AMHERST, Mass. -- One gray squirrel, its bushy tail twitching, barked a warning as another scrounged for food nearby.
It was an ordinary spring day at Hampshire College, except that the rodent issuing the warning was powered by amps, not acorns.
Dubbed "Rocky" after the cartoon character, the robo-squirrel is working its way into Hampshire's live-squirrel clique, controlled by researchers several yards away with a laptop computer and binoculars.
Sarah Partan, an assistant professor in animal behavior at Hampshire, hopes that by capturing a close-up view of squirrels in nature, Rocky will help her team decode squirrels' communication techniques, social cues and survival instincts.
Rocky is among many robotic critters worldwide helping researchers observe animals in their natural environments rather than in labs. The research could let scientists better understand how animals work in groups, court, intimidate rivals and warn allies of danger.
In Indiana, for instance, a fake lizard shows off its machismo as researchers assess which actions intimidate and which attract real lizards. Pheromone-soaked cockroach counterfeits in Brussels, meanwhile, exert peer pressure on real roaches to move out of protective darkness. In California, a tiny video camera inside a fake female sage grouse records close-up details as it's wooed — and more — by the breed's unusually promiscuous males.
The research may even help explain similar instinctive behaviors in humans, researchers say.
"Animals and humans are all affected by behaviors, body postures and signals from each other that we may not be aware of," Partan said.
The use of fake critters to infiltrate real groups of animals is so new that few companies build or sell such tools to researchers.
Many of the scientists using animal doppelgangers have modified toy animals or, like Partan and her students, cobbled together their own with fake fur, small motors, circuits and other material. Partan, who created Rocky a few years ago with students when she taught at the University of South Florida, is constantly refining its actions and updating its technology.
Rocky's movement is controlled by basic computer programs, and it has tiny speakers inside that play recordings Partan purchased from an animal-sounds library at Cornell University.
One recent afternoon, she and students Maya Gounard, 20, and Andrew Fulmer, 19, brought Rocky out for field testing and placed him near real squirrels. Mounted on a board, he was shielded by a camouflage hood and a long cord connected him to the researchers' laptop.
After the computer's program flipped the hood open, Rocky went into a sequence of tail-flagging, barking and other motions squirrels recognize as warnings of danger.
The most successful experiments are when the real squirrels respond by "flagging" their own tail, halting their foraging to check for danger, scamper up a tree or take other actions that show they picked up on the signals, Partan said.
"We watch for a trade-off in their behavior," she said, pointing out a squirrel that jerked to its hind legs and froze, its eyes scanning the area as it heard Rocky's barks. "He gave up foraging to focus on being vigilant, so that's something we'd note as a discernible response."
They focus on whether squirrels react more strongly to Rocky's noises or movements or a combination that researchers call multi-modal signals.
Although animal behavior has been studied for years, much remains unknown about instinctive responses.
A particular sound may be the courting equivalent of, "Come over here, you sexy beast." But a tiny change can alter the message entirely, making it something akin to, "You're about to be torn to shreds if you don't get out of my territory."
"Whether it's a bunch of squirrels in a field or humans in a mall, there are general principles of behavior that seem to hold up across species lines," said Greg Demas, director of Indiana University's Center for the Integrative Study of Animal Behavior and an associate professor of biology.
Robot critters also can help researchers discover how far a species can be pushed beyond its survival instincts.
Researchers at the Free University of Brussels, for instance, found that fake roaches doused in familiar pheromones became so accepted among their cockroach compatriots that the real bugs succumbed to the interlopers' peer pressure to move out of dark areas into the light.
In other experiments, a robotic lizard developed by Indiana University researcher Emilia Martins uses energetic push-ups to trigger similar displays of courtship, power and machismo among real lizards.
Depending on the fake lizard's prompting, the real critters react as if they're being taunted, threatened or titillated — all of which gives researchers a chance to study the tiniest movements of their legs, eye flaps and other quirks.
"There's been the old, classic trade-off for years between the ecological relevance you get (researching) in the field, versus those studies in the lab where you can control the environment while knowing they're not going to react as much," Demas said. "Having these models out in the field is taking us to the next steps of the research."
Researchers say the applicability of fake animals in research can depend on the intelligence, size, eyesight and sense of smell in the real species.
"The bigger the animal is and the more complicated it is, the harder it is to have a proper robot that mimics the signals and has the right visual cues," said Cornell ornithology professor Jack Bradbury.
Bradbury's research has ranged from vocal mimicry in wild parrots to the sexual choices of hermaphroditic sea slugs. He hasn't used robots but does use sound cues — emanating from speakers hidden in bushes — to manipulate animals in the wild by "talking" with them or playing noises they recognize.
"Wild parrots are pretty smart, but I've gone on for hours interacting with them that way," he said. "They come up to the bush and look at it and don't see the birds, but they keep communicating with the belief there's another parrot in there somewhere."
He said mechanical animals aren't used "just to be clever."
"The real issue from a scientist's point of view is, 'Can I come up with a robot that will help me answer a question that I couldn't answer otherwise?'"