対連合学習における媒介連合の効果I

Four experiments were performed to examine the facilitation effect and interference effect of mediated assciation with the paradigm A-B, B-C, A-C. (Exp. I and II for facilitation effect, Exp. IV for interference effect and Exp. III for both.)<br>The stimulus list consisted of pairs of standard nonsense syllables prepared by Umemoto and his co-workers and having 30%-70% non-association values. Exp. I used 4 lists of 8 pairs (Table 1), Exp. II, lists of 4 pairs (Table 3), Exp. III, 7 lists of 6 pairs (Table 5) and Exp. IV, 5 lists of 4 pairs (Table 10). Each list contained two or three subsets for different experimental purposes, that is, for facilitation, interference and control. The Ss were 10 college students in Exp. I, 10 students in Exp. II, 15 students in Exp. III, and 10 students in Exp. IV. In Exp. I, II and IV, the Ss were divided into two groups; in Exp. III, into three groups at random. The Ss learned three lists successively, at 24hr. intervals in Exp. I and at 1-min. intervals in Exp. II, III and IV. Each pair was presented to the Ss for 2sec., until 5 successive correct responses were attained. Two kinds of measures were used; number of trials for each pair and its rank order in learning. The main findings are as follows:<br>1. In Exp. I, the facilitation effect of mediated association was observed not in terms of the number of trials, but in terms of the rank order of learning, which was statistically significant (Table 2).<br>2. In Exp. II, the facilitation effect was significant in terms of both measures (Table 4). It can be said, therefore, that if the number of pairs in a list and the length of list intervals are both appropriate, the facilitation effect can be unfailingly observed in such experiment as this.<br>3. In Exp. III, the facilitation effect was observed in both measures but the interference effect did not appear in either measure (Tables 7, 8, 9).<br>4. In Exp. IV, in which the interference effect alone was examined, the effect was not observed in terms of either measure (Table 11). Consequently, the interference effect must be examined in a different context from this experiment.