By closing your eyes and gently pressing against the corner of your eyelid, you can conjure colorful shapes -- steel-blue circles with brilliant yellow edges that follow the movements of your finger. Just rubbing your eyes, as everyone knows, can produce a light show worthy of the psychedelic Sixties.

The sensation of "seeing" these lights is called phosphene, from the Greek words phos for "light" (as in phosphorus) and phainein for "to show" (the root of "fantasy"). But the patterns that seem to dance behind the shutters of your eyes are neither light nor fantasy.

Phosphenes are the brain's interpretation of signals from the optic nerve, a kind of illusion that the conscious brain interprets as something seen, even though no light has entered the eye.

Phosphenes have long intrigued scientists as a phenomenon that might be exploited to give sight to the blind.

By tweaking certain brain cells with a tiny amount of electricity, scientists have found they can trigger phosphenes. They hope to develop a device that can trigger scores of phosphenes simultaneously to create an image, much like a pointillist painter uses thousands of dots of color to create a picture.

Scientists in the National Institutes of Health's Neural Prosthesis Program say they hope to develop a phosphene prosthesis that will someday allow blind people to read. Eventually, they say, it may be even be possible to give the blind enough phosphene vision to identify objects and people.

The idea is not new. It began growing about 30 years ago from the knowledge that light, when it impinges on the retina, is translated into electrical impulses that are sent to a region of the brain in the back of the head. The signals create a map on the surface of the visual cortex that corresponds to the image mapped across the surface of the retina.

Most of the 600,000 Americans who are totally or legally blind retain a healthy visual cortex that lies dormant from lack of input from their eyes.

Efforts years ago to artificially stimulate the visual cortex achieved modest success. By placing electrodes on the surface of the brain researchers were able to trigger phosphenes in blind volunteers. They described "seeing" images that would flicker and hover even after the electricity was turned off. And although the scientists could reliably generate phosphene points, efforts to make them appear in predictable locations failed. Controlling the Phosphenes

Recent work has overcome many of these problems, and the new generation of implant can elicit more and "better" phosphenes.

The major advance has been the development of a thinner electrode that can stimulate small groups of brain cells in the visual cortex. The microelectrode was tested last year in three patients already undergoing brain surgery for epilepsy. "We were very encouraged by the experiments," said F. Terry Hambrecht, head of the NIH's prosthesis program. "The phosphenes these people saw did not flicker. They held steady and disappeared when the electricity stopped." The researchers could control the brightness of the phosphenes by varying the current.

Instead of placing electrodes on the surface, Hambrecht's group inserted the microelectrodes about two millimeters into the brain. This led to phosphenes that would reappear in the same location, even when more than one electrode at a time was fired. The researchers were able to summon crisp, non-flickering phosphenes that could be turned on and off at will.

Surprisingly, there were blue, yellow and red phosphenes, while in earlier experiments electrodes placed at the surface of the brain produced mostly white phosphenes. "We think our microelectrodes may have stimulated columns of brain cells that receive only certain colors from the eyes," Hambrecht said. These groups of brain cells, called "color blobs," are each thought to process information about a single color.

The phosphenes that Hambrecht's patients saw resembled points of light, each the size of "the head of a pin held at arm's length." Approximately 100 such phosphenes in a 10 by 10 grid would be necessary for a functional prosthesis, Hambrecht estimated.

The microelectrodes, made of silicon, are on the cutting edge of technology. They are thinner than human hairs and come with built-in electronic circuitry. Because they are so thin and cannot be enclosed in protective casing -- like a pacemaker is -- the microelectrodes are prone to corrosion in the salty environment of the brain. Researchers are creating new, protective compounds that can be sprayed in a thin layer on the electrodes to combat the corrosion.

A silicon microelectrode array has yet to be tested in human beings. "We don't know whether or not we can stimulate many phosphenes to occur simultaneously," Hambrecht said. Designing a Reading Aid

The first prosthesis would be designed as a reading aid. A camera mounted on a pair of glasses would relay information to a computer chip, which would translate the information into electric currents -- one-thousandth the amount needed to power a pocket radio -- directed into the appropriate electrodes.

If the camera were focused on an "R," the result would be a phosphene map of an "R" that would look like a dot-matrix printout. Other scientists have likened the image to that displayed by a stadium scoreboard screen.

But, as Hambrecht points out, "We don't read one letter at a time, we read groups of letters or whole words at a time." A prosthesis able to present images of entire words would have to stimulate about 2,000 phosphenes.

Reading whole words at a time would be one of many advantages to a 2,000-phosphene prosthesis. Such a prosthesis might even create a view of one's immediate vicinity by which a blind person could navigate. By using a camera with a 2,000-phosphene map, "one could have a wide-angle view of the world, then zoom in on a face and get a pretty good image of it," Hambrecht said. Eventually, he hopes, phosphenes of different colors could be incorporated into a color map.

As promising as a visual prosthesis sounds, many questions must be answered and technological problems surmounted before a working model becomes available. One problem is that brains vary from one person to the next. Chances are, if microelectrodes were inserted into the same location of the brain in two different people, the phosphenes would be slightly different.

Also, the researchers do not know whether the visual cortex can handle sustained stimulation without damage. Inserting a foreign body into the brain is risky, too.

The idea of a visual prosthesis using phosphenes has even spread from the laboratories into television. The fictitious VISOR (visual input sensory optical reflector) worn by the blind Lieutenant LaForge in "Star Trek: The Next Generation," for example, can detect not only light, but a wide range of other radiations such as gamma rays and infrared.

Hambrecht and his colleagues have set their sights on a more modest light-detecting prosthesis. The gamma rays and infrared can wait.