Abstract
Previous research has suggested two ways in which the eyes move during scene viewing (Henderson & Smith, 2009): (i) process monitoring proposes that decisions regarding when to move the eyes are based on the continuous processing of the viewed scene, and (ii) autonomous control suggests that these decisions are driven by some general internal strategy such as a timer. We investigated these views by presenting global scene transitions. Participants’ eye movements were monitored as they were presented with two images depicting natural scenes, each for 4.5 s, thus introducing a global transition as one scene changed to the other. They were then presented with an object from one of the two scenes, and asked to identify the scene that contained the object. We observed two distinct groups of fixations when the scene transition was introduced. The End-Early group (32% of the fixations) ended less than 100ms after the transition, exhibited shorter fixation durations (mean±SEM of 197.06±16.23ms vs. 249.53±11.16ms for baseline) in the first fixation, and moved towards the center of the scene in the second fixation. The End-Late group ended more than 100ms after the transition, exhibited longer fixation durations (329.19±17.29ms vs. 252.36±9.01ms) during the transition, and moved towards the center in the first fixation. This suggests that immediately after a transition, End-Late fixations aborted the previously-programmed saccade, and reprogrammed to the center, which may serve as a vantage point to re-start scene exploration (Tatler, 2007). For the End-Early fixations, the transition probably occurred too late for saccade reprogramming, which caused the pre-planned eye movement to be executed, before moving to the center. This suggests that eye movements were optimized during viewing, i.e. they were planned and executed while continuously monitoring the visual scene, which supports process monitoring. We did not find any evidence of timer control in our data.
Meeting abstract presented at VSS 2012