Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
Dynamic functional connectivity underlying predictive sentence comprehension
Poster B21 in Poster Session B, Tuesday, October 24, 3:30 - 5:15 pm CEST, Espace Vieux-Port
Meichao Zhang1,2, Yaji He1,2, Ximing Shao3, Xiaoqing Li1,2; 1Institute of Psychology, Chinese Academy of Sciences, 2Department of Psychology, University of Chinese Academy of Sciences, Beijing, China, 3Department of Psychology, York Neuroimaging Centre, University of York, United Kingdom
The human brain is remarkably efficient in language comprehension, which is capable of quickly integrating perceived information and simultaneously predicting upcoming information based on available semantic context. Language prediction is considered to be actively generated in a top-down manner (Friston, 2005, 2010; Kuperberg & Jaeger, 2016), with cognitive control playing an important role in it (e.g., Bonhage et al., 2015; Shao et al., 2022). For example, left inferior frontal gyrus (LIFG), an important control site, exhibits stronger activation during the prediction of incoming semantic information (Shao et al., 2022). Contemporary work has also shown the implication of default mode network (DMN) in semantic integration (e.g., Lanzoni et al., 2020; Wang et al., 2021), with a key role of temporo-parietal junction (TPJ) in sentence comprehension (e.g., Alexandrou et al., 2017; Matchin et al., 2019). While it has been well-documented the neural basis underlying semantic prediction in language comprehension, it remains unclear how these control and integration-related sites connect with other brain areas to support the anticipation and integration phases of processing during online language comprehension. To explore that question, a generalized psychophysiological interaction (gPPI) analysis was performed on our previously published fMRI dataset (Shao et al., 2022). In that sentence comprehension study, the semantic constraint of sentence context was manipulated (Strong tool, Strong building, and Weak), and both the anticipation and integration phases of processing were recorded. Two ROIs (6-mm sphere) were selected: (i) semantic integration ROI of right TPJ (coordinate: 54 -54 28) based on a comprehensive analysis of functional connectivity of DMN (Andrews-Hanna et al., 2010); (ii) cognitive control ROI of LIFG (coordinate: -44 22 24) defined by using the search term “cognitive control” in Neurosynth (Yarkoni, Poldrack, Nichols, Van Essen, & Wager, 2011). For right TPJ seed, an interaction between Processing phase (Anticipation vs. Integration) and Semantic constraint (Strong vs. Weak) was observed in parahippocampal gyrus (PPA), fusiform gyrus, posterior cingulate cortex (PCC), supramarginal gyrus, inferior parietal lobe and insula; these voxels largely fell within visual network (41.64%), ventral attention network (VAN; 25.08%), and DMN (9.45%). When the semantic constraint was strong, the functional coupling of right TPJ to PPA/fusiform/PCC/insula was stronger during anticipation phase, and its coupling to all the identified sites was weaker during integration phase, compared with when the semantic constraint was weak. For LIFG seed, an interaction effect was also observed in fusiform gyrus and PCC, which mainly fell within visual network (95.49%). When the semantic constraint was weak, the functional connectivity of LIFG with these sites was stronger during integration phase, compared with when the semantic constraint was strong. No significant semantic constraint effect was found during anticipation phase. Taken together, these results suggest that right TPJ-to-visual/VAN/DMN functional connectivity support top-down prediction based on available semantic context, while the functional coupling of LIFG to visual system is important for the bottom-up integration when the upcoming linguistic input cannot be efficiently predicted. We conclude that top-down prediction and bottom-up integration relies upon dynamic functional coupling of memory, control, and visual systems.
Topic Areas: Meaning: Lexical Semantics, Reading