Auditory Processing with spatio-temporal codes

Abstract

A B S T R A C 7 We are currently developing a model of auditory processing including several specialized modules connected partly in series, partly in parallel. Signal is first decomposed between frequency channels in the cochlea, and transduced into spike trains which are then directed towards auditory centres where they are processed by neurons with various response characteristics, with either a preference for tonic behavior synchronized on the frequency components of the incident stimulation, or for phasic responses. A number of signal characteristics are exhaustively mapped, such as frequency, amplitude modulation, intensity, interaural delays or timing between acoustic events, and these intermediary representations further converge towards decoding networks. We insist all along this pathway on the necessity to cope with the intrinsic temporal characteristics of the spike trains, and we introduce processing mechanisms based on coincidence computations, which could dcal with both time and space in a natural \sa)’.