Mute

Pump sits close to me and quietly pants, gazing at me: she wants something. On our walks she tells me when we've gone far enough and she is ready to go back: she hops up, pivots on her rear legs, then beelines back from where we came. I turn on the bathwater, turn to her with a smile, and her tail drops and wags low, her ears flattening on her head. All this talking and yet no talking at all.

There is a certain poignancy in describing animals as our "dumb friends"; in noting the "blank bewilderment" of a dog; in nodding at their "uncommunicating muteness." These are familiar ways of talking about dogs, who never respond in kind as we speak to them. No small amount of dogs' winsomeness is the empathy that we can attribute to them as they silently contemplate us. Still, these characterizations, while evocative, seem to me to be outright flawed in two ways. First, it is not the animals who desire to speak and cannot, I suspect; it is that we desire them to talk and cannot effect it. Second, most animals, and dogs in particular, are neither blank of expression nor in fact mute. Dogs, like wolves, communicate with their eyes, ears, tail, and very posture. Far from pleasantly silent, they squeal, growl, grunt, yelp, moan, whine, whimper, bark, yawn, and howl. And that's just in the first few weeks.

Dogs talk. They communicate; they declare; they express themselves. This comes as no surprise; what is surprising is how often they are communicating, and in how many ways. They talk to each other, they talk to you, and they talk to noises on the other side of closed doors or hidden in high grasses. This gregariousness is familiar to us: having a large roster of communications is consistent with being social, as humans are. Those canids such as foxes, who do not live in a social group, appear to have a much more limited range of things to say. Even the kinds of sounds foxes make are indicative of their more solitary nature: they make sounds that travel well over long distances. Dogs' staunch unmuteness is expressed through making sounds bellowed and whispered. Vocalizations, scent, stance, and facial expression each function to communicate to other dogs and, if we know how to listen, to us.

OUT LOUD

Two human beings stroll through a park chatting. They move with ease from commenting on the warmth of the air, to the nature of humans in positions of power, to expressions of mutual adoration, to reflections on past expressions of mutual adoration, to admonishment to observe the tree straight ahead. They do this primarily by making small, strange contortions of the shape of the cavities of their mouths, the placement of their tongues, by pushing air through the vocal tract and squeezing or widening their lips. Theirs is not the only communication going on. Over the course of a walk, the dogs by their sides may scold one another, confirm friendships, court each other, declare dominance, rebuff advances, claim ownership of a stick, or assert allegiance to their person. Dogs, like so many non-human animals, have evolved innumerable, non-language-driven methods to communicate with one another. Human facility at communication is unquestionable. We converse with an elaborate, symbol-driven language, quite unlike anything seen in other animals. But we sometimes forget that even non-language-using creatures might be talking up a storm.

What animals have are whole systems of behavior that get information from a sender (speaker) to a recipient (listener). That is all that is needed to call something a communication. It needn't be important, relevant, or even interesting information, but between animals it often is. Communication is only sometimes within our range of hearing, or even vocal: it is often made through body language—using limbs, head, eyes, tails, or the entire body—or even through such surprising forms as changing color, urinating and defecating, or making oneself larger or smaller.

We can spot a communication by noticing if, after one animal makes a noise or does an action, another responds to it by changing its behavior. Information has been imparted. What we'll miss, since we don't know the language of, say, spiders or sloths (though there are currently researchers trying to learn these communication systems), is those utterances that fall on deaf ears. Still, animals are constant gabbers. The discoveries of natural science over the last one hundred years have shown us the variety of guises in which this gabbing can appear. Birds twitter, peep, and sing songs—so do humpback whales. Bats emit high-frequency clicks; elephants, low-frequency rumbles. The wiggling dance of a honeybee communicates the direction, quality, and distance to food; a monkey's yawn conveys a threat. A firefly's flashes indicate his species; a poison-dart frog's coloration identifies his toxicity.

The kind we notice first is the one that most closely matches our own language: communication out loud.

DOG-EARED

Thunder outside. Pump's ears, velvet equilateral triangles that fold perfectly along the side of her head, prick into long isosceles. Head up, eyes to the window, she identifies the sound: a storm, a frightful thing. Her ears pivot back, flattened along her skull as if to hold them shut by their own force. I coo to her consolingly and watch her ears for feedback. The tips soften but she relaxes only slightly, still holding them tight against the roar.

Without prominent ears ourselves, we can envy dogs' proud ears. They come in a dazzling array of equally adorable variants: extremely long and lobular; small, soft, and perked; folding gracefully alongside the face. Dogs' ears may be mobile or rigid, triangular or rounded, floppy or upright. In most dogs, the pinna—the outer, visible part of the ear—rotates to better open a channel from the sound source to the inner ear. The practice of cropping ears, severing the pinnae to make floppy ears stand upright, long mandated in many breed standards, is becoming less popular. This designing of dogs, sometimes defended as reducing infections, has unknown consequences in auditory sensitivity.

By natural design, dogs' ears have evolved to hear certain kinds of sounds. Happily, that set of sounds overlaps with those we can hear and produce: if we utter it, it will at least hit the eardrum of a nearby dog. Our auditory range is from 20 hertz to 20 kilohertz: from the lowest pitch on the longest organ pipe to an impossibly squeaky squeak.* We spend most of our time straining to understand sounds between 100 hertz and 1 kilohertz, the range of any interesting speech going on in the vicinity. Dogs hear most of what we hear and then some. They can detect sounds up to 45 kilohertz, much higher than the hair cells of our ears bother to bend to. Hence the power of the dog whistle, a seemingly magical device that makes no apparent sound and yet perks the ears of dogs for blocks around. We call this sound "ultrasonic," since it's beyond our ken, but it is within the sonic range for many animals in our local environment. Don't think for a moment that apart from the occasional dog whistle, the world is quiet for dogs up at those high registers. Even a typical room is pulsing with high frequencies, detectable by dogs constantly. Think your bedroom is quiet when you rise in the morning? The crystal resonator used in digital alarm clocks emits a never-ending alarm of high-frequency pulses audible to canine ears. Dogs can hear the navigational chirping of rats behind your walls and the bodily vibrations of termites within your walls. That compact fluorescent light you installed to save energy? You may not hear the hum, but your dog probably can.

The range of pitches we are most intent on are those used in speech. Dogs hear all sounds of speech, and are nearly as good as we are at detecting a change of pitch—relevant, say, for understanding statements, which end in a low pitch, versus questions, which in English end in a raised pitch: "Do you want to go for a walk(?)" With the question mark, this sentence is exciting to a dog with experience going on walks with humans. Without it, it is simply noise. Imagine the confusion generated by the recent growth of "up-talking," speech that ends every sentence with the sound of a question?

If dogs understand the stress and tones—the prosody—of speech, does this hint that they understand language? This is a natural but vexed question. Since language use is one of the most glaring differences between the human animal and all other animals, it has been proposed as the ultimate, incomparable criterion for intelligence. This raises serious hackles in some animal researchers (not thought of as a hackled species, ironically), who have set about trying to demonstrate what linguistic ability animals have. Even those researchers who may agree that language is necessary for intelligence have nonetheless added reams of results to the growing pile of evidence of linguistic ability in non-human animals. All parties agree, though, that there has been no discovery of a humanlike language—a corpus of infinitely combinable words that often carry many definitions, with rules for combining words into meaningful sentences—in animals.

This is not to say that animals might not understand some of our language use, even if they don't produce it themselves. There are, for instance, many examples of animals taking advantage of the communicative system of nearby unrelated animal species. Monkeys can make use of nearby birds' warning calls of a nearby predator to themselves take protective action. Even an animal who deceives another animal by mimicry—which some snakes, moths, and even flies can do—is in some way using another species's language.

The research with dogs suggests that they do understand language—to a limited degree. On the one hand, to say that dogs understand words is a misnomer. Words exist in a language, which itself is product of a culture; dogs are participants in that culture on a very different level. Their framework for understanding the application of the word is entirely different. There is, no doubt, more to the words of their world than Gary Larson's Far Side comics suggest: eat, walk, and fetch. But he is on to something, insofar as these are organizing elements of their interaction with us: we circumscribe the dog's world to a small set of activities. Working dogs seem miraculously responsive and focused compared to city pets. It is not that they are innately more responsive or focused, but that their owners have added to their vocabularies types of things to do.

One component in understanding a word is the ability to discriminate it from other words. Given their sensitivity to the prosody of speech, dogs do not always excel at this. Try asking your dog on one morning to go for a walk; on the next, ask if your dog wants to snow forty locks in the same voice. If everything else remains the same, you'll probably get the same, affirmative reaction. The very first sounds of an utterance seem to be important to dog perception, though, so changing the swallowed consonants for articulated ones and the long vowels for short ones—ma for a polk?—might prompt the confusion merited by this gibberish. Of course humans read meaning into prosody, too. English does not give the prosody of speech syntactical leverage but it is still part of how we interpret "what has just been said."

If we were more sensitive to the sound of what we say to dogs, we might get better responses from them. High-pitched sounds mean something different than low sounds; rising sounds contrast with falling sounds. It is not accidental that we find ourselves cooing to an infant in silly, giddy tones (called motherese)—and might greet a wagging dog with similar baby talk. Infants can hear other speech sounds, but they are more interested in motherese. Dogs, too, respond with alacrity to baby talk—partially because it distinguishes speech that is directed at them from the rest of the continuous yammering above their heads. Moreover, they will come more easily to high-pitched and repeated call requests than to those at a lower pitch. What is the ecology behind this? High-pitched sounds are naturally interesting to dogs: they might indicate the excitement of a tussle or the shrieking of nearby injured prey. If a dog fails to respond to your reasonable suggestion that he come right now, resist the urge to lower and sharpen your tone. It indicates your frame of mind—and the punishment that might ensue for his prior uncooperativeness. Correspondingly, it is easier to get a dog to sit on command to a longer, descending tone rather than repeated, rising notes. Such a tone might be more likely to induce relaxation, or preparation for the next command from their talky human.

There is one celebrated dog whose word usage is exceptional. Rico, a border collie in Germany, can identify over two hundred toys by name. Given an enormous heap of all the toys and balls he has ever seen, he can reliably pull out and retrieve the one his owner requests. Now, putting aside why a dog might need two hundred toys, this ability is impressive. Children are hard-pressed to do the same task (and are only sometimes helpful in bringing things back). Even better, Rico can quickly learn a name for a new object, by process of elimination. Experimenters put a novel toy among familiar ones and asked him, using a word he had never heard before, to retrieve it. Go get the snark, Rico. One would be sympathetic if he looked bemused, and wandered back with a favorite toy in his chops. Instead, though, Rico reliably picked out the new toy: naming it.

Rico was not using language, of course, in the way we, or even young children, do. One can debate how much he was understanding, or if he was even doing anything other than showing a preference for the new object. On the other hand, he was showing an astute ability to satisfy the humans making various sounds by picking up the referents of those sounds. His ability might not indicate that all dogs are so able: Rico might be an unusually skilled word user*

—and he is definitely unusually motivated by the praise received on retrieving the right toy. Still, even if he were the only dog who does this, it indicates that the dog's cognitive equipment is good enough to understand language in the right context.

It is not only the express content or sound of speech that carries meaning. Being a competent language user means understanding the pragmatics of usage: how the means, form, and context of what you say also affect the meaning of what you say. Paul Grice, a twentieth-century philosopher, famously described various "conversational maxims," known to us implicitly, that regulate language use. Their use marks you as a cooperative speaker; even their express violation is often meaningful. They include the charming maxim of relation (be relevant), the maxim of manner (be brief and clear), and maxims of quality (tell the truth) and quantity (say only as much as you need to).

On a good day, dogs mind all of Grice's maxims. Consider a dog who espies a roguish-looking fellow down the street. The dog may bark (relevant: the guy is roguish-looking) sharply (quite unambiguous), but only as long as the fellow is around (so the warning bark is currently true), and not more than a few times (relatively pithy). While dogs hardly qualify as competent language users, it is notably not because of their violation of the pragmatics of communication. It is only the smallness of their vocabulary and restricted use of words in combination that disqualifies them.

Many owners lament that, by contrast to Rico, their dogs are not terrific listeners—despite their broad range of audition. To be fair, canids do not rely on hearing as their primary sense. Relative to even our hearing, their ability to pinpoint where a sound is coming from is imprecise. They hear sounds unmoored from their origins. And just like us, they must bring attention to a noise to hear it best—first apparent in the familiar tilt of the head, to direct the ears slightly toward the sound source, or in radar-dish adjustments of the pinnae. Instead of being used to "see" the source of the sound, their auditory sense seems to serve an ancillary function: helping dogs find the general direction of a sound, at which point they can turn on a more acute sense, like olfaction or even vision, to investigate further.

Dogs themselves make a variety of sounds across a range of pitches or differing only by subtle alterations in tempo or frequency. They are downright noisy.

THE OPPOSITE OF MUTE

Her slow, light panting, mouth open partway, tongue purple and wet and perfect. Pump's panting was a conversation in and of itself—I always felt talked-to when she panted at me.

The cacophony of a packed dog run seems at first pass to be an undifferentiated racket. With closer attention, though, one can distinguish shouts from cries; yelps from barks; and play barks from threatening barks. Dogs make sounds both intentionally and inadvertently. Both kinds may hold information, the minimum requirement to call an aural disturbance a "communication" rather than simply "noise." What is interesting for scientists is determining the meaning of that information. Given the way dogs wield these noises, there is no doubt that they have different meanings.

Countless hours of researchers' lives spent listening to animals shout, coo, click, groan, and scream has led to the discovery of some universal features of sound signals. They either express something about the world—a discovery, a danger—or something about the signalers themselves—their identity, sexual status, rank, membership in a group, fear, or pleasure. They effect a change in others: they may decrease social distance between the signaler and those around him, calling someone closer; or increase social distance, frightening someone away. In addition, sounds may serve to cohere a group (in defense from a predator or intruder, for instance) or they may elicit maternal or sexual affiliation. Ultimately, all these purposes for making sounds make evolutionary sense: they aid the animal in securing its survival or the survival of its relatives.

What, then, are dogs saying, and how are they saying it? The what is answered by looking at the context of making a sound. The context includes not just the sounds around it but also the means: a screamed word winds up meaning something different than one intoned with a sultry whisper. A sound a dog makes while wagging merrily means something different than the same sound delivered through bared teeth.

The meaning of an uttered sound can also be identified by looking at what those who hear it do. Although human responses to an utterance (say, How are you?) may range from the appropriate (I'm well, thanks) to the seeming non sequitur (Yes, we have no bananas), there is reason to believe that dogs, and all non-human animals, respond ingenuously. In many cases, a sound will have a reliable effect on those in the vicinity: think Fire! or Free money!

The how of sound signaling is simple with dogs. Most of the sounds dogs make are oral: using or coming out of the mouth. At least, these are the sounds that we know about. These vocal sounds might be voiced, with vibration in the larynx—the airway used for breathing—or may be expiratory—part of an exhalation. Others are entirely unvoiced but use the mouth, such as the mechanical sound of tooth-snapping. Vocal sounds vary from one another along four easily audible dimensions. They vary in pitch (frequency): whines are nearly always high-pitched, while growls are low-pitched. Try and squeal out a growl and it becomes something else. They vary in duration: some are uttered once, quickly, lasting less than half a second; others are protracted sounds or are repeated again and again. Sounds vary in their shape: some are pure tones while others are more fractured, fluctuating or rising and falling. A howl has little variation for long periods, while barks are noisy, changeable sounds. Finally, they vary in loudness or intensity. Moans don't come in loud and yelps don't come in a whisper.

WHIMPERS, GROWLS, SQUEAKS, AND CHUCKLES

She sees I'm almost ready. With her head fixed on the ground between her paws, Pump follows me with her eyes as I cross the room gathering my bag, a book, my keys. I scratch her around her ears in consolation and break for the door. She lifts her head and makes a sound: a plaintive yelp. I freeze. A look back and she hurries over, wagging. Okay, then; I guess she'll come with me.

The paradigmatic dog sound is the bark, but barks do not form the preponderance of most dogs' daily noisemaking, which includes high and low sounds, incidental sounds, even howls and chuckles. High-frequency sounds—cries, squeals, whines, whimpers, yelps, and screams—occur when the dog is in sudden pain or needs attention. These are some of the first sounds a puppy will make, which clues us in to their meaning: they tend to attract attention of the mother. A yelp might come out of a puppy who was just stepped on, or who has wandered off. Deaf and blind, it is easier for mom to find her pup than vice versa. Having been reunited, some continue to yelp, winding down off their crying jag, when carried by their mothers. Yelps are different than screams, which in wolves prompt the mother to groom the pup, providing the contact that is necessary for normal development. Cries and squeals may be ignored by the mother, and so a particular squeal may be a less specifically meaningful utterance and instead a general-purpose sound used simply to see how others respond.

Low moans or grunts are also very common in puppies, and seem not to be signs of pain but rather a kind of dog purr. There are snuffling moans and sighing moans—what some call "contentment grunts," and they all seem to mean about the same thing. Pups moan when they are in close contact with littermates, their mother, or a well-known human caretaker. The sound might be simply a result of heavy, slow breathing, which indicates it might not be intentionally produced: there is no evidence that dogs moan on purpose (neither is there evidence that they do not; neither has been proven). But whether they do or not, moans probably function to affirm the bond between family members, whether heard as a low vibration or felt through skin-to-skin contact.

The rumble of a growl and the steady ominous snarl, you won't need to be told, are aggressive sounds. Puppies do not tend to produce them, as puppies do not tend to initiate aggression. Part of what makes them aggressive is their low pitch: they are the kind of sounds that would come out of a large animal, rather than the high-pitched squeals of a small one. In an antagonistic (what in biology is called agonistic) encounter with another animal, a dog wants to appear to be the bigger, more powerful creature—so he makes a big-dog sound. By making higher-pitched sounds, an animal sounds, simply, smaller: a friendly or appeasing noise, by contrast. Though aggressive in intent, growls are still social sounds, not just utterances produced when a dog feels fear or anger: for the most part, dogs do not growl at inanimate objects,* or even at animate objects that aren't faced or directed toward them. They are also subtler than we think: distinct growls, from rumble to nearly roaring, are used in different contexts. The growl of tug-of-war may sound fearsome, but it is nothing like the possessive warning snarled over a treasured bone. Play these growls back over a speaker set up right in front of a desirable bone and dogs in the vicinity will avoid the bone—even with no dog in sight. But if the speaker growls only play or stranger growls, nearby dogs go ahead and grab the unguarded bone.

Incidental sounds of dogs are sometimes produced so reliably in certain contexts that they have become effectively communicative. The play slap, an audible landing on the two forefeet at once, is an inevitable part of play. It conveys sufficient exuberance that it can be used by itself to ask a dog to play with you. Some dogs chatter their teeth in anxious excitement, and the clicking of teeth serves as a warning that the dog is wary. An exaggerated shriek on being nosed or bitten roughly in play can even become a ritualized deception, a way to get out of a social interaction that is making the dog uncertain. The snuffling sound created when reaching the head vertically up and sniffing for food around a human mouth can become not just a search for food but also a request for food. Even the noisy breathing created by lying so close as to have the nose pressed against another body comes to indicate a state of contented relaxation.

If you live with a hound, you are familiar with the howl. From a staccato baying to a mournful wail, howling in dogs seems to be a behavior left over from their ancestors, living in social packs. Wolves howl when separated from the group, and also when setting out with the group for a hunt or in reunion afterward. A howl when alone is a communication seeking company; howling together may be simply a rallying cry or celebration of the group. It has a contagious component, leading others in the vicinity to pick it up in an impromptu fugue. We do not know what they are saying, to each other or to the moon.

The most social of human sounds is the cackling laugh rumbling across the room. Do dogs laugh? Well, only when something is terrific fun. Yes, dogs have what has been called a laugh. It is not identical to human laughter, the spontaneous sounds spit out in response to something funny, surprising, or even frightening. Nor is it as variable as the cackles, giggles, and twitters that we produce. The dog laugh is a breathy exhalation that sounds like an excited burst of panting. We could call it social panting: it is a pant only heard when dogs are playing or trying to get someone to play with them. Dogs don't seem to laugh to themselves, off sitting in the corner of the room, recollecting how that tawny dog in the park outsmarted her human this morning. Instead, dogs laugh when interacting socially. If you have played with your dog, you have probably heard it. In fact, doing your own social panting toward a dog is one of the most effective ways to elicit play.

Just as our laughs are often inadvertent, reflexive responses, so may dog laughs be: simply the kind of panting that results when you're throwing your body around in play. Though it might not be under the control of the dog, social panting does seem to be a sign of enjoyment. And it may induce enjoyment—or at least alleviate stress—in others: playing a recording of the sounds of dog laughter at animal shelters has been found to reduce barking, pacing, and other signs of stress in the dogs housed there. Whether mirth feels like what it does in humans is yet to be studied.

WOOF

I can remember the first time a bark came out of Pump, when she was maybe three years old. She'd been so quiet until then, and then one day, after spending time with her barky German shepherd friend, a bark popped out of her. It was bark like more than a bark, as though a sound that stood for a bark but wasn't itself the real thing: a well-articulated rurph! accompanied by a little leap off her front legs and a madly wagging tail. She refined this splendid display somewhat through the years, but it always felt like a new dog thing she was trying on.

It is regrettable that barks tend to be such loud affairs. The bark is shouted. While a calm conversation between the two strollers in the park might register about 60 decibels, dog barks begin at 70 decibels and a stream of barks may be punctuated with spikes to 130 decibels. Increases in decibels, the unit of measurement of the loudness of sounds, are exponential: an increase of 10 decibels describes a hundredfold rise in the experience of the strength of a sound. One hundred and thirty decibels is up there with thunderclaps and plane takeoffs. The bark is momentary, but it is a moment of displeasure for our ears. The reason this is regrettable is that there is, most dog researchers agree, much information in those barks. Given the relative scarcity of barking in wolves, some theorize that dogs have developed a more elaborate barking language precisely in order to communicate with humans. If we consider barks as all cut from the same cloth, though, they are likelier to annoy than to communicate.

Researchers might not call barks "annoying," but they call them "chaotic" and "noisy." "Chaotic" is a good description for the variability in the kinds of sounds within each bark; "noisy" means not just disagreeably loud but also having fluctuations in its structure. Barks are loud, and different barks have varying numbers of harmonic components, depending on the context in which the bark is used.

Still, of the sounds dogs make, barks come closest to speech sounds. The dog's bark is, like the phonemes of speech, produced by vibrations in the vocal folds and air flowing along the folds and through the mouth cavity. Perhaps because they are in overlapping frequencies—from 10 hertz to 2 kilohertz—with the sounds of speech, we are inclined to look for speechlike meaning in them. We even name the bark using phonemes from our language: the dog "woofs," "rufs," "arfs," or (although no dog I know says it) "bow-wows." The French hear the dog "ouah-ouah"; Norwegian dogs voff-voff; Italian dogs go bau-bau.

Some ethologists think that barking is not fundamentally communicating anything, though: that it is "ambiguous" and "meaningless." This view is encouraged by the difficulty of deciphering what the meaning of the barks could be, since sometimes dogs bark without an obvious prompt or audience, or continue to bark long after any message therein would have been conveyed. Think of the dog barking continuously, dozens of times in a row, in front of another dog: If there is meaning in that bark, would not one or two repetitions do to convey it?

This strikes at the heart of the trouble in determining the subjective experience of an animal of which you cannot ask questions. Each moment of an animal's behavior is scrutinized for its meaning. Surely few human actions could bear such scrutiny and yield a correct assessment of the human. If you were to videotape me practicing, at home in front of my dog, a speech I am to deliver later in the day, you might well conclude that (a) I believe the dog can understand what I'm saying, or

(b) I am talking to myself. In either case, (c) holds: the noises I am making appear not to be classically communicative—for I do not have an audience who can understand me. Similarly, examples of poor communication by a dog may seem to undermine the notion that dogs communicate at all. But most researchers think that barks do have meaning, albeit one dependent on the context and even on the individual. Barking, especially alarm barking, is one of the clearest distinctions between dogs and other canid species. Wolves bark to convey alarm, but rarely, and they make more of a "woof" sound than anything like the protracted dog barking with which we are familiar. Dogs do not just bark more than wolves; they have developed numerous variations on the theme.

There are a handful of distinguishable barks, used reliably in a handful of distinguishable cases. Dogs bark to get attention, to warn of danger, in fear, as a greeting, in play, or even out of loneliness, anxiety, confusion, distress, or discomfort. The meaning is in the context of their use, but not only in the context: spectrograms of dog barks show that they are mixtures of the tones used in growls, in whimpers, and in yelps. By altering the prevalence of one tone over the others, the bark takes on a different character—a different gist.

Early research into dogs' vocalizations concluded that all dog barking was attention-getting barking. In fact they do attract attention, assuming someone is close enough to hear them. But recent studies have made more subtle discriminations between barks. While in some way all barks come down to some manner of "attention-getting," one might as well say that we speak in order to be heard: true, but incomplete. For instance, when experimenters analyzed the spectrograms of thousands of dog barks during one of three contexts—a stranger ringing the doorbell, being locked outside, or in play—they found three distinct types of barks.

Stranger barks were the lowest in pitch and the harshest: they are nearly spat out. Less variable than the other types, stranger barks are well designed to send a message over a distance, something necessary if caught in a threatening situation alone. They can also be combined into "superbarks," concatenations of barks that together last much longer than the duration of barking in other contexts. The end result is a bark that most human listeners find to be aggressive.

The isolation barks tended to be higher-frequency and more variable: some ranged from loud to soft and back again, some went from high to low. These barks are lobbed into the air one by one, sometimes with great intervals between them. They sound "fearful," people tend to say.

Play barks, too, are high-frequency, but they happen more often one after the other than the isolation barks. They're directed at someone else, unlike isolation barks: at a dog or human playmate. There is considerable individual variation, of course: not every dog barks alike. The stranger bark of a small dog may come out as rar, rar or raoaw, raoaw, while a larger dog emits a capital-r Rumph.

These differences between bark types make evolutionary sense: the lowest sounds are used in threatening situations (again to appear bigger); higher sounds are entreaties—to friends, for companionship—and as such are submissive requests, not warnings. Differences between individual barkers indicate that barks might be used to affirm a dog's identity, or reveal an association with a group (even the group me and the woman at the end of my leash, rather than these dogs I'm frolicking among). And barking together with others may be a form of social cohesion. Barking can be contagious, like the howl: one dog barking might prompt a chorus of barking dogs, all joined in their shared noisiness.

BODY AND TAIL

When we approached people on the street, Pump set all her senses to looking; if she recognized them, her head would lower ever so slightly—looking up coyly as though over reading glasses—and she would wag her tail low. This was quite different from her approach of a dog she was smitten with: all upright, tail high, posture impeccable, wags soldierlike in their rhythm—or a dog friend—a looser,

janglier approach, and even an open-mouthed grab toward their face, or a gentle bump with her hip along their body.

You may be sitting down right now, folded into a comfy chair—or perhaps you're standing, straphanging on a train to work, book scrunched against another commuter's back. Most likely you don't mean anything by your sitting or standing, or when you walk or lie supine: it's just a posture of convenience or comfort. But in other contexts our very posture conveys information. A catcher crouches: he's prepared for a pitch. A parent crouches and opens his arms: he's inviting a child for a hug. Running when someone you know approaches, you suddenly stop and greet them; standing still when someone you know approaches, you suddenly turn and run. There can be meaning simply in the vigor or slouch of your body. For an animal with a limited vocal repertoire, posture is ever more important. And it appears that dogs use specific postures to make very specific statements.

There is a language of the body, formed of phonemes made from rumps, heads, ears, legs, and tails. Dogs know how to translate this language intuitively; I learned it after watching hundreds of hours of dogs interacting with each other. We must look like such stiffs to dogs, who can express everything from playfulness to aggression to amorous intent by changing the shape of their body and its altitude. By contrast, we are inhibited straight-backeds, mostly stationary or traveling forward with little excess movement. Occasionally—heavens—we turn a head or arm flamboyantly to the side.

But man himself cannot express love and humility by external signs, so plainly as does a dog, when with drooping ears, hanging lips, flexuous body, and wagging tail, he meets his beloved master.—Charles Darwin

For dogs, posture can announce aggressive intent or shrinking modesty. To simply stand erect, at full height, with head and ears up, is to announce readiness to engage, and perhaps to be the prime mover in the engagement. Even the hair between the shoulders or at the rump—the hackles—may be standing at attention, serving not just as a visual signal of arousal but also releasing the odor of the skin glands at the base of the hairs. To exaggerate the whole effect, a dog might stand not just up but over another dog, head or paws on his back. That's about as declarative a statement as you can make that you are feeling dominant. The opposite body posture, crouching with head down, ears down, and tail tucked away, is submissive. To lie all the way down and expose the belly is even more so.*

This principle of antithesis—that opposing postures communicate opposing emotions—describes much of the expressive scope of dogs. Facial expressions, most visible in the mouth and ears, mind this principle, too. The mouth sweeps from closed to open and relaxed, to open with lips raised, snout wrinkled, and teeth bared. A dog's "grin," with jaw closed, is submissive; as the mouth is opened, the arousal increases; and if the teeth are exposed, the look gets aggressive. Coming full circle, a wide-open mouth with teeth mostly covered—a yawn—is not a sign of boredom, as often assumed by analogy with our own yawning; instead, it may indicate anxiety, timidity, or stress, and is used by dogs to calm themselves or others. The ears can also go through these gymnastics: they can be pricked, relaxed and down, or folded tightly along the head. Eyeballing another dog directly can be threatening or aggressive; by contrast, looking away is submissive—an attempt to quell one's own anxiety and the other dog's arousal. In other words, in each case there is a range from one extreme to another, representing variation in intensity along an emotional continuum, from relaxed to aroused in fear or alarm.

None of these is a static symbol—or, if it is, its being static is meaningful. Holding an erect posture, motionless, is a quiet way of putting an exclamation point on the posture. It exaggerates the tenseness of the communication. For the most part, postures are taken, and moved through. The tail, especially, is a limb of movement. It is to science's great discredit that no one has done a thorough investigation of the meaning of every wag of the dog's tail.

As a puppy her tail was trim, an arrow of soft black fur. This turned out not to be the destiny of her tail at all: it grew into an incredible banner of a tail, with extravagant feathering that matted and gathered leaves. It was bent at the tip from a disagreement with a car door when she was young. She brandished it when excited or delighted, curved to a sickle with the tip pointing at her back. When lying down, she drummed it happily on the ground at my approach. Her tail registered her exhaustion in a low-hanging straight pose; her disinterest in a nosy dog by tucking between her legs. Most of the time as we walked together, it hung loosely down, curving jauntily toward its tip, and merrily swished to and fro. I loved to approach her slowly, stalking her, and prompt her tail to quiver into wagging.

One of the difficulties in deciphering the language of tails is the great variation in tails among dogs. The flamboyant plumage of a golden retriever contrasts mightily with the tight corkscrew of the pug. Dogs wear tails long and rigid, stumpy and curled, hanging heavily or perpetually perked. The wolf tail is in some ways an average of the various breed tails: it is a long, slightly feathered tail, held naturally slightly down. Early ethologists who did a reckoning of wolf tail postures identified at least thirteen different tail carriages, conveying thirteen distinguishable messages. As per the antithesis thesis, tails held high indicate confidence, self-assertion, or excitement from interest or aggression, while low-hanging tails indicate depression, stress, or anxiety. An erect tail also exposes the anal region, allowing a bold dog to air his odor signature. By contrast, a long tail held so low as to curl back between the legs, closing off the rump, is actively submissive and fearful. When a dog is simply waiting around, his tail is relaxed, hanging low, dropped down but not rigid. A tail gently lifted is a sign of mild interest or alertness.

But it is not as simple as tail height, for the tail is not just held, it is wagged. Wagging does not translate as simple happiness. A high, stiffly wagging tail can be a threat, especially when accompanied by an erect posture. Quickly wagging a dropped, low tail is another sign of submission. This is the tail of the dog who has just been caught finishing off the last of your shoes. The vigor of the wag is roughly indicative of the intensity of the emotions. A neutral tail wagging lightly is interested but tentative. A loose, lively whisking tail accompanies the noseled search for a ball lost in high grasses or an odor trail discovered on the ground. The familiar happy wag is incredibly different from all of these: the tail is held above or out from the body and strongly draws rough arcs in the air behind it. Unmistakable delight. Even non-wagging is meaningful: dogs tend to still their tails when attending carefully to a ball in your hand or waiting for you to tell them what's happening next.

Researchers interested in the brain of the dog accidentally discovered something about the tail of the dog along the way: the dog wags asymmetrically. On average, dogs' wags tend more strongly to the right when they suddenly see their owners—or even anything else of some interest: another person, or a cat. When presented with an unfamiliar dog, dogs still wag—more that tentative wag than the happy wag—but tending to the left. You might not be able to see this in your own dog unless you watch them wag in slow-motion video playback (which I highly recommend)—or unless your dog is one of those who wag less back-and-forth than round-and-round, inclining to the side. Consider yourself lucky to be wagged at with such clear-cut enthusiasm.

Pump does a full-body shake: it starts in her head and rolls down her body, shimmering out through her tail. It is like a punctuation mark that has yet to be discovered. She shakes to end an episode, when she's unsure, and sometimes when she's just ambling along.

The dog uses his body expressively: communication writ through movement. Even the moments between interactions are marked by movement: as when a dog does a full-body shake, his skin twisting over his frame, to indicate his finishing one activity and moving on to another. Not all dogs have hackles that visibly raise with pique, long tails to ostentatiously wag, or ears that raise with interest. The fabulously ropy-furred komondor approaches other dogs with what we must assume is his head, but neither eyes nor ears are visible underneath his long locks. In breeding dogs to have particular looks that we find agreeable, we are limiting their possibilities for communicating. Just as we might expect, but would rather not confront, a dog with a docked tail has, thereby, a docked repertoire of things he can say.

Research looking at the range and rate of signals used by ten physically dissimilar breeds found just this. Comparing the behavior of dogs from the Cavalier King Charles spaniel to the French bulldog to the Siberian husky, there was a clear relationship between the breed appearance and the number of signals they used. Those animals that had been most changed physically in domestication from wolves—the King Charles, at the extreme—sent the fewest signals. These pedomorphic or neotenous dogs, who retain more features of juvenile members of canid species into adulthood, look least like adult wolves. The huskies, which have the most wolflike features and are genetically closer to Canis lupus, do the most wolflike signals.

Given that many bodily signals provide information about one's status, strength, or intent, the necessity for dogs to send these signals is presumably diminished in a world where humans chaperone dogs through life. But the same signals used to convince a dominant animal of benign intent may also be used to communicate information to humans. Walking through the city, I turn a blind corner and nearly step on an unfamiliar dog pulling on a long leash. Seeing me, she crouches, wags her tail furiously between her legs, and licks toward my face. It may have begun as a submissive gesture, but now it is adorable.

INADVERTENT AND INTENT

After sleeping late and enduring the slow pace of my morning rituals, Pump's first move when we get outside never varies. She takes two steps out the door and unceremoniously squats. She crouches deeply, fully committed to the pose, with only her tail—curled high out of the way—pulling her body up. The torrent of urine released (surely record-breaking this time) seems accompanied by a relaxing of the muscles of her face—and by a growing guilt on my own that I made her wait so long. She watches the stream of her own urine wandering by her as it finds cracks in the sidewalk in which to divert itself cleanly.

As much as is being said with a bark, snarl, or wagging tail, vocalization and posture are not the only media of communication for dogs. Neither is a match for the informational possibilities of smells. Urination, as we saw earlier, is the means of odorant communication most conspicuous to us. It might be hard to believe that the release of the bladder is a "communicative act" right up there with a polite conversation between friends or a politician orating before his constituents. At some level, it is like both of these: it is part of normal dog sociality, and it can also be a bellowing self-promotion writ on a hydrant.

You might balk at calling the moist message left high on a lowly fire hydrant the same kind of communication that humans use—and not just because they are talking out of their rumps, not their faces. Crucially, we communicate (most of the time) with intention: rather than yammering out loud to our own left hand, we tend to direct our communications to other people—people who are near enough to hear us, not otherwise distracted, who know the language, and can understand what we're saying. Intention distinguishes communication done with others in mind from the automatic oof! uttered on being hit in the belly, blushing at a compliment, a mosquito's constant buzzing, or the unmindful bit of information imparting done by traffic lights and flags at half-mast.

Urine marking is peeing with intention. The morning's blissful release relieves the strain on the bladder, but most of the time some is held back for later use in marking. Presumably the urine is the same urine: there's no evidence of an independent channel or means by which to modify the odor they are emitting. But marking behavior differs in a few key ways. First, in most adult males, and some gender-bending females, marking is characterized by a prominently raised leg. There are individual and contextual variations on the so-called "raised-leg display," from a modest retraction of the rear leg up toward the body, to hoisting the leg up above the hip points, above vertical—certainly also a visual display for any other dogs in the vicinity. Both allow for a directional flow of urine, aimed to land at a conspicuous site. (One can squat and mark, too, although it is a quieter affair, perhaps for messages better whispered than shouted.)

Second, the bladder is not emptied when marking; urine is doled out a little at a time, allowing for greater distribution of scent over the course of the dog's travels. If you've left your dog indoors long enough that he races outside to squat, this urgency might preempt his ability to cache some urine for later marking. Thus the fruitless raised-leg displays you may witness, waving dryly at bushes, lampposts, and trash cans.

Finally, dogs usually urine mark only after spending some time sniffing the area. This is what elevates the odor exchange from Lorenz's notion of flag planting to a kind of conversation. Researchers keeping careful tally of dogs' marking behavior over time found that who has marked before them, the time of year, and who is nearby all affect where and when they mark.

Interestingly, these message bouquets aren't left indiscriminately: not every surface is marked. Watch a dog sniff his way down the street: he will sniff more locations than he will squirt. This indicates that not every message is the same—and the message this dog will leave may be intended for certain audiences only. Countermarking—covering old urine with new—is a common behavior of male dogs, when the old urine is that of less dominant male dogs. Everyone's marking increases when there is a new dog around.

If it is not territorial, what is the message in the mark? The first hint is that puppies don't urine mark: the communication must have to do with adult concerns. From the position of the anal glands and the compounds in the urine, we know that they are at least saying something about who they are: their odor is their identity. This is a fine message, but it is probably fairly unintentional. I may communicate something about who I am by merely walking into a room and being seen, but the very fact of my person is not a continuous, intentional communication about my identity (except when I was a kid and dressed to be seen).

What does look intentional in this communication is that dogs don't bother to say it if there's no one else around. Dogs who are kept penned by themselves spend very little time marking. The males rarely lift a leg to urinate, and neither sex bothers to deposit just a small amount. Dogs kept in similar-sized enclosures with other dogs mark much more frequently, and they mark regularly, every day. The Indian feral dogs marked to audiences—and audiences of the opposite sex. This makes sense if the message conveyed is about sex: seeking it oneself, or declaring oneself fit to be seeked. They did the most raised-leg displays (even without urinating) when other dogs were present. A leg held high only gets someone else's attention if someone else is already there to attend to it.

It also makes sense if the mark is communication for communication's sake: a comment, an opinion, a strongly held belief. There is no scientific evidence that it is so, but it is consistent with communication done only to an audience. Researchers have found that dogs raised in isolation make many fewer communicative noises than those raised with other dogs. When finally around others, though, they begin producing vocalizations at the same rate that the socialized dogs do. In other words, they speak when there is someone to speak to.

Just as they mark with intention, so too do dogs read intention in our markings: in our gestures. As we will see in the next chapters, they interpret the body language of humans with the attention they bring to reading each other. As a young child toddles toward a treasured toy, a dog can see where she is going and get there first. A turn of the head in thought garners little attention, but a turn of the head that looks at the door—there is intention in that turn. And dogs know it. They realize that there is a difference between gazing toward the door and turning to look at the clock on the wall; they can distinguish a finger pointing toward hidden food, and a point done while lifting one's arm to check a watch. We speak loudly with our bodies.

A confession: a dog has dictated this entire chapter to me. She sat by my chair, head on my foot, and patiently waited while I struggled to translate her words to the page. It is from her that the insights of the book come, from her that evocations spring, from her that the scenes and images and umwelt emerge.

Alas, it is not quite so. But to see the remarkable number of volumes purportedly written by dogs one must imagine that this is what we all want: the story straight out of the dog's mouth—but in our native tongue, of course. At the end of the nineteenth century, a peculiar kind of autobiography began to appear in bookshops: it was the "memoir" of your cat, your old dog, or the animal gone missing in that winter storm. This form, narrated by talking animals, could be considered the first prose attempt to get at the point of view of the dog. When I read one of these—and there are many to choose from, among even such writers as Rudyard Kipling and Virginia Woolf—a strange discontent washes over me. It's a sham: there is no perspective of the dog in them. Instead, it is a dog with the human's voice box transplanted to the dog's snout. Imagining that dogs' thoughts are but cruder forms of human discourse does the dog a disservice. And despite their marvelous range and extent of communication, it is the very fact that they do not use language that makes me especially treasure dogs. Their silence can be one of their most endearing traits. Not muteness: absence of linguistic noise. There is no awkwardness in a shared silent moment with a dog: a gaze from the dog on the other side of the room; lying sleepily alongside each other. It is when language stops that we connect most fully.

Dog-eyed

It takes all of six seconds for Pump to go from masterful to maladroit. In the first five she flawlessly navigates the brambles and bushes and thick-trunked trees that web the opening of the forest into field to catch a fast-moving tennis ball. It thonks off a tree and she's there to nearly vacuum it into her mouth. An apparition of a dog tears in out of nowhere, a racing blur of white fur and bark. Pump notes him and hurtles away, evading this stealer of tennis balls. In that sixth second, she stops, suddenly adrift. She's lost track of me. I watch her search: erect body, head high. I'm within sight; I smile at her. She looks at and past me, not seeing me. Instead, she spots the large, limping, heavy-coated man who came through with the white tear. She takes off after him. I must run to retrieve her. The moment before Pump was all-seeing; now she's a fool.

There is an intrinsic ranking of the modes by which we humans sense the world—and vision is winner by a long shot. Eyes arouse great interest in human psychologists; they betoken much more than one might imagine just from their physical form. However pretty a nose one might have, however close the forehead is to the brain, neither our noses nor foreheads nor cheeks nor ears are granted such importance.

We are visual animals. There's barely a challenge for second, either: audition is part of nearly every experience we have. Olfaction and touch might duke it out for third, and taste runs a distant fifth. Not that each of these isn't important to us on any particular occasion. The loveliness of presentation of, say, a tiered wedding cake would be undercut if vinegar replaced the anticipated taste of unmitigated sweetness. Or if any odor besides that of baked goods emanated from the cake—or if the first bite was not soft and yielding, but crunchy or slimy. Still, on most occasions we first direct our gaze to a new scene or object. If we notice something unusual or unexpected on the sleeve of our jacket, we turn to examine it with our eyes. Vision would have to really fail to provide any information before we decide to learn about it by inhaling it closely or taking a bold lick.

The order of operations is turned upside-down for dogs. Snout beats eyes and mouth beats ears. Given the olfactory acuity of dogs, it makes sense that vision plays an accessory role. When a dog turns his head toward you, it is not so much to look at you with his eyes; rather, it is to get his nose to look at you. The eyes just come along for the ride. You may be a recipient of an imploring gaze by a dog across the room right now. But can dogs even see what we do?

In many ways, the visual system of a dog—a subsidiary means to look at the world—is very much like our own. Its demotion behind other senses, in fact, may allow dogs to see details with their eyes that we overlook with ours.

One might well ask what a dog would even need eyes for. They can navigate and find food with their remarkable noses. Anything that needs closer examination goes right in the mouth. And they can identify each other through that sensory apparatus squished between their mouth and nose, the vomeronasal organ. As it turns out, they have at least two critical uses of their eyes: to complement their other senses and to see us. The natural history of the dog eye, seen in the story of their forebears, wolves, explains the context in which their vision evolved. It is a happy and transformative side effect that this has made them good watchers of human beings.

Just one element of the lives of wolves goes a long way to explaining the eyes they have evolved: eating. Most of their food runs away. Not only that, it is often camouflaged or living in the relative safety of herds. It is active—and thus findable—at dusk, dawn, or night. So wolves, like all predators, have evolved in response to their prey. As important as scent is, it cannot serve as the only indicator of the presence of prey, as air currents send odors on circuitous paths before they reach the nose. Odors are volatile: if smell lies on a surface, a sensitive nose can track it specifically; but if it is on the wind, the odor appears more like a cloud that could have come from one of a thousand sources. Rapidly moving prey outrun their own odor. Light waves, by contrast, are transmitted reliably through open air. So after catching a whiff, wolves use their sight to locate their prey. Many prey animals are camouflaged to blend in with their environment. This camouflage is betrayed in motion, however. So wolves are adept at spotting a change in the visual scene that indicates that something is moving. Finally, prey animals are often active at dusk or dawn, a compromise of lighting: easier to hide, harder to see. In response, wolves developed eyes that are especially sensitive in low light, and are especially good at noticing motion in that light.

Her eyes are deep pools of brown and black. It is hard to see which way they gaze, they are so dark—but it also makes any glimpse of her irises delightful, as though seeing inside her soul. Her eyelashes only became apparent when they grayed. Her eyebrows are also essentially invisible, but the effect of their moving—as with her head on the floor, to follow me walking across the room—is visible. In sleep, in dreams, her eyes scanned the world under their lids. Even closed, the lids reveal a bit of pink peeking out, as though she were keeping prepared to open her eyes at once should something important happen nearby.

At first glance these prey-tracking eyes look much like ours: viscous spheres fitted in sockets. Our eyes are about the same size as dogs'. Despite the fact that dogs' heads vary so significantly in size (four Chihuahua heads would fit in a wolfhound's mouth—not that anyone would deign suggest verifying such a thing), eye size barely varies between breeds. Small dogs, like puppies and infants, have large eyes relative to the size of their heads.

But small differences between the eyes of humans and dogs immediately become apparent. First, our eyes are smack in the front of our face. We look forward, and images in the periphery fall away to darkness around our ears. While there is variation among dogs, most dogs' eyes are situated more laterally on their heads in the manner of other quadrupeds, allowing a panoramic view of the environment: 250–270 degrees, as contrasted to humans' 180 degrees.

If we look a little closer, we discover another key difference. The superficial anatomy of our eyes gives us away: it shows where we're looking, how we're feeling, our level of attention. While dog and human eyes are similarly sized, our pupil size—the black center of the eye that lets light in—varies considerably when we are in a darkened room or are aroused or fearful (expanding to up to 9 millimeters wide) or in the bright sun or are highly relaxed (contracting to 1 millimeter). Dogs' pupils, by contrast, are relatively fixed at about 3 to 4 millimeters, regardless of the light or the dog's level of excitement. Our irises, the muscles that control pupil size, tend to be colored to contrast with the pupil, blue or brown or green. Not so in most dogs, whose eyes are often so monochromatically dark that they remind me of bottomless lakes, repository of all attributions of purity or desolation we might ascribe to them. And the human iris sits amidst the sclera—the white—of the eye, while many (but not all) dogs have very little sclera. The anatomical sum effect is that we can always see where another person is looking: the pupil and iris point the way, and the amount of sclera revealed underlines it. Without a prominent sclera or a distinct pupil, the eyes of a dog don't indicate the direction of their attention nearly as much.

Closer still, and we begin to see serious species differences. Dogs manage to gather more light than we do. Once light enters the eye of a dog, it travels through the gel-like mass that holds nerve cells to the retina (we'll get to that in a moment), then through the retina to a triangle of tissue, which reflects it back. This tapetum lucidum, in Latin "carpet of light," accounts for all the photographs you have of your dog with brilliant light shining out where their eyes should be. Light entering the dog's eye hits the retina at least twice, resulting not in the redoubling of the image but in a redoubling of the light that makes images visible. This is part of the system enabling dogs to have such improved night and low-light vision. While we might make out a match being brightly struck in the distance on a dark night, the dog could detect the gentle flame on the lit candle. Arctic wolves spend a full half of the year living in utter darkness; if there is a flame on the horizon, they have the eyes to spot it.

EYES OF THE BALL-HOLDER

It is inside the eye—at that retina receiving light twice—that, one by one, characteristic habits of dogs can be traced to their anatomy. The retina, a sheet of cells on the back of the eyeball, translates light energy into electrical signals to our brains, which leads to our feeling that we've seen something. Much of what we see is given meaning only by our brains, of course—the retina just registers the light—but without the retina, we would experience only darkness. Even slight changes in the conformation of the retina can radically change vision.

There are two slight changes in the canine retinae: in the distribution of photoreceptor cells and in the speed with which they operate. The former leads to their ability to chase prey, retrieve a tossed tennis ball, to their indifference to most colors, and to their inability to see something right in front of their noses. The latter leads to their being uninterested in daytime soap operas left on for them when their owners are out of the house. We'll look at these in turn.

Go get the ball!

Some of the most important things for humans to see are any other humans situated within a few feet of their face. Our eyes face forward, and our retinae have foveae: central areas with an extra abundance of photoreceptors. Having so many cells in the center of the retina means that we are very good at seeing things right in front of us in high detail, great focus, and strong color. Perfect for identifying that blob of color and form coming at you as your boyfriend or your mortal enemy.

Only primates have foveae. By contrast, dogs have what is called an area centralis: a broad central region with fewer receptors than a fovea, but more than in peripheral parts of the eye. Things directly in front of a dog's face are visible to him, but they are not quite as sharply in focus as they would be for us. The lens of the eye, which adjusts its curvature to focus light onto the retina, doesn't accommodate to nearby sources of light. In fact, dogs might overlook small things right in front of their nose (within ten to fifteen inches), because they have fewer retinal cells committed to receiving light from that part of the visual world. You need no longer puzzle at your dog's inability to find the toy that he is nearly stepping on: he's not got the vision to take note of it until he takes a step back.

Breeds of dogs differ so much in their retinae that they see the world differently. The area centralis is most pronounced in those breeds with short noses. Pugs, for instance, have very strong areas centralis—almost fovea-like. But they lack a "visual streak," which dogs with long noses (and wolves) have. In Afghans and retrievers, for instance, the area centralis is less pronounced, and the photoreceptors of the retina are more dense along a horizontal band spanning the middle of the eye. The shorter the nose, the less visual streak; the longer the nose, the more visual streak. Dogs with the visual streaks have better panoramic, high-quality vision, and much more peripheral vision than humans. Dogs with the pronounced area centralis have better focus in front of their faces.

In a small but significant way, this difference explains some breed-based behavioral tendencies. Pugs are not typically so-called "ball dogs" but long-nosed Labrador retrievers are. Not because of their long noses per se. In addition to their ability to put their millions of olfactory cells to good use, Labradors are visually equipped to notice, say, a tennis ball traveling across the horizon, without having to shift their gaze. For a short-nosed dog (as for all humans of any nose length), a ball tossed to the side just disappears into the periphery if they don't follow it with their head. Instead pugs are probably better at bringing close objects into focus—say, the faces of their owners on whose laps they sit. Some researchers speculate that this relatively blinkered vision makes them more attentive to our expressions, and seem more companionable.

Go get the green ball!

Dogs are not color-blind, as is popularly believed. But color plays a much less important role for them than it does for humans, and their retinae are why. Humans have three kinds of cones, the photoreceptors responsible for our perception of details and of colors: each fires to red, blue, or green wavelengths. Dogs have only two: one is sensitive to blue and the other to greenish-yellow. And they have fewer of even those two than humans do. So dogs experience a color most strongly when it is in the range of blue or green. Ah, but a well-scrubbed backyard pool must seem radiant to a dog.

As a result of this difference in cone cells, any light that looks to us like yellow, red, or orange simply doesn't look the same way to a dog. Consequently, they seem perfectly oblivious when you ask them to bring back grapefruits from the store and grow irritated when they bring back tangerines. Still, orange, red, and yellow objects might still look different to them: the colors have different brightnesses. Red may be seen by them as a faint green; yellow a stronger one. If they seem to be able to discriminate red and yellow, they are noticing a difference in the amount of light these colors reflect toward them.

To imagine what this might be like, consider the time of day when our color system breaks down: in the dusk right before night. If you're outside in a park, in your yard, or anywhere nature lives, take a look around you. You might notice the wash of exuberant green leaves above you subtly dulling to a more unassuming hue. You can still see the ground underfoot, but details—the distinctness of blades of grass, the layers of petals—are reduced. Depth of field squashes somewhat. I tend to stumble more than usual on protuberant gray rocks that blend with the earth. The reason for this loss of visual information is anatomical. Cones, clustered toward the center of the retina, are not sensitive to low light, so they don't fire as often at dusk or night. As a result, our brains get signals from fewer cells detecting colors. And the near world flattens out a little: we can still see that there is color, and we still detect lights and darks, but the richness of colors has fallen away; colors are grainier, less detailed. So it might be for dogs, even at noontime.

As they do not experience a great range of distinct colors, dogs rarely show color preferences. Your clashing choice of red leash and blue collar affects your dog not at all. But a deeply saturated color may get more attention from a dog, as will an object placed in a background of contrasting colors. It may be meaningful that your dog attacks and pops all the blue or red balloons left over after the birthday party winds down: they are most distinct among a sea of pastels.

Go get the green bouncing ball … on the TV!

Dogs make up for their dearth of cones with a battery of rods, the other kind of photoreceptor in the retina. Rods fire most in low-light situations and at changes in light densities, which is seen as motion. In human eyes, rods cluster at the periphery, helping us notice something moving out of the corner of our vision, or when the cones slow their firing at dusk or night. The density of rods in dogs' eyes varies, but they have as much as three times as many rods as we do. You can make that ball your dog is not seeing directly in front of him magically appear by giving it a little shove. Acuity greatly improves for close objects when they are bouncing.

All these differences in the dog's perception, experience, and behavior result from some small changes in the distribution of cells on the back of the canine eyeball. And there is another small change that results in a large difference—potentially more far-reaching than a change in focal area or color vision. In all mammalian eyes, rods and cones make electrical activity out of light waves by means of a change in the pigment in the cells. The change takes time—a very small amount of time. But in that time, a cell processing light from the world cannot receive more light to process. The rate at which the cells do this leads to what is called the "flicker-fusion" rate: the number of snapshots of the world that the eyes take in every second.

For the most part, we experience the world as smoothly unfolding, not as a series of sixty still images every second, which is our flicker-fusion rate. Given the pace at which events that matter to us happen, this is usually plenty fast. A closing door can be grabbed before it slams; a handshake received before it is withdrawn in annoyance. To create a simulacrum of reality, films—literally "moving pictures"—must exceed our flicker-fusion rate only slightly. If they do, we do not notice that they are just a series of static pictures projected in sequence. But we will notice if an old-fashioned (pre-digital) film reel slows down in the projector. While ordinarily the images are being shown to us faster than we can process them, when it slows we see the film flickering, with dark gaps between the frames.

Similarly, fluorescent lights are so annoying because they operate too close to the human flicker-fusion rate. The electrical devices used to regulate the current in the light function right at sixty cycles per second, which those of us with slightly faster flicker-fusion rates can thus see as a flicker (and hear as a buzz). All indoor lights fluorescently flicker to houseflies, with extremely different eyes than ours.

Dogs also have a higher flicker-fusion rate than humans do: seventy or even eighty cycles per second. This provides an indication why dogs have not taken up a particular foible of persons: our constant gawking at the television screen. Like film, the image on your (non-digital) TV is really a sequence of still shots sent quickly enough to fool our eyes into seeing a continuous stream. But it's not fast enough for dog vision. They see the individual frames and the dark space between them too, as though stroboscopically. This—and the lack of concurrent odors wafting out of the television—might explain why most dogs cannot be planted in front of the television to engage them. It doesn't look real.*

One could say that dogs see the world faster than we do, but what they really do is see just a bit more world in every second. We marvel at dogs' seemingly magical skill at catching a Frisbee on the fly, or following a rapidly bouncing ball. Their Frisbee-catching procedure, as has been documented with microvideo recording and trajectory analysis, turns out to match nicely the navigational strategy naturally used by baseball outfielders to line themselves up with the arc of an incoming ball. Excepting a few phenomenal outfielders, dogs actually see the Frisbee's, or the ball's, new location a fraction of a second before we do. Our eyes are internally blinking in those milliseconds that a flung Frisbee is moving along its course toward our heads.

Neuroscientists have identified an unusual brain disorder in some humans called "akinetopsia." These akinetopsics have a kind of motion blindness: they have difficulty integrating a sequence of images into the normal perception of motion. A person with akinetopsia may begin pouring a cup of tea and then not register a change until many images later, by which time the cup is overflowing. As non-brain-damaged persons are to akinetopsics, dogs are to us: they see the interstices between our moments. We must always seem a little slow. Our responses to the world are a split second behind the dogs'.

VISUAL UMWELT

With age, Pump suddenly became reluctant to enter the elevator, perhaps not seeing it well in the darkness after being outside. I encourage her, or jump in myself first, or throw something light-colored on the elevator floor for her to see. Finally, every time, she rallies and leaps in, as though crossing a great crevasse, brave girl.

So dogs can see some of the same things we do, but they don't see in the way that we do. The very construction of their visual capacity explains a broad swath of dog behavior. First, with a wide visual field, they see what is around them well, but what is right in front of them less well. Their own paws are probably not in terrific focus to dogs. What wonder then how little they use their paws, relative to our reliance on the end of our forelimbs, to manipulate the world. A small change in vision leads to less reaching, grabbing, and handling.

Similarly, dogs can bring our faces into focus, but detect eyes less well. This means they will catch your full facial expression better than a meaningful glare, and they will follow a point or a turn better than a surreptitious glance out of the corner of the eye. Their vision complements their other senses. While they can locate a sound in space only roughly, their hearing is good enough for them to turn their eyes in the right direction, so they can search further visually … and then examine closely by nose.

For instance, dogs recognize us by our smells—but they also clearly look at us. What are they seeing? If your smell is not available—you are downwind or you're covered in perfume—they can use visual cues exclusively. They will hesitate if they hear your voice calling them, but it is not your face atop the approaching person, or your particular walk, or your mouth moving to call their name. Recent research confirmed this by examining dogs' behavior when they heard their owner's voice or a stranger's voice, accompanied by either a picture of the owner's face (on a large monitor) or of a stranger's face. The dogs looked longer at the incongruous faces: the owner's face, when paired with a stranger's voice, and the stranger's face, when it appeared with the owner's voice. If it were just that the dog preferred the owner's face, they would have always gazed at that face the longest. Instead they looked longest when there was something surprising: a mismatch.

The physical elements of vision define and circumscribe what the dog experiences. There is a further element of that experience: the role vision plays in the hierarchy of senses. For visual creatures like humans, there is particular delight when we encounter something through one of our non-visual senses first. To arrive outside my apartment door and smell something wonderful—to open the door and hear the sounds of sizzling in pots, the clank of silverware; to be instructed to taste a forkful of the pot's contents with my eyes closed—renders a familiar experience new. I only come forth to verify the scene with my eyes: my boyfriend in front of me with dinner in messy preparation around him.

Coming to something through the secondary senses first discombobulates, then introduces a feeling of novelty to the ordinary. As dogs have their own hierarchy of senses, I imagine that they too might feel the mystery of coming at something by means other than their nose. This may explain both the difficulty dogs have in understanding some of our first requests to them (off the couch! I said to my new puppy, as she looked at me searchingly), and the pride they seem to take in learning a distinction from our visual world.

Though our visual worlds overlap, dogs attach different meanings to the objects seen. A Seeing Eye dog must be taught the umwelt of the human: the objects that are important to the blind person, not those of interest to the dog. Try yourself getting your dog to even acknowledge the existence of a sidewalk curb. What is a curb to a dog? With persistence, dogs can be taught, but most dogs simply do not see a curb: it is not that the curb is invisible, but that it lacks any important meaning to them. The surface below their feet may be rough or soft, slippery or rocky, it may hold the scent of dogs or of men; but the distinction between the sidewalk and the street is a human distinction. A curb is but a slight variation in altitude of the hardened mass with which we cover the dirt, which only has a meaning to those who concern themselves with such concepts as roads, pedestrians, and traffic. The guide dog must learn the importance of the curb to his companion. He must learn the significance of a speeding car, a mailbox, other people approaching, a doorknob. And he will: he may begin to associate the curb with the distinctive striping of a crosswalk, with the dark, smelly rain gutters that run along them, or with the change in brightness from the concrete to the asphalt. Dogs are much better at learning about things that are important to us in our visual world than we seem to be in understanding theirs. I still can't tell you why Pump became excited at the mere sight of a husky-shaped dog appearing around the corner. But after a dozen years I began to notice that she did. She, on the other hand, was quicker to recognize the importance I placed on certain objects—the distinction between the frayed sofa and my favored armchair with respect to her chance of sitting on it; the slippers whose fetching made me laugh versus the running shoes whose delivery made me scold.

There is a final, unexpected facet of the visual experience of the dog: they see details that we cannot. The fact of dogs' relatively weak visual capacity turns out to be a boon to them. Since they are not trying to take in the whole world with their eyes alone, they may see details that we don't notice. Humans are gestalt lookers: every time we enter a room, we take it all in in broad strokes: if everything is more or less where we expect it to be … yes … we stop looking. We don't examine the scene for small, or even radical changes; we might miss a gaping hole in the wall. Don't believe it? At every moment of our lives we are not noticing a gaping hole: one in our visual field caused by the very construction of our eyes. The optic nerve, the neural route conveying information from the retinal cells to brain cells, tunnels right through the retina on its way back to the brain. Thus if we hold our eyes still, there is a part of the scene in front of us that is not captured on our retina—as there is no retina there to capture it. It's a blind spot.

We never notice this gaping hole in front of us because our imaginations fill in that spot with what we expect to be there. Our eyes dart back and forth constantly and unconsciously—movement called saccades—to further complete the visual scene. We never experience the missing spot. In the same way, we also have a blind spot for those things that are slightly different—but close enough—to what we expect to see. As well-adapted visual creatures, our brains are equipped to find the sense in the visual information sent it, despite holes and incomplete information.

We are maybe too well adapted. Some of what we overlook, animals see. The celebrated autistic scientist Temple Grandin has demonstrated the reality of this with cows, for instance. Often cows being led along wending chutes into the slaughterhouse balk, kick, and refuse to proceed. As far as we know, this is not because they understand what will happen in the slaughterhouse. Instead there were small visual details that surprised or frightened them. The reflection of light in a puddle; an isolated yellow raincoat; a sudden shadow; a flag flapping in the breeze: seemingly insignificant details. We are certainly able to see these visual elements—but we do not notice them as cows do.

Dogs are closer to those cows than to us. Humans quickly label and categorize a scene. Walking to work along a Manhattan street, the typical commuter is perfectly oblivious to the world he is passing. He notices neither beggars nor celebrities; startles neither to ambulances or parades; simply sidesteps a crowd gathered to gape at … well, whatever it is crowds gape at: I rarely stop to see. On most mornings, the route is reduced to its landmarks; nothing else needs attending to. There is good reason to believe that this is not how dogs think. The walk to the park becomes familiar over time, but they don't stop looking. They are much more struck by what they actually see, the immediate details, than what they expect to see.

Given how dogs see, how do they apply their visual ability? Cleverly: they look at us. Once a dog has opened up his eyes to us, a remarkable thing happens. He starts gazing at us. Dogs see us, but the differences in their vision also seem to allow them to see things about us that even we do not see. Soon it seems they are looking straight into our minds.

Seen by a Dog

I am startled and a little flustered to look up from my work and see Pump watching me, her eyes trained on mine. There is a powerful pull to a dog who looks you in the eyes. I am on her radar: it feels that she is looking not just at me, but to—and into—me.

Look a dog in the eyes and you get the definite feeling that he is looking back. Dogs return our gaze. Their look is more than just setting eyes on us; they are looking at us in the same way that we look at them. The importance of the dog's gaze, when it is directed at our faces, is that gaze implies a frame of mind. It implies attention. A gazer is both paying attention to you and, possibly, paying attention to your own attention.

At its most basic level, attention is a process of bringing forward some aspects of all the stimuli bombarding an individual in a moment. Visual attention begins with looking; auditory attention with hearing: both are possible for all animals with eyes and ears. Just having the sensory apparatuses isn't sufficient to do what we generally mean by paying attention, though: considering what it is one has turned to look or hear.

When invoked by psychologists, attention is treated not just as turning the head toward a stimulus, but as something else in addition: a state of mind that indicates interest, intent. In attending to someone else's head turns, one may be demonstrating an understanding of the psychological states of other people—a distinctively human skill. We attend to others' attention because it helps to predict what that someone other will do next, or what he can see and what he might know. One of the deficits that many people with autism have is an inability, or lack of inclination, to look at other people's eyes. As a result, they aren't instinctively able to understand when other people are paying attention—or how to manipulate others' attention.

The simple ability to focus on some things while ignoring others is crucial for any animal: objects one sees, smells, or hears may be more or less relevant for survival. Attend to those that are relevant; ignore the rest of the visual landscape or the confusion of sounds. Even with survival no longer our most pressing concern, humans are constantly trying to direct, divert, or attract attention. Some attention mechanism is required to do all the ordinary things of our days: to listen to someone talking to us, to plan a walking route to work, even to remember what one was thinking a moment ago.

Dogs, social animals like us, and also more or less relieved of survival pressures, surely have some interesting mechanisms with which to attend to the world. By virtue of their different sensory abilities, though, they are able to attend to things we never notice, such as how our odor changes through the day. Likewise, we focus carefully on things that dogs do not even detect, such as subtle differences in language use.

But what distinguishes dogs from other mammals, even other domesticated mammals, is the way that their attention overlaps with ours. Like us, they pay attention to humans: to our location, subtle movements, moods, and, most avidly, to our faces. A popular conception of animals is that if they look at us at all, it is from fear or appetite, monitoring us as possible predator or prey. Not true: the dog looks very particularly at humans.

Just how particularly is the subject of a mad rush of contemporary research into dog cognitive abilities. This research uses as markers the landmarks in the development of human infants into human adults, which is well documented, and which result is obvious: by adulthood, we all understand what it means to pay attention. What the dog research is revealing is that dogs have some of the same abilities that we do.

THE EYES OF A CHILD

For dogs and humans both, it all begins with a few innate behavioral tendencies. Having and understanding attention is not automatic, but it develops naturally from these instincts. Human infants, like most animals, have a basic orienting reflex: move, as best or as much as you can, toward a source of warmth, food, or safety. Newborns turn their faces toward warmth and suck: the rooting reflex. At that age, infants can do little more. Ducklings, more precocious, relentlessly chase after the first adult creature they see.* In both ducklings and humanlings, this reflex relies on an early perceptual ability: having at least noticed the presence of others. It is an ability that helps us, in our first few years, learn about the important fact of others' attention.

For humans, there is a reliable course of development through infancy of certain behaviors associated with this growing understanding of other people. It is all about learning to attend to the right—human—things in the world, and beginning to understand that others are attending, too. And it begins as soon as they open their eyes. Newborn babies can see, although not much. They are incredibly nearsighted: peering, cooing faces brought just inches from their own may be clear, but that is about the extent of clarity of the world. One of the first things infants notice is any faces nearby. In fact our brains have specialized neurons that fire when we see a face. Infants can detect, and prefer to look at, a face or something facelike—even three points forming a V—rather than other visual scenes. From early in their lives, infants stare longer* at that which interests them, the mother's face being among the first items of interest. Soon infants also learn to distinguish a face looking toward them from one looking away. This is a simple skill, but not a trivial one: out of the visual cacophony of the world, they must start noticing that there are objects, that some of those objects are alive, that some of those alive objects are of particular interest, and that some of those interesting live objects attend to you when they face you.

Once that is established, and their own visual acuity improves, infants focus on the details in that face. They delight at peekaboo: a game playing simply with the importance of eyes. As psychologists have shown by sticking out their tongues and making faces at infants, very young infants can imitate simple expressions. Of course, these expressions don't have the meaning that they will later (we must assume that the infant is not actually sticking his tongue out spitefully at the psychologist, though one might wish it so). Infants are simply learning to use their facial muscles. By three months, they've got it, and they start reacting to others by making faces and smiling socially. They move their heads to look at other faces nearby. By nine months old they are tracking other people's gaze and seeing where it lands. They might use that gaze to find some object that they have asked about, or that has been hidden from them. Soon they extend the line of gaze into a point with finger, fist, or arm to request an object, and by their first birthday, to show or share.

These behaviors reflect the infant's burgeoning understanding that other people have attention, attention that can light on objects of interest: a bottle, a toy, or them. Between twelve and eighteen months, they begin to engage in bouts of joint attention with others: locking eyes, then looking to another object, then back to eye contact. This marks a breakthrough: to achieve full "jointness" the infant must on some level understand that not only are they both looking together, they are attending together. They are understanding that there is some invisible but real connection between other people and the objects that are in their line of vision. Once they do this, all hell can break loose. Infants can start manipulating others' attention simply by gazing someplace. They check where other people are looking and pointing, and they begin to notice if adults are looking at them while they are doing activities they want to share (or conceal). They will give an anticipatory glance at an adult before pointing or showing themselves. They work very hard to get attention looking at them. And they may begin avoiding attention: going out of a room at key moments, or concealing objects from an adult's view. (This prepares them well for being difficult adolescents.)

We all become characteristically human by this same developmental route. Within a few years an infant goes from aimlessly looking out of new eyes, to looking meaningfully, to gazing at others, to following the gaze of others. They happily hold eye contact. Before long they are using gaze to get information, to manipulate the gaze of others—by distraction, gaze avoidance, or pointing—and to get attention. At some point, they come to a realization about the fact of the mind behind someone else's gaze.

THE ATTENTION OF ANIMALS

She comes within an inch of me and starts panting at me, eyes wide and unblinking, to tell me that she needs something.

Step-by-step, cognition researchers have been tracing this developmental course with a new subject: non-human animals. How much of the infant's trajectory is followed by animals? After they open their eyes, do they look with intention? Do they notice others' eyes? Do they understand the importance of attention?

This is one facet of the study of animal cognition, which asks what an animal subject understands about the "mental states" of others. Most of the experimental tests run with animals are of the kind we feel sure we humans excel at: tests of physical and social cognition. Captive animals from sea slugs to pigeons to prairie dogs to chimpanzees have been set into mazes; presented with counting, categorizing, and naming tasks; asked to discriminate, learn, and remember series of numbers and pictures. Tasks are devised to see if they recognize, imitate, or deceive others—or even recognize themselves. And in some tests, the question is even more characteristically human: of the kind of social thinking going on when animals interact—with members of their own species, and with those of other species. When a caged chimpanzee looks at a human attendant, is he considering anything about the attendant? Does he wonder how to get her to open the door (does he wonder anything at all?), or is he simply waiting to see what this colorful, animate object nearby does that might be relevant or interesting? Does a cat consider that mouse as an agent, as an animal with a life—or does he see the mouse as a moving meal that must be stopped and dismantled?

As we've touched on already, the subjective experience of animals is notoriously difficult to get at scientifically. No animal can be asked to relate its experience in voice or on paper,* so behavior must be our guide. Behavior has its pitfalls, too, since we cannot be positive that any two individuals' similar behavior indicates similar psychological states. For instance, I smile when I am happy … but I may also smile out of concern, uncertainty, or surprise. You smile back at me: it too might be happiness—or ironic detachment. To say nothing of the near impossibility of determining whether your "happiness" feels like mine does.

Still, even without having constant verification of others' mental states, behavior is a good enough guide to allow us to predict an animal's future behavior well enough to interact peacefully and productively. Thus we study what animals do—in particular, what they do that is like what humans do. Since using and following attention is so important in human social interaction, animal cognition researchers look for behaviors that indicate that an animal is using attention.

Dogs have recently trotted gamely into experimental labs, controlled outdoor facilities, and onto data sheets meant to gather information about their abilities at using attention. The dogs are put in controlled settings, usually with one or more experimenters present, and a hidden, desirable object: a toy or a food treat. By varying the cues that they use to inform the dogs about the location of the treat, the experimenters aim to determine which ones are meaningful to the dogs.

The question for researchers is just how far along these stages of the child's development of attention dogs go. Attention begins with gaze, and gaze requires visual capacity. We have already established what dogs can see; we know that they look. Do they understand attention?

Mutual gaze

A gaze is more than it seems: by gazing at someone, one very nearly acts upon him. As my students discover in their field experiments, eye contact comes close to feeling like actual tactile contact. There are undiscussed and yet widely shared rules governing eye contact with others—violation of which may be seen as an act of aggression or of intimacy. We may stare down someone in an attempt to subdue them, or, alternatively, use a long, steady gaze to indicate a more lustful interest.

With a little variation, this could as easily describe how many non-human animals use eye contact. Between apes, eye contact is steeped with importance: it can be used as an aggressive action, and will be avoided by a submissive member of a troop. To stare at a dominant animal is to invite yourself to be attacked. Not only do chimpanzees avoid staring, they avoid being stared at. Subordinate chimps carry themselves despondently, looking down at the ground or their own feet and only furtively glancing around them. In wolves, too, a direct stare may be taken as a threat. So the "aggressive" element of eye contact is the same as with humans. The variation is this: all non-human animals with any meaningful visual capacities will turn their eyes to something of interest—but if the thing-of-interest is a member of their species, the social pressure of gazing usually deflects the gaze of interest.

Thus we can expect that dogs might act somewhat differently than we do with regards to mutual gaze. As dogs evolved from a species in which a stare is most often a threat, we might do best to consider their avoidance of eye contact less an inability than a result of their evolutionary history. But wait! Dogs do look at our faces. They look at each other in the center of the face: at eye level. Most dog owners will report that their dogs gaze at them directly in the eyes.*

So something changed with dogs. While the threat of aggression prevents mutual gaze among wolves, chimpanzees, and monkeys, for dogs the information to be gained by looking us in the eyes is worth enduring any residual, ancient fear that a stare might cause an attack. That humans respond well to a dog gazing at them is a happy circumstance—and our bond with them is thereby strengthened.

To be sure, it may be less "eye contact" than "face contact."* Because of the superficial anatomy of the dog eye—the lack of distinct iris and whites of the eye—specific eye direction can often only be confirmed from closer range than scientists' video cameras have gotten. Generations of dog breeders have tended to prefer the trait of dark eyes in their charges. Dogs with light-colored irises are often thought to look volatile or sneaky—ironically because we can see clearly when they avoid eye contact. By breeding out the light irises we do not eliminate shiftiness, just our awareness of the fact that dogs shift their gaze. Darting eyes become less conspicuous. We sleep better at night with a calm-featured dog at the foot of our bed than with a nervous eye-darter. For all intents and purposes, though, we can say that a dog and human "mutually gaze" when we turn our faces toward each other.

The primal pull of gaze still affects dogs' behavior. If you stare unblinkingly at your dog, he may look away. Approached by a dog who appears overly aggressive or overly interested, a dog can diffuse some of that excitement by glancing to the side. Your chastisement or accusation of your dog accompanied by a glare may also provoke a demure averral of the dog's gaze. Given the easily recognized shifty look of the guilty man confronted by his accuser, it is no wonder that we attribute the same to the gaze-avoiding dog. The refusal to look us in the eyes contributes to a look of guilt—especially when we are already certain they have done something to inspire it. Whether they are themselves feeling guilt or atavism is not obvious.

But the fact that dogs will look us in the eyes allows us to treat them as a little more human. We apply to them the implicit rules that accompany human conversations. It is not uncommon to see a dog owner pause from scolding a "bad dog" to physically turn the dog's head back toward the owner's face. We want dogs to look at us when we are talking to them—just as we use gaze in human conversation, in which listeners look at the face of the speaker more than the reverse. (Notably, we do not stare at each other non-stop in conversation, and might feel unsettled if someone did.) There is more direct eye contact among humans speaking intimately or honestly, and we tend to extend that conversational dynamic to our dogs. We call their names before speaking to them, treating them like willing, if taciturn, interlocutors.

Gaze following

It doesn't happen at once, but not long after bringing a dog or puppy to your home for the first time you might notice something: nothing in the house is safe. Dogs train humans to become suddenly tidy: putting away shoes and socks almost as soon as they are removed; taking out the trash well before it heaps high; leaving nothing on the floor that could fit into the gaping mouth of a teething, excited, unrestrained pup. A temporary peace may ensue. After all, you can put things behind closed doors, into shut cabinets, and onto high shelves. Dogs look, baffled, in the area from which the (shoe, takeout container, hat) has mysteriously gone missing. But soon you will notice that the dog has learned something new: you are the source of the mysterious relocations—and you have a tendency to tip your hand.

How? You look. When we pick up the sock and set it down, we're not just attached to it by a hand; the action is accompanied by a gaze. We look where we are going. Later, we may look again at that safe sock spot when discussing the dog's earlier thievery. Again our gaze reveals the location of the sock: the gaze is itself information. We have already met this ability to use the direction of another's gaze, so-called gaze following, which infants do before reaching one year old. Dogs do it even sooner.

A gaze that intends to share information is simply a point done without hands. Following a point is a slightly simpler ability. Certainly dogs see a lot of pointing and gesturing as they observe human members of their families. This may be the source of their gaze-following ability, or it may simply bring out an innate ability to glean any information possible from our behavior. Researchers have tested the limits of their ability, natural or learned, in various experiments that put dogs in a context where they can get information from a person's pointing gesture. For instance, a biscuit or other desired food might be hidden under one of two inverted buckets while the dog subject is out of the room. When all odor cues are masked, the dog has to make a decision which bucket to choose. If he chooses correctly, he is rewarded with the food; if not, he is rewarded with nothing. A person who knows which bucket to choose is standing nearby.

Chimpanzees have been given variations of this task in captive research settings. Surprisingly, though they seem to follow points, they don't always do well at following gaze alone.

Dogs perform admirably. They follow points, points that reach across the pointer's body, points from behind the body, and are even better if the point includes a finger further signifying the baited bucket.* They haven't simply learned the importance of an outstretched arm. Pointing with elbows, knees, and legs also serves as information. Given even a momentary point—a glance of a point—the information is theirs. They can follow the pointed cue given by a life-sized video projection of their owner. Though they have no arms with which to point themselves, they outperform the chimps who have been tested. Best, dogs can use simply the person's head direction—her gaze—to get information. You may be able to hide that sock from your sock-coveting chimpanzee, but your dog will spot it.

Where dogs' use of attention really gets interesting is in less overt cases. Not just when we point and they look, but when they have to decide how to inform us that they need to go outside—or want a ball tossed to them. Or they need to tell us some very important news about where a tasty treat fell out of their reach while we were out of the room. Play with humans is a rich context for the possible appearance of some of these abilities; experimental paradigms also manipulate the information that can be gleaned from others' attention. All signs indicate that dogs seem to understand how to get attention, how to make requests of us using attention, and what kind of inattention allows them to get away with bad behavior.

Attention-getting

The first of these abilities, when seen in children, is called "attention-getting." Informally, you may know it as anything that your dog does to interfere with what you are currently trying to do. More formally, these are behaviors that are sufficient to change the focus of someone else's attention, by stepping into his visual field, making a discernible noise, or making contact. Suddenly jumping on you is a familiar dog attention-getting behavior, if not one that is well loved by the jumped-upon. Barking is another. Their attention-getting means aren't restricted to the quotidian, though. Less recognized means include bumping, pawing, or simply orienting oneself right up in front of someone else: what I have called an in-your-face in my data of dog play behaviors. Guide dogs use "sonorous mouth licking"—audible slurps—to get the attention of their visually impaired charges when needed. The excitement of play sometimes leads them to come up with novel techniques, too. My favorite sessions to observe are those in which an eager but frustrated dog mirrors the behavior of the object of his unrequited play interest: approaching and drinking from the bowl from which the first dog is drinking—and using it as a means toward licking his face; or grabbing a stick of his own when another dog finds a good stick to be sufficient company.

Dogs use attention-getters regularly with us, and are often rewarded with our attention. But unless they show some subtlety in application of these behaviors, their use does not prove their full understanding of our attention. It may be that they are simply throwing all the tools they have at the problem of needing you to look at them. A child hollers, you come racing to his side: an attention-getter is born. Observations of dogs playing with humans show just how crude or subtle their use of these behaviors is. There are dogs who will stand barking continuously over a retrieved tennis ball, while their owners socialize with members of their own species. While a good attention-getter, a bark is not well applied if it continues to be used after it has failed to get attention. On the other hand, there is also evidence of very subtle visual attention-getters by dogs in reaction to the divided attention of their owners. By changing posture, as from a seated posture to standing, or from standing to approaching, dogs are able to reengage the owner enough to toss a ball or lunge playfully.

You are regularly a witness to the flexibility dogs show in attention-getting. If your dog did not rouse you from your armchair and novel by simply approaching you, he may have wandered off only to return carrying a shoe or another verboten item. Probably this causes you to chastise him gently and return to your book. More serious tactics are needed, he sees. Next may come whining, or a tentative woof; a tactile intervention—a slight push with a wet nose, nuzzling, or jumping; even a loud drop to the floor at your feet with a sigh. They are trying their best with you here.

Showing

So far, the dogs have kept pace with the developing child: gazing, following a point, following gaze, and using attention-getters. Do they also point, as best they can, with their bodies? Do they point with their heads to show you something?

Here again, experimenters set up a situation that they presumed would prompt the behavior, if the ability exists. The scenario is the gaze-following task, inverted. Instead of being the naïve ones, in these cases the dogs are informed but impotent: they alone witness an experimenter hiding a treat, damnably out of their reach. Their owners then wander into the room, and experimenters train their cameras on the dogs: Do they see the owners as tools which can help them? If so, do they communicate the location of the treat?

In these cases, it seems the only obtuse animals in the room are the humans, who may not see the dog's behavior as potentially showing them anything. That behavior consists of a lot of attention-getters (such as barking), followed, critically, by looking back and forth between the owner and the location of the treat. In other words, pointing with that gaze: showing.

This is visible daily in non-experimental settings. Ball dogs crazy about retrieving generally deliver their slobbered spheres to the front side—the face—of the ball tosser, not to her back. And, if the ball is mistakenly dropped at the unresponding backside of the owner, the dog has an arsenal of attention-getters to employ, followed by relentless gaze alterations—looking at the face of the human holding the ball, and looking back at the ball in quick succession. The restless, attention-starved dog is never satisfied dropping found socks at your back; they are left within your sight, if not right on your lap.

Manipulating attention

Finally, dogs use the attention of others as information, both to get something they want and, more remarkably, to determine when they can get away with something.

Research has determined this by asking if dogs choose intelligently when given a choice about whom to request food from. If every person is an equally good source of food, one would expect that dogs would approach all persons with that same beguiling expression—half entreaty, half expectation. There are dogs who do so, of course,* and those who reserve their begging for butchers, or owners who stuff their pockets with liver treats. But most dogs make a distinction that is important to us when we desire something: between possible and impossible collaborators. We make requests appropriate to the state of knowledge and capacity of our audience. You do not ask the baker to explain string theory and the physicist for a loaf of seven-grain, sliced.

In experimental settings mining the same four elements of dog, experimenter, food, and knowledge, dogs seem to distinguish between humans who might be helpful to them and humans who will likely not. When a person with a sandwich is either blindfolded or facing away, dogs suppress the urge to stay as close as possible to the sandwich. Instead, if there is a non-blindfolded person nearby, they go beg to him instead. Let this be a lesson that begging at the table is probably encouraged by your eye contact toward the dog—even just long enough to tell him no begging! Alternately, set up one person as the responsive, looking beggee and all the dog's attention will go to him. (Children are good for this role.)

Dogs also approach the blindfolded persons warily—as befits the situation, if one isn't let in on the fact of being a subject in an experiment. These experiments using unresponsive, oddly outfitted characters are typical of psychological tests. At some level, they are useful in order to avoid the possibility that the subject has had experience with the setting they are about to encounter. In other words, the tests aim to get at what the dogs intuitively understand about the knowledge states of the human, not what the dog might have learned about what to do when you see someone who is blindfolded. Still, the dog is confronted with what must be a strange couple of hours.

Variations of the begging trials were first run with chimpanzees. In that context, the attentional state of the human was taken to indicate something about her knowledge. Someone who sees food baited in one of two hidden bins is "knowledgeable"; someone who stands idly by in the same room, but has a bucket over her head, is not. Did the chimps then beg to the knowledgeable person or to the one who is guessing at the location of the food (by chance guessing correctly once in a while)? Over time, chimps learn to beg to the knowledgeable informant—but only when the guesser has been out of the room, or has her back turned when the bin is baited. When the guesser simply has her eyes blocked—with a bucket, paper bag, or blindfold—the chimps begged to her, too.

Dogs have gone through trials with odd humans wearing buckets, blindfolds, or holding books in front of their eyes, blocking their vision. They outperform chimps: dogs preferentially beg to the looking—to those whose eyes they can see. This is just how we act, preferring to talk, cajole, invite, or solicit those whose eyes are visible. Eyes equal attention equals knowledge.

Best, dogs use this knowledge for manipulative ends. Researchers have found that dogs not only understand when we are attentive, but are sensitive to what they can get away with at different levels of their owners' attention. In one experiment, after being instructed to lie down (and dutifully so doing), dogs were observed in three trials. In the first condition, an owner stood and stared at her dog. The result? The dog stayed lying down: perfectly obedient. In the second condition, the owner proceeded to sit down and watch television: here the dog paused, but shortly disobeyed and got up. And in the third condition, the owner didn't just ignore the dog but left the room entirely, leaving the dog alone with his owner's command still echoing in his ears.

Apparently the echo was not long-lasting, for in these trials the dogs were quickest and likeliest to disobey the same command so well heeded when the owner was around. What is surprising is not that the dogs disobeyed when the owner left. It is, instead, that dogs do what two-year-olds, chimps, monkeys, and no other animals seem to do: simply notice exactly how attentive someone is, and vary their own behavior accordingly. The dogs methodically used the level of their owners' attention to determine under what circumstances they were free to break the owners' rules—just as they used the information from other dogs to get attention back toward them in play.

The dogs' attention-reading is highly contextual, however. When the same experiment was run using food, that great motivator to perform at their best, the threshold to disobedience was lowered: dogs disobeyed more quickly, and at lower levels of owner distraction. When the owner's attention was harder to gauge—when she was talking with someone else, or sitting quietly with closed eyes—the dogs' behavior was mixed. Some sat patiently, but, seemingly gathering steam, were prepared to spring up at once as soon as the owner left the room. Other dogs took even longer to disobey when the owners left the room than when they were in it but otherwise engaged. This illogic might be explained by a developmental fact, one that would vary dog to dog. Some owners establish a routine of a sequence of commands: sit! stay! (long torturous pause), okay! In that routine, one might have to wait an awfully long time before being given the okay to go at the food. Dogs put up with this game of ours with admirable self-possession. But if the owner starts chatting with someone else in the room—busying himself with someone else's attention—why, the game is off.

Lest you think that you can use this knowledge to trick your dog into behaving himself while you are at work by simply pretending to be home with him—over speakerphone or video—one experiment brings very disappointing news. When a life-sized video image (in visible digital) of the owner was displayed before dogs, they disobeyed at levels befitting being home alone with no supervision. While they could use their video-owners' pointing hints to help find food, they didn't bother to follow many of their verbal commands. Dogs are dutiful, but more selectively dutiful when the owner is reduced to a videotape. You cannot hope to reduce your dog's lonely wailing by telling him to stop over the answering machine—but you might be able to tell him where he can find that treat you left out for him. When you next visit the zoo, check in on the monkey cages. Maybe there are capuchin monkeys, quick-moving, tail-flaunting animals who leap easily and shriek piercingly. Or colobus, slow-moving leaf-eaters whose black-and-white coats often hide a small colobus clinging inside. Watch the male snow monkeys as they follow around the red-bottomed females. There is much to recognize here in our distant evolutionary cousins. We see their interests, their fears, their lusts. And most will notice and respond to you—most likely by moving farther away, or turning their heads to avoid your gaze. What is surprising is that dogs, so much less humanlike than these primates, are so much better at realizing what is behind our gaze, how to use it to get information or to their advantage. Dogs can see us as our primate cousins cannot.

Canine Anthropologists

I am I because my little dog knows me.

—GERTRUDE STEIN The dog's gaze is an examination, a regard: a gaze at another animate creature. He sees us, which might imply that he thinks about us—and we like to be considered. Naturally we wonder, in that moment of shared gaze, Is the dog thinking about us the way we are thinking about the dog? What does he know about us?

We are known by our dogs—probably far better than we know them. They are the consummate eavesdroppers and peeping Toms: let into the privacy of our rooms, they quietly spy on our every move. They know about our comings and goings. They come to know our habits: how long we spend in the bathroom, how long we spend in front of the television. They know whom we sleep with; what we eat; what we eat too much of; whom we sleep too much with. They watch us like no other animal watches us. We share our homes with uncounted numbers of mice, millipedes, and mites: none bothers to look our way. We open our door and see pigeons, squirrels, and assorted flying bugs; they barely notice us. Dogs, by contrast, watch us from across the room, from the window, and out of the corner of their eyes. Their watching is enabled by a subtle but powerful ability that begins with simple vision. Sight is used to pay visual attention, and visual attention is used to see what we attend to. In some ways this is similar to us, but in other ways it surpasses human capacity.

The blind and the deaf sometimes keep dogs to see or hear the world for them. For some disabled persons, a dog may enable movement through a world they cannot navigate alone. Just as for the physically impaired dogs can act as eyes, ears, and feet, so also do they act as readers of human behavior for some autistic individuals. Persons with any kind of autism spectrum disorder are united by their shared inability to understand the expressions, emotions, and perspectives of other people. As the neurologist Oliver Sacks describes, for an autistic person who keeps dogs, the dogs may seem to be human-mind-readers. While an autistic person cannot parse a brow furrowed with concern, or interpret the rising tone indicating someone's fright or worry, the dog is sensitive to the mind-set behind them.

Dogs are anthropologists among us. They are students of behavior, observing us in the way that the science of anthropology teaches its practitioners to look at humans. As adults, we walk among other humans largely without examining them closely, socially trained to keep to ourselves. Even with those we know best, we might stop attending to the minute changes in their expressions, their moods, their outlooks. The Swiss psychologist Jean Piaget suggested that as children we are little scientists, forming theories about the world and testing them by acting. If so, we are scientists who hone our skills only to later neglect them. We mature by learning how people behave, but eventually we pay less attention to how others are behaving at every instant. We outgrow the habit of looking. A curious child stares with fascination at the stranger limping down the street: he will be taught this is not polite. A child might be enraptured by a swirl of fallen leaves on the pavement; by adulthood, he will overlook it. The child wonders at our crying, monitors our smiles, looks where we look; with age we are all still able to do all this, but we fall out of the habit.

Dogs don't stop looking—at the gimpy walk, at a rush of leaves tumbling down the sidewalk, at our faces. The urban dog may be bereft of natural sights, but he is rich in the odd: the drunken man swerving through the crowd, the shouting sidewalk preacher, the lame and destitute. All get long stares from the dogs who pass them. What makes dogs good anthropologists is that they are so attuned to humans: they notice what is typical, and what is different. And, just as crucially, they don't become inured to us, as we do—nor do they grow up to be us.

DOGS' PSYCHIC POWERS DECONSTRUCTED

This attunement to us feels magical. Dogs are able to anticipate us—and, it seems, to know something essential about us and others. Is this clairvoyance? A sixth sense?

I am reminded of the story of a horse. At the turn of the twentieth century, the actions of the horse Hans, whose ironic sobriquet, "Clever Hans," has come to stand both for what he could not do, and as a warning against overattributions to animals, helped shape the course of animal cognition research for the next hundred years.

Hans, his owner claimed, could count. Shown an arithmetic problem written on a blackboard, Hans tapped out the sum on the earth with his hoof. Though he had been encouraged and reinforced, using straightforward conditioning, for tapping, this was not a rote response to predetermined questions: he was excellent at all sums, with novel problems, and even when the questioner was someone other than his trainer.

Such was the tenor of the time that this discovered, presumed latent cognitive ability of horses created a small furor. Animal trainers and academics alike were stumped as to how Hans was doing it. It almost looked as though there were no other explanation than that he was actually doing arithmetic.

Finally, the trick—an inadvertent trick, unknown even to his owner—was discovered by a psychologist named Oskar Pfungst. When the questioner himself was prevented from knowing the answer to the problem, Hans's math was wildly off. Hans was not counting, and he was not psychic; he was simply reading the behavior of his questioners. They unconsciously tipped him off to the answer through small bodily movements: leaning forward or away from the horse when he'd tapped the correct sum; relaxing their shoulders and muscles of their face; inclining ever so slightly until the answer was reached.

Clever Hans stood, and still stands, as a cautionary tale against assigning to animals abilities that could be explained by simpler mechanisms. But thinking about the dog's use of attention reminds me of Hans's skill. While Hans was not clever in the way advertised, he was remarkably clever at reading the inadvertent signals given by the persons quizzing him. Before an audience of hundreds, only Hans noticed his trainer leaning forward, his body tensing and relaxing, which Hans had figured out meant that he was to stop tapping his hoof. He attended to the very cues that had information: an attention far greater than the human spectators brought to the event.

Hans's preternatural sensitivity may have ensued, paradoxically, from other deficits. Since he presumably didn't have any notion about numbers or arithmetic, he was not distracted by those stimuli. By contrast, our attention to the seemingly salient details would lead us to miss the one clear indication of the answer.

An experimental psychologist I've met who does research with pigeons demonstrates this phenomenon when teaching his undergraduate class. He shows the students a series of slides of bar graphs with blue bars of various lengths set against a white background. The slides fall into one of two categories, he says: those that have some unspecified "x-ness" feature, and those that don't. He points out which slides are members of the x-ness category. He then puts it to the students to use the example slides to figure out what the conditions for x-ness are.

After many minutes of frustrated and failing attempts by the students, he reveals that pigeons trained on a set of members with x-ness can without fail tell of a new graph whether it satisfies this elusive criterion or not. The students shift uncomfortably in their seats. Still, not a person comes up with the answer. Finally, their professor fills them in: those slides that are mostly blue are members of the x-ness category; those that are more white are not.

The students are outraged: they've been outsmarted by pigeons. In running this test with my psychology classes, I find they also decry the task. Though no student has ever come up with the answer, they all later complain that the answer is unfair. They were looking for some complicated relationship between the bars—one consistent with the kinds of relationships between features that bar graphs are meant to represent. But there is none. "X-ness" is simply "more blue." Only pigeons, blissfully unaware of bar graphs, saw them for their colors and perceived the true categories.

What dogs do is a version of what Hans and the pigeons did. Anecdotal tales of this kind of phenomenon abound. One trainer of search-and-rescue dogs put his hands on his hips in exasperation when the dog was following the wrong path. Another rubbed his chin uneasily. In both cases, the dogs learned to use the cues of their trainers as information that they were off the trail. (The trainers had to be trained to tone down their cuing.) As we look for the more complex explanation for an event, or for others' behavior, we may overlook clues that dogs see naturally. It is less extrasensory perception than the well-added sum of their ordinary senses. Dogs use their sensory skills in combination with their attention to us. Without their interest in our attention, they would not perceive the subtle differences in our strides and body postures and stress levels as important bits of information. It allows them to predict us and to reveal us.

READING US

The dog observes us, thinks about us, knows us. Do they then have some special knowledge about us, born of their attention to us and to our attention? They do.

In a nonverbal way, dogs know who we are, they know what we do, and they know some things about us unknown to ourselves. We're knowable by look, and even more so by our smells. Over and above that, how we act defines who we are. Part of my recognition of Pump is not just of her visage; it is of her walk: her slightly off-kilter, jaunty trot with her droopy ears bouncing in step. For dogs, too, the identity of a person is not just how she smells and looks; it is how she moves. We are recognizable by our behavior.

Even our most ordinary behavior—walking across a room in our characteristic style—is chock full of information that the dog can mine. All dog owners watch their pups' growing sensitivity to the rituals that precede going for what in many dog-peopled households is called a W-A-L-K.*

Dogs quickly learn to recognize shoe donning, of course; we come to expect that grabbing a leash or a jacket will clue them in; a regular walk time explains their prescience; but what if all you did was look up from your work or rise from your seat before your dog was on to you?

If done suddenly, or if you cross the room with a purposeful stride, an attentive dog has all the information he needs. Habitual watcher of your behavior, he sees your intent even when you think you are giving nothing away. As we've seen, dogs are very sensitive to gaze and thus to changes in our gaze. The difference between a head lifted up or angled down, away from them or toward them, is large for an animal so sensitive to eye contact. Even small movements of the hands or adjustments of the body attract notice. Spend three hours looking at a computer screen, hands tethered to the keyboard, then look up and stretch your arms overhead—this is a metamorphosis! The redirection of your attention is clear—and a hopeful dog can easily interpret it as a prelude to a walk. An acute human observer would notice this, too, but we rarely let others oversee us so closely in our daily affairs. (Nor do we find it terribly interesting to so oversee.)

Their facility at anticipation of our actions is part anatomy and part psychology. Their anatomy—all those rod photoreceptors—allows them a millisecond head start on noticing motion. They react before we see that there is something to react to. The critical psychologies are of anticipation—predicting the future from the past—and of association. Familiarity with your typical movements is necessary in order to so anticipate you: a new puppy might not be tricked by a feigned tennis-ball toss, but with age he will be. Even without familiarity, dogs are skilled at making associations between events—between the arrival of one's mother and the delivery of food; between a shift in your focus and the promise of a walk.

Dogs pick up the theme of our quotidian habits, and thus are especially sensitive to variations in them. Just as we often take the same route to our cars, to work, to the subway, we take our dogs on similar walks. Over time, they learn the route themselves, and can anticipate that we turn left past the hedgerow and make a sharp right at the corner with the fireplug. If we introduce a new detour on our way home, even an unnecessary one—circling around the block an additional go—dogs adjust to the new route after just a few outings. And they even begin heading in the direction of the detour before their owner makes any movements in that direction. This makes them fine, cooperative walking companions—better than many humans I perambulate the city with, whom I constantly knock into as I lead them on a preferred route.

The complement of dogs' anticipatory prowess is their purported character-reading ability. Plenty of people let their dogs choose their potential romantic partners. Others declare their dog a good judge of character, able to spot a duplicitous person, a bad sort, on first meeting. They may seem to recognize someone who is not to be trusted.* What this ability might come down to is their close looking at our looking. If you feel hesitant about an approaching stranger, you reveal it, however unintentionally. Dogs are, as we have seen, sensitive to the olfactory changes that come with stress; they can also notice tensing muscles and the auditory change of quick breathing or gasping. (These physiological changes are among those measured by lie detectors: one might imagine that a trained dog could substitute both for the machine and its technician.) But they will let their visual ability trump these notes when assessing a new person or trying to solve a problem. We all have characteristic behaviors we display when angry, nervous, or excited. "Untrustworthy" people often glance furtively in conversation. Dogs notice this gaze. An aggressive stranger may make bold eye contact, move unnaturally slowly or quickly, or veer oddly from a straight path before doing any actual aggression. Dogs notice the behavior; they react viscerally to the meeting of eyes.

One winter we took a trip north, to a place of assertive winter and genuine cold, and were treated to a large snowstorm. We pulled out sleds, found a great big hill, and proceeded to plow an erratic track down it. Pump was suddenly overcome, and ferociously pursued us on each ride downhill, biting, grabbing, and growling at our faces. When it was my fast-moving, snow-covered face being attacked, I couldn't stop her for all my laughing. She was playing, but it was a play I have not seen before: tinged with real aggression. When I managed to rise and shake off the covering of snow I'd gathered on the way down, she calmed at once.

Does this clairvoyance mean that dogs can't be fooled? No. They are astute watchers, but they are not mind readers, nor are they immune from being misled. I was changed for Pump when I leapt on that sled: I was horizontal; I was enrobed in snow gear and snow; and most critically, I was moving entirely differently. I was suddenly a smoothly moving, high-speed prey animal, not an upright, ambling companion.

My dog may have a particular interest in sledders, but her behavior is similar to many other dogs' chasing behaviors. Dogs often chase bicyclists, skateboarders, RollerBladers, cars, or runners. The general-purpose answer given for why they do this is usually that they have an instinct to chase prey. This answer is not entirely wrong, but it is mightily incomplete. It is not quite that dogs think of these objects or persons as "prey," per se. Your motion reveals another dimension to you: you roll! quickly! It is an attribute that alters you in the dog's eyes, which are especially responsive to a certain kind of motion. Mounting and riding a bicycle, you have not turned to prey—as indicated by the fact that your dog greets you, not eats you, on dismount. Their responsive sensitivity probably evolved as a prey-detection tactic, but it will be applied variously. It lends to the dog's experience an additional way to interpret objects and animals in the environment. That way is by the quality of their motion.

There are shared components of sledding, bicycling, or running: a person is moving in a certain way—smoothly and quickly. Walkers are moving, but not quickly: they are not chased. Pump did not recognize me sledding because ordinarily, much as I would like to think otherwise, I am not particularly smooth nor quick in my motions. There is an excess of vertical movement in my walk; I weave to and fro; I gesture a lot—all frivolous in making forward progress.

To stop a dog pursuing a bicycle with a predatory glint in his eye, one can simply interrupt the illusion: stop the bike. The chasing impulse triggered by the visual cells that detected the motion will itself let up. (The hormones involved in the arousal of barking and chasing such a smooth and quick mover may still be coursing through his system, though, for a few minutes.)

Science has confirmed the importance of behavior in identity. Our identities, who we are, are defined partly by our actions, so we can examine how actions inform recognition of personal identity. In one experiment, dogs showed that they have no difficulty distinguishing friendly and unfriendly strangers: those demonstrating different identities. To do this, the experimenters divided participants into two groups and asked members of each group to behave in a prescribed manner. Friendly behavior included walking at a normal speed, talking to the dog in a cheerful voice, and gently petting the dog. Unfriendly behavior included actions that could be interpreted as threatening: an erratic, hesitating approach combined with staring at the eyes of the dog without talking.

The main result of the experiment is not all that surprising: the dogs approached the friendly and avoided the unfriendly. But there's a hidden gem in the experiment. The key trial was this: How did the dogs act when a formerly friendly person suddenly acted threatening? The dogs acted variably: For some, the person was now a different kind of person altogether—an unfriendly one, her identity changed. To others, the olfactory recognition of the stranger who had been friendly trumped this new odd behavior.

These people began as strangers to the dog, but over the course of the sessions, dogs became familiar with the various people: they became "less strange." Their identity was defined in part by their smell and in part by their behavior.