Objective assessment of the functional role of spike train correlations using information measures

Abstract

The functional role of correlations between neuronal spike trains remains strongly debated. This debate partly stems from the lack of a standardized analysis technique capable of accurately quantifying the role of correlations in stimulus encoding. We believe that information theoretic measures may represent an objective method for analysing the functional role of neuronal correlations. Here we show that information analysis of pairs of spike trains allows the information content present in the firing rate to be disambiguated from any extra information that may be present in the temporal relationships of the two spike trains. We validate and illustrate the method by applying it to simulated data with variable degrees of known synchrony, and by applying it to recordings from pairs of sites in the primary visual cortex of anaesthetized cats. We discuss the importance of information theoretic analysis in elucidating the neuronal mechanisms underlying object identification.