Improvements to computer code are making a mind-controlled mouse cursor more accurate than ever, according to a new study.
Although it sounds like science fiction, research labs have worked for years to develop mind-controlled cursors and prosthetic limbs for people with disabilities. Researchers have tested devices in monkeys and made working limbs for people who cannot control their arms. One research group, BrainGate, has begun seeking U.S. Food and Drug Administration approval for its mind-controlled arms, a process that could take years.
BrainGate may benefit from the latest improvements. BrainGate is running Phase I tests for FDA approval at Stanford University, according to the engineer who led research into the new cursor controller. "These findings could lead to greatly improved prosthetic system performance and robustness," Stanford engineer Krishna Shenoy said in a statement.
Shenoy and his colleagues tested their new computer program in two rhesus monkeys that had silicon chips implanted in their brains. The chip read electrical activity in individual brain cells in the monkeys, while the program interpreted what the chip read. The interpretation happened while the monkeys tried to control cursors on a screen using their thoughts. (The monkeys were rewarded with juice if they successfully moved the cursor over colored dots on-screen.)
Before this, mind-control programs generally worked by gathering data about testers' brain activity, then analyzing the data afterward, not simultaneously, Vikash Gilja, a scientist in Shenoy's lab, said. The real-time data analysis helped monkeys drag cursors twice as quickly as Shenoy's previous mind-reading system, the researchers reported. The monkeys could control the cursors 75 to 85 percent as quickly as they could operate a mouse with their arms.
The researchers also found the system works well over time. Their two monkeys have used the new computer program over the past four years, they reported.
Shenoy, Gilja and their collaborators published their findings Nov. 18 in the journal Nature Neuroscience.