Abstract
Spatial relation perception (e.g., finding a plus above a dash, among dashes above pluses) is an inefficient process (Logan, 1994; Wolfe, 1998; Franconeri et al., 2012), but dramatically improves when the target is a single object, such as a short rectangle and tall rectangle with 0 degrees of spacing between them, compared to 1 degree between them (Nothelfer & Franconeri, VSS 2017). However, it is unclear whether this improvement stems from the target simply being a single object, or because the target's components are closer together and subsequently grouped more easily by proximity. An object-based hypothesis of spatial relation processing predicts that 0-degree spacing is a special case, such that finding this target relation among 0-degree distractors is easier than finding a non-0-degree target among non-0-degree distractors, independently of the actual degree of spacing. The purely space-based (i.e., object-blind) view predicts that search efficiency declines linearly as inter-object spacing increases from 0. Participants were asked to find a particular relation among the opposite arrangement (e.g., small/large rectangle among large/small rectangles). The rectangles (0.63 degrees in width) were separated by a space 0, 0.06, 0.19, or 0.57 degrees in width. Displays were divided into quadrants each containing 1-5 rectangle pairs with always the same degree of spacing within a trial, for a total set size of 4-20 rectangle pairs. Participants were asked to quickly indicate which quadrant contained the target relation. Visual search was more efficient when participants searched for two adjacent rectangles (0 degrees separated), rather than two separated rectangles (0.06, 0.19, or 0.57 degrees separated). Critically, additional space between the rectangles did not further degrade search, supporting an object-based view. These results show that visual processing of spatial relations is substantially better when judging a single object, and this is not simply due to closer proximity of its components.
Meeting abstract presented at VSS 2018