Using stimulus shaping and fading to establish stimulus control in normal and retarded children

The present study was an effort to investigate whether, in addition to his IQ level, the child's ability to identify all relevant stimulus components would affect the frequency of overselective responding. Children of different IQ levels (i.e. normal, educably retarded, and trainable retarded children) participated. Subjects were trained to learn the meanings of four sets of fictitious words, i.e. two sets containing words printed in Roman letters (Roman words), and two sets containing words printed in Hebrew letters (Hebrew words). All subjects could identify the words of each set. The normal and educably retarded subjects could read aloud the Roman words, whereas the trainable retarded subjects could not. None of the subjects could read the Hebrew words. Two training procedures were used, one requiring transfer of stimulus control (fading), and one which did not (stimulus shaping). The results indicated that, firstly, the discrimination learning of the normal and educably retarded subjects covaried with the IQ level and their ability to read the words. The learning rate was not affected by the training procedures. Secondly, the trainable retarded subjects learned much better through stimulus shaping than through fading. Their acquisition rates were slow and not affected by the types of letters. Thirdly, the training procedures had no effect on the breadth of stimulus control. Instead, it covaried as a function of the IQ level (all groups) and of the child's ability to read the words (normal and educably retarded subjects). Fourthly, the training procedures had, however, considerable effect on which letters controlled the discriminations. When overselective selective responding was evident, the letters that had been associated with the prompts were more often functional than the other letters, but only for the words trained through stimulus shaping.