An assessment of hypothesis testing in mentally retarded adolescents.

Groups of educable and trainable institutionalised retarded adolescents were tested on a discrimination learning problem with a modified blank trials procedure in an attempt to measure their hierarchy of hypotheses. Results indicated that the hierarchy and size of initial hypothesis sets varied as a function of degree of retardation. Trainable retarded subjects had fewer hypotheses, and initially chose position hypotheses predominantly. Under discriminate reinforcement, most of these were readily switched to stimulus dimension hypotheses, which they retained during additional non-discriminate (100 per cent) reinforcement trials. Educable retarded subjects predominantly chose stimulus dimension hypotheses initially, and most of these switched to position hypotheses during either discriminate or non-discriminate reinforcement trials.