The results are shown in
Figure 5. With the Regular expanding and contracting background (
Movie 3A), the magenta square looked smaller than the green square and had to be increased in size by 46% (leftmost bar,
Figure 5) to match the apparent size of the green square. Each square's contours are shifted in the direction of the background motion that comes directly after it making the green square appear bigger and the magenta square appear smaller. Since the background undergoes a four-fold change in size (a 300% increase), the combined size shifts driven by the regular background motion is about one sixth of that. With the Limited Lifetime background motion (
Movie 3B), there was less motion energy but also no persisting shapes in the random dots that could be followed. The result was slightly less effect: the magenta square needed only 36% increase in size to match the green square in size. With the Contrast Reversing motion, there was little or no apparent size difference between the green and magenta squares (
Movie 3C). The reverse apparent motion suppressed the effect of the motion but did not reverse it as was the case in
Experiment 1. In the Size Only condition (
Movie 3D), the random patterns increased and decreased in size, but there was no local motion signals because the random pattern was different on each frame. The result again was little or no effect on the relative apparent sizes of the two test squares. Finally, with the reintroduction of the explicit Banded contours in the background (
Movie 3E), the combined influence of the contours and background dot fields reinstated the large, more than 100% effect (a doubling in size) seen in the original article (
Anstis & Cavanagh, 2017). This is one third of the physical size change of the stimulus, which quadrupled in size (4:1), a 300% increase.