Abstract
Perceptual grouping (via closure) degrades stereoacuity and depth magnitude estimation e.g., McKee (1983) showed thresholds are elevated when two lines are connected to form a rectangle. Experiments that show these effects have all used line targets presented using stereoscopic displays with other depth information eliminated or held constant. However, there is evidence that when physical targets are used, the detrimental effects of grouping are eliminated. In these experiments we evaluate the extent to which the presence of ‘supportive’ monocular depth cues impacts the forementioned grouping effect. We used line configurations similar to those used previously in isolated and ‘closed object’ configurations with and without consistent perspective and motion parallax information. Four vertical lines were presented in two configurations 1) isolated lines and 2) with horizontal lines connecting the inner pair (grouped). Both configurations were rendered with consistent and inconsistent perspective foreshortening and were viewed while either stationary or moving. Stimuli were displayed using an Oculus Rift such that correct motion parallax was available in the motion condition. Observers indicated how much depth they perceived between the two central lines using a touch-sensitive sensor strip ranging between 0 to 4cm in 1cm steps. Our results confirmed that when only binocular disparity information was present, depth judgements were more accurate for isolated lines than closed objects. However, the disruptive effect of grouping was reduced when perspective cues were present (viewed without moving) and eliminated when motion parallax was also available. Taken together, these results support the hypothesis that presentation of binocular disparity information under unnatural, cue-restricted conditions makes it susceptible to contextual effects that may not represent how processing occurs in natural cue-rich environments. While the impact of mid-level perceptual grouping on isolated depth information provides important insights into visual processing, caution should be taken in generalizing the results to natural environments.