Modeling Loudness Perception in Electrical Hearing with a Phenomenological Auditory Nerve Model
* Presenting author
The development of new coding strategies for electrical stimulation in cochlear implants (CI) motivates the demand for models of stimulated auditory nerve fibers (ANF), which can be fitted to individual subjects. With the intention of predicting patients’ loudness percepts for biphasic pulse trains, a phenomenological ANF model based on a stochastic leaky integrator approach (Takanen and Seeber, 2017), which takes into account a variety of temporal (refractoriness, rate adaptation, accommodation, facilitation) and spatial (spread of excitation) response properties of the neurons, was combined with a widely accepted loudness model for electrical stimulation (McKay et. al., 2001). The latter considers both temporal and spatial loudness summation effects and takes the neuronal response spread over the whole cochlea as input. It was adapted to incorporate the stochastically fluctuating output of the leaky integration and includes a definition of loudness threshold and maximum comfortable level (MCL) based on the firing probability width along the electrode array. In past work at AIP the ANF model was fitted to individual patients using measurements of the electrically evoked compound action potential (ECAP). The present work extends the approach to reproduce psychophysical loudness data, individual loudness growth and effects on loudness of stimulation rate and phase duration.