Science

What Makes People Look Like Their Pets?

The eyes are a window to the ego.

Illustration by Mark Alan Stamaty.

Illustration by Mark Alan Stamaty

If ever you overhear someone comparing you to a dog, chances are it’s not a compliment. Yes, there’s the famous loyalty of dogs, their unbridled enthusiasm for life, their boundless love and devotion, their fierce protectiveness—qualities that any of us would be lucky to possess at even a modicum of their standard manifestation in the canine. Typically, though, it’s meant as a slight and a reference to some especially animalistic aspect of our four-legged friends. That assertive woman people call a “bitch,” for instance (a term that has always struck me as being dubious; some of my kindest, gentlest companions in this world have been female dogs), or the lowlife “cur” who cheated you in that game of poker the other night.

As much as we might quibble over the virtues and vices of Canis domesticus, however, and over whether human nature is any better or worse than dog nature, even dog fanciers don’t usually want to look like a dog. The hair of a poodle, the jowls of a bulldog, the bug eyes of a pug, the wrinkles of a Shar-Pei, the profile of a collie, the street-sweeping udders of a lactating mongrel … none of these traits are considered beautiful when incarnated in our own species. Still, if we look in the mirror, each of us can expect to find a certain doggy je ne sais quoi staring back at us. Those of us who own a dog, anyway. And we don’t resemble just any old dog, either. Rather, we look somehow, in a can’t-quite-put-your-finger-on-it kind of way, like our own dogs.

It’s one of those curious observations that’s had scientists scratching their heads for decades. When shown a photo lineup of random people and random dogs, people are able to match the pets with their owners at a rate greater than chance. At first, researchers thought there must be something obvious going on here, something that boils down to a simple, perhaps implicit, heuristic. Maybe men are more likely than women to own large breeds, for example, and women to own toy breeds. Or women with long hair are more likely to own dogs with floppy ears rather than perky ears. Or perhaps obese people overfeed their dogs, and thus we’d expect fat owners to have fat dogs (a correlation that does, in fact, exist). Yet the ability to match strangers with their own dogs holds up even when these more obvious superficial characteristics are carefully ruled out by the research design.

So what is it, exactly, that enables us to correctly link owners and their dogs? That’s the mystery that Sadahiko Nakajima, a psychologist from Kwansei Gakuin University in Japan, set out to solve in a recent study published in the journal Anthrozoös. This wasn’t Nakajima’s first stab at it. In prior research, he and his colleagues had shown that research participants could match photos of owners and their dogs by facial appearance alone. People could also recognize that photos of dogs and owners that the investigators had arbitrarily coupled were fake pairs. Impressive! Still, that just told him that people are surprisingly adept at knowing which pooch goes with which person on the basis of their facial appearance. So in this latest study, Nakajima teased apart the various possibilities to find out which facial features people use to make their bizarrely accurate judgments.

Here’s how it worked. The researcher presented 502 Japanese undergrads with two test sheets. Each sheet included 20 photo sets of dog-human pairs, showing their faces together side-by-side. To eliminate extraneous factors, the photos were very basic color headshots cropped at the shoulders and shown against a plain white background. Nakajima writes that the portraits were taken earlier at a “dog-lovers’ field festival” and that the pet owners were instructed to look straight at the camera and smile slightly. Presumably these instructions worked for the dogs as well—the photos show them with the same Mona Lisa grins as their masters. These resulting 40 human faces and 40 dog faces were digitally rendered equal in size (11 to 12 millimeters “from the vertex [highest point of the forehead] to the chin”). The photos were then randomly assigned to one of those two test sheets. On one test sheet, the images included a set of 20 real-life dog-owner pairs; the other sheet featured 20 randomly matched pairs. These photo sets included an equal number of female and male human owners. It’s not entirely clear to me why mutts weren’t included (perhaps there was a bit of snobbery at that dog-lovers’ fest), but nonetheless there was still a healthy variety of breeds represented in the portraits, everything from the relatively rare Belgian tervuren to that popular pint-sized terror, the Yorkshire terrier, to papillons and golden retrievers.

The judges’ task was simple: “Choose the set of dog-owner pairs that physically resemble each other,” they were told, “Set A or Set B.” Ah, but there was more to it than that. The participants had also been randomly assigned to one of five different “masking” photo conditions. The fundamental difference among these conditions was the way in which the photo sets were presented to the judges on the two sheets: no-mask (in which the participant saw the full unobstructed faces of humans and dogs); eye-mask (the humans’ eyes were covered by black rectangular bars … just think crime-scene photos); mouth-mask (the humans’ mouths were covered in this same way); dog-eye-mask (now it’s the dogs’ eyes that are covered by the creepy black bar); or eye-only (only the thin rectangular slices of the eye regions for both human and dog are shown).

Human dog look alike study.

Courtesy of Sadahiko Nakajima

Just as in Nakajima’s earlier study, the people in the no-mask condition—that’s to say, those who saw the full faces of both the dogs and the owners—were strikingly good at sniffing out the fake domestic bonds. It’s rather amazing, actually, to think that being asked to make a forced choice on “physical resemblance” resulted in 49 of the 61 judges (80 percent) selecting the set of images showing the real-life pairs. Those who saw the same photo sets but with the owners’ mouths concealed (mouth-mask) were only slightly less impressive (73 percent correct). By contrast, simply covering the eyes of either the humans or dogs made the judges’ performance fall to statistically chance levels. So it’s all in the eyes, it turns out.

The most striking finding from Nakajima’s experiment comes from the performance of those participants assigned to the eye-only condition. These judges, you’ll recall, were shown just those thin slices of the human and dog eye regions. Nothing else. Yet 40 of these 54 students (74 percent) still chose the set of true pairs. Nakajima was so surprised by their ability to do this by seeing just the eye regions that he tested a new batch of subjects on the eyes-only condition, just to be sure that the findings weren’t some strange fluke. But this fresh group of judges nailed it, too. This time, 42 of the 55 judges (76 percent) picked the image set with the real-life dog-human pairs.

One value of this study is its ability to tell us which physical cues people aren’t using to correctly match dogs with their human owners. It’s not about hairstyles, obesity, gender, height, or even eye color. As Nakajima points out, since all of the human models were Asian dog-owners, they all had similarly dark-colored eyes. Instead, it’s clearly something that’s being conveyed in the shared look about the eyes of dogs and their people. I’d add something romantic here about the eyes being the window to the soul, and therefore how this all makes sense given that our pets are of course—of course!—our soulmates, except I’m afraid that I don’t believe in souls of either the human or canine variety.

More likely is a logical scientific explanation about our apparently superhuman (or at least subconscious) ability to extract meaningful psychological cues from eyes. Nakajima is just as stumped as the rest of us about the underlying mechanism. And similar riddles exist, too. The psychologist Nicholas Rule and his colleagues, for instance, have found that naive judges can discern the sexual orientation of strangers from their eyes alone. How they’re going about this remains unclear, however, even to the judges themselves.

The good news is that we’ve narrowed it down to the eyes. What it is about the eyes is anyone’s guess at this point. I’m sure you can come up with your own hypotheses.

And while you do that, I’m going to take my border terrier, Gulliver, who’s now begun to shove his head under my hand to keep me from finishing this piece, out for a nice walk. By the way, Gulliver, has anyone ever told you you have the most beautiful eyes?  

Jesse Bering's dog!
He has the author’s eyes.

Courtesy of Jesse Bering