Extinction after a Small Number of Partial Reinforcement Trials

Abstract

An earlier study (McCain, Love, & Gruer, 1962) indicated that the partial reinforcement effect could be produced in rats following three training trials. The present study was an attempt to extend this finding to humans. Two previous studies (Lewis & Duncan, 1958; Yarczower, Vlases, & Friedman, 1960) have been addressed to this same problem. The findings, although not significant, have been in the same direction as those on rats. Ss were 40 first grade children. The test apparatus was similar to the Wisconsin General Test Apparatus except that it was adapted for children. Small toys and balloons were used for reinforcement. Ss were told they were playing a game, that they were to lift the lid of the box and could keep anything they found. The partial reinforcement group (P) received a reinforced, a nonreinforced, and a reinforced trial, in that order. The consistent group (C) received chree consecutive reinforced trials. After the first block of chree trials Ss were asked to guess whether a toy would be in the box before each of the 10 succeeding extinction trials which were separated only by 5 to 10 sec. for resetting the apparatus. Group P guessed that a reinforcement would be present more often than Group C did throughout extinction. The data were analyzed in blocks of two trials. The difference between the groups increased from early to late blocks of extinction trials. When the last four blocks of trials were compared, the difference was significant (t = 2.72, df = 38, p < .01), as it was also for the last two blocks (t = 1.80, p < .02). These results indicate that the greater resistance to extinction normally associated wich partial reinforcement can be obtained wich humans given small numbers of trials. The primary importance of this study is the possibility of studying partial reinforcement in the very early stages of training withouc complications from extended training. Children's past social experiences with schedules of reinforcement, based on day-to-day living, enforces caution in weighing conclusions without replication.