A Study of Set Effects in the Verbal Conditioning Experiment
Frances Brown, Wilse B. Webb · Psychological Reports · 1960
The effect of set on the prediction of uncertain alternative events was studied in this experiment in relation to the statistical association theory of learning. The standard verbal conditioning was provided in which Ss were required individually to predict which of two lights, right or left, would flash on following a signal light. All Ss were given 160 trials. Two patterns of trials were set up with the reinforcing event randomly spaced, and having a probability of occurrence of .15. One pattern placed the reinforcing event on the left; the other placed it on the right. These patterns were used alternately within groups. Ss (college students) were divided into three groups of 10 each. Group I was told that there was a perfect solution to the problem, and that they should make no errors if possible. Group I1 was told that there was no perfect solution to the problem, and that their task was simply to make as few errors as possible. ' Group I11 was told only to say the first word they could think of when the ready signal flashed and then to predict which light would appear. They were given no instructions concerning possible solution or errors. The appropriate instructions were restated to each S at the end of the 40th, 80th, and 120th trials. A statistical analysis of the data was performed by a Type I, mixed group analysis of variance. The results showed no significant effect by treatments or a significant interaction. The terminal response of all three groups was approximately .I5 with no significant difference in the last 20 trials. Conclusion: The three conditions of set used in this experiment were not effective in changing the lawful probability of response as outlined by the statistical association theory of learning.