Comparison of three methods to train a size discrimination with profoundly mentally retarded students

Three procedures for training a size discrimination to profoundly mentally retarded adults were compared: (a) the size of the incorrect choice (S-) was progressively increased in discrete steps to criterion size (criterion related); (b) a prompt, unrelated to the criterion of size (the trainer pointing to the correct stimulus), was attenuated in discrete steps (noncriterion related); and (c) the stimuli were presented at criterion level on all trials without additional prompts (trial and error). Differences between groups were evaluated for the percentage of errors during training. Results indicated that the criterion-related method was most effective, noncriterion related was next, and trial and error was the least effective. These results were compared with previous findings.