Abstract
An important task of the visual system is to extract speed estimates from the two-dimensional image plane (2D) and from motion in depth (3D), or, as in the majority of scenes, from a combination of both. However, it remains unclear from available research how 2D and 3D speeds are combined. We addressed this issue using a square (5×5deg) Global Dot Motion (GDM) stimulus consisting of 100 circular dots (radius: 8minarc, 90 cd/m2) undergoing random motion on a grey computer screen (45cd/m2). Concurrently, the entire GDM stimulus was displaced in depth (by systematically changing binocular disparity) towards the observer. Using this stimulus, we determined how the 3D speed of the dot-defined square is affected by systematic manipulation of 2D dot speed. A two-interval forced-choice procedure was used. One interval presented the reference stimulus: a dot defined square moving in depth at speeds of 4, 6 and 9cm/sec with no 2D motion, while the other presented the test stimulus: an identical dot defined square, but containing dots moving randomly at speeds of: 0 0.4, 0.8, 1.6 and 3.2cm/sec on the image plane. The task of the observer was to adjust the test stimulus 3D speed to match the reference stimulus. Results show that increasing 2D speed increased the apparent 3D speed of the dot-defined square in a manner consistent with the vector sum of 2D and 3D velocities. We discuss our findings in light of the functional properties of global motion detectors tuned for speed information.
This research was funded by the University Grants Council of Hong Kong (CERG number HKU7426/05H) awarded to A. Hayes and S.K. Khuu