Slide Slam M16
Semantic conflict is resolved by semantic and multiple demand networks
Curtiss Chapman1, Gesa Hartwigsen1; 1Max Planck Institute for Human Cognitive and Brain Sciences
How do our brains deal with irrelevant information? Neuroimaging studies show that a variety of brain regions help to regulate meaning processing so that only correct, relevant information is focused upon. The current study provides new information on which brain areas are active when dealing with irrelevant, conflicting information. Moreover, we elucidate how these regions interact—with one another and with other brain networks—to drive successful resolution of such conflict. Forty participants performed two semantic tasks in which they made yes/no judgments on simultaneously presented pairs of line drawings while undergoing fMRI. In the taxonomic task, they judged whether items were in the same category; in the thematic task, they judged whether items were associated thematically but not in the same category. Both tasks included the same items, which could be taxonomically related, thematically related, or unrelated. The semantic brain network for the tasks was investigated by contrasting performance across both semantic tasks with performance on a visual control task, in which participants judged whether two scrambled images were mirrored versions of each other. Regions dealing with semantic conflict were measured by comparing items that were related in a way that was irrelevant to the current task (e.g., thematically related items in the taxonomic task) compared to unrelated items across tasks. All fMRI statistical analyses were performed with GLMs in SPM 12. Results from semantic and conflict contrasts were thresholded at FWE-corrected p < .05. Regions from the tasks' semantic network were used as seed regions for effective connectivity (PPI) analyses, which were performed with the gPPI toolbox. One-sample t-tests were then performed on subject-level results and thresholded at p < .001, uncorrected. Our crucial PPI contrast was the semantic conflict contrast cited above. The observed semantic network included the bilateral angular gyrus (AG) and adjacent lateral occipital cortex, bilateral precuneus and posterior cingulate (PCC), left inferior frontal gyrus, left posterior middle temporal gyrus (pMTG), and left fusiform gyrus, extending medially. Regions more active in the conflict condition overlapped the semantic network in the bilateral AG, bilateral precuneus/PCC, and left pMTG and additionally included the bilateral medial superior frontal regions, left frontal pole, and right middle frontal gyrus. In PPI analyses, connectivity increased between ROIs from our tasks' semantic network (right AG, bilateral precuneus/PCC) and regions associated with multiple demand and semantic control networks in trials where items were related in an irrelevant way compared to when items were unrelated. Our results thus provide evidence that resolution of semantic conflict relies on domain-general and domain-specific functional brain networks. Interestingly, the seeds revealing these networks suggest a crucial role for bilateral multiple-demand regions in driving the resolution of semantic conflict, in contrast to the predominately left-lateralized semantic control network. Future planned analyses will reveal whether behavioral effects are impacted by the strength of our observed connectivity and whether resting-state connectivity in the same participants reveals the same semantic networks as our effective connectivity analyses.