Abstract
Two experiments examined whether scene-specific memory guides attention to relevant objects and/or locations within a scene. In both experiments, subjects performed an initial block of flicker change detection trials in which they detected the addition of an object to natural scenes. After a 30 minute filled delay, subjects performed an unanticipated second block of change detection trials. In experiment one, the second block consisted of trials with entirely new scenes and new changes, trials that were identical to the first block, and trials with identical scenes but the item that changed was a different object and it appeared in a different location than the change in the original block. Results: Identical changes were detected quickly, suggesting that memory of the scene guided attention to the change. However, new changes in previously viewed scenes were detected no faster than changes in entirely new scenes, suggesting that the memory was limited to the relevant aspects of the original change location and/or object. In experiment 2, the second block of trials independently varied the location of the change and the identity of the change item to determine the extent to which faster change detection was due to memory of a location and/or the identity of the change object. Results: Changes that occurred in the original location were detected quickly even when a new object changed at that location, suggesting that attention was allocated to the behaviorally relevant location in the scene. However, detection was slow when the original object changed in a new location, suggesting that the identity of the object was not used to allocate attention. Results are consistent with the view that incidental memory is created while viewing a scene and can be used to guide attention to locations, but not specific objects, which were behaviorally relevant during previous experiences with environments.