The study was designed to test what is the effect of criterion of learning on extinction of responding in successive operant discrimination. The method used in this experiment was similar to that of my previous study. Six pigeons were trained for pecking at a monochromatic light as the positive stimulus and an achromatic light as the negative one. the criterion of learning set up at 100 per cent correct response with no errors for one session a day. After the criterion of training was reached, testing effect of learning was carried out under the extinction trials in one session similar to that of training trials. The rate of responding in extinction was compared to the number of errors per reinforcement in the last but one session and average number of errors per reinforcement over the entire learning session. It was found that the more respondent was the extinction, the more errorless was the acquisition in the last but one session, while no correlation was between the rate of responding in extinction and average number of errors over the entire learning session. This result suggests that total correct (or incorrect) responses in the last session or over the entire learning session are only one aspect of criterion of learning and the other aspect, the slope of learning curve at the point where training ceased, needs to be considered.

横山松三郎先生古稀記念論文集