What Does a Dog See in a Mirror?

Photo: David Ward/Getty Images

Serious question: When your dog looks in the mirror, what does he see? A playmate, maybe. An adversary, sometimes. But could he possibly recognize himself? For that matter — does a dog have a sense of “self”?

These are abstract questions, and it is too bad for us that dogs can’t answer for themselves. And so, instead, scientists who study animal cognition have devised a test for measuring these hard-to-comprehend concepts, involving the use of one commonplace household object: the mirror. It’s called the mirror self-recognition test, and since its inception in the 1970s it has been considered be the “gold standard” of determining whether or not a creature possesses self-awareness. Elephants, chimpanzees, and dolphins are among the creatures who have passed, suggesting that these animals have a sense of self. But plenty of other primates, along with highly intelligent creatures like octopuses, are either confused by or totally uninterested in the mirror.

Dogs, for example — as you no doubt know if you have ever kept one for a pet — fail the mirror test. They do not seem to know, or perhaps they don’t really care, what is going on with the mirror’s scentless, two-dimensional representation of a dog. They may try to fight the dog in the mirror, or play with it, but very few dogs demonstrate any behaviors that would signal they recognize the reflection as their own. According to the standards set by this test, this means that dogs must lack a self-concept. And yet, in recent years, some prominent scientists have begun to question the authority of this test. “People say, ‘This species has no self-awareness because we tested it in the mirror,’” primatologist Frans de Waal said in a recent interview with Science of Us. “But I would argue that self-awareness is a broader concept than that. And I cannot imagine that a cat or a dog — even though they don’t recognize themselves in the mirror — I find it hard to imagine that they have no awareness of themselves.”

It’s possible, then, that an animal’s failure to pass the mirror test says more about a human lack of imagination than the animal’s lack of a self-concept, as de Waal argues in his book, Are We Smart Enough to Know How Smart Animals Are? Because, think about it: Why should a mirror mean anything to most animals? Some newer takes on the mirror self-recognition test are suggesting that self-awareness is less of a yes-or-no question, and instead exists on something like a spectrum — and some of the most creative work being done in this area happens to involve dogs.

The mirror test is said to have been inspired by an offhand observation jotted down by Charles Darwin as he watched a London Zoo orangutan named Jenny gaze at her own reflection in a mirror; the experience seemed to leave her “astonished beyond measure.” In the 1970s, Tulane University psychologist named Gordon Gallup Jr. came across these notes, and from it devised his mirror self-recognition test — or, what he called the “first experimental demonstration of a self-concept in a subhuman form.”

The classic version goes like this: The experimenter places some colored but odorless dye somewhere conspicuous on the animal’s body, and then places a mirror in front of the animal. If the creature picks at the dye in some way, this is supposed to suggest self-recognition — that the animal knows that this is his own reflection and also knows something is off about the way he looks. As science writer Virginia Morell phrased it in her book Animal Wise, an animal that passes the test reacts to the unfamiliar mark “just as you or I would scrape at a blog of mustard we notice on our lapel when looking in a mirror.”

The link to self-awareness borrows from human developmental psychology. “In human development, when humans start to recognize themselves in the mirror at the age of about 2, that’s also the age in which they become very self-conscious,” de Waal explained. They start to show so-called self-conscious emotions, like pride or shame or embarrassment; they also begin to use personal pronouns, such as I and me. Recognizing yourself in the mirror, the theory goes, signals that you are at least beginning to grasp your you-ness.

At least, that’s what it means for humans, and perhaps for other animals with brains like ours, which are highly attuned to social intelligence. But, then again, not every mind on Earth works exactly like a human mind. The human experience of the world is informed largely by what we see; for other animals, though, other senses are far more important. To return to our canine pals: Dogs have okay vision (they may not see colors as vividly as humans, but they are not color-blind), but they don’t understand the world through their eyes — they understand the world through their noses. Which is why, a few years ago, one rather inventive researcher created a twist on the mirror test, one that would make sense for dogs and their superior sense of smell.

The experiment has come to be known as the Yellow Snow Study. It is as gross as it sounds. About 15 years ago, an animal cognition researcher named Marc Bekoff was dissatisfied with the notion that a dog’s failure to pass the mirror test meant that the animal lacked a self-concept. And so he devised a new test for his dog, Jethro. “Over the course of about five winters, whenever I walked my dog, Jethro, I would put on gloves and I would see where he peed, and I would move it. And then I would move the pee of other dogs,” said Bekoff, now a professor emeritus of ecology and evolutionary biology at the University of Colorado, Boulder. The idea was that if Jethro showed less interest in his own urine, that might be a sign of self-recognition in an olfactory creature. This is exactly what happened, and Bekoff later published his findings in 2001 in Behavioural Processes. “It was really simple to do,” Bekoff said. “Though people thought I was weird.”

A problem with the Jethro study, though, was that it involved only Jethro. And so earlier this year, Alexandra Horowitz — who studies animal cognition at Barnard and runs its Dog Cognition Lab — ran a version of the yellow snow test in her lab. In it, about three dozen dogs were presented with canisters that held either their own urine, or the urine of another dog, or a sample of their own urine to which the experimenters had added an additional scent. Horowitz is still analyzing her findings, but early results are suggesting that, like Jethro, the dogs in her study were least interested in their own scent — though Horowitz is not so sure what, exactly, this can tell us about their self-concept. “I don’t think it’s a self-awareness test, exactly,” she said. “But it does say something about identity.”

It also speaks to the necessity of finding new ways of testing or observing the concept of self-recognition in animals, ones that are tailored to the individual species’ interaction with their worlds. And some scientists are already taking on the challenge posed by de Waal in his book in some exciting ways. Octopuses, for example, have sticky suckers on their eight long arms that tend to attach to any surface they come in contact with — any surface, that is, except for their own skin, which could be the way an octopus demonstrates self-recognition. Or consider bats, which live in caves with thousands of other bats. “They fly out of these caves in the dark, and so they need to do echolocation while there are thousands of other bats doing echolocation at the same time,” de Waal said. “So all these shrieks are coming out, and they need to pick out the sound of their own.” This, too, could be a form of self-recognition.

To return to our original question, though: What does a dog see in the mirror? Judging from their behavior, they likely perceive from their own reflection an unsettlingly scent-less image of a dog, which may be why some dogs try to fight it, and others simply seem to dismiss or ignore it. What that — along with the more, uh, creative, urine-centric versions of the mirror test — actually means about a dog’s sense of self is less clear. But it does suggest that the idea of self-awareness may be a “spectrum,” with some creatures having none or very little of it, and others having rich autobiographical memories and experiences. “What does a dog know about themselves? I think they have this sense of mine-ness and this sense of body-ness. ‘This is my body; this is my bone,’” Bekoff said. “I don’t know that they have a rich sense of self like you and I might have.”

And that in itself is probably a form of self-awareness. “If you define self-awareness a bit more broadly, I would say every species has to have some of it,” de Waal said. “It’s a vague concept, and the boundaries are also probably vague.” Zooming out, the failures of the mirror test are a small example of the larger failures de Waal sees in the traditional ways of testing all sorts of intelligence in animals. “That’s really the challenge of our field,” de Waal added. “To find ways of testing animals on their terms, and their strengths.” This is about them, in other words— not about us.

What Does a Dog See in a Mirror?