In this paper, we analyse the Event-Related Potentials (ERPs) produced by cuts where the scenes before and after the cut are narratively related. In tests with 6 participants and 930 cuts from 5 Hollywood feature movies we found that cuts produce a large negative ERP with an onset 100 ms after a cut and a duration of 600 ms, distributed over a very large region of the scalp. The real-world nature of the stimuli makes it hard to characterise the effects of cuts on a trial-by-trial basis. However, we found that aggregating data across all electrodes and averaging the ERPs elicited by cuts across all participants (a technique we borrowed from collaborative brain-computer interfaces) produced more reliable information. In particular we were able to reveal a relationship between the length of shots and the amplitude of the corresponding ERP with longer scenes producing bigger amplitudes. We also found that amplitudes vary across and within movies, most likely as a consequence of movie directors and editors using different choices of cutting techniques. In the future, we will explore the possibility of turning these findings into a collaborative brain-computer interface for aiding test screening by evaluating whether specific cuts have their intended effect on viewers.
展开▼
