Abstract
Spatially discontiguous sections of a partially occluded visual object are often easily identified as sharing a common source. For example, when two collinear lines abut opposite sides of a circle, we perceive a single line beneath a disc-shaped occluder. In the present study we investigated mechanisms involved in such amodal completion. Without an occluding surface, it is well established that observers are less sensitive to misalignments in oblique line segments than vertical line segments. We asked whether observers' vernier acuity for vertical and oblique lines predicts sensitivity to misalignments when the same lines are separated by an apparently occluding disc. We expected worse sensitivity when an occluder was present, due to additional visual noise in the positional mechanisms used to judge alignment, and that this noise might exacerbate the vernier oblique effect. Using a novel 2IFC method of constant stimuli that deconfounds sensitivity from biases such as the Poggendorff illusion, we compared observers' objective vernier acuity thresholds with and without an occluder. We found a large oblique effect, but, contrary to our predictions, no evidence that occluders affected observers' objective thresholds. We next measured observers' subjective thresholds when judging the apparent continuity of line segments under the same conditions and stimulus parameters. We reasoned that if observers' subjective judgments are informed by the same mechanisms used to detect misalignments, occluders should not influence appearance. However, observers reported that lines appeared continuous over a wider range of misalignments with an occluder than without, but only for oblique lines. This dissociation between objective and subjective thresholds suggests that the appearance of an occluded object involves mechanisms other than those used to detect the positions of the object parts.
Meeting abstract presented at VSS 2016