EEG deconvolution modelling of naturalistic reading reveals surprisal effects modulated during mind wandering

Mind wandering (MW) during reading impairs comprehension and learning, and understanding its neural mechanisms is crucial for theory and educational applications. While electroencephalography (EEG) experiments using controlled stimuli identified neural correlates of semantic processing (e.g. N400) and suppressed activity during MW, studying their interaction requires naturalistic reading of long connected texts, rather than words and single sentences. This introduces methodological hurdles including the temporal overlap of neural responses and oculomotor artefacts. Our work addresses these by using co-registered eye-tracking and EEG with linear deconvolution to extract fixation-related potentials (FRPs) from reading a challenging 5000-word text. We quantified contextual predictability effects on the FRP using word-level surprisal, a robust predictor of N400 amplitude. Consistent with our hypothesis, we find a significant disruption in the surprisal effect on the N400 during MW, thereby contributing to the neural correlates underlying reading comprehension research and attentional state tracking in more ecological contexts.