Toward a Monistic Theory of Science the “Strong Programme” Reconsidered

10.1177/0048393103252780PHILOSOPHYKempARTICLE / STRONG OF PROGRAMMETHE SOCIAL SCIENCES RECONSIDERED / September 2003 Toward a Monistic Theory of Science The “Strong Programme” Reconsidered

STEPHEN KEMP University of Sussex

This article considers the “Strong Programme” account of scientific knowledge from a fresh perspective. It argues that insufficient attention has been paid to the Strong Programme’s monistic intent, that is, its aim to unify considerations of instrumental adequacy and social interests in explanations of the development of scientific knowledge. Although sharing the judgment of many critics that the Strong Programme approach is flawed, the article diverges from standard criticisms by suggesting that the best alternative is not a dualistic framework but a more adequate monistic approach.

Keywords: Strong Programme; interests; monism; finitism; classification

1. INTRODUCTION

The Strong Programme (SP) analysis of scientific knowledge is generally acknowledged to have been a crucial inspiration for developments in the sociological study of science over the past 25 years. David Bloor and Barry Barnes, sometimes in collaboration with Don- ald MacKenzie and Steven Shapin, rigorously argued that social factors must be incorporated into any account of the dynamics of scientific knowledge. Crucially, they claimed that the commitments of scientists to specific beliefs cannot be explained without reference to social interests, even if those beliefs are judged to be rational or adequate. These arguments stimulated many responses, including sple- netic attacks (e.g., Slezak 1994), careful critical appraisals (e.g., Keat

Received 1 November 2001

My thanks go to the British Academy for supporting me during the completion of this article through their Postdoctoral Fellowship scheme. Thanks also to Gillian Haddow, Gregor McLennan, Maureen O’Malley, Sharani Osborn, Irene Rafanell, and especially John Holmwood for their comments on this piece. Philosophy of the Social Sciences, Vol. 33 No. 3, September 2003 311-338 DOI: 10.1177/0048393103252780 © 2003 Sage Publications 311 312 PHILOSOPHY OF THE SOCIAL SCIENCES / September 2003

1989; Turner 1981), and sympathetic defences (e.g., Manicas and Rosenberg 1985, 1988). Even though the basic tenets of the SP approach have remained the same since the 1970s, having been subject to careful elaboration rather than serious revision, new and interesting critical appraisals continued to appear throughout the 1990s (e.g., Roth 1994; Mermin 1998; Pels 1996). Despite the large amount of critical scrutiny given to the SP, I would argue that a fundamental aspect of this approach has not been given the attention that it deserves. Specifically,serious consideration has not been afforded to the SP thesis that the pursuit of social interests and the pursuit of instrumentally adequate knowledge are not antithetical to one another (Barnes 1982; Bloor 1976). To put this thesis another way, the SPers claim that the logics of interest and of instrumental adequacy are not separate and potentially conflicting but are one and the same. It is important to analyze this argument for two reasons. First, if we are to properly assess the merits of the theoretical apparatus produced by the SPers, we must adequately comprehend the intent of the theory. Otherwise, critical remarks may be misdi- rected and the true benefits and problems of the approach obscured.1 Second, there are wider issues at stake in a reassessment of the SP approach. The SP argument for the consonance of social interests and instrumental adequacy is of particular interest because it goes against dominant tendencies within philosophy and sociological theory. In recent times, many approaches have either implicitly or explicitly offered a dualistic analysis of science that separates issues of interest from those of adequacy, suggesting that commitments to belief based on social interests may diverge from commitments based on the rationality or adequacy of belief. Such a position is found in those philosophical responses to the SP that argue that a plausible explanation for scientists’ commitments to irrational beliefs is that they are pursu- ing social interests that conflict with rational appraisals of the adequacy of knowledge (see, e.g., Hollis 1982; Newton-Smith 1981). Dualist theories are also popular within British social science. Realists such as Margaret Archer and Roy Bhaskar argue that the pursuit of social interests may lead actors to commit to beliefs other than those that are the most demonstrably adequate at the time (Archer 1988, 1995; Bhaskar 1986). Against this tendency, the SPers offer a monistic approach, in which no distinction can be made between the pursuit of social interest and the pursuit of adequate belief. Rather, commitments based on social interests and commitments based on the instrumental adequacy of belief are said to be the same. Kemp / STRONG PROGRAMME RECONSIDERED 313

An assessment of the SP framework is thus relevant to the debate between monistic and dualistic approaches. Within the course of this article, I shall be arguing that there are serious difficulties within the SP approach. However, unlike many critics, I shall suggest that we should look to resolve these problems by adopting a more consistent monism rather than shifting to a dualist perspective. To offer a coherent analysis of scientific knowledge, it is necessary to successfully unite the analysis of social interests and the adequacy of knowledge. To draw out the monistic intent of SP theorizing, I shall focus on their theory of classification, which is based on the philosophical doctrine of meaning finitism. This is outlined in some detail in Section 2. Section 3 then critically assesses this theory, lauding its attempt to unify analysis of interests and adequacy but suggesting that this ultimately fails due to untenable accounts of both factors. Section 4 offers some concluding remarks.

2. SCIENTIFIC CLASSIFICATION: A SOCIOLOGICAL APPROACH

This section introduces the SP theory of categorization. Processes of categorization are fundamental to the development of knowledge in two respects. Firstly, it is through categorization that actors produce a differentiated understanding of the world and identify the characteristics of the entities that they encounter. Second, such categorizations are called upon in other forms of knowledge, including generalizations, laws, and predictions. As a result, our understanding of processes of categorization has ramifications for these other forms of knowledge. The theory of classification2 offered by the SPers is intended to demonstrate the importance of social and instrumental influences in the development of classifications. To show this, they build upon a theory of the meaning and content of categorizations called meaning finitism, developed from the work of Mary Hesse (1974) and Ludwig Wittgenstein (1968), among others.3 Although the essentials of the SP approach to classification were developed in the early 1980s (e.g., Barnes 1982), interest in this approach has been renewed by the release of a textbook expounding the SP approach titled Scientific Knowledge: A Sociological Analysis (Barnes, Bloor, and Henry 1996) (for recent discussions, see Mermin 1998; Forge 1996; Miller 1996). 314 PHILOSOPHY OF THE SOCIAL SCIENCES / September 2003

Meaning finitism is an account of categorization processes that is intended to apply to all categorizations, whether these are part of our wider culture or lodged in specialized bodies of knowledge. Put simply, meaning finitism argues that the meaning or content of any term is given by the finite set of particulars that the term designates (Bloor 1996). To grasp the meaning of any term is to learn the set of particular entities that are classified under it and that are thus instances of it. For example, to learn the meaning of the term dog is to learn which particular creatures fall under that categorization. It is the set of instances that conveys the sense in which all members of the category are similar to each other, that is, resemble each other, and are thus appropri- ately classified together (Barnes, Bloor, and Henry 1996, 50). So the sense in which dogs are similar is conveyed by an examination of the instances of the term dog. This understanding of the similarity of one set of entities may also provide a sense of their difference from other sets of entities. By identifying a group of particulars as dogs, we differentiate them from another group of particulars classified as cats. Having made these basic points, the SPers then make a radical move. They argue that, formally speaking, existing categorizations of similarity do not provide any guidance whatsoever for continuing acts of classification, that is, attempts to decide if a new case should be treated as an instance of an existing category. As they put it,

Any classification of the next case can be reconciled with what has already been learned...however other things have been classified so far, the next thing can be classified in any way without any formal inconsistency with earlier practice. (Barnes, Bloor, and Henry 1996, 51)

The reasoning behind this move is fairly simple. The SPers argue that a comparison of any and every pair of entities will find them to be both similar to each other in some respects, and different in other respects (Barnes 1982, 28). As such, if we compare a new particular to members of some existing category, we will always find that it has both similarities and differences to members of that category (Barnes, Bloor, and Henry 1996, 50). This gives us no general grounds for deciding whether it should enter that category. It could be objected that particulars can still be classified by their greater or lesser degrees of resemblance. However, Barnes, Bloor, and Henry (1996) argue that there is no possible “metric for resemblance,” and this being the case, there is no objective basis for claiming that “sameness here outweighs difference there” (Barnes, Bloor, and Kemp / STRONG PROGRAMME RECONSIDERED 315

Henry 1996, 51). Every particular resembles every other particular, and there is no way of measuring the degree of similarity between them in order to establish the formal validity of a categorization. Whenever classifiers try to decide whether a new particular should fall into category C, they find that there are grounds based on C’s current membership to classify the particular as a C or a not-C (Barnes 1981a, 313). In other words, there is no sufficient basis that can be drawn from the existing members of C to assert whether a new particular is an instance of C or not. For example, even when a language user is well-versed with the instances that are classified under dog, this categorization does not supply enough information for them to unproblematically categorize new, previously unexperienced particulars as dog or not-dog. One response to this finitist claim would be to argue that the meaning of a term is not merely in its set of members but also in its relations to other terms. Such relations might then be used to evaluate whether a new particular can properly be assigned to a category, cutting down the indeterminacy claimed by finitists. Certainly, the finitist position accepts the relevance of other classifications to the meaning of a term (Barnes 1981a). Such relations can be of various kinds. A term may be related by being connected through a generalisation such as “dogs have paws,” linking the membership of one category (dogs) with the possession of an attribute defined by another category (paws). In another kind of connection, one category may be connected to another because of certain similarities and contrasts. So dogs are similar to cats in that they are both household pets, but they can also be distinguished by their eating habits and responsiveness to commands. Connections between categories are thus relevant to the meaning of those categories. However, finitists argue that this does not provide any further determinacy to acts of classification, as the further categories called on are also indeterminate (Barnes, Bloor, and Henry 1996, 52). Using category connections, classifiers might attempt to decide if a new particular is a dog by checking to see if it has paws. They would then come across the same problem as before, because the category “paws” is constructed from a finite number of instances of resemblance, and the items on the end of the new animal’s legs bear both a similarity and a difference to those items categorized under paws. This move merely generates a further question: are they paws or not?4 The attempt to connect up terms cannot remove the indeterminacy of classification. No matter how many secondary descriptive catego- 316 PHILOSOPHY OF THE SOCIAL SCIENCES / September 2003 ries we attempt to apply, we have to make a judgment about the resemblance of members of a category to a new particular (Barnes, Bloor, and Henry 1996, 52). These judgments of resemblance are always problematic because there is no formal way of measuring whether the similarity of a new case is close enough to established cases to induct it into the category. The SPers do not limit the scope of their arguments to new acts of categorization but also apply them to those acts of classification that created the current categories. Just as there can be no definitive decision as to whether a newly encountered particular is similar to those in an existing category, there can be no definitive claim that the existing members of the category are all correctly described as similar. It is perfectly possible for classifiers to take any particular placed within an existing category and argue that it does not actually resemble the other members of that category (Barnes, Bloor, and Henry 1996, 57). There are no formal grounds to demonstrate that any such reclassification would be inappropriate (and thus, no formal grounds to demonstrate that any existing act of classification is appropriate). For example, any member of the existing category “dog” could be ejected from that category by classifiers claiming its dissimilarity to other members, and there would be no general grounds to object to this reclassification. Not only are future instances of a category underdetermined, but the current content of a category is formally unstable as well. It is important to emphasize the generality of the finitist claims. The examples that I have given are from commonsense animal terms. However, meaning finitism applies to all concepts, including those of a more precise, scientific nature. Thus, it is as true of carbon as it is of dogs that new cases of classification are always open-ended, and arguments can be made both ways for whether a new particular should be considered carbon or not.5 For example, diamond and carbon black are both classified as carbon even though their chemical behavior is distinct in a way that problematizes their classification as the same element (Barnes, Bloor, and Henry 1996, 67). In other words, diamond and carbon black are both similar to and different from each other, and there is nothing in nature or logic that instructs us to categorize them as the same or different. All beliefs or generalizations that we hold are constructed out of concepts of a finite, open character. Just as any new particular may reasonably be included in a specific set, formally speaking, any event or experience may be reasonably incorporated into specific beliefs and generalizations. Kemp / STRONG PROGRAMME RECONSIDERED 317

Meaning finitism is largely a negative doctrine, arguing that no act of classification can be definitively justified. Although the SPers support these claims, they do not conclude that the decision to classify one way rather than another is a random one. Rather, they insist that to make classification intelligible, we must bring another factor into the equation: the socially defined goals and interests of actors. How, therefore, does reference to the goals and interests of actors bring determinacy to an instance of classification? Essentially, groups of actors have goals and interests that may be forwarded by classifying particulars into one of the categories out of the many formally possible alternatives. These interests provide the motivation for the group to push for one kind of classification over another. As a result, alternative classifications that are equally reasonable in a formal sense will not be equally attractive to “interested” or goal-driven actors. By including reference to social interests, one can make intelligible classificatory decisions that were not explicable by nature or logic alone. Barnes and Bloor’s approach is thus instrumentalist, emphasizing that knowledge is an instrument used by actors in pursuit of their aims and purposes. A useful example of interested classification is outlined by Barnes, Bloor, and Henry (1996, 121-4) in Scientific Knowledge: A Sociological Analysis. The particular classificatory dilemma involved was whether certain types of red dye, derived from aniline, were to be categorized as “the same” or “different.” This became a salient issue during a court case in which the company Renard Freres argued that its patent was being broken by manufacturers selling the “same substance” as that which it had patented. In accordance with meaning finitism, the SPers argue that, formally speaking, the classification could have gone either way. On one hand, the dyes of competing manufacturers sometimes had different practical applications and different names. On the other hand, calling on a certain chemical tradition, it could be argued that the dyes were identical in their chemical composition, the differences being due to superficial additives not relevant to the essential structure. However, when social interests are at stake in the outcome of a decision, actors have good reason to commit to one classification or other. Unsurprisingly, in this case, the members of Renard Freres, its lawyers, and supporters held that all the aniline dyes were the same, and their opponents, with attendant lawyers and supporters, held that they were importantly different. What was undecidable using nature alone became decidable to the groups as they pursued their interests. 318 PHILOSOPHY OF THE SOCIAL SCIENCES / September 2003

From the account of the case given so far, it could be argued that although each side had its own interests to pursue, and thus had clear motivation to classify in one way or another, it might be possible to appeal to neutral scientists to make a decision in a disinterested fash- ion. However, Barnes, Bloor, and Henry (1996) argue that this could not have helped to settle matters. Although independent experts have no specific vested interests in the outcome of the case, they have other commitments and interests located in their disciplinary practice. So it is quite feasible to imagine that a “realist” chemist from one tradition and a “positivist” chemist from another might disagree on the classification of the dyes because of their approaches to classifying sub- stances. In such a case, different kinds of disciplinary practice, involv- ing different aims and interests, would provide the motivation for opposed classifications. Summarizing the issue, Barnes, Bloor, and Henry argue that there is no way for scientists to be entirely disinterested. They state,

In the absence of any project at all, in a situation where goals and interests have no role, the question of the sameness or difference of the aniline reds is undecidable and would be experienced as a meaningless one. (P. 124)

Thus, processes of classification and the construction of knowledge can only be properly understood when the motivating goals and interests of actors are taken into account. This leads us to a crucial move made by the SPers in their attempt to avoid a dualistic approach. Barnes (1982, 106-8) argues that many accounts of knowledge subscribe to a “Manichean myth” about the nature of the interests involved in knowledge generation. When the interest is taken to be instrumental, that is, oriented to “prediction and control,” this is seen as legitimate, expressing as it does the general- ized human need to deal successfully with the environment. On the other hand, when the interest is taken to involve specific ideological or sociopolitical aims, this is seen as illegitimate, and such interests are held to distort or bias the knowledge produced. Barnes (1982), however, does not subscribe to this myth. He argues that there is no essential difference between sociopolitical interests and those related to prediction and control. In other words, sociopolitical interests are always also to be understood as predictive and oriented to dealing successfully with the environment. Likewise, there can be no issue of purely predictive interests aside from the Kemp / STRONG PROGRAMME RECONSIDERED 319 socially sanctioned concerns about which predictions are to be considered important and which are to be discarded as irrelevant. Impor- tantly, then, Barnes argues that instrumental adequacy and social interest are not separate factors potentially operating in tension with one another but are intrinsically bound up with one another. To illustrate these arguments, let us consider two of Barnes’ examples. To show the social aspect of predictive interests, Barnes discusses the development of the existing category “male” on the basis of a new interest or goal that emerges in a subculture (Barnes 1981a, 326- 8). This new goal is to make predictions about the physiological states of “males.” To achieve this, existing exemplars of “male” are sub- jected to empirical investigation to discover as yet unknown features of this group. However, such investigations may not unproblematically reinforce the preexisting sense of who should be in the category “male.” For example, it may be that when investigating males, scientists discover that nearly all people categorized as male have the XY chromosome. By altering the category of male so that it includes only those who have XY chromosomes, a particular predictive goal is served, and generalizations such as “males secrete testosterone” or “males grow beards” are made more successful (Barnes 1981a, 326). To the SPers, of course, this particular development of the category “male” is a contingent one. There is no formal reason why “male” must be developed in a way that emphasizes correlations with factors such as the secretion of testosterone and the growing of beards. When a group argues that this is how the category should develop, they do so on the basis of particular social interests that may not be shared by all. For example, there may be other subgroups who believe that developing “male” in relation to personality or status characteristics is a better option (Barnes 1981a, 326). Furthermore, each of the possible developments has its own predictive advantages, making a “purely technical” decision an impossibility. As Barnes states,

Similarity relations can appear in any number of generalisations, and what makes one better to guess with will invariably make some other worse. . . . There is no logic to determine the relative technical advantages of the alternative strategies of concept application: people simply have to agree which generalisations they will take account of, and agree in their practice how they will be taken account of. (Barnes 1982, 109)

Another example provided by Barnes illustrates the way in which social and political interests influence classification. In this case, he 320 PHILOSOPHY OF THE SOCIAL SCIENCES / September 2003 discusses the dispute between “conservative hereditarians” and “liberal environmentalists” over gender differences (Barnes 1982, 105-9). He suggests that each group could be said to have interests in sustaining opposed beliefs about the source of gendered characteristics. The hereditarians argue that gendered differences in behavior, such as degree of aggressiveness, are the result of biological factors. Likewise, environmentalists argue that differences in gendered behavior are learned and that such characteristics are not based in nature but society. According to Barnes, no matter what evidence is produced in the course of this dispute, each party can reasonably continue to defend its own position (Barnes 1982, 106). He states,

Conceptual fabrics can always be maintained by Duhem-type strategies so that, whatever their form, they remain both internally consistent and also consistent with experience. . . . Hence the protagonists of hereditarian and environmentalist ideologies could indubitably keep their respective systems consistent in just this way. (Barnes 1982, 106)

Any evidence that emerges provides just one more new “particular” to be classified, and formally speaking, there is no reason why it must be classified in one way rather than another. For example, evidence that shows the variability of male aggressiveness in relation to environmental factors can be construed differently. Environmental- ists will claim that it strongly supports their case, whereas hereditarians will argue that it merely illustrates the ways in which innate male aggression can be masked. Each side may be said, then, to be learning from the evidence and making reasonable classifications, without needing to give up its core claims. To Barnes, this suggests that neither group should be interpreted as defending a social interest instead of an interest in prediction and control (Barnes 1982, 108-9). Rather, although divergent classifications serve different social interests, both groups can be said to be learning from the evidence to improve their predictions while continuing to defend their central beliefs. This completes my exposition of the SP account of categorization. Barnes and Bloor argue that processes of categorization are made explicable by reference to the social and instrumental interests that motivate them. They also reject the dualist idea that the pursuit of social interests can be antithetical to the pursuit of instrumentally suc- Kemp / STRONG PROGRAMME RECONSIDERED 321 cessful knowledge. Instead, they suggest that the logic of social interests is inseparable from that of instrumental success. The next section turns to a critical evaluation of their approach.

3. INSTRUMENTAL ADEQUACY AND SOCIAL INTERESTS

Having described the SP’s attempt to offer a nondualistic theory of classification, it is now time to evaluate it. This section will argue that despite the best of intentions, the SP approach does not successfully theorize the drive toward instrumental adequacy in knowledge or the influence of social interests. First, I will suggest that the SPers ultimately remove issues of instrumental success and failure from their account of the dynamics of knowledge. Second, I will suggest that the SP theory of social interests fails to account for the production of the resources that actors are supposed to have an interest in securing. On the face of it, the instrumentalism espoused by Barnes and Bloor seems modest and defensible in its claims. They favor an instrumentalist perspective on the basis that it is not possible to assess the correspondence of knowledge with reality. As Bloor (1976) writes,

At no stage is this correspondence ever perceived, known, or, conse- quently,put to any use. We never have the independent access to reality that would be necessary if it were to be matched up against our theories. (P. 40)

Instead, the assessment of a theory is internal, concerned with creating a coherent account of experience and avoiding anomalies6 (Bloor 1976, 39). The coherence of a theory and its ability to avoid anomalies are relevant because they are signs of its ability to produce instrumentally successful interactions between actors and the environment. Acknowledging this, the SPers suggest that the theories employed by actors may be more or less instrumentally successful. As Barnes (1977) puts it,

Knowledge arises out of our encounters with reality and is continually subject to feedback-correction from these encounters, as failures of prediction manipulation [sic] and control occur. We seek to eliminate such failures, but so far reality has sustained its capacity to surprise us and dash our expectations. (P. 10) 322 PHILOSOPHY OF THE SOCIAL SCIENCES / September 2003

In other words, practical success is never guaranteed by our desire to achieve a certain goal; there is always a question of better or worse knowledge. Bloor (1991) concurs with this point, arguing that since our beliefs are created as instruments, “we must distinguish those which work from those which do not” (p. 40). To cite one of Barnes’s examples, although cancer research has been strongly supported in the postwar era, the small returns from this research suggest that the will to solve a problem may not easily transform into its solution (Barnes 1974, 103). When it comes to analyzing the dynamics of knowledge, the important factor is that the failure of knowledge to be instrumentally successful drives actors to develop more adequate theories. Thus, the failure of existing approaches to curing cancer motivates the development of new beliefs in the hope for more instrumentally successful approaches. We can see, then, that at the level of stated commitment, the SPers offer a plausible account of the influence of instrumental success on dynamics of knowledge. However, it is important to ask whether their theoretical system of analysis can properly incorporate such a concern. I shall argue, in fact, that this system is set up so that instrumental success, or a lack of it, does not have an explanatory role. The first issue to consider here is whether a social group can achieve an interest or goal no matter what experiences arise. If a social group sets up a certain classification or knowledge claim in order to pursue an interest of theirs, is it always the case that this classification successfully serves that interest? In other words, can an interested classification be successful no matter what experiences the world throws up for the social group to deal with? In line with the SPers’ explicit pro- nouncements, we should expect the answer to these questions to be “no.” The desire to reach a particular outcome does not necessarily produce that outcome. With reference to cancer research, just because researchers set up categorizations and classifications to try to predict and control the causes and development of cancer, it does not mean that they successfully manage to do so. Somewhat surprisingly,when it comes to the SP theoretical system, we find that this instrumental orientation is dropped. Instead, the SPers argue that any experience whatsoever can be made consistent with an interested classification, and thus any interested classification can always be seen as successful. The clearest statement of this position is to be found in Barnes’s (1982) T.S. Kuhn and Social Science. In this work, Barnes employs Duhemian arguments to suggest that if a social group has an interest in a particular classification or generalization,7 Kemp / STRONG PROGRAMME RECONSIDERED 323 nothing found in experience has to be taken to undermine the success of their knowledge claim (Barnes 1982, 73-76). Talking of the consequences of Duhem’s claims, Barnes states,

Not only, as a result, is reality incapable of giving lie to an isolated hypothesis; it is no more capable of indicating the existence of defi- ciency in a whole set of connected hypotheses. Awhole conceptual fab- ric can always be made out as in perfect accord with experience, if the community sustaining it is of a mind to do so. (Barnes 1982, 75, emphasis added)

That is to say, a group can always reasonably claim that experience is confirming its classifications, no matter what actually occurs.8 As such, it can always claim that its knowledge is “instrumentally successful,” whatever the outcome. A question still remains as to how this can actually be done. After all, we might accept that although all categorizations are “formally” open, once they are employed for a practical purpose they have a determinacy that cannot be ignored. For example, if we are interested in correlating mobile phone use and the development of a form of cancer, research outcomes will either support this view or challenge it.9 According to Barnes, however, this is not the case. He argues that when applying an interested classification to new data, those phenomena that are found to fit with the category may be inducted into it. Likewise, whatever appears to be anomalous to the existing classification can be placed under a new category, leaving the previous knowledge as it was. Thus, Barnes (1982) states,

Whenever anything of nuisance value arises out of experience, it can always be deemed a new kind of thing or event, and assimilated under a new concept, leaving the existing structure unaltered. (P. 75, emphasis added)

To return to the above example, if one wishes to assert that mobile phone use causes cancer, this can always reasonably be done, and any evidence that appears not to fit with these claims can be justifiably placed in another category that does not problematize them.10 It seems fairly clear that on such an approach, instrumental success and instrumental failure are no longer distinguishable. If one cannot differentiate between what is anomalous to a claim and what coheres with it,11 any claim may be treated as correct (or wrong), and any interest may be said to be successfully (or unsuccessfully) pursued. Instru- 324 PHILOSOPHY OF THE SOCIAL SCIENCES / September 2003 mental success does not need to be striven for by investigation: it can be defined by fiat, regardless of what is experienced. This clearly has important consequences for a sociological approach to knowledge. If the SPers are correct, then an explanation of why actors hold certain beliefs will require no consideration of the practical success or failure of belief. Whatever the interest pursued by actors, this can be said to be fulfilled by their current beliefs and classifications. Furthermore, the instrumental improvement of a category cannot be a legitimate explanation for changes in a belief, given that there are no grounds for imputing pragmatic failure to the belief at its earlier stage. I would argue that these claims cannot be sustained. Some of the difficulty with the SP arguments can be demonstrated by returning to one of Barnes’ examples, the debate between hereditarians and environmentalists over the causes of sexual difference. Here, I would like to consider Barnes’s aforementioned claim that regardless of the evidence brought to bear in this debate, the hereditarians (or the environmentalists) can claim that their theory allows them to successfully deal with the subject matter in question. Barnes asks us to imagine that evidence of environmental influences has emerged in such a way that the hereditarians appear to be on the back foot, desperately try- ing to protect their thesis against refutation (Barnes 1982, 109). He suggests, however, that such maneuvers can equally be characterized as “learning” for the hereditarians, in which they come to comprehend those environmental factors that can mask innate characteristics such as male aggressiveness. For Barnes, the hereditarian theory can always be defended as a successful one, whatever the evidence produced by inquiry. It seems likely, however, that Barnes misdescribes the process of this debate. It is true that in the face of powerful evidence for the environmentalist position, the claim that heredity has some influence on behavior may still be defended. However, this involves shifting from a “strong” hereditarian view, in which most or all behavior is caused by heredity, to a more moderate position that accepts both environmental and hereditary influences. If the evidence pushes hereditarians to accept that some behavior is influenced by environmental factors, then they can no longer legitimately defend the strong view, regardless of any putative interest in doing so. Although rhetor- ically the environmental factors may be described as nonprimary, if hereditarians admit that they have to be employed to successfully describe human behavior, then evidence has produced a shift in the Kemp / STRONG PROGRAMME RECONSIDERED 325 debate.12 In other words, the “success” of their attempt to describe behavior as caused mostly by hereditary factors is thrown into doubt. Even using an example chosen by Barnes, we can see that actors may not easily be able to construct knowledge that achieves their aims and purposes. Nevertheless, we need to tackle more directly Barnes’s argument that those particulars (or pieces of evidence) that appear anomalous to interested classifications can always be dismissed by not classifying them at all or by inventing new categories that remove their problematic character. If this argument was indeed adequate, then it could be said that the hereditarians had needlessly capitulated from their “strong” position. Any group could always reasonably argue that its socially constituted aims and purposes were being successfully achieved. To address this issue, it will be helpful to consider what it means for a piece of knowledge to be “useful” to an actor. Essentially, it means that certain classifications or generalizations are constructed for the purpose of reaching a desired goal. Knowledge is thus designed to successfully produce a certain outcome for the actor. This should give us pause for thought in relation to Barnes’s argument. It is central to his position that actors can neutralize particulars that are anomalous to a category by inventing new categories for them or ignoring them altogether (Barnes 1982, 75; Barnes 1981a, 322). However, the very purpose of the classification is to deal successfully with these kinds of particulars. If it does not do this (as apparent when anomalies emerge), then merely shunting off the anomaly to another category does not help matters. In fact, it is precisely a failure to deal with particulars that actors wished to successfully relate to. I would argue that when the genuinely instrumental concern of action is taken into account, the kind of reclassifications advocated by Barnes cannot make sense to actors. To illustrate this point, we can consider Barnes’s (1981a) discussion of the classification of whales in “On the Conventional Character of Knowledge and Cognition.” Barnes discusses a community who classify creatures into fishes and animals13 in a similar way to us. However, he asks us to imagine that they have never encountered a whale before. When they do so, the question arises as to whether this new particular conflicts with their existing system and forces a reworking of the community’s categories. After all, the whale has characteristics split across existing classifications, being “live-bearing” and “air-breathing” like animals, but “finned” and “sea-dwelling” like fish (Barnes 1981a, 322). 326 PHILOSOPHY OF THE SOCIAL SCIENCES / September 2003

According to Barnes, the community has a range of options in its classificatory response. It can classify the whale as a fish, although this could threaten its categorization of fish, as the whale breathes air; it can classify it as an animal, although this conflicts with generalizations about animals living on land; or it can designate it as a new type of creature all together, a “whale,” which is neither fish nor animal. Barnes argues that this last approach is particularly interesting as by creating a new category, “no existing generalizations need be dis- turbed” (Barnes 1981a, 322). In other words, an instance that is problematic to existing categories can be made unproblematic by creating a new category for it. There is something rather fishy about these arguments. First, it is not the case that a new category has no bearing upon existing classifications. By creating a new category of creature, we alter understandings about how many kinds of creatures there are and how the characteristics of creatures are distinguished. For example, now the categorization “sea-dwelling creature” will include not only fish but whales as well. Some part of the actors’ network of theories is necessarily disrupted by the incorporation of an anomaly, contrary to Barnes’s arguments. More important, as acknowledged within the SP approach, we must argue that classifications are made to achieve a certain purpose. This means that the categories used to classify creatures exist in relation to generalizations found useful and important by the social group. For example, one difference between fish and animals is their method of breeding, and knowledge of this difference is of interest to humans in relation to their husbandry of resources. Likewise, it may be useful to know the domains of creatures to avoid the dangerous ones. (To avoid sharks, one should stay on the land.) It is therefore in the interests of the actors to put the characteristics of whales into a proper relation to existing categories so that they can successfully relate to them also. To avoid classifying whales, or to put them in a separate category that does not disturb existing categories, is to avoid dealing with a whale in a way that is useful to the community. Unless the whale can be fitted in with the understandings of the community, it will be unpredictable and difficult to deal with. Although successful classification may require the reconstruction of understandings, it can hardly be in the community’s interest not to improve its understanding of phenomena that impinge upon it. And yet this is precisely what Barnes suggests in relation to anomalous occurrences.14 The strategies suggested by the SPers to avoid prob- Kemp / STRONG PROGRAMME RECONSIDERED 327 lematic phenomena necessarily result in the restriction of actors’ knowledge and capacity in relation to their interests. Such strategies would not be pursued by actors, as practical success is a genuine issue for them. To understand shifts in hereditarian claims or the way in which a community classifies a whale, analysts must incorporate a concern with the success of the categorization in relation to the interests involved. These points are relevant to the more general analysis of classification. We can accept the meaning finitist point that at some general level, there are no grounds for validating one categorization over its competitors. However, when actors are attempting to achieve a certain result, a categorization can be demonstrably better or worse at achieving that result. It may systematically allow the production of desired effects or may fail in this task and leave interactions with the world unpredictable. By denying this, the SPers take the general argument about the indeterminacy of categorizations and apply it to specific socially interested cases. It is logically possible for actors to ignore anomalies. However, it is in their interest to reconstruct their categories so that these categories facilitate successful interaction with the environment.15 Summarizing the argument so far, it seems that the SP theory of classification cannot account for the dynamic aspects of knowledge connected with instrumental success and failure. Specifically, the SP analysis implies that changes in actors’ beliefs cannot be explained by a shift in actors’ allegiance to the most instrumentally successful beliefs currently available. Instead, the SPers argue that no such dynamic exists because any categorization produced by actors can be defended as successful through the employment of strategies to deflect the disruptive power of anomalous instances. I have argued that this is not the case and that actors will attempt to improve the instrumental adequacy of their beliefs by reconstructing them to remove anomalies. The pursuit of instrumental success thus contrib- utes to an explanation of changes in the state of knowledge. It is crucial to note that these criticisms are not intended to establish the importance of considerations of instrumental success instead of those pertaining to social interests. Certainly, it has been necessary to show that a defensible analysis of instrumental adequacy does not cohere with Barnes’s and Bloor’s conception of interests, which removes issues of instrumental success from consideration. However, this need not be taken to vindicate a dualist separation of considerations of interests and adequacy (as in Archer 1988). Rather, it sug- 328 PHILOSOPHY OF THE SOCIAL SCIENCES / September 2003 gests that the conception of social interest employed by Barnes and Bloor is as problematic as their conception of instrumental adequacy. By criticizing this account of interests, we can indicate what a successful monistic theory might look like, a theory in which considerations of instrumental success and interest are united rather than divided. Once again, in dealing with the SPers work, we will have to distinguish the tenets that are explicitly subscribed to from the notions that are actually employed in their system. In their general statements, the SPers emphasize that the interests ascribed to actors are “instrumental interests” or “interests in prediction and control” (see, e.g., Barnes 1977, 10-19; Barnes 1982). Having an interest involves wishing to control the course of events so that a particular desired outcome or class of outcomes result. The viability of an interest and its continued pursuit by actors must depend on their ability to achieve this result. It can only be in the interests of actors to hold beliefs that allow them to produce the outcomes that they desire. If these general points about interests were adhered to in the SP explanatory framework, the result would be a monistic analysis of knowledge in which social interests are not understood to pull in a different direction to concerns with adequacy but are oriented to its pursuit. However, this is not the case. We have considered already the SPers’ argument that actors can always be said to successfully achieve their aims, whatever experience throws up in relation to their categorizations. With the adequacy or otherwise of a classification no longer a matter of concern in the SP system, interest statements are not connected to the particular modes of prediction and control that make them viable. Instead, they are treated as if they are independent of such concerns. Because the content of categorizations is treated as indeterminate, categorizations can be unproblematically manipulated to achieve whatever social interest is being pursued at the time. Nothing within a categorization can cast into doubt the viability of the social interest being pursued. An example of this style of analysis can be found in Bloor’s (1991) Knowledge and Social Imagery. In this work, Bloor discusses a case well- known in the philosophy of social science, that of the Azande and the institutionalized mode of classification whereby they attribute the status of “witch” to clan members (Bloor 1991, 138-46; see also Evans- Pritchard 1937; Winch 1970; MacIntyre 1970). Witches are considered to be troublemakers by the Azande. They are identifiable by the witchcraft substance that they have in their stomachs, the existence of which can be confirmed by a postmortem. This substance is inherited Kemp / STRONG PROGRAMME RECONSIDERED 329 from parents, and any parent bearing it will hand it down to children of the same sex. Given the nature of witches, it might be inferred that once one witch has been located, this status must also be attributed to all members of the same sex in that family line. However, the Azande do not do so. When the suggestion was made to them that they should make this inference, the Azande saw the sense in it but did not accept that this conclusion had to be drawn. Some analysts have suggested that this demonstrates a disinterest in logic on the part of the Azande, particularly in a logical inference that might problematize the institution by creating too many witch attributions. Bloor believes that such analysts are wrong. He states that the apparent logical contradiction inherent in this institution has no force, because it can always be “negotiated” away (Bloor 1991, 141). For example, the Azande sometimes argue that just because a person has witchcraft substance, it does not mean that the person is a witch. Rather, witch-hood is a potential that may or may nor be actualized, and only in the former case is a person to be ascribed the status witch. As Bloor (1991) states, “Logic poses no threat to the institution of witchcraft, for one piece of logic can always be met by another” (p. 141). To Bloor’s mind, this illustrates the general point that actors’ support for an institutionalized mode of classification has nothing to do with a reasoned analysis of the content of that institution. Arguments about the validity of an institution are always indecisive and cannot produce a rational conviction one way or the other. Importantly,Bloor argues that it is our commitment to an institution that leads us to defend it as “reasonable” rather than our belief in the “reasonable” nature of the institution that makes us committed to it (Bloor 1991, 143). The same point applies to the power of arguments directed against institutions. Bloor (1991) argues that “in so far as we feel the force of . . . logical inferences it is because we are already critical of the institutions” (p. 143). Although this argument is not couched in terms of interests, it clearly fits with the SP model to “explain” an actor’s commitment to an institution in these terms. Furthermore, it illustrates the way in which the SPers view social factors, such as commitment to an institution, as entirely independent from a reasoned and knowledgeable analysis of the institution. Having an interest in defending an institution has nothing to do with reasoned arguments for justifying the institution. 330 PHILOSOPHY OF THE SOCIAL SCIENCES / September 2003

The problem with this separation arises when one tries to specify what it means for an actor to have an interest in an institution. Draw- ing on the preliminary remarks above, it means that through prediction and control, an institution successfully produces some desired result or benefit for the actor. To continue receiving this benefit, the actor supports the institution and its continuation. Likewise, if actors have an interest in altering institutional arrangements, it is because they believe that their goals will be better served in a new configura- tion. Locating an interest thus means identifying the structure of the institution and how it produces and distributes the resources (benefits) that are pursued by actors in their interested activity. This means that actors have interests because of the effects or outcomes that an institution can produce for them. However, as we have seen, no outcomes can be achieved unproblematically. An institution may be more or less successful in producing the resources desired by actors, depending on the instrumental adequacy of the knowledge embodied in it. Importantly,this suggests that debate about the validity of an institution is a reflection upon its practical adequacy for pro- viding benefits to the actors involved. It is not a form of sophistry in which a claim of institutional validity can be defended whatever the evidence, but an argument about the instrumental success of an institution from the location of particular actors. Actors’ interests, on the one hand, and knowledge of the adequacy or validity of an institution, on the other, should thus be seen as intrinsically connected. Actors do not have interests in defending or attack- ing an institution regardless of its instrumental success in producing resources for them. Rather, they have interests only because of the success or otherwise of an institution. It is incoherent for analysts to attribute interests to actors that are not located in relation to an institutional structure of knowledge. Such attributions fail to explain how the resources to which that interest is oriented are produced and thus what knowledgeable activities the institution embodies in order to make a certain interest possible. It cannot be that actors have an interest in supporting or challenging an institution regardless of the instrumental adequacy of that institution in dealing with the material world. When interests are reconnected with knowledgeable institutional frameworks, they are subject to alteration and development in the light of changes in assessments of the instrumental adequacy of understandings. As an example of this we can return once more to the debate between liberal environmentalists and conservative Kemp / STRONG PROGRAMME RECONSIDERED 331 hereditarians. It is possible, as Barnes suggests, that evidence will come in that supports the environmentalist account of behavior rather than the hereditarian theory. As this evidence accrues, two shifts may happen. First, conservatives may decide that it is no longer in their interest to defend their social and political resources using a theory that is shown to be weak in its account of human behavior. If the theory is used to justify existing status and influence, its explanatory paucity may undermine such justifications, and conservatives will wish to look for these elsewhere. Their interest in supporting hereditarianism will alter as this comes to undermine their political claims. Second, (some) conservatives may change their political affiliation on the basis that hereditarian claims are no longer defensible. If the greater adequacy of an environmentalist account has been established, then this suggests that existing identities and their associated inequalities are not natural but can be modified. Conservatives may then come to believe that it is in their interest to participate in a society that removes inequalities and actualizes the possibilities of all actors. That is to say, they may decide that their interest in maintaining existing status and influence while living in a society that restricts human possibilities is less than their interest in bringing about an egalitarian society with fully realized human beings. Of course, such a serious alteration in political ideology is unlikely to be brought about by a shift in one specific debate. However, a number of such shifts may make a change in interest a plausible outcome.16 To take another example, we can consider the case of “narrow professional vested interests,” which are a standard kind of interest referred to in SP accounts (see, e.g., Barnes 1982, 114-18; Bloor forthcoming). A typical account of professional vested interests suggests that scientists seek to defend a particular theory or practice because of the resources of various kinds (status, financial, technical) that accrue to them when this theory is sanctioned by the larger collective. To gain such resources, scientists are taken to have an interest in defending their particular procedures and knowledgeability regardless of how successful these procedures actually are. Or to put it in SP terms, a group of scientists can always give reasons for the claim that their work is successful.17 To respond to such an account, it is necessary to reiterate that procedures, and the knowledge that they embody, cannot always be made to appear successful at dealing with a given subject matter. When comparing competing approaches, it is often possible to decide 332 PHILOSOPHY OF THE SOCIAL SCIENCES / September 2003 which is better at handling the phenomenon in question. If a theory has to resort to the kind of protectionism discussed by Barnes, it is necessarily limiting the scope of its application, and competing theories may exist that can handle the anomalies that it does not deal with. If so, judgments about relative success can be made. Of course, scientists within a research group wish to present their own approach in the best possible light so that they can gain the resources at stake in the scientific enterprise. However, it will not be in their long-term interest to continue defending a procedure or theory that is manifestly less successful than that of its rivals. In such a case, their interest in defending their theory diminishes, even when it has well-established successes behind it. When resources come to be allo- cated to more practically successful theories, there is no longer an interest in defending the theory in question. This kind of shift is part of the dynamic of science. Such points bring us on to the territory explored by Andrew Pickering (1995) in The Mangle of Practice. Pickering emphasizes that there can be no such thing as a “purely” social interest that stands out- side of the practical success or otherwise of scientists’ interactions with the material world.18 He argues that goals and interests are reconfigured as investigation continues and are interrelated with the outcome of scientific work and its more or less successful character. Pickering then uses these arguments to criticize the SPers for their lack of concern with how practical activity can generate a change in interests. He states,

The tendency is to write as if the substantive interests of actors were present and identifiable in advance of particular passages of practice, setting them in motion and structuring outcomes without being them- selves at stake. (Pickering 1995, 64)

Bloor has responded to these arguments, suggesting that they mis- represent the SP position. To show that the SP can deal with the practical and reconstructible nature of interests, he considers how we might analyze the development of a scientific paradigm (Bloor forthcoming). Bloor asks us to imagine that this paradigm is becoming “pro- gressively consolidated.” When this occurs, it is likely that a diver- gence of interest will result between those who founded the paradigm and those who adopted it later in order to develop and articulate its details. The founders will have an interest in making sure that their work continues to be used in the development of the paradigm. Kemp / STRONG PROGRAMME RECONSIDERED 333

Newer recruits, however, will see their interests as being served by breaking the existing paradigm in order to search for a radical new approach, of which they would be recognized as founders. The con- solidation of a paradigm thus results in a shift in interests for those involved. Bloor is correct to assert that this kind of SP analysis is a dynamic one, in that it does not rely on social interests remaining unchanged. However, the important issue here is not simply whether interests change or not but whether these changes are adequately theorized. Bloor’s example, in keeping with the SP approach, analyses shifts in interests as if they could be located without consideration of the adequacy of the scientific activity they are bound up with. Questions of adequacy are relevant, however, to understanding the interests of the actors involved. If, for example, the development of a paradigm along traditional lines continues to provide new insights and illumination, then it is in scientists’ interests to keep working along these lines. Likewise, if it appears that such lines are being exhausted, it is in scientists’ interests to break away from the tradition. Without an analysis of the paradigm’s success, however, such interests cannot be correctly identified. The implication of Bloor’s analysis is that it is always in the interests of actors to be “founders” of a paradigm. This seems funda- mentally implausible and implies that novel analysis will always be rewarded by the scientific community, no matter how successful it is in accounting for phenomena. Surely it is more likely that credit is given to successful new developments, whether this involves found- ing a paradigm or developing it. Bloor’s claim that later generations have an interest in rejecting a paradigm is only defensible in cases where a paradigm’s insights are close to exhaustion and a successful alternative can be developed. However, neither of these factors can be incorporated into the SP account.

CONCLUSION

The article has argued that the SPers’ account of scientific knowledge needs to be reappraised in light of their avowed intention to avoid dichotomizing the influence of social and instrumental factors in the development of knowledge. To some readers, the care taken to precisely locate the aims and parameters of the SP account may have seemed excessive, given that the final judgment offered is similar to 334 PHILOSOPHY OF THE SOCIAL SCIENCES / September 2003 that of many other critics: that the SP approach does not successfully account for the dynamics of scientific knowledge. However, unlike other writers (Archer 1988; Bhaskar 1986; Hollis 1982), I have not suggested that this licenses a dualistic account of knowledge that separates the influence of social interests from considerations of adequacy. Rather, I have suggested that a more consistent monism would provide an appropriate resolution to the difficulties found in the SP approach. Given the critical orientation of this article, the parameters of such a monism were only gestured at rather than fully developed. However, I believe that a more extensive monistic analysis would offer a coherent theory of the relation between social interests and the adequacy of knowledge, as well as avoiding the difficulties inherent in dualistic approaches (for an extensive critique, see Holmwood and Stewart 1991). This would not only be a gain at the theoretical level but would contribute to the production of more coherent empirical analyses, which is a goal that should surely be strived for by sociological theorists.

NOTES

1. A recent example of this kind of misunderstanding can be found in Philip Sullivan’s critique of Donald MacKenzie’s Strong Programme (SP)–inspired analysis of the development of statistics in Britain (Sullivan 1998; the work he discusses is Mac- Kenzie 1981). Sullivan believes that MacKenzie emphasizes social factors at the expense of technical ones (Sullivan 1998, 91), missing the point that SP analyses do not view these factors as having separate and competing logics (MacKenzie 1998). 2. I use the terms classification and categorization as synonyms throughout this article. 3. The meaning finitist approach and its sociological development are laid out in a number of different places, including Barnes (1981a); Barnes (1982); Barnes, Bloor, and Henry (1996); Bloor (1997). The essentials of the position are present in all of these expo- sitions, with no significant divergences. 4. Writers defending a “natural kind” approach argue that categories can be given determinacy if investigators locate an attribute (or set of attributes) that all and only members of this category possess (see, e.g., Forge 1996). However, this claim is vulnera- ble to the form of argument given above. No definitive decision can be made as to whether a particular has the relevant attribute because this in itself requires a judgment of similarity or difference that is ultimately problematic. 5. Scientific concepts frequently refer to processes rather than objects, but this causes no problems for the SP approach. Process concepts are learned through ostension and analogy with other inductively produced concepts, whereby initial commonsense experiences are refined and reflected upon to produce the more technical understandings in question (Barnes, Bloor, and Henry 1996, 60-61). Kemp / STRONG PROGRAMME RECONSIDERED 335

6. Bloor’s understanding of theory assessment as internal, and focusing on coherence, is shared with postpositivist philosophers of science such as Kuhn (1970) and Lakatos (1978). 7. For example, a generalization that predicts connections between two categories such as “geese eat grass” (Barnes 1982, 74-75). 8. In a debate with Woolgar over the status of interests, Barnes (1981b) makes a similar remark: “Almost everyone who accepts the Duhem-Quine hypothesis will rec- ognize that any theory can be maintained compatible with any findings by appropriate strategies of application and interpretation, and that the strategies involved are just those which maintain our actual accepted theories as our accepted theories” (p. 493). 9. Of course, the determinacy here cannot be specified with absolute precision. The precise level of correlation required can be debated in relation to other similar research, and so on. However, to acknowledge a certain amount of flexibility is not to view this as total. Furthermore, the specificity of one’s interest gives more determinacy to what counts as “cancer” and a “mobile phone” than the total formal openness that exists when no particular concern is involved. 10. Logically speaking, the converse claim is equally plausible. One can reasonably defend the claim that mobile phone use does not cause cancer without ever running into problems with existing or future experience. 11. See Barnes (1982, 100) for the claim that anomalies are not forced upon theories but are taken to be such for other reasons. In making this argument, the SPers diverge from writers such as Lakatos (1978) who argue that anomalies are not optional and demonstrate flaws in existing theories. For Lakatos, the existence of an anomaly does not necessitate the abandonment of a theory, but it does show that reconstruction of some aspect of the theoretical framework is necessary. 12. Furthermore, the required shift has not been on the periphery of the theory, as Barnes suggests, but very much relates to its core premises. 13. I have followed Barnes’s usage here, in which fish and animal are alternative and exclusive classifications. However, it should be noted that fish are standardly considered to be a subcategory of animals. Barnes’s arguments make more sense if we take him to mean mammal instead of animal. I thank Maureen O’Malley for pointing this out to me. 14. Barnes makes similar moves when discussing how actors relate to earlier classifications. His argument is that these classifications come with no particular goals and interests attached but are available to actors as a “resource for those willing to take them as a resource” (Barnes 1982, 113). I would argue that this is somewhat misguided. After all, those classifications were developed for some purpose and were more or less successful in achieving it. Any actors who wish to achieve a purpose to which the classification relates can only choose to ignore it at the peril of not successfully achieving their goals. 15. It is worth noting that I have not called upon a theory of natural kinds to critique the SP approach (for natural kind accounts, see Bhaskar 1997; Kripke 1980). Such theories produce difficulties of their own (see for discussion Shapere 1984), and I would suggest that the instrumentalist position offered here is sufficient for understanding the dynamics of categorization. 16. The connection of interests to instrumental adequacy thus provides a dynamic analysis that can comprehend the transformation of interests and thus interest groups. This goes some way to answering the criticisms of writers such as Steve Woolgar and 336 PHILOSOPHY OF THE SOCIAL SCIENCES / September 2003

Brian Wynne, who claim that interest explanations fail to account for the capacity of agents to transform possibilities (see Woolgar 1981, 374-75; Wynne 1992, 577). 17. The idea that scientists can always reasonably claim their own approach to be successful is commonly found within studies in the sociology of scientific knowledge, as it removes the possibility that success may be an explanation for the triumph of one theory over another. This makes room for other forms of account, whether they be interest explanations, micropolitical explanations (e.g. Collins 1985), or explanations that resist giving an explanation (e.g., Ashmore 1993). 18. Pickering’s (1995) arguments about interests are derived from the actor-network approach created by Latour and Callon (see, e.g., Latour 1987; Callon 1986). For a recent exchange between Bloor and Latour, see Bloor (1999a, 1999b) and Latour (1999).

REFERENCES

Archer, Margaret. 1988. Culture and agency: The place of culture in social theory. Cam- bridge, UK: Cambridge University Press. Archer, Margaret. 1995. Realist social theory: The morphogenetic approach. Cambridge, UK: Cambridge University Press. Ashmore, Malcolm. 1993. The theatre of the blind: Starring a Promethean prankster, a phoney phenomenon, a prism, a pocket, and a piece of wood. Social Studies of Science 23:67-106. Barnes, Barry. 1974. Scientific knowledge and sociological theory. London: Routledge and Kegan Paul. Barnes, Barry. 1977. Interests and the growth of knowledge. London: Routledge and Kegan Paul. Barnes, Barry. 1981a. On the conventional character of knowledge and cognition. Phi- losophy of the Social Sciences 11:303-33. Barnes, Barry. 1981b. On the “hows” and “whys” of cultural change (response to Woolgar). Social Studies of Science 11:481-98. Barnes, Barry. 1982. T.S. Kuhn and social science . London: Macmillan. Barnes, Barry,David Bloor, and John Henry.1996. Scientific knowledge: Asociological analysis. London: Athlone. Bhaskar, Roy. 1986. Scientific realism and human emancipation. London: Verso. Bhaskar, Roy. 1997. A realist theory of science, 2d ed. London: Verso. Bhaskar, Roy.1998. The possibility of naturalism: A philosophical critique of the contemporary human sciences, 3d ed. London: Routledge. Bloor, David. 1976. Knowledge and social imagery, 1st ed. London: Routledge and Kegan Paul. Bloor, David. 1991. Knowledge and social imagery, 2d ed. Chicago: University of Chicago Press. Bloor, David. 1996. Idealism and the sociology of knowledge. Social Studies of Science 26:839-56. Bloor, David. 1997. Wittgenstein, rules and institutions. London: Routledge. Bloor, David. 1999a. Anti-Latour. Studies in the History and Philosophy of Science 30 (1): 81- 112. Bloor, David. 1999b. Reply to Bruno Latour. Studies in the History and Philosophy of Sci- ence 30 (1): 131-36. Kemp / STRONG PROGRAMME RECONSIDERED 337

Bloor, David. Forthcoming. The sociology of scientific knowledge. In Handbook of Epis- temology, edited by I. Niiniluoto, M. Sintonen, and J. Wolenski.Dordrecht: Kluwer. Callon, Michel. 1986. Some elements of a sociology of translation: Domestication of the scallops and the fisherman of St Brieuc Bay.In Power, action and belief: A new sociology of knowledge?, edited by John Law. Sociological Review Monograph 32. London: Routledge. Collins, Harry. 1985. Changing order: Replication and induction in scientific practice. Lon- don: Sage Ltd. Evans-Pritchard, E. E. 1937. Witchcraft, oracles and magic among the Azande. Oxford, UK: Oxford University Press. Forge, John. 1996. Untitled contribution to review symposium “Finitism Interruptus? ‘Interests’ and the Foundations of SSK.” Metascience 11:34-41. Hesse, Mary. 1974. The structure of scientific inference. London: Macmillan. Hollis, M. 1982. The social destruction of reality. In Rationality and Relativism, edited by Martin Hollis and Steven Lukes. Oxford, UK: Basil Blackwell. Holmwood, John, and Alexander Stewart. 1991. Explanation and social theory. London: Macmillan. Keat, Russell. 1989. Relativism, value-freedom, and the sociology of science. In Relativ- ism: Interpretation and confrontation, edited by Michael Krausz. Notre Dame, IN: Uni- versity of Notre Dame Press. Kripke, Saul A. 1980. Naming and necessity. Oxford, UK: Basil Blackwell. Kuhn, Thomas S. 1970. The structure of scientific revolutions, 2d ed. Chicago: University of Chicago Press. Lakatos, Imre. 1978. The methodology of scientific research programmes: Philosophical Papers Volume 1, edited by J. Worrall and G. Currie. Cambridge, UK: Cambridge University Press. Latour, Bruno. 1987. Science in action: How to follow scientists and engineers through society. Milton Keynes: Open University Press. Latour, Bruno. 1999. For David Bloor...andBeyond: A reply to David Bloor’s “Anti- Latour.” Studies in the History and Philosophy of Science 30 (1): 113-29. MacIntyre, Alasdair. 1970. Is understanding religion compatible with believing? In Rationality, edited by Bryan R. Wilson. Oxford, UK: Basil Blackwell. MacKenzie, Donald. 1981. Statistics in Britain, 1865-1930: The social construction of scientific knowledge. Edinburgh, UK: Edinburgh University Press. MacKenzie, Donald. 1998. The zero sum assumption. Social Studies of Science 29 (2): 223- 34. Manicas, Peter, and Alan Rosenberg. 1985. Naturalism, epistemological individualism and “the Strong Programme” in the sociology of knowledge. Journal for the Theory of Social Behaviour 15 (1): 76-101. Manicas, Peter, and Alan Rosenberg. 1988. The sociology of scientific knowledge: Can we ever get it straight? Journal for the Theory of Social Behaviour 18 (1): 51-76. Mermin, N. David. 1998. The science of science: A physicist reads Barnes, Bloor and Henry. Social Studies of Science 28 (4): 603-23. Miller, David P. 1996. Untitled contribution to Review Symposium “Finitism Interrup- tus? ‘Interests’ and the Foundations of SSK.” Metascience 11:22-30. Newton-Smith, W. 1981. The rationality of science. London: Routledge. Pels, Dick. 1996. The politics of symmetry. Social Studies of Science 26:277-304. Pickering, Andrew. 1995. The mangle of practice: Time, agency and science. Chicago: Uni- versity of Chicago Press. 338 PHILOSOPHY OF THE SOCIAL SCIENCES / September 2003

Roth, Paul. 1994. What does the sociology of scientific knowledge explain? Or, when epistemological chickens come home to roost. History of the Human Sciences 7 (1): 95- 108. Shapere, Dudley. 1984. Reason and the search for knowledge: Investigations in the philosophy of science. Lancaster, UK: D. Reidel. Slezak, Peter. 1994. Asecond look at David Bloor’s Knowledge and Social Imagery. Philoso- phy of the Social Sciences 24 (3): 336-61. Sullivan, Philip. 1998. An engineer dissects two case studies: Hayles on fluid mechanics and MacKenzie on statistics. In A house built on sand: Exposing postmodernist myths about science, edited by Noretta Koertge. Oxford, UK: Oxford University Press. Turner, Stephen. 1981. Interpretive charity, Durkheim, and the “Strong Programme” in the sociology of science. Philosophy of the Social Sciences 11:231-43. Winch, Peter. 1970. Understanding a primitive society.In Rationality, edited by Bryan R. Wilson. Oxford, UK: Basil Blackwell. Wittgenstein, L. 1968. Philosophical investigations, 3d ed. Oxford, UK: Basil Blackwell. Woolgar, Steve. 1981. Interests and explanation in the social studies of science. Social Studies of Science 11:365-94. Wynne, B. 1992. Representing policy constructions and interests in SSK. Social Studies of Science 22: 575-80.

Stephen Kemp is a British Academy postdoctoral fellow based at the University of Sus- sex, United Kingdom. His areas of interest include philosophy of social science, science and technology studies, and conceptions of criticism in social theory.