Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams
Two spatiotemporally distinct cortical networks for sentence reading
Poster A74 in Poster Session A, Thursday, October 6, 10:15 am - 12:00 pm EDT, Millennium Hall
Oscar Woolnough1, Cristian Donos2, Elliot Murphy1, Patrick Rollo1, Zachary Roccaforte1, Stanislas Dehaene3, Nitin Tandon1; 1University of Texas Health Science Center at Houston, 2University of Bucharest, 3College de France
Reading a sentence entails inferring and integrating meaning at multiple levels by combining the information conveyed by each word. This highly rapid and complex human behavior is believed to be distributed between the inferior frontal gyrus (IFG) and middle temporal gyrus (MTG) in the language dominant hemisphere, yet the relative and distinct contributions of these regions to sentence comprehension are still debated. An important open question remains - how do these two closely interconnected hubs dynamically interact to integrate information across a sentence? To create a high spatiotemporal resolution map of activation across the reading and language networks we used direct intracranial recordings in 27 people, with >2,400 electrodes in the language dominant hemisphere, to measure induced broadband gamma activity (70-150Hz). Each patient read eight-word normal sentences, Jabberwocky sentences, and lists of words and pseudowords presented in rapid serial visual presentation. We tracked activation across mid-fusiform cortex, lateral temporal cortex, IFG and the medial frontal operculum and used linear mixed effects modelling to dissociate contributions from multiple lexical and sentential factors. Examination of the evolution of activity across consecutive stimuli revealed two functionally and spatiotemporally distinct frontotemporal networks, each sensitive to distinct aspects of lexical and syntactic composition. The first distributed network, in which the posterior IFG precedes MTG, demonstrated a slow ramping-up of activity over the duration of a sentence, which was reduced or absent during Jabberwocky and lists, suggesting a role in the composition of sentence-level meaning. In the second network activity in superior temporal gyrus preceded activity in anterior IFG, and this network showed greater phasic activation for each word presented in a list relative to those in sentences, suggesting that sentential context affords a reduction in semantic and/or phonological processing effort. We also tracked the spread of sensitivity to lexical features. Word frequency was strongly represented in mid-fusiform cortex first and later in anterior IFG whereas posterior IFG and lateral temporal cortex did not show substantial modulation by lexical level features. In conclusion, we have identified and characterised at least two spatiotemporally separable frontotemporal language networks utilised during sentence reading. This spatiotemporal dissociation of subregions of the IFG and lateral temporal cortex sheds light on the layered semantic processes evoked by sentence processing and confirms that ramping-up of activity in IFG and MTG constitutes a plausible signature of semantic compositionality.
Topic Areas: Reading, Syntax