Moving robots with the mind

In November 2000, at Duke University, an owl monkey named Belle moved a robot arm 800 miles away in a lab at MIT with her mind.

Despite the sci-fi results, the setup used in the study seems crude two years later. The monkey wore a cap that put 100 tiny wires in direct contact with motor neurons in her brain. The cap was connected to a black box that recorded the electrical signals from each neuron. The black box was cabled to a computer running software to translate them into machine commands. The computer was attached to a robot arm. Only the ethernet connection from the Duke computer to the MIT computer was wireless.

Miguel Nicolelis, professor of neurobiology and biomedical engineering at Duke, describes the now-famous experiment, the research that led up to it, and the rapid advances since then in an article in Scientific American.

"The most important message is that basic science in brain research is capable these days to make advances in the clinical arena," he says. What began with recording the activity of single neurons in animals' brains is now leading to the development of brain implants that will wirelessly control robotic prostheses for paralyzed patients. Nicolelis hopes patients will be using such prostheses within a decade.

The primates in today's experiments at Duke have tiny, harmless brain implants that record the activity of hundreds of neurons for months to years. The electronic apparatus, computer, and software are now contained in a wireless neurochip implant the size of a pinky fingernail. Scientists at Duke are now shrinking both the electrodes and the neurochips so much smaller that we would need a microscope to show them to you instead of a TV camera.

Numbers of neurons

When Belle moved the robot arm with her mind in 2000, "only 50 to 100 neurons randomly sampled from tens of millions were doing the needed work," wrote Nicolelis.

Experiments with rats in collaboration with John Chapin's lab at SUNY Health Science Center in Brooklyn are zeroing in on the number of neurons needed for much more precise control of a robot.

"Right now, we can monitor the activity of about 300 brain cells," says Nicolelis. "We're very close to get to 500, and that's a very important landmark."

Fortunately, that number seems to be optimal for useful clinical applications. If the number were in the millions, the electrodes would be impossibly invasive. On the other hand, says Nicolelis, recording only tens of neurons from each motor area wouldn't keep up with the brain's plasticity over time. He’s betting that the optimal number of neurons will turn out to be between 500 and 1000.

A vision of the future

The Scientific American article features a sidebar with a graphic showing how implanted neurochips might one day be designed to move a patient's muscles. Nicolelis says that is possible in the future, but not imminent.

And while mind control of robots is imminent, there are major hurdles. Nicolelis says the devices will need to be as safe and effective as today's heart pacemakers. "You would have to demonstrate very clearly without any doubt that this technology would enhance the quality of life of your patients, to a point that it pays to suffer whatever surgical or other procedures that may be required to create a brain-machine interface." That will take many more animal experiments as well as human clinical trials.

Will stem cells and other research into spinal cord regeneration make those efforts obsolete? Nicolelis hopes they will one day. But he believes that before that, neuroprostheses will have benefited millions of patients.

Nicolelis and other Duke scientists are also developing brain pacemakers to control seizures in patients with severe epilepsy, as well as neurochips that will give neurosurgeons a better view of how brain tissues are responding during surgical procedures.

"If this technology and this concept proves to be feasible, there will be a series of new developments, some of which we cannot even imagine right now," says Nicolelis, "because we will be gaining a new entry, a new view of how the brain produces behavior."

Nicolelis' research group is funded by The Defense Advanced Research Projects Agency (DARPA) Brain Machine Interfaces Program, The National Science Foundation, the National Institutes of Health and Duke University.