人間の弁別学習における過剰訓練の役割(2)

概要

論文の詳細を見る
Continued from the previous paper, recent experimental researches on discrimination shifts were briefly summarized. Two main trends in research efforts were found in this period ; (a) some predictions from the Zeaman-House theory (1963) were tested, especially concerning the prediction on effects of overtraining trials in the original discrimination problem on discrimination shifts, and (b) characteristics of mediating processes were specified by relating with mediating responses and observable experimental variables. In place of the traditional one-stage S-R theories, the so-called two-stage or mediating-response theories of discrimination learning have recently been developed to account for the findings in researches of discrimination shift problems. The theories were arbitrarily divided into the verbal mediational theory and the nonverbal or perceptual mediational theory, (a) The Kendlers' theory (1962), which represents verbal mediational theories, was briefly described. The theory assumes the relation with the existence of mediating processes and the age or developmental level. Mackintosh (1965) criticized the Kendlers' theory mainly with regard to the theory construction, which was followed by the Kendlers' comments (1966) on the selective attention versus the mediation. Shepp and Turrisi (1966a) pointed out the procedural shortcomings in Kendlers' experiments. Kendler's lecture (1963) on development of mediating responses in children were discussed by well-known psychologists. The discussion was concentrated upon two points. One was the question of whether the mediating process is primarily a verbal one and therfore available only to older children. The second question dealt with whether verbal mediation involves dimensional identification or antonym-association. These materials are found to be very important to future research efforts, (b) The Zeaman and House's attention or observing-response theory (1963) assumes the learning and transfer of both the observing responses to relevant dimension and the instrumental responses to cues within the dimension, and they formulated a mathematical model of discrimination learning. The theory predicts the interaction with degree of original learning and two types of discrimination shift, where an intradimensional shift is facilitated in terms of overtraining trials but an extradimensional shift is hindered. Shepp and Turrisi (1966a) examined the validity of the theory based on the experimental evidences and suggested some methodological problems in transfer studies. It was shown from the available data that the prediction on the effect or overtraining was verified under intradimensional shifts but not necessarily confirmed under extradimensional shifts.Discussions were made with respect to dimensionnal preferences, and measurements and characteristics of mediating processes, (a) Recently, dimensional preferences were found to be influential to ease of learning discriminatinn and shift problems. The concept of dimensional preferences will be very useful as a measure of independent variables of mediating responses, (b) It seems not to be strategic for the measurement of mediating responses to rely upon the transfer-of-discrimination method only. A control group must be employed for the exact assessment of mediating responses in discrimination shifts, and transfer effects of mediating responses will be distinguished from how-to-learn effects or learning sets by the use of the control group. Some procedural faults were found in the typical device of optional shift problems, (c) The nature of mediating processes may be either verbal or perceptual, but it is more important to specify the relationship with the mediating process in each case and observable independent variables, and to analyse the conditions where the processes act upon the learning of discrimination problems. Although mediating responses are generally regarded as those to the stimulus dimension, it is possible to assume the mediating responses to the single stimulus value, and which mediation is available for the learning seems to be dependent upon both the degree of discrimination learning and the age or developmental level of the subjects. It was also suggested that mediating responses to two or more stimulus dimensions may be learned incidentally after overtraining, which may account for the faster learning of extradimensional shift problems after extensive training of the original problem.
奈良教育大学の論文
1968-02-29

著者

杉村健
奈良教育大学心理学教室

人間の弁別学習における過剰訓練の役割(2)

スポンサーリンク

概要

著者

関連論文

スポンサーリンク