Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
Task-dependent functional connectivity of language-selective regions
Poster A83 in Poster Session A, Tuesday, October 24, 10:15 am - 12:00 pm CEST, Espace Vieux-Port
Arielle Moore1, Terri Scott2, Tyler Perrachione1; 1Boston University, 2University of California, San Francisco
A cortical network for spoken language processing encompasses a consistent array of regions in the frontal, temporal, and parietal lobes, the precise location of which varies uniquely across individuals. Much of the neuroimaging work characterizing these regions has used univariate response magnitudes or multivoxel pattern analyses to investigate differences in local selectivity for language vs. nonlinguistic stimuli or tasks. However, little work has asked how the connectivity of these language areas – either to each other or to the rest of the brain – is affected by task demands. Here, we used generalized psychophysiological interaction (PPI) models to examine how functional connectivity between individually defined language areas and the rest of the brain changed during linguistic vs. nonlinguistic processing. Using functional magnetic resonance imaging (fMRI), we measured whole-brain BOLD responses from N=24 neurotypical adults during an auditory language localizer task that contrasted passive listening to intact speech vs. incomprehensible degraded speech (Scott et al., 2017). Using group-constrained, subject-specific (GCSS) analyses, we obtained a probabilistic parcellation of the intact > degraded speech contrast, yielding bilateral anterior and posterior STG, IFG pars opercularis, IFG pars triangularis, MFG, SFG, and right cerebellum parcels. Within each of these parcels, we defined subject-specific functional regions of interest (fROIs) corresponding to each participant’s maximally selective (top 10%) voxels. We used the mean BOLD timeseries extracted from each of these fROIs to construct generalized PPI models that analyzed task-dependent changes in connectivity between that fROI and the rest of the brain when listening to comprehensible intact vs. incomprehensible speech. We analyzed these results in two ways: (1) in univariate group analysis examining task-dependent connectivity from the parcelwise fROIs across the whole brain, and (2) by extracting the models' interaction term coefficients within the individually defined language-network fROIs for each participant. The whole-brain group-average analysis revealed that listening to comprehensible speech was associated with a broad reduction in functional connectivity from language fROIs. Notably, much of this reduced connectivity was localized to tissue proximal to, but distinct from, core language areas. Analysis of task-dependent functional connectivity within individually-defined nodes of the language network revealed a more complicated pattern, with some areas tending to show increased connectivity during spoken language processing (e.g., left IFG), while others tended to show decreased connectivity (e.g., left MFG), and while still others showed a mix of increased and decreased connectivity (e.g., left STG) to the rest of the language network. Taken together, these results suggest that cortical language network dynamics are more complicated than just a wholesale increase in response during language processing and decreased response to nonlinguistic stimuli or domain-general cognitive tasks. Instead, activity in the cortical language network appears to become broadly decoupled from other areas, especially spatially proximal areas, during language comprehension. Furthermore, the complex dynamics of increasing and decreasing connections between language areas hint that, despite homogeneous patterns of univariate responses in these regions, they may nonetheless have functionally dissociable roles during online language processing.
Topic Areas: Speech Perception, Control, Selection, and Executive Processes