Noninstrumental Punishment and Shock Intensity in Observational Learning

Abstract

It was hypothesized that direct negative reinforcement to observers, solely contingent upon errors in a model's behavior, would facilitate or interfere with the observer's subsequent efforts to learn the task as a function of intensity of punishment (electric shock). 80 Ss were run in a factorial design varying shock intensity, the number of model errors and shock contingency. The task was a complex temporal maze which explicitly required the learning of a series of correct switch selections and implicitly required avoidance of punished switches. The findings indicated that relatively severe noninstrumental punishment disrupted vicarious acquisition of the task insofar as avoidance of punished decisions was concerned. The effects were independent of whether the shock was contingent or noncontingent. Marginal support was found for the contention that mild punishment would facilitate performance beyond the level to which no punishment would influence performance.