Abstract
Purpose: We examined the extent to which recognition of a structured object influenced stereo discrimination.
Method: Our stimuli were static frames of a point-human figure (13 points) in stereo. Of the 13 points that were otherwise green, two were red depicting one forearm (elbow and wrist), and two blue depicting the other. Subjects discriminated, in a staircase procedure and without feedback, whether the 3D distance between the red points was longer than between the blue ones. The 2D distances of the two pairs, in contrast, were always kept the same in each trial. The static human figure was presented upside-down so no naïve subject recognized it. After Session 1, the upright human figure in stereo was presented in a movie so that all subjects recognized it. The next session repeated Session 1 except that the human figure was now upright.
In the first control condition, the human figure movie was not shown after Session 1, while everything else remained unchanged.
In the second control, the 3D distance between the shoulder and wrist was used instead. This distance remained the same on average as that between the elbow and wrist in the experimental condition.
Results: In the experimental condition, although subjects' threshold decreased from Session 1 to 2, this decrease was smaller than in the first control (p<0.05). No subject in the first control recognized the human figure. In the second control, in contrast, the threshold increased (p<0.025).
Discussion: Stereo distance discrimination was impeded when two pairs of points, which were otherwise meaningless, were recognized as a human figure's two forearms that were presumed equal in length. Discrimination was also impeded when the recognized context distracted an otherwise direct comparison (an elbow in between shoulder and wrist).