The Role of Response-Unit Formation in the Operant Partial Reinforcement Extinction Effect

Abstract

<p>In both instrumental and operant situations, the number of responses emitted in extinction is greater following partial than following continuous reinforcement. This partial reinforcement effect (PREE) has been the subject of considerable theoretical interest.</p> <p>The response unit hypothesis proposed by Mowrer and Jones (1945) accounts for the operant PREE by arguing that when reinforcement follows a sequence of responses, that entire sequence forms a single functional response, or response unit. The hypothesis predicts that the number of individual responses emitted in extinction will increase with response-unit length, but the number of response units emitted will be independent of response-unit length.</p> <p>A response unit terminates when the subject approaches the foodcup. The response-unit hypothesis argues that this terminal response is controlled by internal stimuli produced by execution of the unit. When such internal control is fostered by the elimination of external cues to response-unit completion, extinction performance is consistent with the response-unit hypothesis: The number of responses emitted increases with acquisition unit length, while the number of units emitted is constant across a wide range of unit lengths.</p> <p>Fixed ratio (FR) reinforcement schedules provide external cues to ratio completion during acquisition. Several studies have shown that the presence of such stimuli in acquisition does not prevent the development of internal stimulus control over response-unit termination, but that extinction performance may be affected by the availability of such stimuli in extinction.</p> <p>In the studies reported in the thesis, the length and number of response units emitted in extinction were examined as functions of the extinction scheduling of external stimuli associated with reinforcement during acquisition.</p> <p>An early study revealed that extinction performance following FR reinforcement differed considerably in the presence or absence of external cues to ratio completion. In the absence of such cues, extinction unit length exceeded acquisition unit length, and the number of units emitted was independent of acquisition unit length. In the presence of such cues, acquisition and extinction unit lengths were identical, the number of units emitted was much greater, ans decreased in a negatively accelerated manner with increasing acquisition unit length.</p> <p>Subsequent studies, in which the extinction schedule of external stimulus presentation was manipulated, eliminate several alternative accounts of these results, and suggested a modified version of the response-unit hypothesis which rejects the assumption that the number of response units emitted in extinction is independent of response-unit length.</p> <p>According to this revised response-unit hypothesis, the number of units emitted in extinction depends on the length of both the unit reinforced during acquisition, and the unit emitted in extinction. For any given acquisition unit length, the number of units emitted in extinction decreases in a negatively accelerated manner as extinction unit length increases. This relationship can be expressed as a power function of the form U = M x L^(-B), where U is the number, and L the length, of the units emitted in extinction. As acquisition unit length increases, the function shifts upward and become flatter; thus the number of units emitted in extinction increases with acquisition unit length.</p> <p>This model accounts for the effect of external stimuli on extinction performance in terms of their capacity to determine extinction unit length by eliciting response-unit termination. The model was shown to be capable of accounting for the results of a number of experiments involving manipulations of both acquisition and extinction unit length.</p> <p>The results of these studies provided considerable support for a response-unit analysis of the operant PREE, and suggested that such an account might provide the basis for a general theoretical account of both the operant and the instrumental PREE.</p>