Visual probability learning of institutionalized profoundly retarded adult males.

Eighteen profoundly retarded men were given repeated presentations of a two-choice visual discrimination using a modified Wisconsin General Test Apparatus and two probabilistic reinforcement schedules counterbalanced for order. Individuals received 240, 70:30 trials and up to as many 100:0 trials; when schedules were shifted, majority/minority (A1/A2) designations were reversed. The 100:0 results indicated preshift mean terminal rates that approached the reinforcement level and postshift rates that were inhibited by prior 70:30 training. The 70:30 results indicate learning only for those given a particular stimulus as their a1: overmatching prior to the shift that was reduced by previous 100:0 training to the matching level. Learning failures were explaiined in terms of attention to a preferred stimulus and the need for more trials, and areas for further inquiry were suggested.