Stimulus generalization after equal training on two stimuli.

Publisher Website
Google Scholar
Abstract
To ex-plore the problem of the summation of stimulus-generalization gradients, three groups of pigeons were reinforced for key-pecking at a disk illuminated by monochromatic light. One group was given an equal number of reinforcements with wave lengths of 540 and of 550 m[mu]; a 2d group was trained with 530 and 560 m[mu], and a 3d group with 520 and 570 m[mu] Generalization gradients for all groups were obtained over the range from 500 to 590 m[mu] from measures of response rate during extinction. The gradients around each conditioned stimulus generally resemblec those obtained after training to a single stimulus. Results were compared with implications of alternative hypotheses (Hullian exponential summation versus nonsummation) regarding the manner in which overlapping gradients combine. A set of empirical functions relating the form of the single-stimulus generalization gradient to response strength ("generalization surface") was used to derive the 2-stimulus gradients required by the hypotheses. The 2 theories of combination appeared to account for the results almost equally well. Extension of the method of the present study to afford a more conclusive answer to the problem was suggested. With additional data, the generalization surface was found to be more in accord with the notion that changes in the generalization gradient with changes in response strength can be described by means of a simple multiplicative relation.
Keywords

This publication has 1 reference indexed in Scilit: