Abstract
When comprehending visual narratives, the human brain constantly accumulates and integrates incoming information to make a cohesive representation of event structures. Recent studies have suggested that the brain fluctuates between globally segregated and integrated states, and such fluctuation is modulated by the cognitive process engaged at the moment (Shine et al. 2016). Here, we ask whether the dynamic reconfiguration of brain networks can be driven by changes in cognitive states during visual narrative comprehension. Using fMRI, we investigated network- and regional-level changes when subjects were engaged in comprehending scrambled versions of the silent films. Whole-brain time-resolved FCs were mapped, and the graph indices were compared between the time points when understanding of the narrative actively occurred, and the time points when understanding occurred to a lesser degree. Global modularity was low during times of high narrative understanding, implying that the modular structure of the brain was tightly integrated when the degree of information integration was high. Regions within the default mode network (DMN) and frontoparietal network significantly increased modular connections during times of high narrative understanding; specifically, medial prefrontal cortex, posterior cingulate cortex and precuneus increased its across-modular connections, whereas frontal regions increased within-modular connections. Similar patterns of changes were not shown when subjects repeatedly watched the same film, indicating that these FC changes were not driven by the physical characteristics of visual stimuli. Furthermore, we examined whether the moment-to-moment pattern of regional FCs can predict the degree of narrative understanding. Decoding analysis using selective FCs revealed that it was possible to predict the degree of understanding at a specific time point of a novel subject, and even for a novel film. Our findings suggest that the brain adaptively regulates its modular structure by coordinating connections of the DMN and frontoparietal network, upon dynamically updating representation of visual narrative structures.
Meeting abstract presented at VSS 2018