Characterizing cochlear mechano-electric transduction using a nonlinear systems identification procedure

A nonlinear systems identification technique [J. S. Bendat, Nonlinear System Analysis and identification from Random Data (Wiley, New York, 1990)] provided the frequency-domain parameters of a third-order system polynomial equation describing cochlear mechano-electric transduction (MET) in Mongolian gerbils. The magnitude of the linear system term was the largest followed by the cubic and quadratic terms. The phase of the linear and cubic system terms differed by approximately 180 deg and the phase of the quadratic term lay between. Between-animal and within-animal variability was smallest for the linear and cubic terms, and largest for the quadratic term. Summing linear and nonlinear coherence functions revealed that the third-order system polynomial equation characterized approximately 83% of MET for low frequencies and 92% for high frequencies. Animals exposed to a 4-kHz pure tone at 100 dB SPL for 20 min showed changes in the magnitude and phase of the parameters of the third-order polynomial equation. An increase in linear coherence and a decrease in nonlinear coherence occurred at frequencies centered at the exposure frequency. Below the exposure frequency, linear coherence decreased and nonlinear coherence increased. Summation of the coherence functions showed that the third-order polynomial equation characterized MET better after exposure to the pure tone. (C) 1996 Acoustical Society of America.