Effects of stimulus level and background noise on vowel representations in the auditory brain stem of cats

Abstract

Effects of stimulus level and background noise on vowel representations in the ventral cochlear nucleus (VCN) were investigated by recording single-unit discharge rates in anesthetized cats. The quality of vowel encoding by populations of neurons can be evaluated by plotting vowel-driven rates as a function of each unit’s best frequency (BF, the most sensitive frequency); the resulting rate profiles are assumed to show good representations of formant structure if there are high driven rates at BFs near formant frequencies and low rates at BFs near spectral troughs. Among the VCN unit types recorded in the present study, chopper units exhibited the most detectable formant-to-trough rate differences across vowel levels and in background noise. By contrast, primarylike units produced vowel representations that reflected the limited dynamic range of auditory-nerve fibers [May et al., Aud. Neurosci. 3, 135–162 (1996)]. Primarylike units with low spontaneous rates (SR<18 sp/s) showed better representations than high SR primarylike units, but failed to respond at low vowel levels. These results suggest that peripheral representations of speech sounds are sharpened by patterns of auditory nerve convergence in the cochlear nucleus. [Work supported by NIDCD Grant No. DC00109.]