Poster E51, Monday, March 26, 2:30-4:30 pm, Exhibit Hall C
Spatiotemporal Dissociations associated with Fulfilling and Violating Predictions at Multiple Levels of Representation: A multimodal approach
Gina R. Kuperberg1,2, Lotte Schoot1,2, Lin Wang1,2, Edward Alexander2, Nate Delaney-Busch2, Eddie Wlotko2, Minjae Kim1,2, Lena Warnke2, Sheraz Kahn1, Matti Hamalainen1; 1Martinos Center for Biomedical Imaging, Massachusetts General Hospital, 2Tufts University
We used ERP, MEG, and fMRI to ask whether and when distinct neuroanatomical networks are engaged to inputs that fulfill or violate strong contextual predictions generated at the level of specific lexical items and/or semantic-thematic structure. Thirty-two participants read and judged the acceptability of three-sentence scenarios in fMRI and ERP/MEG sessions. Scenarios varied in their lexical constraint and in whether critical nouns fulfilled or violated lexical predictions and/or the selection restrictions of their preceding verbs. ERPs revealed reduced activity on the N400 to predictable nouns, relative to all other conditions. MEG localized this activity to the left inferior frontal cortex (IFG) and the left anterior superior/middle temporal cortex (ant-S/MTG). fMRI revealed most robust modulation within left IFG. These findings support a functional role of left IFG/ant-S/MTG in mediating semantic retrieval during sentence comprehension. Relative to unpredicted words in low-constraint contexts, words that violated strong lexical predictions evoked a larger late anteriorly-distributed positivity, which MEG localized to the left IFG and posterior inferior temporal cortex. fMRI confirmed this pattern. We suggest that these regions mediated prolonged attempts to retrieve unpredicted lexical items, and infer the event dictated by the bottom-up input. In contrast, words that violated selection restrictions evoked a larger late posteriorly-distributed positivity, which MEG localized to the left anterior superior DLPFC and the post-ITG/fusiform cortex. Again, fMRI confirmed this pattern. We suggest that these regions reflected reanalysis and prolonged efforts to infer novel event structures dictated by the bottom-up input.
Topic Area: LANGUAGE: Semantic