Has the Wheel Turned Full Circle? Fifty Years of Learning Theory, 1946–1996

Abstract

Fifty years ago, learning theory, whose principles were derived from experiments on conditioning in animals, was a central focus of much of experimental psychology. But the cognitive revolution that swept through human experimental psychology in the 1960s, especially when it was taken up by many animal psychologists themselves, seemed to consign traditional learning theory to the scrap heap. Liberation from the shackles of old-fashioned behaviourism, however, should not be bought at the price of dismissing associative learning theory. Suitably modi ed and extended, associative theory is capable of explaining much of the behavior of animals and of pointing precisely to the ways in which that behaviour is complex. And under some circumstances, at least, the behaviour of people mirrors that of other animals and is equally amenable to an associative analysis. preface to Munn' s Handbook of psychological research in the rat, commended the volume as essential reading for undergraduate and graduate students as well as professional psychologists [who] will nd in these pages a treatment of some of the most important aspects of general modern scienti c psychology (Carmichael, 1950).And anyone who has read Hull' s or Skinner's autobiographies (Hull, 1952; Skinner required three volumes for his : Skinner, 1976: Skinner, , 1979: Skinner, , 1983 will not be left long in doubt about the value they placed on their work.If this was hubris, divine retribution was swift and harsh. The decline of animal learning theory from its pinnacle of importance was rapid and, at one time, seemed complete. Two sets of statistics give some measure of the change. In 1946, the American Psychological Association published two main journals devoted to reports of empirical research in mainstream experimental psychology, Journal of Experimental Psychology and Journal of Comparative Psychology. In that year, just over 20% of the articles appearing in the former, and 50% of those appearing in the latter were reports of experimental studies of animal learning (narrowly de ned-excluding, for example, research on animal behaviour, ethology, or physiological psychology, which made up the other 50% of the Journal of Comparative Psychology). Thirty years later, in 1974, the last year in which the Journal of Experimental Psychology was published as a single journal, less than 5% of its articles were on animal learning, and the contents of Journal of Comparative and Physiological Psychology, which had long since replaced Journal of Comparative Psychology, were now largely (over 75% ) devoted to physiological rather than comparative psychology.More recently, Thomas and Blackman (1991) reported that between 1977 and 1989 there had been a decline of over 20% in the number of British psychology departments with an animal laboratory. They also reported that although there had been a signi cant increase in the overall number of research students in psychology over this 12-year period, the number undertaking research requiring a Home Of ce licence had halved.What is t...

Associations and propositions: The case for a dual-process account of learning in humans

“…We took the view that the cognitive system dealt in propositional knowledge and would therefore have beliefs, and reasons for those beliefs that could be verbalized, whereas the associative system would extract the statistical regularities from experience. Subsequently, Mackintosh (1997), starting from a rather different position, also argued for the importance of associative processes in human learning.…”

Human place learning in a virtual Morris water task: some important constraints on the flexibility of place navigation

“…distal cues) and responses (e.g. approach behaviors) and it is assumed that place learning follows similar rules of acquisition to those operating in classical and instrumental conditioning [5,14]. Associative explanations of novel, or flexible, spatial behaviors are typically limited to those based upon stimulus generalization and do not involve appeals to representations such as a cognitive map, although spatial behaviors may come under the control of potentially complex stimulus relationships or configurations [30,35].…”

Discrimination and generalization along a simple dimension: Peak shift and rule-governed responding.

“…Hanson's pigeons, for instance, showed a sharp decline in responding for 530nm and shorter wavelengths. Mackintosh (1997; see also Wills and Mackintosh, 1998) has focussed attention on this observation in discussing relational learning in animals, as it suggests that discriminative behavior is often controlled by elementary physical stimulus properties rather than abstract relationships between stimuli, even when the stimuli are conceptually related in a very simple manner. The extreme stimulus values which elicit less responding are less similar to S+ in terms of their absolute physical properties, even though, in many instances they are easier to classify in terms of their abstract or symbolic relationship to the training stimuli.…”

“…Similarly, when the dimensional relationship between stimuli is masked by task requirements, the results appear to be more in favor of a peak shift effect. In an unpublished study, Aitken (1996; also reported briefly in Mackintosh, 1997) showed that a categorization task which normally produced monotonic gradients in response accuracy could be made to yield peak-shifted gradients by changing the task to one involving speeded, cued reactions, thereby making discrimination learning incidental to performance of the task. However, the question remains whether there is any evidence for associative processes influencing the classification of stimuli that are more simply related along a single physical dimension.…”