Effects of nonlinearities on speech encoding in the auditory nerve

The representation of steady‐state vowels in terms of both average rate and temporal aspects of the discharge patterns of populations of auditory‐nerve fibers is discussed. The effects of auditory‐nerve nonlinearities on this representation is emphasized. Aspects of two rate‐related nonlinearities, rate saturation and two‐tone suppression, are reviewed. At low sound levels, profiles of discharge rate versus characteristic frequency in populations of auditory‐nerve fibers show well‐defined peaks at frequencies corresponding to the formants of a vowel stimulus. At levels above about 60 dB SPL, these peaks are not seen because of a combination of rate saturation and two‐tone suppression. Units with spontaneous rates less than 1/s show the effects of suppression more dramatically than do units with higher spontaneous rates; nonetheless, this population can retain formant peaks in its rate profiles up to levels at least 20 dB higher than does the higher spontaneous rate population. Aspects of phase‐locking to tones and two‐tone combinations, including synchrony suppression, are reviewed. Temporal patterns of responses to vowels are analyzed in terms of interval histograms. This analysis shows that in general, phase‐locking to the formant frequencies is stronger than locking to other harmonics of the stimulus. As sound level increases, synchrony to the formant frequencies saturates. However, responses to nonformant harmonics are suppressed by responses to the formant harmonics. This synchrony suppression allows the phase‐locking to the formant frequencies to remain dominant even at high levels. The exception to this behavior is locking to harmonics and intermodulation products of the first two formant frequencies (2F₁, 3F₁, F₂‐F₁, etc.) which increases rapidly in amplitude at higher sound levels. These distortion products are, to some extent, a consequence of the rectification inherent in hair cell/nerve fiber transduction and may also reflect the presence of propagating combination tones.