Method and theory in the study of avoidance.

Abstract

Two-factor theories of avoidance were conceived to explain responding in avoidance procedures that closely resemble the Pavlovian paradigm in superficial features, although differing in the fundamental contingency of reinforcement. Both typically involve an arbitrary conditioned stimulus and a trial-by-trial sequence of pairings between the conditioned and unconditioned stimuli. According to two-factor theory, the instrumental reinforcement of avoidance is based on the Pavlovian reinforcement of a drive state in the presence of the conditioned stimulus. It has been shown, however, that the presence of the conditioned stimulus is not necessary for the occurrence of avoidance responding. A procedure in which the sole effect of the avoidance response was a reduction in the average frequency of occurrence of an aversive electric shock proved to be fully adequate to maintain lever pressing in rats, thereby suggesting that not all avoidance requires two factors. Further experiments with various new procedures suggested that the conditioned stimulus may function as a discriminative stimulus for the avoidance response, rather than as a stimulus whose removal is inherently reinforcing, as two-factor theory requires. The conditioned reflex was to I. P. Pavlov (1928, pp. 59-60) the final answer to the problem of biological adaptation. As a mechanist, Pavlov sought a naturalistic explanation for everything an animal did, which had come to mean an explanation in terms of physical processes that could be isolated by the vivisectionist techniques of nineteenth-century physiology. But the behavior of many animals, for example, the dog, precluded any such simple machinery. Dogs clearly differed in what they did and seemed to know even though they might share virtually identical inheritances. The psyche of 1 Preparation of this paper, as well as the conduct of the previously unpublished experiments described herein, was supported by grants from the National Science Foundation to Harvard University. The author wishes to express thanks to P. N. Hineline for his generosity in allowing use of some of his as yet unpublished data and for help in formulating some of the notions here advanced. An early and much reduced version of this paper was presented at the 1966 meeting of the American Association for the Advancement of Science, Washington, D. C., as part of a symposium on Aversive Control. The author owes thanks to J. V. Brady for having organized the symposium and for inviting him to participate in it. To another of the participants, D. Anger, special thanks are owed for his vigorous and insightful criticisms of many of the author's theoretical ideas. the dog was, in other words, a sizable obstacle to the progress of a science of adapta