Visual discrimination learning in mentally handicapped adults: comparative effects of two-choice and multiple-choice training methods on stimulus generalization performance.

The hypothesis that in mentally handicapped subjects a multi-choice discrimination learning method would result in superior generalization performance to the traditional two-choice method was investigated experimentally. Twenty-four mentally handicapped adults matched for age, intelligence and duration of institutionalization were divided into two equal groups and given visual discrimination training using a differential reinforcement, prompt-fading procedure. The groups were allocated separately to a two-choice and four-choice training condition in which the discriminative stimulus was associated with one and three non-discriminative stimuli respectively. Following training, stimulus discrimination performance was compared on a generalization test comprising increasing levels of distraction. The results showed a clear advantage for multi-choice training over two-choice. However, it was found that during generalization testing discriminative performance deteriorated as a function of increasing distraction in both groups, suggesting an underlying selective attention defect. Such a defect may be characteristic of stimulus generalization performance in the mentally retarded. Multi-choice discrimination learning appeared partly to counteract major generalization failure when the ratio of non-discriminative to discriminative stimuli was high.