Abstract
Observers remember seeing beyond the edges of a view (boundary extension: BE). A multisource model suggests that visual scenes are rapidly set within a spatial framework; at test observers misattribute top-down-generated expectations to having been seen (Intraub 2011). This misattribution (BE) is associated with activity in scene-selective brain areas (PPA & RSC: Park et al., 2007). In other research, inverted scenes yielded decreased PPA activity compared to upright (Epstein et al, 2005). Might BE similarly decrease for inverted scenes under time pressure? If the view must be mentally rotated to fit an "upright" spatial framework, this could disrupt development of BE in a speeded task. In contrast, if view-taking is fast and flexible, disorientation may have no effect. Observers were assigned to one stimulus-duration condition: 125, 250 or 375 ms (N=72 in each). On each of 36 trials a square photo-image was presented for the designated duration, interrupted by a 250-ms mask, and then immediately reappeared to be rated as the same, much closer or much farther (5-pt scale) than the brief view. Across trials, upright, sideways and inverted photos were intermixed. Significant BE occurred in all but the upright-125-ms condition. A 3 (Stimulus Durations) x 3 (Orientation) Mixed ANOVA on mean boundary ratings revealed greater BE as duration lengthened, F(2,213)=3.877, p<.025, and greater BE for disoriented views, F(2, 426)=8.089, p<.001 (supported also by contrasts): no interaction. Increased BE with increased viewing time suggests that the putative top-down surrounding representation becomes stronger with viewing, so that slightly more surrounding space is later misattributed to vision. Unexpectedly, disorientation slightly increased the BE error. As suggested by the inversion effect (Yin, 1969), disorientation may limit the quality of visual encoding, thus increasing similarity between visual memory and memory for the surrounding top-down scene structure, inducing more BE for brief, disoriented views.
Meeting abstract presented at VSS 2012