Writing Through Sympathetic Arousal: A Perceptual Active Inference Model on the Effects of Sympathetic Arousal Interoception on Written and Spoken Grammatical Complexity
摘要
This study investigates the role of sympathetic arousal (SA) interoception in shaping the grammatical complexity of language production across writing and speaking modalities. Drawing from the active inference framework, we hypothesized that SA interoception drives the deployment of complex linguistic structures during language production. Under laboratory conditions, university undergraduate students produced short written and spoken texts while their galvanic skin response (GSR) was continuously recorded. Grammatical complexity was scored through a natural language processing algorithm, and sympathetic arousal was estimated using non-linear dynamic causal models fitted to the GSR data. A perceptual Markov decision process (MDP) model was then applied to capture the inferred interoceptive states and their influence on discourse production with high and low grammatical complexity. The results revealed that participants a priori believed that SA onsets would occur at the beginning of the discourse production task, regardless of the modality. However, when they produced written discourses, they showed higher interoception precision as to when the sympathetic arousal would occur—which was not accounted for by the time when they started to write. Furthermore, during the observation of a late SA onset, this higher precision more likely drove the production of registers with high grammatical complexity in the writing condition than in the speaking condition. While the reported MDP model has limitations when explaining the complex process of writing and its neural underpinnings, it is sensitive to capturing differences in SA interoception across language production modalities and grammatical complexity levels.