Information coding by ensembles of resonant neurons

Abstract

In the present paper, we propose a novel neu- ral procedure for signal processing and coding based on the subthreshold oscillations and resonance of the neural membrane potential that could be used by real neurons to perform frequency spectra analysis and information cod- ing of incoming signals. Taking into account the biophys- ical properties of the neural membranes, we note that the subthreshold resonant behaviour they exhibit can be used to analyse incoming signals and represent them in the fre- quency domain. We study the reliability of the representa- tion of signals depending on the biophysical parameters of the neurons, the fault-tolerance of this coding scheme and its robustness against noise and in the presence of spikes. The principal characteristics of our system are the use of the physical phenomenon of neural resonance (rarely con- sidered in the literature for signal coding); it fits well with the biophysical parameters of most neurons that exhibit subthreshold oscillations; it is compatible with experimen- tal data; and it can be easily integrated in a more general model of information processing and coding that includes communication between neurons based on spikes.