Purchase this article with an account.
Susan Carrigan, Evan Palmer, Philip Kellman; Differentiating Local and Global Processes in Amodal Completion Through Dot Localization. Journal of Vision 2015;15(12):1123. doi: https://doi.org/10.1167/15.12.1123.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Competing theories of perception of partly occluded objects (amodal completion) have emphasized local contour interactions or global influences of symmetry and/or familiarity. These theories may reflect two different processes: a contour completion process and a more global recognition process. The two could be distinguished experimentally if only the former gives rise to precise boundary representations. Using a dot localization paradigm, we assigned participants to either a local or a global condition, which determined how the participant was instructed to complete objects with divergent local and global interpretations. On each trial, a small red dot was flashed on top of an occluder. Subjects reported whether the dot fell inside or outside of the occluded object’s boundaries. Interleaved adaptive 2-up, 1-down staircases were used to estimate two points on the psychometric function: the point, in terms of pixels away from the boundary, where the probability was .707 that the dot would be seen as outside of the occluded object’s boundaries (outer threshold), and the point where the probability was .707 that the dot would be seen as inside the occluded object’s boundaries (inner threshold). We examined imprecision, measured as the distance between these two thresholds. We also examined perceived location, defined as the mean of the inner and outer thresholds, as well as location error, measured as the absolute value of location. The results reveal that local contour interpolation produces precise and accurate representations of occluded contours, but completion according to mirror symmetry does not. This supports our hypothesis of two separate processes in contour completion: A low-level, local process completes contours according to local geometry and outputs a precise representation of interpolated contours; whereas a global recognition process outputs at least a decision regarding the likely shape of an object but does not produce precise representations of occluded contours.
Meeting abstract presented at VSS 2015
This PDF is available to Subscribers Only