Cochlear compressive nonlinearities introduce level-dependent effects in sound processing by the inner ear, which are perceptually relevant for normal hearing and altered with sensorineural hearing loss. Electrocochleography may provide a means to assess nonlinearities of the human cochlea with moderate invasiveness. We recently developed a model for forward-masked Compound Action Potential (CAP), which was successful on predicting CAP waveforms recorded at the round window of chinchillas associated with notched-noise maskers of different notch widths and attenuations. The method relied primarily on the estimation of masking as a function of intensity at the cochlear-filter output. Based on these functions, cochlear 'excitation patterns' were defined and convolved by a unitary response to predict the waveforms of forward-masked CAPs. However, if we remove the low frequency suppressor of the notched-noise maskers (i.e., using derivatives of high-pass noise maskers), differential suppression effects become too important and the linear model acting as a cochlear filter bank becomes insufficient. In this preprint, we illustrate these shortcomings and we show that the inclusion of compression in our model can mitigate these issues.

Funding:Fondation Pour l'Audition: FPA RD-2019-3National Institutes of Health: 5R01DC009838National Institutes of Health: 5T32DC016853