The role of response-reinforcer contingency on acquisition and maintenance of responses learned by observation

Abstract

Results are reported of two experiments designed to evaluate the effects of different response-reinforcer relationships, both during the modeling phase and in the later maintenance phase, with respect to the performance of a novel response in observer pigeons. In experiment 1, three groups of observer pigeons were exposed, during the twelve-trial modeling phase, to a model trained to open a tube in order to receive food. A control group was not exposed to the model. Later, the groups were exposed to different responsereinforcer relationships: in one group, the relationship was contingent; in another, the relationship was non-contingent; in a third group, the relationship was negative and in the control group, the relationship was contingent. In experiment 2, three groups of observers were exposed, during the modeling, to a relationship that was non-contingent between the responses of the model and the reinforcer. Subsequently, one of the groups was exposed to a contingent relationship, while the other two groups were exposed to a non-contingent relationship; the number of trials being varied between those two groups. The results showed that the acquisition of the response by observers was superior when the relationship between the responses of the model and the presentation of the reinforcer was contingent. With respect to maintenance of the responses, the data showed that the responses acquired by observation only remained in the repertoire of the observers when the contingent relationship between responses and reinforcers was maintained. The foregoing allows the argument that learning by observation, as with instrumental learning, is sensitive to variations in response-reinforcer contingency.