Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session B: Wednesday, August 13, 1:00 – 4:00 pm, de Brug & E‑Hall

The relationship between pupil dilation and neural surprise in natural language comprehension

Quirin Gehmacher¹, Juliane Schubert, Aaron Kaltenmaier¹, Nathan Weisz, Clare Press¹; ¹University College London, University of London

Presenter: Quirin Gehmacher

Predictive processing theories propose that language comprehension involves generating and updating context-based expectations. We tested whether such predictions are reflected not only in neural activity but also in pupil-linked responses. Using GPT-2, we derived contextual predictions and analysed MEG and pupil data recorded during audiobook listening. Replicating prior work (Heilbron et al., 2022), we find that MEG responses are modulated by both lexical surprise and semantic prediction error. Extending this, we show that pupil dilation selectively tracks semantic prediction error, suggesting sensitivity to meaning-level violations. We assess the mapping function from surprise to these pupil and MEG measures, focusing on linear vs non-linear response profiles and discuss their relation with respect to current predictive processing theories.

Topic Area: Predictive Processing & Cognitive Control

Extended Abstract: Full Text PDF