Mapping Contributions of the Anterior Temporal Semantic Hub to the Processing of Abstract and Concrete Verbs

Abstract

Multiple representation theories of semantic processing propose that word meaning is supported by simulated sensorimotor experience in modality-specific neural regions, as well as in cognitive systems that involve processing of linguistic, emotional, and introspective information. According to the Hub and Spoke Model of Semantic Memory, activity from these distributed cortical areas feeds into a primary semantic hub located in the ventral anterior temporal lobe (vATL). Though a substantial amount of research has tested this model in terms of concrete noun representation, there is less known about how this model can account for the representation of verb meaning, and in particular the meaning of abstract verbs which convey important information, for example, about socioemotional dynamics. In the present pre-registered study, we examined whether different types of abstract verbs (mental, emotional, nonembodied) and concrete (embodied) verbs all engage the vATL, and also whether they differentially recruit a broader set of distributed neurocognitive systems (consistent with multiple representation theories). Finally, we investigated whether there is information about different verb types distributed across the broader ATL region, consistent with a Graded Semantic Hub Hypothesis. We collected data from 30 participants who completed a syntactic classification task (is it a verb? Yes or no) and a numerical judgement task which served as an active but less semantic baseline task. Whole brain univariate analyses revealed consistent BOLD signal throughout the canonical semantic network, including the left inferior frontal gyrus, left middle temporal gyrus, and the vATL. All types of abstract verbs engaged the vATL except for mental state verbs. Finally, a multivariate pattern analysis revealed clusters within the ATL that were differentially engaged when processing each type of abstract verb. Our findings extend previous research and suggest that the hub-and-spoke hypothesis and the graded semantic hub hypothesis provide a neurobiologically constrained model of semantics that can account for abstract verb representation and processing.