Boundary extension: Findings and theories

Abstract

A view of a scene is often remembered as containing information that might have been present just beyond the actual boundaries of that view, and this is referred to as boundary extension. Characteristics of the view (e.g., scene or nonscene; close-up or wide-angle; whether objects are cropped, static, or in motion, emotionally neutral or emotionally charged), display (e.g., aperture shape and size; target duration; retention interval; whether probes of memory involve magnification/minification or change in physical distance), and observer (e.g., allocation of attention; age; planned fixation, gaze direction, and eye movements; monocular or binocular viewing; prior exposure; neurological correlates) that influence boundary extension are reviewed. Proposed mechanisms of boundary extension (perceptual, memory, or motion schema; extension–normalization; attentional selection; errors in source monitoring) are discussed, and possible relationships of boundary extension to other cognitive processes (e.g., representational momentum; remembered distance and remembered size; amodal completion; transsaccadic memory) are briefly addressed.

Keywords:

• Boundary extension
• Scene perception
• Spatial memory
• Perceptual schema
• Picture perception

Notes

¹ One of the most obvious differences between boundary extension and representational momentum discussed in Intraub (2002) and that led to a rejection of the notion that boundary extension and representational momentum might be related was in the apparent time course of each type of displacement; at that time, representational momentum had been shown to increase in magnitude rapidly during the first few hundred milliseconds after a stimulus vanished (for review, Hubbard, 2005), but there was no evidence that boundary extension exhibited such a rapid time course. However, Dickinson and Intraub (2008) subsequently reported that boundary extension exists after just 42 ms and was larger after 250 ms, and this is much more compatible with the time course of representational momentum. Such a rapid time course for boundary extension is also essential if boundary extension precedes representational momentum (as suggested by Munger et al., 2005).

² Hubbard (2005, 2006b, 2009) suggested that any account of representational momentum should begin with consideration of the underlying computational theory. Following Marr (1982), such a computational theory would focus on initially defining a problem that an organism needed to solve; in the case of representational momentum, the problem involves locating objects in and navigating through the environment. Interestingly, boundary extension could be considered a different solution to this same general problem. Thus, it is possible that ideas regarding a computational theory of representational momentum could be extended to include boundary extension in a broader computational theory of displacement in spatial representation. This would be consistent with Hubbard's (2006a) suggestion that boundary extension and representational momentum (a) reflect operation of a more general mechanism that biases representation in ways consistent with past experience, and (b) help bridge the gap between perception and action (and compensate for delays in perception due to neural processing times) by allowing observers to anticipate what would be likely to be seen in the next fixation or moment of time. It would not be surprising to find that boundary extension and representational momentum use different implementations or algorithms to implement their respective displacements, but such considerations are independent of a more general computational level theory that unites these two types of displacement as solutions to the same more general problem.