Orientation masks 3-Gabor alignment performance.

Abstract

Several workers have concluded that Gabor alignment tasks are performed by using central tendencies of the micropatterns as a cue. One reason for this conclusion was that the 3-Gabor alignment task is performed equally well whether the orientations of the patches are collinear or orthogonal to the group orientation. We wished to find out if the orientation of the micropatterns has any effect on performance. We tested subjects in 3-micropattern alignment tasks using a variety of orientational conditions. If three vertically-aligned Gabor patches were vertical, horizontal or both, or if bullseye or Gaussian blobs were used, no difference in performance was found. If, however, the orientation of the patches was randomized, performance became much worse. Similarly, if the three patches were at 45 deg, thresholds were raised. The effect of orientation was maintained across different spatial frequencies. Control conditions involving randomization of the phase of the sinusoidal carrier, or jitter on the size of Gaussian blobs, confirmed that a central tendency of the micropatterns was indeed being used by subjects, indicating that the role of orientation in this task is that of a mask, rather than of a cue.