John J. Furedy (2000) Is Instrumentalism the Right Method for Psychological Science?. Psycoloquy: 11(066) Ai Cognitive Science (6)

Volume: 11 (next, prev) Issue: 066 (next, prev) Article: 6 (next prev first) Alternate versions: ASCII Summary
PSYCOLOQUY (ISSN 1055-0143) is sponsored by the American Psychological Association (APA).
Psycoloquy 11(066): Is Instrumentalism the Right Method for Psychological Science?

Commentary on Green on AI-Cognitive-Science

John J. Furedy
Department of Psychology
University of Toronto
Toronto, Ontario M5S 3G3



Green's assertion that "the computer model... lends itself to realism about mental states, whereas behaviorism encourages mental instrumentalism" is consistent with most accounts of the cognitive revolution in psychology, but here I contend that it is wrong. My arguments are based on distinctions between realism and instrumentalism, theories and models, and cognitive and non-cognitive psychological functions, as well as on an account of the S-S vs. S-R, Tolman vs. Hull learning theory arguments about what is learned (psychology's then central issue), culminating in the 1953 and 1952 papers, respectively, by Kendler and Ritchie. I conclude that it is not the behaviorist, but the instrumentalist tactic that is at the root of the problem, and that it is not only AI but the instrumentalist approach that is not the right method not only for cognitive science, but also for the science of psychology.


artificial intelligence, behaviorism, cognitive science, computationalism, Fodor, functionalism, Searle, Turing Machine, Turing Test.
1. Green's (1993/2000) assertion that "the computer model ... lends itself to realism about mental states, whereas behaviorism encourages mental instrumentalism" (para. 1) is consistent with most current accounts of the cognitive revolution in psychology, but I question this account. A realist approach would require a computer THEORY and not just MODEL of the mind (in an instrumentalist view, there is no clear distinction between theories and models, since the true/false category has been given up--see e.g. Furedy, 1988), and no such theory would ever be presented seriously, because it is patently false. It not only fails to account for mentation that is at least partly influenced by affective and connative factors, but it even ignores all cognition that is not computational (i.e., computer-like). For example, although computers can now play chess at a grand-master level, the mentation involved in computer chess is nothing like that involved in human chess. The former is purely computational, whereas the latter involves a combination of computation, perception, and intuition, as well as quite a lot of noncognitive factors like affect (i.e., feelings about the chess position which can exhibit varying degrees of beauty, feelings about oneself [e.g. fear of the blunder, inadequacy], the opponent [hostility, etc.]), and different degrees of connation (i.e., varying effort put into the game depending on one's physical conditioning, and the importance of the game). Moreover, to continue with this relatively trivial example, to illustrate the task of any adequate psychological explanation, there must be an account of all important cognitive AND non-cognitive phenomena if one wants to develop a science of psychology, rather than merely more computer models of the mind.

2. Contrary to current accounts of the cognitive revolution which Green appears to echo, the shift from (S-R) behaviorism to the (S-S) cognitive approach in psychology was not, primarily, in terms of the relevant evidence (which are realist grounds), but rather in terms of theoretical fruitfulness (which are instrumentalist grounds). Contemporary cognitivists like Segal and Lachman (1972) gave such an instrumentalist account (although they did not use the term) of the triumph of this modern cognitive approach, although the instrumentalist aspect of the approach was systematically put forward two decades before by Kendler (1952), who then was still defending the then dominant Hull-Spence S-R position.

3. As has been detailed elsewhere (Furedy & Riley, 1987), the context of Kendler's (1952) paper was learning theory, which occupied such a prominent position in experimental psychology that it was often dubbed as being equivalent to behavior theory, i.e., the theory of psychology. And the most fundamental question of that period was that of "what is learned?". The answer of the S-R theorists of the Hull-Spence school to this question was that all learning was S-R learning, although the explanatory S-R concepts included hypothetical entities like the fractional anticipatory goal response (see e.g. Spence, 1956, for a relatively recent formulation) which were not only not directly observable, but sometimes even dubbed as "incorporeal". The contrary S-S view of Tolman and his students was that sign-significate, propositional, S-S learning (e.g. "cognitive maps") also occurred, and they strove to gather evidence for this cognitive view by running experiments that showed phenomena such as latent learning to occur. It is important to note that these cognitive theorists were still behaviorists, as they agreed with their S-R opponents that the way to decide issues of this sort was through observation of behavior, often in the "little white test-tube" (Osgood, 1953), i.e. the laboratory-bred rat. And, of course, the fractional anticipatory goal response and the cognitive map are equally "insubstantial", the difference being that the latter S-S construct is propositional or cognitive, because it is not a category mistake to apply the notion of truth and falsity to it. It is, in other words, a construct that can be formulated propositionally, whereas the fractional anticipatory goal response cannot (see Furedy & Riley, 1987, for an elaboration of this propositionally-based distinction between cognitive and non-cognitive processes).

4. Kendler's (1952) paper, as suggested by the title, argued that by the standard of "modern methodology" (which was really the approach of instrumentalism), the S-R versus S-S debate was not over an empirical issue, and that fruitfulness (which at that time clearly favored the S-R camp) should decide the matter. Tolman's former student, Ritchie (1953), provided the reductio ad absurdum retort to Kendler's (1952) position. The explicit focus of this paper is the defence of the S-S cognitive view, but the mode of argument used clearly marks it as a realist refutation of Kendler's (1952) instrumentalism. The implication of Ritchie's (1953) humorous paper (the only funny paper in the history of this Journal, one likely reason why Ritchie lost the debate-- serious scientists do not make jokes) is that fundamental psychological issues should be decided (though never with absolute certainty, since all empirical assertions are open to error) on empirical grounds, more specifically in terms of observed behavior in controlled experiments designed to provide empirical arbitration between conflicting theories. Tolman and other cognitivists like Ritchie were behaviorists, although their brand of behaviorism was a cognitive or S-S one, in contrast to the S-R behaviorism of the Hull-Spence school.

5. Until Kendler (1952) supplied the sword of instrumentalism (what he called the principles of "modern methodology") to cut the Gordian knot of "what is learned?", the accepted approach of both S-R and S-S behaviorists was that of realism. That is, this and other issues in psychology were considered to be empirical issues about which genuine theories and not just models or analogies were to be constructed. And, consistent with the (implicit) instrumentalism of his 1952 article, Kendler later joined the cognitive "paradigm shift" which, as characterized by Segal and Lachman (1972), was a shift based on fruitfulness (of mainly the computer analogy for mind) rather than evidence. On the other hand, the realist-cognitivist Ritchie disappeared from experimental psychology and did not share in the "fruits" of the new cognitivism. So Kendler rather than Ritchie won the "hearts and minds" of most experimental psychologists not as regards the psychological S-R versus S-S issue, but rather as regards the philosophy-of-science issue between the instrumentalist and realist approaches.

6. It is the adoption of the instrumentalist approach, the substitution of models for theories that lead to putting the problems of psychology into a "black box". This is regardless of whether the box is not talked about at all (as Skinner recommends) or whether it is talked about only in terms of models, analogies, and metaphors, but never in terms of hypothetical processes that are assumed by the theory really to occur. So it is not the "behaviorist" (Green, para. 10), but the instrumentalist tactic which is the root of the problem. And it is not simply AI (which is just a particular version of cognitive instrumentalism) but the instrumentalist approach that is not the right method not just for cognitive science (and I am not at all sure whether this is an area that has a clear referent), but for also for the science of psychology.


Furedy, J. J. (1988). On the relevance of philosophy for psychological research: A preliminary analysis of some influences of Andersonian realism. Australian Journal of Psychology, 24, 93-100.

Furedy, J. J., & Riley, D. (1987). Human Pavlovian autonomic conditioning and the cognitive paradigm. In, G. Davey, G. (Ed). Cognitive Processes and Pavlovian Conditioning in Humans, Wiley, London.

Green, C.D. (2000) Is AI the Right Method for Cognitive Science? PSYCOLOQUY 11(061)

Kendler, H. H. (1952). What is learned?--A theoretical blind alley. Psychology Review, 59, 269-77.

Osgood, C.E. (1953). Method and Theory in Experimental Psychology. Oxford University Press, New York.

Ritchie, B. (1953). On the circumnavigation of cognition. Psychology Review, 60, 206-211.

Segal, E. M., & Lachman, R. (1972). Complex behavior or higher mental process: Is there a paradigm shift? American Psychologist, 27, 46-55.

Spence, K. W. (1956). Behavior Theory and Conditioning. Yale Univ. Press, New Haven, CT.

Volume: 11 (next, prev) Issue: 066 (next, prev) Article: 6 (next prev first) Alternate versions: ASCII Summary