Inhibitory S-R Associations Influence the Partial Reinforcement Extinction Effect

Abstract

The Partial Reinforcement Extinction Effect (PREE) is the finding of more rapid extinction to a continuously reinforced cue than to a partially reinforced cue. Despite its well-documented existence, it has challenged many theoretical accounts of learning that assume the likelihood of responding is correlated with the strength of the learned Stimulus (S) – Response (R) association. While a number of explanations (notably by Amsel and Capaldi) of the PREE have been proposed, Rescorla (1999) reported a PREE using an entirely within-subjects design—a finding that runs counter to these proposed explanations. We describe how the PREE might be explained by the formation of inhibitory S-R associations that develop during the acquisition of partial reinforcement, but not continuous reinforcement. To test this proposal, we trained pigeons to peck a response key in the presence of one partially reinforced stimulus but also omitted the response key during the training of an alternative partially reinforced stimulus—thus preventing the development of inhibitory S-Rs to that latter stimulus. We found evidence for the PREE only for the partially reinforced stimulus where inhibitory S-Rs were capable of being formed, thus supporting our proposal. Alternative explanations of our procedure and its relevance to broader computational rules of learning is discussed.