A replication of Hovland's study of generalization to frequencies of tone.

Abstract

48 Ss were tested for generalization with 4 tones of the same frequencies as in Hovland''s classic study. The 16 tones were presented in a 4 x 4 Latin square during adaptation and testing for generalization. 1/2 of the Ss were conditioned to a tone at 1 end and 1/2 to a tone at the other end of the dimension. U-shaped curves of generalization were found for all testing cycles. No significant difference was found between electrodes on and off during generaliza-tion, nor between the exosomatic and the endosomatic method. It was concluded that testing for generalization is complicated by alerting reactions and stimulus-disparity effects, and that the overall evidence for monotonic decremental gradients of primary stimulus generalization of the GSR in humans is highly questionable, apart from a consideration of the exact shape of the curve.