Related Issue
September–October 2024
ILLUSTRATION: Mark Ulriksen
Science & Tech
What’s Fido Thinking?
Brown’s canine cognition lab is not just a walk in the park.
By Julianna Chang ’26 / September–October 2024
August 27th, 2024
What makes a dog smart? When we take our dogs out on a walk, what do they look at? Do dogs actually understand when we point under the couch and say the ball is right there, so please go get it and stop barking?
Tucked away on the third floor of Metcalf Research Building, the Brown Dog Lab is answering these questions. Founded by Daphna Buchsbaum ’02, assistant professor of cognitive, linguistic, and psychological sciences, it’s one half of the Computational Cognitive Development Lab (the other half studies children).
Studying canines, Buchsbaum explains, is becoming increasingly popular among cognitive scientists because they’re smaller, more domesticated, and less dangerous than chimpanzees—humans’ most closest genetic relatives. Also, dogs are pretty good at communicating with humans, Buchsbaum says, as “we’ve arguably selected them to fit more naturally into the human physical and social environment.” The lab recruits volunteer subjects but has also become popular with its academic neighbors: on good days, dogs from around the building stop in for attention and treats.
Dogs and distrust
It’s undeniable that dogs trust some people more than others. But why? Past research shows some trust people based on familiarity or how nice they are to their owners, says Madeline Pelgrim ’23 ScM, a fourth-year PhD candidate. But Buchsbaum and Pelgrim had another idea: past accuracy.
Using dogs’ understanding of human pointing, the lab had two informants point repeatedly in opposite directions—the “trustworthy” one would always point in the direction of a treat, while the other would point away from it. When faced with these informants again and again, would dogs adapt?
The idea stemmed from previous research with children: kids are able to distinguish the trustworthy from the untrustworthy with verbal testimony, but they will often follow inaccurate points until age four or five. In their 2021 study, Pelgrim and Buchsbaum found that dogs are able to selectively trust informants based on their past accuracy. And though they struggled with the task, the dogs were able to use pointing cues to determine the informants’ accuracy—unlike young children—which indicates that pointing may be particularly powerful for dogs.
Dog’s eye view
Many child researchers use eye tracking goggles to understand what’s drawing infants’ attention and how that connects to the developmental concepts they’re learning. Buchsbaum says these researchers often claim that “this ability to engage with other people in this way is uniquely human.”
Buchsbaum and Pelgrim disagreed. “I had the thought of going to some of the people who make the eye tracking goggles and asking, ‘Could you make some of these for dogs?’” Buchsbaum said. “Luckily, they said yes.”
No one knew if it would work, but with slight adjustments, dogs were soon coming into the lab to try on a pair of stylish black frames. Thus adorned, the canine subjects—and their owners—went on a half-mile walk around campus while researchers recorded surroundings and eye movements.
Dogs were soon coming into the lab to try on a pair of stylish black frames.
In the soon-to-be published results, Pelgrim found that “dogs are actually fairly consistent in terms of how they move their heads through the world.” Like humans, they spend a lot of their time wayfinding—looking forward so they don’t trip and fall. But the objects that interested them most? Construction equipment, nearby people and, of course, the Brown bus.
Next Pelgrim will focus on social interaction. Using the goggles in collaboration with researchers in computer science, she and Buchsbaum hope to understand how humans and dogs interact with each other visually—especially through pointing gestures.
Good Robot!
How can humans effectively communicate and cooperate with robots? Human-human interaction is often extremely complicated—involving advanced thought processes, goals, and false beliefs. Robots just aren’t there yet. But what if you think of the robot as a dog—another non-
human species that frequently and successfully partners with humans?
Collaborating with researchers in the computer science department and other cognitive labs on campus, the Dog Lab is starting to adapt many human-dog interactions—like airport security and drug detection—to robots to see how humans’ perceptions of their task partner change.
Canine Einsteins
“My dog is too dumb to participate”—Miriam Ross ’24 ScM, a second-year PhD student in the lab, hears that a lot. The owners explain that their dogs can’t untangle themselves from their leash or find a ball under the couch or grab a treat from right under their nose. Sometimes, the opposite occurs: some owners are immensely confident in the intelligence of their dog, when “from our anecdotal observation,” Buchsbaum says, “their dog was, in fact, not an Einstein.”
When owners with so-called “dumb” dogs decline to participate, Ross and Buchsbaum wonder if it skews the data: Do these “dumb” dogs actually possess a particular characteristic that would correlate with worse performance in these tasks? Or are they just dogs whose owners think they’re dumb?
Inspired by these questions, Ross’s current project focuses on how humans perceive dogs’ abilities and intelligence, and what specific factors make us think a dog is “particularly smart or socially savvy or physically able,” Buchsbaum says. Ross anticipates the findings could inform how we select guide dogs for training—a costly and time-consuming process. But for now, when owners insist that their dog is too dumb, Ross reassures them: “Your dog doesn’t need to be the Einstein of the task. We’re just curious about what comes to them intuitively.”
For Buchsbaum, one of the best parts about studying dogs is how much interest the research generates: “Everyone is always wanting to hear more.” But there’s also the fact that the dogs continually surprise all the researchers. “I’ve never had a study in which I’ve predicted all of the ways that the dogs will behave,” Pelgrim says. “We can all talk dog forever,” Buchsbaum adds, and Ross agrees, laughing: “Maybe that’s why we’re doing this.”
