Abstract
When watching a movie, viewers are likely to find a cut from one shot to another more jarring if the two shots differed by only a small margin compared to two very different shots. According to Walter Murch, the small change may not be sufficient to compel viewers to re-evaluate the scene, but sufficient to indicate that something has changed. One interpretation is that the failure to detect a global context change, and the detection of a local change produce discomfiture. In the current study, we conducted two experiments that examined how abrupt scene transitions affected eye fixation patterns. Pairs of natural scene images were presented such that the first scene was shown for several seconds before a blank screen appeared for 15 ms during a fixation, followed by the second scene. In a subset of trials, the first scene was re-presented after the blank screen. In Experiment 1, we observed a shift of the eyes towards the scene center in the first saccade following the transition. This shift was greater when the scene changed, but was present to a smaller degree when the scene remained the same, suggesting rapid identification of global context changes before the first saccade. In Experiment 2, we varied the magnitude of scene changes by presenting pairs of scenes that overlapped to different degrees horizontally. At very small levels of scene change (1%), there were no changes in the shift of the eyes to the center, indicating a failure in detecting a global context change. At larger levels of scene change (>10%), the shift to the center increased significantly. Our results suggest that the detection of a global context change may be required to trigger a re-evaluation of the scene, and explains the discomfort when local changes are not accompanied by global context changes.
Meeting abstract presented at VSS 2015