Theories of learning that predict a negative exponential growth curve are based on the assumption that learning is a process of replacement : incorrect response tendencies are replaced with correct ones. In this article, an alternate model is presented, which describes learning as a process of accumulation, where incorrect response tendencies remain constant and correct response tendencies increase with practice. These two approaches were compared in their ability to describe the shapes of the learning curves for several experiments in free recall and perceptual and motor learning. In nearly every case, the predictions of the accumulation model were superior, and the predictions of the best fitting replacement equations failed in systematic ways. The relationship between the accumulation model and a reinforcement theory of steady-state responding was discussed. It was argued that the accumulation model should be given greater consideration as a variable model of the learning process. (Author/publisher)
Abstract