There is more social in semantics! A brief commentary and reanalysis of Balgova et al. (2024)

Balgova et al. (2024) recently conducted a large-scale meta-analysis on mentalizing and on semantic cognition, to investigate the degree to which the neural correlates of these two processes are overlapping. The study found consistent neural overlap between the two processes, especially in the bilateral anterior temporal lobe (ATL) and the left temporoparietal junction (TPJ), although they also identified many areas of activation specific to mentalizing. Although we agree with their general conclusion, we investigated to what extent the semantic dataset was actually devoid of social content, and if not, how this would change the results. After careful screening and categorization of the “semantic” material, we found experiments that contained elements of social mentalizing (N = 36) and social action observation (N = 16), apart from nonsocial semantics (N = 46). ALE analyses on the social mentalizing and nonsocial semantic subsets from the original “semantic” full dataset, confirmed that semantic brain areas are activated when processing both social mentalizing and nonsocial semantic content, while mentalizing brain areas are uniquely activated when processing social mentalizing content. Specifically, semantic and mentalizing content activated the left inferior frontal gyrus (IFG), left middle temporal gyrus (MTG) and posterior medial frontal cortex (pmFC); and also the left ventral temporal lobe, supporting the graded multimodal hub model of semantic cognition. Critically, as we claimed, mentalizing content uniquely activated the temporal pole (TP), medial prefrontal cortex (mPFC), although activation in the left TPJ was also shared with semantic processes. We conclude that a more careful distinction between social and nonsocial datasets guarantees more sensitive and valid analyses.