Response Generalization

Abstract: In Experiments 1 and 2, rats received initial training in which two stimuli (A and N) were either followed by the same consequence (food) or by different consequences (food and no food). Subsequently N was paired with electric shock and the generalization of conditioned suppression to A was assessed. Suppression to A was more marked when A and N had both been followed by food than when they had had different outcomes. In Experiment 3, 3 stimuli (A, B, and N) were presented initially. For one group, A and N were paired with food and B was nonreinforced; for a second group, B was paired with food and A and N were nonreinforced. Generalization of suppression was found to be more substantial to A than to B for both groups. These results indicate that the extent to which stimuli are treated as being equivalent is partly determined by their reinforcement histories. The observation that a conditioned response that has been established to one stimulus will also be elicited, to some extent, by other stimuli is known as stimulus generalization. A widely held interpretation of this phenomenon relies on the assumption that the presentation of a stimulus will unconditionally excite a number of representational elements. It is supposed that though some of these elements (unique elements) will be activated only by the trained stimulus itself, others (common elements) will be activated by a range of stimuli (see, e.g., Atkinson & Estes, 1963; Mackintosh, 1974; Rescorla, 1976). To the extent that two stimuli excite common elements, animals will perceive them as alike and will respond to them in similar ways. This analysis provides a ready explanation of stimulus generalization and of a range of other phenomena (Rescorla, 1976).