Abstract
To perceive a symmetry pattern, the observer has to determine whether two parts of an image are mirrored copies of each other about a symmetry axis and then to integrate the image feature pairs about the same axis to form the percept of a symmetry pattern. We investigated the role of color in such integration process in symmetry detection with a noise masking paradigm. In each 2AFC trial, a symmetric target was randomly presented in one interval while a random dot control was presented in the other and a random dot noise mask was presented in both intervals. The task of the observer was to determine which interval contained the symmetric target. The target density threshold was measured with PSI staircase method at 75% correct response level. In Experiment 1, the target was either red or blue while the masker was one of six possible chromaticity on the isoluminant plane. The symmetry detection threshold was greatest when the target and the mask were of the same color and decreased as the difference in chromaticity between the target and the mask increased. This suggests that the symmetry detectors are color-selective. In Experiment 2, both the target and the masker contained one, two or four colors. The axis orientation of the symmetric target was either 45° or 45° away from the vertical. The noise dot density ranged from 0 to 10%. The symmetry detection threshold increased with the noise densities. The symmetry detection thresholds decreased with the number of the colors in the stimuli at low and median noise densities but not at high densities. This result cannot be explained by independent color selective channels which would predict the threshold decreases with the number of colors. However, the result can be accounted for by a nonlinear integration across mechanisms.
NSC96-2413-H-002-006-MY3.