Abstract
This study examined whether iconic memories for natural scenes, similar to those used in Rensink et al.’s (1997) change blindness experiments, exist. A change detection procedure was used which included 120 scenes. Changes consisted of a deletion to the scene. Scenes were shown to each subject in one of four ways: change with cue (30 trials), change without cue (30 trials), no change with cue (30 trials), and no change without cue (30). There were two main conditions, each with twelve observers who were shown every scene (500 msec), followed by an inter-stimulus interval (1500 msec), and the same scene with or without a change (500 msec). In condition 1, on half of the trials, a red arrow (cue) appeared 0 msec after offset of the first scene, pointing to the location of a possible change. In condition 2, the cue appeared 300 msec after offset of the first scene. Observers reported change and the identity of the pre-change item. In both conditions, change detection and identification benefitted from the cue. In the first, the mean frequency of change detection with a cue (35%) was more than twice that without the cue (14%), and the mean frequency of change identification with a cue (29%) was almost three times that without a cue (10%). In the second, the mean change detection with a cue was 34% and 17% without it. Change identification with a cue (24%) was double that without it (11%). These results indicate that more information is available from a brief presentation of a natural scene than change blindness suggests, and that this information is available at least 300 msec after scene offset. Clearly the information in the iconic memory of scenes is sufficiently processed to afford some identification of the pre-change objects.
Meeting abstract presented at VSS 2015