Abstract
People tend to remember images with more information than was in the original image, a phenomenon known as boundary extension. Most studies of boundary extension have used small sets of images, often with a single object dominating the frame. Also, expanded and contracted versions consisted of separate photographs taken with different optical zoom settings on a camera. These methodological factors could contribute to the effect and constrain the generalizability of the phenomenon. In three experiments, we show that boundary extension in recognition memory generalizes to a large and varied set of scene photographs, with different versions created from the same digital original. In Experiment 1, subjects viewed 148 images for 4s each at study and then were given an old/new recognition test with images that were either identical to the studied images, boundary-extended to include more information, or boundary-contracted to include less information. Unlike earlier studies, the extension and contraction were accomplished by using smaller (768×576) or larger (1200×900) central sub-regions of an original digital image (2048×1536) than were used at study (960×720). All images, both at study and test, were resized and presented at 1024×768. False recognition was greater for extended scenes than for contracted scenes, generalizing the boundary extension phenomenon to a large set of digitally rather than optically modified images. Experiment 2 reduced the memory load by using multiple study and test blocks with fewer images and produced the same results. Experiment 3 eliminated the possibility that resizing contributed to the effect because extended scenes were reduced in size to 1024×768 whereas contracted scenes were increased in size in Experiment 1 and 2. In Experiment 3, all images were reduced in size to 720×540 with no reduction in boundary extension.
Supported by a grant from the University of Illinois Research Board.