Automatic learning of synchrony in neuronal electrode recordings

摘要

Synchrony among neuronal impulses (or spikes) plays, according to some of the most prominent neural coding hypotheses, a central role in information processing in biological neural networks. When dealing with multiple electrode recordings (i.e., spike trains) modelers generally characterize synchrony by means of a maximal time span (since exact spike-time coincidences cannot be expected): two or more spikes are regarded as synchronous if they lie from each other within a distance at most this maximal time span. Such time span is determined by the modeler and there is no agreement about how long it should be. In this paper we present methodology to learn this time span automatically from spike-train data that involves the assessment of the amount of synchrony in the database (relative to that expected if spike trains in it were uncorrelated) and a learning process that looks at the time span that maximizes it (over all those considered).