Lecture by Roeland Hancock "Neurochemistry of speech oscillations"
Roeland Hancock is from the University of California, San Francisco, USA
On April 7, 2016 Roeland Hancock gave a talk on "Neurochemistry of speech oscillations".
Abstract: The acoustic speech signal contains a mixture of quasi-periodic, linguistically meaningful information at multiple timescales. Recent models of speech processing (e.g. Giraud & Poeppel, 2012) propose that a hierarchy of oscillatory neural processing can provide a mechanism for de-multiplexing and coding such sensory signals. In particular, gamma band oscillations, nested within theta oscillations, may provide a basis for discretizing the speech signal. Theta-gamma coupling and the frequency of gamma oscillations reflect a delicate balance of regulated periods of excitation and inhibition. As a complement to neurally plausible computational models of theta-gamma sensory processing, I examine the neurochemical correlates of speech and auditory processing in vivo. Multimodal imaging experiments, integrating measures of GABA and glutamate concentrations within human auditory cortex (using magnetic resonance spectroscopy [MRS]) and magnetoencephalographic (MEG) measures of auditory processing, are used to investigate the possible neurochemical correlates of speech processing. Preliminary results from behavioral and auditory steady state response (ASSR) paradigms suggest that GABA—presumably a marker of tonic neural inhibition—is closely associated with gamma band frequency tuning of auditory processing. A mutual information analysis of naturalistic speech suggests that GABA-related shifts in auditory tuning for non-speech stimuli also impact the neural coding of natural speech within a theta-gamma hierarchy. These results support hierarchical oscillatory models of speech processing.