“Confidence-Rating” Analysis on Response Amplitude Scores of Classical Discrimination Conditioning Data

Abstract

Signal-detection theory was used to analyze classical discrimination-conditioning data. A “confidence-rating” type of analysis of response amplitudes was used to test the signal-detection-theory assumptions of underlying normal distributions with equal variances. The data showed that S behaved as if he could hold several criteria simultaneously and gave larger amplitude conditioned responses when the sensory response exceeded larger criterion values. In addition, d' did not vary with conditioned-response amplitude. Plotting the receiver-operating-characteristic curves generated by the analysis of conditioned-response amplitude on double-probability paper supported the assumptions of signal-detection theory of underlying normal distributions with equal variances.