Statistical analysis of multipath neural systems

Publisher Website
Google Scholar
Abstract
Peripheral sensory and motor systems may be characterized by models consisting of multiple parallel convergent pathways, each described by the same set of equations, but having different parameter values in each path. Such models, although deterministic, are best analyzed using a statistical approach, which is illustrated here by analysis of several simple multipath models composed of linear dynamic elements and static non-linear elements. Relationships between instantaneous means of signals at different points in such systems are used to show that a multipath system can exhibit behaviour which would not be expected from observations of individual pathways. Mechanisms for linearization of static non-linearities are briefly described. Important implications for neurophysiologists are discussed.