Conversation in fMRI: DMN Activates During Live Mentalizing

Abstract

Although language might have evolved in a conversational setting, neural language processing is rarely studied in this setting. It is also important to delineate the roles of networks interacting with the language network, such as the default mode network and networks subserving social cognition or cognitive control. We studied the fundamental question of how conversational production and comprehension processes may differ. By acquiring fMRI data of participants engaging in conversations on ethical dilemmas with a confederate, we aimed to study extra-linguistic areas during live mentalizing, crucially with long (10 min) conversations. A subject-specific network analysis was used to complement group-averaged whole-brain analysis. The results replicate asymmetries between production and comprehension already observed using short conversations. This included left inferior frontal gyrus preferences for production and superior and middle temporal gyri preferences for comprehension. Activation in the posterior dorsal medial prefrontal cortex was specific to production, a result that is stable across conversational durations. Notably, we found preferential activity of the default mode network (DMN) during comprehension, specifically during longer conversations. The frontoparietal network for control (FPN) showed the same pattern, although it was less robust. This might indicate simultaneous self-reflective, Theory of Mind, and/or situational processing during comprehension, potentially due to lower conversational pressures, compared to production. The subnet-work DN-B was more strongly activated than DNA, across both comprehension and production, suggesting that Theory of Mind remains continuously active throughout conversation. Our findings thus advance the understanding of how several networks contribute to production vs comprehension during conversation.