The object-based representation of partially occluded surfaces in short-term visual memory: evidence from image combination

Distinct short-term and longer term visual representations of form allow people to combine mental images of separately presented pictures in order to identify a novel object (see, e.g., Brandimonte, Hitch, & Bishop, 1992; Hitch, Brandimonte, & Walker, 1995; Walker, Hitch, Dewhurst, Whiteley, & Brandimonte, 1997). The present study focuses on the short-term representation and asks whether it describes the 2-D features in a picture or the depicted surfaces in 3-D space (see Nakayama, He, & Shimojo, 1995). To-be-combined figures were sometimes partially occluded by irrelevant forms, and it was determined whether image combination was contingent on the directly visible regions of the figures or on the perceptually completed figures. Results showed that image combination was not impeded by partial occlusion, though it was impeded if an occluder was removed from a picture and only the previously visible regions of the figure were presented. It is concluded that the short-term visual representation supporting image combination is not a pictorial code, but is an object-based description of completed surfaces in 3-D space.