Parameter estimation for a resonant tectorial membrane cochlear model

Abstract

One of the basic problems of cochlear modeling is in systematically determining the parameters of the model from neural data. Frequency domain models give the response along the basilar membrane for pure‐tone inputs, whereas the data that are desirable to match are families of tuning curves as a function of frequency for many positions along the basilar membrane. This computational problem has been sidestepped by transforming the neural tuning curve data to place, forming neural excitation patterns. This allows for a major simplification of the data fitting process. If the neural data are accurately fit in the place domain for several well‐separated frequencies, then the frequency tuning data should fit as well. Using this approach, fits to neural tuning curves using passive cochlear models were greatly improved over the entire range of CFs. The model assumptions will be discussed and the results will be compared to measured data.