Evidence for multiple processes contributing to the Perruchet effect : response priming and associative learning

The Perruchet effect constitutes a robust demonstration that it is possible to dissociate conditioned responding and expectancy in a random partial reinforcement design across a variety of human associative learning paradigms. This dissociation has been interpreted as providing evidence for multiple processes supporting learning, with expectancy driven by cognitive processes that lead to a Gambler's fallacy, and the pattern of conditioned responding (CRs) the result of an associative learning process. An alternative explanation is that the pattern of CRs is the result of exposure to the unconditioned stimulus (US). In 3 human eyeblink conditioning experiments we examined these competing explanations of the Perruchet effect by employing a differential conditioning design and varying the degree to which the 2 conditioned stimuli (CS) were discriminable. Across all of these experiments there was evidence for a component of the CRs being strongly influenced by recent reinforcement, in a way that was not demonstrably influenced by manipulations of CS discriminability, which suggests a response priming mechanism contributes to the Perruchet effect. However, the complete pattern of results and an analysis of the results from previously published studies are also consistent with there being an associative contribution to the effect.