Abstract
A large body of research has linked the ‘event structure’ of experience to core cognitive processes such as attention. For example, when subjects step through the frames of a movie in a self-paced slideshow, they dwell longer on certain frames, and these moments of greater attentional engagement turn out to be the same moments that subjects later mark as the ‘boundaries’ between successive events. An assumption in this method—and indeed, in most past discussions of event perception—is that event representations are perfectly contiguous, with each passing moment reflected in one representation or another. By contrast, here we asked whether event representations might actually omit stretches of time. Subjects placed markers on a movie timeline to indicate not the boundaries between events, but rather the start and end times of each event. Although subjects were encouraged to mark every event that they saw, large portions of each movie’s timeline (21%; around 8s of a 40s movie) were not included in any event interval. Why should this be? In a slideshow (completed prior to the event marking task), subjects dwelled less on frames that they later omitted from any interval, compared to frames that they later marked as starts, suggesting lower-information moments may be omitted. We also investigated another form of event segmentation—comic artists’ depictions of these same movies as sequences of drawn images. A second group of subjects indicated the first and last frames that each comic panel represented, again by placing markers on a timeline. This revealed that comics are a similarly sparse form of event representation, omitting on average 33% of their target movie, with omitted frames again occasioning lower dwell times than frames corresponding to the start of an event/panel. We conclude that event representations may omit time to a much greater extent than is typically acknowledged.