Abstract
When processing partial views of scenes (e.g., photographs) viewers activate expectations about the spatial layout that may exist outside the view's boundaries. As a result viewers remember having seen a larger expanse of the scene than had been presented (“boundary extension”). We examined how the distribution of attention across the scene affects the extrapolation processes. Participants viewed photographs depicting two objects in naturalistic scenes. One of the objects, the front object, was close to the camera. The other, the back object, was farther away from the camera. In Experiment 1, viewers were told to pay attention to one of the objects. The location of the “important” object was cued before each picture was presented. Eight photographs were presented for 350 ms each. A memory test showed that viewers extended the scene around both objects. However, more extension was obtained for the front objects than for the back objects independent of whether they were the attended objects or not. In Experiment 2, viewers were presented with the same photographs but were told that both objects were important. In this experiment viewers showed equal extension around both the front and the back objects. It is suggested that focusing attention on one object has the effect of segmenting the scene expanse into two sections, one around each object. Each section was extended differently depending on the size and location of the main object in it. However, when attention was more global, extension was global as well.