The Brain-Computer Interface That Let a Quadriplegic Woman Move a Cup

Two severely paralyzed people operated robotic arms and prosthetics using thoughts captured by implants in their brains, a new study disclosed.


Inside the brain of a test subject known as S3, a symphony of neurons fired in her motor cortex one day in April last year. Paralyzed by stroke 15 years earlier, this 58 year-old woman with a bright smile and wearing a flashy red shirt imagined that her arm was working again and that it was moving in space. She ordered it to pick up a cup filled with a morning dose of Joe.

Read an exclusive excerpt from The Atavist about subject S3, Cathy Hutchinson.

Near her wheelchair a blue and gray robotic arm swung into action. Swiveling towards patient S3, it reached out its mechanical hand and grabbed a container of coffee with a lid and straw, lifting it up to S3's waiting lips. Arriving in just the right spot she sucked on the straw, a simple act for most people that she had been unable to do on her own for over a decade.

In a video of the maneuver, S3's beaming face registers her delight.

According to the researchers, whose work was published today in Nature, this is the first peer-reviewed study of a person with severe paralysis controlling a robotic or prosthetic arm directly in three-dimensional space using thought.

"We now show that people with longstanding, profound paralysis can move complex real-world machines like robotic arms, and not just virtual devices, like a dot on a computer," said Brown University neuroscientist John Donoghue, one of the lead researchers.

In 2006, Brown and colleagues published a paper in Nature on an earlier version of this device and process - called Braingate - that demonstrated in a man paralyzed from the neck down that it was possible to move a cursor on a computer using thought. The updated version used by S3 is Braingate II.

In the video one can see what looks like a plug attached to the top of S3's head, trailing wires that run to a computer. The plug is attached to a tiny bed of 96 hair-sized electrodes the size of a child's aspirin that was implanted by a surgeon directly on her brain. This bed of sensors is connected to a nickel-sized pedestal tucked into her scalp - which serves as the outlet for Braingate's plug. Wires from the plug were connected to a computer.

"Using a sophisticated algorithm years in the making, the fuzzy noise of the brain was sorted out and the critical signals identified," said Donoghue. "The computer translated these signals into commands for the artificial arm." S3 and a 66 year-old male subject designed as T2 performed a number of grabbing tasks, including the snatching of the coffee.

"We actually demonstrated two types of robots--one that is an assistive technology and another that is made to be a prosthetic limb for amputees," said Donoghue. "They could both position the hand anywhere in 'reach' space and close the hand to grab a ball at will."

The project was a collaboration of neuroscientists, neurologists, and experts in robotics, computers and algorithms.

Braingate was approved by the Food and Drug Administration as an early stage clinical trial for a handful of people, though the ethics of putting a person through surgery, which always has some risk, remains a concern. This is ameliorated some by the successful implantation of tens-of-thousands of deep brain devices that are used to treat the tremors of patients with Parkinson's disease.

Critics of the team's earlier work called the work in humans premature, and were dismissive of what was then a commercial approach through a company called Cyberkenetics - which since has "closed", according to Donoghue. The current effort is academic, he added.

In 2005, I saw the original Braingate in action outside of Boston, where a 25 year-old man paralyzed from the neck down was hooked up to a sizable bank of computers and processors. (Check out my NPR Morning Edition story, "Thinking is Doing With Cyborg Technology" - also written a version in Technology Review). During the demonstration the patient, Mathew Nagle, was able to move a cursor on a computer and play the simple video game "pong". He also was able to open and close the metal hand of a prosthetic arm.

The system, however, took a long time for the technician operating the equipment to calibrate - to match up Nagle's thought-signals with the computer. Nagle seemed frustrated at times that on a computer animation he couldn't catch a small bag of money with the cursor.

"I can't get it today, not even close," he told me, though on other days he said he was more successful. He was able to change channels on a television using the Braingate.

Critics at the time wondered if the experiments on humans were premature. "The movements they're
 getting are crude," said University of Pittsburg neuroscientist Andrew Schwartz, another leader in the field, in 2005. "It's 
not clear how good the human recordings of the neural signals are. To be useful, it will have to be much better, to do more things," he said.

Donoghue says that the latest version of Braingate is greatly improved over the original.

He and colleagues at Brown are working to eliminate the wires and to create a wireless system. They are conducting work on monkeys, he said, but still need FDA approval for human testing.

The work is still years away from being ready for routine use, said Leigh Hochberg, a neurologist at the Massachusetts General Hospital in Boston and another principal of the Braingate project. "It has to make a difference in people's lives, and be affordable," he said. The scientists also need to replicate the data on more people over a longer period of time.

Braingate II seems to take us one baby crawl closer to an age of true brain-machine interfaces (BMIs) that Donoghue and company believe will lead to reconnecting damaged limbs with the brain.

It's also a potential step towards beginning to unravel the millions of individual notes in the neuronal symphony of the brain. Indeed, I asked Donoghue if humans ever disentangle what now is mostly a blur of electrical static.

"Yes", he answered, but that does not mean we will ever be able to reproduce individual consciousness. "Every thought leads to a unique pattern widely spread across millions of neurons. The details of your thoughts, in my opinion, are in that particular pattern and they won't be able to be fully read, potentially ever. They are as unique as the clouds you'll see tomorrow. We can name the general style, but not ever see that exact pattern twice."