In a research laboratory at St. Elizabeths Hospital, brain tissue is growing inside a one-inch-square microchip. This curious marriage of advanced electronics and living nerve cells was dreamed up by a psychiatrist, Dr. Richard Jed Wyatt, chief of the National Institute of Mental Health's neuropsychiatry branch. Wyatt hopes microchips made to interact with living cells someday may be inserted into damaged human brains to connect with remaining undamaged nerve cells and take over functions destroyed by injury or disease.

Implantable computer chips are a long way from reality, but Wyatt and other brain scientists are struggling to develop a variety of simpler brain implants that could be used within a decade to help brain-damaged people regain at least some of their lost ability to walk, see, hear or remember. Some of these implants would be clumps of cells that produce missing brain chemicals. Others would be tiny electronic devices.

Though brain implants are still in the experimental stage, "the whole field is now an exploding area," Wyatt says. "It has caught people's imagination -- not only because there will be some clinical applications, but also because it offers an opportunity to learn a great deal about how the brain works and about immunology."

Patients who suffer from advanced Parkinson's disease and can no longer control their movements may be among the first beneficiaries of the new techniques.

Parkinson's disease -- a degenerative illness of old age often characterized by uncontrollable shaking, rigidity and, eventually, paralysis -- develops when the cells in a small part of the brain, the substantia nigra, degenerate and can no longer produce an important chemical called dopamine, which converts thoughts into voluntary movement.

Physicians currently treat Parkinson's victims by restoring the missing chemical in the form of L-dopa. Dopamine cannot be given directly because it does not cross the blood-brain barrier, a chemical barrier that prevents foreign molecules from getting into the brain from the blood. L-Dopa, however, can cross the blood-brain barrier and once in the brain is converted into dopamine. The treatment is not always effective and can cause severe side effects, including shaking fits, especially as the disease progresses.

In an attempt to prevent the side effects caused by L-dopa treatments, Wyatt's group, about six years ago, began to implant dopamine-producing cells into the brains of rats and monkeys. The brain is one of the few "privileged sites" in the body where foreign cells or tissue are seldom rejected by the immune system. The researchers wanted to find out whether the transplanted cells would survive and whether they would go on making dopamine.

To mimic Parkinson's disease in rats, Dr. William J. Freed, a neuroscientist at St. Elizabeths, chemically damaged the dopamine-producing cells in the substantia nigra on just one side of their brains, leaving the other side intact. With one side of the brain producing no dopamine, one side of the body became rigid as control of voluntary movement was lost. This made the rats walk unnaturally, in circles.

Then the researchers tried to undo the damage by implanting fresh substantia nigras, taken from the brains of rat embryos, into the damaged side of the brain.

Each brain implant was done through a small hole drilled in the rat's skull. The implants were placed in the proper position with a squirt of a syringe.

"We have been moderately successful with these embryonic brain grafts," says Wyatt. "We've been able to correct the Parkinson-like behavior in a proportion of the rats. After several months, we found that the rotational behavior had stopped in most cases. And when we looked at their brains under the microscope, we saw that the grafts had grown and thrived. Their cells seemed to be producing enough dopamine to make up for the original loss."

In looking ahead to the treatment of Parkinson's disease, the team realized that there would be ethical problems arising from the use of human embryos to provide the dopamine-secreting cells for implantation. To solve that problem, the team began looking for a different source of tissue.

"We began to think about where dopamine was present in the body," Wyatt says. Could dopamine-making tissue be transplanted from some other part of the body to the brain?

They soon focused on the adrenal glands, of which there are two -- one above each kidney. One adrenal gland can be removed from the body -- rat or human -- without much damage, as long as the other gland remained in good condition.

The researchers learned that when they took cells from a particular portion of this gland and implanted them into the same animal's brain -- which had been damaged on one side to simulate Parkinson's -- the transplant would correct the rats' circling behavior in at least half the cases.

The adrenal gland cells, when removed from the kidney and placed in the environment of the brain, were able to survive and produce low concentrations of dopamine that were sufficient to restore normal movement.

The next step was to try such implants in monkeys, which are closer to humans on the evolutionary ladder. This time the team did not try to produce or correct any strange behavior. "We were just looking for the survival of cells when one adrenal gland was transplanted to the brain," Wyatt says. "Unfortunately, we've had a relatively poor survival of cells -- far less than in the rat."

One reason for this poor survival may be that monkeys' brains are 100 times larger than rats' brains and may require implants at several different sites. Nevertheless, "we are gradually getting increased survival, and by now about 600 cells out of 1,000 survive the implant," says Wyatt, whose group is continuing the monkey experiments.

Meanwhile, in Sweden, scientists at Stockholm's Karolinska Institute took an adrenal gland out of a patient who suffered from a severe case of Parkinson's disease and implanted cells from this gland in the patient's brain in 1982. They tried the same thing with another patient in 1983. While this did not appear to harm the patients, it did not relieve their symptoms, either. The Swedish researchers plan two more attempts, with minor changes in the position of the implants, early this year.

Wyatt believes that much more work needs to be done on monkeys before beginning such experiments with humans. But he thinks brain tissue transplants may well be used to cure Parkinson's disease within a decade or two.

The many scientists who are now working on brain tissue transplants are conscious of the need to find a good, steady source of donor tissue, especially in cases where using "autografts" of the patient's own cells is not practical.

"From work on rodents, we know that fetal tissue would be optimal, but there are practical, ethical and legal problems involved," says Dr. Donald M. Gash of the University of Rochester Center for Brain Research. "So we have been looking at several alternate sources. I have the highest hope for cell lines that are grown in culture. The advantage is that you can grow as many cells as you want."

Gash has been growing cells taken from a cancer of the human nervous system and implanting them, after treatment, into the brains of African green monkeys. "We treat the cells with anticancer drugs, which cause them to mature and stop dividing," says Gash. Once the cells have matured, they produce an important brain chemical, acetylcholine, which transmits nerve impulses that are essential to memory.

Alzheimer's disease, which causes a loss of memory, is known to involve the deterioration of brain cells that produce acetylcholine, and therefore might be treated with implants of acetylcholine-producing cells. In recent experiments, Gash's group removed some memory-related tissue from the brains of four monkeys and then implanted the treated human cells in the monkeys' brains.

"The cells survived and made some connections," Gash says, but he is not yet sure how effective these connections are or whether the animals' ability to learn and remember has improved.

Other attempts to rejuvenate middle-aged or aged animals through implants of brain cells have had mixed results.

At Clark University in Worcester, Mass., Dr. Donald Stein has found that rats whose frontal cortex was removed, destroying their ability to learn their way in mazes, regained that ability when he implanted brain tissue taken from the frontal cortex of rat fetuses.

But this worked only on "young adult" rats -- not on aging ones. Noting that old people are the most likely to suffer brain injuries from strokes, tumors or disease, Stein is now investigating whether injections of growth factors -- chemicals that makes cells grow -- might help his implants "take" in older rats' brains.

Swedish researchers at the University of Lund have worked with normal, aged rats that could no longer walk along narrow rods without falling off. Two weeks after these rats were given brain implants of cells taken from rat embryos' brains, the researchers reported, the old rats had regained their ability to walk along the rods with ease.

Some engineering-minded scientists are taking a radically different approach as they try to develop brain implants that contain no living cells -- only electronics.

Most of these devices are designed for people who have lost their hearing or sight. The person would wear a microphone or a camera that picks up sounds or light patterns. The signals from these instruments are then translated into electrical impulses that can be transmitted to specific brain cells via implanted electrodes.

The Food and Drug Administration approved the cochlear implant, a type of artificial ear, late last year. The cochlear implant is actually not a true brain implant since it is inserted into auditory nerve in the skull deep within the ear.

The single-channel devices currently approved use an electrode to stimulate the auditory nerve just a few millimeters away from where the nerve goes into the brain; this gives totally deaf persons a chance to hear the difference between sound and no sound, or to distinguish between a telephone ringing and a knock on the door through the difference in their rhythms.

More advanced cochlear implants, with eight sound channels, are now being tried at the University of California at San Francisco. They are about the size of a quarter and receive signals from a transmitter that the patient carries on a belt. Two of the patients with these implants have succeeded in distinguishing some sounds through the difference in their frequency. Eventually such implants may have more sound channels and produce better sound discrimination.

"We estimate that about 200,000 persons would be candidates for cochlear implants," says Dr. Terry Hambrecht of the National Institute of Neurological and Communicative Disorders and Stroke, which supports the tests. These persons are totally deaf because of damage to their middle ear, but their auditory nerve -- which connects to the brain -- is still functioning.

The next step will be to put hair-thin electrodes directly into the brains of people whose auditory nerve has been damaged, bypassing the nerve. This is now being tried out in cats.

Any electrodes that are inserted into the brain would have to be extremely flexible and lightweight, as well as thin, so that they can float in the brain together with brain tissue, Hambrecht says. More rigid electrodes would cut through the soft brain tissue whenever the person's head moved suddenly. Researchers are now developing such electrodes made of pure iridium, a metal that does not react with other substances.

"Probably in two or three years we'll start to do feasibility studies of these electrodes in human beings," Hambrecht says. Once such electrodes are known to be safe and available, other researchers will want to try out some "visual prostheses" that are now on the drawing board -- brain implants that may give blind people some form of sight, he says.

Electronic implants designed for the deaf or blind would provide only one-way communication with brain cells as they transmit signals from microphones or cameras outside the brain to specific cells inside the brain. Wyatt's group at St. Elizabeths has a far more ambitious goal: developing an implant that could carry on two-way communication with brain cells.

Wyatt would like this implant to receive signals from cells in other parts of the brain, integrate them, and send out signals of its own to neighboring cells, much as normal brain cells do. Then it could replace brain cells that are dead or not functioning.

A veteran brain researcher of 45, Wyatt has spent many years thinking about how brain tissue might "interface" with an implanted microchip to establish two-way communication. Two years ago he began to collaborate with Dr. Martin Peckerar of the Office of Naval Research, who produced a new type of microchip that could play host to brain cells.

Wyatt expects that when such a chip is implanted in the brain, undamaged brain tissue would attach itself to the chip and begin the communication process. Eventually a tiny computer, preprogrammed to do what normal brain cells do, would be built on to the microchip, Wyatt says. But scientists do not yet fully understand how brain cells respond to experience, so the software for this computer is still a long way off.

The one-inch square microchip on which such dreams rest looks quite unprepossessing. It has a small sunken area in its center in which the researchers have planted several hundred nerve cells taken from the cerebellum of a rat.

Nourished by a drop of fluid in the well, the cells go on living and producing the small electrical discharges through which they once communicated with other cells in the rat's brain. Each discharge now activates one of 30 hair-like gold wire electrodes embedded in the microchip.

These electrodes are the link with an adjoining computer, the precursor of the miniaturized computer that may eventually be implanted. If Wyatt is right, someday a built-in computer may integrate the many signals it receives from such electrodes in the brain and formulate its own signals to other cells.

Suppose somebody had a stroke or lost part of his cerebral cortex through disease, or suppose some brain function had never developed normally because of a genetic error. Then a preprogrammed microchip might take over the missing function, Wyatt suggests. It could also be designed to learn from its experience, he says. In effect, it would be a piece of artificial brain.

"It's a futuristic idea," Wyatt admits, "but it's probably something that will happen -- perhaps 30 or 50 years from now . . . With all that's going on now in brain research, I see good reason to be optimistic."