Abstract
When remembering occluded objects, it is possible visual memory fills in missing information, representing more of the object than was actually presented. This amodal completion would be similar to a boundary extension bias where scenes are remembered from a greater distance with extensions of the edges and completion of objects on the edge. If boundary extension operates in visual memory for occluded objects, we should demonstrate a bias to remember objects as less occluded than presented. Participants were shown images of 120 objects that were occluded 40% to 60% (randomly intermixed) with the object occluded on a single side or in alternating visible and occluded stripes. After the study phase, participants were given a five-alternative memory test in which they had to select the previously seen level of occlusion (40-60%). Boundary extension would be evident if memory errors were skewed toward selecting foils with less occlusion than was presented. We restricted the analysis to the 45%, 50%, and 55% occluded conditions because they had memory foils to either side of the correct answer. A boundary bias score was created where a correct answer was given a 0, boundary extension was given a -1, and boundary constriction (remembering a less visible object) was given a 1. Overall, we failed to find consistent boundary extension bias. For the 50% occlusion condition, the boundary bias score was not different than 0. For the 45% occlusion condition, we found a boundary constriction bias, but in the 55% occlusion condition, we found a boundary extension bias. In other words, participants appeared biased to guessing more toward the middle level of occlusion rather than remembering the object as more visible. These results indicate that, although boundary extension may be operating in scene memory, the same bias is not found when remembering occluded objects.
Meeting abstract presented at VSS 2018