A pair of partially paralyzed monkeys regained the ability to move their wrists when researchers wired individual neurons directly to the monkey’s arm muscles, according to a study published online in Nature on Wednesday.

“This approach would be very relevant to spinal-cord injury patients,” says Northwestern University neuroscientist Lee Miller, who was not involved in the work.

The researchers, led by Eberhard Fetz, a professor of physiology and biophysics at the University of Washington, temporarily paralyzed each monkey’s arm. Then they rerouted brain signals around the blocked nerve pathway by running wires from a single neuron in the motor cortex–the brain area responsible for movement–through a computer and into a muscle in the arm. Whenever the neuron fired above a certain rate, the computer translated the signal into a jolt of electricity to the arm muscle, causing it to contract.

As a test of the rewiring, the researchers had each monkey play a simple video game. By moving its wrist, the monkey could manipulate a cursor on a computer screen. Moving the cursor into a box at the side of the screen earned the monkey a reward. Even though the rewired brain cell was chosen at random, the monkeys quickly learned to move their paralyzed wrists.

“We found, remarkably, that nearly every neuron that we tested in the brain could be used to control this type of stimulation,” says Chet Moritz, a senior research fellow at the University of Washingtonand coauthor of the paper. “Even neurons which were unrelated to the movement of the wrist before the nerve block could be brought under control and co-opted.”

Normally, arm movement–even the contraction of a single arm muscle–would not result from the firing of a single neuron but from the coordinated action of many neurons in the motor cortex. Those neurons would initiate an electrical signal that propagated down the spinal cord and through peripheral nerves to trigger arm movements tailored to the monkey’s intention.

Other groups have recorded those complex neuron-firing patterns and used computer algorithms to translate them into action–for instance, moving a computer cursor. Instead, the University of Washington group linked a single neuron to a single muscle. “Our approach is to create the raw connectivity between single neurons in the brain and muscles, or groups of muscles, and let the monkey learn how to use that connectivity.”

Using a single neuron has its advantages, says Moritz. Translating one cell’s firing rate into an electrical shock is a straightforward computation, easily accomplished with a device the size of a cell phone. Translating simultaneous measurements into a suite of coordinated muscle movements takes far more computing power.

But for the single-neuron approach to be useful to a paralyzed patient, it will need to successfully scale up. Contracting one arm muscle offers little practical reward; movements like reaching and grasping require many muscles to work in concert. The researchers have already taken steps in this direction. First, they showed that a single cell could work two different muscles: a high firing rate triggered the wrist to flex, while a low firing rate caused it to extend. Next, they hooked up two rerouted connections at the same time, with one neuron wired to the wrist-extending muscle and another to the wrist-flexing muscle.

But Andrew Schwartz, a professor of neurobiology at the University of Pittsburgh, is skeptical. A moving arm, says Schwartz, is “a very complicated mechanical system.” Any given sophisticated arm movement requires not only a large number of precisely coordinated muscles acting across several complex joints, but also the propagation of forces along the limb. “If your intention is to generate a movement, you have to somehow calculate the effect of all these forces across the arm,” says Schwartz. “It’s not just, ‘Activate a muscle and the arm goes where you want.’ There’s a lot of math involved.”

According to the University of Washington group, it may be possible to circumvent the question of how to generate intricate movements by wiring a single brain cell directly to a specific region of the spinal cord. “Stimulating a single location in the spinal cord will often activate 10 to 15 different muscles in a precise balance,” says Moritz.

Beyond any theoretical shortcomings of the single-neuron strategy, there are a number of technological hurdles to overcome before it could be used in patients. Electrode readings from an individual brain cell can degrade over time, potentially destroying the rerouted connection. As a result, says Moritz, any long-term setup would need some degree of redundancy.

Also, he adds, the system would ideally be fully implantable. Whenever wires protrude through the skin, as they did in the monkey experiments, they introduce risks of infection and disruption. The group plans to tackle this problem with miniaturized components and wireless technology.

Because their approach requires relatively little computing power, says Moritz, “we think we may be one step closer to low-power, fully implantable systems.”