Notice that when the rotating bar is in a horizontal position, a flash is presented at each end of the bar. The flashes appear to lag slightly behind the position of the bar; this is called the flash-lag effect. This particular stimulus is not optimal, but most people observe at least a small flash-lag effect. If the bar's motion is not smooth, try devoting more memory to your web browser, using Internet Explorer, or quitting other applications.
So what is interesting about this visual illusion? Consider that the visual system is something like an astronomer. When an astronomer studies stars, galaxies, etc. he/she is actually looking back in time--it takes so long for light to travel from celestial bodies to the earth that the light we see was actually emitted hundreds, thousands, or millions of years ago. The visual system faces a somewhat similar problem that is not due to light itself, but to the neural delays involved in processing visual information. When light hits the eye, we do not perceive that light instantaneously; it takes about one-tenth of a second for the visual system to process an image before we can perceive it. In that tenth of a second, however, objects out in the world could have changed dramatically.
For example, think of a baseball thrown by a pitcher. The image of the ball hits the eye and is processed for one-tenth of a second before the ball can be perceived. Yet, in that tenth of a second, the ball has continued to move, so that by the time we perceive the baseball, it is actually in a different location. For a fastball moving at 100 mph, neural delays of one-tenth of a second would cause the ball to appear as if it lagged about 15 feet behind its actual location. Like the astronomer, we are looking at the history of the moving object--we see where it was, not where it is.
One recent hypothesis has suggested how the visual system may be able to overcome this problem. If an object is moving continuously, then maybe the visual system could predict where the object should be after the neural delays. For example, if the visual system were able to shift or extrapolate the baseball's perceived position 15 feet forward, along the trajectory of motion, then we could perceive the ball where it really is. This kind of extrapolation process could explain why the flash-lag effect (in the demo above) occurs: since the perceived position of the moving object is shifted forward, it is perceived ahead of the flash (which is perceived where it was presented). This may explain how we are able to successfully deal with fast moving objects, such as the baseball.
Another explanation for the flash-lag effect has suggested that neural delays are different for moving and flashed stimuli (like the bar and flash in the demo above). First, consider the situation where the moving bar and flash have identical neural delays (it takes the same amount of time to perceive both the flash and bar). In this case, the bar and flash should appear as if they are aligned. What if the neural delays were reduced just for the moving bar, but not for the flash? In this case, the bar is perceived first, and by the time the flash is perceived, the bar looks as if it's ahead of the flash (which is exactly what people perceive in the flash-lag effect above--the bar appears to lead the flash). So, reduced neural delays for moving stimuli could produce the flash-lag effect. This could be very advantageous, since, as shown in the baseball example above, long neural delays cause us to perceive moving objects inaccurately; reducing these delays may therefore improve our ability to interact with moving objects.
Clearly we are able to deal with fast moving objects, suggesting that the brain must somehow overcome or compensate for its physical limitations. There is still a great deal of debate over whether this compensation takes the form of reduced neural delays, an extrapolation process, motor predictions, or something else.