Reinforcement in Human Learning
Abstract
A review of the literature on reward in human learning reveals major inadequacies in the law of effect interpretation which become accentuated as experimental designs depart from classical paradigms. New types of experiments indicate that the associative learning process is independent of rewarding or punishing aftereffects. The alternative theoretical framework outlined in the paper embodies the following principal assumptions: (1) Associative learning is a function solely of conditioning by contiguity. (2) Recognition performance depends solely upon this learning by contiguity. (3) Modification of learned performance by rewards involves two stages. Firstly, an association must be established between the stimulus member of a stimulus-response event and the subsequent reward. Secondly, upon the subsequent occurrence of the stimulus member, the representation of the reward must be brought into memory by activation of the learned association. This anticipation of reward generates facilitatory feedback. If at the same time the previous response is recalled, the effect of the feedback is to cause the recalled response to be made overtly. (4) The immediate effect of facilitatory feedback following presentation of a stimulus is to increase the probability that the associated response will be made overtly during any given interval of time, and thus on the average to reduce the reaction time. (5) In a choice situation, in which the subject must select the stimulus to which to respond from a set of two or more, the subject is assumed to scan the available stimuli and, in general, to respond overtly to the one which the highest reward has been associated.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 20, 1967
- Accession Number
- AD0664458
Entities
People
- W. K. Estes
Organizations
- Stanford University