Eye Movements and Memory for Objects in Scenes
As we move our eyes from object to object in a scene, what information do we retain from each object we have looked at? And, what effect does the order in which we have looked at those objects have on our memory for them? Research on change blindness, in which changes made to real world or pictorial scenes go unnoticed, even when a person is looking at the scene, has suggested that people remember little of what they see in scenes—otherwise they would surely notice the change(!) However, our research in collaboration with Gregory Zelinsky, of SUNY Stony Brook (Zelinsky & Loschky, 2005), has shown that memory for an object in a scene can last over several intervening fixations on other objects in the scene, suggesting that memory is better than some have suggested base on change blindness studies. Nevertheless, our research has also shown that memory for an object rapidly decreases as you fixate at the first two to three subsequent objects. This recency effect is based on fixation sequence, not simply decay of memory over time. Thus, the change blindness phenomenon my in part be explained by this gradual loss of memory as one looks from object to object in a scene. We propose that a prominent theory for explaining memory for objects in scenes, Object File Theory, must be modified to take into account the rich body of theory and research in the classic memory literature. This research should also be applicable to areas in which it is important to know what information a person attends to and remembers during task performance, for example information displays for airplane cockpits, air traffic control, driving simulators, etc.
In a recent collaboration with Dr. Donald Varakin of Knox College (Varakin & Loschky, 2010) we have also investigated how memory for objects in scenes is combined (or not) with memory for the specific views of the scenes in which those objects were seen. Previous research has shown that visual long-term memory (VLTM) stores detailed information about object appearance. Our studies investigated whether object appearance information in VLTM is integrated within representations that contain picture-specific viewpoint information. In three experiments using both incidental and intentional encoding instructions, we found that participants were unable to perform above chance on recognition tests that required recognizing the conjunction of object appearance and viewpoint information . However, we found that performance was better when object appearance information or picture-specific viewpoint information alone was sufficient to succeed on the memory test. Thus, our results suggest that object appearance and viewpoint are not episodically integrated in VLTM for scenes.