Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Intracortical recordings from inferior frontal gyrus reveal encoding of linguistic and nonlinguistic operations, not speech production
Poster Session D, Thursday, October 1, 4:30 - 6:30 pm, Wangari Maathai
Erin M. Kunz1, Sasidhar Madugula1, Hannah Cui1, Alisa Levin1, Ryan Z. Wang1, Alessandro Marin Vargas1, Donald Avansino1, Foram Kamdar1, Christina Vo1, Akansha Singh1, Nick Hahn1, Erxiao Wang1, Yuhan Zhang1, Leigh Hochberg2,3,4, Shaul Druckmann1, Cory Shain1, Francis R. Willett1, Jaimie M. Henderson1; 1Stanford University, 2Brown University, 3VA Providence Healthcare System, Providence, RI, 4Harvard Medical School
Inferior frontal gyrus has long been implicated in speech and language production, but its precise role remains poorly understood. As part of a communication neuroprostheses study, we recorded single-neuron activity from intracortical arrays implanted in the inferior frontal gyrus of two BrainGate2 participants with severe dysarthria due to amyotrophic lateral sclerosis. Across arrays, we found little evidence that IFG neurons were tuned to phoneme production or single-word output. Instead, neural activity reflected higher-order linguistic and cognitive processes. In one participant, IFG activity was strongly modulated during sentence reading and production. In a follow-up controlled sentence task, most of the variance underlying the sentence-specific neural patterns could be accounted for by syntactic structure largely independent of semantic content, although not all of it. Across arrays, neural activity in metalinguistic tasks such as analogy solving tracked the operation being performed more strongly than the answer word itself. Further controlled studies revealed distinct IFG neural patterns for syntactic, semantic, and morphological computations, regardless of the content to which they were applied. Similar modulation was also present in some nonlinguistic tasks, including melody reversal, indicating sensitivity to abstract transformations that extend beyond language. Differences in large-scale network affiliation accounted for some, but not all, of the diversity of responses across arrays. Together, these findings argue for the existence of language-relevant regions of IFG that are not primarily organized around speech motor planning or response output, but instead support higher-order transformations that are important for language but are not limited to it. These results refine theories of Broca’s area and may help guide the design of communication brain-computer interfaces.
Topic Areas: Language Production, Syntax and Combinatorial Semantics