Where do objects become scenes?

Abstract

A pair of objects can be depicted side-by-side or interacting. W. Epstein et al. (1960) showed that the depiction of interacting objects (e.g., a pipe on a clock) facilitated cued recall compared to when the same images were shown side-by-side. The saliency of object interactions, readily described by prepositions or gerunds, is immediately appreciated. What are the neural loci of these interactions? Does novelty of the interaction matter? In an fMRI block-design, subjects viewed a series of minimal scenes, each consisting of a pair of objects. The objects were presented either side-by-side (e.g., a bird and a birdhouse) in some blocks or interacting with each other (e.g., the bird perched on the birdhouse) in other blocks. To examine the effects of novelty, on some blocks one of the objects in each pair was switched with an object from another pair (e.g., an ear for the birdhouse) so now the interaction would depict the bird on the ear. Interactions produced greater BOLD activity both in the lateral occipital cortex (LOC) and, only under some conditions, in the parahippocampal place area (PPA) as well. Novelty of the interactions magnified this gain, an effect that was absent in the side-by-side depictions. Controls ruled out eccentricity, relative size, task influences, or feed-forward activity from earlier visual areas as possible sources of the interaction effect. The results thus suggest that objects can become scenes as early as LOC. The distinctive relation of each interacting pair likely elicited additional associations—more so for novel relations—and it is that activity that might have magnified the BOLD responses in these higher visual areas.