An operant analysis of the overtraining reversal effect

Abstract

This dissertation presents a discussion of various theories that have been proposed to account for the known facilitative effects of overtraining on single or serial reversals of a discrimination. The advantages of applying operant methodology in the analysis of these effects is discussed, and an experiment is reported in which the effects of amount of training on discrimination reversal learning are analysed in context of successive reversals of a free-operant discrimination. Three groups of pigeons were trained on a red-green discrimination which was reversed 24 times. Groups were given 1, 2, or 4 hrs. of training on each discrimination. The results showed three separated effects of overtraining. The changes in performance over a series of reversals were qualitatively similar to the changes in performance produced by overtraining. These results are not accounted for by any of the theories so far proposed, nor does the present study suggest an obvious alternative to these. It is argued that speculation on the cause of the overtraining reversal effect should best await a more complete description of the phenomenon along the lines of the present study, but exploring the effect over a greater range of values of the independent variable.