The purpose of the present experiment was to discuss effects of overtraining on nonreversal shift learning in Intellectually Handicapped children. (Experiment I) Effects of overtraining on nonreversal shift learning were examined under stimulus arrengement conditions. The principal hypotheses were as follows; 1. Overtraining tends to hinder nonreversal shift learning when the initial irrelevant dimension was still kept to exist in the transfer situation. 2. Nonreversal shift learning is more facilitated in CW (constant-irrelevant within trial) condition. A 2×2 factorial experiment design was used, with degree of the initial learning (criterion training and 100% overtraining) and the stimulus arrangement conditions (VW and CW). The Ss were 32 mentally retarded children, with a mean MA of 5 years and 7 months and were divided into 4 groups in considering with CA, MA and IQ. They were trained on the two-choice simultaneous discrimination tasks to achieve the criterion of 5 successive correct responses. Immediately after completing the initial discrimination learning, half of the Ss were shifted under the VW condition, and the another half under the CW condition. The correct responses were informed by saying "you got", and the incorrected responses by "you missed". The main results were as follows; (1) With respect to VW condition, there was no significant different between the criterion group and the overtrained. Therefore, the hyposesis 1 was not demonstrated. The result of CW condition was considered to be compatible with that of VW condition. (2) No significant effect was found in different conditions of VW and CW. The hypothesis 2 was not demonstrated. (3) For the criterion group and the overtrained, the correct response rate was significantly higher in UP than in CP in the first pair trial block. (4) In terms of CW condition, the correct response rate was considered to be compatible with PNP and PPP in the first pair trial block. (Experiment II) Effects of overtraining on nonreversal shift learning were examined in mentally retarded children (MR), autistic children (AU) and normal children (NR) with the same MA levels (MA 5-6 years). A 2×3 factorial design was used, with degree of the initial training (criterion training and 100% overtraining) and types of subject (MR, AU and NR). The procedure was the same with that of Experiment I. The main result were as follows; (1) Trials and errors tendency was significantly increased with overtraining according to an order of MR, AU and NR. It was found that the probability of attention to relevant dimension increased with overtraining in the order above mentioned. (2) It was, as awhole, shown that correct response rate was more increased in UP than in CP. (3) It was found that the probability of spontaneous reversal in overtrained group increased in accordance with an order of MR, AU and NR. The result were discussed in considering with two-stage theories of discrimination learning, especially with respect to the Attention Theory (the Sutherland and Mackintosh's analyzer theory and the Zeaman and House's attention theory).
View full abstract