When you look at a picture, you are probably generally focused on the central objects, though the overall style might catch your eye. But what about your memory for the background of the picture? Put another way, how accurate are you at recalling the exact borders of a particular view? For that matter, how accurate are you at remembering the borders of the objects depicted in that picture? Carmella Gottesman and Helene Intraub of the University of Delaware were curious about these questions, and so took some special pictures that had objects stacked on objects ("Constraints on spatial extrapolation in the mental representation of scenes: View-boundaries vs. object boundaries," Visual Cognition, 2003). I've made a simple cartoon example, where my picture has some orange scissors on top of a parchment scroll, all on a teal background (Gottesman and Intraub used real objects, like a flip-flop sandal and towel resting on some grass).
There are two borders for us to consider: the view's border (or how much teal is between the scroll and the edge), and the border around the scissors (or how much parchment is between the scissors and the edge of the scroll). To find out what people were remembering, Gottesman and Intraub provided puzzle piece cut-outs of the stacked objects in various sizes, and asked individuals to select the pair they had studied. It turns out that people remember seeing more background than was really there--something Intraub and her colleagues call boundary extension. In other words, even though you might have seen the picture above, you remember seeing this picture (notice that is has more teal).
What about the other border in the picture--that of the scroll? It turns out that you don't get boundary extension for objects in the scene, just for the borders of the view. Gottesman and Intraub's results highlight that there is something different about how we view scenes, and that by extending the boundaries of the view we are anticipating what might lie just out of sight.