Dissociating neural sensitivity to target identity and mental state content type during inferences about other minds

Predicting and inferring what other people think and feel (mentalizing) is central to social interaction. Since the discovery of the brain's "mentalizing network," fMRI studies have probed the lines along which the activity of different regions in this network converges and dissociates. Here, we use fMRI meta-analysis to aggregate across stimuli, paradigms, and contrasts from past studies in order to definitively test two sources of possible sensitivity among brain regions of this network with particular theoretical relevance. First, it has been proposed that mentalizing processes depend on aspects of target identity (whose mind is considered), with self-projection or simulation strategies engaging disproportionately for psychologically close targets. Second, it has been proposed that mentalizing processes depend on content type (what the inference is), with inferences about epistemic mental states (e.g., beliefs and knowledge) engaging different processes than mentalizing about other types of content (e.g., emotions or preferences). Overall, evidence supports the conclusion that different mentalizing regions are sensitive to target identity and content type, respectively, but with some points of divergence from previous claims. The results point to fruitful directions for future studies, with implications for theories of mentalizing.