Abstract
After viewing a picture of a scene, our memory of it typically extends beyond what was presented—a phenomenon called boundary extension (Intraub et al, 1992). However, recent work has found evidence for boundary contraction as well (Bainbridge & Baker, 2020), proposing different outcomes for object-focused and scene-focused images. In naturalistic visual experience, there is a continuum of views that naturally transition between scene-focused and object-focused, e.g., as an observer enters an indoor environment and moves toward an object. Here, we examined how memory for a view of an environment changes along this object-scene continuum. We constructed 3D indoor environments using a virtual-reality game engine, allowing for tight control of their spatial dimensions and object content, and captured a series of images at systematic distances from a center object within each environment. We found continuous progression from boundary-extension in close-up views to boundary-contraction in distant views, with a consistent transition point across environments, and a clear relationship between the magnitude of the memory distortion and the position along the continuum. In a series of subsequent experiments, by manipulating the number of objects in the environment, we found that the memory transition point is not linked to a specific depicted distance per se, e.g., the transition between reachable and non-reachable space. Rather, the position at which boundary extension transitions to boundary contraction seems to track with the views that “looked best,” as judged by independent observers. Taken together, these results demonstrate that there is indeed a systematic link between the direction of memory distortion and the depicted scale of the space, but where object content has a clear influencing role. Further, the correspondence between the best view and the transition point suggests a common underlying mechanism, providing new empirical inroads for articulating the mechanisms of visual scene encoding.