Estimates of internal noise

Abstract

The concept of internal noise appears frequently in the literature of psychoacoustics. It has been used to explain the differences between the performance of human subjects and various ideal detectors and also to explain the variability in human performance. Internal noise has also appeared as a free parameter in many models of auditory processing. Watson [Ph.D. thesis, Indiana U. (1962)] estimated the ratio of internal to external noise by comparing the decisions of multiple subjects on a trial by trial basis. Swets et al. [J. Acoust. Soc. Am. 31, 514–521 (1959)] and Green [Psychol. Rev. 71, 392–407 (1964)] were able to estimate the ratio of internal to external noise for individual subjects by using reproducible noise as a stimulus. The present paper evaluates the performance of multiple subjects in a single interval detection task using computer generated reproducible noise as a masking stimulus. Multiple-subject and individual-subject methods of estimating internal noise are compared. General problems with estimates of internal noise are discussed. [Research supported by NSF research grant BNS-77-17308.]