Predictive synchrony organized by spike-based Hebbian learning with time-representing synfire activities

Abstract

In this chapter, we introduce a computational model to give a theoretical account for a phenomenon experimentally observed in neural activity of behaving animals. Pairs of neurons in the primary motor cortex exhibit significant increases of coincident spikes at times when a monkey expects behavioral events. The result provides an evidence that such a synchrony has predictive power. To investigate the underlying mechanism of such a predictive synchrony, we construct a computational model based on two known characteristics in the brain: one is the synfire chain, the other is spike-timing-dependent plasticity. The synfire chain is a model to explain a precisely firing spike sequence observed in frontal parts of the cortex. Synaptic plasticity, which is commonly believed a basic phenomenon underlying learning and memory, has been reported to depend on relative timings of neuronal spikes. In the proposed model, occurrence times of events are embedded in synapses from the synfire chains to time-coding neurons through spike-timing-dependent synaptic plasticity. We also discuss the robustness of the proposed mechanism and possible information coding in this cortical region.