The effectiveness of social reinforcement as a function of changes in rate of reinforcement

Abstract

One hundred and twenty subjects were divided into two groups, receiving either 100% or 33% reinforcement for the emission of the “correct” emotional-labeling response during an acquisition phase of 60 trials. This phase was followed by 60 trials during which conditions of none, or low change, moderate change, and substantial change from the base-line rates were established, thereby giving us three positive-discrepancy groups and three negative-discrepancy groups of 20 subjects each. Changes in the effectiveness of social reinforcement were assessed by comparing, through the use of difference scores, the average number of correct responses made during the last two blocks of the shift phase with the average number of correct responses made during the last two blocks of the acquisition series. As hypothesized, a significant interaction between the initial rate of reinforcement and amount of change was obtained. A negatively accelerated linear trend was found for the 100% groups and a quadratic trend peaking at the moderate-change level for the 33% groups. These results were interpreted as indicating the importance of direction as well as of the magnitude of change. Contrary to a second hypothesis, positivity of subjects' mood changes was significantly affected only by magnitude of change. It was also found that subjects who valued the experimenter's approval and who encountered a substantial decrement in rate of approval were more likely to be resistant to extinction and to volunteer for an unpleasant experiment than were their counterparts who received a substantial increase in rate of approval. These reactions were taken as supporting the notion, derived from the senior author's SRS model, that subjects may use ingratiation tacties to reduce disparities between current and past levels of reinforcement.