Spectral cues for sound localization in cats: a model for discharge rate representations in the auditory nerve

Neural representations of pinna-based spectral cues for sound localization were modeled by simulating auditory nerve discharge rates to noise bursts that had been shaped by filtering properties of the cat's head-related transfer functions (HRTFs) at 179 locations in the frontal field. The auditory nerve model transformed spectral differences between HRTFs into simulated neural rate differences. Linear equations for this transformation were developed from actual auditory nerve responses to a limited subset of HRTF-filtered noise bursts [Rice et al., J. Acoust. Soc. Am. 97, 1764-1776 (1995)]. Signal detection methods were used to investigate simulated neural responses to pairwise changes between HRTFs. The quality of neural representation for these changes, in terms of d' values, declined when the reference HRTF was moved from a central location (0 degree AZ, 0 degree EL) to a large positive azimuth in the horizontal plane (75 degrees AZ, 0 degree EL) or a high elevation in the median plane (0 degree AZ, 75 degrees EL). Most simulated responses exhibited large d' values for comparisons of contralateral versus ipsilateral azimuths, or eccentric versus frontal elevations. This rate information resulted from directionally dependent changes in the overall gain of HRTFs. In addition, fibers with best frequency (BF: the frequency of greatest sensitivity for individual fibers) between 5 and 18 kHz showed large d' values for HRTF contrasts in the immediate frontal field because of the effects of spectral notches (i.e., sharp drops in gain over a narrow frequency range). Spectral notches also played a prominent role in simulations that required identification of HRTF location in the absence of a fixed reference stimulus. These modeling results correspond well with previously described patterns in the cat's localization behaviors.