Abstract
After viewing thousands of images presented for a few seconds apiece, an observer can successfully distinguish old images from new ones (Standing, 1973). How is this accomplished? Do observers remember a list of objects? Do they remember the background (e.g. a beach)? In a variation of the standard picture memory experiment, we created photorealistic scenes of natural settings, urban places and indoors spaces. These consisted of a background and three added objects consistent with that background. Subjects were presented with 60 of these scenes for 3 seconds each, and then immediately tested for their memory. Test stimuli could have the same background, a similar background (e.g. a different beach with a similar spatial layout), or a different background. The 3 objects, too, could be the same or all different. Subjects made two responses; one about the novelty of the objects, the other about the background. Even though these stimuli resemble each other more than the usual picture memory stimuli, overall performance was quite good (objects: 74% correct, d′ = 1.37, backgrounds: 68% correct, d′=1.15). Poor performance was found in very specific situations. Whenever objects were incorrectly classified, performance on the background task dropped to chance. Whenever observers reported that an old background was new, they were poor at distinguishing old from new objects (d′ = .41). These results would be consistent with observers simply forgetting some scenes entirely. However, when observers erroneously reported that a new background was old, classification of objects remained quite good (d′=1.02 or 0.85 depending on the type of new background). Clearly, observers can recall the global background and the local objects in a briefly presented scene. These sources of information interact asymmetrically. Forgotten objects predict a forgotten background, but a forgotten background may accompany remembered objects.
Supported by NIMH MH56020.