The purpose of the present experiment was to ask whether the ease of post-shift discrimination learning after overtraining was related to the rate of extinction of original responses or to the mediating verbal or attentional process to the relevant dimension, as was proposed by twostage theorists of discrimination learning (e.g., Kendler & Kendler, 1962; Mackintosh, 1965). A 2×2 factorial design was used, with two degrees of original discrimination learning and two types of shift (reversal and nonreversal). In each of the four groups, extinction trials were inserted between original and shift problems. The Ss were young children ranged in age from four to six years, who were divided into each of four groups equating in age and sex. They were trained on a twodimensional (color and form), simultaneous discrimination task either to a criterion of nine correct responses out of 10 successive trials or to additional 30 overtraining trials after the criterion. Then, the Ss were subjected to extinction of the original responses until the successive perseverative responses to S+ were extinguished and additional four extinction trials were further given. Shift discrimination was continued without further instructions to a criterion of nine correct responses out of 10 successive trials. The correct responses were reinforced by saying “Hit”, but no information was given after the incorrect responses under both discrimination learnings. In the extinction phase no information was given after each response. The main findings were summarized as follows; (a) Performances during the extinction trials were not influenced by the overtraining of original discrimination (Table 2). (b) No significant correlation was found between rate of extinction and ease of shift discrimination. (c) No significant effect of overtraining was found in case of reversal shift, while nonreversal shift was hindered by the overtraining of original discrimination (Table 3). (d) Fast learners in the original discrimination extinguished their original esponses more slowly than slow learners. (e) Reversal shift was faster than nonreversal shift for the fast learners, but a difference in rate of both shifts was negligible for the slow learners (Table 4). The findings were discussed with reference to two-stage theories of discrimination learning, especially with respect to the Kendlers' mediating response theory and the Mackintosh's attention theory.
This experiment was designed to examine the effects of verbal reinforcement combinations on discrimination learning and discrimination shift learning in normal and mentally retarded children. The apparatus was the same as in our previous study. We set 9 trials as one block and defined the criterion of learning as the state of all correct responses in a block. After the Ss had reached the criterion in the discrimination learning, in which one of three colors was the correct stimulus, they learned the discrimination shift learning to the criterion, in which another color was the correct stimulus. Besides three combinations of verbal reinforcement, that is, Right-Wrong (RW), Right-Nothing (RN), and Nothing-Wrong (NW), but there were also RNw and NrW that were RN and NW in which S was instructed about meaning of Nothing before the begining of learning. The results as were follows: 1. In the discrimination learning, mentally retarded children learned faster under NrW, RW and NW conditions than under RN, RNw. For normal children, there were no differences in the speed of learning among 5 conditions. Under RNw, RN and NW mentally retarded children learned more slowly than normal children, but under NrW and RW they learned as fast as normal children (see Table 3 and Table 4). In the discrimination shift learning there were no differences among 5 conditions and between normal and mentally retarded children (see Table 4). 2. In the discrimination learning for mentally retarded children, the mean rate of error responses was smaller under NrW than under NW and RNw, and it was smaller under RW and RN than under RNw. In the discrimination shift learning, it was smaller under NrW than under RNw and RN, and it was smaller under RW than under RN (see Fig. 1, Table 5 and Table 6). In the discrimination learning for normal children, the mean rate of error responses was not significantly different among 5 conditions, but in the discrimination shift learning it was smaller under RW than under any other conditions (see Fig. 1, Table 7 and Table 8). 3. In the discrimination learning for mentally retarded children, the rate of error responses decreased largely toward the end of learning (see Fig. 1 and Table 5). For normal children it decreased rather early in the period of learning under NrW, decreased linearly under RW and RNw, but under RN the curve of decreasing was like that of mentally retarded children (see Fig. 1 and Table 7). In the discrimination shift learning it decreased rather early in the period of learning under all conditions for both groups of subjects (see Fig. 1, Table 6 and Table 8). These findings suggest that N is a positive reinforcer at the start of the experiment and its positive reinforcement value is increased under NW, while it is weakened to turn into a negative reinforcer under RN. In other words, the positive reinforcement value of N is stronger than its negative reinforcement value. This is especially true for mentally retarded children. The acquirement of negative reinforcement value of N seems more difficult for mentally retarded children because of their functional rigidity. The instruction as to the positiveness of N under NrW was perhaps easy to accept for both normal and mentally retarded children, but the instruction as to be negativeness of N under RNw might have been difficult for mentally-retarded children because of N's strong possitiveness for them.
The present study was designed to investigate (1) whether or not the prediction of academic records in high school on the basis of intellectual apptitude is influenced by personal emotional adjustment as defined by Group Inkblot Test, and (2) the relationship between personal adjustment and a) level of intellectual ability and b) school records. A hypothesis to be tested was that “better” adjusted students would be more predictable than the maladjusted. Predictability here was determined by correlation coefficients between test scores of apptitude-general intelligence, verbal and mathematical and 3rd-year school records-Japanese language, social science, mathematics, natural science and foreign language. The subjects consisted of 257 high school male students who were classified into three groups “better”, “average” and “worse” adjusted according to the number of check points on the ten maladjustment scales of Group Inkblot Test. Comparisons of the groups on correlations between apptitude test scores and high school records confirmed that “better” adjusted students were more predictable than the maladjusted. However, simple analysis of variance among the groups on both levels of intellectual ability and of high school achievement revealed that the “better” adjusted group was superior to the average group only in verbal ability and foreign language. But a further analysis of data indicated that in mathematics, natural science and foreign language the “better” adjusted group obtained higher records than the group which was checked on FC′+C′F+C′+C′sym (emotional factor in personal adjustment) in the maladjustment scales of Group Inkblot Test, although mean differences for each of apptitude test scores between these groups were not statistically significant. These results indicated that (1) no clear relationship existed between total degree of adjustment diagnosed on the ten maladjustment scales and a) levels of intellectual ability and b) high school records, but (2) there were positive relationships between emotionality scale and high school records, and (3) the prediction of high school achievement was markedly influenced by personal adjustment as defied by Group Inkblot Test.