Abstract
Convexity context effects (CCEs) have been demonstrated in figure-ground perception, where the figural bias of convexity increases from 57%-89% as the number of alternating convex and concave regions increases from 2 to 8 in 100-ms displays (Peterson & Salvagio, 2008). In 2-Region displays Ghose and Palmer (2010) found that when convexity conflicted with Extremal Edges (EE), the EE-side was perceived as closer on ~98% of trials (2000-ms exposures). We examined whether CCEs are obtained when convexity and EE cooperate (Congruent condition) or conflict (Incongruent condition). Observers reported whether a red probe located on the region to the left or right of a central border appeared to lie on the region they perceived as figure or ground; they were instructed that figures were both shaped and closer near the border. Convex regions were perceived as figures more often in 8-Region (92%) than 2-Region (77%) Congruent displays, p<.0001. Thus, EE benefits from context when combined with convexity. In contrast, convex regions were not perceived as figures more often in 8-Region (45%) than 2-Region (37%) Incongruent displays (p=0.089). Thus, when EE and convexity conflict, neither benefits from context. EE regions were perceived as figures on the majority of trials with Incongruent 2-Region displays (70%), replicating Ghose and Palmer, yet our observers perceived EE regions as figures less often than their observers perceived them as closer. Task or exposure duration differences may contribute to this result. A novel factor that affected 2-Region displays only also contributes: EE regions were less likely to be perceived as figure when they lay on the right (56%) versus left (84%) of the central border (p<.0001). We attribute this new finding to an influence on EE from a light-from-left bias that was overcome by context in 8-Region displays. Figure assignment is a complex interaction among a variety of cues.
Meeting abstract presented at VSS 2014