What induces capture in motion capture?

The phenomenon of "motion capture" has been demonstrated by presenting, one after the other, two identical Kanizsa squares spatially separated and superimposed on a regular matrix of dots. For appropriate temporal intervals, one illusory square is seen to jump from one location to the other and the dots in it appear to move with it even though they are physically stationary. The standard explanation of the effect is that motion signals from the subjective figure are spontaneously attributed to the static elements laying on it. We have found, however, that if alternative removal of right-angle sectors (required to obtain apparent motion of the illusory square) is not accompanied by alternative appearance and disappearance of a few dots, motion capture does not occur. This suggests that the basic mechanism underlying capture is not the motion of the subjective figure per se, but the spreading of motion signals arising from those texture elements that alternately go on and off between frames. On the other hand, subjective contours do play a role by confining the spreading of motion signals to the texture elements located on the figure.