Investigating the Neural Basis of Audiovisual Speech Perception with Intracranial Recordings in Humans
Author ORCID Identifier
Date of Graduation
Doctor of Philosophy (PhD)
Speech is inherently multisensory, containing auditory information from the voice and visual information from the mouth movements of the talker. Hearing the voice is usually sufficient to understand speech, however in noisy environments or when audition is impaired due to aging or disabilities, seeing mouth movements greatly improves speech perception. Although behavioral studies have well established this perceptual benefit, it is still not clear how the brain processes visual information from mouth movements to improve speech perception. To clarify this issue, I studied the neural activity recorded from the brain surfaces of human subjects using intracranial electrodes, a technique known as electrocorticography (ECoG). First, I studied responses to noisy speech in the auditory cortex, specifically in the superior temporal gyrus (STG). Previous studies identified the anterior parts of the STG as unisensory, responding only to auditory stimulus. On the other hand, posterior parts of the STG are known to be multisensory, responding to both auditory and visual stimuli, which makes it a key region for audiovisual speech perception. I examined how these different parts of the STG respond to clear versus noisy speech. I found that noisy speech decreased the amplitude and increased the across-trial variability of the response in the anterior STG. However, possibly due to its multisensory composition, posterior STG was not as sensitive to auditory noise as the anterior STG and responded similarly to clear and noisy speech. I also found that these two response patterns in the STG were separated by a sharp boundary demarcated by the posterior-most portion of the Heschl’s gyrus. Second, I studied responses to silent speech in the visual cortex. Previous studies demonstrated that visual cortex shows response enhancement when the auditory component of speech is noisy or absent, however it was not clear which regions of the visual cortex specifically show this response enhancement and whether this response enhancement is a result of top-down modulation from a higher region. To test this, I first mapped the receptive fields of different regions in the visual cortex and then measured their responses to visual (silent) and audiovisual speech stimuli. I found that visual regions that have central receptive fields show greater response enhancement to visual speech, possibly because these regions receive more visual information from mouth movements. I found similar response enhancement to visual speech in frontal cortex, specifically in the inferior frontal gyrus, premotor and dorsolateral prefrontal cortices, which have been implicated in speech reading in previous studies. I showed that these frontal regions display strong functional connectivity with visual regions that have central receptive fields during speech perception.
Intracranial EEG, Electrocorticography, Speech Perception, Audiovisual Integration