Abstract
Edited movies are sequences of different shots connected by cuts every couple of seconds. In spite of these abrupt image changes, professional editing often results in seamless perceptual experience. We hypothesized that after cuts, attention is primed for features fixated immediately before the cut, enabling instantaneous recognition of related image content. In a first series of experiments we presented two edited sports movies in parallel on the same screen and instructed observers to attend to only one of them. At each cut we interrupted playback for a central fixation. Playback then resumed either at the same or at a switched position. Observers' performance in identifying the previously attended movie was significantly better when visual similarity of the two sequential shots was high than when it was low, and when the movies' positions were the same. In a second series of experiments we investigated eye movements following cuts in edited full-screen movies. We produced 240 Full HD clips of street scenes and presented them in sequences of two in smaller 4:3 crop frames (taken from the left or right border of the original movie). The first clip ended with a blank and central fixation. This was followed by a second clip showing either (1) the complementary crop frame of the same source movie, or (2) the identical crop frame of the same source movie, or (3) an unrelated movie. Assignment of the videos was balanced across participants, allowing us to gauge the contribution of prior visual experience to gaze guidance after within-scene cuts [as in (1) and (2)] against scene exploration after between-scene cuts [as in (3)]. Our results provide strong empirical evidence for movie editors' intuitions and help to understand how memory and attention enable the visual system to combine sequential shots into coherent perceptual experiences.
Meeting abstract presented at VSS 2015