In Breakthrough, Monkey Think, Computer Do

An experimental brain implant the size of an M&M; has allowed a monkey to control a computer cursor by thought alone, Brown University researchers announced Wednesday.

It is the latest advance by scientists trying to perfect a link between mind and machine in the hope that thousands of patients who are unable to move or speak can resume communication with the world around them.

The development heralds a future when the paralyzed and infirm may send e-mail, surf the Web and command other computer resources simply by thinking about them.

The new device, described in research published today in Nature, uses a special mathematical formula to translate signals from a few motor neurons on the surface of the monkey’s brain via cable into movement on a computer screen. There is no need for the extensive training previous experimental techniques have required.

“We substituted thought control for hand control,” said Brown neuroscientist John Donoghue, the project’s senior researcher.

Lead researcher Dr. Mijail D. Serruya and his colleagues at Brown tested the device by having a monkey play a simple video game, in which the animal used the cursor to chase a moving target on a computer screen.

The monkey was able to move the cursor “instantly” with almost as much control as if it were using a computer mouse or a joystick, Serruya said. The monkey wills the cursor to move. The cursor moves.

The animal’s hands-free cursor control was almost as fast and accurate as when it used its hands, the researchers reported. So far, three monkeys have received the implant.

Linked to a personal computer, the cursor control device “would work for anything you can do or you can imagine doing by pointing and clicking. This includes reading e-mail,” Donoghue said. “Or imagine an on-screen keyboard that someone can use to type sentences or issue commands by pointing and clicking.”

The Brown implant system uses an array of 100 tiny electrodes to detect electrical activity from a pinprick of neurons--between seven and 30 of the untold thousands of nerve cells in the motor cortex, the section of the brain that controls movement--and relay it to a personal computer.

There, a formula the Brown group designed turns the brain activity into instructions a computer can use to plot the cursor movements. That software interpreter uses a fairly straightforward application of textbook algebra, the scientists said.

The system is so small and draws so little power that any future device developed for human use could easily be made wireless, said biomedical engineer Sandro Mussa-Ivaldi at the Rehabilitation Institute of Chicago.

“In clinical terms, that is quite important,” he said.

It may be a decade or more before any clinical product is ready for testing, however. So little is known about tapping the neural activity of the human brain that researchers are likely to proceed slowly. No one knows, for example, what the effect would be of having such an electrode in the brain for years or whether over time the implant might lose its ability to function.

“This implant is potentially one that is very suitable for humans,” Serruya said. “It shows enough promise that we think it could ultimately be hooked up via a computer to a paralyzed patient.

“We want to be careful that the implant is suitable and safe,” Serruya said. “There are a few technical details that we are still working on.”

The Brown researchers have filed for a patent on the technique and formed a company called Cyberkinetics to develop medical applications.

Other research groups in recent years have announced encouraging successes in experiments with devices designed to turn brain commands into computer instructions. Until now, these were limited in their potential usefulness by technical constraints.

In 1998, neuroscientists at the University of Tuebingen in Germany tested an experimental cursor control device in human volunteers, as did researchers at Emory University in Atlanta.

Those devices allowed several patients paralyzed by advanced strokes or amyotrophic lateral sclerosis to spell words on a computer screen by modulating their brain waves.

Both implant systems, however, required intensive practice to master--up to 150 training sessions over more than a year. The implanted electrodes themselves were bulky.

In related work, researchers led by neurobiologist Miguel Nicolesis at Duke University successfully used the motor control signals from monkeys to move a robot arm. During an experiment last year, Nicolesis linked the monkey to the Internet and used its brain signals to control a robot arm 600 miles away.

To mimic arm movements, however, that system requires extensive computer analysis after the scientists have recorded the brain signals during many repetitions.

What makes the newest effort to link brain and machine remarkable is the ease with which it can be learned and the small number of neurons it requires to operate, Mussa-Ivaldi said.

“There are a mountain of things we still need to know,” he said. “But this is progress.”