08/30/2001 07:00:00

New research shows that brain is involvedin visual afterimages

If you stare at a bright red disk for a time and then glance away, you'll soon see a green disk of the same size appear and then disappear. The perceived disk is known as an afterimage, and has long been thought to be an effect of the "bleaching" of photochemical pigments or adaptation of neurons in the retina and merely a part of the ocular machinery that makes vision possible.

But a novel new experimental procedure by psychophysicists shows that the brain and its adaptive change are involved in the formation of afterimages.

Reporting in the August 31 issue of the journal Science, a joint team from the California Institute of Technology and NTT Communication Science Laboratories, led by Caltech professor Shinsuke Shimojo, demonstrates that adaptation to a specific visual pattern which induces perception of "color filling-in" later leads to a negative afterimage of the filled-in surface. The research further demonstrates that this global type of afterimage requires adaptation not at the retinal, but rather at the cortical, level of visual neural representation.

The Shimojo team employed a specific type of image (see image A below) in which a red semi-transparent square is perceived on top of the four white disks. Only the wedge parts of the disks are colored, and there is no local stimulus or indication of redness in the central portion of the display, yet the color filling-in mechanism operates to give an impression of filled-in red surface.

If an observer were staring only at the red square for at least 30 seconds, then he or she would see a reverse-color green square for a few seconds after refixating on a blank screen (as in the image at the top of C).

However, an observer who fixates on the image at left (in A) and then refixates on a blank screen will usually see four black disks such as the ones at the bottom of C, followed by a global afterimage in which a green square appears to be solid.

The fact that no light from the center of the original square was red during adaptation demonstrates that the effect was not merely caused by a leaking-over or fuzziness of neural adaptation, because the four white disks are at first clearly distinct as black afterimages. Thus, the global afterimage is distinct from a conventional afterimage.

One possibility is that local afterimages of the disks and wedges—but only these—are induced first, and then the color filling-in occurs to give an impression of the global square, just as in the case of red filling-in during adaptation. The researchers considered this element-adaptation hypothesis, but eventually turned it down.

The other hypothesis is that, since neural circuits employing cortical neurons are known to cause the filling-in of the center of the red square, then perhaps it is this cortical circuit that undergoes adaptation to directly create the global negative green afterimage. This is called the surface-adaptation hypothesis, which was eventually supported by their results.

The researchers came up with experiments to provide three lines of evidence to reject the first and support the second hypothesis. First, the local and the global afterimages were visible with different timing, and tended to be exclusive of each other. This argued against the first hypothesis that the local afterimages are necessary to see the global afterimage.

Second, when the strength of color filling-in during adaptation was manipulated by changing the timing of the presentation of disks and colored wedges, the strength of the global afterimage was positively correlated with it, as predicted by the surface-adaptation hypothesis but not by the element adaptation hypothesis.

For the last piece of evidence, the researchers prepared a dynamic adapting stimulus designed specifically to minimize the local afterimages, yet to maximize the impression of color filling-in during adaptation. If the element-adaptation hypothesis is correct, then test subjects would not observe the global afterimage. If, on the other hand, the surface-adaptation hypothesis is correct, the observers would see a vivid global afterimage only. The result turned out to be the latter.

The study has no immediate applications, but furthers the understanding of perception and the human brain, says Shimojo, a professor of computation and neural systems at Caltech and lead author of the study.

"This has profound implications with regard to how brain activity is responsible for our conscious perception," he says.

According to Shimojo, the brain is the ultimate organ for humans to adapt to the environment, so it would make more sense if the brain, as well as the retina, can modify their activity—and perception as a result—due to experience and adaptation.

The other authors of the paper are Yukiyasu Kamitani, a Caltech graduate student in computation and neural systems, and Shin'ya Nishida of the NTT Communication Science Laboratories in Atsugi, Kanagawa, Japan.

Contact: Robert Tindol (626) 395-3631