Training the brain to control a single neuron 's activation could restore motion in paralyzed limbs, according to a study to be published tomorrow in Nature. The study represents a novel approach for developing neuroprosthetics.

"This paper demonstrates that simple methods can be very useful," said Leigh Hochberg, a clinician and researcher at Brown University and other institutions, who was not involved with the study.

Previous attempts at designing neuroprosthetics have focused on decoding the activity of a group of neurons associated with movement. Chet Moritz and colleagues from the University of Washington, however, went for individual cells --- by teaching monkeys to control the firing of a single neuron in their motor cortex, using biofeedback in order to power electrodes implanted in a paralyzed hand.

The work fuses two different approaches in neuroprosthetic design -- direct muscle stimulation and brain-machine interfaces. Robert Kirsch and Hunter Peckham and others at the Cleveland Functional Electrical Stimulation Center at Case Western Reserve University have been working on implanting electrodes into muscles surrounding paralyzed limbs, allowing humans to perform tasks like standing, walking, or reaching out and grasping a cup. Meanwhile, in the BrainGate project, Hochberg, John Donoghue and colleagues implanted electrodes directly into the brain of a paralyzed man, allowing him to control a computer cursor and a robotic arm.

"We use the simplest possible conversion from nerve activity to muscle activity," Moritz told The Scientist. They avoided using complex algorithm to interpret averaged neuronal activity, because averaged information "could confuse the patient."

The researchers implanted sensors into the brains of two monkeys to record the activity of individual neurons within the motor cortex. These recordings were routed to computers outside of the monkey's body, which translated the brain signals into electrical impulses that activated electrodes implanted into muscles in the monkey's wrist.

The researchers then temporarily paralyzed the animals' wrists with anaesthetic -- the only way the monkeys could move the muscles that activated their wrists was to learn to control those electrical pulses. The reward for successfully controlling the brain-to-wrist muscle connection: applesauce.

It took the monkeys between 10 to 30 minutes to gain control of a single neuron while doing the task; not only were they able to control their muscle contraction, but the monkeys were also able to control the force of that contraction by regulating neuron firing rates.

Surprisingly, the biofeedback training also worked for motor cortex neurons that didn't normally control wrist movement, said Moritz. "We found that the motor neuron system is remarkable in learning to use this system."

Moritz estimated that it will take about five to ten years before the technology is ready for human trials. First, the sensor will have to be made fully implantable (in the monkeys, part of the electrode stuck through the skin to allow researcher to manipulate it). Also, due to the brain's natural movement and the build-up of scar tissue, implanted sensors usually lose connection to their target neurons within a few weeks, at best. Researchers are working on a new generation of sensors that would keep the connection longer or move to a new neuron when a connection is lost.

Another major hurdle, said Moritz, will be to scale up the system to be able to control more muscles, thus increasing movement complexity. That means users would have to practice "each day for several hours," in order to make some of those brain-muscle associations automatic. "You can only pay attention to about seven things at once," he said.