An analysis of error-correction procedures during discrimination training

Mechanisms involved in error-correction procedures during behavioral acquisition were examined. Seven developmentally delayed subjects participated in match-to-sample discrimination training, consisting of three conditions arranged in a multielement design. Correct responses in all conditions were followed by praise and either food or pennies. In the baseline condition (differential reinforcement), an error produced no consequences. In the practice condition, an error was followed by repetition of the trial until a correct response occurred. In the avoidance condition, an error was followed by additional trials consisting of irrelevant stimuli; this condition separated the effects of repeated exposure to the same task from those of negative reinforcement, both of which existed in the practice condition. All 7 subjects made noticeable progress in the baseline condition. However, 5 of the 7 performed better in one of the error-correction conditions: 2 performed better in the practice condition, and 3 performed better in the avoidance condition. These data indicate that error-correction procedures may serve multiple functions and suggest that the practice requirement in this study included both avoidance and stimulus control components. More generally, the data indicate that additional control procedures should be included in acquisition studies to identify the relevant behavioral mechanism(s).