New Brain-Computer Interface Has Monkeys 'Typing' Up a Storm

Check out this video.

Okay, so it has the look and feel of a circa-1980 video game—maybe something they let high school English students play in an attempt to foster an appreciation of Shakespeare.

But not so fast. It's a lot more interesting when you know the action, such as it is, is being driven by the thoughts of monkeys.

In a paper published in the journal IEEE Monday, Stanford University researchers say they have achieved the equivalent of a "typing" rate of up to 12 words per minute in monkeys. The monkeys moved a cursor to a green dot with their minds by thinking about the green dot. (The primates, alas, were not thinking about actual letters of the alphabet.)

The monkeys' cerebral cortexes had been implanted with electrode arrays, which read their brain signals and translated them into cursor movements. The animals received a reward in the form of a beverage, delivered electronically, after each successful trial.

The experiments outlined in the paper were primarily conducted in 2011 and 2012.

Researchers hope the system could eventually be used by patients with locked-in syndrome and other illnesses that render people incapable of moving or speaking, in order to help them communicate more quickly.

We put the word "typing" in quotes because, first off, the monkeys were doing it virtually, and secondly, the keyboard you see in the video was actually overlaid in post-production, according to Paul Nuyujukian, a Stanford bioengineering department faculty member, who along with engineering professor Krishna Shenoy developed the system.

Trained to Select the Dots

What the monkeys were actually trained to do was navigate the cursor to the green dots as they appeared on a screen, selecting each dot by mentally dwelling on it for about a half-second, or by mentally intending to select it.

In human trials taking place now, Nuyujukian says, participants see the actual letters, as in a keyboard. "We don’t have these results published, but this system can work for a person as well," he says.

The words-per-minute results the monkeys achieved do not account for the actual time it might take a human, who is not simply responding to colored lights, to type similar passages, the researchers acknowledge. "However, for a copy typing task as is commonly used for measuring typing performance, this may be a close approximation," they wrote.

The trials achieved a typing rate up to three times as fast as recorded in other brain-computer interface studies, Nuyujukian says. However, the comparison was calculated using a measure called bit rate, and then converted into words per minute. That was necessary in order to compare these results with those recorded in previous studies.

Multiple Systems Available

If you've seen the film "The Diving Bell and the Butterfly," you know that other systems of communication have long been used by severely disabled patients to communicate.

Perhaps the most famous method was used by physicist Stephen Hawking, who communicated using a sensor on his glasses to detect movement in his cheek muscle.  (That has since been upgraded.) There are also eye-gaze systems that scroll letters across a screen and sense when patients fix their gaze on particular characters.

Those systems can be fatiguing, says Nuyujukian. Another problem, he says: You can’t look away without throwing the system off. And there can be reflection-related problems for those who wear eyeglasses. Though using an eye-gaze system is fast, he says, "It’s not gotten as much adoption as we’d hope from a technology."

Another type of brain-computer interface reads brain signals using sensors arrayed across the scalp. The interfaces are less accurate than an invasive system like the one from Stanford, says Melanie Fried-Oken, a professor of neurology at Oregon Health & Science University in Portland.

"You sneeze, you yawn, you blink, they affect the system more than if you have a circuit board on your brain," says Fried-Oken, who works with locked-in patients.

While she welcomed the system described in IEEE, she thought the researchers' claim of comparative speediness with other systems was a stretch.

"They’re looking at bit rate and humans are looking at words per minute," Fried-Oken said. "That translation from bit rate to words per minute is very difficult. The monkeys are not conceptualizing words as they type."

She also said locked-in patients with whom she has worked think efficiency is more important than speed. "They are happy to communicate slowly," she says.

Fried-Oken said that although some implanted systems have the best signal quality, "you look like a tree's coming out of  your head," although this is beginning to change.

"Every system is difficult," she said. "There will never be a system that’s not difficult."

Eduardo Miranda, who has developed brain-computer interface technology related to music, gave this assessment of the study in an email:

"Even though the work does not address the actual cognitive processing required to form words and sentences, this certainly is a significant development. The great challenge of this invasive approach to BCI research is to translate what we learn from experiments with macaques to real-world applications for humans: using implanted electrodes is not practical, as it involves surgical procedures and the risk of infection is extremely high. Nevertheless the results are exciting.

"The significance of this research will probably become more apparent, and indeed useful, when far more sophisticated non-invasive technology to read the EEG becomes available."