NEURAL PATTERN-FORMATION IN NETWORKS WITH DENDRITIC STRUCTURE

We present a detailed analysis of a recently proposed model of neural pattern formation that is based on the combined effect of diffusion al ong a neuron's dendritic tree and recurrent interactions along axo-den dritic synaptic connections. For concreteness, we consider a one-dimen sional array of analog neurons with the dendritic tree idealized as a one-dimensional cable. Linear stability analysis and bifurcation theor y together with numerical simulations an used to establish conditions for the onset of a Turing instability leading to the formation of stab le spatial patterns of network output activity. It is shown that the p resence of dendritic structure can induce dynamic (time-periodic) spat ial pattern formation. Moreover, correlations between the dendritic lo cation of a synapse and the relative positions of neurons in the netwo rk are shown to result in spatially oscillating patterns of activity a long the dendrites of each neuron. Copyright (C) 1998 Elsevier Science B.V.