Mapping content and dynamics in the stream of consciousness through latent brain state analysis.
iScience November 21, 2025 Yuhua Yu, Quentin Raffaelli, Delaney Freveletti et al.
The brain supports unprompted, stream-of-consciousness thoughts through a multistage neural architecture. Using fMRI while people spoke their thoughts aloud, recurring brain states were identified that track moment-to-moment thought content. A brain state involving core default mode network regions tracked the early emergence of internally oriented thought. A state engaging executive control and dorsal attention networks tracked externally oriented thought. A third state involving the medial temporal lobe and dorsal attention network engaged later during internal thought verbalization, suggesting roles in elaboration. This latter state was also linked to a large-language-model derived measure of surprisal, signaling novelty and internal event boundaries. Default mode network subsystems interact with attention and control networks to scaffold the fluid progression of spontaneous thoughts.