Click to viewTwiddling our thumbs while waiting for files to download is one of the great frustrations of our high-speed, technologically interconnected world. Computer scientists are unlikely to eliminate the waiting any time soon, but they've done the next best thing.

Chris Harrison at Carnegie Mellon University in Pittsburgh, Pennsylvania, working with Zhiquan Yeo and Scott Hudson, has shown that animated pop-up download progress bars which use visual illusions make the process seem around 10 per cent faster than it really is.

Apple already uses a a basic visual trick of this kind in its Mac OS X operating system, but Harrison's research suggests such techniques could be used to greater effect.


Previous research suggests that rhythmic stimuli can seem to create time-warping effects (PDF) and that the way we perceive motion depends on its context (PDF). With this in mind, Harrison's group generated a series of animated progress bars: some pulsated between pale and dark blue at varying speeds, and others had pale blue ripples moving either left or right, also at different rates, as the bar crept forward (see video).

They then showed different pairs of either pulsing or rippling animations to 20 volunteers, with each mocked-up download lasting exactly 5 seconds. Many participants said that progress bars which pulsated increasingly quickly made the download time seem shorter than those that pulsated increasingly slowly.


A significant number also said downloads were faster when ripples in the progress bars moved to the left rather than to the right.

Apple uses left-moving ripples that travel at a constant velocity for progress bars in Mac OS X. However, many of the volunteers in Harrison's study said that this type of animation appeared slower than one in which the left-moving ripples slowed down as the download neared completion.

In a second part of the study the researchers sought to quantify the time-bending effect of the visual illusions by asking the volunteers to compare a progress bar with slowing ripples with a bar that neither pulsated nor rippled. During these tests the time it took the rippling progress bar to complete the mock download was gradually lengthened, while the download time for the standard bar was left the same, until the volunteers felt that both had completed the operation in the same length of time.

They found that, on average, a 5.61-second rippling progress bar appeared to take the same amount of time as a 5-second standard bar, and a 16.75-second-long rippling bar seemed to take as long as a 15-second-long standard bar – illusory speeding-up effects of 10.9 per cent and 10.4 per cent respectively.

That could matter to computer users, says Harrison. "A good number of progress bars are between 5 and 15 seconds," he says. "For longer operations, you might go on to other tasks and not look at the progress bar."

"It is not uncommon for [such] illusions to have a measured magnitude of around 10 per cent," says George Mather, a psychologist at the University of Sussex, UK.

Harrison and colleagues will present their work at the Association for Computing Machinery 2010 Conference on Human Factors in Computing Systems in Atlanta, Georgia, next week.