Abstract
The phenomenon of “motion capture” has been described as a suppression of the motion of high spatial-frequency (SF) elements in favor of the motion of low SF components, allowing larger forms to capture their smaller features when they move (e.g., Ramachandran, 1990). However, recent evidence suggests that motion capture might not reflect active suppression of high SF signals, but instead is simply an instance of weighted combination of high and low SF signals (Durgin, Freeman, Huk, VSS 2005). Here we show that increasing the contrast (and thus the motion energy) of high SF components allows them to “capture” the low SF components, thus arguing against the need for the specialized motion capture mechanism originally posited. We utilized displays in which we independently manipulated the contrast and speeds of two superimposed motion components, consisting of randomly scattered 2D Gaussian elements of two different sizes (low-SF and high-SF). Participants were instructed to judge the speed of either the high-SF or low-SF elements relative to a standard speed (2.5 deg/sec). The irrelevant (unjudged) elements moved either 75%, 100%, or 125% of the standard speed. The contrast of the low-SF elements was always 20%; the contrast of the high-SF elements was either 20% or 80%. We measured the actual speed of the relevant elements necessary to perceptually match the standard speed (for each irrelevant element speed). When nominal contrasts were equal, the low-SF elements exerted larger effects on the perceived speed of the high-SF elements than vice versa. But when the high-SF elements were higher in contrast the relative weights reversed. A generic model of weighted motion integration accounts for both traditional motion capture as well as for the reversed capture of low SF components by high SF elements.