Computational modeling of otoacoustic emissions

Abstract

The mammalian ear is a remarkable sensory system with wide dynamic range and sharp frequency selectivity. These impressive characteristics are products of an active feedback mechanism within the cochlea. A consequence of this active feedback mechanism is that, in addition to detecting sound, the cochlea can also generate sounds, called otoacoustic emissions, that can be measured in the ear canal to provide a noninvasive assessment of hearing. There are several types of otoacoustic emissions, including distortion product otoacoustic emissions (that are generated in response to a stimulus of two tones) and spontaneous otoacoustic emissions (self-sustained oscillations that occur without any applied stimulus). Distortion product otoacoustic emissions are commonly used clinically and in research labs; however, key aspects of how these emissions are generated and propagate within the cochlear are debated. In this thesis, the properties of distortion product generation and propagation are investigated using a computational model of the mammalian cochlea. Additionally, this computational model is used to demonstrate that altering the viscoelastic properties of the tectorial membrane (a structure in the cochlea) affects spontaneous otoacoustic emission generation and the ability of the cochlea to detect low level sounds.