Propagation of synchronous spiking activity in feedforward neural networks

Abstract

'Synfire' activity has been proposed as a model for the experimentally observed accurate spike patterns in cortical activity. We investigated the structural and dynamical aspects of this theory. To quantify the degree of synchrony in neural activity, we introduced the concept of 'pulse packets'. This enabled us to derive a novel neural transmission function which was used to assess the role of the single neuron dynamics and to characterize the stability conditions for propagating synfire activity. Thus, we could demonstrate that the cortical network is able to sustain synchronous spiking activity using local feedforward (synfire) connections. This new approach opens the way for a quantitative description of neural network dynamics, and enables us to test the synfire hypothesis on physiological data.