How brain-machine interfaces work

First aired on Quirks & Quarks (06/04/13)

Download Flash Player to view this content.

brain-machine-220.jpgImagine being able to operate a computer with your thoughts -- no need for a mouse or a keyboard -- and navigate a virtual world or control a vehicle or a robot avatar. Imagine what that would mean to someone who is paralysed: imagine them controlling a motor arm or a motorized walker with their mind alone.

This might sound like something out of science fiction, but Dr. Miguel Nicolelis is working to make it a reality. He's a leading researcher in the field of brain-machine interfaces. These systems involve connecting living brains to computerized devices so that they can be controlled by thought alone. Nicolelis is a professor of neurobiology and co-director of Duke University's Center for Neuroengineering; he's also the author of Beyond Boundaries: The New Neuroscience of Connecting Brains with Machines -- And How It Will Change Our Lives. In a recent interview on Quirks & Quarks, he described his research, and how brain-machine interfaces work.

"We started all this research way back in the early 1990s, developing a technique that allows us to record the electrical signals produced by neurons simultaneously," Nicolelis told host Bob McDonald. In 1999, they published the first paper on the subject and coined the term "brain-machine interface," by demonstrating that rats could use brain activity to control a simple robot. Since then, they've moved to primates.

Nicolelis went on to describe how the experiments worked with monkeys. Electrodes are implanted in the monkey's brain, and used to record the electrical activity in the cells. "We can decode simple motor programs produced by this electrical brain storm, and transform them into digital commands that can be used by mechanical, electronic or even virtual devices," he said. The monkey can learn to control the movement of these devices using thinking alone. "They can also receive artificial tactile feedback directly into the brain, so they can feel what these devices are touching when they're performing a particular task."

In the latest experiment, monkeys used an avatar hand to explore the surface of virtual objects that looked identical, but had different textures. They learned to distinguish between the different textures and control the virtual hand using the electrical signals of the brain. Nicolelis pointed out that, surprisingly, even though hundreds of thousands of neurons are involved in this kind of motor activity, recording just a small sample of the overall population of neurons supplies enough information "to control artificial devices to reproduce some of these behaviours."

The neurons' signals are recorded through electrodes that are a fraction of a human hair in diameter. Hundreds of these microfilaments are implanted, but because they are so tiny, they don't damage the brain. "What we are producing are really tiny little insertion points, just the track of the electrodes, and that doesn't compromise the overall functioning of the brains we've studied -- and that includes mice, rats, monkeys and more recently, humans."

After the neurons' signals are recorded, they are digitally decoded by computer programs "that play the role of interpreting the messages embedded in the electrical brain storm," Nicolelis said. Because of the feedback loop set up between the brain and the artificial device, the brain learns how to control something in a new way, and the computer learns how to understand the signals Beyond-Boundaries_140.jpgthe brain produces.

The brain's concept of the body it inhabits changes "because it has to incorporate artificial devices that can be controlled directly by the brain," Nicolelis said. "It's almost like the brain is liberating itself from the physical constraints of the body."

The finer the motion, the more complicated it is, and the more neurons that need to be sampled. "The whole game, in my opinion, is about how many neurons you can record simultaneously and from many areas at the same time, because nothing in the brain is happening in just one location," Nicolelis said. He believes that one day it will be possible for someone who is paralysed to control a full body exo-skeleton, "a hollow robot that you actually wear," with thought alone. But it will require recording signals from a huge number of neurons, as many as 50,000 or 100,000.

When asked how close we are to actually developing those kinds of devices, Nicolelis acknowledged that "it's not going to be some magic jump." But he predicted that in the next few years, "we are going to move steadily toward a variety of applications that will make a mark in medicine in the next decade."

Related links: