Immobilised monkeys with electrode filaments inserted into their cerebral cortex learned in only days to reach out with the free-standing prosthesis, pluck a tasty morsel with a pincer-like claw, and pop it in their mouths.
When the path of the arm - positioned next to the shoulder – was deliberately blocked, the animals simply willed it around the obstacle with their minds, says the study, published in Nature.
"The entire task is now performed with brain control," Andrew Schwartz, the lead researcher and a professor at the University of Pittsburgh, told AFP.
Study findings a 'first'
In preliminary experiments, also with Macaca mulatta monkeys, computers assisted with various parts of the task, he explained.
The study's findings are the first reported use of a so-called "brain-machine interface" (BMI) to perform a practical action in three dimensions - in this case feeding oneself - purely via brain control of a computerised arm, noted John Kalaska, an expert on the central nervous system at the University of Montreal.
Strokes, spinal cord injuries and degenerative neuromuscular diseases cripple tens of thousands of people every year, rendering the simplest of actions - opening a door, scratching an itch, drinking a glass of water - frustratingly difficult or impossible.
Those afflicted with the most severe form of paralysis, known as locked-in syndrome, are fully-conscious prisoners inside a body that no longer responds to the most basic of commands.
"These patients are still able to produce the brain activity that would normally result in voluntary movements," explained Kalaska.
"But their condition prevents those signals from either getting to the muscles or activating them," he said in a commentary, also in Nature.
Experiments offer hope to paralysis victims
Schwartz's experiments provide the most tantalising hope to date that paralysis victims can one day short-circuit their own nervous system by hardwiring their brains directly to a computerised robot.
"Hopefully we will be implanting microelectrode arrays [in humans] in the next two years," Schwartz said. "At that point it should be relatively easy to perform this kind of task," he said.
In the meantime, Schwartz and his team are making improvements on the robotic arm, adding points of articulation in the wrist and hand to the five already built in - three at the shoulder, one at the elbow, and one at the hand.
This does not mean that "neuroprosthetic robots will soon be available at the local rehabilitation clinic," cautions Kalakska, who says several barriers remain before such devices can be easily deployed.
The long-term reliability of the electrodes - about the breadth of a human hair - must be vastly improved.
"Patients will need to use this technology for many years, but the quality of the recorded neural activity often deteriorates within weeks or months," he said.
Problems to be ironed out
Portability is also a problem. In the laboratory, the successful experiments depend on a vast array of electronic and robotic equipment under the constant supervision of a trained technician.
Rendering these operations more compact and more automatic are not impossible, but will take time.
Finally, without tactile sensors to provide a more complete control of the artificial appendage, it is nearly impossible for the user to gauge the force with which an object is grasped.
The aim is "to pick up an object with a strong enough grip to prevent it slipping from the robotic hand, but not so strong as to crush it," Kalakska said. – (Sapa, May 2008)