We propose to build upon our extensive body of results by exploring the biophysical bases and functional utility of various peripheral nonlinear phenomena. Direct measurements of basilar membrane motion demonstrate that, at low sound levels, the response can be sharply tuned, becoming more broadly tuned at high levels. This nonlinear tuning is due to the outer hair cells, which can both sense mechanical stimuli, and respond to changes in membrane potential with changes in length. This novel form of electromotility is piezoelectric in nature allowing it to achieve very high velocities [. Mountain Hubbard 1994.]. We have developed (computationally intensive) hydromechanical active models in which the outer hair cells interact with the membrane mechanics to create a traveling-wave amplifier [.Hubbard 1993.]. To simulate them efficiently, these biophysical models are usually abstracted as digital filters with or without feedback depending on the biophysical accuracy desired [. Carney 1993 , Hubbard Mountain 1996, Yang Shamma Wang 1992.].
As part of this MURI project we propose to compare a range of filter-bank models to the hydromechanical models in order to find the best compromise between physiological accuracy and computational efficiency. This effort will be closely coordinated with the psychoacoustically-based model developed in Thrust area III(A). We will also continue the development of new formulations of the coupling between the mechanical motion of the basilar membrane and of the inner hair cell bundles [. Mountain Cody 1989, Shamma Wilbur Rinzel Chadwick 1986.], and of the adapting hair cell synapse to the auditory-nerve [. Carney 1993, Zagaeski Cody Russell Mountain 1994.]. These models accurately predict the driving forces on the inner hair cells and auditory-nerve, but must be optimized for computational efficiency, and compared to the minimal models described next.