Kitcher\'s Compromise

Pergamon

Stud. Hist. Phil. Sci., Vol. 28, No. 2, pp. 319-338, 1997 0 1997 Elsevier Science Ltd All rights reserved. Printed in Great Britain 0039-3681/97 $17.00+0.00

Kitcher’s Compromise: A Critical Examination of the Compromise Model of Scientific Closure, and its Implications for the Relationship Between History and Philosophy of Science Timothy Shanahan* In The Advancement of Science (1993) Philip Kitcher develops what he calls the ‘Compromise Model’ of the closure of scientific debates. The model is designed to acknowledge significant elements from ‘Rationalist’ and ‘Antirationalist’ accounts of science, without succumbing to the one-sidedness of either. As part of an ambitious naturalistic account of scientific progress, Kitcher’s model succeeds to the extent that transitions in the history of science satisfy its several conditions. 1 critically evaluate the Compromise Model by identifying its crucial assumptions and by attempting to apply the model to a major transition in the history of biology: the rejection of ‘naive group selectionism’ in the 1960s. I argue that the weaknesses and limitations of Kitcher’s model exemplify general problems facing philosophical models of scientific change, and that recognition of these problems supports a more modest vision of the relationship between historical and philosophical accounts of science. 0 1997 Elsevier Science Ltd. 1. Introduction 1.1. Lakatos’

Dictum

In the famous opening lines of a famous paper, Imre Lakatos declared: ‘Philosophy of science without history of science is empty; history of science without philosophy of science is blind’.’ He articulated, in a pithy and memorable fashion, an extremely appealing vision of the relationship between historical and philosophical inquiries into science. His own detailed account of this relationship is well known. According to Lakatos, philosophy of science provides ‘normative methodologies’ in terms of which the historian of science *Department of Philosophy, Loyola Marymount University, U.S.A. Received 20 August 1995; in revisedform 25 January 1996.

Los Angeles,

‘I. Lakatos, ‘History of Science and its Rational Reconstructions’, (eds), Boston Studies in the Philosophy of Science VIII (1971), 91-136:

PII: SOO39-3681(%)00023-4 319

CA 90045-2699,

in R. Buck and R. Cohen p. 91.

Studies in History and Philosophy of Science

320

constructs a ‘rational explanation of the growth of objective knowledge’.2 Historical inquiry into science, in turn, provides the data against which philosophical conceptions of science are to be adjudicated.’ Lakatos’ own work aspires to exemplify this mutual reliance between the history of science and the philosophy of science.4 The overall tone of his writings is optimistic: history of science and philosophy of science are mutually necessary--each contributes essential guidance or data for the other. An approach that properly utilizes both, therefore, promises to provide the most satisfying general account of science of which we are capable. 1.2 Troubles fix

Optimism

Despite its attractions, Lakatos’ optimistic conception of the relationship between history of science and philosophy of science has not met with universal acceptance. Some have questioned the value of history of science (as opposed to science simpliciter) for the philosophy of science.5 Others have attacked Lakatos’ attempt to demonstrate the correctness of his particular model of the growth of scientific knowledge. His practice of ‘rationally reconstructing’ important episodes from the history of science has come under special criticism.6 Although these critiques focus on Lakatos’ work, what is at issue, ultimately, is the question of the relationship between historical and philosophical accounts of science. 1.3. Aim and Plan of this Paper

This paper explores further the issues raised by these critiques. Rather than (re-)examining the work of Lakatos, however, I wish to consider what is perhaps the most ambitious recent attempt to use history of science to validate a philosophical model of science. In The Advancement of Science (1993) Philip Kitcher develops what he calls the ‘Compromise Model’ of the closure of scientific debates. The ‘compromise’ in question is between ‘Rationalist’ and ‘Antirationalist’ accounts of science. ‘Rationalists’ view scientific agents as moved solely by epistemic goals, and attribute the closure of scientific *Lakatos, op. cit., note I, p. 91. ‘For his fullest account, including - applications to some examples, see I. Lakatos, ‘Falsification __ and the Methodology of Scientific Research Programmes’, in I.-Lakatos and A. Musgrave (eds), Criticism and the Growth of Knowledge (Cambridge University Press, 1970). nn. 91-196. ‘See, for example, I. Lakatos and E. Zahar, ‘Why Did- Copernicus” Research Programme Supersede Ptolemy’s?, in R. Westman (ed.), The Copernican Achievement (Berkeley/Los Angeles: University of California Press, 1975) pp. 354383. ‘See, for example, R. Giere, ‘History and Philosophy of Science: Intimate Relationship or Marriage of Convenience?‘, British Journal for the Philosophy of Science 24 (1973), 282-297. ‘The best example of this is N. Thomason, ‘Could Lakatos, Even with Zahar’s Criterion for Novel Fact, Evaluate the Copernican Research Programme?‘, British Journalfor the Philosophy of Science 43 (I 992), 161-200. Thomason argues convincingly that Lakatos and Zahar’s analysis of the success of Copernican astronomy in the 17th century must do significant violence to the historical material in order to succeed.

Etcher’s

Compromise

321

controversies to the triumph of the superior cognitive processes employed by ‘the winners’. ‘Antirationalists’ view scientific agents as significantly influenced by non-epistemic goals, and attribute the closure of scientific controversies to the exclusion from the community of one group (‘the losers’) by a more powerful rival. Kitcher’s Compromise Model is explicitly designed to capture significant aspects of each of these views without succumbing to the one-sidedness of either. It consists of five related theses:

(Cl) The community decision is reached when sufficiently many sufficiently powerful subgroups within the community have arrived at decisions (possibly independent, possibly coordinated) to modify their practices in a particular way. (C2) Scientists are typically moved by non-epistemic as well as epistemic goals. (C3) There is significant cognitive variation within scientific communities, in terms of individual practices, underlying propensities, and exposure to stimuli. (C4) During early phases of scientific debate, the processes undergone by the ultimate victors are (usually) no more well designed for promoting cognitive progress than those undergone by the ultimate losers. (CV Scientific debates are closed when, as a result of conversations among peers and encounters with nature that are partially produced by early decisions to modify individual practices, there emerges in the community a widely available argument, encapsulating a process for modifying practice which... is markedly superior in promoting cognitive progress than other processes undergone by protagonists in the debate; power accrues to the victorious group principally in virtue of the integration of this process into the thinking of members of the community and recognition of its virtues.’ Because the Compromise Model is presented as part of a naturalistic account of scientific progress, it succeeds to the extent that major transitions in the history of science satisfy its conditions. 8 Kitcher illustrates his model by applying it to three episodes in the history of science: the triumph of Darwin’s ‘P. Kitcher, The Advancement of Science: Science Without Legend, Objectivity Without Illusions (Oxford University Press, 1993) p. 201. ‘More precisely, the Compromise Model succeeds to the extent that it captures the salient features of major transitions in the history of science better than alternative models do. In his discussion of historical cases, Kitcher is concerned to show how each case can be fitted to the theses of his Compromise Model. Absent is a direct, point-by-point comparison of the relative merits of the three models. But little is gained from showing that a given philosophical model is able to plausibly represent selected details from a variety of historical episodes. The underdetermination thesis, so celebrated in philosophy of science, must be applied reflexively to its own practices. Any number of different philosophical (or sociological, etc.) models may be able to accommodate the details of a range of historical episodes. In order to derive interesting conclusions, however, these models must be critically compared and their relative merits assessed. This is to demand no more of metascientific models than is demanded within science itself. For the specific claims of the Rationalist and Antirationalist models, see Kitcher op cit., note 7, pp. 196200.

322

Studies in History and Philosophy of Science

theory of evolution in the period 185967; the emergence of a consensus concerning Copernican astronomy between 1543 and 1633; and the resolution of ‘the Great Devonian Controversy’ between 1834 and 1848. In each case he tries to show that the conditions of the Compromise Model are satisfied. Kitcher’s project stands squarely in the tradition of Kuhn, Lakatos, Laudan and Shapere.g Like his predecessors, Kitcher uses historical case-studies of science in order to validate a philosophical model of scientific change. Like them he intends his model of scientific change to be generally applicable to cases beyond those used to illustrate the model. Like Lakatos, especially, Kitcher’s account is designed to show that scientific change is a rational process that results in objective knowledge. By offering a thoroughly naturalized account of scientific consensus formation, however, in terms of the acceptance of cognitively progressive psychological processes, Kitcher intends to avoid the difficulties facing Lakatos’ explicitly rationalistic account. My aim in what follows is to show that Kitcher’s model repeats, rather than evades, these difficulties. The fundamental problem facing his account is a failure to take the (messy) details of history seriously enough, with the result that what was intended as a naturalistic account of the dynamics of actual historical episodes rapidly assumes the form of Lakatosian rational reconstruction. Rather than re-analyze the cases that Kitcher discusses, I will evaluate his model by applying it to a major transition in the recent history of biology, one that is well known, at least in outline, to most biologists, historians of biology and philosophers of biology. 10 Prior to the 1960s explanations of biological phenomena appealing to the ‘good of the group’ or the ‘preservation of the species’ were common in the scientific literature. By the end of this decade such explanations had become rare. A number of factors intersected to bring about this development, but one especially salient event was the controversy surrounding the claims of V. C. Wynne-Edwards. Critical reactions to his claims marked the end of ‘naive’ group selectionist theorizing, and paved the way for the ‘gene-centered’ approach common to much current explanatory practice in biology. In the next section I provide a very brief account of this development, an account that is comparable in degree of historical detail to the historical case-studies presented in The Advancement of Science. I then attempt to apply Kitcher’s model to this account, drawing particular attention to the kinds of requirements that must be satisfied in order to do so. I argue that in order to ‘The tradition in question is represented in the following texts: T. Kuhn, The Structure qf Scientzjic Revolutions,-second edn (University of Chicago Press, 1970); I. Lakatos, op. cit., note j; L. Laudan. Proaress and Ifs Problems (Berkelev: Universitv of California Press. 1977): and D. Shapere, Reasonand the Growth of Knowledge (Dordrecht: Reidel, 1984). ” “See for example R. N. Brandon and R. M. Burian, Genes, Organisms, and Populalions (Cambridge, MA: MiT Press, 1984); M. Ridley, Evoluzion (Cambridge: Blackwell, 1993); Elliott Sober, The Nature of Selection (Cambridge, MA: MIT Press, 1984).

Kiteher’s Compromise

323

successfully instantiate the Compromise Model one must either have a level of access to historical, psychological and social processes that is likely to remain permanently beyond our cognitive grasp, or else one must be willing to forego a genuinely naturalistic account of scientific change in favor of a rational reconstruction of historical episodes. Fitting historical materials into the model thus requires a compromise of historiographical standards no less radical than that required by Lakatos’ model. The sense in which such philosophical models enhance our understanding of the actual development of science, therefore, remains unclear. The implications of this conclusion for the relationship between historical and philosophical accounts of science are discussed in the last section. I suggest that whereas philosophy of science can be useful in discerning patterns of reasoning underlying the publicly-available arguments in particular historical episodes, and may, to some extent, help us to see the broader implications of highly focused historical investigations, history of science is philosophically useful primarily because, in addition to providing content for philosophical reflection, it reveals the severe limitations of idealized philosophical models of science and thereby points us toward more realistic accounts. 2. Wynne-Edwards’ Theory of Population Regulation and Its Reception 2.1. The Theory of Animal Dispersion A basic fact about nature, the importance of which had been recognized since Darwin, is that although even relatively ‘slow’ breeding animals are physiologically capable of increasing their numbers at a stupendous rate, under normal circumstances their populations remain remarkably stable. Why? In 1962 V. C. Wynne-Edwards, Regius Professor of Natural History at the University of Aberdeen, published Animal Dispersion in Relation to Social Behavi0ur.l’ In over 600 smoothly flowing pages he amassed a large body of evidence intended to show that animals actively regulate their population densities. Why such population regulation is necessary is explained early on. Wynne-Edwards writes that ‘It is easy to appreciate that if each species maintains an optimum population-density on its own account, not only will it be providing the most favourable conditions for its own survival, but it will automatically offer the best possible living to species higher up the chain that depend on it for food’. ‘2 Such considerations led Wynne-Edwards to conclude that ‘it must be highly advantageous to survival, and thus strongly favoured by selection, for animal species (1) to control their own population-densities, and (2) to keep them as near as possible to the optimum level for each habitat they occupy’.i3 ’ 'V C Wynne-Edwards, Animal Dispersion in Relation to Social Behaviour (Edinburgh: Boyd; 1462). “Wynne-Edwards, op. cit., note 11, p. 9. ‘3Wynne-Edwards, op. cit., note 11, p. 9.

Oliver &

324

Studies in History and Philosophy of Science

Wynne-Edwards realized that natural selection operating at the level of individual organisms would not be effective in bringing about the kinds of social adaptations central to his theory, adaptations that benefit the group by subordinating the reproductive interests of individuals. Selection operating on individuals would favour organisms that maximize individual reproductive success without regard for group welfare, leading to overexploitation of the habitat and population crash. Because such occurrences are rare, another evolutionary force must be operative. The mechanism for promoting the evolution of density regulation is identified as ‘group selection’. Groups in which social conventions are not honoured suffer from overcrowding, overexploitation of resources, and eventually population crash and extinction. Groups in which individual reproductive rates are sensitive to overall population density will tend to persist longer, and may colonize areas left vacant by groups not exhibiting such reproductive restraint. Wynne-Edwards thought such group selection to be pervasive in nature, and indeed to lie at the base of all social behaviour. 2.2. Critical Reactions to Animal Dispersion In the Preface to Animal Dispersion Wynne-Edwards remarks that ‘It has turned out to be an agreeable and characteristic feature of the theory not to keep butting against widely held, pre-existing generalisations, but to lead instead into relatively undisturbed ground’.14 This was a reasonable assessment for someone situated in the explanatory tradition that then dominated ecology. Wynne-Edwards’ emphasis on the subordination of individual advantage to group benefit reflected a long-standing and widely accepted explanatory tradition in ecology. Biologists in this tradition routinely assumed that selection operates to insure the well-being of biological entities more inclusive than individual organisms. A perusal of the biological literature of the 1950s and early 1960s confirms that this approach was not uncommon and certainly not considered theoretically controversial, at least within ecology.15 Wynne-Edwards’ complacency about the cogency of his approach reflects the relative insulation of ecology from other biological disciplines. It was precisely because it conflicted with ‘widely held, pre-existing generalizations’ in the tradition of mathematical, genetically-centred population biology that his book had the impact it did. Whereas ecologists tended to talk about the adaptations associated with whole organisms, groups and even multi-species ecological “‘Wynne-Edwards, op. cit., note 11, p. v. “Prominent examples include W. C. Allee et al., Principles of Animal Ecology (Philadelphia: W. B. Saunders Co., 1949); M. J. Dunbar, ‘The Evolution of Stability in Marine Environments: Natural Selection at the Level of the Ecosystem’, The American Naturalist 94 (1960), 129-136; A. E. Emerson, ‘The Evolution of Adaptation in Population Systems’, in S. Tax (ed.), The Evolution of Life (University of Chicago Press, 1960), pp. 307-348. On ecology at mid-century, and at the University of Chicago in particular, see G. Mitman, The Sfnte of Nature (University of Chicago Press, 1992).

Kitcher’s Compromise

325

communities, quantitatively-oriented population biologists focussed on the precise numerical values of selection coefficients in single-locus genetic models. For such biologists, selection acts at the level of individual organisms, no higher, and is best represented by models depicting changes in gene frequencies in idealized (e.g. infinite, banmictic) populations.16 Not surprisingly, the strongest criticisms of Wynne-Edwards’ theory came from biologists influenced by this tradition. The most important critiques, in terms of their lasting impact on evolutionary biology, were those by John Maynard Smith, David Lack and George C. Williams. A basic problem with Wynne-Edwards’ scheme, Maynard Smith argued, is that it postulates groups consisting entirely of individuals who abide by social conventions and altruistically limit their own reproduction for the sake of achieving population homeostasis. I7 But whenever a group of altruists is ‘infected’ by an individual (or gene) pursuing the anti-social (selfish) strategy of seeking to maximize individual reproduction, such an individual will have an advantage over its altruistic rivals and hence its strategy (by being passed on to its more numerous offspring) will quickly spread through the group. Groups consisting of altruists are thus always vulnerable to subversion from an anti-social invader (a ‘free-rider’) who benefits from the social arrangement but contributes nothing toward its maintenance. Social arrangements of the kind required by Wynne-Edwards’ theory are inherently unstable, and are thus unlikely to be realized very frequently in nature. In criticizing Wynne-Edwards’ view that animals limit their reproductive output in order to preserve food resources, David Lack drew heavily upon results derived from his own research into the factors governing clutch-size in birds.18 Individuals of each species lay a characteristic number of eggs in each breeding cycle. For example, the species-specific clutch-size for swifts is three. Because the reproductive rate would seem to be directly related to the number of eggs in each clutch, it may be asked why the individuals of particular species do not lay more eggs than they typically do. Lack argued that breeding pairs of at least some species produce as many young as possible. His experiments showed that even under very favourable conditions, swifts that laid clutches of three eggs fledged more offspring than those who laid four eggs. Further experiments showed that the upper limit of clutch-size is set by the fact that 16For a more detailed, and nuanced, discussion see W. Provine, ‘The Role of Mathematical Population Genetics in the Evolutionary Synthesis of the 1930s and 194Os’, Studies in the History of Biology 2 (1978) 167-192. “5 Maynard Smith, ‘Group Selection and Kin Selection: A Rejoinder’, Nature 201 (1964), 1145Ll147. ‘sD . Lack , ‘Significance of Clutch-Size in Swift and Grouse’, Nature 203 (1964) 98-99. Lack continued the attack in his 1965 Presidential Address to the British Ecological Society‘Evolutionary Ecology’, Journal of Animal Ecology 34 (1965) 223-231-and finally, in more detail, in a special appendix to his book Populution Studies of Birds (London: Oxford University Press, 1966).

326

Studies

in History and Philosophy of Science

when more young than this are produced, parents are unable to find enough food for all, so that increased mortality reduces the total number of offspring reaching maturity. Lack concluded that there is a sense in which animals limit their own numbers, but they do so, not for the good of their group, as Wynne-Edwards supposed, but rather to enhance their own individual fitness. Population ‘regulation’, Lack argued, is simply an effect of organisms striving to maximize their individual fitness in resource-limited environments. Finally, George C. Williams delivered what many biologists consider to have been the fatal coup de grace to Wynne-Edwards’ theory.‘9 He argued that the sort of group adaptations Wynne-Edwards thought needed to be explained in terms of group selection could be explained more parsimoniously as the statistical effects of selection operating on individual organisms. Schooling in fish, for example, should be explained, not as a means for fish to assess the density of their population and to adjust their reproduction accordingly, as Wynne-Edwards supposed, but rather simply as the cumulative effect of the selfish behaviour of individual organisms, each of which uses the bodies of its schoolmates to create a buffer between itself and any predatorsvlurking nearby. Other supposed group adaptations could be disposed of in similar ways. But if so, then group selection, which was introduced to explain group adaptations, could be dismissed without further adieu. 2.3. Closure of Controversy Although not the last critique of Animal L&persion to appear in the 1960s Williams’ arguments convinced many biologists that group adaptations of the sort that Wynne-Edwards considered pervasive in nature were chimerical, and consequently that postulating group selection was unnecessary. As a result of such criticisms, ecologists became more self-conscious about the evolutionary mechanisms postulated for ecological phenomena, and ecology gradually joined the ranks of the other (quantitative, genetically-informed) biological disciplines. One casualty of this advance was the idea of ‘group selection’. At the beginning of the 1960s group selectionist explanations of biological phenomena were common. By the end of the decade, as one commentator has noted, ‘group selection rivaled Lamarckianism as the most thoroughly repudiated idea in evolutionary theory’.20 Animal Dispersion is widely acknowledged as having been important in jolting biologists out of their vague, “George C. Williams, Adaptation and Natural Selection: A Critique of Some Current Evolutionary Thought (Princeton University Press, 1966). Although Wynne-Edwards’ theory receives more explicit attention in the book than the views of other biologists, Williams’ book was begun as a response to the views of A. E. Emerson. Williams reports that as a postdoctoral student in the mid-1950s he heard a lecture by Emerson in Chicago and left convinced that something had to be done about what seemed to him to be such misguided evolutionary thinking. Wynne-Edwards’ book appeared as the writing of Adaptation and Natural Selection was already underway (Williams, personal communication). “D. S. Wilson, ‘The Group Selection Controversy: History and Current Status’, Annual Review of Ecology and System&es 14 (1983), 159.

Kitcher’s Compromise

321

‘group-selectionist’ slumbers, but is now cited by neo-Darwinists chiefly as a shining example of how not to frame evolutionary explanations.21 3. Critical Evaluation of the Compromise Model 3. I. Kiteher’s Ladder

This brief historical sketch will allow us to begin probing Kitcher’s model. The Compromise Model is clearly a highly idealized account of the structure of consensus formation in science. In order to instantiate this model with a specific historical episode, it is necessary to make a number of simplifying assumptions. A process of ‘purification’ must be undertaken in order to ‘prepare’ a case-study for use in confirming the model.“2 Borrowing a metaphor that Kitcher himself employs in another context, we might describe this process as climbing ‘Kitcher’s Ladder’.23 For any scientific debate to be analyzed in terms of the Compromise Model, a number of ‘rungs’ must be negotiated. Among the steps to be taken are the following:

(1) Select an episode (a ‘debate’) from the history of science, individuating it from the larger context(s) in which it occurs. Partition the debate into ‘early’ and ‘late’ phases. (2) (3) Identify the key players and positions in the debate. (4) By examining subsequent developments, identify the (eventual) ‘winners’ and ‘losers’. (5) Identify the crucial argumentative moves in the debate. (6) Assume that specific argumentative moves represent specific underlying cognitive processes. (7) Postulate specific underlying cognitive processes. (8) Establish a metric for assessing the cognitive progress-promoting properties of cognitive processes. evaluate the relative cognitive progress-promoting (9) Comparatively properties of the specific cognitive processes implicated in the debate. (10) Make a link between the specific cognitive processes undergone by participants in the debate and their expression in arguments, articles, books, and other modes of communication available to the relevant scientific communities. “See, e.g. R. Trivers, Social Evolution (Menlo Park, CA: Benjamin/Cummings, 1985); R. Dawkins, The Se&h Gene, revised edn (Oxford University Press, 1989); and H. Cronin, The Ant and the Peacock (Cambridge University Press, 1991). ‘*The process of making historical materials suitable for use in a philosophical model bears a striking resemblance to the operations that must be performed upon natural materials in order to transform them into publishable scientific results. See B. Latour and S. Woolgar, Laboratory Life: The Construction of Scientzjic Facts (Princeton University Press, 1986). In both cases the finished products differ significantly from the raw materials from which they are constructed. Kitcher describes. and systematically % the process of discrediting ‘pop sociobiology’, demolishes, what he calls ‘Wilson’s Ladder’. See P. Kitcher, Vaulting Ambition: Sociobiology and the Quesrfor Human Nuture (Cambridge, MA: MIT Press, 1985), pp. 1618, 125-132.

328

Studies in History and Philosophy of Science

an examination of the trajectories of belief in the relevant (11) Through scientific communities, assess the relative influence of the various cognitive processes

undergone

by participants

epistemic

factors in producing

scientific

communities.

in the debate

the emergence

with

of consensus

salient

non-

in the relevant

(12) Show that the scientific debate in question reached closure in the particular way that it did principally through the integration into the community’s thinking of a process for modifying practice in promoting cognitive progress to other protagonists

that is markedly superior processes undergone by

in the debate.

3.2. Perils of the Climb At each step of this ‘ladder’ a number of relatively unconstrained decisions must be made, each of which is likely to profoundly affect the final outcome of the analysis. Which episode from the history of science one selects for analysis, how one chooses to individuate this episode from the larger scientific and cultural contexts of which it is a part, and how one partitions the debate encompassed

in this episode into ‘early’ and ‘late’ phases are all likely to have

major effects on the kinds of conclusions one reaches regarding consensus formation in science .24 Likewise, although a number

the nature of of individuals

are typically involved in any major scientific debate, it is practically impossible to fully analyze the complex roles played by all (or even any) of them. So it is necessary to restrict the analysis to just those individuals deemed to be most important. strategy

How

are these

individuals

to be identified?

is to focus on the most ‘visible’ of the disputants,

participants outcome

to the shadowy of the analysis

periphery.

in important

Clearly, ways.

do, the identity of the eventual ‘winners’ identification of the ‘key’ disputants in

The most

such a move

Knowing,

common

and to relegate other may bias the

as we unavoidably

and ‘losers’ makes an objective a historical episode even more

precarious. ?t may be no accident that philosophers who prefer Rationalist accounts of science tend to emphasize fairly uncontroversial historical examples embodying consensus positions that represent established scientific knowledge, e.g. Copernican astronomy, Darwinian evolutionary theory, Wegnerian plate tectonics, etc. In addition to Kitcher, op. cit., note 7, see R. Giere, Explaining Science: A Cognitive Approach (University of Chicago Press, 1988); P. Thagard, Conceptual Revolutions (Princeton University Press, 1992). Constructivists (i.e. ‘Antirationalists’) tend to emphasize more controversial cases, e.g. recent theorizing about quarks, gravity waves, solar neutrinos, etc. See, for example, A. Pickering, Constructing Quarks: A Sociological History of Particle Physics (Edinburgh University Press, 1984); H. Collins, Changing Order: Replication and Induction in ScientiJic Practice (Beverly Hills, CA: Sage, 1985); T. Pinch, Confronting Nature: The Sociology qf Solar-Neutrino Detection (Dordrecht: Reidel, 1986). Different philosophical conclusions are, evidently, more easily derived from some historical episodes than from others. Some implications of this observation for the Rationalist-Constructivist debate are explored by A. Nelson, ‘How Could Scientific Facts be Socially Constructed?‘, Studies in History and Philosophy OJ Science 25 (1994). 535-547.

Kitcher’s Compromise

329

In the recounting of a historical episode in order to illustrate a philosophical model of scientific progress, the inclination to select the best arguments of the eventual winners and the worst arguments of the eventual losers, in order to highlight the differences between them, may be difficult. to resist. When this requires postulating specific ‘cognitive processes’ that are presumed to underlay the specific arguments identified, with little or no independent checks on our constructions, this danger is greatly increased. The situation becomes even more perilous when one is faced with the task of evaluating the ‘cognitive progress-promoting properties’ of these postulated cognitive processes. Because there is no readily accessible metric of cognitive progress against which cognitive processes could be measured, one is forced to rely upon commonsense generalizations derived from the history of science. Because the cognitive processes to be assessed belong to this same history, this procedure risks begging the question. An additional difficulty arises in moving from the specific cognitive processes undergone by the participants in the debate to the effect on their scientific communities. A link must be made between these cognitive processes and their publicly accessible forms. Because these cognitive processes, as such, are no more accessible to the immediate scientific community than they are to retrospective analysts, arguments as embodied in articles, books, lectures and personal discussions must serve as proxies once again. Because the transference of information in a scientific community requires that this information be materially embodied, and scientific communities can be highly viscous media for the flow of information, not all the arguments deployed during the course of a scientific debate will be known equally well to all members of the relevant scientific communities. Consequently, determining which arguments were generally known and assessing the actual persuasive efficacy of each is likely to be a very difficult undertaking requiring considerable sociological as well as historical investigation. A convincing, well-established account is unlikely to emerge in unambiguous form from a casually constructed historical case-study. Finally, in order to reach the top rung of the ladder, we have to be able to determine with confidence the relative importance of various causes in bringing about the final ‘community decision’. Whereas ‘Antirationalist’ accounts maintain that scientific debates close when one group is forced from the community by a competitor commanding more non-epistemic power, according to the Compromise Model, ‘scientific debates are ultimately closed through the articulation and acceptance of decisive arguments’.25 Although ‘various kinds of social factors may operate in the acquisition of power by the victorious group’, the model requires ‘only that the argument emerging when the debate crystallizes should be the principal source of power’.26 In other words, ‘in a 25Kitcher, Z6Kitcher,

op. cit., note 7, op. cit., note 7,

p. 201. p. 201.

330

Studies

in History

and Philosophy

of Science

competition between the social factors and arguments leading in a contrary direction, the acquisition of power should be more affected by the arguments’.27 Establishing these claims requires that it be possible to: (i) clearly distinguish epistemic (argumentative) and non-epistemic (social) factors in scientific debates; (ii) assign values to the ‘power’ possessed by epistemic and nonepistemic factors, respectively; (iii) show that epistemic factors have greater power than the non-epistemic factors; and (iv) demonstrate that in virtue of their greater power epistemic factors constitute the principal cause of scientific closure. Each of these requirements is controversial. The ‘sociology of scientific knowledge’ is premised on the claim that epistemic and social factors are not (or at least not always) easily distinguishable. 28 What is taken to be a significant epistemic virtue for assessing a scientific theory, and how one ranks different epistemic virtues in cases of conflict between theories displaying epistemic virtues to different degrees, are sometimes inseparable from the social context(s) in which these theories are being considered. Assuming that epistemic and social factors can be distinguished in a given context, however, leaves untouched the difficulty associated with assigning ‘power values’ to these different factors. Carrying out such assignments successfully may presuppose a level of access to them that surpasses our cognitive capacities. Finally, determining which kind of factor was not only sufficient to constitute the principle cause of scientific closure, but actually played this role in particular historical episodes, may unwittingly embroil us in unresolvable counterfactual speculations about what would have happened in a given scientific debate had various epistemic and/or social factors been different. While it is true that counterfactual historical claims can sometimes be made convincingly, doing so requires a much more detailed historical investigation than the typical ‘case-study’ approach to the history of science is able to supply. 3.3. Unsteady Steps These general concerns become acute problems when we attempt to apply the Compromise Model to a particular historical episode such as the one sketched earlier. There I described the debate over group selection as transpiring in the 1960s giving the impression that the debate could be neatly bracketed by the appearances of Wynne-Edwards’ (1962) and Williams’ (1966) books, “Kitcher. op. cit., note 7, pp. 201-202. In a footnote (p. 201, n. 27) Kitcher acknowledges that these statements may be entirely congenial to Rationalists, and that the Compromise Model ‘may be viewed as making some rather obvious amendments to classical rationalism once the apsychologistic approach to rutionalism has been abandoned’ (emphasis in original). As I will argue shortly, it is precisely because Kitcher fails to fully abandon the apsychologistic approach to science that his model encounters the difficulties that it does. *‘For the classic statement, see D. Bloor, Knowledge and Social Imagery (University of Chicago Press, 1976; second edn, 1991).

Kitcher’s

Compromise

331

respectively. A broader historical perspective reveals this to be a gross oversimplification necessary in presenting a compact account of a particularly interesting episode in a much longer ongoing controversy. Debates about population regulation-including the issue of the cogency of treating populations as characterized by evolved self-regulating mechanisms-predate the appearance of Wynne-Edwards’ book, and were by no means settled by Williams’ book.29 Despite their eclipse at the end of the 1960s theories of group selection reappeared in the early 1970s and have enjoyed a kind of renaissance in recent years such that it is no longer considered ‘naive’ to employ group selectionist explanations of particular biological phenomena.3O There is presently a heated debate amongst biologists about the proper application of group selectionist explanations that shows little sign of abating any time soon.3l Interestingly, Wynne-Edwards himself has continued to defend his theory to the present, claiming that important evidence, the significance of which his critics fail to appreciate, counts in its favour. 32 Some biologists have come to WynneEdwards’ defence, contending that ‘his critics have argued against the untenability of a position never his own’, and that Wynne-Edwards’ theory can be rehabilitated by showing how the basic account of population dynamics implicit in his theory can be made plausible in light of recent theories of population structuring.33 None of this, of course, is intended to show that Wynne-Edwards was ‘right after all’ (although, in some respects, he may be). Rather, it demonstrates the serious oversimplification involved in talking about ‘early’ and ‘late’ phases of a scientific debate, and of ‘winners’ and ‘losers’, without a more thorough investigation into the shifting boundaries of dispute which often involve later re-evaluations of earlier positions. Such simplification may be essential for applying a philosophical model to historical materials, but it may also mask significant subtleties and developments that a more detailed historical investigation would uncover. ‘?See R. H Tamarin (ed.), Population Regulation (Stroudsburg, PA: Dowden, Hutchinson & Ross, 1978), for a collection of classic papers. For more recent discussions, see M. P. Hassell and R. M. May (eds), Population Regulation and Dynamics (London: Royal Society, 1990). 30See, for example: M. E. Gilpin, Group Selection in Predator-Prey Communities (Princeton University Press, 1975); M. J. Wade, ‘Group Selection Among Laboratory Populations of Tribolium’, Proceedings of the National Academy of Sciences (U.S. A.) 113 (1976), 399417; D. S. Wilson, ‘A Theory of Group Selection’, Proceedings of the National Academy of Sciences (U.S.A.) 72 (1975). 1433146. For a critical evaluation of models of group selection, see T. Shanahan, ‘Group Selection and the Evolution of Myxomatosis’, Evolutionary Theory 9 (1990), 239-254. “See the target article, along with the multiple responses that follow it, by D. S. Wilson and E. Sober, ‘Reintroducing Group Selection to the Human Behavioral Sciences’, Behavioral and Brain Sciences 17 (1994), 585-654. “V . C Wynne-Edwards, Evolution Through Group Selection (Oxford: Blackwell, 1986); ‘Ecology Denies Neo-Darwinism’, The Ecologist 21 (1991), 136141; ‘A Rationale for Group Selection’, Journal of Theoretical Biology 162 (1993). l-22. 33G Pollock, ‘Suspending Disbelief: Of Wynne-Edwards and His Critics’, Journal of Evolutionary Biology 2 (1988), 205-221.

Studies in History and Philosophy of Science

332

The foregoing difficulties might still be regarded as rather minor inasmuch as they could, in principle, be resolved by undertaking more thorough historical investigations and by more carefully qualifying claims about ‘early’ and ‘late’ phases of ‘the debate’, ‘winners’ and ‘losers’, and the like. More difficult problems attend attempts to ascend to the higher rungs of the ladder, where we are required to identify the relevant cognitive processes undergone by the participants in a debate and to comparatively evaluate their cognitive progresspromoting properties. The fundamental problem here is obvious: cognitive processes, as such, are not available for direct inspection. Published arguments are available for direct inspection, and so Kitcher takes them as representing underlying cognitive processes, but it must be emphasized that this inference requires considerable justification, an obligation that is nowhere discharged in The Advancement of Science. Likewise, we are not told how to represent the cognitive processes themselves. To a distressingly large extent, we are left to our own devices. Suppose, however, that we decide to take advantage of this interpretive liberty by representing the cognitive processes of the participants in the debate in question, as ‘suggested’ by their argumentative moves, as cognitive directives for guiding the collection and interpretation of evidence, the framing of arguments, and the drawing of conclusions. For example, a detailed analysis of Wynne-Edwards’ arguments in Animal Dispersion, as well as in papers published before and after the eruption of the controversy,34 might justify postulation of the following cognitive directives: l

Seek simple, phenomena.

unifying,

synthetic

explanations

of complex

biological

l

Treat natural selection as an optimizing agent with vast resources of biological variation and time at its disposal, with the consequence that optimal solutions to biological problems are both possible and frequently instantiated in biological phenomena.

l

Adopt a hierarchical view of the operation of natural selection, according to which selection is capable of producing adaptations among entities at any level of the biological hierarchy exhibiting variation, differential fitness and heritability of traits.

In similar fashion, detailed analysis of the published arguments of WynneEdwards’ critics might yield the following cognitive directives: l

Employ game theoretic considerations when assessing the likelihood of encountering various social organizations in nature, and require that

‘%ee for example, V. C. Wynne-Edwards, ‘The Control of Population-Density Through Social Beha&r: A Hypothesis’, Ibis 101 (1959), 43-I; ‘(Reply to Maynard Smith)‘, Nature 201 (1964). 1147; ‘(Reply to Perrins)‘, Natufe 201 (1964), 1148; ‘Reply to Lack’s “Significance of Clutch-Size in Swift and Grouse” ‘, Nature 203 (1964), 99.

Kitcher’s Compromise

333

proposed accounts satisfy typically realized empirical conditions. (Maynard Smith) l

Insist on experimental confirmation

of ecological hypotheses. (Lack)

l

Recognize adaptation at no higher a level of organization than is demanded by the evidence. Whenever possible explain group characteristics as the statistical effects of the interactions of individual organisms. (Williams)

Suppose that the foregoing cognitive directives do undergird the explicit, publicly-accessible arguments deployed by participants on both sides of the debate. On what grounds are we to comparatively assess the cognitive progress-promoting designs of these directives? Clearly what we cannot do is to suppose (or argue) that the cognitive processes of the (eventual) ‘winners’ were superior to those of the (eventual) ‘losers’ because the former but not the latter eventuated in true (or at least better) beliefs. To argue thus would be to beg the question about the relationship between the cognitive progress-promoting properties of the cognitive processes undergone by the disputants and the epistemic status of the eventual outcome of the debate. This relationship is precisely what is at issue. Instead what we must do is examine the cognitive processes themselves and ask whether those employed by the ‘winners’ are such that they promote cognitive progress better than those employed by the ‘losers’. This requirement raises some additional difficulties. All of the ‘cognitive directives’ listed above can boast the support of distinguished contemporary biologists. All of these directives can be shown to underlie successful (i.e. cognitively progressive) scientific practice. But if so, then one cannot appeal to any record of superior cognitive-progressiveness in order to distinguish one set of directives from the other. This problem is compounded by the fact that all of the directives pertain to the broad context of theorizing or explaining in evolutionary biology and ecology. As such, they are peculiar to just the kind of issues at stake in the debate over group selection. They cannot, therefore, be comparatively assessed in terms of their value for promoting cognitive progress in the disciplines of evolutionary biology and ecology-or even, perhaps, in ‘biology’-without begging the crucial question. If we are to judge of the relative superiority of the cognitive progresspromoting designs of the cognitive processes employed by participants in the debate, we shall have to do so in terms of some standard of evaluation independent of the particular issues at stake in this debate. Perhaps each of the cognitive directives could be rendered more general and hence allow for assessment within a broader cognitive domain. For example, Lack’s insistence on experimental confirmation of ecological hypotheses could be generalized as a cognitive directive insisting on experimental confirmation of scientiJic hypotheses. We could then assess the cognitive progress-promoting characteristics of

Studies in History and Philosophy of Science

334

such a directive

against

a broad

background

of scientific

practices,

including

drawn from the history of science. But this strategy has serious (and obvious) limitations. Scientists do not (typically) disagree on the importance of bringing evidence to bear on scientific hypotheses. Rather, scientific debates are typically characterized by disagreements about what is to count as relevant evidence, disputes about the interpretation and significance of evidence, and differences of opinion about how to weigh bodies of evidence that point in different directions. A cognitive directive like ‘Insist on experimental confirmation of ecological hypotheses’, therefore, will be virtually vacuous. The fact that Wynne-Edwards, no less than Lack, insisted on experimental verification for the scientific hypotheses being considered renders such a move doubly vacuous. For the other cognitive directives listed, formulating them in more ‘general’, non-biological form would amount to gutting them of all content, and would further remove their comparative evaluation from the specific historical context under consideration. We are still without an independent standard of evaluation. Is it possible to escape from this predicament? From whence are our standards of cognitive progressiveness to be derived? As Kitcher notes, ‘As we think about design for cognitive progress it quickly becomes apparent that there are a number of possible ways to construct standards’.35 Kitcher’s solution to this problem is to propose an ‘External Standard’ (ES) that embodies a rigourous criterion for defining cognitive progress: examples

The shift from one individual practice to another is rational if and only if the process through which the shift was made has a success ratio at least as high as that of any other process used by human beings (past, present, and future) across the set of epistemic contexts that includes all possible combinations of possible initial practices (for human beings) and possible stimuli (given the world as it is and the characteristics of the human recipient).36 (ES):

It (almost)

goes without

(ES) presents an extremely demanding ‘For the study of the historical development of be far too demanding’.j’ Besides other problems, to the fact that processes available to subjects at science may not have been available at others, and

saying

standard. As Kitcher notes, science... (ES) may prove to (ES) ‘makes no concessions some stages in the history of

that

it requires that only optimal processes count as rational’.38 Nevertheless, Kitcher believes that ‘(ES) provides a valuable target for the methodologist: pointing to a goal-optimal cognitive design-at which we aim’.39 This may be true, but treating (ES) as a ‘valuable methodological ideal’ is quite different from employing it as a standard of cognitive progressiveness for assessing the 35Kitcher. 3”Kitcher, 37Kitcher, “Kitcher, “Kitcher,

op. op. op. op. op.

cit., note cit., note cit., note cit.. note fir., note

7, p. 188. 7, p. 189. 7, p. 191. 7, p. 189. 7, p. 190.

Kitcher’s Compromise

335

cognitive processes of participants in a concrete, historically specific scientific debate. What we need is not a target at which to aim but rather a standard by which we may make confident judgements about the relative importance of various factors in explaining specific scientific transitions. (ES) is of little help in this regard. 4. Idealization and Empirical Accuracy in History and Philosophy of Science Kitcher notes that ‘The rationalist model is an idealization, and it neglects complications that are regarded as insignificant’.40 Clearly the same is true for the Compromise Model he champions, although he fails to appreciate the implications of this point for his project. At one point Kitcher acknowledges that the success of his model depends on whether ‘the charms of simplicity and historical accuracy can be combined’.41 This is, of course, the crucial issue for Kitcher’s project, and indeed for anyone who wishes to construct idealized models of scientific change. Unfortunately, his explicit methodological reflections on this issue are ambiguous. On the one hand, he thinks that carefully chosen historical examples can obviate the need for more precise philosophical explications of key concepts of the sort that occupied us above, concepts like ‘cognitive progress-promoting processes’. According to Kitcher, ‘Simple claims about rationality can avoid fine-grained resolutions by focusing on the differences between the processes undergone by those whom we retrospectively praise and the processes of their opponents’.42 As an example he cites the case of Galileo and his Aristotelian adversaries. ‘Galileo’s modification of his practice counts as rational because looking at the heavens rather than not looking is, it seems, likely to have a high success ratio in modifying astronomical practice’.43 He concludes that ‘The details of which (cognitive) processes are available, the epistemic contexts we consider, the ways cognitive improvement is measured seem insignificant beside the gross differences we expect to find. However we do the accounting we think that we shall get the same result’.44 On the other hand, he recognizes that conducting a more fine-grained historical analysis introduces complications that threaten to undermine the confident assertions of his philosophical model: ‘Of course, as the history becomes more refined and it is understood why the telescope might have seemed an unreliable instrument, confidence that we can ignore the vagueness begins to waver’.4s These remarks deserve closer scrutiny. If we restrict ourselves to making ‘simple claims’ about those whom we consider to be the eventual winners and losers of scientific debates, then much of the vagueness in assessing the cognitive 4”Kitcher, 4’Kitcher, 42Kitcher, 43Kitcher, “Kitcher, 45Kitcher,

op. cit., note op. cit., note op. cit., note op. cit., note op. cit., note up. cit., note

7, 7, 7, 7, 7, 7,

p. p. p. p. p. p.

197. 202. 190. 190; emphasis 190. 190.

in original.

336

Studies in History and Philosophy of Science

processes employed by these participants does indeed disappear. As Kitcher notes, however, as soon as we begin insisting on a more fine-grained historical account, our confidence that we can ignore the historical details is shaken. This is crucial. ‘Looking or not looking at the heavens’ is unlikely to represent the best historical account we can give of the cognitive options available to Galileo and his intellectual adversaries. As the historical account becomes more refined, its bearing on philosophical claims about the superiority of the cognitive processes of the ‘winners’ becomes increasingly problematic. Clearly what we cannot do, if we wish to avoid circularity, is to construct our historical accounts on the basis of what we ‘expect to find’, and to assume, at the outset, that ‘however we do the accounting... we shall get the same result’. To proceed in this fashion is to allow the requirements of a philosophical model to determine the structure and depth of our historical inquiry. Under these conditions the historical examples can at best serve to ‘illustrate’ a philosophical model constructed on the basis of non-historical considerations. As a consequence, however, the historical material is rendered useless as a source of philosophical insights. This is a rather high price to pay for the validation of a philosophical model.46 It should be evident that this problem is not peculiar to Kitcher’s model. It was precisely the same problem that some critics have focused upon in the work of Lakatos.47 To put the problem in its most general form, there may be an inverse relationship between the degree of historical detail provided in a historical case-study, and the level of confirmation that case-study may provide for a philosophically elegant model of science. That is, it may be the case that we can have a simple, philosophically elegant model of scientific change, or a rich, historically contextualized account, but we cannot combine both in the same description. Such a conclusion, if correct, bears directly on the question of the relationship between history of science and philosophy of science, and challenges Lakatos’ (and Kitcher’s) optimistic view of the relationship between the two disciplines. 4hKitcher’s attempt to support his model by appealing to episodes in the history of science is not as ‘lovingly done’ as it might be, making empirical evaluation of his model difficult. The historical accounts he provides are insufficiently developed to allow us to assess the application of his model properly. There is an irony in this, because in a previous book (op. cit., note 23) Kitcher takes sociobiologist E. 0. Wilson to task precisely for his failure to meet appropriate standards for doing behavioural ecology. While Kitcher has nothing but praise for Wilson’s earlier work on social insects, and on ants in particular, when it comes to the application of sociobiological principles to Homo sapiens Kitcher argues that Wilson abandons his usual caution and begins to extend his claims far beyond what his empirical base can properly support. In his most recent book Kitcher is guilty of precisely the same sort of lapse that he accuses Wilson of. Kitcher’s earlier books are exemplary, setting standards for others to emulate. In The Advuncement of Science the same caution is not observed. The philosophy of science presented is brilliant. The history of science is less so. But the philosophy of science is supposed to be assessed in terms of how well it captures the historical development of science. The sum is therefore extremely interesting and provocative, but ultimately only as compelling as its weakest part. 47See especially Thomason, op. cit., note 6.

Kitcher’s Compromise

337

Recall that whereas Lakatos was content to show how his model of scientific change could be illustrated by ‘rational reconstructions’ of episodes in the history of science, Kitcher is aiming at a naturalistic (e.g. psychologically grounded) account of science. Despite his naturalistic pronouncements, however, his practice strongly resembles the rational reconstructionist strategy of Lakatos.48 Fitting historical materials into his model requires a severe compromise of historiographical standards no less radical than that required by Lakatos’ model. The sense in which such philosophical models enhance our understanding of the actual development of science, therefore, remains unclear. This gap between Kitcher’s prqject and its actual products is not entirely surprising. It is a truism that philosophers of science are typically tempted to privilege the ‘big picture’ over the messy details of particular historical episodes, whereas historians of science sometimes take it as a mark of the maturity of their profession that their detailed historical investigations need not acknowledge, much less directly address, the relationship of their inquiries to broader philosophical concerns (e.g. the issue of whether scientific knowledge is ‘progressive’, and how such progress should be measured). It is not difficult, however, to see how history of science could bear on philosophical claims. Detailed historical studies of particular scientific episodes are the ideal instrument for dissecting naturalistically-committed philosophical models and exposing their hidden presuppositions and commitments. But this amounts to a very different, and much more limited, vision of the relationship between history and philosophy of science than that comprised in the confident pronouncements of philosophers like Lakatos and Kitcher. What is less clear is whether, or how, historical and philosophical approaches can be combined to produce a convincing, positive characterization of science. Our discussion above suggests that whereas philosophy of science can be useful in discerning and evaluating patterns of reasoning underlying the publicly-available arguments in particular historical episodes, and may, to some extent, be able to make general claims about scientific reasoning, history of science is philosophically useful primarily because, in addition to providing content for philosophical reflection, when properly executed it reveals the limitations of idealized philosophical models of science. Lakatos, it will be recalled, once declared that ‘Philosophy of science without history of science is empty; history of science without philosophy of science is blind’. The foregoing considerations suggest a less pithy but, I think, more accurate formulation of the disciplinary relationships at issue: ‘Philosophy of science without serious history of science provides little insight into science as it has been and is being practiced; history of science, without being related to the general problems and concerns about the nature of science that it is the business 48See M. Solomon ‘Legend Naturalism and Scientific Progress: An Essay on Philip Kitcher’s Advuncement of Sciek’, Studies in History and Philosophy of Science 26 (1995), 205-218.

The

338

Studies in History and Philosophy oj’science

of philosophy of science to investigate, cannot properly assess the implications of its detailed findings about particular historical episodes for our understanding of the scientific enterprise’. The aspiration to develop an approach that integrates the considerable resources of history of science and philosophy of science, without compromising the internal standards of either, is as yet unrealized, but still eminently worthy of pursuit. would like to thank Philip Kitcher, Edward Manier, Robert Westman, and the anonymous referee(s) of this journal for helpful comments on the ideas expressed in this paper. Acknowledgements-I