In less than 60 days, Brazil will begin hosting soccer’s 2014 World Cup, even though workers are still hurrying to pour concrete at three unfinished stadiums. At a laboratory in São Paulo, a Duke University neuroscientist is in his own race with the World Cup clock. He is rushing to finish work on a mind-controlled exoskeleton that he says a paralyzed Brazilian volunteer will don, navigate across a soccer pitch using his or her thoughts, and use to make the ceremonial opening kick of the tournament on June 12.
The project, called Walk Again, is led by Miguel Nicolelis, a 53-year-old native of Brazil and one of the biggest names in neuroscience. If it goes as planned, the kick will be a highly public display of research into brain-machine interfaces, a technology that aims to help paralyzed people control machines with their thoughts and restore their ability to get around.
“It’s going to be like putting a man on the moon—it’s conquering a level of audacity and innovation that the people outside Brazil aren’t used to associating with Brazil,” Nicolelis has told audiences. The kick, he has said, “will inaugurate a new era of neuroscience, [that of] neuroengineering.”
But the Walk Again project is drawing doubters. Saying the demonstration is as much publicity stunt as science, they question whether it will illustrate any real degree of thought control. That’s because it relies on a fairly old, imprecise brain-recording technology called EEG, or electroencephalography.
At least three other research groups have recently published reports of EEG-controlled exoskeletons. Yet so far, none have managed to do much more than send a start or stop signal. They let the robotic harness do the rest of the work on a preset trajectory, with plenty of outside assistance in balancing.
That suggests that the level of brain control could be disappointingly minimal, even if it’s presented as a breakthrough on TV. “What would happen if a gust of wind moved the ball by three centimeters right before the demo began?” asks Andrew Schwartz, a neuroprosthetics researcher at the University of Pittsburgh. “Everything you’ll see in the demo will be fancy robotics, not brain control, and it will probably all be preprogrammed.”
The demo is expected to take place ahead of the opening match between Brazil and Croatia. To pull it off, the Walk Again team says it intends to combine EEG signals with other (as yet undisclosed) strategies for giving the patient dynamic control over walking and kicking the ball.
“We are well aware of the limitations of EEG, but we decided to show what could be done,” says Alan Rudolph, vice president for research at Colorado State University and manager of the project. “People will see a control system that is new in terms of using and exploiting brain signals as well as body signals.”
Brazil’s federal innovation agency gave Nicolelis $15 million to carry out the project, a large sum by the standards of any country. The ceremonial kick will be seen by 70,000 people in São Paulo’s Itaquera stadium, and according to Brazil’s government, “billions of TV viewers will be able to follow what could become one of the biggest achievements in Brazilian and world science.” (In actuality, World Cup audiences seldom exceed 250 million people, and a far smaller number tune in to the opening ceremonies.)
Nicolelis has a well-developed knack for mixing science with show business and, in Brazil, with politics. He did not respond to a request for an interview but has been posting updates to a fan page on Facebook. “It’s a lot of flash, and that rubs some people the wrong way,” says Daniel Ferris, who studies EEG and the kinetics of walking at the University of Michigan. “I think Nicolelis is a very good scientist, and it doesn’t bother me how he gets in the media.”
Until now, Nicolelis has been an ardent defender of an entirely different approach to brain-machine interfaces: directly recording neurons inside the brain, using implantable chips.
In 2008, he transmitted signals from a monkey’s brain over the Internet to Kyoto, causing a bipedal robot to ambulate in Japan (see “The Power of Thought”). That experiment was the origin of the Walk Again project, which now involves about 125 people from five continents, says Rudolph.
Nicolelis initially hoped to equip a human volunteer with a tiny implanted recording cube studded with 1,000 or more electrodes, each capable of directly recording an individual neuron inside the brain. By recording from many neurons at once, his research in monkeys had shown, it would be possible to capture fine details of intended arm and leg movements and reproduce them in a robot.
Writing in Scientific American in 2012, Nicolelis said that to control the robotic legs of an exoskeleton, it would be necessary to “implant electrodes into the brain to manipulate the robot.” He also firmly rejected the idea of using less-accurate EEG, which relies on recording from outside the skull, saying in 2011 that “low spatial resolution makes using EEG in interfaces like those created in our laboratory inviable.” In a book he published that same year, Nicolelis went further, describing “an almost insurmountable chasm” separating his intellectual goals from those of EEG researchers.
In the end, it was the pressures of the World Cup deadline and the difficultly of organizing a brain implant that proved insurmountable, says Rudolph. That motivated a second-half substitution of the more accessible EEG technology. Brazil’s government agreed to the demonstration only in January 2013, Nicolelis has said, leaving little time to prepare.
“We decided to show what could be done now [and] portend what is coming,” says Rudolph, adding that the Walk Again project still intends to adopt intracortical recordings at a future date. “It’s an aggressive project moving very quickly. Any time you have to demonstrate a new technology at a specific time and date, that is a challenge.”
Even so, the choice puts Nicolelis’s demonstration at the center of a debate over what EEG signals can actually do. Some, like Schwartz, say it’s a stalled technology that “hit a hard wall 20 years ago.” Others say it’s a practical avenue for brain control whose promise is still unfolding.
The clear advantage is that recording an EEG signal doesn’t require brain surgery, just an electrode-studded cap. The problem is that the electrical signal is collected outside the skull, the combined product of billions of neurons firing at once. That’s a bit like listening to an orchestra from a noisy avenue outside a concert hall. The notes of individual instruments are lost.
But if enough players raise their volume or tempo, that change can be detected—and used as a simple control signal.
Last year, biomedical engineer Bin He of the University of Minnesota published a video of a person controlling the movement of a toy helicopter. The subject did so by imagining clenching one fist or the other to move it left or right, or sweeping both hands upward to give it lift. With training, computer algorithms recognized the resulting changes to the EEG wave and used them to control the copter.
But there are practical limits. Generating and decoding willful changes to the EEG wave is slow and produces very small amounts of useful information—often just five to 50 bits per minute. It’s not nearly enough data to control anything as complex as the Walk Again exoskeleton, which Rudolph says has 17 degrees of freedom.
Researchers have hoped to get more information from the EEG signal but have been so far been disappointed, says Guy Chéron, a researcher at the Université Libre de Bruxelles in Belgium, who led a European initiative that spent $7 million developing an EEG-controlled exoskeleton and wrapped up last year.
Chéron’s group did get the exoskeleton to move forward on a command from a brain wave, evoked by a person thinking about walking. But Chéron considers the result something of a flop. “We are not convinced it’s linked to real neuronal signals controlling locomotion, and not an artifact,” he says. “I think we are far from this. It’s my dream, but it remains a dream. And Nicolelis knows very well what the problem is.”
In Brazil, Nicolelis has been facing down similar criticisms, leveled by former colleagues and by the Brazilian media. Last month, he told the Brazilian newspaper Folha de São Paulo that the switch to EEG “doesn’t diminish the scientific or clinical relevance of our initiative.”
In part, that’s because mind control isn’t the only technology involved. The World Cup demonstration is expected to pull together several technologies that are now pointing to new possibilities for paralyzed patients, particularly recent advances in exoskeletons.
Because of fast developments in robotics, it’s unclear what role brain-derived signals will ultimately play. Experts note that there are much easier ways than EEG to give a simple command—for example, pushing a button or using one’s voice. “Twenty years from now, there is going to be a whole sensor suite,” says Ferris of the University of Michigan. “Maybe some EEG, maybe muscle recordings, then inertial sensors, and things you control by leaning into them like a Segway. You should be able to come up with a pretty decent controller for a paraplegic. Coming up with something for a quadriplegic is a lot harder.”
Nicolelis thinks his demo will shed light on all these promising developments. But there’s not much time. Tests with patients started in February, according to the team’s Facebook page. Two exoskeletons arrived by plane in Brazil only in March. Nicolelis, posting a video of the exoskeleton taking its first steps on April 7, chose to add the soundtrack from Mission Impossible.
Whether or not the demo proceeds, and whatever its technical importance, Nicolelis may have already achieved his desired legacy for Brazil. “Believe it not … in Brazil, people are talking about neuroscience when they talk about World Cup,” he told an audience in Chicago this February. “That is already enough of an accomplishment. To have kids stop and talk about science instead of football.”
The big new idea for making self-driving cars that can go anywhere
The mainstream approach to driverless cars is slow and difficult. These startups think going all-in on AI will get there faster.
Inside Charm Industrial’s big bet on corn stalks for carbon removal
The startup used plant matter and bio-oil to sequester thousands of tons of carbon. The question now is how reliable, scalable, and economical this approach will prove.
The hype around DeepMind’s new AI model misses what’s actually cool about it
Some worry that the chatter about these tools is doing the whole field a disservice.
The dark secret behind those cute AI-generated animal images
Google Brain has revealed its own image-making AI, called Imagen. But don't expect to see anything that isn't wholesome.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.