Abstract
Change-detection (CD) studies and eye-tracking studies have largely used static images rather than videos as stimuli; adding dynamism allows investigation of the influence of an unfolding event on behaviour. The study reported here is the first of a series investigating how task goals and the nature of the event in the video affects CD and the way people inspect one kind of video: CCTV footage. Two 2-minute videos, matched except for a critical 10 sec, showed a waiting room with views alternating between two cameras’ viewpoints every 5 sec. In one video a crime occurred (theft of a dropped phone); in the other, no crime occurred, and so there was less of a plot. In both videos, a target change occurred starting concurrent with the start of the crime: two people depicted in different camera views switched locations. One of those two people was the thief in the crime video. Fifty participants were instructed to look for a crime; fifty to look for something unusual occurring and fifty simply to watch the video. Eye movements were recorded. CD (24% overall) was unaffected by event type and task instruction. Eye movements, however, were affected by both factors and by whether or not a change was detected. Observers looking for a crime made more fixations on the thief and fixations of shorter duration overall than other observers. Observers of the crime video made fewer fixations before the crime and fixated the thief more during the crime than other observers, particularly when looking for a crime. Participants demonstrating CD had longer fixations on the people who changed before, during and after the change. They made more fixations and for longer durations, which we interpret as indicating more cognitive processing. Future studies will dissociate the change from the crime both spatial and temporally.
Meeting abstract presented at VSS 2013