Poster Presentation

Poster F in Poster Session F: Thursday, August 6, 1:45 – 3:30 pm, Kimmel Center, Shorin & Rosenthal Rooms

Selective attention modulates the flow of information between visual- and lexical-semantic networks in the human brain

Emily X. Meschke¹, Jen D. Holmberg², Jack L. Gallant³; ¹Columbia University, ²University of Minnesota, ³University of California, Berkeley

Presenter: Emily X. Meschke

Goal-directed behavior relies on the ability to selectively prioritize relevant streams of information. For example, when communicating about the visual world, both visual and lexical information are relevant, whereas during introspection, only lexical information is relevant. Given the brain flexibly supports both these behaviors, we hypothesized that selective attention acts as a gating mechanism, controlling the information flow between visual and lexical functional networks. To assess this, we designed a functional MRI (fMRI) experiment in which eight participants watched compilations of movie clips while either describing the content of the movies (attend-external), or while ignoring the movies and describing the content of their thoughts (attend-internal). Visual and lexical features were extracted from the movies and the participant’s speech, and these features were used to fit voxelwise encoding models to the fMRI data separately for each attention condition and for every participant. The prediction accuracy of the semantic models revealed that task demands dramatically affected the interaction between cortical visual- and lexical-semantic networks. In both conditions, regions in anterior visual cortex were well predicted by the visual-semantic models, and regions in higher-order association cortex were well predicted by the lexical-semantic models. When participants described the movies (attend-external), both semantic models predicted activity in regions that border both networks. However, when participants ignored the movies and described their thoughts (attend-internal), these border regions represented one or the other modality, but not both. These results suggest that attention flexibly modulates the flow of information between functional brain networks in service of behavioral goals.

Topic Area: Methods, Tools, Theory & Neural Coding