Take a look at the following two movies. Your job is to determine whether the yellow square is moving faster in the first movie or the second movie.
If you’re like me, you’re probably cynical enough to guess that they were moving at the same speed. But if you’re honest and you just go with your initial impression of what you’re seeing, it’s hard not to perceive the second movie as much slower than the first one. What you’re witnessing is Michotte’s launching effect paradigm, first observed in 1946 by Albert Michotte: when two objects collide, we ascribe the motion of the second object to the motion of the first. Even though both yellow squares were moving at the same speed, since the blue square was moving faster in the first movie, it appeared as if the yellow square was too.
One way to see the effect in action is to measure it indirectly by using another phenomenon. Tim Hubbard and Susan Ruppel of Texas Christian University designed an experiment to do that (“A Possible Role of Naive Impetus in Michotte’s ‘Launching Effect': Evidence from Representational Momentum,” Visual Cognition, 2002). In their experiment, instead of asking participants about the velocity of the object, they tested participants for “representational momentum.”
Representational momentum was first observed in 1984 by Jennifer Freyd and Ronald Finke. The basic phenomenon is simple: watch an object move across the computer screen and disappear, trying to remember the exact spot where you last saw it. You will remember it as having moved farther across the screen than it really did. It’s as if the image in your head kept on moving, even after the object disappeared—as if your mental representation of the image has actual momentum. If an image is moving faster, the effect is larger, just like real, physical momentum. There are some cases in which the analogy to momentum breaks down—for example, we don’t have more momentum for heavier objects, even though heavier objects are harder to stop in real life—but otherwise, the phenomenon is very robust, and can even be observed for changes in brightness and for sounds.
Hubbard and Ruppel showed observers movies like the two at the beginning of this article, but had the target square disappear before it hit the edge of the screen. Then the observers had to click the cursor where they thought the objects had disappeared. As Freyd and Finke had found before, the observers consistently believed that the objects had disappeared farther along their path than the really had. Hubbard and Ruppel showed four different events—a fast and slow launcher, paired with a fast and slow target. The amount of displacement in the target didn’t vary according to the speed of the target, instead what mattered was the speed of the launcher:
Whether the speed of the target was fast or slow, the trials with the fast launcher had a larger displacement than the trials with a slow launcher. The difference in displacements between the fast and slow targets was not significant, but the difference in displacements between the fast and slow launchers was, regardless of the speed of the target. So the Michotte effect works even when people are being tested for something else—and we have another indication that what’s going on in our heads isn’t real momentum. We behave based on what we think has happened—our minds make a “best guess,” and usually that best guess is good enough. In this case, it’s not quite.