Stochastic resonance and spike-timing precision in an ensemble of leaky integrate and fire neuron models

We analyze the transmission of sinelike periodic signals by an ensemble of leaky integrate-and-fire neuron models in the presence of additive noise. W e observe that when the number of units in the ensemble is large enough, th e point process formed by pooling the spike trains of all units is an inhom ogeneous Poisson process. We obtain the intensity of this process, i.e., th e instantaneous discharge rate of the ensemble, from the cycle histogram of the discharge of a single unit. This enables us to link measures of the re gularity of the output discharge rate and the transmission of the periodic input. such as the signal to noise ratio and the input-output power norm an d normalized power norm directly to the shape of the cycle histogram. Furth ermore, we also show that firing precision in response to subthreshold stim ulation is maximized at some intermediate noise value, and argue that in th is regime the ensemble can reliably transmit fast periodic signals below th e resolution of the individual units. Our analysis clarifies the conditions whereby noise enhances signal transmission and detection in ensembles. [S1 063-651X(99)09803-7].