Kindergarten children performed in pairs on a double lever (Rl and R2) apparatus. In group 50:100, Child A received 50% reward for RI responses and Child B received 100% reward for R2 responses. In group 100:100 Child A and Child B received 100% reward for RI and R2 responses, respectively. The R I data were contrary to previous findings with rats and children in that Ss rewarded on a 50% schedule responded more slowly than those who were continuously rewarded.Higher asymptotic performance for partially as compared with continuously rewarded Ss has been demonstrated with infrahuman (Spence, 1960, Chap. 6) and human Ss (Ryan, 1965). A similar finding has been noted when children perform in the presence of a passively observing peer (Rosenbaum & Brunning, 1966) .In one study (Ryan, 1965), children performed two successive lever responses, R1 and R2' On R1' Ss who received 50% reward responded faster than those who received 100% reward; on R2' where all Ss in both groups received 100% reward, those who had received 50% reward on R1 responded faster as compared with the 100% group. The present investigation replicated the reward conditions of Ryan's study with the exception that the children were seen in pairs. Thus, each child performed in the presence of an observer who actively performed a similar response.
MethodThe Ss were 32 male and 32 female children (64-76 months). With one exception, the double-lever apparatus was as described by Ryan (1965). A unique feature was a common goal box which collected marbles from both levers.Pairs of Ss were matched for sex and age. Child A performed R1 and Child B performed R2 throughout the experiment. The Ss were told that this was a team game which required filling the common goal box with marbles in order to win a toy. Ss were instructed not to converse during the game. 3 There were two rewarded practice trials and 36 test trials. Interresponse and intertrial intervals were 5 and 15 sec., respectively.Each pair of Ss was assigned to one of two reinforcement conditions. In group 100;100, each member of a given pair received 100% reward. In group 50;100, Child A received 50% reward for Rl responses and Child B received 100% reward for R2 responses. For partially rewarded Ss, three reward and three nonreward trials occurred in each block of six trial s. Measures of starting time, from onset of a stimulus light to initial movement of the lever, and movement time, duration of the 15-in. lever pull, were taken on each trial. Results R1 starting and movement times were converted to speeds (l/t sec.). In a separate analysis for each measure, the between-Ss variables were reward condition (50;100 vs. 100;100) and sex of S. The within-Ss main effect was trial blocks (six blocks of six trials each). The Rl movement speeds revealed a significant interaction ( Fig. 1)