This spring, "roborats" were all the rage: The rodents climbed, jumped and moved as directed by researchers, who stimulated the rats' pleasure centers by sending radio signals to receivers on the animals' backs. Media speculation abounded about using the rats to locate people trapped in collapsed buildings or to sniff out bombs.
While the technology the researchers used to train the animals was new, the basic technique is as old as psychology itself.
Indeed, at the heart of the experiment is the workhorse of psychology: conditioning. The rats learned to respond to cues--in this case the feeling of having their whiskers touched--in a very specific manner all for a reward. Over the years, behavioral scientists have trained dogs to rescue people lost in the wilderness, dolphins to recognize hostile movements near U.S. military operations and pigeons to pick out defective drug capsules.
But the potential is even greater than what's been tapped, say psychologists who work with conditioning. Many animals possess sensory systems far more sophisticated than humans' or anything we can build, and they're often able to maneuver in places humans can not. Conditioning could allow us to harness those systems to do anything from screening bags at airports to helping with search and rescue operations.
"If you've got the right technology to arrange reinforcement, you can get animals to do a lot of things," says University of Pennsylvania experimental psychologist Robert Rescorla, PhD. And the conditioning paradigm is pretty much fail-proof.
"It's not much appreciated by people outside the field," says APA's Senior Scientist Susan Brandon, PhD, who uses conditioning in her own work at Yale University. "I always tell the kids in my lab that if the session data look strange, then search for researcher error or an error on the part of the apparatus because the error is virtually never in the animal's conditioning, unless the animal is ill."
A little history
As one of the first psychologists to condition animals for human tasks, B.F. Skinner worked on a plan to train pigeons to help guide missiles to their targets during World War II. Although Skinner's training worked well at getting pigeons to recognize battleships on radar screens as well if not better than humans, the U.S. military had little confidence in the project. Overcoming this lack of confidence--or what some call the "laugh factor"--has plagued many of these kinds of projects.
Navy scientist Jim Simmons, PhD, who, in the late 1970s and early 1980s, used conditioning to train pigeons to conduct search and rescue from Coast Guard helicopters, knows the problem first hand. "That's never going to go away when you start thinking about something like this," he says. "It's hard to believe that these kinds of systems can be reliable."
But eventually the Coast Guard recognized the value of Simmons' pigeons and called his project "the best daylight search system" they had worked with. The pigeons--trained to recognize objects floating in the water and communicate with helicopter pilots by pecking a key that would help guide the pilots to the targets--were 93 percent accurate at locating objects floating at sea and their false positive rates were extremely low. Human flight crews were accurate 38 percent of the time. When combined with human searchers, the pigeons' success rate went up to nearly perfect. In addition, they didn't get bored--Simmons trained them to respond even when there were hours between sightings.
Despite its success, the program was plagued by problems--two helicopters crashed, either destroying or damaging the pigeon system. Then funding was cut when the Coast Guard's priorities changed under President Ronald Reagan.
Now, admits Simmons, the pigeons might not compete with computer-aided searching. But the program demonstrates how powerful the conditioning paradigm can be.
He believes the same techniques could be used to train pigeons--which have excellent abilities to recognize visual patterns--to scan bags at an airport or for other quality-control types of tasks. In fact, the late psychologist Thom Verhave, PhD, trained the birds to pick out defective drug capsules for a drug company in 1966. Published accounts of his work show a remarkable success rate--99 percent accurate after only a week's training--but the program never got off the ground, reportedly because the drug company was afraid of image problems.
In the end, "it's a societal question whether people will accept animals in these kinds of roles," says Rescorla. "History seems to suggest they won't."
Would he trust a pigeon to screen baggage at our airports? "I sure would," he says. "I think humans are more easily distracted. You can make it part of an animal's training to not be distracted, to ignore extraneous stimuli."
Roborats to the rescue?
Whether roborats can overcome the image problems other animals have faced is an open question, admits Sanjiv Talwar, PhD, who led the roborat research team at the State University of New York Downstate Medical Center. He realizes that much of the media attention stems more from a fascination with the novelty of watching rats run around with backpacks than the actual science.
But that doesn't take away from the fact that his team has developed some powerful technology--they create both the conditioning cues and rewards inside the rats' brains by using electrodes to stimulate specific brain areas. What's more, they can trigger the electrodes wirelessly, through a computer up to 500 meters away.
For "cues," Talwar and his team inserted two electrodes in the brain region that controls sensory stimulation to their right and left whiskers. A jolt to one or the other made the rats think something was touching their right or left whisker.
As a reward, the researchers also inserted an electrode in the rats' pleasure center. Over a series of training sessions inside a maze, the researchers rewarded the rats for moving correctly in response to whisker "touches" with jolts to their pleasure center.
Within 10 days, Talwar had the rats moving left in response to a "touch" on the left whiskers and right in response to a "touch" on the right whiskers. In fact, he could get the rats to run, climb or jump even outside the maze and in settings where rats normally avoid, simply by rewarding them for their work.
When he saw how well this particular paradigm worked in the rats he immediately thought of applications outside the original intent--the work is part of a much larger research program within the lab of behavioral neurobiologist John Chapin, PhD, with the goal of returning sensory and motor function to people who have lost limb use.
"It seemed wild to me that from almost a kilometer away you could wirelessly control where the rat was moving," he says. "It seemed incredibly powerful and potentially effective as a humanitarian tool."
Right now, researchers need to be able to see the rats to steer them. But Talwar envisions the day whey they can use heat sensors, video cameras, global positioning satellite devices and other technology to move them around spaces hidden from sight. Another idea they're working on that might be highly effective, says Talwar, is to use the rat's sophisticated sense of smell to gain insight into what the animal is sensing in its environment.
"First we need to know how to translate those signals," says Talwar. "If we can identify the neural signal for certain smells we might be able to use that information to understand what the animal is facing--perhaps explosives, perhaps human bodies--whatever we want."
Practicality and ethics
Most conditioning researchers agree that animals are best used in situations that take advantage of some highly developed sensory system that outperforms what humans can do or simulate.
"Where [animals] radically outperform anything we can create technologically now or for the foreseeable future is in navigating over rough terrain and in image processing/pattern recognition," says Rutgers University conditioning expert Randy Gallistel. Rats, for example, are useful for their small size, nimble motor skills and keen sense of smell. Pigeons have an excellent visual range and can be taught to discriminate specific objects as well as objects within certain categories. And goldfish have been found to have a remarkable ability to detect pollutants in water.
"A fish can be conditioned to make use of its behavioral sensitivity to a large number of substances to indicate that a new substance is present," explains Kurt Salzinger, PhD, APA's executive director for science who, with S.P. Fairhurst, PhD, S.J. Freimark, PhD, and D.D. Wolkoff, PhD, capitalized on the fish's talent in 1973. "Chemical tests can then follow to determine what that substance might be. But chemical tests are for specific substances while fish would show sensitivity to many different substances all at once."
And, unlike the canary in the coal mine, says Salzinger, his fish responded with behavioral cues rather than their lives. Of course, other animals trained for human work might not be so lucky. In fact, the ethical questions of using animals for human benefit was one of the main issues Talwar and his colleagues had to address when the media crush hit.
But, insists Talwar, his rats are extremely well cared for and far more stimulated than the average lab rat. Yes, they have transmitters implanted in their brains, but these devices are implanted the same way as in humans: since the brain has no pain receptors, there is no pain associated with the device once it is in place.
Whether this technology could be misused to manipulate higher functioning animals, or even humans, is another issue, he admits. One he's sure people will grapple with as the technology gets more sophisticated.Beth Azar is a writer in Portland, Ore.