Abstract
In previous studies, we reported that subjects use scene-specific memory to guide attention in a repeated change detection task. In those experiments, different sources of information seemed to be hierarchical in that attention seemed to be allocated to previously relevant locations in a scene before previously relevant objects. However, in those experiments subjects indicated the presence of a change by clicking on the spatial location of the change. In experiment 1, we now show that even when object identity is made salient by requiring a response based on object identity rather than location, the allocation of attention based on location information continues to occur prior to allocation based on object information. In experiment 2, we show that location-based information is still effective at cuing relevant locations in a scene even when the scene's viewpoint is altered between two viewings. This lack of viewpoint dependence suggests that the scene-specific memory that is utilized when allocating attention in real world photographs is not identical to the implicit memory system underlying contextual cuing, which previous research has shown to be viewpoint dependent (Chua & Chun, 2003).