Inside of a Dog: What Dogs See, Smell, and Know (26 page)

The best interpretation is that the dogs' performance on the task makes a methodological point. There may be other cues the dogs are using to make their decisions that are, to them, just as strong as the presence or absence of the guesser is to us. Consider, for instance, that all humans are on the whole highly knowledgeable about the sources of food, from a dog's point of view. We are regularly around food, we smell like food, we open and close a cold box filled with food all day long, and sometimes we even have food dribbling out of our pockets. This is such a well-learned feature of us that it might be hard to overturn on the basis of a few trials one afternoon. This hypothesis is borne out by the fact that the dogs did use the
people
to make their decision: they never chose a third box, unselected by either the guesser or the knower.

However we interpret the results, though, the dogs are not going out of their way to prove to us that they have a theory of mind. Of course, one of the difficulties of designing experiments for any animal is that, as the procedure grows more complicated in order to test for a very specific skill, it risks becoming an exceedingly strange scenario for the animal. One might suggest that massive confusion on the part of the subjects is not unreasonable. They are often thrust into situations that are bizarre: that are, in fact, intentionally unlike anything they've seen before. People appear with buckets over their heads; trials go on endlessly; it is in every way not normal. Dogs nonetheless sometimes manage to perform well at the tasks in front of them.

Still, their natural behavior—in a natural setting—is a better indication. What do dogs do without the peculiarities of baited and locked boxes and uncooperative humans to puzzle over? Their most representative behavior will appear in dealing naturally with other dogs or with humans. If it is socially helpful for a dog to consider what other dogs are thinking, the ability to do so may have evolved—and may still be visible in social interactions. This is why I spent a year watching dogs play: playing in living rooms and veterinary offices, down hallways and pathways, on beaches and in parks.

Pump appears in the corners of all the videos: in one, she hops nimbly to avoid collision with a dog approaching too fast—then pursues him as he rushes out of the frame. In another, she lies prone with another dog, feigning bites with open mouths. In a third she tries and fails to join two dogs in play; as they run off she is left wagging alone in the camera's eye.

I should correct myself: I was
lucky enough
to spend a year watching dogs play. What is called, appropriately, "rough-and-tumble" play between two competent, athletic dogs is a gymnastic marvel to witness. The playing dogs seem to give a perfunctory greeting to each other before they suddenly mutually attack, teeth bared; tumbling together in precarious free fall; jumping on and over each other; bodies bent and tangled. When they stop, suddenly, at a noise nearby, they may be the pictures of quiet. It takes only a look or a paw raised in the air to engage in their shared havoc again.

Play might seem just like
that thing dogs do,
but it has a very particular scientific definition. Animal play, science intones, is a voluntary activity incorporating exaggerated, repeated behaviors, extended or truncated in duration, varied in fortitude, and atypically combined; and using action patterns that have identifiable, more functional, roles in other contexts. We don't just define play this way to take the pleasure out of it: we define it to reliably recognize it. Play also has all the attributes of a good social interaction: coordination, turn taking, and, if necessary, self-handicapping—playing at the level of one's play partner. Each partner takes the abilities and behavior of the other into account.

The function of animal play is a bit of a puzzle. Most animal behaviors are described by how they function to improve the survival of the individual or species. The search for a function of play is paradoxical, as it looks like behavior which is clearly function
less:
at the end of play, no food has been gained, no territory secured, no mate wooed. Instead two dogs pantingly collapse on the ground and wag their tongues at each other. One might thus suggest that the function is
to have
fun
—but this is frowned upon as a true function, because the risks are too great. Play takes a lot of energy, can cause injuries, and, in the wild, increases an animal's danger of predation. Play-fighting can escalate into true fights, causing not just injury but social upheaval. Its riskiness makes the case for a real, undiscovered function of play even more compelling: it must be terribly useful to play, if this behavior survived the evolutionary process. It might serve as practice: a context in which to hone physical and social skills. Strangely, though, studies have shown that play is not essential to adult proficiency at the skills practiced in play. Maybe play serves as training for unexpected events. It does seem that volatile and unpredictable play is deliberately sought. In humans, play is part of normal development—socially, physically, and cognitively. In dogs, it may be the result of having spare energy and time—and owners who live vicariously through their dogs' tumblings.

Play among dogs is particularly interesting because they play more than other canids, including wolves. And they play into adulthood, which is rare for most playing animals, including humans. Although we ritualize play into team sports and solo video game marathons, as sober adults we rarely spontaneously blindside and tackle our friends, tag them and run, or make faces at each other. The hobbling, slow-moving fifteen-year-old dog on the block looks warily at the enthusiasm of young puppies approaching him, but even he occasionally play-slaps and bites at a younger dog's legs in play.

In my study of dog play I shadowed dogs around with a video camera rolling, and controlled my own delighted laughter at their fun long enough to record bouts of play, from a few seconds to many minutes long. After a few hours of this the fun stopped, the dogs would get packed into the backs of cars, and I would walk home, reflecting on the day. I'd sit down in front of my computer and play back the videos, at an extremely slow rate: slowed enough to see each frame—thirty of which fill a second—individually. Only at this speed could I really see what had happened in front of me. What I saw was not a repeat of the scene I'd witnessed at the park. At this speed I could see the mutual nods that preceded a chase. I saw the head-jockeying, open-mouth volleys that blurred into unrecognizability in real time. I could count how many bites it takes, over the course of two seconds, before a bitten dog responds; I could count how many seconds it takes for a paused bout to resume.

And, most important, I could look to see what behaviors dogs do, and when. Watching the play deconstructed into these subsecond moments enabled me to record a long catalog of the behaviors of each dog: a transcript of the play. I also noted their postures, their proximity to one another, and which way they were looking at every moment. Then, so deconstructed, the play could be reconstructed to see what behaviors match what postures.

In particular, I was interested in two kinds of behaviors: play signals and attention-getters. Attention-getters, as we've seen, are obvious things: they serve to get attention. Specifically, they are acts that alter the sensory experience of someone else—someone whose attention you're keen to have on you. They can be an interruption of the visual field, as when Pump suddenly puts her head between me and the book I'm holding. They can interrupt the auditory environment: a car's honk is so intended, and dogs' barks are so as well. If these methods fail, attention can be gotten by interacting physically: a hand on the shoulder; a paw on the lap; or, between dogs, a bump with the hip or a light bite on the rump. Clearly, many things we do are in some way attention-getting, but not every behavior is equally good at the task.

Play signals, the other behaviors, are requests for play or announcements of interest in playing: they could be translated as saying something like
Let's play
or
I
want to play
or even
Ready? because I'm about to play with you.
What the specific words are is not as important as their functional effect: play signals are reliably used to begin and to continue play with others. They are a social requirement, not just a social nicety. Dogs typically play together rambunctiously and at a breakneck pace. Since they are doing all manner of actions that could easily be misinterpreted—biting each other on the face, mounting from behind or fore, tackling the legs out from under another dog—the playfulness of their actions has to be manifest.* If you fail to signal before biting, jumping on, hip-slamming, and standing over your playmate, you are not in fact playing; you are assaulting him. A bout wherein only one participant thinks it's play is no longer playful. All dog owners who walk their dogs among others know what then happens: a play bout becomes an attack. Without the play signal, a bite is a bite, worthy of rancor or retribution. With it, a bite is just part of the game.

Nearly every play bout begins with one of these signals. The quintessential signal is the
play
bow,
in which the dog's body genuflects in front of a desired play partner. A dog bent on his forelegs, mouth open and relaxed, with his rump in the air and tail high and wagging is pulling out all the stops to induce someone to play. Even tailless, you can mimic this pose yourself; expect a response in kind, a friendly nip, or at least a second look. Two dogs who are regular playmates may use a bow shorthand: familiarity allows abbreviations in formality, just as between human acquaintances. Just as
How do you do?
became
Howdy,
the play bow can be shortened into the aforementioned
play slap,
the front legs clapping the ground at the beginning of the bow; the
open mouth display,
the mouth opened but without the teeth bared; or the
head
bow,
a bobbing of the head with opened mouth. Even panting in quick bursts can be a signal to play.

It is how dogs might use these play-signal and attention-getting behaviors together that could reveal or refute that dogs have a theory of mind. In just the way the false belief task shows that some children are thinking about what other people know, and some are not, one's use of attention in communicating is meaningful. The key question I asked of my data of playing dogs was this: Did they communicate, using play signals, intentionally—with attention to the attention of their audience? And did they use attention-getters when they didn't have their play partner's attention? Just how were those bumps, barks, and bows of play used?

It's hard to give a good account of what's happened in a bout of play you have just watched. Sure, I could create a very simplistic story line between two dog protagonists—
Bailey and Darcy ran around together … Darcy chased Bailey and
barked … they both bit at each other's faces … then they split
—but it glosses over the details, such as how often Darcy and Bailey self-handicapped, intentionally throwing themselves on the ground on their backs to be bitten, or using less force in a bite than they could. Whether they took turns in biting and being-bitten; chasing and being-chased. And, most critically, whether they signaled to each other when the signal could be seen and responded to—with play or by hightailing it out of there. For this, you need to look at the moments between the seconds.