PROJECT SUMMARY / ABSTRACT The overall goal of this project is to test and refine a neurobiological systems model of speech-induced sensory modulation (SISM). Previous studies used magnetoencephalography (MEG) and electroencephalography (EEG) to reveal that auditory cortical responses evoked when speaking are reduced compared to passive listening, a phenomenon known as speaking-induced suppression (SIS). This effect has been proposed to reflect the action of learned forward models that predict sensory consequences of speech movements. While EEG/MEG studies of SIS have provided important insights, these methods have limited anatomical precision. For this reason, the auditory regions modulated by forward model predictions in speech, and the circuits that mediate them, remain relatively poorly understood. To close this knowledge gap, this project uses 7T fMRI to measure brainwide activity at exquisite resolution while participants complete a series of related auditory-motor tasks across multiple sessions. Guided by contemporary models of speech production, we hypothesize that three functional circuits link frontal speech planning areas with superior temporal auditory areas via cortical and trans-cerebellar pathways, providing sensory predictions at different levels of representation. In Aim 1, one EEG and three fMRI experiments will be used to map the sensory cortical areas modulated by self-generated speech. These studies will define subject-specific regions-of-interest (ROIs) that differentially respond to overt speech vs. passive listening to oneself or another speaker producing the same syllables or sentences. Multivariate encoding models test hypotheses about the speech features represented in these ROIs. In the same speakers, we will determine the extent and specificity of modulations in these ROIs during movement planning, prior to motor output. Aim 2 tests how these ROIs participate in error monitoring. Two fMRI experiments are designed to test hypotheses about responses to auditory errors that are either (1) “low-level” acoustic-phonetic errors induced by unexpected shifts in auditory feedback, or (2) “high-level” sound substitution errors elicited during a phonological error priming task. Because Aim 2 is carried out in the same speakers as Aim 1, within-subject comparisons will enable high- resolution individual-specific models and interpretations across observed effects. In Aim 3, we will determine the role of the cerebellum (CB) in implementing learned forward models that drive SISM. Individuals with cerebellar lesions and controls will be recruited to test the hypothesis that the CB is critical in learning predictive models for speech. In these participants, and those tested in Aims 1 and 2, 7T fMRI will measure neural activity changes during a speech motor adaptation task and a non-vocal auditory-motor learning task. Individuals with lesions in Lobule VI are predicted to show reduced learning, reduced SISM, and reduced associations between CB and auditory cortical activity. Together these studies will accelerate understanding of sensory-motor interactions in speech through in-depth analysis of individual brains, enable the refinement of detailed neurocomputational models, and set the stage for precision approaches to diagnosis and intervention in speech disorders.

PROJECT NARRATIVE This project will improve our understanding of how the brain’s motor system interacts with its sensory systems via the cerebellum as we plan and produce speech. It will employ state-of-the-art non-invasive, ultra-high field functional magnetic resonance imaging to provide maps of brain activity during speaking tasks in unprecedented detail. The foundational knowledge gained through this project is expected to lead to new insights about the mechanisms underlying speech disorders.