Shell game: Quigley, a curly-coated retriever, follows cues to find hidden treats. Credit: Chris Hildreth.
Shell game: Quigley, a curly-coated retriever, follows cues to find hidden treats. Credit: Chris Hildreth.

Why Dogs Love Us

In his new Duke Canine Cognition Center, evolutionary anthropologist Brian Hare explores the bonds that, over thousands of years, have linked dogs and humans.
January 31, 2010

Buffy is having a good day. Just four years old, she isn't shy around strangers. She shows the energy of a wind-up toy, bumping into chairs with abandon, and the inquisitiveness of an alien dropped into a strange environment, sniffing out the scene in one end of the room and then the other. Originally from sun-speckled Hawaii, she's more at home outdoors than here in the sub-basement of Duke's Biological Sciences Building. In fact, she's an avid walker. She is strong-willed and independent; as young as she is, she enjoys taking the leading role at home.

Buffy, whose distinguishing characteristic is a thick, shaggy, glistening white coat, is a sixty-five-pound mixed-breed "Hawaiian sled dog," as her owner, Cheryl Miller, describes her—some Samoyed, some German shepherd, and some indeterminate other influences. She is playing in what is likely the only Duke lab stocked with boxes of Bow Wow Bites and Fido's Favorites, a Spill-less Smart Bowl, Bouncy Bone dog toys, and a stylized "Dog Xing" sign.

Brian Hare and friend

Chris Hildreth.

This is the Duke Canine Cognition Center, newly created by Brian Hare, a dogged (truly) assistant professor of evolutionary anthropology. He regularly tells visitors, "We have the cheapest tuition and the best acceptance rate at Duke." For each study the lab runs, researchers will test between thirty and fifty dogs; some 550 dogs are in its database. "We've sort of jokingly named the first experiment, 'Does your dog love you more than a stranger?' Really the experiment is all about trust," Hare says. "It's trust or tolerance that's at the core of almost everything we do."

For an hour on this November afternoon, Buffy is completing a food-finding task presented under various conditions. First she's tested to see whether she grabs for tasty treats placed in front of her owner, in front of an experimenter, or at some remove from both. Next she's tested to gauge her understanding of a very basic concept: A green plastic cup is placed over the treat, obscuring it from view; will she realize the treat is still to be found? She gets it—that is, she consistently pushes the cup with her snout in search of her edible objective. "No problem for her. She's 100 percent," says a lab researcher.

From there it gets harder. Two cups are put on the floor; an experimenter may supply one or both, surreptitiously, with the treat. Sometimes the owner points to the one treat-bearing cup. Sometimes an assistant, a "stranger" to Buffy, does. And sometimes, when both cups are provided with the treats, both the owner and the stranger are pointing, one to the left, one to the right. How will the treat-seeking Buffy—deprived of any verbal signals or eye contact—respond to the pointing of her owner versus that of a stranger?

A dog can have a short-term side bias. That is, Buffy might ignore the pointing and favor one side of the room over the other, if that side seems to be producing tasty rewards. So the researchers introduce frequent breaks to, essentially, get the bias out of Buffy's mind.

The casual, non-canine observer might figure that Buffy could simply sniff out the treat. But lab director Kara Schroepfer points out that there are enough residual smells in the room—including those left by previous dog subjects—to confuse a dog in search of a particular telltale smell. Schroepfer says the lab uses a sophisticated videotaping system. Cameras, placed around the ceiling of the room, capture multiple angles, making it easy to see where Buffy looked first and showing whether the experimenters made any mistakes in prompting Buffy.  

"I thought my dog was pretty smart," says Miller after the session. "I wanted to validate that." At home, Buffy is very adept at finding hidden treats, Miller adds. She was curious to see how that knack might be expressed in a lab setting.

Hare adapts well to multiple settings, including the classroom. On a dreary November morning, he's firing up his students in "Human Cognitive Evolution" with a reference to Ferris Bueller's Day Off—though he's at a different end of the evolutionary spectrum from the movie's teacher, the monotonic Ben Stein. Lecturing on mating strategies and the evolution of mate preferences, he mentions a survey of Playboy pinups that highlighted a constant standard of female beauty; the survey pointed to a particular ratio between hip and waist radius. He shows images of Madonna and Denzel Washington to demonstrate the innate appeal of the "incredible symmetry in their facial features." And he explains why females seem to be attracted to men with symmetrical faces that are like their father's face. "That's seriously spooky, and I'm sure Freud would be thrilled at this finding," he tells the students.

At a lecture for graduate students and departmental colleagues—given in a seminar room where coffee, donuts, and bagels are set out alongside early-human skulls and jawbones—he good-naturedly contends with constant technical glitches afflicting his PowerPoint presentation. He wryly encourages the asking of informed questions before he's offered any scientific speculations, imagines a dialogue between a dominant and a submissive chimpanzee competing for food, celebrates the fact that his data appear so small on the projected screen that they can't be contested, describes with mock affection being confined to three acres on an island research station for three months, and shouts out congratulations to Sesame Street on its fortieth anniversary.

While Hare may be adept at creating good feelings, the same is true for his canine subjects. From earlier studies, researchers know that "if you have a stranger play with a dog for twenty minutes, they have—the dog and the human—a reduction in cortisol, a spike in oxytocin, a spike in norepinephrine, and a spike in prolactin," he says. "All those neurotransmitters are things that are involved in making you feel less anxious, very happy, and very social. Cortisol is a stress hormone, and if you have very high levels, it means something is negatively arousing you. So interacting with a dog makes you feel really good. And it makes the dog feel really good."

"Dogs love us," Hare says. "They're obsessed with humans. They're fascinated with us, and they've been bred to be so. It's a little bit artificial for me to have a social interaction with a chimpanzee and make conclusions about its social cognition. With a dog, the best social stimulus you can have is a human."

But humans haven't necessarily been adept at understanding dogs, a phenomenon that presents a scientific opportunity. "Where dogs have been selected to be obsessed with humans, humans have not been selected to be obsessed with dogs," he says. "When I'm with my dog, he's watching me constantly. He wants to be in the same room. He wants to know where I'm going, he wants to know what I'm doing, he wants to know what I'm touching. I'm not watching him that way. That means I miss a lot of stuff that he's doing."

That human-canine dynamic suggests we're too quick to interpret dog behavior in terms of human behavior. A dog's view of the world, though, isn't our view. Imagine your dog, fresh from shredding your living-room couch or gulping down the entire cake on the dining-room table. He goes slinking off with his lowered tail. Is he feeling guilt? It's more likely, Hare says, that he's feeling anxiety as a result of your threatening posture.  

Hare had an early dog obsession. Oreo, a black labrador, was his first dog. "He was my best friend growing up. It was a 'Mary had a little lamb' story: Anywhere I went, Oreo went, too—short of school, of course. We hung out all the time."

Shell game: Quigley, a curly-coated retriever, follows cues to find hidden treats.

Shell game: Quigley, a curly-coated retriever, follows cues to find hidden treats. Chris Hildreth

As they hung out, Hare became fascinated with the dog's ball-fetching skill. "He was totally obsessed, just like any labrador. He didn't want to chase ducks; he wanted to chase tennis balls. He had a really big slobbery mouth, where he could get three tennis balls in at the same time." Oreo was so driven to grab the first ball that he couldn't locate the next couple of balls Hare threw in rapid succession. The balls were somewhere out there, Oreo knew, but where?

Hare, then, had to signal the position of the other balls. "He would use my pointing gesture to locate the ball. I never thought about it. Then I went to college, and I realized that was something that was interesting."

An Atlanta native, Hare went to Emory University. In high school he had interned at Zoo Atlanta, concentrating on drills, baboon-like primates with colorful bottoms—an adaptation that helps their cohorts follow them through the forest. So it was easy for him to talk his way into the primatology lab of Michael Tomasello '72, who treated him as a fellow researcher from the start. Tomasello was investigating the use of gestures to communicate in chimpanzees, one of humankind's closest relatives. Scientists think that at fourteen to eighteen months of age, humans start to pay attention to social cues; we use those cues to read someone else's intentions. If chimps couldn't be shown to replicate that behavior, then, as Hare puts it, the lab would be "onto something really big about what makes us human."

Shell game: Quigley, a curly-coated retriever, follows cues to find hidden treats.

Chris Hildreth

One of Hare's Emory experiments was an early version of his work with dogs: It turned out that if you hide food from chimps and then point to show them where to find it, they just don't get it. "The chimps really are not good at this," says Hare. And it's surprising because they are so good at so many other things. They're good at cooperating. They're good at learning how to use tools. They're good at having coalitions and alliances."

That insight into primate limitations was important, Tomasello assured Hare. "At some point it dawned on me that my dog does that," Hare recalls. "I played with the dog a million times with tennis balls, and I knew that if I pointed, he'd go and get the ball. And Mike said, 'Well, if you think you're right, let's do an experiment.' " The eventual experiment involved Hare's videotaping his two dogs at home in Atlanta—Oreo, who was still around, and his little brother's dog, Daisy. It validated the idea that dogs could locate hidden food by a point alone.

"Everybody would think that there would be hardly any tasks that a dog could do that a chimp couldn't," says Tomasello. "But evolution doesn't work that way. Even insects can do very clever things in their own domain. Evolution is not a ladder up to human beings. It's a tree with all kinds of branches on it. And dogs occupy a very special branch."

Today Tomasello heads the department of developmental and comparative psychology at the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany. He says Hare earned credit on a professional article about his very first project with the Emory research team, when he was just a sophomore. Tomasello's group was looking at the tendency of primates to follow the gaze of others.

"We would wait until two animals were aligned such that one was looking in our direction and the other had its back to us, "Tomasello says. "We would hold up some fruit, for example, and the first one would kind of look over excitedly. And we wanted to know if the one who had has back to us but saw the guy looking over excitedly would take that cue and turn around and look back."

After that, Tomasello recalls, "Brian said, 'Well, I wonder what less-social primates would do, like lemurs?' And of course, the great place to study them is at Duke, at the Duke Lemur Center. It turned out that because a lot of prosimians are nocturnal, they actually didn't even see the stimulus very well when we waved the fruit around. So it didn't really work. But I never had another undergraduate who would drive six hours and stay for some time at a new place to go collect new data."

Some half-dozen more publication credits followed during Hare's time as an undergraduate. "Brian's number of projects and publications was unprecedented," says Tomasello.

Primates have longstanding status as lab subjects. Cognitive research with dogs is a recent phenomenon. "Primates are so sophisticated and so human-like, and we expect them to be so complex in how they solve problems in their rich social groups," says Hare. "How could dogs possibly be interesting?" But cognitive scientists were missing out on something almost literally just under their feet.

Part of what makes them interesting is that dogs, unlike primates, have been selected to interact with humans; they can communicate with humans in ways that other animals can't. They are the first animal to be domesticated, something that happened between 15,000 and 40,000 years ago. A common assumption about dog evolution is that "people just went and grabbed a wolf baby, and it was kind of fun to have a wolf baby around," says Hare. "So we just started having wolves as pets, and somehow we selected them to do stuff like going hunting with us.

"Having worked with wolves, I know that is highly improbable. You can tame a wolf, you can socialize a wolf, but it's still a wolf. It will still bite you. It's still really nervous around you. And it doesn't read your social cues without explicit training. But for dogs, even those with less experience with humans, it doesn't take a lot to get good at all this."

The more compelling evolutionary story is that dogs adopted us—or at least evolved to get along with us. Fifteen thousand to 20,000 years ago, humans were becoming less nomadic; among other things, that meant they produced, in a single place, lots of food leftovers. Proto-dogs found an evolutionary imperative for becoming full-fledged dogs: "If you're going to be a wolf feeding out of human trash, then you had better be a wolf that's not so anxious, not so fearful, and not expending a lot of energy running away from humans," Hare says. "You had better not be a wolf that's aggressive toward humans. Otherwise the humans are going to kill you. So you have selection against aggression and for lower anxiety in wolves. And you end up with dogs."

A first wave of selection for Buffy and her canine cohorts, then, started as an accident—wolves learning where their bread might be buttered, in effect, and adapting to become human companions. But there was a second wave, beginning about 500 years ago, when humans started selecting specifically for dogs that could read social cues.

"Today we've got toy dogs and dogs that are pulling sleds and dogs that find bombs, and there are lots of specialized breeds that are really good at certain jobs"—or, Hare adds, that have really cute features. In problem-solving ability, he says, no dog breed has demonstrably outperformed any other breed. "People write popular books and say, here's the scale of the top-twenty most intelligent breeds. That's based on nothing."

While still at Emory, Hare went pretty much to the ends of the Earth—Novosibirsk, Siberia—for research based on an unusual animal-human interaction. Embarking on the Trans-Siberian Railway from Moscow, he spent three weeks dealing with language barriers, cultural differences in doing the work of science, and the stresses of a desolate setting. "I basically took the same path that anybody banished to Siberia takes. It was one species of grass and one species of tree for days on the train. Very restful, but you're going nuts by the time you get there." 

His purpose for getting there traces back to the Cold War era. In 1959, a group of scientists in Siberia had begun a long-running experiment on domestication. Two separate populations of foxes have been maintained: a control-group population and an experimental population selectively bred based on a single criterion—they fearlessly and non-aggressively approached humans.

Shell game: Quigley, a curly-coated retriever, follows cues to find hidden treats.

Shell game: Quigley, a curly-coated retriever, follows cues to find hidden treats. Chris Hildreth

"Not only does their behavior change—of course, they end up with foxes that run toward people, because that's what they were selected for—but they end up with foxes that have floppy ears, curly tails, multicolored coats, and a smaller cranium, which is what you see in all domesticated animals." If domestication shaped dogs' skills in reading our social cues, foxes that were experimentally domesticated would show the same ability. And that's exactly what Hare found. The specially bred foxes didn't just look like dogs; they were adept at responding to social cues, including the pointing cue, in seeking out hidden food. Hare's experiment found that young domesticated foxes responded to human gestures similarly to puppies.

Shell game: Quigley, a curly-coated retriever, follows cues to find hidden treats.

Chris Hildreth

For his doctoral work at Harvard University, Hare returned to primates, collaborating with Richard Wrangham, Ruth Moore Professor of anthropology and a leading researcher in primate behavior and human evolution. Hare had written to Wrangham as an Emory undergraduate, asking to spend a summer observing in his field camp in Uganda's Kibale National Park. He wanted to test his aptitude for research in the field. But days after his arrival in Kibale, he was told that for over a decade or so, a chimpanzee (or chimpanzees) had been attacking children in neighboring villages. That set him on a different path.

Hare left the camp, rented a room in a village several miles away, and "became deeply involved in the lives of local farmers and the quest to understand why the attacks were happening," Wrangham says. "I think he got my permission, but basically it was clear that he was going to do this regardless."

Today, Hare laughs at the memory. There he was, a twenty-one-year-old kid who barely had traveled outside the U.S., in pursuit of killer chimps. Hare interviewed villagers who said they had been attacked; he interviewed their neighbors as well. "Many times the stories did not agree, or they explicitly said the victims had made up the story to extort money from the park," he recalls. He tracked the movement of chimps, aiming to figure out the extent to which the park boundaries were permeable (meaning that an outsider could be the culprit). And he climbed into more than 250 chimp nests to collect hair samples; the DNA-driven chimp census could point to DNA matches with the attack sites. "All the while my roommate was a chicken. The owner of my apartment gave me a good rate if I let his chicken sleep in my room. I had to take her out each morning at 6:30 to go to the bathroom, until her chicks hatched."

After Hare ended his summer stint, what turned out to be the single culprit struck again: The chimp snatched a baby being looked after by a five-year-old sister. Wrangham's research team found the chimp, wounded by the villagers' spears, in a nest near the village, and they shot him.

Wrangham sees Hare's current scholarly interest in dogs as a natural evolution of his early work with primates. Dogs are "definitely coming to the fore as objects of cognitive study," he says, in part because of what they tell us about domestication and adaptation, and in part for very practical reasons: They are cheap as research subjects, with no need to keep them in a lab; they are numerous; they have breed differences; and wolves are available, if a bit more unruly as subjects, for the study of ancestral traits.

If there's an overarching question that drives Hare, "it's just to try to figure out what makes humans unique, what makes humans human," says Tomasello, his college mentor. There are only two species that have grown up to be adapted to human culture, he says—humans and dogs. Dogs have been domesticated to operate in human society. Understanding dogs, then, is an avenue into understanding humans.

Back in the Duke Canine Cognition Center, Buffy is having a hard time understanding what she's facing. She's being howled at, first by her owner and then by an experimenter. Buffy rubs up against the howler, wanders in tight circles, looks back curiously, and barks a couple of times. She's being put through a variation on the hidden-food experiment. In wolves, howling keeps pack members together. Though it's not scientifically certain, howling also may enforce social bonds. So the lab is testing whether a dog is more likely to howl with its pack member—in this case, its owner—than with a stranger.

One organization that is watching the canine cognition lab carefully is Canine Companions for Independence. (Hare has visited its headquarters in Santa Rosa, California, and met with staff members.) Canine Companions has placed some 1,500 assistance dogs nationwide, mostly Labrador retrievers, golden retrievers, and crossbreeds of the two. Its dogs work with the physically disabled, including injured veterans of the Iraq and Afghanistan wars, and also with hearing-impaired individuals. But the success rate is only about 40 percent, says Paul Mundell, national director of canine programs for Canine Companions. That's one reason the waiting period for eligible individuals is about two years. Most of the failures, he says, are for behavioral reasons. "They have too much energy, they're too nervous in new environments, or they're indifferent to learning. The one drive we really want to see is their wanting to be with people."

Mundell says the training of dogs has long hinged on the very mechanistic notion of behavioral psychology: If a dog shows a certain behavior and you reinforce it, you can get more of that behavior. "That has nothing to say about how dogs are really motivated or tuned into humans. But can we tell which puppies are more likely to work better or learn better with humans? How can we bring out those qualities? As fascinating as it is to know how dogs' cognitive abilities evolved, that could be a practical outcome of Brian's work."

It's not a scary thing to realize that dogs "actually aren't little humans," Hare says. "It's exciting, because they're not identical to us. They have different minds. And the fact that they have these different minds but they still can manage to get along with us and be incredibly successful—I mean, they're the most successful carnivore there is—is remarkable."