Temporal coding of resonances by low-frequency auditory nerve fibers: single-fiber responses and a population model

1. We recorded responses of low-frequency auditory nerve fibers (characteristic frequency (CF) less than 3 kHz) in the cat to resonant stimuli with varied natural frequencies, damping coefficients, and sound pressure levels. Responses to resonances were synchronized to frequencies lying between the peak frequency of the stimulus spectrum and a frequency near the fiber's CF. The frequency of the dominant synchrony in the response varied systematically as a function of the stimulus parameters. 2. More lightly damped resonances, which have sharp spectral peaks, elicited synchrony closer to the peak frequency, whereas the broader peaks of more highly damped resonances elicited synchrony closer to the fiber's CF. Thus as the stimulus was varied from an undamped tone to a highly damped transient, the dominant component of the synchronized response moved from the peak frequency of the stimulus toward the CF of the fiber. The trajectory of the dominant component varied as a function of stimulus level, with higher levels resulting in synchrony biased toward the peak of the stimulus spectrum over a wider range of damping. 3. The frequency tuning and synchronization characteristics of a fiber, along with the stimulus parameters, determined the temporal properties of its response to complex stimuli. Using reverse correlation (revcor) filters to characterize the tuning and synchronization of auditory nerve fibers, we were able to predict the temporal properties of responses to resonant stimuli. 4. A parametric model was fit to measured revcor functions derived from responses of auditory nerve fibers to wideband noise. In this way, a bank of model revcor filters was developed based on our population of measured filters. 5. The filter bank was used to model the response of a population of auditory nerve fibers to resonances. Temporal patterns present in the response of a population of fibers encoded the parameters of resonant stimuli. 6. The model revcor filter bank provided a means of studying temporal response patterns of the population of fibers to other complex sounds. 7. The output of the population model is a representation of the temporal information provided by the auditory periphery to the central nervous system; thus it provides a potentially useful tool for testing hypotheses concerning the processing of temporal information by the central auditory system.(ABSTRACT TRUNCATED AT 400 WORDS)