Rats turned into remote controlled robots

Scientists for the first time have managed to remotely direct the movements of rats by using implanted electrodes to control their behavior — in effect transforming living animals into robots.


The technique has potentially important implications for activities ranging from land mine detection, earthquake recovery and spying to the emerging field of “neural prostheses” — using electronics to bridge nervous-system gaps caused by spinal injuries, strokes or other physical infirmities.

“It’s really just conditioning behavior,” said physiologist John K. Chapin of the State University of New York’s Downstate Medical Center, noting that training an animal to do human bidding is as old as teaching dogs to fetch. “But it’s different in that you can do it all with remote control,” he added. “In theory, you could guide the animal anywhere.”

In fact, after implanting the electrodes and training a “roborat” for eight to 10 days in a figure-eight-shaped maze, the Chapin-led team could steer it through any three-dimensional route. It could induce the animal to climb ladders, descend ramps, walk on a pipe or navigate through uneven terrain. The rat would even climb trees or wander around a brightly lit room — alien behaviors for the untrained.

“I really like the results,” said Northwestern University physiologist Sandro Mussa-Ivaldi. “People have been doing conditioning with reflex behaviors for a long time, but this is the first time where you have control of a whole complex animal.”

Chapin said the research, reported yesterday in the journal Nature, was inspired by his own and many others’ efforts to use electronics to help the disabled bridge the gap between what their brains want to do and what their damaged bodies are able to do.

Chapin and others demonstrated in earlier research that rats could be trained to retrieve rewards by operating a robotic arm with neural motor impulses captured by electrodes.

But Chapin wanted to do the same thing with sensory impulses. “Imagine a paralyzed person grasping a glass of water and bringing it to his mouth,” Chapin said. “But he can’t feel the glass when he grasps it, so now we have to work on the sensory side. The rat is the first experiment. Can it discriminate between one stimulus and another?”

The Chapin team implanted three electrodes in the rat’s brain. One was placed in a “generic” pleasure center that records satisfaction whenever needs — for food, water or warmth — are satisfied. The others were implanted to stimulate the whisker bundles on either side of the rat’s nose.

By triggering one of the whisker implants and then stimulating the reward center, the researchers were able to make the rat turn in one direction or the other and move forward — much as a sled driver can order his dogs to “gee” or “haw.”

After up to 10 days of training, the rat could navigate practically any landscape, wearing a receiver and a power pack on its back and being steered by a technician issuing commands from a laptop computer up to 550 yards away, Chapin said.

The rat thus becomes a living robot, controlled remotely by a human handler but able to go anywhere a rat can go. And its supple anatomy gives it a huge and — at least for now — insurmountable advantage over any mechanical robot, which can be confounded by a pair of shoes lying on a carpet.

“This trumps that problem,” Chapin said. “The rat is much more adept than a robot at getting around difficult terrain — and it has a nose.”

The military and public service potential of the project has won funding from the Defense Advanced Research Projects Agency, the Army’s research arm. Camera-equipped rats may have a future as land mine detectors, or as couriers or scouts searching for human victims trapped in collapsed buildings or mine shafts, Chapin said.

They could also be used as the “rat on the floor” equivalent of the “fly on the wall,” providing a real-time ability to eavesdrop on sensitive conversations taking place behind closed doors.

The implications of the experiment for neural prostheses are far less clear. “The idea is wonderful, and it’s really interesting that John has accomplished this,” said Rutgers University neuroscientist Gyorgy Buzsaki.”But robotic control is relatively easy, and if you want to achieve complex patterns of behavior” — by training a human brain to react instinctively to an electronically transmitted stimulus — “it gets very complicated in a hurry.”

Also, Chapin noted, “we’re trying to avoid using bigger animals” because of the “big brother” ethical issues involved in developing a technique that in many cases overrides an animal’s natural instincts.

“The rat looks normal and isn’t feeling any pain because he’s getting rewards for doing the right thing,” Chapin said. “They get very tame. They love to get picked up, and they don’t even have to be sacrificed because the longer we use them the better they get. We have one old lady rat that received an implant at the beginning of last September.”

SOURCE: © 2002 The Washington Post Company
Guy Gugliotta, Washington Post Staff Writer