HOW WOLVES BECAME DOGS

Though we tend not to think much about it, the history of dogs, well before you got your dog, bears more on what your dog is like than the particulars of his parentage. Their history begins with wolves.

Wolves are dogs before the accoutrements. The coat of domestication makes dogs quite different creatures, however.* While a pet dog gone missing may not survive even a handful of days on his own, the anatomy, instinctual drive, and sociality of the wolf combine to make it very adaptable. These canids can be found in diverse environments: in deserts, forests, and on ice. For the most part, wolves live in packs, with one mating pair and from four up to forty younger, usually related wolves. The pack works cooperatively, sharing tasks. Older wolves may help raise the youngest pups, and the whole group works together when hunting large prey. They are very territorial and spend a good amount of time demarcating and defending their borders.

Inside some of these borders, tens of thousands of years ago, human beings began to appear. Homo sapiens, having outgrown his habilis and erectus forms, was becoming less nomadic and beginning to create settlements. Even before agriculture began, interactions between humans and wolves began. Just how those interactions played out is the source of speculation. One idea is that the humans' relatively fixed communities produced a large amount of waste, including food waste. Wolves, who will scavenge as well as hunt, would have quickly discovered this food source. The most brazen among them may have overcome any fear of these new, naked human animals and begun feasting on the scraps pile. In this way, an accidental natural selection of wolves who are less fearful of humans would have begun.

Over time, humans would tolerate the wolves, maybe taking a few pups in as pets, or, in leaner times, as meat. Generation by generation, the calmer wolves would have more success living on the edge of human society. Eventually, people would begin intentionally breeding those animals they particularly liked. This is the first step of domestication, a remaking of animals to our liking. With all species, this process typically occurs through a gradual association with humans, whereby successive generations become more and more tame and finally become distinct in behavior and body from their wild ancestors. Domestication is thus preceded by a kind of inadvertent selection of animals who are nearby, useful, or pleasing, allowing them to loiter on the edges of human society. The next step in the process involves more intention. Those animals who are less useful or liked are abandoned, destroyed, or deterred from hanging about with us. In this way, we select those animals who more easily submit to our breeding of them. Finally, and most familiar, domestication involves breeding animals for specific characteristics.

Archeological evidence dates the first domesticated wolf-cum-dog at ten thousand to fourteen thousand years ago. Dog remains have been found in trash heaps (suggesting their use as food or property) and in grave sites, their skeletons curled up aside human skeletons. Most researchers think dogs began to associate with us even earlier, maybe many tens of thousands of years ago. There is genetic evidence, in the form of mitochondrial DNA samples,* of a subtle split as long as 145,000 years ago between pure wolves and those that were to become dogs. We could call the latter wolves protodomesticators, since they had themselves changed behaviorally in ways that would later encourage humans' interest (or merely tolerance) of them. By the time humans came along, they might have been ripe for domesticating. The wolves taken up by humans were probably less hunters than scavengers, less dominant and smaller than alpha wolves, and tamer. In sum, less wolfy. Thus, early in the development of ancient civilizations, thousands of years before domesticating any other animal, humans took this one animal with them inside the walls of their fledgling villages.

These vanguard dogs would not be mistaken as members of one of the hundreds of currently recognized dog breeds. The short stature of the dachshund, the flattened nose of the pug—these are the results of selective breeding by humans much later. Most dog breeds we recognize today have only been developed in the last few hundred years. But these early dogs would have inherited the social skills and curiosity of their wolf ancestors, and would then have applied them toward cooperating with and appeasing humans as much as toward each other. They lost some of their tendency toward pack behavior: scavengers don't need the proclivity to hunt together. Nor is any hierarchy relevant when you might live and eat on your own. They were sociable but not in a social hierarchy.

The change from wolf to dog was striking in its speed. Humans took nearly two million years to morph from Homo habilis to Homo sapiens, but the wolf leapfrogged into dogness in a fraction of the time. Domestication mirrors what nature, through natural selection, does over hundreds of generations: a kind of artificial selection that hurries up the clock. Dogs were the first domesticated animals, and in some ways the most surprising. Most domestic animals are not predators. A predator seems like an unwise choice to take into one's home: not only would it be difficult to find provisions for a meat eater, one risks being seen as meat oneself. And though this might make them (and has made them) good hunting pals, their main role in the last hundred years has been to be a friend and nonjudgmental confidant, not a worker.

But wolves do have features that made them terrific candidates for artificial selection. The process favors a social animal who is behaviorally flexible, able to adjust its behavior in different settings. Wolves are born into a pack, but only stay until they are a few years old: then they leave and find a mate, create a new pack, or join an already existing pack. This kind of flexibility to changing status and roles is well suited to dealing with the new social unit that includes humans. Within a pack or moving between packs, wolves would need to be attentive to the behavior of packmates—just as dogs will need to be attentive to their keepers and sensitive to their behavior. Those early wolf-dogs meeting early human settlers would not have benefited the humans much, so they must have been valued for some other reason—say, for their companionship. The openness of these canids allowed them to adjust to a new pack: one that would include animals of an entirely different species.

UNWOLFY

And so some wolflike ancestor of both wolves and dogs took the plunge, loitered among human loiterers, and was eventually adopted and then molded by humans instead of solely by the caprice of nature. This makes present-day wolves an interesting comparison species to dogs: they likely share many traits. The present-day wolf is not the ancestor of the dog; though wolves and dogs share a common ancestor. Even the modern wolf is likely quite different than the ancestral wolves. What is different between dogs and wolves is probably due to what made some protodogs likely to be taken in, plus whatever humans have done in breeding them since.

And there are many differences. Some are developmental: for instance, dogs' eyes don't open for two or more weeks, whereas wolf pups open their eyes at ten days old. This slight difference can have a cascading effect. Generally, dogs are slower to develop physically and behaviorally. The big developmental milestones—walking, carrying objects in the mouth, when they first engage in biting games—come generally later for dogs than for wolves.* This small difference blossoms into a large difference: it means that the window for socialization is different in dogs and wolves. Dogs have more leisure to learn about others and to become accustomed to objects in their environment. If dogs are exposed to non-dogs—humans or monkeys or rabbits or cats—in the first few months of development, they form an attachment to and preference for these species over others, often trumping any predatory or fearful drive we might expect them to feel. This so-called sensitive or critical period of social learning is the time during which dogs will learn who is a dog, an ally, or a stranger. They are most susceptible to learning who their peers are, how to behave, and associations between events. Wolves have a smaller window during which to determine who is familiar and who is foe.

There are differences in social organization: dogs do not form true packs; rather, they scavenge or hunt small prey individually or in parallel.* Though they don't hunt cooperatively, they are cooperative: bird dogs and assistance dogs, for instance, learn to act in synchrony with their owners. For dogs, socialization among humans is natural; not so for wolves, who learn to avoid humans naturally. The dog is a member of a human social group; its natural environment, among people and other dogs. Dogs show what is called with human infants "attachment": preference for the primary caregiver over others. They have anxiety at separation from the caregiver, and greet her specially on her return. Though wolves greet other members of the pack when they reunite after being apart, they don't seem to show attachment to particular figures. For an animal who is going to be around humans, specific attachments make sense; for an animal who lives in a pack, it is less applicable.

Physically, dogs and wolves differ. While still quadrupedal omnivores, the range of body types and sizes among dogs is extraordinary. No other canid, or other species, shows the same diversity of body types within a species, from the four-pound papillon to the two-hundred-pound Newfoundland; from slender dogs with long snouts and whiplike tails to pudgy dogs with foreshortened noses and stubs of tails. Limbs, ears, eyes, nose, tail, fur, haunches, and belly are all dimensions along which dogs can be reconfigured and still be dogs. Wolves' sizes, by contrast, are, like most wild animals, fairly reliably uniform in a particular environment. But even the "average" dog—something resembling a prototypical mutt—is distinguishable from the wolf. The dog's skin is thicker than wolves'; while both have the same number and kind of teeth, the dog's are smaller. And the whole head is smaller on a dog than on a wolf: about 20 percent smaller. In other words, between a dog and a wolf of similar body size, the dog has the much smaller skull—and, correspondingly, a smaller brain.

This latter fact has continued to be promulgated, perhaps an indication of the ongoing appeal of the claim (now debunked) that brain size determines intellect. While erroneous, the smoothness of the shift from talking about brain size to brain quality trumped evidence to the contrary. Comparative studies with wolves and dogs on problem-solving tasks initially seemed to confirm dogs' cognitive inferiority. Hand-raised wolves tested on their ability to learn a task—to pull three ropes from an array of ropes in a particular order—well outperformed the dogs tested. The wolves more quickly learned to pull any rope to begin and then proceeded to be more successful at learning the order in which the ropes were to be pulled. (They also tore more ropes to pieces than the dogs did, although the researchers are silent about what this indicates about their cognition.) Wolves are also great at escaping from enclosed cages; dogs are not. Most canid researchers agree that wolves pay more attention than dogs do to physical objects and handle these objects more capably.

From results like these comes the notion that there is a cognitive difference between wolves and dogs: usually, that wolves are insightful problem solvers, and dogs simpletons. In actual fact, historically theories have oscillated between claiming dogs to be more intelligent, or wolves to be the smarter of the two. Science is often contingent on the culture in which it is practiced, and these theories reflect the then-prevailing ideas about animal minds. The accumulated data of dog and wolf behavior, though, leads to a more nuanced position. Wolves seem to be better at solving certain kinds of physical puzzles. Some of this skill is explainable by looking at their natural behavior. Why did wolves readily learn the rope-pulling task? Well, they do a lot of grabbing and pulling on things (like prey) in their natural environment. Some of the difference can be traced to dogs' more limited requirements for living. Having been folded into the world of humans, dogs no longer need some of the skills that they would to survive on their own. As we'll see, what dogs lack in physical skills, they make up for in people skills.

AND THEN OUR EYES MET …

There is a final, seemingly minor difference between the two species. This one small behavioral variation between wolves and dogs has remarkable consequences. The difference is this: dogs look at our eyes.

Dogs make eye contact and look to us for information—about the location of food, about our emotions, about what is happening. Wolves avoid eye contact. In both species, eye contact can be a threat: to stare is to assert authority. So too is it with humans. In one of my undergraduate psychology classes, I have my students do a simple field experiment wherein they try to make and hold eye contact with everyone they pass on campus.

Both they and those on the receiving end of their stares behave remarkably consistently: everyone can't wait to break eye contact. It's stressful for the students, a great number of whom suddenly claim to be shy: they report that their hearts begin to race and they start sweating when simply holding someone's gaze for a few seconds. They concoct elaborate stories on the spot to explain why someone looked away, or held their gaze for a half second longer. For the most part, their staring is met with deflected gazes from those they eyeball. In a related experiment, they test gaze in a second way, verifying our species' tendency to follow the gaze of others to its focal point. A student approaches any publicly visible and shared object—a building, tree, spot on the sidewalk—and looks fixedly at one point on it. Her partner, another student, stands nearby and surreptitiously records the reactions of passersby. If it's not rush hour and raining, they report finding that at least some people stop in their tracks to follow their gaze and stare curiously at that fascinating sidewalk spot: surely there must be something.

If this behavior is unsurprising, it is because it is so human: we look. Dogs look, too. Though they have inherited some aversion to staring too long at eyes, dogs seem to be predisposed to inspect our faces for information, for reassurance, for guidance. Not only is this pleasing to us—there is a certain satisfaction in gazing deep into a dog's eyes gazing back at you—it is also perfectly suited to getting along with humans. As we will see later in this book, it also serves as a foundation for their skill at social cognition. We not only avoid eye contact with strangers, we rely on eye contact with intimates. There is information in a furtive glance; a gaze mutually held feels profound. Eye contact between people is essential to normal communication.

Hence a dog's ability to find and gaze at our eyes may have been one of the first steps in the domestication of dogs: we chose those that looked at us. What we then did with dogs is peculiar. We began designing them.

FANCY DOGS

The label on her cage said "Lab mix." Every dog in the shelter was a Lab mix. But surely Pump was born of a spaniel: her black, silky hair fell against her slender frame; her velvet ears framed her face. In sleep she was a perfect bear cub. Soon her tail hairs grew longer and feathery: so she's a golden retriever. Then the gentle curls on her underbelly tightened; her jowls filled a bit: okay, she's a water dog. As she ages her belly grows until she has a solid, barrel-like shape—she's a Lab after all; her tail becomes a flag needing trimming—a Lab/golden mix; she could be still one moment and sprinting the next—a poodle. She is curly and round-bellied: clearly the product of a sheepdog who'd snuck into the bushes with a pretty sheep. She's her own dog.

The original dogs were mongrels, in the sense that they didn't come from a controlled lineage. But many of the dogs we keep, mutts or not, emerged from hundreds of years of strictly controlled breeding. The consequence of this breeding is the creation of what are nearly subspecies, varying in shape, size, lifetime, temperament,* and skills. The outgoing Norwich terrier, ten inches high and ten pounds strong, is but the weight of the calm, sweet, enormous Newfoundland's head. Ask some dogs to retrieve a ball and you get a puzzled look; but a border collie doesn't need to be asked twice.

The familiar differences between modern breeds aren't always the result of intentional selection. Some behaviors and physical features are selected for—retrieving prey, smallness, a tightly curled tail—and some just come along for the ride. The biological reality of breeding is that the genes for traits and behaviors come in clusters. Mate a few generations of dogs with particularly long ears and you might find that they all share other characteristics: a strong neck, downcast eyes, fine jowls. Coursing dogs, bred to gallop swiftly or long, are leggy—their leg length matches (in the husky) or surpasses (in the greyhound) the depth of their chests. By contrast, dogs who track on the ground (as the dachshund) wound up with legs much shorter than their chest is deep. Similarly, selecting for one particular behavior inadvertently selects for accompanying behaviors. Breed dogs who are very sensitive to motion—who probably have an overabundance of rod photoreceptors in their retinae—and you may also get a dog whose acute sensitivity to motion leads to their being temperamentally high-strung. Their appearance might change, too: they may have large, globular eyes for seeing at night. Sometimes what comes to be desirable in a breed is a trait that first appeared inadvertently.

There is evidence of distinct dog breeds as early as five thousand years ago. In drawings from ancient Egypt, at least two kinds of dogs are depicted: mastiff-looking dogs, big of head and body, and slim dogs with curled tails.* The mastiffs may have been guard dogs; the slender dogs appear to have been hunting companions. And so the designing of dogs for particular purposes began—and continued along these lines for a long while. By the sixteenth century, there were added other hounds, bird dogs, terriers, and shepherds. By the nineteenth century, clubs and competitions sprouted, and the naming and monitoring of breeds exploded.

The various modern breeds have probably all emerged with this proliferation of breeding in the last four hundred years. The American Kennel Club now lists nearly one hundred fifty varieties, grouped according to the purported* occupation of the breed. Hunting companions are distributed into "sporting," "hound," "working," and "terrier" categories; there are, in addition, the working "herding" breeds, the plainly "nonsporting" breeds, and the rather self-explanatory "toys." Even among dogs bred to join the hunt, there are subdivisions, by the very kind of assistance they provide (pointers point out the prey; retrievers retrieve it; Afghan hounds exhaust it); by the specific prey they're after (terriers are ratters, and harriers go after hares); and by the medium preferred (beagles chase on land; spaniels will swim in water). Worldwide, there are hundreds more breeds still. Breeds vary not just by our uses of them but physically: by body size, head size, head shape, body shape, type of tail, coat kind, coat color. Go searching for a purebred dog and you'll confront a new-car-worthy list of specs, detailing everything from the ears to the temperament of your future pup. Want a long-limbed, short-haired, jowly dog? Consider the Great Dane. In more of a short-nosed, rolled-skin, curly-tailed kind of mood? Here's a nice pug for you. Choosing between breeds is like choosing between anthropomorphized option packages. You not only get a dog, you get one who is typically "dignified, lordly, scowling, sober and snobbish" (shar-pei); "merry and affectionate" (English cocker spaniel); "reserved and discerning with strangers" (chow chow); with a "rollicking personality" (Irish setter); full of "self-importance" (Pekingese); having "heedless, reckless pluck" (Irish terrier); "equable" (Bouvier des Flandres); or, most surprisingly, "a dog at heart" (briard).

The dog fanciers will be surprised to hear, perhaps, that the grouping of breeds based on genetic similarity does not result in the same groupings as the AKC. Cairn terriers are closer to hounds; shepherds and mastiffs share much of their genomes. The genome belies most people's assumptions about dogs' similarities to wolves, too: the long-haired, sickle-tailed huskies are closer to wolves than the long-bodied, skulking German Shepherd. Basenjis, who bear almost no physical resemblance to wolves, are closer still. This is yet another indication that, for most of their domestication, the dog's appearance was an accidental side effect of his breeding.

Dog breeds are relatively closed genetic populations, meaning that each breed's gene pool is not accepting new genomes from outside the pool. To be a member of a breed, a dog must have parents who are themselves members. Thus any physical changes in the offspring can only come from random genetic mutations, not from the mixing of different gene pools that usually occurs when animals (including humans) mate. Mutations, variation, and admixtures are generally good for populations, though, and help to prevent inherited disease: this is why purebred dogs, though they come from what is considered "good stock" in that the ancestry of the dogs is traceable through the breeding line, are more susceptible to many physical disorders than are mixed-breed dogs.

One boon of a closed gene pool is that the genome of a breed can be mapped, and in fact it recently has: a boxer's genome was the first, around nineteen thousand genes' worth. As a result, scientists are starting to make an accounting of where on the genome the genetic variations are that lead to characteristic traits and disorders, such as narcolepsy, the sudden and total fall into unconsciousness to which some dog breeds (particularly Dobermans) are susceptible.

Another advantage of a closed gene pool of a breed discussed by researchers is that it feels as though one is getting a relatively reliable animal when one selects from it. One can pick a "family-friendly" dog or one advertised as a skillful house guardian. But it is not so simple: dogs, like us, are more than their genome. No animal develops in a vacuum: genes interact with the environment to produce the dog you come to know. The exact formulation is difficult to specify: the genome shapes the dog's neural and physical development, which itself partially determines what will be noticed in the environment—and whatever is noticed itself further shapes continued neural and physical development. As a result, even with inherited genes, dogs aren't just carbon copies of their parents. On top of this, there is also great natural variability in the genome. Even a cloned dog, should you be tempted to replicate your beloved pet, will not be identical to the original: what a dog experiences and whom he meets will influence who he becomes in innumerable, untraceable ways.

So although we have tried to design dogs, the dogs we see today are partly creatures of serendipity. What breed is she? is a question I've been asked about Pump more than any other—and I in turn ask of other dogs. Her mongrelness encourages the great game of guessing her heritage: the resulting hunches are satisfying, even though none could ever be verified.*

THE ONE DIFFERENCE BETWEEN BREEDS

Though there is an extensive literature of dog breeds, there has never been a scientific comparison of breed behavior differences: a comparison that controls each animal's environment, giving them the same physical objects, the same exposure to dogs and humans, the same everything. It's hard to believe, given that such bold statements are made about what each breed is like. This is not to suggest that the differences are minimal or nonexistent. Dogs of various breeds will doubtless behave differently when, say, they are presented with a nearby, running rabbit. But it would be a mistake to guarantee that a dog, bred or not, will inevitably act a certain way on seeing that rabbit. This is the same mistake that is made when we wind up calling some breeds "aggressive" and legislating against them.*

Even without knowing the specific differences in the Labrador retriever's and the Australian shepherd's reaction to that rabbit, there is one thing that may account for the variability in behavior between breeds. They have different threshold levels to notice and react to stimuli. The same rabbit, for instance, causes different amounts of excitement in two different dogs; similarly, the same amount of hormone producing that excitement causes different rates of response, from raising a head in mild interest, to a full-on chase.

There is a genetic explanation behind this. Though we call a dog a retriever or a shepherd, it is not the behavior retrieving or shepherding that was selected for. Instead, it was the likelihood that the dog would respond just the right amount to various events and scenes. However, there is no one gene we can point to here. No gene develops right into retrieving behavior—or into any particular behavior at all. But a set of genes may affect the likelihood that an animal acts in a certain way. In humans, too, a genetic difference between individuals may appear as different propensities to certain behaviors. One might be more or less susceptible to becoming addicted to stimulant drugs, based partly on how much stimulation one's brain needs to produce a pleasurable feeling. Addictive behavior is thereby traceable to genes that design the brain—but there is no gene for addiction. The environment is clearly important here, too. Some genes regulate expression of other genes—which expression might depend on features of the environment. If raised in a box, without access to drugs, one never develops a drug problem, regardless of one's propensity to addiction.

In the same way, one breed of dog can be distinguished from others by its propensity to respond to certain events. While all dogs can see birds taking flight in front of them, some are particularly sensitive to the small quick motion of something going aloft. Their threshold to respond to this motion is much lower than for dogs not bred to be hunting companions. By comparison to dogs, our response threshold is higher still. We humans can certainly see the birds taking off, but even when they are directly in front of us, we might not notice them. In hunting dogs, the motion is not only noticed, it is directly connected to another tendency: to pursue prey that moves in just that way. And, of course, one must have birds or birdlike things around for this tendency to lead to bird chasing.

Similarly, a sheepdog who will spend his life herding sheep is one who has a certain set of specific tendencies: to notice and keep track of individuals of a group, to detect the errant motion of a sheep moving away from the herd, and to have a drive to keep the herd together. The end result is a herding dog, but his behavior is made up of piecemeal tendencies that shepherds direct toward controlling their sheep. The dog must also be exposed to sheep early in his life, or these propensities wind up being applied not to sheep, but in a disorganized way to young children, to people jogging in the park, or to the squirrels in your yard.

A dog breed that is called aggressive, then, is one that might have a lower threshold to perceive and react to a threatening motion. If the threshold is too low, then even neutral motion—approaching the dog—may be perceived as threatening. But if the dog is not encouraged to follow through on this tendency, it is quite likely that he will never exhibit the aggression that his breed is notorious for.

Knowing the breed of a dog gives us a first-pass entry into understanding something about the dog before we have even met the dog. But it is a mistake to think that knowing a breed guarantees that it will behave as advertised—only that it has certain tendencies. What you get with a mixed-breed dog is a softening of the hard edges seen in breeds. Temperaments are more complex: averaged versions of their bred forebears. In any event, naming a dog's breed is only the beginning of a true understanding of the dog's umwelt, not an endpoint: it doesn't get to what the dog's life is about to the dog.

ANIMALS WITH AN ASTERISK

It's snowing and dawn is breaking, which means we have about three minutes for me to get dressed and get us into the park to play before the snow is trammeled by other merrymakers. Outside, well bundled, I plow clumsily through the deep snow, and Pump hurtles herself through it with great bounds, leaving the footprints of a giant bunny. I plop down to make a snow angel, and Pump throws herself down beside me and seems to be making a snow-dog angel, twisting to and fro on her back. I look to her with complete joy at our shared play. Then I smell a horrible odor coming from her direction. The realization is quick: Pump's not making a snow-dog angel; she's rolling in the decaying carcass of a small animal.

There is a tension between those who consider dogs wild animals at their core and those who consider dogs creatures of our own making. The first group tends to turn to wolf behavior to explain dog behavior. The recently popular dog trainers are admired for their full embrace of the wolf side of dogs. They are often seen mocking the second group, which treats their dogs as quadrupedal, slobbery people. Neither has got it right. The answer is plumb in the middle of these approaches. Dogs are animals, of course, with atavistic tendencies, but to stop here is to have a blinkered view of the natural history of the dog. They have been retooled. Now they are animals with an asterisk.

The inclination to look at dogs as animals rather than creations of our psychology is essentially right. To avoid anthropomorphizing, some turn to what might be called unsympathetic biology: a biology free of subjectivity or such messy considerations as consciousness, preferences, sentiment, or personal experiences. A dog is but an animal, they say, and animals are but biological systems whose behavior and physiology can be explained with simpler, general-purpose terminology. Recently I saw a woman leaving the pet store with her terrier, who himself was newly shod in four tiny shoes—to prevent his bringing the street filth into her house, she explained, as she pulled him skating on rigid limbs down the filthy street. This woman could benefit from more reflection on her dog's animal nature, and less on his resemblance to a stuffed toy. In fact, as we'll see, understanding some of the dogs' complexities—the acuity of their noses, what they can see and cannot see, their loss of fearfulness, and the simple affect of a wag—goes a long way to understanding dogs.

On the other hand, in a number of ways, calling a dog just an animal, and explaining all dog behavior as emerging from wolf behavior, is incomplete and misleading. The key to dogs' success living with us in our homes is the very fact that dogs are not wolves.

For instance, it is high time we revamp the false notion that our dogs view us as their "pack." The "pack" language—with its talk of the "alpha" dog, dominance, and submission—is one of the most pervasive metaphors for the family of humans and dogs. It originates where dogs originated: dogs emerged from wolflike ancestors, and wolves form packs. Thus, it is claimed, dogs form packs. The seeming naturalness of this move is belied by some of the attributes we don't transfer from wolves to dogs: wolves are hunters, but we don't let our dogs hunt for their own food.* And though we may feel secure with a dog at the threshold of a nursery, we would never let a wolf alone in a room with our sleeping newborn baby, seven pounds of vulnerable meat.

Still, to many, the analogy to a dominance-pack organization is terribly appealing—especially with us as dominant and the dog submissive. Once applied, the popular conception of a pack works itself into all sorts of interactions with our dogs: we eat first, the dog second; we command, the dog obeys; we walk the dog, the dog doesn't walk us. Unsure how to deal with an animal in our midst, the "pack" notion gives us a structure.

Unfortunately, it not only limits the kind of understanding and interaction we can have with our dogs, it also relies on a faulty premise. The "pack" evoked in this way bears little resemblance to actual wolf packs. The traditional model of the pack was that of a linear hierarchy, with a ruling alpha pair and various "beta" and even "gamma" or "omega" wolves below them, but contemporary wolf biologists find this model far too simplistic. It was formed from observations of captive wolves. With limited space and resources in small, enclosed pens, unrelated wolves self-organize, and a hierarchy of power results. The same might happen in any social species confined with little room.

In the wild, wolf packs consist almost entirely of related or mated animals. They are families, not groups of peers vying for the top spot. A typical pack includes a breeding pair and one or many generations of their offspring. The pack unit organizes social behavior and hunting behavior. Only one pair mates, while other adult or adolescent pack members participate in raising the pups. Different individuals hunt and share food; at times, many members together hunt large prey which may be too large to tackle individually. Unrelated animals do occasionally join together to form packs with multiple breeding partners, but this is an exception, probably an accommodation to environmental pressures. Some wolves never join a pack.

The one breeding pair—parents to all or most of the other pack members—guides the group's course and behaviors, but to call them "alphas" implies a vying for the top that is not quite accurate. They are not alpha dominants any more than a human parent is the alpha in the family. Similarly, the subordinate status of a young wolf has more to do with his age than with a strictly enforced hierarchy. Behaviors seen as "dominant" or "submissive" are used not in a scramble for power, they are used to maintain social unity. Rather than being a pecking order, rank is a mark of age. It is regularly on display in the animals' expressive postures in greeting and in interaction. Approaching an older wolf with a low wagging tail and a body close to the ground, a younger wolf is acknowledging the older's biological priority. Young pups are naturally at a subordinate level; in mixed-family packs pups may inherit some of the status of their parents. While rank may be reinforced by charged and sometimes dangerous encounters between pack members, this is rarer than aggression against an intruder. Pups learn their place by interacting with and observing their packmates more than by being put in their place.

The reality of wolf pack behavior contrasts starkly with dog behavior in other ways. Domestic dogs do not generally hunt. Most are not born into the family unit in which they will live: with humans the predominant members. Pet dogs' attempts to mate are (happily) unrelated to their adopted humans'—supposedly the alpha pair's—mating schedules. Even feral dogs—those who may never have lived in a human family—usually do not form traditional social packs, although they may travel in parallel.

Neither are we the dog's pack. Our lives are so much more stable than that of a wolf pack: the size and membership of a wolf pack is always in flux, changing with the seasons, with the rates of offspring, with young adult wolves growing up and leaving in their first years, with the availability of prey. Typically, dogs adopted by humans live out their lives with us; no one is pushed out of the house in spring or joins us just for the big winter moose hunt. What domestic dogs do seem to have inherited from wolves is the sociality of a pack: an interest in being around others. Indeed, dogs are social opportunists. They are attuned to the actions of others, and humans turned out to be very good animals to attune to.

To evoke the outdated, simplistic model of packs glosses over real differences between dog and wolf behavior and misses some of the most interesting features of packs in wolves. We do better to explain dogs' taking commands from us, deferring to us, and indulging us by the fact that we are their source of food than by reasoning that we are alpha. We can certainly make dogs totally submissive to us, but that is neither biologically necessary nor particularly enriching for either of us. The pack analogy does nothing but replace our anthropomorphisms with a kind of "beastomorphism," whose crazy philosophy seems to be something like "dogs aren't humans, so we must see them as precisely unhuman in every way."

We and our dogs come closer to being a benign gang than a pack: a gang of two (or three or four or more). We are a family. We share habits, preferences, homes; we sleep together and rise together; we walk the same routes and stop to greet the same dogs. If we are a gang, we are a merrily navel-gazing gang, worshiping nothing but the maintenance of our gang itself. Our gang works by sharing fundamental premises of behavior. For instance, we agree to rules of conduct in our home. I agree with my family that under no circumstances is urination on the living room rug allowed. This is a tacit agreement, happily. A dog has to be taught this premise for habitation; no dog knows about the value of rugs. In fact, rugs might provide a nice feeling underfoot for some bladder release.

Trainers who espouse the pack metaphor extract the "hierarchy" component and ignore the social context from which it emerges. (They further ignore that we still have a lot to learn about wolf behavior in the wild, given the difficulty of following these animals closely.) A wolfcentric trainer may call the humans the pack leaders responsible for discipline and forcing submission by others. These trainers teach by punishing the dog after discovery of, say, the inevitable peed-upon rug. The punishment can be a yell, forcing the dog down, a sharp word or jerk of the collar. Bringing the dog to the scene of the crime to enact the punishment is common—and is an especially misguided tactic.

This approach is farther from what we know of the reality of wolf packs and closer to the timeworn fiction of the animal kingdom with humans at the pinnacle, exerting dominion over the rest. Wolves seem to learn from each other not by punishing each other but by observing each other. Dogs, too, are keen observers—of our reactions. Instead of a punishment happening to them, they'll learn best if you let them discover for themselves which behaviors are rewarded and which lead to naught. Your relationship with your dog is defined by what happens in those undesired moments—as when you return home to a puddle of urine on the floor. Punishing the dog for his misbehavior—the deed having been done maybe hours before—with dominance tactics is a quick way to make your relationship about bullying. If your trainer punishes the dog, the problem behavior may temporarily abate, but the only relationship created is one between your trainer and your dog. (Unless the trainer's moving in with you, that won't last long.) The result will be a dog who becomes extra sensitive and possibly fearful, but not one who understands what you mean to impart. Instead, let the dog use his observation skills. Undesired behavior gets no attention, no food: nothing that the dog wants from you. Good behavior gets it all. That's an integral part of how a young child learns how to be a person. And that's how the dog-human gang coheres into a family.

CANIS UNFAMILIARIS

On the other hand, let's not forget that it is only tens of thousands of years of evolution that separate wolves and dogs. We would have to go back millions of years to trace our split from chimpanzees; appropriately, we do not look to chimpanzee behavior to learn how to raise our children.* Wolves and dogs share all but a third of 1 percent of their DNA. We see occasional snatches of wolf in our pets: a glimpse of a growl when you move to extract a beloved ball from your dog's mouth; rough-and-tumble play in which one animal seems more prey than playmate; a glimmer of wildness in the eye of a dog grabbing for a meat bone.

The orderliness of most of our interactions with dogs clashes mightily with their atavistic side. Once in a while it feels as if some renegade ancient gene takes a hold of the domesticated product of its peers. A dog bites his owner, kills the family cat, attacks a neighbor. This unpredictable, wild side of dogs should be acknowledged. The species has been bred for millennia, but it evolved for millions of years before that without us. They were predators. Their jaws are strong, their teeth designed for tearing flesh. They are wired to act before contemplating action. They have an urge to protect—themselves, their families, their turf—and we cannot always predict when they will be prompted to be protective. And they do not automatically heed the shared premises of humans living in civilized society.

As a result, the first time your dog tears from your side, running maniacally off the trail after some invisible thing in the bushes, you panic. With time, you will become familiar with each other: they, with what you expect of them; you, with what they do. It is only off the trail to you; to the dog it is a natural continuation of walking, and he will learn about trails in time. You may never see the invisible thing in the bush, but you learn, after a dozen walks, that invisible things are in bushes, and the dog will return to you. Living with a dog is a long process of becoming mutually familiar. Even the dog bite is not a uniform entity. There are bites done out of fear, out of frustration, out of pain, and out of anxiety. An aggressive snap is different than an exploratory mouthing; a play bite is different than a grooming nibble.

Despite their sometime wildness, dogs never revert to wolves. Stray dogs—those who lived with humans but have wandered away or been abandoned—and free-ranging dogs—provisioned with food but living apart from humans—do not take on more wolflike qualities. Strays seem to live a life familiar to city dwellers: parallel to and cooperative with others, but often solitary. They do not self-organize socially into packs with a single breeding pair. They don't build dens for the pups or provide food for them as wolves do. Free-ranging dogs may form a social ordering like other wild canids—but one organized by age more than by fights and strife. Neither hunts cooperatively: they scavenge or hunt small prey by themselves. Domestication changed them.

Even when wolves have been socialized—raised from birth among humans instead of other wolves—they do not turn into dogs. They strike a middle ground in behavior. Socialized wolves are more interested in and attentive to humans than wild-born wolves. They follow human communicative gestures better than wild wolves. But they are not dogs in wolves' clothing. Dogs raised with a human caretaker prefer her company over that of other humans; wolves are less discriminating. Dogs far outpace hand-raised wolves in interpreting human cues. To see a wolf on a leash, sitting and lying down on request, one could be convinced there is little difference between the socialized wolf and dog. To see that wolf in the presence of a rabbit is to see how much difference there still is: the human is forgotten while the rabbit is relentlessly pursued. A dog near that same rabbit may patiently wait, gazing at his owner, to be permitted to run. Human companionship has become dogs' motivational meat.

MAKING YOUR DOG

As you choose a new dog from among a litter or a loud shelter of baying mutts and bring him home, you begin to "make a dog" again, recapitulating the history of domestication of the species. With each interaction, with each day, you define—at once circumscribing and expanding—his world. In the first few weeks with you, the pup's world is, if not entirely a tabula rasa, awfully close to the "blooming, buzzing confusion" that a newborn baby experiences. No dog knows, on first turning his eyes on the person who peeks at him in his shelter cage, what the person expects of him. Many people's expectations, at least in this country, are fairly similar: be friendly, loyal, pettable; find me charming and lovable—but know that I am in charge; do not pee in the house; do not jump on guests; do not chew my dress shoes; do not get into the trash. Somehow, word hasn't gotten to the dogs. Each dog has to be taught this set of parameters for his life with people. The dog learns, through you, the kinds of things that are important to you—and that you want to be important to him. We are all domesticated, too: inculcated with our culture's mores, with how to be human, with how to behave with others. This is facilitated by language, but spoken language is not necessary to achieve it. Instead we need to be alert to what the dog is perceiving and to make our perceptions clear to him.

The first-century Roman encyclopedist Pliny's prodigious Natural History includes a confident statement of fact about the birth of bears. The cubs, he wrote, "are a white and shapeless lump of flesh, little larger than mice, without eyes or hair and only the claws projecting. This lump the mother bears slowly lick into shape." The bear is born, he was suggesting, as nothing but pure undifferentiated matter, and, like a true empiricist, the mother bear makes her pup a bear by licking it. When we brought Pump into our home, I felt I was doing just this: I was licking her into shape. (And not just because there was a lot of licking between us—after all, it was exclusively she who was licking.) It was our way of interacting together that made her who she was, that makes dogs that most people want to live with: interested in our goings and comings, attentive to us, not overly intrusive, playful just at the right times. She interpreted the world through acting on it, by seeing others act, by being shown, and by acting with me on the world—promoted into being a good member of the family. And the more time we spent together, the more she became who she was, and the more we were intertwined.

Sniff

First sniff of the day: Pump wanders into the living room in the morning while I am dishing out her food. She's looking sleepy but her nose is wide awake, stretching every which way as though it's doing morning exercises. She reaches her nose toward the food without committing her body, and sniffs. A look at me. Another sniff. A judgment has been levied. She backs from the bowl and forgives me by nosing my outstretched hand, her whiskers tickling while her moist nose examines my palm. We go outside and her nose is gymnastic, almost prehensile, happily taking in smells that gust by …

We humans tend not to spend a lot of time thinking about smelling. Smells are minor blips in our sensory day compared to the reams of visual information that we take in and obsess over in every moment. The room I'm in right now is a phantasmagoric mix of colors and surfaces and densities, of small movements and shadows and lights. Oh, and if I really call my attention to it I can smell the coffee on the table next to me, and maybe the fresh scent of the book cracked open—but only if I dig my nose into its pages.

Not only are we not always smelling, but when we do notice a smell it is usually because it is a good smell, or a bad one: it's rarely just a source of information. We find most odors either alluring or repulsive; few have the neutral character that visual perceptions do. We savor or avoid them. My current world seems relatively odorless. But it is most decidedly not free of smell. Our own weak olfactory sense has, no doubt, limited our curiosity about what the world smells like. A growing coalition of scientists is working to change that, and what they have found about olfactory animals, dogs included, is enough to make us envy those nose-creatures. As we see the world, the dog smells it. The dog's universe is a stratum of complex odors. The world of scents is at least as rich as the world of sight.

SNIFFERS

… Her ungulate-grazing sniff, nose deep in a patch of good grass, trawling the ground and not coming up for air; the examinatory sniff, judging a proffered hand; the alarm-clock sniff, close enough to my sleeping face to tickle me awake with her whiskers; the contemplative sniff, nose held high in the wake of a breeze. All followed by a half sneeze—just the CHOO, no AH—as though to clear her nostrils of whatever molecule she'd just inhaled …

Dogs don't act on the world by handling objects or by eyeballing them, as people might, or by pointing and asking others to act on the object (as the timid might); instead they bravely stride right up to a new, unknown object, stretch their magnificent snouts within millimeters of it, and take a nice deep sniff. That dog nose, in most breeds, is anything but subtle. The snout holding the nose projects forth to examine a new person seconds before the dog himself arrives on the scene. And the sniffer is not just an ornament atop the muzzle; it is the leading, moist headliner. What its prominence suggests, and what all science confirms, is that the dog is a creature of the nose.

The sniff is the great medium for getting smelly objects to the dog, the tramway on which chemical odors speed up to the waiting receptor cells along the caverns of the dog nose. Sniffing is the action of inhaling air, but it is more active than that, usually involving short, sharp bursts of drawing air into the nose. Everyone sniffs—to clear the nose, to smell dinner cooking, as part of a preparatory inhale. Humans even sniff emotively, or meaningfully—to express disdain, contempt, surprise, and as punctuation at a sentence's end. Animals mostly sniff, as far as we know, to investigate the world. Elephants raise their trunk into the air in a "periscope sniff," tortoises slowly reach and open their nostrils wide in a sniff, marmosets sniff while they nuzzle. Ethologists watching animals often take note of all these sniffs, for they may precede an attempt to mate, a social interaction, aggression, or feeding. They record an animal as "sniffing" when it brings its nose close to—but not touching—the ground or an object, or an object is brought close to—but not touching—the nose. In these cases, they are assuming that the animal is in fact inhaling sharply—but they may not be able to get close enough to the animal to see the nostrils moving, or the tiny vortex of air that stirs the area in front of the nose.

Few have looked closely at exactly what happens in a sniff. But recently some researchers have used a specialized photographic method that shows air flow in order to detect when, and how, dogs are sniffing. They have found that the sniff is nothing to be sniffed at. In fact one could make the case that it is neither a single nor a simple inhalation. The sniff begins with muscles in the nostrils straining to draw a current of air into them—this allows a large amount of any air-based odorant to enter the nose. At the same time, the air already in the nose has to be displaced. Again, the nostrils quiver slightly to push the present air deeper into the nose, or off through slits in the side of the nose and backward, out the nose and out of the way. In this way, inhaled odors don't need to jostle with the air already in the nose for access to the lining of the nose. Here's why this is particularly special: the photography also reveals that the slight wind generated by the exhale in fact helps to pull more of the new scent in, by creating a current of air over it.

This action is markedly different from human sniffing, with our clumsy "in through one nostril hole, out through the same hole" method. If we want to get a good smell of something, we have to sniff-hyperventilate, inhaling repeatedly without strongly exhaling. Dogs naturally create tiny wind currents in exhalations that hurry the inhalations in. So for dogs, the sniff includes an exhaled component that helps the sniffer smell. This is visible: watch for a small puff of dust rising up from the ground as a dog investigates it with his nose.

Given our tendency to find so many smells disgusting, we should all celebrate that our olfactory system adapts to an odor in the environment: over time, if we stay in one place, the intensity of every smell diminishes until we don't notice it at all. The first smell of coffee brewing in the morning: fantastic … and gone in a few minutes. The first smell of something rotting under the porch: nauseating … and gone in a few minutes. The sniffing method of dogs enables them to avoid habituation to the olfactory topography of the world: they are continually refreshing the scent in their nose, as though shifting their gaze to get another look.

THE NOSE NOSE

I crack open her window in the car—just enough to fit a dog-sized head (remembering the time she threw herself completely out the open window after that squirrel hitchhiking on the side of the road). Pump props herself up on the armrest and pokes her muzzle out of the car as we race along in the night. She squints her eyes tight, her face is streamlined in the wind, and she projects her nose deep into the rushing air.

Once a smell has been vacuumed in, it finds a receptive welcome from an extravagance of nasal tissue. Most purebreds, and nearly all mutts, have long muzzles in whose noses are labyrinths of channels lined with special skin tissue. This lining, like the lining of our own noses, is primed to receive air carrying "chemicals"—molecules of various sizes that will be perceived as scents. Any object we encounter in the world is cast in a haze of these molecules—not only the ripe peach on the counter but the shoes we kick off at the door and the doorknob we grasp. The tissue of the inside of the nose is entirely blanketed with tiny receptor sites, each with soldiers of hairs to help catch molecules of certain shapes and pin them down. Human noses have about six million of these sensory receptor sites; sheepdog noses, over two hundred million; beagle noses, over three hundred million. Dogs have more genes committed to coding olfactory cells, more cells, and more kinds of cells, able to detect more kinds of smells. The difference in the smell experience is exponential: on detecting certain molecules from that doorknob, not single sites but combinations of sites fire together to send information to the brain. Only when the signal reaches the brain is it experienced as a scent: if it is we doing the sniffing, we'd say A-ha! I smell it.

More than likely, though, we won't smell it. But the beagle will: it's been estimated that their sense of smell may be millions of times more sensitive than ours. Next to them we are downright anosmic: smelling nothing. We might notice if our coffee's been sweetened with a teaspoon of sugar; a dog can detect a teaspoon of sugar diluted in a million gallons of water: two Olympic-sized pools full.*

What's this like? Imagine if each detail of our visual world were matched by a corresponding smell. Each petal on a rose may be distinct, having been visited by insects leaving pollen footprints from faraway flowers. What is to us just a single stem actually holds a record of who held it, and when. A burst of chemicals marks where a leaf was torn. The flesh of the petals, plump with moisture compared to that of the leaf, holds a different odor besides. The fold of a leaf has a smell; so does a dewdrop on a thorn. And time is in those details: while we can see one of the petals drying and browning, the dog can smell this process of decay and aging. Imagine smelling every minute visual detail. That might be the experience of a rose to a dog.

The nose is also the fastest route by which information can get to the brain. While visual or auditory data goes through an intermediate staging ground on the way to the cortex, the highest level of processing, the receptors in the nose connect directly to nerves in specialized olfactory "bulbs" (so shaped). The olfactory bulbs of the dog brain make up about an eighth of its mass: proportionally greater than the size of our central visual processing center, the occipital lobes, in our brains. But dogs' specially keen sense of smell may also be due to an additional way they perceive odors: through the vomeronasal organ.

THE VOMERONASAL NOSE

What specificity of image the name "vomeronasal" conjures up! Evoking the displeasure of getting a good sniff of fresh vomit, the "vomer" is actually a description of the part of the small bone in the nose where the sensory cells sit. Still, the name seems somehow fitting for an animal that is notorious for coprophagia (feces eating) and that may lick another dog's urine off the ground. Neither act is vomitous for dogs; it's just a way of getting even more information about other dogs or animals in the area. The vomeronasal organ, first discovered in reptiles, is a specialized sac above the mouth or in the nose covered with more receptor sites for molecules. Reptiles use it to find their way, to find food, and to find mates. The lizard who darts out its tongue to touch an unknown object is not tasting or sniffing; it is drawing chemical information toward its vomeronasal organ.

These chemicals are pheromones: hormonelike substances released by one animal and perceived by another of the same species, and usually prompting a specific reaction—such as readying oneself for sex—or even changing hormonal levels. There is some evidence that humans unconsciously perceive pheromones, perhaps even through a nasal vomeronasal organ.*

Dogs definitely have a vomeronasal organ: it sits above the roof (hard palate) of the mouth, along the floor of the nose (nasal septum). Unlike in other animals, the receptor sites are covered in cilia, tiny hairs encouraging these molecules along. Pheromones are often carried in a fluid: urine, in particular, is a great medium for one animal to send personalized information to members of the opposite sex about, say, one's eagerness to mate. To detect the pheromones in that urine some mammals touch the liquid and do a distinctive, mortifying, lip-curling grimace called flehmen. The face of a flehmening animal is notably unlovable—but it is the face of an animal who is on the hunt for a lover. The flehmen pose seems to hurtle the fluid toward the animal's vomeronasal organ, where it is pumped into the tissue, or is absorbed through capillary action. Rhinos, elephants, and other ungulates flehmen regularly; so do bats and cats, which have their own species variations. Humans may have vomeronasal organs, but we do not flehmen. Neither do dogs. But a regular observer of dogs will notice an often very intense interest in the urine of other dogs—sometimes an interest which lures them right … up … into … wait, gross! Stop licking that! Dogs may lightly lap up urine, especially urine of a female in heat. This could be their version of flehmen.

Even better than flehmen is keeping the outside of the nose nice and moist. The vomeronasal organ is probably why a dog's nose is wet. Most animals with vomeronasal organs have wet noses, too. It is difficult for an airborne odor to land squarely on the vomeronasal organ, since it is situated in a safe, dark interior recess of the face. A hearty sniff not only brings molecules into the dog's nasal cavity; little molecular bits also stick onto the moist exterior tissue of the nose. Once there, they can dissolve and travel to the vomeronasal organ through interior ducts. When your dog nuzzles against you, he is actually collecting your odor on his nose: better to confirm that you're you. In this way, dogs double their methods of smelling the world.

THE BRAVE SMELL OF A STONE

When Pump got her nose into a good smell in the grass—when she really dug her nose deeply into the earth—I came to know what was going to happen next. She'd hop around, resniff the smell from different angles, then take a tentative swipe at it, upending a dollop of turf. More deep sniffing, some licking, smushing her nose into the ground—and then the climax: an unrestrained dive into the smell, nose first, throwing her whole body down after it, and wriggling madly back and forth.

What, then, do these noses enable the dog to smell? What does the world look like from the vantage of a nose? Let's start with the easy stuff for them: what they smell of us and of each other. Then we might be ready to challenge them to smell time, the history of a river stone, and the approach of a thunderstorm.

The smelly ape

Humans stink. The human armpit is one of the most profound sources of odor produced by any animal; our breath is a confusing melody of smells; our genitals reek. The organ that covers our body—our skin—is itself covered in sweat and sebaceous glands, which are regularly churning out fluid and oils holding our particular brand of scent. When we touch objects, we leave a bit of ourselves on it: a slough of skin, with its clutch of bacteria steadily munching and excreting away. This is our smell, our signature odor. If the object is porous—a soft slipper, say—and we spend a lot of time touching it—putting a foot in it, clutching it, carrying it under an arm—it becomes an extension of ourselves for a creature of the nose. For my dog, my slipper is a part of me. The slipper may not look to us like an object that would be terribly interesting to a dog, but anyone who has returned home to find a ravaged slipper, or who has been tracked by the scent they've left thereon, knows otherwise.

We needn't even touch objects for them to smell of us: as we move, we leave behind a trail of skin cells. The air is perfumed with our constant dehumidifying sweat. Added to this, we wear in odor what we've eaten today, whom we've kissed, what we've brushed against. Whatever cologne we put on merely adds to the cacophony. On top of this, our urine, traveling down from the kidneys, catches odorous notes from other organs and glands: the adrenal glands, the renal tubes, and potentially the sex organs. The trace of this concoction on our bodies and our clothes provides more uniquely specific information about us. As a result, dogs find it incredibly easy to distinguish us by scent alone. Trained dogs can tell identical twins apart by scent. And our aroma remains even when we've left, hence the "magical" powers of tracking dogs. These skilled sniffers see us in the cloud of molecules we leave behind.

To dogs, we are our scent. In some ways, olfactory recognition of people is quite similar to our own visual recognition of people: there are multiple components of the image responsible for how we look. A different haircut or a newly bespectacled face can, at least momentarily, mislead us as to the identity of the person standing before us. I can be surprised what even a close friend looks like from a different vantage or from a distance. So too must the olfactory image we embody be different in different contexts. The mere arrival of my (human) friend at the dog park is enough to set me smiling; it takes another beat before my dog notices her own friend. And odors are subject to decay and dispersal that light is not: a smell from a nearby object may not reach you if a breeze carries it in the other direction, and the strength of an odor diminishes over time. Unless my friend tries ducking behind a tree, it's hard for her to conceal her visual image from me: a wind won't conceal her. But it might conceal her from a dog momentarily.

When we return home at a day's end, dogs typically greet our cocktail of stink promptly and lovingly. Should we come home after having bathed in unfamiliar perfume or wearing someone else's clothes, we might expect a moment of puzzlement—it is no longer "us"—but our natural effusion will soon give us away. Dogs are not alone among animals in seeing in scent. Sharks have been seen to follow the same zigzaggy path through water that an injured fish took some time before: through not just its blood but also its hormones, the fish has left a bit of itself behind. But dogs are unique in being encouraged and trained by people to use scent to follow someone who is visually long gone.

Bloodhounds are one of the supersmellers among dogs. Not only do they have more nose tissue—more nose—but many features of their body seem to conspire to enable them to smell extra strongly. Their ears are terrifically long, but not to enable better hearing, as they fall close to the head. Instead a slight swing of the head sets these ears in motion, fanning up more scented air for the nose to catch. Their constant stream of drool is a perfect design to gather extra liquids up to the vomeronasal organ for examination. Basset hounds, thought to be bred from bloodhounds, go one step further: with their foreshortened legs, the whole head is already at ground—scent—level.

These hounds smell well naturally. Through training—rewarding them for attending to certain scents and ignoring others—they are easily able to follow a scent left by someone one or many days before, and can even specify where two individuals parted ways. It doesn't take very much of our odor: some researchers tested dogs using five thoroughly cleaned glass slides, to one of which a single fingerprint was added. The slides were put away for a few hours or up to three weeks. Dogs then got to examine the array of slides, and tried to choose the human slide: they were rewarded with a treat if they guessed correctly, which is sufficient motivation for them to stand and sniff at glass slides. One dog was correct on all but six of one hundred trials. When the slides were then placed outside on the building roof for a week, exposed over the course of the seven days to direct sun, rain, and all manner of blowing debris, the same dog was still correct on almost half the trials—well above chance.

They track not just by noticing odors, but by noticing very small changes in odor. Each of our footsteps will have more or less the same amount of our scent in it. In theory, then, if I saturate the ground with my scent, by running chaotically to and fro, a dog who tracks by noticing that smell won't be able to tell my path—only that I've definitely been there. But trained dogs don't just notice a smell. They notice the change in a smell over time. The concentration of an odor left on the ground by, say, a running footprint, diminishes with every second that passes. In just two seconds, a runner may have made four or five footprints: enough for a trained tracker to tell the direction that he ran based just on the differences in the odor emanating from the first and fifth print. The track you left as you exited the room has more smell in it than the one right before it; thus your path is reconstructed. Scent marks time.

Conveniently, instead of becoming inured to smells over time, as we do, the vomeronasal organ and the dog nose may regularly swap roles, to keep the scent fresh. It is this ability that is exploited when training rescue dogs, who must orient themselves to the odor of someone who has disappeared. Similarly, scenting dogs who trail a criminal suspect are trained to follow what is delicately called our "personal odor generation": our natural, regular, and entirely involuntary butyric acid production. This is easy for them, and they can then extend this skill to smelling other fatty acids, too. Unless you are wearing a body suit made entirely of scentproof plastic, a hound can find you.

You showed fear

Even those of us who are not fleeing a crime scene, or in need of rescue, have a reason not to underestimate just how good a sniffer the dog is. Not only can dogs identify individuals by odor, they can also identify characteristics of the individual. A dog knows if you've had sex, smoked a cigarette (done both these things in succession), just had a snack, or just run a mile. This may seem benign: except, perhaps, for the snack, these facts about you might not be of particular interest to a dog. But they can also smell your emotions.

Generations of schoolchildren have been admonished to "never show fear" to a strange dog.* It is likely that dogs do smell fear, as well as anxiety and sadness. Mystical abilities need not be invoked to account for this: fear smells. Researchers have identified many social animals, from bees to deer, who can detect pheromones emitted when one animal is alarmed, and who react by taking action to get to safety. Pheromones are produced involuntarily and unconsciously, and through different means: damaged skin may provoke release of them, and there are specialized glands that release chemicals of alarm. In addition, the very feeling of alarm, fear, and every other emotion correlates with physiological changes, from changes in heart rate and breathing rate, to sweating and metabolic changes. Polygraph machines work (to the extent that they work at all) by measuring changes in these autonomic bodily responses; one might say that animals' noses "work" by being sensitive to them as well. Laboratory experiments using rats confirm this: when one rat is given a shock in a cage, and learns to be fearful of the cage, other rats nearby pick up on the shocked rat's fear—even without seeing the rat being shocked—and themselves avoid the cage, which was otherwise not distinguishable from nearby cages.

How does that strange, menacing-looking dog smell our apprehension or fear as he approaches us? We spontaneously sweat under stress, and our perspiration carries a note of our odor on it: that's the first clue to the dog. Adrenaline, used by the body to gear up for a good sprint away from something dangerous, is unscented to us, but not to the sensitive sniffer of the dog: another hint. Even the simple act of increased blood flow brings chemicals more quickly to the surface of the body, where they can be diffused through the skin. Given that we emit odors that reflect these physiological changes accompanying fear, and given the budding evidence of pheromones in humans, chances are that if we've got the heebie-jeebies, a dog can tell. And as we'll see later, dogs are skilled readers of our behavior. We can sometimes see fear in other people in their facial expressions; there is sufficient information in our posture and gait for a dog to see it, too.

In these ways, the fleeing criminal being tracked by dogs is doubly doomed. Dogs can be trained to track based not just on pursuit of a specific person's odor, but also based on a certain kind of odor: the most recent odor of a human in the vicinity (good for finding someone's hiding place), or a human in emotional distress—fearful (as one running from the cops might be), angry, even annoyed.

The smell of disease

If dogs can detect trace amounts of chemicals we leave behind on a doorknob, or in a footprint, might they be able to detect chemicals indicating disease? If you're lucky, when you come down with a disease difficult to diagnose, you'll have a doctor who recognizes, as some have, that a distinctive smell of freshly baked bread about you is due to typhoid fever, or that a stale, sour scent is due to tuberculosis being exhaled from your lungs. According to many doctors, they have come to notice a distinctive smell to various infections, or even to diabetes, cancer, or schizophrenia. These experts come unequipped with the dog's nose—but more equipped to identify disease. Still, a few small-scale experiments indicate that you might get an even more refined diagnosis if you make an appointment with a well-trained dog.

Researchers have begun training dogs to recognize the chemical smells produced by cancerous, unhealthy tissues. The training is simple: the dogs were rewarded when they sat or lay down next to the smells; they weren't rewarded when they didn't. Then the scientists collected the smells of cancer patients and patients without cancer, in small urine samples or by having them breathe into tubes able to catch exhaled molecules. Although the numbers of trained dogs are small, the results were big: the dogs could detect which of the patients had cancer. In one study, they only missed on 14 out of 1,272 attempts. In another small study with two dogs, they sniffed out a melanoma nearly every time. The latest studies show trained dogs can detect cancers of the skin, breast, bladder, and lungs at high rates.

Does this mean your dog will let you know when a small tumor develops in you? Probably not. What it indicates is that dogs are able to do so. You might smell different to them, but your changing smell might be gradual. Both you and your dog would need training: the dog to pay attention to the smell, you to pay attention to behaviors indicating your dog has found something.*

The smell of a dog

Since odor is so conspicuous to a dog, it gets great use socially. While we humans leave our scents behind inadvertently, dogs are not only advertent, they are profligate with their scents. It is as though dogs, realizing how well the odor of our bodies comes to stand for ourselves (even in our absence), determined to use this to their advantage. All canids—wild and domestic dogs and their relations—leave urine conspicuously splashed on all manner of object. Urine marking, as this method of communication is called, conveys a message—but it is more like note-leaving than a conversation. The message is left by one dog's rear end for retrieval by another's front end. Every dog owner is familiar with the raised-leg marking of fire hydrants, lampposts, trees, bushes, and sometimes the unlucky dog or bystander's pant leg. Most marked spots are high or prominent: better to be seen, and better for the odor in the urine (the pheromones and affiliated chemical stew) to be smelled. Dogs' bladders—sacs that serve no known purpose except as a holding pen for urine—allow for release of just a little urine at a time, allowing them to mark repeatedly and often.

And having left smells in their wake, they also come right up to investigate others' smells. From observations of the behavior of the sniffing dogs, it appears that the chemicals in the urine give information about, for females, sexual readiness, and for males, their social confidence. The prevailing myth is that the message is "this is mine": that dogs urinate to "mark territory." This idea was introduced by the great early-twentieth-century ethologist Konrad Lorenz. He formed a reasonable hypothesis: urine is the dog's colonial flag, planted where one claims ownership. But research in the fifty years since he proposed that theory has failed to bear that out as the exclusive, or even predominant, use of urine marking.

Research on free-ranging dogs in India, for instance, showed how dogs behave when left entirely to their own devices. Both sexes marked, but only 20 percent of the markings were "territorial"—on a boundary of a territory. Marking changed by seasons, and happened more often when courting or when scavenging. The "territory" notion is also belied by the simple fact that few dogs urinate around the interior corners of the house or apartment where they live. Instead, marking seems to leave information about who the urinator is, how often he walks by this spot in the neighborhood, his recent victories, and his interest in mating. In this way, the invisible pile of scents on the hydrant becomes a community center bulletin board, with old, deteriorating announcements and requests peeking out from underneath more recent posts of activities and successes. Those who visit more frequently wind up being at the top of the heap: a natural hierarchy is thus revealed. But the old messages still get read, and they still have information—one element of which is simply their age.

In the annals of animal urine marking, dogs are not the most impressive players. Hippopotami wave their tails as they spray urine, better to scatter it, sprinklerlike, in all directions. There are rhinoceroses who follow their high-powered urination onto bushes with destruction of the same bushes with horn and hoof—to ensure, presumably, that their urine is spread far and wide. Pity the owner whose dog is the first to discover the spreading-efficiency of high-powered, whirling-sprinkler urination.

Other animals also press their rear ends against the ground to release fecal and other anal odors. The mongoose does a handstand and rubs itself against a high perch; some dogs do what gymnastics they can, seemingly deliberately relieving themselves on large rocks and other outcroppings. Although secondary to urine marking, defecation also holds identifying odors—not in the excreta itself but in the chemicals dolloped on top. These come from the pea-sized anal sacs, situated right inside the anus and holding secretions from nearby glands: extremely foul, dead-fish-in-a-sweatsock kind of secretions with apparently individual-dead-fish-in-individual-sweatsock odors for each individual dog. These anal sacs also release involuntarily when a dog is afraid or alarmed. It may be no wonder that so many dogs fright at their veterinarian's office: as part of a routine examination, vets often express (squeeze to release the contents of) the anal sacs, which can get impacted and infected. The smell, covered for us by the familiar scent of veterinary antibiotic soaps, must be all over the vets: they reek of epic dog fear.

Finally, if these mephitic calling cards are insufficient, dogs have one other trick in their marking book: they scratch the ground after defecation or urination. Researchers think that this adds new odors to the mix—from the glands on the pads of the feet—but it may also serve as a complementary visual cue leading a dog to the source of the odor for closer examination. On a windy day, dogs may seem friskier, more likely to scratch the ground; they may in fact be leading others to a message that otherwise would waft away.

LEAVES AND GRASS

Science, out of decorum or disinterest, has not definitively explained Pump's mad wriggling in a funky spot of grass. The odor may be of a dog she's interested in, or of a dog she recognizes. Or it may be the remnants of a dead animal, rolled in not so much to conceal her own smell as enjoyed for its sumptuous bouquet.

We respond pithily and with soap: by giving our dogs frequent baths. My neighborhood has not only its fill of dog groomers, but is visited by a mobile grooming van that will come to your home to pick up, suds, fluff, and otherwise de-dog your dog for you. I'm sympathetic to owners who have a lower tolerance for detritus and dust around their homes than I do: a well-walked, thoroughly played-out dog is an efficient spreader of dirt. But we deprive our dogs of something by bathing them so much—to say nothing of our culture's overenthusiastic cleaning of our own homes, including our dogs' bedding. What smells clean to us is the smell of artificial chemical clean, something expressly non-biological. The mildest fragrance that cleansers come in is still an olfactory insult to a dog. And although we might like a visually clean space, a place rid entirely of organic smells would be an impoverished one for dogs. Better to keep the occasional well-worn T-shirt around and not scrub the floors for a while. The dog himself does not have any drive to be what we would call clean. It is no wonder that the dog follows his bath by hightailing it to roll vigorously on the rug or in the grass. We deprive dogs of an important part of their identity, temporarily, to bathe them in coconut-lavender shampoo.

Similarly, recent research found that when we give dogs antibiotics excessively, their body odor changes, temporarily wreaking havoc with the social information they normally emit. We can be alert to this while still using these medicines appropriately. So too with the laughable Elizabethan collar, an enormous cone collar typically used to prevent a dog from chewing at stitches closing a wound: it is useful to prevent self-mutilation, but consider all the ordinary interactive behavior it prevents—looking away from an aggressive dog; seeing someone's loping approach from the side; the ability to reach and sniff another dog's rump.

Pity the urban dog, subjected to the remnants of an old society-wide terror that odors themselves caused disease. Urban planning shifted in the eighteenth and nineteenth centuries toward elaborate "deodorization" of cities: paving streets and replacing dirt paths with concrete to trap odors. In Manhattan it even prompted a grid-based street system that, it was thought, would encourage odors to race out of the city to the rivers, instead of settling into pleasant nooks and alleys. This surely

reduces the dog's possible enjoyment of the smells inside the crevices of every fallen leaf and blade of grass paved over.

BRAMBISH AND BRUNKY

I used to be fooled by Pump's motionless posture when we sat outside together. One time, looking more closely at her, I saw that she was motionless but for one part: her nostrils. They were churning information through their caverns, ruminating on the sight before her nose. What was she seeing? The unknown dog who just turned the corner off the block? A barbecue down the hill, with perspiring volleyballers circling grilling meats? An approaching storm, with its fulminating bursts of air from distant climes? The hormones, the sweat, the meat—even the air currents preceding the arrival of a thunderstorm, upwardly moving drafts leaving invisible scent tracks in their wake—are all detectable, if not necessarily detected or understood, by the dog's nose. Whatever it was, she was far from the idle creature she'd seemed to be.

Knowing the importance of odor in a dog's world changed the way I thought about Pump's merry greeting of a visitor in my house by heading directly for his groin. The genitals, along with the mouth and the armpits, are truly good sources of information. To disallow this greeting is tantamount to blindfolding yourself when you open the door to a stranger. Since my guests may be less keen on the dog umwelt, though, I advise visitors to proffer a hand (undoubtedly fragrant), or kneel and let their head or trunk be sniffed instead.

Similarly, it is peculiarly human to chastise a dog for greeting a new dog in the neighborhood by smelling his rump. Our distaste for the notion of rump-smelling as a human social practice is irrelevant. For dogs, by all means, the closer the better. Dogs will communicate to each other if they are uninterested in being so intimately examined; interference may agitate one or both of them.

To understand the dog umwelt, then, we must think of objects, people, emotions—even times of day—as having distinctive odors. That we have so few words for smells restricts our imagination of the brambish, brunky diversity that exists. Perhaps, a dog can detect what a poet evokes: the "brilliant smell of water, The brave smell of a stone, The smell of dew and thunder …" (and definitely "… The old bones buried under …"). Probably, not all smells are good smells: as there is visual pollution, so is there olfactory pollution. Definitely, those who see smells must remember in smells, too: when we imagine dogs' dreaming and daydreaming, we should envisage dream images made of scents.

Since I've begun to appreciate Pump's smelly world I sometimes take her out just to sit and sniff. We have smell-walks, stopping at every landmark along our routes in which she shows an interest. She is looking; being outside is the most smelly, wonderful part of her day. I won't cut that short. I even look at photographs of her differently: where she once looked to be pensively staring in the distance, I now think what she's really doing is smelling some new exciting air from a far-flung source.

But I'm happiest of all to receive her greeting sniff of me, prompting her wag of recognition. I nuzzle into the scruff of her neck and sniff her right back.