Signal transformation and coding in neural systems

Abstract

The subject of signal transformation and coding in neural systems is fundamental in understanding information processing by the nervous system. This paper addresses this issue at the level of neural units (neurons) using nonparametric nonlinear dynamic models. These models are variants of the general Wiener-Bose model, adapted to this problem as to represent the nonlinear dynamics of neural signal transformation using a set of parallel filters (neuron modes) followed by a binary operator with multiple real-valued operands (equal in number to the number of modes). The postulated model constitutes a reasonable compromise between mathematical complexity and current neurophysiological evidence. It incorporates nonlinear dynamics and spike generation mechanisms in a fairly general, yet parsimonious manner. Although this study has objectives limited to a single unit and represents a small contribution in a vast and complex research area, it is hoped that it will facilitate progress in the systematic study of the functional organization of neural systems with multiple units.