Abstract
How sensitive is the visual system to the spatiotemporal continuity of the incoming visual stream during dynamic real-world scene viewing? We investigated this question using film. Participants watched 35 one-minute clips without scene cuts from the movie 1917 . We disrupted their experience of spatiotemporal continuity in these clips by shifting the clips forward or backward in time by 500ms or 1000ms during saccadic eye movements. Due to saccadic suppression, transients associated with the jumps themselves were not perceived, and we asked participants to report any disruptions in their experience by pressing a button. We found that on average participants failed to detect about 27% of the discontinuities overall (compared to a false alarm rate of 0.01%). Participants were more likely to miss a discontinuity forward in time (32%) compared to a backward in time (22%), and were more likely to miss discontinuities of 500ms (32%) compared to 1000ms (23%). These variables interacted such that shifts backward in time were more influenced by the magnitude of the discontinuity. The results demonstrate that people frequently fail to notice abrupt disruptions in spatiotemporal continuity during an unfolding visual event. The decreased sensitivity to forward vs backward discontinuities raises an interesting possibility that prediction could play a role in this insensitivity, with forward prediction more accommodating of a forward displacement. In this presentation we will explore this and other factors that may explain these results.