Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)




The purpose of tills study was to investigate the reversal and nonreversal shift learning of retardates as a function of mental age and the difficulty level of the cue dimension® Reversal shift learning is generally more rapid if the learner is capable of using a mediational approach to problem solving. Nonreversal shift learning occurs more readily if the learner employs a single-unit S-R approach to problem solving.

In the present study retarded subjects at three different ability levels learned a two-choice discrimination on the size dimension or the color dimension during original learning. The subjects were then randomly assigned to learn a control shift, a reversal shift, or a nonraversal shift*

The writer hypothesised that retardates at the high and middle ability levels would mediate while learning a dis crimination on the size dimension, but that few, if any, would mediate on the more difficult color dimension. Form differences were irrelevant for all subjects.

The subjects consisted of 108 retarded persons,102 of whom were institutionalised. The subjects were placed into mental ability groups on the basis of performance on the Stanford-Binet Intelligence Scale. The chronological age of the subjects ranged from eight to twenty-one years.

The basic experimental design was of the 3x2 factorial type with three types of shifts and two cue dimensions at each ability level.

The apparatus for displaying the discriminanda was a rectangular turntable with two depressions on one side. A wooden screen separated the two sides of the turntable. The discriminanda- were placed in the depressions for presentation to the subject. The discriminanda were blocks of three shapes and colors. The blocks were stacked in order to vary the size of the discriminanda.

Fifty-four subjects had original learning (OL) on the size dimension and fifty-four had OL on the color dimension. As soon as the subjects learned to a criterion of 7 of 8 correct choices during OL, they began with shift learning. In shift learning the same criterion for learning was used. The dependent variable was the number of errors to criterion.

The combining of the subjects of high and middle ability resulted in a significant interaction effect because of the extremely small mean value for the group with KR shift learning on the size dimension.

The consistency in the order of the magnitude of the mean values with the predictions for the high ability groups was viewed as being of theoretical importance. The consistent order of the magnitude of the values became more marked for the data following a logarithmic transformation.

The position that some of the retardates would be able to mediate on the size dimension, but that few, if any, would be able to mediate on the more difficult color dimension •was strongly supported by the findings for the subjects of middle ability. Mediation and non-mediation by retardates does seem to be a function of the difficulty level of the cue dimension.

For the low ability subjects the prediction of R-1IR differences across cue dimensions was not supported. The predictions of a cue dimensional- effect and of specific group differences were not supported for the subjects of low ability.

Although the mean values for the transformed data did not differ significantly for the low ability subjects, the relative magnitude of the values was even more clearly in line with predictions.

The writer concluded that the position that the mediation and non-mediation of retardates on a simple problem solving task may usefully be viewed in terms of the difficulty level of the cue dimension involved was strongly supported, The research which has produced conflicting evidence when the difficulty of the cue dimensions had been ignored might yield more clear-cut results if this effect were considered.