Differential effects of positive and negative social reinforcement on juvenile delinquents and Sunday School students

Two experiments were conducted to investigate the hypothesis that, as a result of different histories of social reinforcement, juvenile delinquents respond differently to such reinforcement than do Sunday School students. In an early review of the learning theory literature, Young (1936) reported studies which showed that a combination of positive and negative reinforcement (PR and NR) produced more learning than PR or NR alone, with the former producing better results than the latter, and with all three conditions leading to more improvement than a no-reinforcement (NoR) condition. In a recent pilot study, however, Martinez (1967) found that institutionalized delinquent girls did not conform to this pattern when reaction time (RT) responses were verbally reinforced by the statements 'Good' following fast RTs and 'That's bad' following slow RTs. (In an earlier (1966) study, he had found RT to be modifiable by verbal reinforcement). For these delinquent Ss, NoR was found to produce more learning than NR, with the PR condition producing more learning than the combination contingency. On the basis of these findings, it was hypothesized that juvenile delinquents, because of their histories of experience with social reinforcement, have become oversensitized to NR so that it is detrimental to learning. In the first experiment, then, a sample of 72 delinquent Ss and one of 60 Sunday School students were each equally divided into four groups. All Ss received 40 RT trials using the classic RT apparatus and procedure, excluding speed instructions, and the same verbal reinforcements employed by Martinez (1967). One group in each sample received no reinforcement after any of the trials, one group was positively reinforced for fast responses, one was negatively reinforced for slow responses, and one received both PR and NR for appropriate responses. The results of this experiment were in accordance with previous reports in the literature of the relative effectiveness of the four reinforcement contingencies for the normal Se. For the delinquents, however, while the NoR condition produced the least improvement and the combination contingency produced the greatest, the curves for the PR and NR groups were almost identical. Because of the significantly greater degree of improvement produced in these S-s by the combination condition, it was decided that the number of trials should be extended to 60 in Experiment II to investigate the hypothesis that, under this contingency, the performance of the delinquents is more closely equal that of the normals. Each reinforcement group in Experiment II was composed of 10 delinquent Ss who received 60 RT trials under the same conditions as in Experiment I. While the results duplicated those of the first experiment with regard to the NOR, PR, and NR groups, the combination group failed to show any improvement over these latter two groups; all three produced almost identical curves. The overall results were interpreted as indicating that, while delinquents have some tendency to perceive PR and NR as cues for directing their behavior, in the manner of the normals, they apparently also react to reinforcement on the basis of some need not present in normals to such a degree. It was suggested that possible needs include attention (love), the production of an influence on the environment (self-assertion), and the solicitation of external controls upon their behavior. It was further suggested that a psychotherapeutic approach with these Se, while focusing on the development of feelings of worth, 'positive' self- assertiveness, and internalized controls, should also include a learning experience as to the nature of and intent behind PR and NR.