A Model of Response-Reinforcer Contingency

Abstract

<p>The experiments reported in this thesis investigated the effect of response-reinforcer contingency on response differentiation. Since the failure to control response probability in previous studies had led to difficulties. the present experiments employed percentile schedules to control response probability. Response-reinforcer contingency was indexed by the measure of statistical association between two dichotomous variables known as the phi coefficient (ϕ). and a model of this independent variable was developed to permit a systematic investigation of contingency in operant conditioning. This model was tested using rats in a spatial response differentiation paradigm. The results of three experiments revealed that the higher the value of ϕ. the more effective the shaping of response location to a target location. Despite differences between experiments in the way the independent variable was manipulated. across all three experiments there was a very orderly relationship between asymptotic conditioning and ϕ. These experiments demonstrate the importance of response-reinforcer contingency in response differentiation and provide support for a model of contingency in operant conditioning based on ϕ.</p>