Purchase this article with an account.
Scott N J Watamaniuk; The perceived speed of global flow. Journal of Vision 2003;3(9):396. doi: 10.1167/3.9.396.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
When an observer views a random-dot cinematogram (RDC) in which dots are assigned new directions of motion from a distribution spanning up to and even greater than 180 deg, the perception is that of global flow (Williams & Sekuler, 1984). The direction of this global flow generally corresponds to the mean direction of the underlying direction distribution. However, though every dot is displaced the same distance, and thus moves at the same speed, the perceived speed of global flow slows as the width of the underlying direction distribution increases. Here, we investigated the perceived speed of global flow for different width direction distributions.
Observers viewed RDCs comprised of 256 dots, each assigned a direction from a uniform distribution of directions with a mean direction upwards (frame rate=60 Hz). Dots were displaced the same distance regardless of direction (constant step size). Each trial, observers were shown one interval containing a standard stimulus in which all dots moved in the same direction and speed (direction distribution width=0 deg) and another interval containing a test stimulus in which the underlying distribution of directions spanned 30, 60, 90, 120, or 180 deg. The speed of the standard stimulus was fixed at 12, 15, 18, or 21 deg/sec and the speed of the test varied from trial-to-trial. The point of subjective equality (PSE) was computed for each distribution width and standard speed.
The perceived speed of global flow closely matched the speed of the computed average vector in the mean direction of the underlying direction distribution. Thus, the individual dot displacement for stimuli with wide direction distributions had to be larger than that of the standard for the global flow speed to match that of the standard. These results suggest that for global flow, local motion vectors (direction and speed) rather than just local directions are integrated.
This PDF is available to Subscribers Only