Synaptic noise and other sources of randomness in motoneuron interspike intervals.

Abstract

Listen

Translate article

IT IS GENERALLY AGREED that most long-distance information transmission in the nervous system is accomplished by the propagation of all-or-none spikes along axons, and further that this information is carried in a single axon by the timing of spikes. When a neuron is discharging with a constant mean frequency, information is probably contained in the length of interspike intervals which, however, are notoriously variable. Much recent attention has been focused on the statistics of this variability. In part, this interest has reflected a concern for the accuracy of nervous system function (interval variability, from this point of view, represents noise in the signal being transmitted (8)), but perhaps a larger body of the literature (19) has been concerned with using the theory of stochastic processes to gain insight into the processes generating the spike trains. To work backward from the statistics of interspike-interval variability to the mechanisms generating spike trains, however, involves making inferences about two separate mechanisms. Properties must be proposed for 7) the synaptic input of the neuron, and 2) the mechanism which converts synaptic input into spike trains. Variability in the interspike interval thus could be due to fluctuations in 7) the synaptic input, and 2) the mechanism converting synaptic currents into spike trains. Although Junge and Moore (15) were able to establish that intracellularly observed fluctuations were consistent with the observed interval variability, they were forced to pool data from many A&vsia neurons whose firing frequencies were drifting. Thus none of the models in the literature has utilized the measured properties of either the steady-state synaptic input or of the spike-generating mechanism of the very neuron whose interval variability they wished to predict.