Percolation approach to formation of synfire chains in two dimensional neural networks

Abstract

We consider the propagation of spike packets in two dimensional networks consisting of locally coupled neural pools. The dynamic attractors of this model, synfire chains, appear for some values of network parameters. The synfire chain formation exhibits behavior, which may be discribed with the percolation phase transition. Using finite-size scaling method, we obtained the critical probabilities and the critical parameter ratio ß / v for different sets of refractoriness and synaptic weights, connecting neighbouring neural pools. © 2006 IEEE.