SPIKE TRAIN PROCESSING BY A SILICON NEUROMORPH - THE ROLE OF SUBLINEAR SUMMATION IN DENDRITES

A dendritic tree, as part of a silicon neuromorph, was modeled in VLSI as a multibranched, passive cable structure with multiple synaptic si tes that either depolarize or hyperpolarize local ''membrane patches,' ' thereby raising or lowering the probability of spike generation of a n integrate-and-fire ''soma.'' As expected from previous theoretical a nalyses, contemporaneous synaptic activation at widely separated sites on the artificial tree resulted in near-linear summation, as did neig hboring excitatory and inhibitory activations. Activation of synapses of the same type close in time and space produced local saturation of potential, resulting in spike train processing capabilities not possib le with linear summation alone. The resulting sublinear synaptic summa tion, as well as being physiologically plausible, is sufficient for a variety of spike train processing functions. With the appropriate arra ngement of synaptic inputs on its dendritic tree, a neuromorph was sho wn to discriminate input pulse intervals and patterns, pulse train fre quencies, and detect correlation between input trains.