Visual attention using spiking neural maps

Visual attention is a mechanism that biological systems have developed to reduce the large amount of visual information in order to efficiently perform tasks such as learning, recognition, tracking, etc. In this paper, we describe a simple spiking neural network model that is able to detect, focus on and track a stimulus even in the presence of noise or distracters. Instead of using a regular rate-coding neuron model based on the continuum neural field theory (CNFT), we propose to use a time-based code by means of a network composed of leaky integrate-and-fire (LIF) neurons. The proposal is experimentally compared against the usual CNFT-based model.