Does Norwich's Entropy Theory of Perception avoid the use of mechanisms, as required of an information-theoretic model of auditory primary-afferent firing

Abstract

Norwich et al.'s Entropy Theory (1975-present) stipulates that (1) the auditory receptor is uncertain about the stimulus' intensity, (2) uncertainty, given by the “entropy” H, relates to primary afferent neuronal firing-rate F as F=constant×H. H was calculated according to Garner & Hake (Psychological Review 58, 1951; after Shannon, Bell System Technical Journal 27, 1948) using Information Theory. H is available any time events, their outcomes, and the appropriate probabilities of occurrence are known, precisely because Information Theory ignores mechanisms, such as sensory transduction. Hence, any “information-theoretic” hearing model must avoid physiological mechanisms, and Norwich et al. claim to comply. But do they? The Entropy Theory was scrutinized for avoidance of physiological mechanisms. Norwich et al. stipulated throughout that (1) “We can envisage a steady sensory stimulus as a stationary stochastic sequence of microscopic sensory events” (Norwich & Wong, Annales de la Fondation Louis de Broglie 22, 1997), (2) those events (representing different intensities) are repeatedly sampled by the receptor. Sampling is a mechanism, not separable from transduction, the event's (presumed) “outcome”. Also, Norwich et al.'s attempts to separate events from outcomes produces inconsistent identification of “events.” The Entropy Theory is not an information-theoretic model of primary-afferent firing.