A stroke in certain parts of the brainstem, the place where brain meets spinal cord, can leave a patient aware of surroundings but able to move few if any voluntary muscles. The most advanced neurotechnologies attempt to get around the disconnection by piping electrical signals directly from a higher-level brain area, the motor cortex that initiates movement, to a robot arm.

Development of technologies for brain control of robotic limbs raise the prospect of practical substitutes for the biological appendages. One advance comes this week with the publication in the May 17 edition of Nature of a report about two patients, paralyzed and unable to talk, who succeeded in moving a robotic arm with signals transmitted directly from their motor cortex. (Scientific American is part of Nature Publishing Group.)

The patients imagined moving the robot arm, which activated a sensor implanted in the cortex. The signals moved to a computer that decoded them and relayed instructions for the positioning of the arm. The BrainGate collaboration involved the Department of Veteran Affairs, Brown University, Massachusetts General Hospital, Harvard Medical School and the German Aerospace Center.

One of the patients, a 58-year-old woman identified in the paper as S3, directed a five-fingered robot hand attached to the mechanical arm to grasp a straw-equipped canister of coffee and bring it to her lips, the first time she had been able to perform that action since she suffered a stroke 14 years earlier. Watch the event itself and an explanation of the technology.

The Brown University team that has led the BrainGate demonstration is one of a series of several groups involved in a highly competitive and sometimes vituperative competition to move this technology forward. Last year Miguel Nicolelis and colleagues at Duke University reported on a monkey that used thoughts to pick up and feel the texture of virtual objects. Andrew Schwartz at the University of Pittsburgh headed a project last year in which a quadriplegic man used a robotic arm to give his girlfriend a high five. Schwartz praised the most recent work as showing the viability of the technology: the electrodes that picked up the brain signals were implanted in one of the patients five years ago, a suggestion that the technology may persist intact for extended periods. “This was a good demonstration of how a useful task could be carried out in a locked-in patient who had a long-term microelectrode implant lasting five years,” Schwartz says.

The idea of translating brain signals into commands that can control a robot goes back decades and has proceeded at times in fits and starts. Cyberkinetics Neurotechnology Systems, the company that started commercializing the BrainGate technology, ceased operations in 2009 before the current clinical trial reported on in Nature was initiated under the aegis of Massachusetts General Hospital. "The most frustrating thing is to see great technology for which there’s not a lot of interest in funding because there’s not enough of a commercial market," says James Cavuoto, editor of the Neurotech Reports newsletter. "There’s only 10 or 11, 000 instances of spinal cord injury each year. It’s just not a big enough market for investors to get interested in.” Nonetheless, competition may have its benefits. Schwartz believes that progress in the field will eventually allow paraplegics to emulate the “smooth, skilled and graceful movement” that comes naturally to the biological appendage when picking up a coffee cup or scratching your nose.