Midbrain sensitivity to spectral cues for sound-source localization: A modeling study

Abstract

The features of head-related transfer functions (HRTFs) are a key topic in the field of spatial auditory displays. Neural encoding of HRTFs is not well understood; their sharp spectral peaks and notches are not well represented by average rates of relatively widely tuned auditory-nerve (AN) fibers, and phase-locking to the temporal fine structure is not adequate to encode features at high-frequencies. However, low-frequency fluctuations in cochlear responses to wideband stimuli are encoded in time-varying rates of AN fibers across all frequency channels. Here, we focus on modeling the profile of amplitude fluctuations in response to spectral cues across the population of AN channels. The fluctuation profile across AN responses sets up an average-rate profile across inferior colliculus (IC) neurons, which are sensitive to envelope-related low-frequency fluctuations. Both unilateral and bilateral IC model responses indicate that rather than responding to spectral peaks in a given HRTF, IC responses are more sensitive to fluctuations in frequency channels near steep spectral slopes. The influence of the spatial location and stimulus level on model IC responses was examined. Using statistical methods, psychoacoustical thresholds of discrimination of sounds that differ in the source location were estimated and compared to trends in the existing perceptual results. The features of head-related transfer functions (HRTFs) are a key topic in the field of spatial auditory displays. Neural encoding of HRTFs is not well understood; their sharp spectral peaks and notches are not well represented by average rates of relatively widely tuned auditory-nerve (AN) fibers, and phase-locking to the temporal fine structure is not adequate to encode features at high-frequencies. However, low-frequency fluctuations in cochlear responses to wideband stimuli are encoded in time-varying rates of AN fibers across all frequency channels. Here, we focus on modeling the profile of amplitude fluctuations in response to spectral cues across the population of AN channels. The fluctuation profile across AN responses sets up an average-rate profile across inferior colliculus (IC) neurons, which are sensitive to envelope-related low-frequency fluctuations. Both unilateral and bilateral IC model responses indicate that rather than responding to spectral peaks in a given HRTF, IC responses are mor...