“…Besides paradigmatic and contextual differences, another layer of complexity that accompanied the relative scarcity of hyperscanning neuroimaging studies, is the relative limited methodologies adopted: most cited fMRI studies deployed univariate (e.g., general linear model or GLM) or region-of-analysis methods (e.g., Abe & Greene, 2014;Ito et al, 2012;Lisofsky, Kazzer, Heekeren, & Prehn, 2014;Sun et al, 2017;Vartanian et al, 2013;Wu, Loke, Xu, & Lee, 2011;Yin et al, 2016), with few recent exceptions utilizing connectivity (Jiang et al, 2015;Speer et al, 2020) and multivariate approaches (Jin et al, 2009;Yin & Weber, 2019). In comparison, in the present study, GLM contrast, psychophysiological interactions (PPI, O'Reilly, Woolrich, Behrens, Smith, & Johansen-Berg, 2012), and multi-voxel pattern analysis (MVPA, Kriegeskorte, Goebel, & Bandettini, 2006) were adopted to identify patterns of task-or seed-driven activity throughout the brain, plus the subsequent principal component analysis (PCA) that reduced the high-dimensionality problems in MVPA analysis, expanding the perspectives of data examination and joint constraints of the result interpretations.…”