An observer viewing a pictorial representation of a three-dimensional (3-D) scene can attend either to the represented scene or to the flat surface of the picture itself. Although these two modes of perception are conceptually independent, we have shown that in practice there is substantial cross-talk between them (Sedgwick and Nicholls, 1993 Perception 22 Supplement, 109). We have also shown that the strength of this cross-talk can be affected by the balance between the visual information specifying the 3-D scene and the visual information specifying the picture surface. In the present research we dissected surface information into several components and attempted to determine their relative effectiveness in decreasing cross-talk from the depicted 3-D scene. We used computer-generated displays both to create the depicted 3-D scene and to simulate the picture surface. Observers made magnitude estimates of the projected size on the picture surface of rectangular objects depicted as standing on a receding platform. We examined four components of surface information: (1) a realistic-looking frame surrounding the picture; (2) a visible surface texture (resembling woven fabric); (3) the pixel resolution of the image (512 versus 2048); (4) a background colour linking the frame to the image. We found that adding texture and adding framing each decreased cross-talk by about the same amount, but that the framing effect interacted with image resolution. Neither image resolution nor background, taken individually, significantly affected the strength of the cross-talk.
