One day last summer, a curious white-faced capuchin encountered a strange contraption in the forest. There, in the middle of the Taboga Forest Reserve in Costa Rica, sat a 15-inch touch screen, mounted in a wooden frame. As the monkey, an alpha male named Papi, began investigating — poking the device here, prodding it there — his fingers landed on the screen. Suddenly, a piece of dried banana dropped into a tray below the frame.
Before long, Papi learned the basic rules of the device: touch the screen, get a dried banana slice. He also provided proof of concept for CapuchinAI, a new device designed to assess the cognitive abilities of monkeys in the wild. The testing apparatus, which the researchers described in a new paper, used A.I.-powered facial recognition software to detect capuchins in real time and record their responses to a simple learning task.
The scientists hope that more sophisticated versions of the device, which they will begin testing in the coming weeks, will shed new light on primate evolution and intelligence, and answer questions that would be impossible to study in a lab, such as how a monkey’s smarts affect its success and survival.
“If we really want to understand how primates make decisions, if we want to understand how they’re using these large brains that they evolved, we have to really put it in the context of the world in which they’re navigating,” said Marcela Benítez, a primatologist at Emory University and an author of the new paper, which was published in the American Journal of Primatology on Tuesday. “But that is a lot easier said than done.”
Dr. Benítez has been studying the white-faced capuchins at Taboga for years, logging their behavior, recording their vocalizations and measuring their hormone levels. For the new study, she and her colleagues used images of some of these monkeys to train an artificial intelligence model to identify capuchins and distinguish them from the other animals roaming the forest. Then they built a portable testing station equipped with a touch screen, webcam and 3-D-printed food dispenser.
The station was designed to detect an approaching capuchin and present it with a simple test of associative learning: Could the animal figure out that touching the screen meant receiving a reward?
Last summer, the researchers put the device to the test, setting it up in the forest and hoping that the monkeys would engage. On the first two days, not a single capuchin appeared.
“And so, you know, in my typical first-time-researcher, Ph.D.-student fashion, I was there biting my nails,” said Federico Sánchez Vargas, a doctoral student at Emory University and an author of the new paper. “‘Oh no, I’m going to have to start from scratch, do something else.’”
But on the third day, Papi showed up. Eventually, other monkeys did, too. And over the course of multiple testing sessions, 16 capuchins explored or interacted with the device.
Not all of them understood the task. One monkey “would bite around the area where the banana would come out,” Mr. Sánchez Vargas said. “He would get on top of the box and slap the box, but he just did not think to touch the screen.”
But 10 of the 16 monkeys learned how to touch the screen to earn a reward, and at least eight of them seemed to retain that knowledge.
This summer, the researchers will pilot test stations with a more sophisticated facial recognition model, one trained to identify and distinguish among individual monkeys. An earlier version of this capuchin identification model, trained on a smaller group of capuchins, was able to identify individual monkeys with more than 97 percent accuracy.
The machines will be loaded with a library of tasks designed to assess a variety of cognitive skills. The tasks include tests of impulse control, in which monkeys are rewarded for touching green squares and ignoring red ones, and of cognitive flexibility, in which the rules of the green-square, red-square task are reversed. The testing stations are designed to keep track of each monkey’s progress, picking back up where it left off on the previous session.
The researchers are especially interested in exploring individual variation in cognitive skills — and both the causes and consequences of that variation. Does experiencing adversity, like a severe drought, early in life affect cognitive development? Does a good short-term memory help monkeys find food?
“Linking the cognition to real-world challenges, outcomes and ultimately fitness — that would be the total end goal,” Dr. Benítez said.
Still, the researchers have no plans to make touch screens permanent fixtures on the landscape. “I don’t want it to be like walking into a casino when you’re walking into the forest,” Dr. Benítez said.
And for now, she added, the stations are not designed to be used without human oversight. “I’m not at the point in this yet where I want it to be fully remote,” she said. “But maybe ask me in 10 years.”
