New EEG method reads story comprehension from film cuts

A team of researchers from multiple Hungarian institutions developed a novel approach to study narrative comprehension using continuous EEG recordings. Instead of the traditional event-related potential (ERP) paradigm that relies on repeated, isolated stimuli, they had participants watch short films while continuously recording brain activity. By aligning EEG data to sharp cinematic transitions (cuts), they extracted what they call transition-related potentials (TRPs). These TRPs exhibit the temporal structure of canonical ERPs, indicating significant information processing at each cut.

To isolate the effect of narrative context, the team compared responses to coherent films with scene-scrambled versions that had identical post-cut sensory input. They found that TRPs were systematically modulated by narrative coherence. Furthermore, they trained a compact deep neural network (DNN) to detect cut-related EEG signatures directly from group-averaged continuous recordings. The DNN generalized across different films and subject groups, reproducing the context-dependent effects of manually annotated cuts. The authors propose that this method can be adapted to other forms of continuous stimulation, providing a general tool for probing naturalistic human experience.