appropriate of the two for the depicting scientific vali- incommensurable; they CAMPBELL advocates an dation. Natural selection, which emphasizes the con- would be ultimately justi- evolutionary epistemol- text, entails a relativist view of science. Artificial fied on self-referential ogy theory which some- selection, insofar as it presents selectors as interested grounds, i.e., internal cri- what differs from natural in empirically supported theories, implies that the sci- teria similar to the ones selection. Section 1 finds entific enterprise is superior to other belief-preserving which sustain belief-pre- that CAMPBELL’s theory tribes. However, artificial selection, like natural selec- serving tribes. remedies the relativist tion, still adheres to the tenet of blind mutation. This CAMPBELL maintains connotation associated tenet defies the fact that scientists—contrary to the that, at least according to with the natural selection core of neo-DARWINISM—propose new theories in light professed norms, the sci- account of scientific vali- of their ex post effect. entific tribe does not op- dation. Section 2 argues erate according to what is that, like natural selection, CAMPBELL’s theory is in- useful but rather according to what is true. The dis- adequate to account for the origin of variety of sci- juncture between usefulness and truth has entific theories. prompted CAMPBELL to argue for a “general selection theory,” where natural selection is only one variety 1. Is the Criterion of Selection of while evolutionary epistemology is another. For CAMPBELL, natural selection (or what he calls “bio- Scientiﬁc Theories Self-Referential? logical selection”) is governed by the usefulness cri- CAMPBELL recognizes the limits of natural selection terion, while epistemological selection is governed when applied to the question of scientific valida- at least partially by the truth criterion. The latter se- tion. The theory of natural selection postulates that lection is evident when members of the scientific a population, abstracting from secondary hurdles, tribe appeal, at least formally, to ‘the way the world ends up with a mix of traits which is optimum given is’ as the ground upon which to adjudicate among the selective environment. Such optimization, competing theories. called ‘adaptationism’, guarantees that the mix of In support of CAMPBELL’s thesis, the disjunction traits which become dominant is the most useful between usefulness and truth informs the difference mix given the initial variety from which a popula- between technology and scientific theorizing or, in tion starts. Given that usefulness is defined by the general, between practical knowledge and theoreti- environment, one cannot apply such a scenario to cal knowledge. The ultimate concern of practical depict the process of scientific validation. It would knowledge, which informs the foundation of voca- entail that the dominant theoretical paradigm is tional training from medicine and engineering to dominant because of the peculiar, faddish tastes of business administration, is how to fix a situation and the selectors, i.e., scientists. If such a relativist view make it more conducive to human welfare. The ulti- of science is correct (KUHN 1970), the dominate par- mate concern of theoretical knowledge, which in- adigm could not, in opposition to alternative para- forms the foundation of disciplines from sociology

Evolution and Cognition ❘ 71 ❘ 1997, Vol. 3, No. 1 Elias L. Khalil and biology to physics, is why are things the way vation that selectors are committed to the norm of they are. Obviously, producers of practical knowl- truth-seeking, evolutionary epistemology still ad- edge may use theoretical knowledge, and producers heres to the three conditions which characterize nat- of theoretical knowledge may use tools and technol- ural selection, viz., “a mechanism for introducing ogies afforded by practical knowledge. Nonetheless, variation, a consistent selection process, and a mech- each one pursues a different end. While the end of anism for preserving and reproducing the selected theoretical knowledge is the satisfaction of curiosity, variations.” First, the blind mechanism for variation the end of practical knowledge is human welfare. is trial and error, where scientists propose diverse The criterion employed to judge a technology is theories to illuminate particular facts. Second, the whether it affords the actor an advantage over com- selection process consists of a scientific tribe whose petitors and, hence, advances human welfare. The members, despite their pre-commitments and prej- assumption that actors are rational does not mean udices, practice at least partially the norm inherited that they have the same technology. It only entails from the ideology of the scientific revolution, viz., that they use the resources and technology they pos- to choose the theory which best depicts ‘the way the sess in the most efficient fashion. It is up to natural world is’. Although CAMPBELL sympathizes with the selection, which can be dubbed “market selection” claims of the advocates of the relativist view of sci- in modern human production activity, to select the entific validation, he maintains that the commit- most useful technology. Like biological selection, ment to judge theory according to ‘facts that speak market selection involves consumers acting as selec- for themselves’ ultimately enhances the “compe- tors. The dominant technology is the one which fits tence of reference” of theories. The competence of the tastes of consumers. As in the case of biological reference makes the scientific community different traits, one cannot determine in the abstract which is from other, belief-preserving, mutual admiration the fittest technology. It is possible that the most tribes. Third, the scientific community reproduces productive technology, measured by units of output the successful theory through the instruction of new per labor-time, is not the fittest. The cost of develop- practitioners. ment and maintenance of the most productive tech- I concur with CAMPBELL that evolutionary episte- nology may not justify it vis-à-vis more simpler mology modified as such still preserves the main el- technology. A technology which can be justified un- ements of natural selection and, hence, can be seen der certain circumstances can become less fit under as a variety of a “general selection theory.” To wit, new circumstances. such evolutionary epistemology can be character- Therefore, market selection, similar to biological ized as “artificial selection”—similar to the selective selection, justifies technology or practical knowl- breeding of domesticated animals and plants accord- edge according to the usefulness criterion. Both mar- ing to the preferences of the breeder. Charles DARWIN ket and biological selections involve the natural commenced the Origin of Species by an extensive dis- selection mechanism: The dominant trait or tech- cussion of artificial selection as a way to elucidate the nology is not valid in the abstract sense, but rather theory of natural selection. However, there is one valid only in relation to the peculiar selective envi- important and obvious difference between the two. ronment. In contrast, according to CAMPBELL, the sci- The difference revolves around who is the benefi- entific enterprise is about validation in the abstract ciary of the selection. In artificial selection, the ben- sense. The market of scientific theories is not sup- eficiary is the selector, whether it is the breeders who posed to generate theories according to fashion or demand a more refined trait or the scientists who practicality but rather, at least formally, according to demand a better theory. In natural selection, the truth. If one applies natural selection to the market beneficiary is the population which is being selected of scientific theories, i.e., treating theories as no dif- so that the succeeding generation is more adapted to ferent from tables and tools, one would be subscrib- its environment. ing to a relativist theory of science (KHALIL 1995). To counteract the connotation of relativism, 2. What is the Origin of Variety of CAMPBELL argues that selection theory should not be identified with natural selection. Evolutionary epis- Scientiﬁc Theories? temology, according to CAMPBELL, posits that the se- CAMPBELL’s recognition of the failing of natural lectors do not weed out theories which are least selection with regard to scientific validation is useful, but rather weed out theories which are least refreshing. CAMPBELL’s theory of evolutionary epis- corroborated by the world of facts. Despite the inno- temology qua artificial selection goes a long way to

Evolution and Cognition ❘ 72 ❘ 1997, Vol. 3, No. 1 Can Artificial Selection Remedy the Failing of Natural Section with Regard to Scientific Validation? remedy this failing. The question pursued in the rest ers do not supply scientific theories in a blind fash- of this commentary is whether CAMPBELL’s theory of ion. Suppliers rather ‘direct’ the generation of artificial selection is adequate with respect to theories, i.e., fashion them in order to meet the ex- another front, viz., the origin of variety of scientific ternal criteria used by selectors. The advocates of sci- theories. entific theories develop their hypotheses in light of CAMPBELL’s model of artificial selection is based on the anticipated criticism of peers, the selectors. Even a core assumption which is shared by all selection if there are no selectors, suppliers of scientific theo- theories within the DARWINIAN tradition. DARWIN ries subject their products to the same criterion used theory distinguished itself from other theories of by the selectors. Suppliers of scientific theories are evolution by its repudiation of teleology (GHISELIN also interested in the pronouncement of statements 1969; see KHALIL, 1997): According to DARWIN, a spe- which they can accept on the same ground which cies did not evolve by the volition of organisms in they use to assess all other theories. the face of environmental challenges. Rather, it If this point is granted, it would be disturbing for evolved irrespective of the intention of organisms, any kind of selection theory in the DARWINIAN tradi- i.e., as a result of the weeding out of the less fit or- tion. It would mean that variety, at least to some ganisms. For such weeding out to take place, the extent, is endogenous; the variety arises from the population should be characterized by variation. anticipation of ex post benefit. It would also entail DARWIN’s theory had to wait for the development of that the selective environment does not play the cru- genetics in the 20th century to give meaning to the cial role portrayed by neo-DARWINISM. origin of variety. DARWIN’s notion of pre-existing va- CAMPBELL would probably have responded by ad- riety became sharpened in the neo-DARWINIAN syn- mitting that originators of scientific theories do not thesis with the notion that if variety changes, it advance them in a blind fashion. He would, how- arises from blind mutation of genes. The notion of ever, probably continue and argue that such evident blind mutation justifies the modeling of change of directed mutation on the part of originators of theo- variety as exogenous. That is, the change in variety ries is the outcome of a “vicarious process.” Similar is not considered to be a function of environmental to the vicarious processes of visual perception, cre- stresses. The idea of blind mutation amounts to the ative thought, belief development through dia- thesis that the genetic variety generated by mutation logue, and so on, the faculty of directing mutation is independent of its ex post effect given the selectors. is short-cut selection, i.e., originally achieved by The temple of neo-DARWINISM rises or falls with its blind-variation-and-selection-retention. central tenet of blind mutation. There are diverse ap- Even if we grant that the vicarious processes which proaches critical of neo-DARWINISM. However, what CAMPBELL has identified had originated from natural determines whether an approach is within or outside selection, this does not mean that directed mutation the neo-DARWINIAN program is whether it subscribes has also risen from natural selection. And even if di- to the hypothesis of blind mutation. Soft neo-DAR- rected mutation has been occasioned by natural se- WINIANS may be critical of the reductionist tendency lection in the past, it does not attenuate the anomaly of orthodox neo-DARWINIAN (WILSON/SOBER 1989; EL- facing CAMPBELL’s artificial selection. The idea that DREDGE 1996) or the treatment of mutation as ran- theories are directed mutations has a radical ramifi- dom, as if there is no ‘structures’ which persist cation. It amounts to the negation of the artificial through evolutionary change (GOULD 1977; GOULD/ selection account as if, to paraphrase Ludwig WIT- LEWONTIN 1979). However, such soft neo-DARWINIANS TGENSTEIN, scientists have kicked off the neo-DAR- refrain from repudiating the central tenet. WINIAN ladder of evolution (viz., artificial selection) The question is whether blind mutation is the ap- once they have started to anticipate the ex post effects propriate description of the origin of variety of sci- of their theories. To begin, why should one be con- entific theories. If selectors, following CAMPBELL, cerned about the origin of directed mutation? The subject theories, at least par- origin could be the result of an tially, to the test of whether act of God, angels, natural se- they are empirically sup- Author’s address lection, or whatever. What ported, the suppliers of scien- matters is the consequence of tific theories must also have Elias L. Khalil, Department of Economics, directed mutation, viz., the subject such theories to the Ohio State University, Mansfield, OH negation of selection theory as same test prior to advancing 44906, USA. the primary explanation of Email: [email protected] them in public. That is, suppli- evolutionary change.

Evolution and Cognition ❘ 73 ❘ 1997, Vol. 3, No. 1 Elias L. Khalil

In short, CAMPBELL’s theory of artificial selection advance theories with ex ante expectations—selec- succeeds in ameliorating the problem of relativism tion theory, even of the artificial variety, fails to cap- which connotes the natural selection account of sci- ture the core issue of scientific validation. entific validation. However, artificial selection, similar to natural selection, adheres to the central tenet Acknowledgments of neo-DARWINISM, viz., mutation of scientific theories are not correlated with ex post effects. Given that I appreciate the help of Carole BROWN. The usual originators of scientific theories are self-critical—i.e., caveat applies.

References co and the Panglossian Paradigm: A Critique of the Adap- tationist Programme. Proceedings of the Royal Society of Campbell, D. T. (1997) From Evolutionary Epistemology Via London, 205: 581–598. Selection Theory to a Sociology of Scientific Validity. Evo- Khalil, E. (1995) Has Economics Progressed? Rectilinear, His- lution & Cognition 3(1). toricist, Universalist, and Evolutionary Historiographies. Eldredge, N. (1996) Ultra-Darwinian Explanation and the Bi- History of Political Economy 27(4): 43–87. ology of Social Systems. In: Khalil, E. L./Boulding, K. E. Khalil, E. (1997) Economics, Biology, and Naturalism: Three (eds) Evolution, Order and Complexity. London: Rout- Problems Concerning the Question of Individuality. Biol- ledge, pp. 89–103. ogy & Philosophy, 12(2). Ghiselin, M. T. (1969) The Triumph of the Darwinian Meth- Kuhn, T. S. (1970) The Structure of Scientific Revolutions. od. Berkeley: University of California Press. Chicago: University of Chicago Press. Gould, S. J. (1977) Ontogeny and Phylogeny. Cambridge, Wilson, D. S./Sober, E. (1989) Reviving the Superorganism. MA: Harvard University Press. Journal of Theoretical Biology, 136: 337–356. Gould, S. J./Lewontin, R. C. (1979) The Spandrels of San Mar-

Evolution and Cognition ❘ 74 ❘ 1997, Vol. 3, No. 1 Kyung-Man Kim

D. T. Campbell’s Social Epistemology of Science

Vicarious Selection, Epistemological Fallibilism, and Sociology of Scientific Validity

Introduction Abstract epistemology. After showing why vicarious se- Unlike his contemporar- The concept of a ‘nested hierarchy of vicarious selec- lection processes are so ies Robert MERTON and tors’ occupies the central place in Donald CAMPBELL’s important in all knowl- Edward SHILS who polem- evolutionary epistemology. After showing why vicari- edge processes, I will pro- ically isolated themselves ous selection process is so important in all knowledge ceed to demonstrate how from the newly emerging processes, I will proceed to show how CAMPBELL incor- this concept can be fruit- sociology of scientific porates this concept into his more recent vehicular fully related to the current knowledge community, theory of knowledge which stresses the embodied debates in the sociology Donald CAMPBELL always character of knowledge. Explicating these two main of scientific knowledge. has been eager to engage pillars of CAMPBELL’s evolutionary epistemology will Examining CAMPBELL’s in a critical dialogue with help us to understand how CAMPBELL develops a so- writings on sociology of the relativist sociologists ciological model of scientific knowledge called a soci- science, I will specifically of science. And during ology of scientific validity which can counter the argue that CAMPBELL’s so- the last fifteen years excessive relativism espoused by the relativist sociolo- ciology of scientific valid- before his death CAMP- gists of scientific knowledge. ity can counter the BELL has been concerned excessive relativism es- with developing a sociological model of scientific poused by the relativist sociologists of scientific knowledge, called the “sociology of scientific valid- knowledge. ity”. In contrast to the relativist model, it is able to explain how science can produce valid beliefs about Knowing as a Vicarious Selection the external world. CAMPBELL often calls his model a Process ‘sociologized version’ of POPPER’s demarcation criterion since it seeks to demarcate science from non- For CAMPBELL, knowing is a profoundly indirect pro- science by asking how the scientific community, in cess in which a punctiform certainty or accuracy in contradistinction to the other belief perpetuating the mapping of the organism’s environment is communities, can improve its “collective” goal of never warranted. Rather, gaining knowledge about validly mapping the physical world. It thus comes the external world requires the organism to inevita- as no surprise that developing such a sociological bly adopt a trial-and-error method. Let me illustrate model of scientific validity takes the first priority in this by citing CAMPBELL’s favorite example. The leg CAMPBELL’s research agenda. As he wrote in his of a lizard, when amputated, regrows to a length notes, his naturalistic epistemology of science even- optimal for locomotion and survival. Such an opti- tually “will also have to be a sociology of scientific mal adjustment process, however, never implies the validity” (editors’ introduction to CAMPBELL 1997). ‘direct fit’ of the organism with its environment. In this short article, I will first briefly discuss the Rather, the leg length is controlled by the internal- concept of the ‘nested hierarchy of vicarious selec- ized ‘vicarious’ selector’ which in turn is the product tors’ which is central to CAMPBELL’s evolutionary of previous natural selection. This vicarious selector

Evolution and Cognition ❘ 75 ❘ 1997, Vol. 3, No. 1 Kyung-Man Kim embodies the ‘already evolved wisdom’ about the and in that sense are cost-effective; second, there will environment obtained through the trial-and-error be a certain systematic bias resulting from the mis- of the whole mutant organism. Such a vicarious match of the vicarious criteria with the changing selector is only contingently related to the environ- environment; third, the increase of vicariousness of ment since the ‘typical’ environment against which selectors in going up the hierarchical ladder means such a vicarious selector has been developed can that, in a certain sense, the higher-level selectors pre- change. In a changed environment, therefore, the suppose, although presumptively, the validity of the current vicarious selector will no longer be optimal. lower level selectors in guiding generation and selec- Take another example; while our eyes seem to give tion of variations at that level. This in turn means us a ‘direct’ and reliable knowledge in the normal the presence of ‘constraints’ on the possible range of environment, they delude us in different environ- variations at that level. Thus, in science, as I shall ments. Consider fog and glass as two such abnormal elaborate below in more detail, while operating with environments. While glass is transparent, it is nev- the currently available vicarious selectors, scientists ertheless impenetrable; and while fog is opaque, it is are groping for more ‘adequate’ or more ‘valid’ vicar- nevertheless penetrable. The presumption that any- ious selectors which can better represent the physi- thing transparent is penetrable (and vice versa) is cal world lurking out there. violated in such abnormal circumstances. Our vision— itself a vicarious selector—sometimes Vehicular Theory of Knowledge deludes us in an ecology which is not typical of the history of human evolution (CAMPBELL 1974). In his William James Lecture delivered at Harvard CAMPBELL argues that such a vicarious selector—or University in 1977 (CAMPBELL 1988), CAMPBELL shortcut—is obtained through the following three spelled out some of the implications which his vicar- steps: ious selection theory has for the naturalistic episte- 1. ‘blind variation’ mology of science. Plugged into the naturalistic 2. selection of variations epistemology, the nested hierarchy of vicarious 3. retention of variation. selection process implies the dependence of the Thus the ‘homeostat’ setting or ‘reference signal’ higher level knowledge processes on the general built into the developmental system of the lizard validity of the lower level ones which are interpreted has been originally obtained through an inductive as developed earlier in the evolutionary time scale. In process which CAMPBELL dubbed the blind-varia- science, we can distinguish at least two kinds of tion-and-selective-retention process. In a changing vicarious selectors (KIM 1992). First, in any given environment, however, such a vicarious selector period of time, we can find certain standards of scien- becomes inappropriate and becomes subject to the tific rationality or acceptability usually represented more fundamental natural selection process. Regard- by the models, theories, inscription devices (LATOUR ing such a ‘nested hierarchy of vicarious selectors’, 1987), and even metaphysical principles (GALISON CAMPBELL (1970, p55) thus argues that “What are 1987), and these constitute the vicarious selectors criteria at one level are but ‘trials’ of the criteria of which contribute to the selective propagation of spe- the next higher, more fundamental, more encom- cific ideas from a larger pool of ideas. These standards passing, less frequently invoked level”. or vicarious selection criteria are in fact the results of Starting from the genetic adaptation as the most the previous selections, and in that sense, have what fundamental level in such a hierarchy, with increas- CAMPBELL calls “historicity” (1997, p15). Having been ing order, CAMPBELL enumerates various levels of vi- selected and retained because they have done better carious selectors. They include non-mnemonic than their conspicuous rivals in the past, these crite- problem solving (i.e, direct locomotion without ria or standards are believed to represent the world in memory), vicarious locomotor devices, habit-in- a fairly valid way. But, as we have seen above, these stinct, visually guided thought, mnemonically criteria are at most imperfect, and has only a heuris- guided or supported thought, social learning, lan- tic value for the representation of the physical world. guage, and finally science, which is but an aspect of They are only tentatively held as true until some sociocultural evolution. From this discussion, we anomalies or counter-examples are encountered. can find that the vicarious selector has at least three Also, as noted above, they tend to have a systematic important implications regarding the process of bias in representing the physical world. knowledge acquisition. First, vicarious selectors are The second type of vicarious selectors are those ‘shortcuts’ which abbreviate more ‘wasteful’ search, which have been used by the sociologists of knowl-

Evolution and Cognition ❘ 76 ❘ 1997, Vol. 3, No. 1 D. T. Campbell’s Social Epistemology of Science

edge to explain the vicissitudes of beliefs observed in BLOOR 1991; COLLINS 1985; LATOUR 1987) in fact different cultures and scientific paradigms. In con- demonstrate that the ‘tribal’ requirements such as trast to the first type, the second type is historically local laboratory culture and practices, social and conditioned by the factors inimical to the selection professional interests, training, and power structure of valid beliefs. In his first paper on the sociology of of the scientific community rather than Nature her- science entitled, “The Tribal Model of Social System self play an important role in the maintenance and Vehicle Carrying Scientific Knowledge” (1979), change of collective beliefs in science. But in his CAMPBELL introduced what he called the ‘vehicular debate with David BLOOR (1989), CAMPBELL (1989) theory of knowledge’. In this theory, knowledge is argues that the relativists fail to specify the role conceived to be the compromise between the phys- played by the natural world, however small, in the ical or vehicular requirements of the system that car- selection of beliefs. Thus, against BLOOR’s argument ries knowledge and the properties of its referents. In that, given a particular need of a group, there is no such a ‘descriptive epistemology’ (CAMPBELL 1977, reason why such a social group cannot invent a 1979, 1988), scientific knowledge is conceived as word for the “three centimeters of a leaf”, CAMPBELL ‘embodied’ in a physical vehicle or carrier which has argues that the “entifiability of the natural world its own physical nature and ‘limitations’ dictated by edits words for a socially pragmatic degree of shared its structure. Just as the knowledge embodied in an reference” (1989, p154; see also KIM 1992). individual organism is a ‘compromise’ between its structural and physical requirements and the exter- Epistemological Fallibilism nal referents which are alien to the organism, so is the knowledge embodied in a social vehicle called The vehicular theory of knowledge discussed above scientific community. In introducing this ‘vehicular is indeed a ‘descriptive epistemology’ delineating theory’ of knowledge, CAMPBELL shows his subscrip- how the current social system of science produces tion to the hypothetical realism in which the opera- knowledge. But, as CAMPBELL repeatedly stresses, his tional independence of knowledge and that-which- descriptive epistemology never precludes the possi- is-to-be-known is emphasized. Such structural re- bility of a specific kind of normative epistemology. quirements that are necessary for the continuity of As I shall discuss below, this is what CAMPBELL the tribal solidarity are however at the same time attempts under the heading “sociology of scientific inimical to the selection of valid beliefs. For CAMP- validity”. Given a descriptive picture of science, the BELL, not only individual scientists have their own next step is to “design” a social system of science in biases and preferences which tend to distort the which the role of the natural world as reflected in valid mapping of the physical world, but also a group the research data can be ‘maximized’. Thus, CAMP- of scientists has its ‘organizational requirements’ BELL argues that, “Descriptive epistemology will which dictate the ‘trade-off’ between keeping the ve- need eventually a physical theory of optimal vehi- hicle intact and validly mapping the physical world. cles … How are these physical requirements … In fact, KUHN’s (1962) stress on the conformity pres- related to the revising, expanding, and improving sure among the members of the paradigm well illus- embodied knowledge? Is total substitution of a dif- trates such vehicular requirements. Thus CAMPBELL ferent portrait [of physical reality] generally more argues that: mechanically feasible procedure than retouching it? [T]he vehicular substance that carries knowledge (CAMPBELL 1997, p17). He recommends the latter is unavoidably alien to the referents of knowl- strategy. edge—it is a different substance with different To illustrate what he has in mind, CAMPBELL intro- structural characteristics. Complete flexibility in duced the notion of doubt-trust ratio by using a bio- depiction, reflection, transmission, or recording, logical example in which the mutations and is precluded by the structural requirements of the recombinations of genes producing new variations vehicle. If the vehicle is completely flexible it lacks in the modal inherited form represents the ‘doubt’ the rigidity to hold together the picture [knowl- (or innovation) and the retained ancestral genes and edge] it carries. These vehicular-structure require- gene complexes represent the trust (or tradition). ments produce not only restrictions on fineness Now, in any given generation, the doubt-trust ratio of detail, but also bias and limitations of aspect is overwhelmingly in favor of trust. Of course, it is (1979, p184). clear that, over a number of generations, all parts of According to CAMPBELL, the ‘strong/constructivist’ the original gene complexes may be replaced. The sociologists of scientific knowledge (BARNES 1982; point here is that, if progress is to be made, both in

Evolution and Cognition ❘ 77 ❘ 1997, Vol. 3, No. 1 Kyung-Man Kim

the biological and in the scientific realm, the doubt- or presuppositions— leads CAMPBELL (1986) to trust ratio should be very small. As is well known, adopt the metaphor of a “hermeneutic circle” or this view of scientific change is widely endorsed by “spiral” to characterize the “validity-enhancing” philosophers including QUINE, POPPER, and HESSE. aspects of science. Granted that knowledge process QUINE’s (1964) rejection of the analytic-synthetic starts from a certain kind of prior understanding— distinction, POPPER’s (1959) background knowledge represented by the presumptive and fallible vicari- in theory testing, and HESSE’s (1974) concept of de- ous selection criteria—more and more hidden pre- gree of ‘entrenchment’ of predicates in a theoretical suppositions and assumptions become known and network all attest to their conviction that any parts articulated in the process of developing and improv- of the theoretical system is in principle revisable. ing knowledge. Of course, we can neither make all They all concur that only by a fallibilistic trust in the of these presuppositions explicit nor can we go back large part of the system, can science grow. to the ‘absolute’ beginning of them. However, Given this, we can ask: at what level of such a something can be gained when we move freely hierarchy of vicarious selectors the most adaptive between different levels of presuppositions and modifications to improve the fit with the physical assumptions. For CAMPBELL, the social system of sci- world are likely to occur? CAMPBELL’s nested hierar- ence must be designed in such a way that it can chy suggests that when an anomaly is found at a facilitate such a hermeneutic tacking procedure. In specific level, it is fed into the more general, deeply such a community of ‘truth seekers’, CAMPBELL entrenched, and less falsifiable level. And if the so- argues, scientists with different theoretical orienta- lutions cannot be found at that more fundamental tions stay together in a focused disputation, attend- level, one should check the still more fundamental ing and monitoring each other’s arguments and vicarious selectors and so on until the most basic and illustrations and keeping each other honest. Such a deeply entrenched level is checked out for its valid- process of argumentation, CAMPBELL argues, will ity. Let me quote CAMPBELL (1977, p51) on this point: maximize the role of the natural world reflected in Many of those who study and theorize about con- the research data, and eventually will contribute to ceptual change in science observe sets of pre- the selection of beliefs in the validity-enhancing sumptions and decision rules which vary direction. Selection theory, when translated into hierarchically in the reluctance with which the the sociology of scientific validity, posits that reality scientific community can or will change them. operates “as one of the selectors among the human The conceptual changes within KUHN’s normal socially generated category systems and hypotheti- science are lower in such a hierarchy than the ba- cal beliefs that are in the contemporary pool of con- sic paradigm changes of scientific revolutions. Of tenders” (CAMPBELL 1994, p xii). course, many more levels are needed than just these two… Perhaps the scientist’s hierarchy of The Conversion of Scientists and the reluctance to change presumptions works back- ward down this developmental sequence, with Sociology of Scientiﬁc Validity the evolutionary earliest being the most tena- While social constructivists have been quite suc- ciously adhered to. I also feel some value in think- cessful in explaining the sustenance of numerous ing about change in science in terms of the controversies in various fields of science by causally metaphor of a nested hierarchy of vicarious selec- linking different interpretations on observational tion processes. Applied to science, most scientists evidence to the starkly different social/professional can be seen as exploring local alternatives holding interests and ideology (i.e., what CAMPBELL calls the a large number of beliefs fixed as vicarious selec- requirements of the tribal continuity) held by tors. Occasionally there are periods of varying opposing theory groups, they are much less success- some of these tentative selectors, and still less fre- ful in explaining how and why such controversies quent variations on still more reluctantly given were closed down. I have argued elsewhere (KIM up selector-beliefs— and so 1991, 1994) that this is inevi- on up to the least frequent table since the relativists, most fundamental varia- Author’s address being preoccupied with ana- tions called scientific revo- lyzing how the ‘tribal’ leader- lutions. Kyung-Man Kim, Department of Sociology, ship and the loyalty to the Such an emphasis on the hier- Sogang University, Seoul, Korea. group determine scientists’ Email: [email protected] archy of vicarious selectors— behavior and arguments, are

Evolution and Cognition ❘ 78 ❘ 1997, Vol. 3, No. 1 D. T. Campbell’s Social Epistemology of Science deprived of any resource that might be used to genetics, authors argue that what is lacking in the explain the conversion of scientists from one theoret- relativist/constructivist analyses of scientific con- ical position to another which is supposedly capable troversy is the detailed explanation of the restructur- of more validly representing the physical world. ing process of the previous network of allies and Assuming that scientific change takes place in a ‘ref- enemies in terms of the changing beliefs of the indi- erential vacuum’, these relativists fail to demon- viduals involved in the controversy. Though such a strate how and why disputants eventually come to a restructuring process of the network necessarily consensus regarding the superiority of one theory involves a description of the changing pattern of over another. allegiance, and hence an account of the converts To demonstrate how ‘reality’ operates as one of who shift their opinions in the course of time, in the the ‘co-selectors’ of beliefs (CAMPBELL 1993), and to relativists’ case analyses, numerous scientists who show how collective belief change in science can be changed their positions and hence contributed to made comprehensible, CAMPBELL argues that the so- the consensus change are simply brushed aside. It is ciologists of scientific validity should attend to: true, as the relativists argue, that some scientists the hundred or so instances (by implication, a may dogmatically adhere to their intellectual off- sampling from thousand) in which ideas, proce- spring. However, these studies show that many dures, and strategies are proposed and rejected, geologists and biologists did change their beliefs in because they won’t work. This “not-working” is the course of time, in response to the persuasive often decided by thought and argument, and of- rhetoric of others based on experimental observa- ten by laboratory efforts. Some of these rejections tions. By challenging the relativists/constructivists no doubt involve blind social conformity to local- to come up with a plausible causal scenario as to the ly preferred belief. But probably only a small pro- ‘role’ played by the natural world in the collective portion. We need an expanded, still more belief change in science, CAMPBELL’s selection the- microprocess, sociology of such idea-winnowing ory teaches us that any plausible social theory of sci- (CAMPBELL 1986, p118). ence should include a careful internal description of Such a proposal for a microsociology of idea-win- how scientists with different locations in an inter- nowing has been actually taken up by several histor- acting network emerge as converts as a result of the ical sociologists of science (e.g., RUDWICK 1985; KIM focused disputations on purported observations 1994, 1996). In these studies of consensus forma- and mutual monitoring of each other’s arguments tion in 19th century geology and early 20th century and illustrations.

References hancing Collective Belief Change and the Problems of the Social Sciences. In: D. Fiske and R. Shweder (eds) Metathe- Barnes, B. (1982) T. S. Kuhn and Social Science. Columbia ory in Social Sciences. University of Chicago Press, Chica- University Press, New York. go, pp 108-135. Bloor, D. (1989) Professor Campbell on Models of Language- Campbell, D. T. (1988) Methodology and Epistemology for Learning and the Sociology of Science: A Reply. In: Fuller, Social Science: Selected Papers. University of Chicago S./DeMay, M./Shinn, T./ Woolgar, S. (eds) The Cognitive Press, Chicago. Turn: Sociological and Psychological Perspectives on Sci- Campbell, D. T. (1989) Models of Language Learning and ence, Dordrecht, Reidel, Netherlands, pp159–165. Their Implications for the Social Constructionist Analyses Bloor, D. (1991) Knowledge and Social Imagery. University of of Scientific Belief. In: Fuller, S./DeMay, M./Shinn, T./ Chicago Press, Chicago. Woolgar, S. (eds) The Cognitive Turn: Sociological and Campbell, D. T. (1970) Natural Selection as an Epistemologi- Psychological Perspectives on Science, Dordrecht, Reidel, cal Model. In: Naroll, R./Cohen, R. (eds), A Handbook of Netherlands, pp153-158. Method in Cultural Anthropology. Natural History Press, Campbell, D. T. (1993) Plausible Coselection of Belief by Ref- New York, pp51–85. erent: All the ‘Objectivity’ That is Possible. Perspectives on Campbell, D. T. (1974) Evolutionary Epistemology. In: P. Science: Historical, Philosophical, Social, 1: 88–108. Schilpp (ed) The Philosophy of Karl Popper, Open Court, Campbell, D. T. (1994) Toward a Sociology of Scientific Valid- La Salle, pp 413–463. ity. In: Kim, K. M., Explaining Scientific Consensus: The Campbell, D. T. (1977) Descriptive Epistemology: Psycholog- Case of Mendelian Genetics. Guilford Press, New York, pp ical, Sociological, and Evolutionary. William James Lec- ix–xxi. ture, Harvard University. Campbell, D. T. (1997) From Evolutionary Epistemology Via Campbell, D. T. (1979) A Tribal Model of the Social System Selection Theory to a Sociology of Scientific Validity. Evo- Vehicle Carrying Scientific Knowledge. Knowledge 1:181– lution and Cognition. 201. Collins, H. M. (1985) Changing Order: Replication and In- Campbell, D. T. (1986) Science’s Social System of Validity En- duction in Scientific Practice. Sage, Beverly Hills and Lon-

Evolution and Cognition ❘ 79 ❘ 1997, Vol. 3, No. 1 Kyung-Man Kim

don. ences 26:3–24. Galison, P. (1987) How Experiments End. University of Chi- Kuhn, T. S. (1962) The Structure of Scientific Revolution, cago Press, Chicago. University of Chicago Press, Chicago. Hesse, M. B. (1974) The Structure of Scientific Inference. Uni- Latour, B. (1987) Science in Action. Harvard University Press, versity of California Press, Berkeley. Cambridge, Mass. Kim, K. M. (1992) The Role of the Natural World in the Theo- Popper, K. R. (1959) The Logic of Scientific Discovery. Hutch- ry Choice of Scientists. Social Science Information 31: 445- ison, London. 464. Quine, W. V. O. (1964) From a Logical Point of View, Harvard Kim, K. M. (1994) Explaining Scientific Consensus: The Case University Press, Cambridge, Mass. of Mendelian Genetics. Guilford Press, New York. Rudwick, M. J. S. (1985) The Great Devonian Controversy. Kim, K. M. (1996) Hierarchy of Scientific Consensus and the University of Chicago Press, Chicago. Flow of Dissensus over Time. Philosophy of the Social Sci-

Evolution and Cognition ❘ 80 ❘ 1997, Vol. 3, No. 1 Marc De Mey

Vision as Paradigm: From VTE to Cognitive Science1

IAGET (in BRINGUIER Abstract would like to be remem- P 1977, p37) once bered as “selection the- complained about fash- Central concepts of CAMPBELL’s “evolutionary episte- ory”. Selection of what by ions and fads dominating mology” or “selection theory” go back to his years as a whom? the study of psychology psychology student when behaviorism was very much A recurrent theme in in big countries like the in fashion. However, VTE (vicarious trial and error) is Don CAMPBELL’s “per- US or the Soviet Union. already a cognitive notion from TOLMAN expanded sonal retrospective his- Apparently like flocks of into a general epistemological category. Linking the tory of ideas” is the role of birds, researchers in those paradigmatic case of vision to action even maintains vision. In his visionary countries all seemed central concepts of an older profound tradition in the view of knowledge pro- heading for the same combined study of knowledge and behavior. This il- cesses, the case of appar- direction, closely sticking lustrates a remarkable steady line of research rising ently simple visual together. All of sudden, above the fluctuations of fashions while remaining perception has always one bird would change sensitive to new findings. had an important posi- direction and, somewhat tion but in this paper its surprisingly, all the others would promptly follow prominence seems more strong than ever. It is pro- and pursue the new direction. He felt these rather moted into the role of prototypical knowledge: “I abrupt changes, coupled to reorientation cycles of now wish to identify ‘knowledge’ with point 2.1.2 roughly ten to fifteen years, as very detrimental to of the Basic Selectionist Dogma, i.e. those vicarious science. A meaningful line of research, once taken, processes which short-cut selection by the life and should not be prematurely ended and should be fol- death of genetic variants. Visual perception is the lowed through tenaciously until fully explored and most important of these.” (p10, italics ours) If vision thoroughly assessed. Though D. T. CAMPBELL might is so important, do recent findings allow us a better occasionally seem to explore apparent evolutionary appreciation of this claim? advantages connected to the type of collective In a forthcoming book on visual perception, behavior that upset PIAGET , it does not seem to Stephen PALMER (in press) introduces the explana- apply to his own line of thinking. From the early tion of vision as the most successful achievement time of his graduate study at UC Berkeley during the of cognitive science. Current research on vision heydays of behaviorism until the post-cognitivism owes much to David MARR. MARR represents for the of connectionism, he remained dedicated to a very study of vision what CHOMSKY means for the study specific line of research in the study of the phenom- of language: an increase in both depth and scope. enon of knowledge. Despite its very wide range, his It was his ambition to develop a general theory of approach to this phenomenon preserved a very per- vision that would accommodate both natural sys- sonal character, combining the rigor of his behav- tems, the eyes and brains of men and animals, as iorist education with the daring speculation of an well as vision devices developed within artificial in- evolutionist interested in that ephemeral organ telligence. His posthumous book Vision (1982) called “mind”. Can one remain a loyal behaviorist brings together the results of various disciplines and be fascinated by the mind during a whole life- dealing with vision, from neurophysiology to arti- time? Donald CAMPBELL managed to do so. Out of ficial intelligence, with the aim of developing an all the tension between these two opposites he forged encompassing scheme for the study of visual per- his remarkable “evolutionary epistemology” that he ception. Though criticised and amended at various

Evolution and Cognition ❘ 81 ❘ 1997, Vol. 3, No. 1 Marc De Mey

points, MARR’s model remains a powerful integra- Trial-and-error is linked to THORNDIKE’s historical tive model to deal with vision as it ranges from ret- research on problem solving by cats. Just before inal sensitivity to conceptual interpretation. Two 1900, THORNDIKE studied how cats managed to es- major features stand out as characteristic for his ap- cape from puzzle boxes with complicated locking proach, one peculiar to MARR, the other widely devices. Trial-and-error is the apparently random be- shared by almost all current models. The generally havior that those cats produce to free themselves accepted feature is the modularity of the visual sys- from the encaged situation. They turn around and tem. The feature peculiar to MARR is the role attrib- push at various handles and buttons, jump up uted to an intermediate integration stage based against the bars of the cage, etc. until, by accident, upon perspective. Supported by neuroanatomical they successfully hit the opening mechanism. At the and neurophysiological findings, the prevailing most primitive level, neither the unsuccessful nor notion nowadays is that vision comprises a con- the successful behaviors are memorised. When the glomerate of many relatively autonomous sub- annoying situation would be installed again by lock- systems working together in intricate ways. The ing up the same cat in the same case, it would mean actual count runs into the thirties! MARR’s achieve- that an entirely new sequence of random behaviors ment is to provide a sequential orchestration for would be produced again until ultimately freedom these subsystems. The skeleton of his system com- is reached once more. As is known, THORNDIKE found prises three stages, 2D, 2.5D and 3D, characterising out that cats do not demonstrate insight but they do the various intermediate results in the sequential exhibit learning. The behavioral segments just pre- and progressive assemblage of the percept. These ceding the escape will increase in probability and the stages show some resemblance to established time needed to get away will gradually decrease. The schemes for drawing and painting with units of Law of Effect corresponds to CAMPBELL’s level 4: analysis such as lines, oriented surfaces and vol- learning. On level 3, the selector is still the environ- umes. Whether one accepts or rejects MARR’s partic- ment. The lock yields only to a specific way of han- ular model, the main challenge in vision studies dling it. On level 4, memory becomes a vicarious seems to be to account for the integration of these selector, imposing a specific probability distribution many relatively independent modules. Despite its upon the behaviors tried. But where is the vicarious apparent simplicity, what is seen in a single eye- selector in vision? glance is the result of a complicated combination. Obviously, the notion of vicariousness came from Visual perception involves a plurality of processes. TOLMAN, CAMPBELL’s mentor at UC Berkeley. A rat Is it in those terms still suitable for its pivotal role learning a maze is not just chaining responses as in selection theory? THORNDIKE’s Law of Effect might suggest. That is In the daring scope of his original hierarchical manifest from the flexibility in behavior when alter- classification of knowledge processes, CAMPBELL native actions are imposed. While being trained (1959) inserted “perception” almost in the middle through various alternatives, the rat constructs a of a series of embedded “vicarious” processes (level “cognitive map” of the maze. Confronted with new 5). The impressive character of that classification challenges (a barrier on an otherwise unobstructed came from the power of coupling random variation path) the rat will demonstrate VTE (Vicarious Trial and selective retention with “vicariousness”. From and Error). That means that the trial and error is to the second level onwards of “sexual reproduction” a substantial extent executed on the cognitive map. up to level 11 of “modern science” and 12 of “ma- But to what degree is this perception and to what chines” (the early AI promises), each new level is degree action? CAMPBELL has always been intrigued reached through some substitute for the variation by an imbedding of perception into action. Contem- mechanism that reduces risk and secures preserva- plation is not the prototypical perceptual act. Per- tion of what has already been achieved. Level 5 perception is subordinated to action. To the degree that ception is a substitute for level 3 trial-and-error. the environment resists specific actions, it is forced Instead of running deadly risks by running around to yield information. The favorite model is the an- in a potentially dangerous environment (level 3), cient image of the blind man with the stick. To per- visual exploration (level 5) allows for the explora- ceive his environment, he has to poke around with tion of that environment by just looking. The gaze the stick and he only “knows” the environment to is a safe substitute for actively walking around in a the degree he is hitting it. CAMPBELL is even more spurious space. Ultimately, this is a combination of fond of the echolocation paradigm. Through some THORNDIKE and TOLMAN. behavior, the organism is acting upon the environ-

Evolution and Cognition ❘ 82 ❘ 1997, Vol. 3, No. 1 Vision as Paradigm: From VTE to Cognitive Science ment and specific knowledge is derived from the way strange items from a philosopher’s point of view: the environment reacts to it. In a more basic sce- Blind trial-and-error exploration (discovery and/or nario, the primitive organism will apply its action to learning on the part of a blind person) is included. whatever it encounters, say eat it, and be vulnerable The resulting ‘knowledge’ is so similar to that pro- to the radical selection by either staying alive or dy- vided by vision that it would be remarkable if no ing. In perception, especially for the distal receptors, philosopher had attended to it. (I await help from the action of the organism is less risky, and the re- historians of epistemology.)” (this volume, p8). sponse of the environment provides the segments of History of science can provide some of the re- the vicarious selector that is knowledge. But where quested help. The more convincing models of vision is the action component in vision? How is echoloca- like EUCLID’s and PTOLEMY’s started out as extramis- tion applied in this celebrated case? sion theories. In their views vision had to be con- Vision is the most remarkable sense because it ceived of as a dynamic force, unfolding with rays makes use, in the truly opportunistic way evolution originating from the eye and propagated like sticks always seems to rely on, of a very reliable component extended into the world until they hit the objects of in our environment: the sun. Given the regularity of vision. But modern robot vision systems too main- its behavior, life on earth can be sure of the action of tain a close affinity to echolocation models with this ultimate battery also as an echolocation emitter. range finders based on reflected beams of laser light. There is no need to poke around with the stick if an When Arab and late medieval scholars found out omnipresent object is immersing the environment that vision could be better seen as based on intromis- in light at predictable time intervals. The sun pokes sion, rays coming in rather than leaving from the the environment with its light and organisms with eye, visual perception maintained its status as proto- eyes figure out how that light affects their world. typical knowledge process. There is a remarkable Reflected light is the echo resulting from the action continuity throughout an erratic trajectory of epis- of life’s most reliable ally. Vision might be so power- temologies from Saint Augustine and thirteenth cen- ful because of the lavish and inexhaustible power of tury theologians to the Age of Enlightenment up to this giant companion. If this “vicarious” action of our own time. All of them long for light as a means the sun is ignored, vision data become epstemolog- for obtaining more knowledge. Enlightenment and ically arcane and shaky. CAMPBELL quotes the case of enlargement of cognition are equated. Even the em- a pioneering vision scientist: blem of CAMPBELL’s university, UC Berkeley carries “DESCARTES got to his kepticism about vision from the phrase “Let there be light”. The metaphor what he took to be the illusory vividness of his own couldn’t be that powerful unless it would in some dreams, from an up-to-date knowledge of the phys- deep sense be true. ics, anatomy, and physiology of vision, and from a As indicated, a characteristic of modern theories pathological need for certainty. But his analysis has of vision is the emphasis on modularity. The action been part of the great tradition of perceptual skepti- of the light is not only powerful but also multiform. cism back to the pre-SOCRATICS. PLATO’s parable of Various modules capture various aspects of the re- the cave (bk. 7 of The Republic) has that theme. The flected light so as to compose a perceptual object “strange prisoner” are “like ourselves.” They see only with genuine “entativity”. Whether along the se- shadows. To them, the truth would be literally noth- quential line of MARR or some more ad hoc combi- ing but the shadows of the images.” In this allegory, nation, the current challenge for theories of vision “the prison house is the world of sight.” Note how is to find out how the modules combine to yield an compatible this is with our modern physics and integrated percept. One wonders to what degree the physiology of perception, in which the brain reifies co-operation between the modules could be handled objects from patterns of light indirectly and superfi- in terms of the social epistemology for which CAMP- cially reflected from them.” BELL became increasingly interested. While relatively But what else can CAMPBELL mean apart from the solid, the entativity of visual objects is ultimately emitter action when empha- also fallible. As the Magic Eye sising vision’s affinity with illustrations show, a few ste- Author’s address blind trial and error in the re- reoscopically integrated mod- vised version of “evolutionary ules can make illusions look Marc De Mey, Blandijnberg 2, 9000 Gent, epistemology”: veridical. The suggestion is Belgium. “This revision still includes Email: [email protected] that what accounts for the en- as knowledge processes some tativity of perceptual objects

Evolution and Cognition ❘ 83 ❘ 1997, Vol. 3, No. 1 Marc De Mey and for the solidity of scientific theories should has undoubtedly been an inspiration for both of somehow be comparable. It is a remarkable coher- them. Before the onset of behaviorism, it was clear ence that makes the various visual features captured that extending psychology with the study of animals by the various modules agglutinate into a single per- and children would ultimately increase the episte- ceptual object. It is a remarkable coherence that mological scope of psychology and the understand- keeps together the achieved in new combinations ing of cognition. At the time behaviorism was toward even superior adaptability. Figuring out the proclaimed, this was understood by Edwin HOLT principles of this grouping, the ability of keeping who accepted behavior as the basic observational together what has been selected, either directly or category but, in contrast to WATSON, did not do away “vicariously”, remains a major challenge for evolu- with knowledge as a psychological entity. PIAGET’s tionary approaches to cognition. In contrast to genetic epistemology might have been tributary to PIAGET, CAMPBELL has not explored logic but social BALDWIN, but CAMPBELL’s “evolutionary epistemol- epistemology along the pursuit of this question. ogy” took the line of Edwin HOLT and TOLMAN who, Nevertheless, they have much in common. while they adhered to behaviorism, never dropped Though CAMPBELL was quite conversant with cognition. That is the line of thought to which PIAGET’s work and his “biological epistemology”, his Donald CAMPBELL remained loyal and to which he name doesn’t figure in the final paper. However, the contributed so much by including the social episte- work of James Mark BALDWIN is mentioned and he mology that cognitive science still has to discover.

Notes

1 This comment was prepared while the author was visiting scholar at the Institute for Cognitive Studies at the Univer- sity of California Berkeley as Peter Paul RUBENS professor during the spring semester of 1997.

References comparative psychology of knowledge processes, Inquiry, 152–182. Bringuier, J.-C. (1977) Conversations libres avec Jean Piaget, Palmer, S. (in press) Vision Science. An Interdisciplinary Ap- Paris, Lafont. proach, Cambridge, Mass., Bradford Books, MIT Press. Campbell, D.T. (1959) Methodological suggestions from a

Evolution and Cognition ❘ 84 ❘ 1997, Vol. 3, No. 1 Erhard Oeser

The Two-stage Model of Evolutionary Epistemology

s Donald T. CAMP- Abstract ysis of evidential systems. A BELL already showed For a long time POPPER was in an essay (1974) dedi- The evolutionary or “phylogenetic” epistemology, as reluctant to accept the cated to Karl POPPER, it is it was called originally by Konrad LORENZ, deals with theory of evolution as a not new to apply the the- the hereditary mechanisms of cognitive organismic natural scientific theory ory of evolution to the systems—valid for man or animal—and is a logical and accused it of having a cognitive capacity of consequence of the biological theory of evolution. circular character. He man. A succession of phi- Called EEM (Evolutionary Epistemology of Mecha- considered DARWINISM to losophers and natural sci- nism) by BRADIE (1986), it was set against EET (Evo- be a medium of battle entists speculated (right lutionary Epistemology of Theories) merely as a against CARNAP´s Induc- after the publication of concept of analogy, as developed by POPPER and later tivism, putting forward DARWIN’s Origin of Species) by TOULMIN and others. Here, however, evolutionary an analogy between DAR- that a priori forms of per- philosophy of science should be recognized as a sec- WINISM and LAMARCKISM, ception and categories of ond stage of phylogenetic epistemology, which doesn’t as well as selection and in- thinking are the product deal directly with the cognitive organismic systems, struction (instruction of phylogenetic develop- but with the scientific method as a goal-orientated through repetition), on ment. Besides, DARWIN cognition activity of man. In this respect MACH and the one hand, and deduc- himself has ascribed this BOLTZMANN had already made use of the evolution- tivism and inductivism, point of view in a most ary-epistemological approaches, found previously by as well as critical elimina- general way to the Darwin himself, and applied them to the ontogeny of tion of mistakes and posi- intended uses of his the- physics, intending to show the evolution of natural tive justification, on the ory of evolution, as can scientific methods of cognition as a self-infringement other. To characterize his be understood clearly of the fundamental cognitive mechanism, which orig- own position POPPER nowadays from his note- inally merely served survival. added that his logic of re- books (BARRET 1980). search contains a theory Since DARWIN it was clear, also, that here a base of knowledge-growth through trial and error—that can be found which establishes in a natural scientific is, through the elimination of mistakes. This means, way the evolutionary dynamics of scientific theo- however, knowledge-growth through DARWINIAN series. It was not the biologists, but physicists like lection, instead of LAMARCKIAN instruction. MACH and BOLTZMANN, who consequently expanded POPPER elaborated this analogy at the end of his the evolutionary approach to the ontology of their life into a four-phase-scheme of the dynamic of the- own discipline. This first attempt, however, became ories (POPPER, 1994, p33): stuck in a metaphorical view, because biology itself, 1. Problem (not observation) and in particular genetics and behavioral science, 2. Attempts for solution = hypotheses had not developed sufficiently to enable it to recog- 3. Elimination = refutation of hypotheses or theories nize hereditary mechanisms of cognition as geneti- 4. New and more sharply defined problem that de- cally conditioned ways of behavior, these being velops out of the critical discussion species-specific characteristics in the same way as an- As POPPER noted himself, this scheme was supported atomical structures. by Konrad LORENZ at the deeper level of comparative In our century the renewal of this attempt behavioral research. Thus it can be said that the old emerged from POPPER’s fundamental objection that ideas of the evolution of science (as anticipated by the really interesting dynamic aspects of evolution MACH and BOLTZMANN) have not only been renewed and variation in science escape the mere logical anal- by POPPER but that modern behavioral science, too,

Evolution and Cognition ❘ 85 ❘ 1997, Vol. 3, No. 1 Erhard Oeser has delivered a natural scientific base which is far like law, state and economy—as already demon- more than a mere analogy. strated by Social DARWINISM. Therefore the real history of evolutionary episte- The evolutionary philosophy of science as a sec- mology as a program of research founded on natural ond phase of evolutionary epistemology however is science starts with Konrad LORENZ, who showed not merely a metaphorical view of the dynamics of through his ethology, based on evolutionary theory, theories in evolutionary disguise, but a further devel- the way back to the unequivocal fundamental simi- opment of the basic trial-error-mechanism related to larities of the cognitive mechanism of all beings to a scientific methods of cognition. Therefore its object homological base, that is back to a real relationship. is not the evolution of theories (as already recognized In this sense three stages of argument, clearly sepa- clearly by BOLTZMANN) but the evolution of the scien- rated from each other, can be differentiated: tific method (OESER 1984) as man’s goal-oriented ac- 1. The biological theory of evolution tivity of cognition, that hastens the experience- 2. Evolutionary epistemology scientific theories. 3. Evolutionary philosophy of science In this respect MACH and BOLTZMANN have already The evolutionary epistemology that deals with employed the evolutionary-epistemological at- the hereditary mechanisms of cognitive organismic tempts, previously traceable to DARWIN himself, for systems—valid for man or animal—is in a narrower the epistemology of physics. Their goal was to show sense a logical consequence of the biological theory that the evolution of natural scientific methods of of evolution. It was called EEM (Evolutionary Epis- cognition is a self-infringement of the fundamental temology of Mechanism) by BRADIE (1986) and put “cognitive mechanism” which originally merely in opposition to EET (Evolutionary Epistemology of served survival. Evolutionary epistemology as phylo- Theories) merely as an analogy, developed by POP- genetic epistemology, however, can merely offer a PER and later by TOULMIN 1963 and others. Here, necessary, but not a sufficient, explanation of the however, the evolutionary philosophy of science functioning as well as the failure of the human per- should be recognized as a second phase of phyloge- ception mechanism in all fields of scientific knowl- netic epistemology, which doesn’t deal directly edge. Therefore, the following must be added: with cognitive organismic systems, but with prod- 1. EE has to be extended by an ontogenetic episte- ucts of epistemology of a certain biological species, mology (in the sense of PIAGET), which should pro- that is homo sapiens, who is the only living being vide the fundamental explanation of the able to practice science. In this respect the evolu- constructive enlargement of hereditary percep- tionary philosophy of science is linked only indi- tional ability in man. This enlargement can be rectly with the biological theory of evolution. The found in all scientific processes of cognition, espe- logical-systematic link between biological theory of cially in the natural sciences. evolution and evolutionary philosophy of science is 2. Furthermore, as both epistemologies are linked evolutionary epistemology in the narrower sense of through the relation of phylogeny and ontogeny a “phylogenetic” epistemology, as it was under- and refer primarily to the macro-level of directly stood by LORENZ. observable behavior, they have to be completed by Every attempt to transfer the biological theory of a causal explanation of the macro-phenomenon- evolution directly to the phenomenon of “science” creating mechanisms on the micro-level of neural has a merely metaphorical or at best analogical char- structures, that is through a “neuroepistemology” acter. Historical as well as recent examples of such (OESER/SEITELBERGER 1988). In this way the alleged metaphorical viewpoints are phrases like “evolution contradiction between adaptionism and construc- of sciences” (TOULMIN 1963) or “evolution of scien- tivism in the sense of competing natural scientific tific hypotheses and theories”, whereby single fields research programs of epistemology (LORENZ and of knowledge or theories are looked upon as biolog- PIAGET) can be abolished. A concrete neurobiolog- ical species, which are able to live and die, compete ical approach is supplied by the “Neural Darwin- with each other (POPPER), ism” of G. EDELMAN (1987), change themselves or are even which is expandable in the Author’s address able to fight against (MACH) field of evolutionary episte- and kill each other. Such met- mology to a “selection-the- Erhard Oeser, Institut für Wissenschaftstheo- aphoric is harmless in this ory of meaning”. With such rie, Universität Wien, Sensengasse 8, A-1090 field, but can become a dan- Wien, Austria. a theory it can be estab- gerous ideology in other fields lished that in “learning

Evolution and Cognition ❘ 86 ❘ 1997, Vol. 3, No. 1 The Two-stage Model of Evolutionary Epistemology

through experience” it is not the environment Such a model is—in the sense of a distinction be- which instructs the cognitive system, but, on the tween the internal and external history—an internal contrary, it is the cognitive system that achieves model of the development of problems, hypotheses meaning or internal so-called mental representa- and theories, where abstraction is made from socio- tions through constructive selection (that is, selec- psychological as well as social-political contexts (in tion and linkage of neural activity which is contrast to CAMPBELL´s “Sociology of Scientific Valid- meaningless on its own). ity”). There is an old saying of ARISTOTLE: “Who ab- Completing the phylogenetic epistemology in that stracts and knows that he abstracts, doesn’t make a way, the simple but fundamental trial-error-mecha- mistake.” But the point is not—as POPPER says—to nism (which is characteristic for all living beings construct a personal PLATONISTIC world of problems, “from amoeba to EINSTEIN” (POPPER) can be ideas and theories way above the individual-social expanded to this model, which mirrors the interplay reality but just to demonstrate the proof of an inter- of theory and experience in the empirical sciences: nal logic of scientific knowledge that is independent T3 from individual social and political decisions only in its result, not in its real temporal course. Whoever denies this kind of independent scientific cognition, T2 denies the autonomy of science and with that, in the end, the truth of scientific cognition, which, in the context and on the foundation of evolutionary epis- T1 temology, in the first as well as in the second phase, primarily has a functional pragmatic character. To the extent that empirical constructs, i.e, hy- H3 H2 H1 P P P 1 2 3 potheses and theories, are used in instructing human actions, they will always be confirmed or disproved. I1 That is to say that there is always a majority of hypotheses and theories, the better functioning theory

I2 with the better explanatory and predicative power will be chosen. This means that science, primarily, is nothing I3 other than a second-order survival machine which is Proceeding from empirical information (I), which built on the fundamental phylogenetically acquired is, however, always determined by the already exist- reaction scheme and leads to more and more coming structural information of our phylogenetically plex, theoretically established instruction schemes created perception mechanism, scientific (that is ex- for successful action, which exceeds indeed the sur- panding the everyday-experience) hypotheses (H) vival of the single individual, but which, in the end, are created that lead to theories (T). From these the- is always linked to the reproductive success of the ories derive prognoses (P), that can be tested reduc- biological species homo sapiens. However scientific tively and can be improved on a higher level through progress should be judged, one thing is absolutely new hypotheses and theories. clear: the enormous reproductive success, that is, the This model of scientific progress, which develops explosive increase of the biomass of the biological spirally through self-correction, mirrors the self-or- species homo sapiens, can be traced back to the ap- ganization of information in the human perception plication of natural scientific knowledge in technol- process, where trial and error, theory and experience ogy, medicine and economics. This success is the are linked together narrowly through formal, more only solid guarantee for that which we call the or less exactly definable, methods like induction, de- “truth” of scientific theories. duction, construction and reduction.

References Schilpp, P. (ed) The Library of Living Philosophers, Vol 14 I and II. The Philosophy of Karl Popper, Vol. I, Lasalle, Barret, P./Gruber, H. (1980) Metaphysics, Materialism and Open Court, pp. 413–463 the Evolution of Mind, Early Writings of Charles Darwin, Edelman, G. M. (1987) Neural Darwinism. The Theory of The University of Chicago Press, Chicago Neural Group Selection. Basic Book, New York. Campbell, D. T. (1974) Evolutionary Epistemology. In: Lorenz, K. (1941) Kants Lehre vom Apriorischen im Lichte

Evolution and Cognition ❘ 87 ❘ 1997, Vol. 3, No. 1 Erhard Oeser

gegenwärtiger Biologie, Blätter für Deutsche Philosophie Wuketits, F.M. (ed) Concepts and Approches in Evolu- 15, pp 94–125. tionary Epistemology, D. Reindel Publishing Company, Lorenz, K. (1974) Die Rückseite des Spiegels. Versuch einer Dortrecht/Boston/Lancester, pp. 149–184. Naturgeschichte menschlichen Erkennens. Piper, Oeser, E. (1987) Psychozoikum. Evolution und Mechanis- München, Zürich. mus der menschlichen Erkenntnisfähigkeit. Paul Parey, Lorenz, K. (1977) Behind the Mirror. Hartcourt Brace Jo- Berlin and Hamburg. vanovich, New York. Oeser, E. (1988) Das Abenteuer der kollektiven Vernunft. Lorenz, K. (1992) Die Naturwissenschaft vom Menschen. Evolution und Involution der Wissenschaft. Paul Parey, eine Einführung in die Vergleichende Verhaltensfors- Berlin and Hamburg. chung. Das “Russische Manuskript”. Piper, München, Oeser, E. (1990) Evolution und Selbstkonstruktion des Re- Zürich. chts. Rechtsphilosophie als Entwicklungstheorie der prak- Mach, E. (1885) Die Analyse der Empfindungen un ddas Ver- tischen Vernunft. Böhlau, Wien. hältnis des Physischen zum Psychischen, G. Fischer, Jena Oeser, E. (1994) Neuroepistemologie und Selbstorganisa- (3rd Edition, 1902). tion, Sofia (Bulgarian). Mach, E. (1905) Erkenntnis und Irrtum. Joh. Ambrosius Bar- Oeser, E./Seitelberger, F. (1988, 2nd Edition 1995) Gehirn, th, Leipzig. Bewußtsein und Erkenntnis. Wissenschaftliche Buchge- Mach, E. (1919) Prinzipien der Wärmelehre, Joh. Ambrosius sellschaft, Darmstadt. Barth, Leipzig. Popper, K. R. (1972) Objective Knowledge: An Evolutionary Mach, E. (8th Edition 1921) Die Mechanik in ihrer Entwick- Approach, Clarendon Press, Oxford/England. lung, F. A. Brockhaus, Leipzig. Popper, K. R. (1982) Ausgangspunkte. Meine intellektuelle Mach, E. (4th Edition 1910) Populär-wissenschaftliche Vor- Entwicklung, Hoffmann und Campe, Hamburg. lesungen. Joh. Ambrosius Barth, Leipzig. Popper, K. R. (1984) Auf der Suche nach einer besseren Oeser, E. (1976) Wissenschaft und Information, Vol. 3: Welt, Piper, München, Zürich. Struktur und Dynamik erfahrungswissenschaftlicher Sys- Toulmin, S. (1963) The Evolutionary Development of Natu- teme. Oldenburg, Wien-München. ral Sciences, in: American Scientist 55. Oeser, E. (1984) The Evolution of Scientific Methods. In:

Evolution and Cognition ❘ 88 ❘ 1997, Vol. 3, No. 1 Henry Plotkin

Knowledge and Adapted Biological Structures

is a tribute to the Abstract evolutionary epistemol- It extraordinarily pro- ogy should be seeking ductive life of Donald contact, not philosophers In this, one of his last pieces of writing, CAMPBELL CAMPBELL that Evolution argues for restricting usage of the terms knowledge for whom the science is ir- and Cognition should be and knowledge processes to vicarious processes that relevant. able to publish posthu- short-cut natural selection. The counterargument is The notion that makes mously so stimulating a presented that it is scientifically counterproductive to the contact is a hierarchy paper by him. As always, do this, whatever virtues it may have for communica- of knowledge-gaining pro- and right to the end of his tion with philosophers. cesses with knowledge as a life, CAMPBELL made us all series of nested products of think. Here I consider just one of the key points to such a hierarchy. This is a most elegant conception, which CAMPBELL addresses himself. This is what whether the hierarchy be that favored by CAMPBELL or CAMPBELL now considers to be the overinclusive one of some other form (for example, PLOTKIN/ usage of the terms knowledge and knowledge pro- ODLING-SMEE 1979; RIEDL 1980). It is at once a concep- cess. Adaptive organic structures, he asserts, should tually simple scheme, and yet it also provides us with not be’ treated as knowledge because “it is a needless an architecture of appropriate complexity for some- obstacle in making contact with the traditions of thing as complicated as human knowledge, whether philosophical epistemology” (section 2.2). As I it be that of the individual or of a collective kind, for understand it, and as CAMPBELL concedes, this is a instance science itself. But an essential part of the con- major shift from his writings of an earlier period. ception is seeing adaptative organic structures as Well, it depends surely on what one wants from knowledge and the processes that gain such knowl- evolutionary epistemology or selection theory. It edge as knowledgegaining processes. must, incidentally, be an error to assume that these Evolutionary epistemology as science also makes are the same thing. Selection theory is a specific form the crucial point that knowledge is a life-or-death of evolutionary epistemology, which I take to be the matter. Why humans have the capacity for acquiring view that the human ability to know is a consequence so much knowledge, much of it seemingly trivial, is of evolution. The former, selection theory, is not an interesting issue. One possible answer is that it is mandated by the latter. But that is to digress. To re- only seemingly trivial and not actually so. Whatever turn to the main point. Evolutionary epistemology the answer, it should not obscure the fact that with- can be made to serve different purposes. For those out knowledge of physical and social causation, with- who want to make the crucial point that a complete out knowledge of spatial relationships, and without scientific understanding of what we as humans know, knowledge of motor and linguistic skills, we would and how we come to have knowledge, must be rooted not survive as we do, and may not survive at all. in evolutionary biology, contact with philosophical The central importance of our capacity to gain such epistemology is not relevant. The reason for this is knowledge for human survival is mounting from the that since evolutionary biology is science, then so too direction of developmental psychologists. There is must be evolutionary epistemology. From DARWIN strong evidence that we come into this world with a and T.H.HUXLEY onwards it has been presented as sci- priori knowledge (see ELMAN et al., 1996 for a minimal- ence, and if it is not science, then surely it is nothing. ist nativist position; and SPERBER et al., 1995 for exam- If this is accepted, then philosophers will not, in any ples of a more explicit content to the knowledge of event, be swayed philosophically by what it has to say. infants at birth or soon after), evidence that must On the other hand, fellow scientists, especially social have delighted CAMPBELL as a proponent of evolution- scientists, are open to persuasion by an argument that ary KANTIANISM. The point of this developmental evi- is rooted in science. That is the audience with whom dence is that no matter how minimal cognitive

Evolution and Cognition ❘ 89 ❘ 1997, Vol. 3, No. 1 Henry Plotkin predispositions are at birth, they must point to genes and hence the order of cause-effect relationships, or as repositories of information leading to such predis- the need for energy resources in the environment to positions. Such genetic information would not exist be exploited in order to maintain a state of negative were it not for a history of selection for such cognitive entropy, both organic structures and knowledge predispositions, and the selection could only have oc- structures are at once assumptions about the unifor- curred if the consequence of such predispositions was mity of at least some aspects of and hence adaptations an increase in fitness. If human knowledge is so com- to both the past and the present. That is why many, plicated as to depend upon genetics and the embodi- perhaps all. human cognitive processes are geneti- ment of such genetic information in neural networks cally and developmentally constrained, hence predis- that are tuned by appropriate developmental circum- posed. Predispositions are constraints that work in stances, then why be enslaved by philosophical con- the present, not the past. ventions that knowledge only begins with the But what is really interesting in CAMPBELL’s paper is “vicarious processes which short-cut selection by the the question of precision. Leaving aside the difficult life and death of genetic variants”. Do that and we are matter of an appropriate metric by which compari- back before the enlightenment of CAMPBELL’s great sons could be made, knowledge, according to CAMP- papers of the 1960s and 1970s with very little gained. BELL, has a greater precision to it than does adapted More interesting, if less substantive, is another ap- organic structure. But is this correct? There is, of proach to the same point which follows on from discourse a wonderful precision, in both simple (the tun- agreement with that basic distinction that CAMPBELL ing of auditory receptors in some species of moth to now seeks to make between adapted biological struc- just that frequency of ultrasound used by the echolo- tures and knowledge deriving from the workings of cating bats that prey upon them) and complex (fa- vicarious processes that short-cut natural selection. In mously, the eye) forms of adaptive biological this case, he expresses the distinction with the asser- structures. After all, the whole of the argument from tion that “… ‘knowledge’ is more indirectly (but more design, the designer being a supreme Creator, took its precisely) selected than are biological structures” (sec- strength from the exquisite precision of adapted bio- tion 5.1). Given acceptance of a hierarchical scheme logical forms. Reject creationism, as virtually all sci- as the appropriate causal explanation for any instance entists of course do, and what we are left with as an or human cognitively-derived knowledge, it being account of such adapted forms is evolution by selec- more indirect than adapted organic structure must be tion. No one now believes otherwise. Yes, there is true, but it is only trivially so. The structure of my “misfitting” (CAMPBELL’s term in this paper) or satis- hand is only marginally more indirectly an expres- ficing (SIMON, 1969). But given that evolution is a gi- sion of the knowledge of a structure capable of fine ant statistical machine that crunches numbers on a manipulation of small objects, as filtered by way of massive scale across geological time, where there are countless selection and transmission episodes and conserved features of the world, like the properties of coded in genes that partdetermine hand form and light or sound, they are adapted to with wonderful which are expressed in phenotypic form in a specific precision. environment of development. than are the neural On the other hand, the degree of precision of network states that represent “mental” or “psycho- knowledge generated by cognitive processes is owed logical” knowledge of my hand. in large part, as CAMPBELL himself argues, to sensitiv- Furthermore, and closely related in his account, ities to “a much narrower ‘specious present’.” My the distinction that CAMPBELL now seeks to maintain point is that the sensitivities to a narrower present between adapted organic structure and knowledge as derive from those self-same constraints on cognition he now defines it in terms of competence for the “im- whose origins lie in past selection and the constitu- mediate environment of behavior” as opposed to or- tion of human genes in the present. However it is ganic structure where “the environment ‘represented’ described, as computational explosion or frame prob- is always a past one” should lem, the search space available also be challenged. Nature is for cognitive mechanisms to never prescient, and while all Author’s address roam in is so huge that the knowledge constitutes a hope- only way of getting cognition Henry Plotkin, Department of Psychology, ful projection into the future of to work rapidly and adap- University College London, London WC1E an experienced past, in many tively, which is what it does. is 6BT, United Kingdom. vital respects, for example, the by pointing it at the right part e-mail: [email protected] direction of the arrow of time of the search space (PLOTKIN

Evolution and Cognition ❘ 90 ❘ 1997, Vol. 3, No. 1 Knowledge and Adapted Biological Structures

1997). Scramble that part of the genome that codes against running too hard the analogy from biological for the central nervous system and human cognition evolution when knowledge is partly a product of hu- if it survives at all, would be reduced to a slow, sham- man social interaction and influence. In other words, bling and utterly imprecise instrument for generating when culture enters the picture, the analogy with bi- knowledge. The precision is a result of the genetic ological evolution becomes tenuous. Partly this is be- constraints on cognition. cause the rate of transmission of information by way As I see it, the crux of the matter is this. Human of language between individuals is so great, partly be- cognition, and the knowledge that it gives rise to, is cause the transmission routes are much more diverse so dependent upon a causally larger, hierarchical than occurs by way of genetic transmission of infor- structure, that it can only be a concession to a differ- mation, and partly because humans are subject to so- ent constituency, in this case a philosophical one, cial forces that significantly affect the way which leads one to carve off one or several levels of information is received and stored. As a great social that hierarchy and say that these constitute knowl- psychologist, Donald CAMPBELL had a better and edge, whereas the others do not. deeper understanding of the importance of these dif- In one respect, though, I do completely agree with ferences than would the rest of us. CAMPBELL’s final judgement, which is to caution

References Plotkin, H. C./Odling-Smee, F. J. (1979) Learning, Change and Evolution. Advances in the Study of Behaviour 10: 1–41. Elman, J. L./Bates, E. A./Johnson, M. H./Karmiloff-Smith, A./ Riedl, R. (1980) Biologie der Erkenntnis. Parey, Berlin. Parisi, D./Plunkett, K. (1996) Rethinking Innateness: A Simon, H. A. (1969) The Sciences of the Artificial. MIT Press, Connectionist Perspective on Development. MIT Press, Cambridge Mass. Cambridge Mass. Sperber, D./Premack, D./Premack, A. J. (1995) Causal Cogni- Plotkin, H. C. (1997) Evolution in Mind. Allen Lane, London. tion. Clarendon Press, Oxford.

Evolution and Cognition ❘ 91 ❘ 1997, Vol. 3, No. 1 Massimo Stanzione

Campbell, Hayek and Kautsky on Societal Evolution

CAMPBELL’s work, Abstract model, inspired by DAR- In along with the WINIAN natural selection main lines of a new evolu- (i.e. analogous but not CAMPBELL’s and HAYEK’s evolutionary views are con- tionary epistemology, fronted, in order to understand their different stances identical to the genetic there is a substantial effort towards methodological individualism. The many process). to reintroduce the evolu- similarities between their interpretation of social cul- 2. that, during this transi- tionary way of thinking in tural evolution are referred to the ideas of the Austrian tional process, the old- the social sciences (see his School of Economics (particularly VON MISES) and to er ‘natural’ control selected papers in OVER- the ideas of an important figure of the Austrian Marx- mechanisms (resum- MAN 1988 plus many oth- ist School, Karl KAUTSKY. Methodological individual- ing into solidarity) ers such as CAMPBELL ism appears as legitimate but non-unique result of the which assured the fit- 1961, 1963, 1965, 1972, evolutionary premises posited by all these authors. ness of ancient societ- 1975, 1979, 1981, 1981b, Yet CAMPBELL’s latest works reassert the principle that ies had to be and 1995). Almost inde- at the biological level, “groups are real”, and that progressively replaced pendently, the same di- there is a nested hierarchy of selective levels which by different and ‘artifi- rection was taken by the goes from individuals to groups as recently stated by cial’ (i.e., cultural) con- Nobel Prize winner E. SOBER and D. S. WILSON. straints, as ethics, law Friedrich August von and market. HAYEK. The Austrian liberal economist soon became 3. that the peculiar character of this replacement is acquainted with CAMPBELL’s ideas, thanks to a sug- a result of human action, but not of human design. gestion by Konrad LORENZ. HAYEK, in his monumen- Therefore, it will be based more on ‘traditions’ tal trilogy (HAYEK 1973, 1976, 1979) as well as in than on rational projects about the best social and HAYEK (1988), approvingly quotes many of CAMP- economic order. Traditions spontaneously (self- BELL’s papers (i. e., CAMPBELL 1963, 1965, 1974, organizationally) emerge from human practices 1974b, 1975, 1977). CAMPBELL, on the contrary, nei- and effectively mediate between instinct and rea- ther in his intellectual biography (see OVERMAN son, without any precise and justified knowledge 1988, pp1–26) nor in his last contributions discussed on the part of agents. HAYEK’s fundamental statement on the role of tradi- Yet CAMPBELL and HAYEK aren’t really in complete tion as the real power behind the moral, legal and agreement. They disagree: economical development in ancient and modern so- 1. on the evaluation of methodological individual- cieties. Why? I’ll try to give a provisional answer to ism (and, more recently, on the applicability of this apparently simple question, because it has many group selection); different theoretical implications, going far beyond 2. on the possibility of favoring social change the political level. through an experimenting society; In fact, there are many similarities between CAMP- 3. on sociobiology. There are also other topics that BELL’s and HAYEK’s conceptual frameworks. They would be interesting to analyze, such as the role both agree : of self-organizational processes in HAYEK’s view, 1. on the possibility to explain the fundamental but I haven’t sufficient time to discuss them. transition which leads from the first human In HAYEK’s view, the collective forms assumed by a groups of hunters and gatherers to the “extended tradition result only (and automatically) from (egoisti- order” of modern (industrial and capitalist) soci- cal and selfish) actions of single individuals. These eties only by adopting a selective evolutionary actions have intended and unintended effects, which

Evolution and Cognition ❘ 92 ❘ 1997, Vol. 3, No. 1 Campbell, Hayek and Kautsky on Societal Evolution are in turn selected through “invisible hand” mecha- when he withholds full advocacy of the utopian ex- nisms (such as the market’s mechanisms). According perimental society: “In the social sciences”, he says, to a central tenet of HAYEK’s (and POPPER’s) “method- “we are scientific by intention and effort, but not yet ological individualism”, it would be a categorical by achievement” (ibid.). error to affirm the ‘ontological’ reality of collective To prove my point, let me summarize the social (institutional) entities. In other terms, there is no space evolutionary assumption of what CAMPBELL here for the reality of groups as proper actors. Groups can calls “my 1960 dogma”. only be selected a posteriori according to the rules they CAMPBELL’s evolutionary theory of sociocultural have traditionally adopted, but they can never be the development took into consideration (as HAYEK also real subject of real action. Therefore, they cannot fig- does) some typical characteristics that the variation- ure in any causal-explanatory reconstruction of the and-selective-retention process acquires when it is social realm. The ideological consequences of this transferred from the biological to the cultural assumption are well-known: any ‘rationalistic’ effort sphere. These differences are due to the active pres- to project (construct) novel political, and economical ence of substitutive and vicarious levels assumed by forms of (present and future) society would be fatal the three basic functions provided by the selective conceit (see HAYEK 1988). model. Already in CAMPBELL (1958) we find that groups CAMPBELL is quick to maintain that social varia- are asserted to be ‘real’, as opposed to methodologi- tion can, and in fact is, not casual (not completely cal individualism. On this matter he never changed blind). For him, the real advantage of maintaining his mind. Very recently, in readdressing the same the term ‘blind’ in characterizing the sources of so- problem, CAMPBELL wrote that methodological indi- ciocultural variation (which may be an alteration of vidualism “is at the base of most social sciences, in- the social group on the whole or an individual devi- cluding economics, psychology and much of ation internal to a single group) lies in the possibil- sociology. It starts from the dogma that all social pro- ity of explaining the rise of certain social cesses are to be explained by laws of individual be- organizations without resorting to conscious plan- havior, that social systems have no separate ning of the change and only pointing out their ontological reality and that all references to social adaptive value—as HAYEK also says. On the other systems are merely convenient summaries for pat- hand, the study of social change cannot exclude the terns of individual behavior” (CAMPBELL 1995, my analysis of those forces that explicitly intend to con- italic). dition social change itself—as HAYEK admits only Nonetheless, in accordance with the prevalent partially. convictions of many evolutionary biologists, for a Turning to the selective mechanisms, CAMPBELL long time CAMPBELL restrained the validity of group honestly admits that in sociocultural evolution, selection mechanisms to cultural evolution (i.e., to “the potential selective systems are … numerous the group selection of ideologies, social organiza- and … intertwined, and the selective criteria … dif- tional traditions, moral indoctrination, and reli- ficult to specify” (CAMPBELL 1965, p29). However, he gious cosmologies—see CAMPBELL 1973, 1975, 1983, was able to identify six different selective systems, 1991). It seems to me that in so doing, he weakend partially integrable and approximately arranged ac- his views of society to the point where it was insuf- cording to a scale of increasing specificity. ficient to contrast HAYEK’s refusal of social planning. The first one is represented by the selective sur- In fact, CAMPBELL’s constant attention to the prob- vival of complete social organizations (groups, lem of ameliorating social intervention was directly tribes, nations). But this selective mechanism, re- “affected by (his) nation’s experience with the great sorting to the conservation or extinction of the so- society, the war on poverty, and social programs dur- cial organism itself, is evidently inadequate to ing the 1960s and early 1970s” (OVERMAN 1988, explain either the internal existence of conservation p240). Also CAMPBELL’s proposal of an “experiment- systems, or their participation in partial adjust- ing society” (characterized as active, open, honest, ments. The second selective system, applicable to accountable, challengeable, due process, decentral- the multilinear variations in the social field, is con- ized, voluntary, etc.) couldn’t be really adequate. stituted by the selective diffusion of traditions and This broad and ambitious linkage of his general the- by the selective adoption of these traditions on the ory of social experimentation with a specific politi- part of other social groups. cal system had a prevalent ideological character. The other four systems are: selective propagation Some skepticism is evident in CAMPBELL himself of temporal variations; selective imitations of inter-

Evolution and Cognition ❘ 93 ❘ 1997, Vol. 3, No. 1 Massimo Stanzione individual variations; selective promotion to lead- ern society there is genetic competition between ership and educational roles; rational selection. group members, which imposes serious limitations They are all applicable to more restricted units and on the survival value of certain ethical tendencies, even to individuals and are based on vicarious and/ such as altruism, formerly favored by biological se- or substitutive selection mechanisms, such as plea- lection. Furthermore, in the case of human societies sure-pain, memory, imitation, conformity (that is, evolved (through a process more social than biolog- the tendency to imitate even those acts where it is ical) beyond the urbanization stage, genetic compe- not possible to observe their success), and, finally, tition gives rise to egotistical and familial tendencies social knowledge itself. that conflict with the group’s need of self-preserva- In societal evolution, retention substitutive sys- tion. These tendencies must therefore be controlled, tems are, on the whole, less rigid than in biological even if not completely eliminated, by means of a evolution. Yet we can expect “the greatest rigidity, social system of moral rules. These moral rules the greatest demand for conformity, on the part of should be based on the substitution of group altru- those societies with the more elaborate adaptive sys- ism for individual altruism. Moreover, there are eas- tems, particularly when these systems demand re- ier tendencies that an individual might adopt, such straint upon individual hedonistic impulse” as choosing a social group where all its members (CAMPBELL 1965, p34), as illustrated, for example, in show altruistic behavior (to the point of offering a tillage society (with its priesthood castes and reli- greater possibilities of procreation to the altruists), gious-ritualistic beliefs) in comparison with a hunt- where selfishness is self-inhibited and where there ing society. Other novelties, such as the greater is maximum cooperation. amount of time devoted to passing on the cumu- As I already said, these assertions are insufficient lated tradition (i.e. the longer educational process) to extend CAMPBELL’s general rejection of method- and the invention of writing, decrease the chances ological individualism to HAYEK’s version of it. of the potential conflict between freedom to change CAMPBELL’s constant reassertion of the reality of and the selective value of retaining tradition. Nev- group beliefs-forming mechanisms notwithstand- ertheless, the retention of some cultural forms (as ing, his evolutionary scheme of moral and social ethnocentric loyalty and willingness for voluntary development remains very similar to the one by self-sacrifice in warfare), positively selected in the HAYEK. Therefore, HAYEK appears to me completely course of preceding sociocultural evolution, may justified in asserting that our moral tendencies to- prove to be inconvenient when new selective forces ward cooperation and within-group solidarity are (such as the development of larger national units only residual products of the genetic fixation (prob- and new weapons) intervene. ably through the so-called ‘BALDWIN effect’) of some The most important implication of the use of a behavioral schemes which were adaptive only in a selective model for the study of sociocultural evolu- prehistoric (cultural) selective environment. In the tion is the possibility of analyzing convergent mul- new “extended order”, that came about by the se- tilinear evolution. “The blind-variation-and- lective elimination of alternative group traditions, selective-retention model would seem to make plau- these instinctual, pro-social attitudes will therefore sible the multiple independent invention of tools, lack any validity. They suffer from what CAMPBELL rope making, fire, spear, bow-and-arrow, monog- has called “clique selfishness”; therefore they will be amy, the avinculate, and headship in social organi- not capable to produce a universally valid set of zation” (CAMPBELL 1965, p38). Among the results of moral (and legal) rules. convergent evolution of social forms we find con- To put it succinctly: both HAYEK’s liberal theory servation of food surplus, division of labor and ur- of the economic and state order and his individual- banization, all due to their clear adaptive istic ideology can truly be presented as logical con- advantages. sequences of these premises, but (as CAMPBELL’s Many years later, in an ideal confrontation with different orientation suggests) they are not the only WILSON’s sociobiology, he goes on to treat the sub- possible ones. In fact, they are a critical reaction ject of the character and role of ethico-social values against the use that some of the most famous Aus- (CAMPBELL 1975, 1979, 1983). There is a real conflict, trian socialists made of these same evolutionary pre- according to CAMPBELL, between the ethical values mises. Let me briefly comment on this point before inherited through phylogenetic evolution and the turning to CAMPBELL’s new idea about reciprocal al- values of modern social organization. As in all soci- truism and the reality of group-selection at the bio- eties composed of superior vertebrates, in our mod- logical level.

Evolution and Cognition ❘ 94 ❘ 1997, Vol. 3, No. 1 Campbell, Hayek and Kautsky on Societal Evolution

When HAYEK was a young student at the Univer- not be considered analogous to individual sity of Vienna, MENGER’s original theories (exposed development. There is no single social-developmen- in his Grundsätze der Volkswirtschaftslehre 1871) were tal law which could predict the ongoing process in being reformulated and updated by the main figures terms of universal and necessary stages. In it there of the Austrian School of Economics: Eugen BOEHM- will be always “effects that were unintended by the BAWERK, Friedrich WIESER, and Ludwig VON MISES. inventors, because they couldn’t be intended”. To- WIESER was a FABIAN socialist and HAYEK became his gether with the division of labor, which occurs also pupil, because he found VON MISES’ anti-socialist po- in some animal societies, the emergence of language sition too radical. In 1913, MISES attended BOEHM- “develops on a higher level what encephalization BAWERK’s famous seminar on the theory of value. had already started”, that is, the capacity to accumu- The marxist interlocutor in the seminar was Otto late and elaborate information. “Something similar BAUER, with whom von MISES become familiar. In occurs in the economy … with the emergence of an 1906, Otto BAUER had written a short paper (BAUER element which mediates circulation. This element is 1905-1906) which criticizes the socialist (and social- money. Now it is possible to sell without immedi- evolutionary) ethic proposed by his friend Karl ately buying, not unlike the brain permiting excita- KAUTSKY (1906). KAUTSKY’s short book was in fact a tions to act on the organism without immediately reassertion of his former interests in DARWINISM and determining a movement”. And so on, until the positivist biological thought. He resorted to this in- emergence of property, religion, customary law and tellectual armament to oppose Eduard BERNSTEIN’s the fundamental human rights. To sum up: if we attempt to mix a (neo-)KANTIAN ethic and MARXIST identify what KAUTSKY calls “abitude” with HAYEK’s ideology (see BERNSTEIN 1988). The fourth chapter of (and CAMPBELL’s 1960) “tradition”; if we refer his (fer- KAUTSKY’s book is titled DARWINIAN ethic. It contains, tile) idea of interlevels selection to the cultural group almost step by step, the same, identical reconstruction selection, and, of course, if we eliminate any call for here exposed as ‘CAMPBELL’s 1960 dogma’ (which is, in the new “proletarian ethic”, the conceptual frame- turn, very similar to HAYEK 1973, 1979 and 1988). work becomes the same. Quoting DARWIN, ESPINAS, AGASSIZ and FOREL, That can be curious, but isn’t too surprising, if we KAUTSKY accords moral consciousness and moral vir- think to the wide circulation which the same sources tues to the animals themselves, soon adding that the (social DARWINISM and MENGER’s theory) obtained in development of human societies changes both the the Austrian cultural climate. In 1922, HAYEK turned strength and the domain of the application of our an- away from FABIAN socialism and began his collabora- cient moral instincts. Furthermore, he says that hu- tion with VON MISES, of whom he stated “There is no man societies, in order to control better old instincts, single man to whom I owe more intellectually”. The produce new cultural inventions which improve concept of ‘praxeology’, methodological individual- group fitness. These novelties (language, division of ism, evolutionary epistemology and antisocialism labor, money, customary law, property) derive from were the main gifts of this heritage: a big research human intentions, but are not the result of human program which HAYEK’s extraordinary mind contrib- conscious planning. They are always provisional and uted to develop in many details. Nonetheless, his imperfect. “Moral law…”, KAUTSKY says, “isn’t a conclusions aren’t completely indisputable. product of wisdom (which here means “conscious Let me quote from CAMPBELL/HEYLIGHEN (1995): knowledge“) and does not generate wisdom”. “It is “Libertarian theorists and laissez-faire economists an error to believe that man could and must follow recommend using market mechanisms as the overall all his instincts without limitations … because they control, while avoiding all centralized interventions often limit themselves reciprocally”. In the fifth except for the protection of private property and in- chapter, second paragraph, KAUTSKY describes in herited wealth. Yet, we believe that market mecha- evolutionary terms “the organism of human soci- nisms cannot operate effectively without the ety” as a new and distinct level of evolution interact- contribution of the other social control mechanisms ing with the previous ones … In particular, legal control constituted by individuals, … is necessary to ensure that Author’s address kins and small groups. Be- positive feedback loops do not cause human society is differ- get out of hand, and that in- Massimo Stanzione, Universita di Salerno, ent both from individual vestments are made in areas Via Ponte Don Melillo, 84048 Fisciano organism and from animal so- (SA), Italy. where the lack of local or short cieties, its development can- term benefits precludes pri-

Evolution and Cognition ❘ 95 ❘ 1997, Vol. 3, No. 1 Massimo Stanzione

vate investment.” How does the new CAMPBELLIAN 1964, MAYNARD-SMITH 1964) … [and] evolutionary theory of these social control mechanisms differ game theory (AXELROD/HAMILTON 1981, MAYNARD- from his old one? In his first essays CAMPBELL SMITH 1982, TRIVERS 1971, WILLIAMS 1966), which adopted the orthodox way of thinking that: explained how altruism could evolve among non-rel- 1. individual selection always dominates group se- atives” (see SOBER/WILSON 1994). According to WILL- lection at the biological level; IAMS and DAWKINS, however, even sexually 2. this notwithstanding, groups are real and self-sac- reproducing organisms do not qualify as units of rificial altruism in the service of human social selection because they, like groups, are too ephem- groups genuinely exists; eral. Therefore DAWKINS (1976) proposed a com- 3. therefore self-sacrificial altruism has to be seen as pletely new concept (except for KAUTSKY), i.e., a result of cultural group selection of ideologies, “vehicles of selection”, which is very similar to social-organizational traditions, moral indoctri- HULL’s concept of “interactors” (see HULL 1980). nation, and religious cosmologies. According to the metaphor of vehicle selection, we A “simple point of view” that CAMPBELL, in the last can say that genes in an individual are like “the part of his life, seemed willing to modify, retaining its members of a rowing crew competing with other relevance for secondary groups, but reinterpreting crews in a race. The only way to win the race is to the role of group selection for social control mecha- cooperate fully with the other crew members”. Yet, if nisms within primary groups. Why? Because, at the individuals can be vehicles of selection, what about biological level, SOBER/WILSON (1994) proposed that the groups? group selection and individual selection could be concurrent, “producing an ambivalence on group B. The plausibility of hypothesizing a nested hi- preservation vs. individual preservation dimension”. erarchy of different levels upon which the selec- To understand CAMPBELL’s new perspective, we have tive process could act. “Taking vehicles seriously to remind ourselves that two classical problems are requires more than acknowledging a few cases of here clearly intertwined: group selection, however; it demands a restructuring of the entire edifice … There is one theory of natural A. The individuation of the real units of selection. selection operating on a nested hierarchy of units, of SOBER/WILSON (1994) reacts against WILLIAMS’ (and which, inclusive fitness and game theory are special DAWKINS’) theory according to which the gene is the cases… Adaptation at any level of the biological hier- fundamental unit of selection because it is a replica- archy requires a process of natural selection at that tor and higher levels of selection are only theoreti- level.” (SOBER/WILSON 1994, italics mine). HEGEL is cally possible but unlikely to occur in nature. They vindicated. In a nested hierarchy of units of selection point out that in WILLIAMS’ scheme, individuals a dialectic process is going on. A few pages later SOBER aren’t real units of selection either. On the contrary, and WILSON say: “Since group selection is seldom the both individuals and groups are environments of only force operating on a trait, the hierarchical the- genes. SOBER and WILSON realize that “WILLIAMS’ case ory explains both the reality of groups that CAMPBELL against group selection was strengthened by two emphasizes, and the genuinely individualistic side of other theories in evolutionary biology: … inclusive human nature that is also an essential part of his fitness theory (also called kin selection; HAMILTON thinking”.

References York, pp. 101–142. Campbell, D. T. (1963) Social attitudes and other acquired be- Axelrod, R./Hamilton W. D. (1981) The evolution of coopera- havioral dispositions. In Koch, S. (ed) Psychology: a study tion. Science, 211: 1390–1396. of a science, vol. 6: Investigations of man as socius. Bauer, O. (1905-1906) Marxismus und Ethik. Die Neue Zeit, MacGraw-Hill, New York, pp. 94–172. 24 (2), pp. 485–499. Campbell, D. T. (1965) Variation and selective retention in Bernstein, E. (1899) Die Voraussetzung des Socialismus und sociocultural evolution. In: Barringer, H. R./Blankstein, G. die Aufgaben der Sozialdemokratie. Stuttgard. I./Mack, R. W. (eds) Social change in developing areas: A Campbell, D. T. (1958) Common fate, similarity and other in- reinterpretation of evolutionary theory. Shenkman, Cam- dices of the status of aggregates of persons as social enti- bridge, Mass., pp. 19–48. ties. Behavioral science 3 (1):14–25. Campbell, D. T. (1972) On the genetics of altruism and the Campbell, D. T. (1961) Conformity in psychology’s theories counter-hedonic components in human culture. Journal of acquired behavioral dispositions. In: Berg, I.A./Bass, Soc. Issues 28: 21–37. B.M. (eds) Conformity and deviation, Harper & Row, New Campbell, D. T. (1974) Evolutionary epistemology. In:

Evolution and Cognition ❘ 96 ❘ 1997, Vol. 3, No. 1 Campbell, Hayek and Kautsky on Societal Evolution

Schilpp, P. (ed) The philosophy of Karl Popper, vol. 1. Dempsey, G.T. (1996) Hayek’s evolutionary epistemology, Open Court, LaSalle, Ill., pp. 413–463. artificial intelligence, and the question of free will. Evolu- Campbell, D. T. (1974b) “Downward causation” in hierarchi- tion and Cognition 2: 139–150. cally organized biological systems. In: Ayala, F. J., Hamilton, W. D. (1964) The genetical evolution ofo social be- Dobzhansky, T. (eds) Studies in the philosophy of biology, havior. (I and II). Journal of theoretical biology, 7: 1–52. Macmillan, London, pp. 179–186. Hayek, F. A. von (1952) The sensory order. The University of Campbell, D. T. (1975) On the conflict between biological Chicago Press, Chicago & London. and social evolution. American Psychologist, 30: 1103– Hayek, F. A. von (1973) Law, legislation and liberty, vol. 1: 1026. Rules and order. Routledge & Kegan Paul, London. Campbell, D. T. (1977) Descriptive epistemology: psycholog- Hayek, F. A. von (1976) Law, legislation and liberty, vol. 2: ical, sociological and evolutionary. William James Lec- The mirage of social justice. Routledge & Kegan Paul, Lon- tures, Harvard University, unpublished. Today in don. Overman, E.S. (ed) Methodology and epistemology for so- Hayek, F. A. von (1979) Law, legislation and liberty, vol. 3: cial sciences. Selected papers of DonalD. T. Campbell. The The political order of a free people. Routledge & Kegan University of Chicago Press, Chicago & London, 1988, pp. Paul, London. 435–486. Hayek, F. A. von (1988) The fatal conceit. Routledge & Kegan Campbell, D. T. (1979) Comments on the sociobiology of Paul, London. ethics and moralizing. Behavioral science 24: 37–45. Hull, D. (1980) Individuality and selection. Annual review of Campbell, D. T. (1981) Levels of organization, selection, and ecology and systematics, 11: 311–332. information storage in biological and social evolution. The Kautsky, K. (1906) Ethik und materialistische Geschichtsauf- behavioral and brain sciences 4: 236–237. fassung. Berlin. Campbell, D. T. (1981b) Variation and selective retention Maynard-Smith, J. (1964) Group selection and kin selelction. theories of sociocultural evolution. Current Anthropology Nature: 145–1146. 22: 603–608. Maynard-Smith, J. (1982) Evolution and the theoy of games. Campbell, D. T. (1983) The two distinct routes beyond kin se- Cambridge University Press, Cambridge. lection to ultrasociality: implications for the humanities Overman, E. S. (ed) (1988) Methodology and epistemology and social sciences. In Bridgeman, D.L. (ed) The nature of for social sciences. Selected papers of DonalD. T. Camp- pro-social development: theories and strategies. Academic bell. The University of Chicago Press, Chicago & London. Press, New York, pp. 11–41. Sober, E./Wilson, D. S. (1994) Reintroducing group selection Campbell, D. T./Heylighen, F. (1995) Selection of organiza- to the human behavioural sciences. Behavioral and brain tion at the social level: obstacles and facilitators of meta- sciences 17 (4): 585–608. system transition. In World futures: the journal of general Trivers, R. L. (1971) The evolution of reciprocal altruism. evolution. Heylighen, F/Joslyn, C./Turchin, V. (eds) Spe- Quarterly review of biology, 46: 35–57. cial issue on: The quantum of evolution: toward a theory Williams, G. C. (1966) Adaptation and natural selection: a cri- of metasystem transition. tique of some current evolutionary thought. Princeton Dawkins, R. (1976) The selfish gene. Oxford University Press, University Press, Princeton. Oxford.

Evolution and Cognition ❘ 97 ❘ 1997, Vol. 3, No. 1 Franz M. Wuketits

Four (or Five?) Types of Evolutionary Epistemology

Donald Campbell and the Constructivist Approach

he meaning of Abstract man’s status as a product T CAMPBELL’s work for of biological and social the evolutionary-episte- This contribution aims at giving a brief discussion of evolution”. However, mology enterprise is CAMPBELL’s attempt to systematize the different ap- there is not simply the undisputed. In his semi- proaches to evolutionary epistemology. It is argued theory of evolution; we nal paper (CAMPBELL that, as CAMPBELL repeatedly stated, ‘evolutionary have to take into account 1974) he systematically epistemology’ covers various conceptions and com- different explanations of outlined—and comprises different views that hold in common only the the processes and mecha- mented on—theories that assumption that all types of cognition and knowledge nisms of evolutionary deserve the name ‘evolu- result from evolution and that, therefore, the study of development and thus to tionary epistemology’ evolution is relevant to an understanding of cognitive distinguish between dif- and are rooted in the evo- phenomena of whatever kind. While CAMPBELL dis- ferent theories (WUKETITS lutionism of the last cen- tinguished between four types of evolutionary episte- 1988). Hence, it should tury. To be sure, this is mology, I think that there are more and that not come as a surprise not to say that CAMPBELL CAMPBELL himself, in his later work, showed some that evolutionary episte- was just compiling the sympathy to the so-called constructivist approach. mology is not a mono- work of other scholars, lith, but a wide field of naturalists and philosophers, who had contributed ramifying theories (WUKETITS 1990), so that we are to the development of evolutionary epistemology. confronted with different types of this epistemol- He himself contributed so much to this develop- ogy. ment that it would be risky to evaluate his efforts in In his posthumous paper, CAMPBELL speaks of four a short paper. However, the aim of the present paper types of evolutionary epistemology and character- is not to give a comprehensive survey of his izes his own model which is a selectionist one. In- thoughts concerning evolutionary approaches to deed, from the beginning on he was engaged in the cognition and knowledge. I shall concentrate on his natural-selection program of evolutionary episte- attempts to systematize these approaches and try to mology. As he once stated: “I began my career as an mark his own place in the evolutionary-epistemol- ‘evolutionary epistemologist’ by applying selection ogy movement. theory to trial-and-error learning, incorporating gestalt models of problem solving…, to visual percep- Theories of Evolution and Evolutionary tion…, and to creative thought” (CAMPBELL 1990, p7). For natural-selection theories have predomi- Epistemology nated evolutionary thinking since DARWIN, one If one takes the theory of evolution really seriously, should expect their impact on evolutionary reflec- then one will come, sooner or later, to the conclu- tions with regard to cognition and knowledge. How- sion that cognitive phenomena too are results of ever, this type of theories ha.s always been evolutionary processes. This is trivial. CAMPBELL compatible with—and has in fact supported—the (1974, p413) was of course right that “an evolution- adaptationist program of evolutionary epi.stemology ary epistemology would be at minimum an episte- (see e.g., LORENZ 1977) according to which cognitive mology taking cognizance of and compatible with structures of organisms are to be understood as ad-

Evolution and Cognition ❘ 98 ❘ 1997, Vol. 3, No. 1 Four (or Five?) Types of Evolutionary Epistemology

aptations to external reality, so that a correspon- ‘true picture’ of what is out there (cf. OESER 1987; dence between the ‘inner’ and the ‘outer’ world of RUSE 1986; WUKETITS 1995). A stricter constructivist any organism can be supposed (see e.g., WUKETITS approach to (evolutionary) epistemology has been 1984 for discussion). formulated by DIETTRICH (1994, p57) who states that General-selection models of cognition—that are “within the constructivist evolutionary epistemol- not necessarily always in tune with strict adaptation- ogy … the regularities which we condense to the ism—have also been proposed to explain the analo- laws of nature are seen as the invariants of phyloge- gies between organic evolution and the evolution of netically formed cognitive operators”. scientific ideas or what HULL (1988) calls the concep- All these interesting extensions of evolutionary tual development of science. CAMPBELL is entirely right epistemology can not be detailed here, for my point that elaborating an appropriate evolutionary theory of reference is CAMPBELL. He refers to his earlier of science requires a broad selection theory in which model (CAMPBELL 1960) as to the ‘1960 dogma’ and both, organic and conceptual evolution, are just is presenting, in the recent paper, a modified view. ‘clusters of exemplars’ (see also CAMPBELL 1988), and He wishes “to identify ‘knowledge’ … with those vi- that simply applying the organic-evolution model to carious processes which short-cut selection by the life the development of science does not suffice. Any- and death of genetic variants”. But already in one of way, his own reflections on these problems and his his previous papers, published jointly with PALLER, proposed solutions are best examples for a refined we can find the following statement: “In the evolu- type of naturalism and have helped us ‘taking the tionary epistemology movement, there is too much naturalistic turn’ (CALLEBAUT 1993). uncritical passing-the-buck to an omnipotent ‘Dear Old Mother Natural Selection’. But clair-voyant vi- The Constructivist Approach sion is mechanically impossible and could not have been produced by evolution. Instead, evolution has There is one type of evolutionary epistemology that opportunistically exploited marvelously effective is not mentioned by CAMPBELL in the present paper, proxy-variables of limited appropriateness” (CAMP- although he himself has contributed much to its BELL/PALLER 1989, p236). Unpacking this passage and development: the constructivist approach. This is, CAMPBELL’s posthumously published manuscript, we in a nutshell, the view that organisms do not simply can formulate the basic assumptions of a construc- ‘represent’ or ‘reconstruct’ their environment, but tivist approach as follows: also—the radical constructivist says: only—con- 1. The cognitive capacities of any organism are lim- struct and interpret it. CAMPBELL (in CALLEBAUT ited; 1993, p298) takes as an example the salamander’s 2. these limitations are mainly the result of organis- leg: “When it regenerates when broken off, does it mic, functional constraints; regenerate until it reaches the ground? No! It regen- 3. as active systems, organisms explore those aspects erates until an internal vicarious monitor for leg of ‘reality’ that are relevant to their survival; length is completed.” Clearly, there must be some- 4. the act of perception always includes an interpre- thing like ‘internal selection’ for living systems are tation of the perceived object; not marionettes hanging on the strings of their 5. this interpretation is based on the organism s own environment. experience that is constrained by the evolution- Those who have recognized the importance of ary pathways of its species. this kind of selection—with organismic constraints This approach to evolutionary epistemology is, of and internal building blocks in the organization of course, not anti-adaptationist, but it is non-adapta- living beings—have pleaded, then, for a constructiv- tionist. It can meet one of the standard counterargu- ist extension of evolutionary epistemology though ments which reads as follows: “The fundamental their starting point was adaptationism (RIEDL error of evolutionary epistemologies as they now 1995). Also, it has been argued that the correspon- exist is their failure to understand how much of dence theory of cognition/ what is ‘out there’ is the prod- knowledge has to be replaced uct of what is ‘in here’. Organ- Author’s address by a coherence theory, this is to ism and environment are say a view according to which codetermined” (LEWONTIN Franz M. Wuketits, Institut für Wissen- getting to know reality is 1982, p169). Yes, they are, schaftstheorie, Universität Wien, Sensen- strongly connected to success gasse 8, A-1090 Wien (Austria) and the constructivist version in life—and not so much to a of evolutionary epistemology

Evolution and Cognition ❘ 99 ❘ 1997, Vol. 3, No. 1 Franz M. Wuketits is in fact at minimum an epistemology taking cog- As a personal retrospective history of evolutionary nizance of and compatible with any organism’s sta- epistemology it shows that CAMPBELL himself was tus as an active system that actively interacts with aware of some shortcomings of the traditional ver- what is ‘out there’. sion of this epistemology, i. e., the adaptationist approach, and that, after all, he developed another Conclusion type of the evolutionary approach to understanding cognition and knowledge. This is a remarkable CAMPBELL’s evolutionary epistemology has a forty- and essential development in a field that some year history and underwent some changes and already tend to consider obsolete. One question, improvements within this period of time. His however, remains: Will—and can—there be a uni- present paper is the best example of these changes. fied evolutionary epistemology?

References 4: 57–74. Hull, D. L. (1988) Science as a Process. The University of Chi- Callebaut, W. (1993) Taking the Naturalistic Turn. University cago Press, Chicago. of Chicago Press, Chicago. Lewontin, R. C. (1982) Organism and Environment. In: H. C. Campbell, D. T. (1960b) Blind Variation and Selective Reten- Plotkin (ed) Learning, Development, and Culture. Wiley, tion in Creative Thought as in Other Knowledge Processes. Chichester. pp. 151–170. Psych. Rev. 67: 380–400. Lorenz, K. (1977) Behind the Mirror. Methuen & Co., Lon- Campbell, D. T. (1974) Evolutionary Epistemology. In: don. Schilpp, P. A. (Ed.) The Philosophy of Karl Popper, vol. 1. Oeser, E. (1987) Psychozoikum: Evolution und Mechanismus Open Court, LaSalle, pp. 413–463. der menschlichen Erkenntnisfähigkeit. Parey, Berlin. Campbell, D. T. (1988) A General ‘Selection Theory’, as Im- Riedl, R. (1995) Deficiencies of Adaptation in Human Reason; plemented in Biological Evolution and in Social Belief- a Constructivist Extension of Evolutionary Epistemology. Transmission-with Modification in Science. Biol. & Philos. Evol. & Cogn. 1: 27–37. 3: 171–177. Ruse, M. (1986) Taking Darwin Seriously. Blackwe]l, Oxford. Campbell, D. T. (1990) Epistemological Roles for Selection Wuketits, F. M. (1984) (ed) Concepts and Approaches in Evo- Theory. In: Rescher, N. (Ed.) Evolution, Cognition, and Re- lutionary Epistemology. Reidel, Dordrecht. alism. University Press of America, Lanham, pp. 1–19. Wuketits, F. M. (1988) Evolutionstheorien. Wissenschaftli- Campbell, D. T./Paller, B. T. (1989) Extending Evolutionary che Buchgesellschaft, Darmstadt. Epistemology to ‘Justifying’ Scientific Beliefs (A Sociologi- Wuketits, F. M. (1990) Evolutionary Epistemology and Its Im- cal Rapprochement with a Fallibilist Perceptual Founda- plications for Humankind. State University of New York tionalism). In: Hahlweg, K. /Hooker, C. A. (eds) Issues in Press, Albany. Evolutionary Epistemology. State University of New York Wuketits, F. M. (1995) A Comment on Some Recent Argu- Press, Albany, pp. 231–257. ments in Evolutionary Epistemology—and Some Counter- Diettrich, 0. (1994) Kognitive und kommunikative Entwick- arguments. Biol. & Philos. 10: 357–363. lung in realitätsfreier Darstellung. Kognitionswissenschaft

Evolution and Cognition ❘ 100 ❘ 1997, Vol. 3, No. 1