Speed and direction of locally-paired dot patterns

Phenomenal transparency in random-dot kinematograms is abolished when two motion directions are 'locally-balanced' by pairing limited-lifetime dots at each location [Qian, Andersen and Adelson (1994). Journal of Neuroscience, 14, 7357-7366]. Qian et al. also report that locally-paired stimuli appear as directionless flicker when the paired dots differ in their directions by 90 degrees or more. They attribute this to local inhibition between motion detectors more than 45 degrees apart. We investigated perceived motion in such displays, by requiring subjects to make direction and speed judgements with locally-paired stimuli containing two directions 60, 90 or 120 degrees apart. Subjects perceived coherent motion in these displays and made reliable direction judgements, indicating that the two motions are combined rather than interfering destructively. Our results show that the judged motion of locally-paired stimuli is in the vector-average direction of the two components. This vector-averaging rule also applies when the two sets of component dots differ in their velocity. Similarly, speed judgements comply with a vector-averaging rule for a range of speeds as well as for mixed-speed stimuli. These results suggest that the abolition of transparency does not necessarily imply abolition of a global motion percept. The local interaction abolishing transparency is not exclusively inhibitory, at least for directions up to 120 degrees apart, but generates a vector combination of the superimposed motions.