Taking boundary extension to the extreme
63.411, Wednesday, May 15, 8:30 am - 12:30 pm, Orchid Ballroom
Ralph Hale1, James Brown1, Benjamin McDunn1, Aisha Siddiqui1; 1University of Georgia
A recent model of boundary extension (BE) suggests a number of factors contribute to why people seemingly always remember seeing more of a studied view than was physically present (Intraub, 2011). According to the model we interact with the world from an egocentric frame of reference which leads us to expect that most images we encounter continue beyond their boundaries. Our strategy is to alter cues about the spatial expanse of the scene to see if BE can be reduced/eliminated. Previously we showed BE is still found even when semantic information about objects and scenes was removed by using abstract shapes on random dot backgrounds (VSS, Siddiqui et al. 2012, 2011). These same abstract stimuli were used here with extremal edges introduced at the image boundaries. Extremal edges are the horizons of self-occlusion on smooth convex surfaces and are a powerful cue to depth and figure-ground organization (Palmer & Ghose, 2008). BE was hypothesized to be reduced for images with extremal edges at their boundaries because the edges indicate the images end there. Two size change conditions (16% & 40%) and two viewing conditions (extremal edges present or not) were tested. Forty images were shown to participants for 15 seconds each. Half the images were close-up and half wide-angle versions. After study the forty images were shown again, half at the same view angle and half at the opposite while subjects rated their perceived size. Without extremal edges normal BE was found for the 16% size change condition while there was reduced BE in the 40% condition. With extremal edges BE was reduced for the 16% and eliminated for the 40% condition. Extremal edges provided such a strong cue about the shape/extent of our images there was little reason to expect them to extend beyond their borders.