Purchase this article with an account.
Lesley Deas, Laurie Wilcox; On the consequences of perceptual organization via good continuation in depth. Journal of Vision 2015;15(12):837. doi: 10.1167/15.12.837.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
We have previously reported that mid-level configural effects – specifically, perceived closure – play a significant role in processing the binocular disparity of line segments (Deas & Wilcox, 2014). Here we demonstrate that the Gestalt cue of good continuity has a stereoscopic counterpart (‘good continuation in depth’) that operates when the relative disparity of neighbouring features varies smoothly. This disparity-based grouping cue negatively influences perceived depth (Experiment 1) but in similar stimuli enhances detectability (Experiment 2). In our first study we assessed the effect of good stereoscopic continuation on perceived depth magnitude using a touch-sensitive sensor. First, the relative separation in depth between two isolated dots was compared to estimates made when intermediate elements were added to form a continuous disparity gradient. We found that the perceived separation in depth between end dots systematically declined as intervening dots were added. Importantly, veridical depth was restored when the disparity of the intermediate dots was jittered. In Experiment 2, the same dot configurations were used in a visual search paradigm to evaluate if disparity-based grouping has a positive impact on search time. Observers searched for a target, defined by good stereoscopic continuation, among distractors which contained depth jitter, or vice versa. By modulating the disparity profile of the target relative to the distractors, we found that detection was dramatically more efficient when the target path contained a continuous disparity gradient. Our results demonstrate the operation of a disparity-based grouping cue that corresponds to the Gestalt principle of good continuity in depth. We posit that this disparity-based grouping may be partly responsible for the well-documented underestimation of perceived slant.
Meeting abstract presented at VSS 2015
This PDF is available to Subscribers Only