Abstract
We present the Distorted Curve Illusion, a new illusion of slant-induced shape distortion. Slanted inducers (parallelograms) distort the perceived profile of adjacent curved shapes (semi-ellipses) by making the minimum-point of the curve appear laterally shifted. The curve appears compressed on one side and extended on the other. We measured the illusion using matching (Experiment 1) and the method of adjustment (Experiment 2). Based on previous illusions of elongation (Coren & Girgus, 1972; Schloss, Fortenbaugh, & Palmer, 2014), we predicted that the illusion would be reduced when the similarity between the inducer and target decreased. In Experiment 1, participants viewed configurations containing a parallelogram inducer (black or white), which shared its bottom edge with the flat side of a black semi-ellipse target. Below, there was a row of nine comparison curves. The center curve was a perfect semi-ellipse (0 pixel shift). The minimum-point of the four curves rightward/leftward of center had +/-4, +/-8, +/-12, and +/-16 pixel shifts, respectively. Participants indicated which comparison curve appeared to match the target curve. We tested three target curvature conditions (0, +/-4 pixel shifts) and 5 inducer slants (0, +/-15°, +/-30°). When the target and inducer were the same color and slants were moderate (+/-15°), the perceived minimum of the target curve shifted to assimilate with inducer slant. The illusion diminished for large slants (+/-30°). Surprisingly, the opposite was true when the target and inducer were different colors (interaction p < .001). The perceived minimum-point shifted in the opposite direction of the inducer slant, with increasing magnitude as the inducer slant increased. Experiment 2 produced similar results when participants matched an adjustable curve to the perceived curvature of the target. The pattern of results can be potentially explained by assimilation of shapes within the same reference frame and contrast between shapes in different reference frames.
Meeting abstract presented at VSS 2016