Peter Narins needed a way to convince real frogs that a male intruder has just hopped into their territory and is croaking boldly. So the animal communication researcher came up with the obvious choice: condoms. In 2000, Narins, a professor at the University of California, Los Angeles, and his colleagues at the University of Vienna glued a condom to the jaw of a robotic frog equipped with an air pump and speaker. It worked: The condom makes such a believable vocal sac that the robot, despite its immobility, can incite a real frog to spar.
"We actually had two casualties," Narins recalls. Robots were broken during wrestling matches, which Narins says can last up to 15 minutes. "They put their fist right through the condom. I hate when that happens."
Condoms will no longer be an issue for Narins' latest version of a frog robot. In August 2007 Narins will embark on a field expedition to study a different species, one he expects to be less focused on condoms and more interested in legs. Staurois natator, the black-spotted rock frog, lives near loud, fast-running streams in forests, and Narins thinks it's unlikely the animals make vocal calls over the din. Rather, he suspects the frogs' communication relies on a graceful movement called foot flagging, where the frog stretches out its rear leg and slowly rotates it in the air. Instead of an inflatable vocal sac, "robo-staurois" will have three miniature motors (each measuring 2x6 mm) to mimic and modify foot-flagging motions, so Narins and colleagues can see what signals frogs respond to in the field.
To get to the field Narins will have to travel across the globe. In his spacious office at UCLA, surrounded by artifacts collected from his work around the world, Narins opens up a map of Southeast Asia and points to a dot on the northern coast of the island of Borneo. "Going to Brunei," says Narins looking up with a smile.
Narins has been on 42 field expeditions. He's uncovered ultrasonic communication in a Chinese frog, multicue processing in a French Guianan dart-poison frog, and ground vibration hearing in the Puerto Rican white-lipped frog. Narins's trip to Brunei will be his first to focus on visual signals, and robo-staurois, still under construction in Austria, is intended to be a perfect look-alike for the tiny, pimply-skinned S. natator (snout to vent, it's 3-5 cm long).
Columbia University professor Darcy Kelley can attest that it doesn't take much resemblance to the real thing to get frogs to respond to stimuli in the field. In her studies, Kelley has found frogs will grow so enamored of a loudspeaker broadcasting frog calls that they will try to mate with it. Kelley witnessed one suitor persist for a half hour. "It's like they don't believe the evidence of their senses," she says.
Narins has employed his robots to fool frogs' senses in more ways than one. Using a "robo-rana" model to study the dart-poison frog Allobates femoralis, Narins and Walter Hö dl's group from the University of Vienna showed that frogs will fall for a ventriloquism illusion. They positioned a robot frog to inflate its vocal sac when a nearby loudspeaker played frog calls. A real frog would hop over to the robot when the loudspeaker was 12 cm away from the robot; at 25 cm, however, the ventriloquist-effect no longer worked and the frog ignored the robot. Narins says these were his favorite experiments using robo-rana. "This is the first example showing multimodal processing in a nonendotherm, and ... it's the first time it was done in a natural habitat," Narins says.
"He's very imaginative," says Kelley of Narins. "The nice thing [about his studies] is they're experimental, not observational." Part of that ability to experiment in the field is helped by the reliability of the frogs' response to the robots. Dart-poison frogs in French Guiana will attack a croaking robo-rana 89% of the time, says Narins. And the robot's likeness to the real deal is so striking, it can even fool a human. "A foot away or two feet away," Narins admits, "I can't tell if it's a real frog or a model frog."