Abstract
Boundary extension is a phenomenon in which people remember seeing more of a recently-viewed image than was actually shown. Typically, close-angle images produce greater amounts of extension than wide-angle images, however, images must be perceived as having a continuous, yet truncated, view of the world. Current theories of boundary extension suggest the phenomenon involves a source monitoring error between mental representations created by sensory information present in the image and representations containing amodal information based on perceptual expectations. If this is the case images/scenes with reduced expectations to amodally continue should not produce or minimally produce boundary extension. In the current study we tested this idea by removing a number of factors thought to contribute to amodal continuation including familiarity, texture gradients, and occlusion information. We did this by using 2-D abstract shapes on random-dot backgrounds where the backgrounds were either occluded by the border or not and where the objects occluded the random-dots or not. We found boundary extension occurred in all conditions created to minimize amodal continuation. These results suggest modification of current boundary extension theories may be necessary to incorporate the present findings.