Computational and statistical analysis of auditory peripheral processing for vowel-like signals

The auditory nerve (AN) transfers information about sound stimuli from the inner ear to the brainstem, which further projects to higher levels of the auditory nervous system. A detailed knowledge of the discharge patterns of AN fibers is important for understanding how sounds are encoded at the input stage of the auditory system. The study presented here investigates neural mechanisms for encoding and processing sound signals, including harmonic complexes and synthesized speech signals. A better understanding of the sound encoding mechanisms in the auditory system can benefit many other related studies, such as the building of speech-signal processors or recognizers as well as the design of hearing aids.; The initial focus of this study is the coding of complex sounds in a quiet background. The analysis is then extended to the coding of complex sounds in a noisy background. Two different approaches, one directly based on the discharge pattern of AN fibers and the other one based on the activity of a coincidence-detection mechanism operating on the AN fiber patterns, are used to evaluate the role of signal processing in the auditory periphery, which includes the outer, middle, and inner ears. Predictions based on average discharge count information (the number of neural action potentials in one trial) are compared with results based on the mechanisms using both count and temporal information (the times of neural action potentials). The results show that the predictions using both count and temporal information are more consistent with the psychophysical data than the predictions based on average-count information. The results of this study also suggest that the prediction based on only a small population of the AN model fibers, which have characteristic frequencies near the signal frequency, can account for the results of the hearing experiments discussed in this project. The trends of the predicted thresholds based on the coincidence-detection mechanism are similar to those based directly on the discharge patterns of AN fibers.