Dissertation Defense Announcement
School of Communication Sciences and Disorders announces the Final Dissertation Defense of
for the Degree of Doctor of Philosophy
June 15, 2018 at 12:00 PM in Dean's Suite, Community Health Building, 4055 North Park Loop
Advisor: Gavin Bidelman
Dissociable Mechanisms of Concurrent Speech Identification in Noise at Cortical and Subcortical Levels
ABSTRACT: When two vowels with different fundamental frequencies (F0s) are presented concurrently, listeners often hear two voices producing different vowels on different pitches. Parsing of this simultaneous speech can also be affected by the signal-to-noise ratio (SNR) in the auditory scene. The extraction and interaction of F0 and SNR cues may occur at multiple levels of the auditory system. The major aims of this dissertation are to elucidate the neural mechanisms and time course of concurrent speech perception in clean and in degraded listening conditions and its behavioral correlates. In two complementary experiments, electrical brain activity (EEG) was recorded at cortical (EEG Study #1) and subcortical (FFR Study #2) levels while participants heard double-vowel stimuli whose fundamental frequencies (F0s) differed by zero and four semitones (STs) presented in either clean or noise degraded (+5 dB SNR) conditions. Behaviorally, listeners were more accurate in identifying both vowels for larger F0 separations (i.e., 4ST; with pitch cues), and this F0-benefit was more pronounced at more favorable SNRs. Time-frequency analysis of cortical EEG oscillations (i.e., "brain rhythms") revealed a dynamic time course for concurrent speech processing that depended on both extrinsic (SNR) and intrinsic (pitch) acoustic factors. Early high frequency activity reflected pre-perceptual encoding of acoustic features (~200 ms) and the quality (i.e., SNR) of the speech signal (~250-350ms), whereas later-evolving low-frequency rhythms (~400-500ms) reflected post-perceptual, cognitive operations that covaried with listening effort and task demands. Analysis of subcortical responses indicated that while FFRs provided a high-fidelity representation of double vowel stimuli and the spectro-temporal nonlinear properties of the peripheral auditory system. FFR activity largely reflected the neural encoding of stimulus features (exogenous coding) rather than perceptual outcomes, but timber (F1) could predict the speed in noise conditions. Taken together, results of this dissertation suggest that subcortical auditory processing reflects mostly exogenous (acoustic) feature encoding in stark contrast to cortical activity, which reflects perceptual and cognitive aspects of concurrent speech perception. By studying multiple brain indices underlying an identical task, these studies provide a more comprehensive window into the hierarchy of brain mechanisms and time-course of concurrent speech processing.