Expertise modifies neural responses to causal inference during language comprehension

Linguistic communication relies on shared common ground, yet different individuals in a community have specialized semantic knowledge. As a key example, physicians acquire rich knowledge about the causes of illness. An open question is whether such "causal expertise" influences the neural basis of language comprehension. Here we tested the hypothesis that when understanding everyday language like, "Holly used a grimy faucet during her overseas plane ride. Now she has the stomach flu," medical experts retrieve richer representations in semantic networks of the brain. Physicians (n=16) and age- and education-matched controls (n=16) undergoing fMRI read two-sentence vignettes that elicited implicit causal inferences about illness or object mechanics as well as control vignettes that did not elicit causal inferences. Vignettes were linguistically matched across conditions and were accessible to experts and non-experts alike. Across both groups, causal vignettes about illness engaged a temporoparietal semantic network previously implicated in thinking about animates, i.e., people and animals (precuneus and left temporoparietal junction), whereas causal vignettes about object mechanics engaged a semantic network associated with places and objects. Critically, physicians' responses in the animacy network were more robust and self-similar, suggesting the activation of richer semantic knowledge. Medical expertise effects were specific to biological inferences rather than mechanical or mental state inferences. By contrast, individually localized frontotemporal language and frontoparietal reasoning networks did not respond preferentially to causally connected sentences or show effects of expertise. Our findings suggest that causal expertise alters neural responses to causal inferences that unfold automatically during language comprehension.