Does narrative drive dynamic attention to a prolonged stimulus?

Attention in the "real world" fluctuates over time, but these fluctuations are hard to examine using a timed trial-based experimental paradigm. Here we use film to study attention. To achieve short-term engagement, filmmakers make use of low-level cinematic techniques such as color, movement and sound design to influence attention. To engage audiences over prolonged periods of time, narrative structure is used. In this experiment, participants performed a secondary auditory choice reaction time (RT) task to measure attention while watching a film. In order to explore the role of narrative on attention, we manipulated the order that film segments were presented. The influence of narrative was then compared to the contribution of low-level features (extracted using a computer-based saliency model) in a multiple regression analysis predicting choice RT. The regression model successfully predicted 28% of the variance in choice RT: 13% was due to low-level saliency, and 8% due to the narrative. This study shows the importance of narrative in determining attention and the value of studying attention with a prolonged stimulus such as film.