Preparations to break the Tyranny of Film: Designing stimuli that change comprehension and moment-to-moment content importance in video

Poster Presentation: Sunday, May 19, 2024, 8:30 am – 12:30 pm, Pavilion
Session: Attention: Spatial selection 2

Prasanth Chandran1 (), Alyssa Smith2, Ashton Lofing2, Lindsey Wilson2, Brayden Wallace2, William Mass2, Joseph P. Magliano3, Lester C. Loschky4; 1Doctoral Student, Kansas State University, 2Undergraduate Student, Kansas State University, 3Professor, Georgia State University, 4Professor, Kansas State University

We know that people's visual attentional selection is strongly influenced both by top-down tasks, and bottom-up salience. But, how does viewers' understanding of what they see, namely their event model, influence their attention? We have tested this in narrative film, through context manipulations that influence viewer's event models. However, in prior studies, large differences in viewers' event models produced small differences in their attentional selection, termed the Tyranny of Film. We hypothesize this is because film stimuli have powerful stimulus features to guide viewers' attention. Based on the Scene Perception & Event Comprehension Theory (SPECT), we propose two necessary conditions for strongly attenuating the Tyranny of Film: 1) There are predetermined differences in viewers’ event models (specified via event indices), focusing on a target of attention. 2) The target of attention has relatively low visual salience. Method: We used 10 pairs of silent commercial video clips that could satisfy the first necessary condition. We created a context manipulation for each clip, and a common viewing period that would be watched 1) without having seen the context, and 2) after having seen the context. After each viewing, participants were asked to watch the clip again and use their mouse to 1) track the most important person/place/thing at each moment, and 2) click at particularly important moments. Results: Participants hovered their mouse longer, and clicked more on the targets of attention when in the Context condition. Thus, the context manipulation influenced viewers' moment-to-moment judgments of importance on the screen, further confirming that our selected videos meet the first criterion for breaking the Tyranny of Film. To test the second necessary condition, we will analyze the common viewing period video clips using the DeepGaze-3 saliency algorithm, and determine whether the targets of attention have relatively low saliency. Future studies will measure eye movements.

Acknowledgements: Research reported in this abstract was partially supported by the Cognitive and Neurobiological Approaches to Plasticity (CNAP) Center of Biomedical Research Excellence (COBRE) of the National Institutes of Health under grant number P20GM113109.