WHY SPIKES - HEBBIAN LEARNING AND RETRIEVAL OF TIME-RESOLVED EXCITATION PATTERNS

Hebbian learning allows a network of spiking neurons to store and retr ieve spatio-temporal patterns with a time resolution of 1 ms, despite the long postsynaptic and dendritic integration times. To show this, w e introduce and analyze a model of spiking neurons, the spike response model, with a realistic distribution of axonal delays and with realis tic postsynaptic potentials. Learning is performed by a local Hebbian rule which is based on the synchronism of presynaptic neurotransmitter release and some short-acting postsynaptic process. The time window o f this synchronism determines the temporal resolution of pattern retri eval, which can be initiated by applying a short external stimulus pat tern. Furthermore, a rate quantization is found in dependence upon the threshold value of the neurons, i.e., in a given time a pattern runs n times as often as learned, where n is a positive integer (n greater- than-or-equal-to 0). We show that all information about the spike patt ern is lost if only mean firing rates (temporal average) or ensemble a ctivities (spatial average) are considered. An average over several re trieval runs in order to generate a post-stimulus time histogram may a lso deteriorate the signal. The full information on a pattern is conta ined in the spike raster of a single run. Our results stress the impor tance, and advantage, of coding by spatio-temporal spike patterns inst ead of firing rates and average ensemble activity. The implications re garding modelling and experimental data analysis are discussed.