Application of a drive and an associative theory to the effects of CR-contingent increments and decrements in unconditioned stimulus intensity

Abstract

An analysis of data from classical aversive conditioning of the nictitating membrane of rabbits with the two-phase model was made in order to test predictions from Spence’s drive theory and the Rescorla-Wagner associative theory. While overall group means were ordered as expected from these theories, the interrelationships among estimates of parameters in the two-phase model showed that neither can account for more detailed analyses of the data. It was shown, for example, that the inferior performance of subjects receiving an avoidance treatment was not, as theoretically predicted, a result of the CR-contingent omission of the US leading to a decrement in response probability. Similarly, it was shown that increments in response probability followed trials on which the CR occurred even when CR-contingent US intensity was lower than that on non-CR trials, an effect which neither theory can predict since both are response independent. These two outcomes are opposite to what has been obtained in human aversive conditioning, which means that the two theories, while accounting for the behavior of one species, are of limited generality across species.