Information-theoretic analysis of neural coding

Abstract

We describe a family of new techniques for analyzing single- and multi-unit discharge patterns. These techniques are based on information theoretic distance measures and on empirical theories derived from work on universal signal processing. They are capable of determining transneuron statistical dependencies even when time-varying responses occur. The response portion contributing most to information coding can be identified and the coding fidelity can be quantified regardless of the neural coding mechanisms-be it timing, rate or transneural correlations.