Modelling Synfire Processing

I give a review of theoretical attempts to understand how synfire processing works, including an estimate of the capacity of a network the size of a cortical column and an analysis of the stability of spike synchronisation. The treatment is at the level of simple signal-to-noise analyses. NMDA-receptor synaptic currents are included in the treatment. They are found not to affect the capacity or the dynamical stability strongly. 1 Introduction There has been a lot of interest recently in the role of spike timing in neural computation. For many years, the working hypothesis of experimental neuroscience was that neuronal spike trains were Poisson processes, so information was conveyed only in average firing rates. In many situations this still appears to be a very good approcximation, provided the period over which the average firing rate is computed is chosen appropriately [1]. However, there is growing evidence that under some circumstances precise spike timing precision may be a cruci..