Abstract
In a fraction of a second people can comprehend novel scenes never experienced previously. To what extent does this capacity derive from familiar relations among clusters of objects? We addressed this question through a search task in which a subject verified a target object in a minimal scene consisting of three objects: two interacting (in physical contact) but the third by itself. Subjects were given a target name, e.g., “watch,” and had to decide whether the target was present in a brief (100 msec) masked presentation of the scene. Half the time the target was present either in the interacting pair or by itself. The critical variable was whether the pair of objects was in a familiar relationship, such as the watch on a wrist, or a novel one, such as the watch on a hammer. If familiar relations reduce scene processing demands, the target, when it is by itself, should be more quickly and accurately identified when the pair is in a familiar relation.