Verbal recollection engages the default mode network more strongly than visual recollection: An fMRI meta-analysis and semantic scaffolding account

Description: Functional neuroimaging studies consistently show that episodic recollection engages the default mode network (DMN), yet the extent to which this engagement varies with the representational content of retrieved information remains unclear. The present meta-analysis examined whether verbal and visual recollection differentially recruit the DMN. Twenty-four verbal and 19 visual studies were included; recollection was operationalized using subjective measures, and network correspondence was assessed using the Yeo 17-network parcellation. Both forms of recollection engaged the DMN, but verbal recollection showed stronger and more widespread recruitment across all three DMN subnetworks. Direct contrasts likewise revealed consistent preferential associations for verbal over visual recollection across all DMN subdivisions. Separate analyses of word-, object-, and scene-based recollection further supported this verbal bias in DMN engagement. This bias may reflect a greater reliance of verbal recollection on DMN-supported conceptual integration, or semantic scaffolding, than visual recollection. More broadly, the representational content of remembered events—particularly along the verbal–visual dimension—may constitute an important organizing principle of DMN function in episodic memory and help account for variability across prior neuroimaging findings, with potential implications for other DMN-supported domains.