Nonlinear feedback models for the tuning of auditory nerve fibers.

Abstract

The tuning of auditory nerve (AN) fibers is generally characterized by an increase in bandwidth and, for mid- to high-frequency fibers, a downward shift in the center frequency as sound level increases. Changes in bandwidth are accompanied by changes in the phase properties of the fibers; thus the timing of neural discharges also changes as a function of sound level. This study focuses on the magnitude and phase properties of models designed to reproduce the nonlinear properties of AN fibers that were studied electrophysiologically. The forward path of each model consisted of a linear second-order resonance, and each feedback path contained a saturating nonlinearity. In model 1, the feedback path was a simple memoryless, saturating nonlinearity. In model 2, a low-pass filter was added after the feedback nonlinearity. The ability of each model to simulate aspects of the nonlinear tuning of AN fibers is discussed. Model 2 was able to simulate a wider range of nonlinear behavior for different AN fibers and thus has promise for use in simulations of populations of fibers tuned to different frequencies.