Ronald N. Giere (1994) Cognitive Science and the Philosophy of Science:. Psycoloquy: 5(62) Scientific Cognition (10)

Volume: 5 (next, prev) Issue: 62 (next, prev) Article: 10 (next prev first) Alternate versions: ASCII Summary
PSYCOLOQUY (ISSN 1055-0143) is sponsored by the American Psychological Association (APA).
Psycoloquy 5(62): Cognitive Science and the Philosophy of Science:

Reply to Bookstein, Catania, Hardcastle, Roitblat,
Shafto, Slater & van Brakel on Scientific-Cognition

Ronald N. Giere
Department of Philosophy and
Center for Philosophy of Science
University of Minnesota
Minneapolis, MN 55455


This reply to reviewers of Cognitive Models of Science (Giere, 1992) focuses on issues raised concerning the overall project. I conclude that the cognitive sciences provide useful resources for a naturalistic philosophy of science. The noted disunity in the cognitive sciences places limitations on how unified a naturalistic philosophy of science could be, but does not detract from the overall project.


Cognitive science, philosophy of science, cognitive models, artificial intelligence, computer science, cognitve neuroscience.
1. Having organized and edited a volume on the role of cognitive models in the philosophy of science (Giere, 1992), I am pleased that a half dozen people have taken the time to review it for PSYCOLOQUY. In this reply I will focus mainly on issues raised concerning the overall project, leaving it to individual contributors to respond to comments on their particular contributions. It should not be assumed that the contributors necessarily share my view of the project as a whole.

2. There are many ways to conceive of relations between the cognitive sciences and the philosophy of science. Most reviewers had some questions about the conception of this connection underlying the volume. Some had their own ideas of what the connection ought to be. Not surprisingly, how one views possible connections depends crucially on how one views the two subjects being connected.

3. The philosophy of science we know as Logical Empiricism (or Logical Positivism) had its origins in a movement for scientific philosophy in the 1920s and early 1930s, mainly in the German speaking areas of Europe. The scientific philosophers rejected all claims to synthetic, a priori knowledge. The only knowledge of the real world, they held, was that provided by scientific inquiry. Scientific philosophy itself consists primarily of the LOGICAL ANALYSIS of scientific concepts and theories, an activity whose status is like that of logic or mathematics. Its conclusions are purely formal and carry no empirical content. Later philosophers of science sometimes portray their activities more softly as CONCEPTUAL ANALYSIS, but this remains autonomous from science itself. Philosophy of science itself is thus logically distinct from science even if it consists solely in the analysis of scientific concepts and theories. Those who practice the logical or conceptual analysis of the sciences often speak of "the philosophical foundations of ...." as in "the philosophical foundations of physics". Some reviewers, for example, van Brakel (1994), seem to think that the proper relationship between philosophy of science and cognitive science would be for philosophers to be investigating the "philosophical foundations of cognitive science" in one of the modes just mentioned.

4. The vision behind Cognitive Models of Science, by contrast, is that there are no separate "philosophical foundations" for any science. There are, of course, deep theoretical problems. But these are always internal to the science in question. There is no way to get at these theoretical problems from the "outside". In particular, philosophers have no special methods for dealing with such problems that are not available to scientists themselves. If one wants to deal seriously with theoretical scientific problems, one must do so as a scientist. This is in practice almost impossible for anyone not trained and socialized in the science in question, and not recognized as a member of that particular scientific community. Book learning is not enough.

5. What then is the positive conception of the philosophy of science behind the project of Cognitive Models of Science? And how might such a philosophy of science forge significant connections with the cognitive sciences as they now exist?

6. The overall conception of the philosophy of science is NATURALISTIC. That is, the claims made by philosophers of science must be empirical claims about science, and thus subject ultimately to appropriate empirical controls. However, the subject matter of the philosophy of science is not that of other sciences such as physics or biology, but the phenomenon of science itself as a human practice. In their general form, its questions are of the following sort: How does science work? What sorts of knowledge does it produce? What forms does this knowledge take? How does it succeed? Why does it often fail?

7. Such a naturalistic philosophy of science makes no initial presumptions that the answers to such questions are the same for all sciences or for all times. They might, for example, be one thing for physics in the seventeenth century and something quite different for biology in the twentieth. Only historical research can reveal the answers to such questions. But one has to ask the right historical questions.

8. Such questions cannot be answered by a formal analysis of the final claims made by scientists. Adequate answers require an understanding of the PROCESS through which the claims made come to be the established claims. And since this process involves human scientists, it would seem that the cognitive capacities and acquired knowledge of scientists is relevant to any accounts one gives. Finally, since our best scientific accounts of the cognitive abilities of human agents are those being developed in the cognitive sciences, the relevance of research in the cognitive sciences seems assured. Only if it were possible fully to abstract from the cognitive characteristics of scientists would the cognitive sciences be rendered irrelevant to the project. So the presumption is that one cannot both abstract from the cognitive abilities of scientists and answer the kinds of questions posed above.

9. From the perspective of a naturalistic philosophy of science, then, the Logical Empiricists both narrowed the questions to be asked and assumed that a formal analysis could provided appropriate answers. They were mistaken on both counts. Carnap (1950), for example, sought to develop a purely formal inductive logic that would provide an index of the rational acceptability of a hypothesis relative to given data. If that project were successful, then no empirical theory of human judgment would be relevant to the understanding of science as a rational enterprise. The fact that it now seems impossible to carry out any such program provides motivation for philosophers of science to bring human judgment explicitly into the picture.

10. So what are the assumed connections between the cognitive sciences and the philosophy of science conceived as a naturalistic enterprise? My introduction and precis (Giere, 1992; 1993) unfortunately provide grounds for both a stronger and a weaker interpretation of this connection. As the stronger interpretation is suggested in the very first paragraph, that is the one that drew most comment. But the weaker interpretation is closer to my own view.

11. In the opening paragraph I said: "The cognitive sciences have reached a sufficient state of maturity that they can now provide a valuable resource for philosophers of science who are developing general theories of science as a human activity. The hope is that the cognitive sciences might come to play the sort of role that formal logic played for Logical Empiricism or that history of science played for the historical school within the philosophy of science." For the cognitive sciences truly to play the role suggested in the second sentence would mean that the philosophy of science would be an application of cognitive science and thus a sub-field within cognitive science -- just as the Logical Empiricists made the philosophy of science a branch of applied logic. Some reviewers (Slater, 1994; Hardcastle, 1994) argue that it is neither possible nor desirable for the cognitive sciences to play so substantial role for the philosophy of science. I agree. I am sorry to have invited that interpretation.

12. The previous sentence speaks of the cognitive sciences providing "a valuable resource" for the philosophy of science. This suggests that there may be other resources as well. And this suggestion is developed in the opening sentences of the penultimate section of the introduction where I wrote: "No one thinks that a cognitive theory of science could be a complete theory of science. The cognitive activities of scientists are embedded in a social fabric whose contribution to the course of scientific development may be as great as that of the cognitive interactions between scientists and the natural world. Thus cognitive models of science need to be supplemented with social models." This seems to rule out the possibility of including the philosophy of science wholly within the cognitive sciences.

13. The only way to sustain the stronger interpretation would be to insist that the social reduces to the cognitive -- or, more precisely, that social interactions can be reduced without remainder to cognitive interactions among cognitive agents. Slater (1994) suggests that some such reduction may be being assumed by the project of Cognitive Models of Science. It fact, I regard it as very doubtful that such a reduction is IN PRINCIPLE possible. The question, however, is moot. The cognitive sciences and the social sciences are IN PRACTICE now sufficiently disjoint that, from the standpoint of a developing naturalistic philosophy of science, they represent separate disciplines providing quite different resources.

14. Nearly every reviewer mentions the lack of unity among the cognitive sciences as a problem for the use of the cognitive sciences by philosophers of science. But this admitted lack of unity is only a problem if one has in mind something like the relationship between logic and Logical Empiricism. For a naturalistic philosophy of science which draws on diverse resources, the lack of unity within cognitive science is no problem. This point is reinforced by realizing that a satisfactory account of how any particular scientific development came to be will have to draw on scientific knowledge in the field being studied. How, for example, could one explain how Watson and Crick discovered the helical structure of DNA without talking about phosphates, proteins, the Chargaff ratios, and so on? So chemistry provides a resource for the study of this episode in the recent history of science. But so do results on the role of analogy or visual imagery in cognitive psychology. That chemistry and cognitive psychology are very different is no deterrent to their both being used in such a study. It is, therefore, no obstacle to a naturalistic philosophy of science that cognitive neurobiology and artificial intelligence are so different.

15. It is a problem that approaches found within the cognitive sciences are sometimes not only different, but actually conflicting. Then one can get conflicting accounts of scientific activities depending on where in the cognitive sciences one looks for resources. This may be the situation which most concerns the reviewers.

16. To some extent the problem of conflicting resources is not at all specific to a naturalistic philosophy of science. In any science, however unified, one is constantly forced to use results of other research in one's own inquiries. The closer to the research frontier one works, the more likely it is that there will exist conflicting results from which to choose. In this case one can only use one's own best judgment about what, or whom, to rely on, and hope one's own results are not too sensitive to the choice made. If they are, and one has chosen badly, then that is the risk one takes. One's own research is always hostage to the resources one employs. In a relatively unified field, however, one can expect that things will sort themselves out in the not too distant future. And that will reduce the risks for other researchers in the field.

17. The worry for a naturalistic philosophy of science is that cognitive science could remain forever a multidisciplinary undertaking. The conflicts one now finds may never get resolved, and neither will the conflicting naturalistic accounts of science which employ these conflicting resources. A prime example is the conflict between cognitive and behavioral analysis in psychology. Among contributors to Cognitive Models of Science, only Houts and Haddock (1992) clearly favor behavioral analyses, as does one reviewer (Catania, 1994). My own sympathies are clearly with the cognitivists. Indeed, I would still maintain that psychology did not provide a sufficient resource for a naturalistic philosophy of science before cognitivism came on the scene. The only prior possibility was not behaviorism, but gestalt psychology and the work of Piaget. Here, as in other sciences, one must make one's choices as best one can, although in this case the prospects for near-term resolution seem bleak. For myself (see Giere, 1994), I continue to find my cognitive resources in the cognitivist literature, for example, the literature on concepts and categories noted by Roitblat (1994). I believe, of course, that the future will confirm this choice, but I could be wrong.

18. What about the cognitive study of science as carried out within the cognitive sciences? What are the implications of the disunity of the cognitive science for this enterprise? This question deserves separate consideration because roughly half of the contributors to Cognitive Models of Science are primarily identified with this cause. It seems to me, however, that the situation here is fundamentally the same as for a naturalistic philosophy of science, except much less severe. It is less severe because, besides cognitive models, a naturalistic philosophy of science also needs social models, and the diversity among approaches to social models may be even greater than the diversity among approaches to cognitive models.

19. In the end, therefore, I agree with those reviewers, particularly Hardcastle (1994) and Slater (1994), who insist that unity in the cognitive study of science, and, by implication, in a naturalistic philosophy of science, is not to be had. Close examination reveals that even the supposedly most unified of sciences, like physics, are not nearly so unified as is often supposed. So lack of unity may not be so bad after all. It might even turn out to be desirable.

20. I turn now to some more specific criticisms of Cognitive Models of Science which concern topics not covered, organization, audience, and implications for other areas of study.

21. Among topics not covered in Cognitive Models of Science are procedural methods (Shafto, 1994) and the use of geometrically organized data (Bookstein, 1993) for evaluating hypotheses. There is no doubt that these subjects can provide valuable resources for a naturalistic philosophy of science. That such topics do not show up in Cognitive Models of Science has nothing to do with their intrinsic value, but only reflects how the participants were selected and the topics organized. The initial organization was by people rather than topics. Participants were selected on several bases. One was their previous involvement in the project of using the methods and results of the cognitive sciences to investigate how science is actually done. Another was a recognized level of prior engagement with topics in the philosophy of science. Finally, it was desirable to end up with a fair representation of people with different perspectives. This process did not yield any papers on procedural methods or geometrically organized data. Ironically, the one person present who had written on these topics from a cognitive perspective was myself. I invite those with such interests to examine chapters six (Scientific Judgment) and seven (Models and Experiments) of Explaining Science (Giere, 1988). When I organized the workshop on Cognitive Models of Science I had just finished my own book and had nothing new to present.

22. Underlying the selection of contributors was a sense of the intended audience. The primary audience was philosophers of science. This is appropriate for a series titled Studies in the Philosophy of Science. One of the major aims of the project is to convince philosophers of science that the cognitive sciences have something to offer to the philosophy of science, at least a naturalistically oriented philosophy of science. It also furthers the general project of developing a naturalistic philosophy of science. The next most important audience consists of historians and sociologists of science. The hope here is that both can come to see the cognitive study of science as contributing to their common interest in understanding the practice of science as an influential cultural activity. Next, of course, are cognitive scientists directly engaged in the cognitive study of science. But it was not expected that this volume would produce any surprises for such an audience.

23. The audiences just mentioned are part of a loose disciplinary grouping that now typically goes under the title Science Studies (or Science and Technology Studies -- STS). Adopting Shafto's (1994) distinction, these are the "insiders" in Science Studies. But relative to most other scientists, for example, physicists and biologists, people in Science Studies are "outsiders". But there are audiences for Cognitive Models of Science outside of the science studies community: other scientists, science educators, and perhaps even science journalists.

24. Contrary to the supposition of some reviewers, for example, Shafto (1994), it was not one of the goals of the project to convince other scientists that they had something to learn from new approaches in science studies. How useful any work in science studies may be for other scientists depends on one's expectations. If one expects methods that can be achieve results in the laboratory tomorrow, one will be disappointed. But if one expects to gain some understanding, perhaps even insight, into the workings of science in general, there may be much to be learned. Everyone has their own theories about what they do. Scientists may be a bit more self-conscious in their off-line theorizing about their own activities than many others. But that does not mean their own theories of science are necessarily good, or even useful ones. Insiders are not necessarily the best placed to acquire a general understanding of what they are doing. Acquiring better theories about how science works may thus be useful in helping scientists do better science. This result, however, can at best only be indirect, and depends on the particular circumstances of individual scientists.

25. It was not the explicit intent of the project to produce something useful for science educators. As it turned out, however, several of the contributors, particularly Carey, Chi, and Nersessian, have explicit interests in science education. It is thus not surprising that implications of their work for science education are relatively close to the surface. In general, one would expect that the better one's general theory of how science works, the more useful it should be for improving science education. Realizing this potential, however, will require the involvement of science educators. By themselves, members of the science studies community are not likely to have any more expertise in science education than any other university professors.

26. In sum, the cognitive sciences, and particularly the explicit study of science by cognitive scientists, provide a useful, even necessary, resource for a naturalistic philosophy of science. The lack of unity in the cognitive sciences themselves places limitations on how unified a naturalistic philosophy of science might be, but does not detract from the overall project. And the results of a naturalistic philosophy of science incorporating resources from the cognitive sciences may be useful to others, including other scientists and science educators.


Bookstein, F.L. (1993) Geometry as Cognition in the Natural Sciences. PSYCOLOQUY 4(65) scientific-cognition.2.bookstein.

Carnap, R. (1950) Logical Foundations of Probability, University of Chicago Press.

Catania, A.C. (1994) Information, Cognition and Behavior: Telling What They Want to Hear. PSYCOLOQUY 5(39) scientific-cognition.5.catania.

Giere, R.N. (1988) Explaining Science: A Cognitive Approach, University of Chicago Press.

Giere, R.N., ed. (1992) Cognitive Models of Science, Minnesota Studies in the Philosophy of Science, Vol. XV, University of Minnesota Press.

Giere, R.N. (1993) Precis of Cognitive Models of Science. PSYCOLOQUY 4(56) scientific-cognition.1.giere.

Giere, R.N. (1994) The Cognitive Structure of Scientific Theories. Philosophy of Science 61:276-96.

Hardcastle, G.L. (1994) Why Don't We Yet Have a Cognitive Science of Science? PSYCOLOQUY 5(43) scientific-cognition.6.hardcastle.

Houts, A.C. and Haddock, C.K. (1992) Answers to Philosophical and Sociological Uses of Psychologism in Science Studies: A Behavioral Psychology of Science. In: Cognitive Models of Science, ed. R.N. Giere, University of Minnesota Press.

Roitblat, H.L. (1994) A Representational View of Science. PSYCOLOQUY 5(44) scientific-cognition.7.roitblat.

Shafto, M.G. (1994) What Can Insiders Learn From Outsiders? PSYCOLOQUY 5(30) scientific-cognition.4.shafto.

Slater, C.W. (1994) Why We Should Not Rely on Cognitive Science to Unify Philosophy of Science. PSYCOLOQUY 5(45) scientific-cognition.8.slater.

van Brakel, J. (1994) Cognitive Scientism of Science. PSYCOLOQUY 5(07) scientific-cognition.3.vanbrakel.

Volume: 5 (next, prev) Issue: 62 (next, prev) Article: 10 (next prev first) Alternate versions: ASCII Summary