Information theoretic analysis of action potential trains. I. Analysis of correlation between two neurons.

A crosscorrelational method of action potential trains has been proposed, based on information theory. Two information theoretic quantities, mutual information and channel capacity, were calculated from a pair of action potential trains for detecting a crosscorrelation and estimating synaptic connectivity. The method was compared with conventional ones, using action potential trains obtained by the simulation of a neuron model. This method was shown to have the advantages to more easily find a weak but significant crosscorrelation and to give better estimation of synaptic connectivity independent of the firing probability of a presynaptic neuron.