Self-organization process in large spiking neural networks leading to formation of working memory mechanism

Abstract

The subject of this work is evolutionary process in initially chaotic and homogenous spiking neural networks leading to formation of the neuron groups with partially synchronized activity (so called polychronous groups) which are not only capable of recognizing input patterns but also can keep information about pattern presentation in form of their specific activity for a long time. This result is demonstrated for very simple neuron - coincidence detector and for standard synaptic plasticity model (STDP). © 2013 Springer-Verlag Berlin Heidelberg.