Detection Theory and Psychophysics

Part I: Traditional psychophysical models and Tanner and Swets' model are critically examined. Some of the weaknesses of these models are found to be eliminable by a more sophisticated analysis in terms of detection theory. Accordingly, psychophysical methods are re-examined, and the two-category forced-choice technique is found to be particularly advantageous on theoretical grounds. Application of detection theory to the problem of auditory masking with gaussian noise as measured by this forced-choice technique leads to the mathematical derivation of the theoretical ("ideal detector") psychophysical function for this situation. Part II: Experiments using the forced-choice method with auditory signals masked by broadband gaussian noise are reported. The aim of these experiments is to determine the extent to which and the manner in which subjects differ from the "ideal detector" of detection theory. It is found that, except for being approximately 13 db less sensitive, subjects behave very much like the ideal detector that is, in accordance with the mathematical predictions of Part I when the signals are pure tones. Results with signals consisting of two-component tones require a somewhat enlarged model; such a model is developed.