Neural coding of categories information efficienc

Authors: Laurent BonnasseGahot JeanPierre Nadal

Publish Date: 2008/01/31

Volume: 25, Issue: 1, Pages: 169-187

PDF Link

Abstract

This paper deals with the analytical study of coding a discrete set of categories by a large assembly of neurons We consider population coding schemes which can also be seen as instances of exemplar models proposed in the literature to account for phenomena in the psychophysics of categorization We quantify the coding efficiency by the mutual information between the set of categories and the neural code and we characterize the properties of the most efficient codes considering different regimes corresponding essentially to different signaltonoise ratio One main outcome is to find that in a high signaltonoise ratio limit the Fisher information at the population level should be the greatest between categories which is achieved by having many cells with the stimulusdiscriminating parts steepest slope of their tuning curves placed in the transition regions between categories in stimulus space We show that these properties are in good agreement with both psychophysical data and with the neurophysiology of the inferotemporal cortex in the monkey a cortex area known to be specifically involved in classification tasksThis work is part of a project “Acqlang” supported by the French National Research Agency ANR05BLAN006501 LBG acknowledges a fellowship from the Délégation Générale pour l’Armement JPN is a Centre National de la Recherche Scientifique member The initial motivation for this work comes from psycho and neuro computational issues in the perception of phonemes we thank Sharon Peperkamp and Janet Pierrehumbert for introducing us to this topic and for valuable discussions LBG is grateful to the members of the Laboratoire de Sciences Cognitives et Psycholinguistique de l’ENS especially to Emmanuel Dupoux for numerous and stimulating discussions We acknowledge useful inputs from the referees and most especially we thank one of them for a detailed list of constructive commentsWe follow the same approach as in Brunel and Nadal 1998 The first step consists in integrating over x Taking the large N limit we show that the leading order of the right term of Eq 34 is zero We then seek for the first correction using Laplace/steepest descent method The last step eventually consists in integrating over rEach cell has an activity r i equal to 1 if x is in θ i θ i + 1 and 0 otherwise The width of the ith tuning curves is thus a i  = θ i + 1 − θ i and we define the preferred stimuli as the center of the receptive fields x i  ≡ θ i  + θ i + 1/2The quantity mathcalH Q mux is zero on a homogeneous domain hence there is no contribution from intervals included in such domain Suppose on the contrary that on the full range under consideration mathcalH Q mux is rapidly varying We expect then the optimal θ i ’s distribution to be dense that is a i  = θ i + 1 − θ i small

Keywords: