Boundary extension effect is larger in tilt-shift photographs
Jiří Lukavský, Filip Děchtěrenko · 2015
In boundary extension (BE) experiments, people report remembering the content, which was originally beyond the edges of a studied view (Intraub & Richardson, 1989). Close-up views are known to yield larger BE errors compared to wide-angle views. We tested the hypothesis that larger BE is not caused by field of view, but by perceived distance. We argue that extrapolation of visual input is more useful in the proximal space, because the view of proximal space changes more profoundly as people move in an environment. However, it is difficult to create comparable stimuli featuring proximal or distant content, because the photographs differ in spatial layout and different objects become attract attention. To overcome this problem we used tilt-shifted photographs, because they have the same layout, evoke same semantic references, but make the impression of toy-like scenery. We presented participants (N=25) with brief study view, followed by mask and asked them to adjust the test scene zoom level to match the study view. The protocol consisted of 12 distant scenes, 12 tilt shift versions of different distant scenes and 12 fillers. We found that irrespective of the initial study view zoom level (+/-10%), the responses for tilt-shift scenes were biased by 3.1% towards more zoom-out view. The observed pattern is consistent with the hypothesis about larger boundary extension found in proximal scenes. In a subsequent control experiment (N=32), we used gray gradient to degrade the information quality around the image boundaries and we found no changes in boundary extension effect.