Generalization of Operant & Autoshaped Responses

Weiss, Stanley

Abstract

Behavior can be brought under the discriminative control of events in an organism's environment through discrimination training. When reinforcement is delivered independently of an organism's behavior in the presence of a particular stimulus, discriminative control is acquired by the conditioned stimulus (CS+) over the conditioned response (CR) through a classical conditioning operation. In comparison, operant discriminative control occurs when responses emitted in the presence of the positive stimulus (S+) are followed by reinforcement and responses emitted in the presence of the negative stimulus (S-) are never followed by reinforcement. Following intradimensional operant discrimination training in which S+ and S- are on the same stimulus continum, a generalization test (where many different values on the continuum are presented) often reveals that maximum responding is not controlled by S+. Rather, the gradient's peak is displaced from S+ in a direction away from S-. This phenomenon - peak-shift - has not been reported following classical intradimensional discrimination training. The proposed research will compare pigeon keypeck generalization gradients after key-pecking is established in one group by operant (response-reinforcer) and in another by classical (stimulus- reinforcer) intradimensional training procedures. Strategies are described to match training parameters over groups, varying only the presence or absence of a response-reinforcer contingency. Several independent assays will be administered to determine the contingency controlling keypecking. If the classically conditioned group produces peak-shift, this will be the first clear demonstration of this theoretically important generalization phenomenon with a respondent. If they do not produce peak-shift, while the matched operant controls do, we will have strong evidence that operant and classical discriminative processes may differ. Types of future systematic parametric investigations that might better define these potential differences are discussed.