Decoding the population responses of retinal ganglions cells using information theory

Abstract

In this paper information theory is used to analyze the coding capabilities of populations of retinal ganglions cells for transmitting information. The study of a single code versus a distributed code has been contrasted using this methodology. The redundancy inherent in the code has been quantified by computing the bits of information transmitted by increasing number of neurons. Although initially there is a linear growth, when certain numbers of neurons are included, information saturates. Our results support the view that redundancy is a crucial feature in visual information processing. © Springer-Verlag Berlin Heidelberg 2001.