In young humans, the upper limit of the hearing range is approximately 18 kHz, but in most mammals the range can extend to 100 kHz. The basic physiological tuning mechanism depends on how sound propagates in the fluid of the cochlear duct: it excites a place specific vibration of the basilar membrane (BM) that is enhanced by a population of sensory hair cells, outer hair cells (OHCs). These cells exhibit what is termed ‘electromotility’. Subsequently inner hair cells (IHCs) signal tonotopic information to the brain. Although qualitatively simple, quantitative descriptions are complex and multiscale because 1) the OHC mechano-transduction step is non-linear; 2) there may be intrinsic bandwidth limits to the sensory process either from filtering by the membrane times constant or from inherent limits of the underlying molecular ‘motor (prestin/SLC26A5) and 3) there is limited access to all points along the cochlea and consequent sparse data, although now improving with better imaging technologies.
The approach here is to show that some of these issues can be overcome using a model that incorporates realistic cochlear geometry (Geisler, 1993) and using parameters from a variety of sources. The essential model features include 1) the OHC length which depends, independent of species, on their characteristic frequency; 2) the orientation of the tectorial membrane (TM) which is angled more steeply to the BM as the characteristic frequency increases; 3) a ‘piezo-electric’ property (reverse electromotility) of OHCs, whereby longitudinal force on the cells generates current that compensates for the membrane time constant; 4) an intrinsic time constant (ca 16 us) of relaxation of the prestin/SLC26A5 driving OHC electromotility (Gale & Ashmore, 1997). A Green’s function approach is used to incorporate the effects of fluid coupling between adjacent sections. Reasonable cochlear frequency tuning curves, agreeing with experimental data, can be obtained for both low and high frequencies up to 70 kHz. Age-related human hearing loss may depend on some of these mechanism being compromised.
All computations are performed in a linear approximation appropriate for threshold measurement. A graphic interface written in Matlab will be available on request.