Hack28.Motion Extrapolation: The
|
Hack 28. Motion Extrapolation: The "Flash-Lag Effect"
If there's a flash of light on a moving object, the flash appears to hang a little behind. How quickly we can act is slow compared to how quickly things can happen to usespecially when you figure that by the time you've decided to respond to something that is moving it will already be in a new position. How do you coordinate your slow reactions to deal with moving objects? One way is to calibrate your muscles to deal with the way you expect things to be, so your legs are prepared for a moving escalator [Hack #62], for example, before you step on it, to avoid the round-trip time of noticing the group is moving, deciding what to do, adjusting your movements, and so on. Expectations are built into your perceptual system as well as your motor system, and they deal with the time delay from sense data coming in to the actual perception being formed. You can see this coping strategy with an illusion called the flash-lag effect.1 2.17.1. In ActionWatch Michael Bach's Flash Lag demo at http://www.michaelbach.de/ot/mot_flashlag1 (Flash). A still from it is shown in Figure 2-23. In it, a blue-filled circle orbits a crosshold your eyes on the cross so you're not looking directly at the moving circle. This is to make sure the circle is moving across your field of view. Figure 2-23. In the movie, the circle orbits the cross and flashes from time to time Occasionally the inside of the ring flashes yellow, but it looks as if the yellow flash happens slightly behind the circle and occupies only part of the ring. This is the flash-lag illusion. You can confirm what's happening by clicking the Slow button (top right). The circle moves slower and the flash lasts longer, and it's now clear that the entire center of the circle turns yellow and the lag is indeed only an illusion. 2.17.2. How It WorksThe basic difficulty here is that visual perception takes time; almost a tenth of a second passes between light hitting your retina to the signal being processed and reaching your cortex (most of this is due to how long it takes the receptors in the eye to respond). The circle in Bach's demo moves a quarter of an inch in that time, and it's not even going that fast. Imagine perpetually interacting with a world that had already moved on by the time you'd seen it. So we continuously extrapolate the motion of anything we see, and our brain presents us with a picture of where the world most likely is now, rather than where it was a fraction of a second ago. This applies only to moving objects, not to stationary ones, and that's why the disparity opens up between the moving blue circle and the static yellow flashone is being extrapolated; the other isn't. Straightforward extrapolation of the path of moving objects is one way in which this effect can take place, and this happens as early as the retina itself during visual processing. The cells in the eye compensate for its slow response by being most active at the front edge of a moving object. (Without this, the most active cells would be the ones that had been exposed to the object the longest, that is, the ones at the back.2) That's one way in which the flash-lag effect could come about, because the delay for visual processing is compensated for with moving objects, but flashes still pay the penalty and are seen later. But that doesn't explain the demonstration movies constructed by David Eagleman and Terrence Sejnowski (http://nba.uth.tmc.edu/homepage/eagleman/flashlag; QuickTime). Essentially the same as Bach's demo, these movies have an erratically moving ring that should confuse the brain's motion prediction. In Experiment 1 (http://nba.uth.tmc.edu/homepage/eagleman/flashlag/r1.html; QuickTime), the ring abruptly changes direction at the same time as the flash. Still we see the flash lag behind the moving ring, even though prediction of the future motion of the ring could not have occurred. Eagleman and Sejnowski's explanation is that vision is postdictive. They argue that the brain takes into account changes in the scene that occur after the flash, for a very short time (less than a tenth of a second), and the motion preceding the flash isn't relevant at all. This is similar to the way two flashing dots can appear to be a single dot apparently moving [Hack #27] smoothly from one position to another, if the timing is right. Your brain must have filled in the interim motion retrospectively, because you can't know what in-between would be before the second dot appears. Similarly, the circle in this flash-lag experiment and the following fraction of a second comprise a period to be assembled retrospectively. The ring is moving smoothly after the flash, so you have to see it moving smoothly, and the flash appears slightly behind, because by the time you've mentally assembled the scene, the ring has moved on. The situation is muddied because flash lag isn't unique to motion. One experiment3 found the same effect with color. Imagine a green dot slowly becoming red by passing through all intermediate shades. At a certain point, another dot flashes up next to it, with the same color for that time. Looking at it, you'd see the flash-lag effect as if the changing dot were moving along the color dimension: the flashed dot would appear lagged. That is, the flashed dot would appear greener than the changing dot. That flash lag appears for phenomena other than motion supports that postdiction position. It could be the case that we don't see the world at an instant, but actually as an average over a short period of time. The moving ring appears to be ahead of the flash because, over a very short period, on average it is ahead of the flash. The colored dot appears to be redder than the flashed dot because it is redder, over the averaged period. 2.17.3. In Real LifeThe effect was first noticed with the taillights of a car in the dark (the car being invisible except for the rear lights). A flash of lightning lets you see the car, and the lights appear to be halfway along it: the carwhich is flashedlags behind the taillightswhich are extrapolated. It should also be evident in reverse. If you're photographed from a moving car, the flash of the camera should appear a little behind the car itself. 2.17.4. End Notes
2.17.5. See Also
|
|
EAN: N/A
Pages: 159