Reversal and extradimensional shifts in abstract and concrete schizophrenics.

AND CONCRETE SCHIZOPHRENICS 331 formulated a mediating-response theory of human discrimination learning. Mediatingresponse theory postulates that, in the process of discrimination learning, an implicit mediating S-R association is forced by which Ss learn to respond to a particular stimulus dimension. Since, in an R shift, the relevant dimension does not change, an R shift should be learned faster than an ED shift. The Kendlers suggested that animals learn ED shifts faster than R shifts because they do not learn mediating responses. It was also implied that human children acquire mediational processes at some stage of their development. In an investigation which dealt with the developmental implication of their theory, Kendler and Kendler (19S9) found no differences between R and ED shift learning in kindergarten children. However, an after-thefact comparison of fast learners and slow learners suggested that, as a group, the children seemed to be in the process of developing mediating processes, but, individually, fast learners tended to respond according to the predictions of the mediating response model, while slow learners responded according to the predictions of a single-stage model. Subsequent studies further seem to support the developmental hypothesis. Kendler, Kendler, and Wells (1960) found that nursery-school children learned ED shifts faster than R shifts. Furth and Youniss (1964) found no difference between R and ED shifts in young grammar school children, paralleling the results reported by Kendler and Kendler (1959), and Kendler, Kendler, and Learnard (1962) found that older grammar school children are progressively more likely to learn R shifts faster than ED shifts. As an alternative to the mediating-response theory, Mackintosh (1965) suggested that all 5s learn attention responses, which facilitate R shifts relative to ED shifts, only if the attention responses do not extinguish as quickly as the choice responses. The latter condition could hold if, as might reasonably be assumed, attention responses are learned under partial reinforcement conditions. Presumably, animals and young children, who do not ordinarily show R shift superiority, learn attention responses more slowly than adult human Ss. Results of studies with animals and young children in which overtraining was given on the preshift task tend to support Mackintosh's contention (e.g., Brookshire, Warren, & Ball, 1961; Furth & Youniss, 1964; Mackintosh, 1962). For example, Furth and Youniss (1964) reported no significant difference between R and ED shift learning speeds in a group of secondand third-grade children. However, in another group which received overtraining, R shifts were learned faster. Similar results were reported by Ohlrich and Ross (1966) with retarded