Neural pattern formation in networks with dendritic structure

Abstract

We present a detailed analysis of a recently proposed model of neural pattern formation that is based on the combined effect of diffusion along a neuron's dendritic tree and recurrent interactions along axo-dendritic synaptic connections. For concreteness, we consider a one-dimensional array of analog neurons with the dendritic tree idealized as a one-dimensional cable. Linear stability analysis and bifurcation theory together with numerical simulations are used to establish conditions for the onset of a Turing instability leading to the formation of stable spatial patterns of network output activity. It is shown that the presence of dendritic structure can induce dynamic (time-periodic) spatial pattern formation. Moreover, correlations between the dendritic location of a synapse and the relative positions of neurons in the network are shown to result in spatially oscillating patterns of activity along the dendrites of each neuron.