Resolving the Abstract-Concrete Paradox in the Angular Gyrus:A Multimethod Investigation

Despite its well-established role in memory-guided cognition, whether and how the angular gyrus (AG) contributes to semantic processing remains unresolved. Connectomic work links the AG to abstract mentation, yet neuroimaging studies paradoxically show greater AG engagement for concrete than abstract semantics. To address this discrepancy, we conducted a multimethod investigation integrating neurostimulation, neuroimaging, and experience sampling across five studies (127 human participants; 53 males; 74 females). Using the contrast between concrete and abstract semantics as a diagnostic test-case, we show that this apparent contradiction reflects multiple interacting neurocognitive factors shaping the AG's functional repertoire. In Study 1, causal disruption of the AG disproportionately impaired abstract semantics and temporary retention of task-relevant information, demonstrating its role in abstract and memory-guided cognition. In Study 2, the apparent concreteness effect was abolished after controlling for performance speed/accuracy, indicating that AG activity is more strongly associated with low-demand mental states than semantic concreteness. In Study 3, the AG showed enhanced functional coupling with regions in the semantic network for abstract words, suggesting sustained integration with the semantic system despite reduced activation. In Study 4, experience sampling showed that concrete words are associated with experiences of mental imagery and automaticity, providing a phenomenological account of heightened AG engagement under low-demand states. Finally, Study 5 clarified how baseline choices influence the interpretation and polarity of AG activity. Together, these findings show that AG engagement is best understood along a continuum of memory-guided cognition, clarifying when and why this region supports abstract versus concrete semantics across situations.Significance Statement The contribution of the AG to semantic processing has long been debated, both regarding whether it is reliably engaged and how such engagement should be interpreted. By integrating evidence from multiple independent datasets using complementary methodologies, the present investigation moves beyond single-process accounts to identify multiple contributors that jointly define AG function. These factors align with contemporary connectomic research that situates the AG at the transmodal apex of a principal cortical gradient. Under this brain-connectomic framework, AG involvement is amplified as representations become abstract, information is buffered, operation becomes automatic, imagery is simulated, reliance on perception is reduced, shifting from perception-guided mode to memory-guided cognition. Together, these findings provide a principled framework for interpreting the elusive patterns of AG functionality.