Modelling the human cochlea

Abstract

One of the salient features of the human cochlea is the incredible dynamic range it possesses—the loudest bearable sound is 10,000,000 times greater than the softest detectable sound; this is in part due to an active process. More than twelve thousand hairlike cells known as outer hair cells are believed to expand and contract in time to amplify cochlear motions. However, the cochlea’s response is more than just the sum of its parts: the local properties of outer hair cells can have unexpected consequences for the global behaviour of the system. One such consequence is the existence of otoacoustic emissions (OAEs), sounds that (sometimes spontaneously!) propagate out of the cochlea to be detected in the ear canal. In this doctoral thesis, a classical, lumped-element model is used to study the cochlea and to simulate click-evoked and spontaneous OAEs. The original parameter values describing the microscopic structures of the cochlea are re-tuned to match several key features of the cochlear response in humans. The frequency domain model is also recast in a formulation known as state space; this permits the calculation of linear instabilities given random perturbations in the cochlea which are predicted to produce spontaneous OAEs. The averaged stability results of an ensemble of randomly perturbed models have been published in [(2008) ‘Statistics of instabilities in a state space model of the human cochlea,’ J. Acoust. Soc. Am. 124(2), 1068-1079]. These findings support one of the prevailing theories of SOAE generation. Nonlinear simulations of OAEs and the model’s response to various stimuli are performed in the time domain. Features observed in the model include the saturation of the forces generated by the OHCs, compression of amplitude growth with increasing stimulus level, harmonic and intermodulation distortion, limit cycle oscillations that travel along the cochlear membranes, and the mutual suppression of nearby linear instabilities.

More information

Identifiers

Catalogue record

Export record