CRCNS Research Proposal: Modeling natural language production in humans to improve artificial language systems

This project aims to better understand the brain mechanisms that support the production and comprehension of language, using advanced neuroimaging and computational methods. Information about the brain networks mediating language will be combined with insights gained from large language models developed recently in Artificial Intelligence to create powerful and efficient algorithms for decoding intended speech from human brain activity. In sum, this project will advance scientific understanding of the brain basis of language production, and provide a solid foundation for brain-machine interfaces that improve communication abilities for people affected by language disorders. The project leverages advanced neuroimaging protocols and computational tools that reveal the function of brain circuits mediating language use, at high resolution and in individual participants. The first aim of the project is to use functional MRI to record brain activity during extended language production tasks, and to create high-resolution, participant-specific maps that characterize the spatial distribution and temporal profile of language-related brain representations across the full hierarchy of brain networks involved in language production. The second aim is to record brain activity during a matched language comprehension task, and to compare high-resolution maps of language representations during production versus comprehension. The final aim will use insights gained from the first two aims to develop advanced, efficient methods for decoding intended speech. This could revolutionize communication aids for individuals with speech impairments, offering them a new avenue for interaction. This project will advance basic scientific understanding of the brain circuits mediating language production, which can in turn be used to improve diagnostic and rehabilitation methods. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.