Discrimination of angular velocity in humans

BARRAZA J.F.; GRZYWACZ N.M.

Fort Lauderdale, Florida, USA

Congreso; ARVO Annual Meeting; 2001

ARVO

Purpose: Humans can perform fine speed discrimination tasks from moving object. However, to evaluate how fast rotational motions are, the angular velocity would be more useful. Werkhoven and Koenderink (1993) suggested that the visual system cannot discriminate angular velocity, because they found a bias in the matching velocity of two rotating annuli of different radii. Unfortunately, their results could not be explained by tangential linear speed discrimination either. Perhaps, angular velocity was not well discriminated because their stimului had regularly spaced points and thus provided poor rotational information. We retook this issue by using richer stimuli. Methods: In one experiment, stimuli consisted of two sequentially displayed random dot annuli revolving around a fixation mark. The reference stimulus had a given radius, while the test’s radius was either half or double that of the reference. The number of dots depended on the radius but was much larger than the number used by Werkhoven and Koenderink. In a second experiment, the stimulus was a random-dot disk in which the dots’ direction of motion were consisten with rotation, but the dots’ speed were constant. The percept was a non-rigid rotation, with angular velocity falling with radius. To compare the velocity of a given radius with the test, two circles were drwan near the desired radius. The test stimulus was the same as in the first experiment. We used a forced choice paradigm using the method of constant stimuli to yield psychometric functions and calculate the matching velocity for a wide range of radii. Results: in the first experiment, we found that subject equalize speed using angular velocity. Like Werkhoven and Koenderink our data showed a bias towards linear velocity, but in our hands the bias was exccedinly small (10%). In the second experiment, the matching velocity fell hyperbolically, which is consistent with using angular velocity rather than linear velocity. Conclusions: in the presence of a rotational motion, humans compute angular velocity. Moreover, even when the reference stimulus is a non-rigid rotation, subjects match angular velocities at the appropriate radii. This means that the mechanisms involved in the computation of rotational motion are metric.