The effect of partial reinforcement on acquisition and extinction of a class of verbal responses.

Abstract

This study tested the effect of partial reinforcement on acquisition and extinction of verbal responses. 124 undergraduates were instructed to estimate movement of an autokinetic stimulus. A range of number responses at the upper end of each S''s distribution of 30 initial estimates was selected for reinforcement. S''s were divided into 100%, 67%, 50% and 0% reinforcement groups. During acquisition each S''s "critical responses" were followed by verbal reinforcement, according to the schedule of his group. After a criterion of six critical responses in ten consecutive trials was reached, extinction was carried on for 75 non-reinforced trials for all experimental groups. It was found that: (1) acquisition with a partial reinforcement procedure requires more trials but fewer reinforcements than acquisition with a continuous reinforcement procedure; and (2) extinction with a continuous reinforcement procedure shows greater initial resistance but faster decline than extinction with a partial reinforcement procedure. During extinction, the total number of critical responses was found to be greatest in the 67% group, although the 50% showed the slowest deceleration in responses. The results of this study were compared with findings in classical human conditioning studies and in animal studies. The difficulties of experimental control in partial reinforcement studies were pointed out, and statistical methods of control which were used in this study were discussed.