Persistent Activation Blobs in Spiking Neural Networks with Mexican Hat Connectivity

Abstract. Short range excitation, long range inhibition sometimes referred to as mexican hat connectivity seems to play important role in organization of the cortex, leading to fairly well delineated sites of activation. In this paper we study a computational model of a grid filled with rather simple spiking neurons with mexican hat connectivity. The simulation shows, that when stimulated with small amount of random noise, the model results in a stable activated state in which the spikes are organized into persistent blobs of activity. Furthermore, these blobs exhibit significant lifetime, and stable movement across the domain. We analyze lifetimes and trajectories of the spots, arguing that they can be interpreted as basic computational charge units of the so called spike flow model introduced in earlier work. Key words: spiking networks, mexican hat connectivity, spike flow model

• Type: Conference
• Year: 2010
• Where: ICAISC
• Authors: Filip Piekniewski
• Keywords: Artificial Intelligence, ICAISC 2010, Mexican Hat Connectivity, Simple Spiking Neurons, Spike flow Model