Share this article on:

Monkeys Move Prosthetic Arm with Help of Brain-Machine Interface: Technical Challenges Remain for Human Use


doi: 10.1097/01.NT.0000333566.96608.40

Electrodes planted in a monkey's motor cortex have enabled the animal to feed itself by using its own neural impulses to operate a prosthetic arm.

However, the electrodes that pick up those impulses cause enough damage to provoke the brain into attacking them vigorously, which means the practical application of a brain-machine interface (BMI) in humans remains a few years away.

“During the first three months the signals are pretty good, but they degrade over time,” said Andrew Schwartz, PhD, professor of neuroscience at the University of Pittsburgh, who led the experiment, which appeared as an advance publication online before print in Nature on May 28. “We're not positive what's going on. The party line is that there's encapsulation — the brain starts to form scar tissue around recording sites, and that's what causes signals to degrade.”



The brain also attacks the electrodes used for deep brain stimulation (DBS), but the devices often continue to work for years because they are sending electrical current to the brain rather than trying to record current coming from the brain. Current delivered to the brain can push through scar tissue or other debris covering the electrodes.

When detecting signals coming from the brain, however, even a slight impediment can make the signal unusable.

“We're talking about microvolts,” said Dr. Schwartz, “so any extra resistance really cuts down on what we're able to see.”

Back to Top | Article Outline


For this experiment, Schwartz implanted in two monkeys a unit known as the Utah array, developed by Richard A. Normann, PhD, professor of bioengineering at the University of Utah, and now manufactured by Cyberkinetics Neurotechnology Inc. (This neuroprosthetic device has received FDA approval for pilot testing on humans.)

The Utah array consists of a grid of 100 insulated silicon needles, each less than the diameter of a human hair, which project from a tiny silicon wafer smaller than the head of a thumbtack. After opening the skull and folding back a flap of the dura, Dr. Schwartz's team inserted the electrodes protruding from the wafer into the arm region of the monkey's motor cortex. Then the unit was covered with the dura and the skull replaced, leaving a small hole for the 100 lead wires to emerge.

Dr. Schwartz implanted an array in each hemisphere of two monkeys. Using signals picked up by the electrodes on either side of the brain, the monkeys, whose own arms were restrained, rapidly learned to use their own neural impulses to make the prosthetic arm reach for food held in front of them, grasp it, and move it to their mouth.



Back to Top | Article Outline


Infection is rarely a problem with these implants, according to John Donoghue, PhD, the founder of Cyberkinetics and a veteran BMI researcher who incorporated the Utah array into the BrainGate, an apparatus that has been implanted in humans. “There appears to be no more problem of infection than with any other implant — about 5 percent,” he said. “In our experience infection has not been an issue.”

But for reasons not fully understood, the brain does not tolerate the intrusion of the electrodes, and this presents a major obstacle to using BMIs in paralyzed humans to enable them to operate a motorized wheelchair, a computer, or other devices with their neural impulses alone.

Although the electrodes themselves are thin, they are rigid. “Think of carved icicles,” Dr. Schwartz said.

The 100 leads running from the back of the wafer tether it firmly in place, and Dr. Schwartz suspects that the slight brain pulsation caused by blood flowing through the arteries might be enough movement to allow the electrodes to irritate the brain, causing microdilation of tissue, swelling, and perhaps an immune reaction. In addition, the dura over the unit sometimes hypertrophies and starts to tear apart the leads attached to it.

While the problem with the dura is conspicuous, the problems caused by the electrode's prongs “are below our resolution at this point,” Dr. Schwartz said. “Anytime you have foreign material in the body, the body will try to cap it off.”

The solution, he believes, will be to develop electrodes that are more flexible, because they would move with the brain and cause less irritation.

“Also, we're working on biocompatible coatings to trick the brain so it won't recognize the electrodes as a foreign body.”

Despite daunting challenges, Dr. Schwartz expects to see electrodes that record clear, durable signals implanted into humans within a few years.

“The electrodes are the biggest hurdle, but already we have electrodes that have lasted for years in animals,” he said. “It's not going to take a whole lot more to get them to where we need them to be for humans.”

Dr. Schwartz and his colleagues represents the first demonstration in which BMI technology has been used to perform a practical behavioral act, according to John F. Kalaska, PhD, professor of neuroscience at the University of Montreal, who wrote a commentary in the same issue of Nature.

The BMI in this case produces enough signals to make the prosthetic arm move at the shoulder, elbow, and hand — movements known as degrees of freedom. A prosthesis that mimics a fully functioning human hand needs about 21 or 22 degrees of freedom, according to Dr. Schwartz. The device operated by the monkeys has five — three at the shoulder, one at the elbow, and one in the gripper, or hand. Dr. Schwartz and his team hope to boost this to 21 degrees of freedom by implanting two arrays, one for the arm and another for the hand alone. They also hope to add pads that will deliver tactile signals from the prosthesis to the brain.

The experiment “is elegant science,” with profound implications for humans suffering from high spinal cord injuries, said William Zev Rymer, MD, PhD, of the Rehabilitation Institute of Chicago.

But, he added, such applications will remain a long way away until methods are found to make the brain accept the electrodes.

“We need electrodes that will last as long as the problem will last,” he said. “You're looking at a lifetime of need, and current electrodes seem to be good for six months to a year. The systems are elegant, but they're not designed for long-term practical use, so I don't think we're within striking distance of a practical application. What's needed now is a radical rethinking about how we connect to the brain.”

Back to Top | Article Outline


Monkeys using a brain-machine interface (BMI) — implanted electrodes that protrude into the arm region of the monkey's motor cortex — were able to make their prosthetic arm move at the shoulder, elbow, and hand. But problems with degradation of the electrodes need to be resolved before the BMI can be used safely in humans.



Back to Top | Article Outline


Visit for these articles:

  • “From Prosthetic Limbs to Robots for Gait Retraining, Rehabilitative Devices Shed New Light on CNS Plasticity,” May 2, 2006.
  • “Biomedical Engineering and Neuroscience Come Together to Foster Computer-Based Movement…But Will Brain-Computer Interfaces Work in Humans,” November 2005.
Back to Top | Article Outline


• Velliste M, Perel S, Schwartz AB, et al. Cortical control of a prosthetic arm for self-feeding. Nature 2008. E-pub May 28, 2008.
    • Kalaska JF. Neuroscience: Brain control of a helping hand. Nature 2008. E-pub May 28, 2008.
      ©2008 American Academy of Neurology