The Impact of Sensitivity to Fast Spectrotemporal Chirps on Speech Encoding in the Mammalian Inferior Colliculus

点击翻译按钮获取中文摘要

Sensitivities of the mammalian auditory midbrain to complex sound features can provide the basis for encoding schemes of critical stimuli such as speech. Recent work in rabbit inferior colliculus (IC) has revealed a novel feature sensitivity to fast spectrotemporal chirps. Individual IC neurons had dramatic differences in response rate to differing chirp directions and velocities. The stimulus used in that work was the Schroeder-phase harmonic complex (SCHR), which contains a chirp within each pitch period because of the phase properties of harmonic components. Speech necessarily contains similar chirps due to phase differences between harmonic components resulting from vocal tract filtering. Given that IC neurons may be as sensitive to chirps in vowels as they are to SCHR chirps, we hypothesize that novel chirp sensitivity in the IC significantly contributes to speech coding at the level of the auditory midbrain. The goal of this proposal is to more precisely define this sensitivity in the mammalian central auditory system to fast chirps using physiological experiments, design an IC computational model that accurately reflects chirp sensitivity, and to assess whether chirp sensitivity plays a role in speech coding. We will design novel sound stimuli to interrogate IC chirp sensitivity more directly, and to disentangle chirp sensitivity from other known IC properties, such as periodicity tuning. Computational models will be proposed using a base assumption that chirp-sensitive IC neurons receive two or more inputs of differing frequencies and respond differently depending on input arrival times. We will analyze chirp cues contained in speech using novel analysis methods. Finally, we will evaluate the importance of chirp sensitivity for speech coding by comparing the ability of a chirp-sensitive and -insensitive models to predict physiological speech responses. This work, when completed, will result in a new understanding of how the subcortical auditory system encodes speech and complex sounds. This work will also represent one of the first attempts to characterize and define this novel midbrain phenomenon. The findings of this proposal will help identify new stimulus features that may be used as cues for next-generation hearing aids to assist speech comprehension in noisy environments. The proposed training plan will concentrate on improving skills such as computational neuroscience, advanced signal processing, data analysis, statistical hypothesis testing and experimental design, all skills necessary for the completion of the proposed research. Training will also address communication skills, such as verbal and written presentations, which will be demonstrated by participation in conferences and generation of journal publications. The University of Rochester represents an ideal location to complete the proposed research, providing an environment with easy access to experts in auditory neuroscience and related fields, as well as many opportunities to participate in career-building activities, workshops, and seminars to help reach professional goals.