1973 年 43 巻 6 号 p. 290-298
The purpose of the present experiment was to examine which of the two major theories, the reinforcement theory or the cognitive theory, could explain verbal conditioning better. If verbal conditioning conforms to the reinforcement theory, the following hypotheses can be demonstrated. (a) The level of reversal learning will be low when the level of original learning is high. (b) The level of reversal learning will be low when the trials of original learning are few.
To test these hypotheses, two experiments were conducted using the procedure of reversal learning of the Taffel-type task. The personal pronouns used were “I”, “YOU”, “HE”, “SHE”, “THEY” and different verbs were combined with each of them. In the original learning (40 trials) of the first experiment, positive reinforcement was given to the pronoun of the lowest operant level (L-O), negative reinforcement to the other four pronouns. Subjects were divided into three groups according to the level of the original learning. In the original learning of the second experiment, the pronoun of the middle operant level (M-O) which was given positive reinforcement in the original learning was used instead of L-O. Subjects were divided into three groups according to the number of trials of the original learning; 20, 40, 60. In the reversal learning of Exp. 1 and Exp. 2, positive reinforcement was given to the pronoun of the highest operant level (H-O). Throughout these experiments, reinforcements were provided by E's saying “Yes” or “No”.
The main results were as follows. (a) The higher the level of original learning, the higher the reversal learning. (b) The reversal learning was higher as the number of trials of the original learning increased.
These results might be interpreted as follows: In the groups which showed the highest level of original learning in Exp. 1 or which were given many trials of original learning in Exp. 2, overtraining might have resulted, so that reversal learning was promoted because of the overtraining reversal effect. To test this possibility, subjects were given 10, 20, or 40 trials of overtraining after reaching a criterion of 5 successive correct responses. Immediately after completing the original learning, they were given the reversal learning with a criterion of 5 successive correct responses. No relationship was found between the magnitude of overtraining and reversal learning.
The present study did not support the hypotheses based on the reinforcement theory, but suggested the interpretation of the cognitive theory.