Rate representation and discriminability of second formant frequencies for /epsilon/-like steady-state vowels in cat auditory nerve

Abstract

Alternate forms of the steady-state vowel /epsilon/ with second formant peaks located at 1400, 1500, 1700, and 2000 Hz were used to study the representation and discrimination of second formant frequencies at the level of the auditory nerve. Recordings from large populations of auditory nerve fibers in response to these stimuli were used to create rate-place plots, which show second formant peaks that resembled the stimulus spectra. Measures of the peak amplitude decreased as sound level was increased and as second formant frequency was lowered. Representation of the spectra was degraded at the higher sound level because of saturation and two-tone suppressive effects. However, formant peaks were clearly represented in plots of rate differences between two vowels. Such plots resemble the ratio of the magnitudes of the two vowel spectra. The results suggest that information concerning the position of formant peaks is present in the average discharge rate of the auditory nerve. A measure of discriminability, d', between vowel pairs was also calculated. Second formants differing by 125-240 hz can be discriminated using the rate responses of individual fibers that are optimally placed on the basilar membrane; the estimated second formant jnd for the whole auditory nerve is approximately 1 Hz.