Spatial Discrimination and Reversal Learning Involving Differential Magnitude of Reinforcement

Abstract

Three experiments are reported in which choice in the acquisition of spatial discrimination and subsequent reversal learning was based on differential magnitude of reinforcement. Exp. I presented evidence suggesting the operation of an important frustration factor on the smaller-magnitude side in this type of discrimination, and revealed a significant effect of temporal distribution of reinforcement on reversal learning. Exp. II showed that, while differential-magnitude discriminations may be originally acquired as readily as reinforcement-nonreinforcement discriminations, the former are not reversed as readily as the latter, and both types are affected by the administration of forced trials guaranteeing equal response frequencies to the two sides as compared with allowing free choice on all trials. Exp. III, concerned with reversal learning as a function of the magnitude ratio and delay of reinforcement, provided further evidence of poor reversal performance under differential-magnitude conditions, but indicated that performance resembling that in traditional reversal learning may be found if the magnitude ratio is very high.