Dog Sense

by John Bradshaw

  These experiments tell us that dogs have a passion for following human gaze, but they do not tell us whether they are born with this obsession, or whether they learn it. The guide dogs that were studied had not always lived with blind people; they had been raised by sighted families for the first twelve to eighteen months of their lives, before starting guide-dog training. Perhaps these are the critical years in which the habit of following a person’s gaze is learned, after which it is very difficult to unlearn. However, it is remarkable that the guide dogs seemed hardly to have altered their behavior at all in the four years, on average, that they had been living with a blind person.

  This apparent overreliance on human eyes and arms raises the question of whether dogs actually have the ability to understand what we’re thinking. They are clearly responsive to what they see us doing, but that’s not the same as understanding that each human has a mind—one that, furthermore, is different from their own. Such an idea must seem like heresy to the majority of dog owners, but science has thus far failed to demonstrate such abilities in any species apart from our own. Even the many experiments that have been done on chimpanzees have failed to provide conclusive evidence that the great apes know other minds exist.

  Part of the problem is that it is very difficult to design experiments that test whether dogs are able to tell what we’re thinking—in other words, experiments that exclude simpler explanations for their behavior. For example, in a study examining whether dogs understand what people can and can’t see,15 eleven dogs that had been trained not to steal food were taken into a room where there was a piece of food on the floor. When the experimenter told the dogs not to eat the food when she was facing them, they usually left it, but when she was faced away from them while issuing the command, they often did take the food. The dogs might have “known” that they couldn’t be seen, but the simpler explanation is that they associated seeing a human face with obeying a command—“If the face is not there, then the verbal command doesn’t mean anything.” Given that dogs are so sensitive to human faces and the expressions on them, it’s very difficult to design experiments that would rule out such an explanation.

  However, a subsequent experiment showed that not all dogs are equally sensitive to being watched.16 In this instance, dogs were invited to take food from behind barriers of various shapes and sizes. One barrier, for example, hid the food but allowed the dog to see the experimenter when she was approaching the food; another was larger but with a small window, blocking the dogs’ view of the experimenter until the dogs were very close to the food. A few of the dogs behaved as if they understood which barriers blocked the experimenter’s view of the food and which didn’t. Others seemed much more inhibited if she could see them start moving toward the food but were oblivious to the fact that they could actually be seen eating it (e.g., through the window in the barrier). Thus it is still not clear whether dogs can work out what people can and can’t see, or whether they’re responding to simple learned “rules” such as “If I can see a person’s face, then I shouldn’t move toward food” or “If I can see a person’s face, then it’s OK to move but I mustn’t actually eat.”

  Such experiments show that dogs are very sensitive to whether people are watching them or not. They do not, however, provide conclusive evidence that dogs know what people are thinking—that they possess what cognitive biologists refer to as a “theory of mind.” Most dogs have been fed all their lives from a bowl placed on the floor by a person’s hand, so it’s hardly surprising that they like to follow the direction that a human hand is pointing to or moving in. Moreover, most dogs have learned that the way a person reacts to them will be quite well predicted by the direction in which the person is facing and moving. Consciously or not, humans expect dogs to be very sensitive to our body-language. This capacity must have been so useful to cementing the bond between man and dog that any dogs who lacked it were probably selected out of the population many, many generations ago.

  Further evidence that dogs do not possess a “theory of mind” lies in their susceptibility to deception. In one recent experiment, dogs were trained to expect that one person (the “truth-teller”) would always point at a container with food in it and that another (the “liar”) would always point at an otherwise identical but empty container.17 More often than not the dogs preferred to go where the truth-teller pointed, but by no means every time. When the people were replaced by a simple association that a white box always contains food (truth) while a black box never does (lie), the dogs marginally preferred the white, “truthful” box. Thus there was no evidence that the dogs had understood the difference between truthful and untruthful people; more likely, they had just learned to associate one of the people with getting food.

  In sum, domestication does not seem to have given dogs the ability to read our minds, or even to understand that humans are capable of independent thought. They must therefore live in a subjective world that is very different from ours, one in which we exist not as independent entities but merely as components of that world (albeit usually the most important components). This is actually unsurprising, given what we now know evolution is capable of. The dog is stuck with a canid brain—and although this has undoubtedly been modified by domestication, it is asking too much to assume that a whole new layer of complexity could have been added during the domestication process.

  So if dogs don’t know what we’re thinking, how come they give the impression that they do? From almost as soon as they can see, dogs seem to be especially sensitive to actions performed by humans. This difference from the wolf is almost certainly due to a genetically programmed change of focus in the dog’s priorities, driven by domestication. Those proto-dogs that happened to possess a predisposition to attend to the humans around them would have been able to learn the significance of specific human gestures. This adaptation, in turn, would give these more sensitive dogs a key advantage over more wolf-like dogs who would have been more focused on their own species and the physical world.

  Today, this almost overwhelming focus on people and what they are doing enables dogs to learn very subtle aspects of human body-language, possibly even actions that we are unaware of ourselves. In addition, they almost certainly gather information about us, using their hypersensitive noses, based on subtle changes in odor that we are entirely unaware of. (This capacity to recognize subtle changes in body odor very likely lies behind the ability of trained dogs to detect impending seizures in diabetics and epileptics.) It is the shift in the focus of attention—from other members of their own species to members of the human race—that is domestication’s primary effect on dogs’ intellect. There has been no step-change in overall ability, just an adjustment of their primary focus. Dogs appear to be no more or less limited than wolves in terms of what they can learn; it’s just that the priorities of what to learn and who to observe have been changed by domestication. Thus, although dogs appear to understand what we are thinking, no evidence has yet been found that suggests they are aware that we even can think. They are merely very well adapted to respond in the most productive way, nine times out of ten. Give them a situation that evolution has not prepared them for, such as an owner who is blind, and they continue to adhere to the standard ways of responding to people.

  The dog’s apparent lack of a “theory of mind,” then, raises the question as to whether dogs even have distinct concepts for “a person” and “a dog.” The level of attachment between dog and owner is different from that between dog and dog, but is there also a qualitative difference? Dogs obviously behave differently toward people than toward other dogs, but could this simply be a consequence of one species walking upright and the other on all fours?

  Play behavior should be a useful window into this aspect of the dog’s mind, since it incorporates a kind of “lingua franca” that dogs and people can use to communicate their intentions equally well. Dogs can of course be readily persuaded to play with other dogs as well as with people, and human volunteers can
be persuaded to play as if they were dogs—for example, by staying on all fours throughout the game. Together with my colleagues at Bristol University, I’ve been making comparisons between dog-dog play and dog-human play in order to gain insight into whether dogs play differently depending on which species they’re playing with and, by inference, whether they have different mental concepts of “person” and “dog.”18 We recruited a dozen Labrador retrievers, all chosen on the basis of their reputation for being particularly playful, as well as for being a popular breed. The dogs were released, one at a time, into a large grass paddock accompanied either by another dog (one it knew well) or by one of its regular carers. We gave the dogs two minutes for an initial exploration of the area and then threw into the paddock a tug-toy consisting of a short length of knotted rope. All the dogs were accustomed to and enjoyed playing tug-of-war games and thus immediately began playing with the rope, regardless of whether their play-partner was another dog or a person.

  The dogs spent most of their time engaged in the game—again, regardless of whether they were playing with a person or with another dog. On the other hand, they clearly played very differently depending on who their partner was. When playing with the person, the dogs were much more likely to surrender the toy, seemingly in order to keep the game going. But when two dogs were playing, each tried to keep possession of the rope, attempting to guard it from the other whenever the other let go.

  This behavioral distinction was even more pronounced when we added another component to the game. Three minutes after the first tug-toy was thrown into the paddock, we threw in a second one. Now each dog had the choice of continuing to play with the original toy or grabbing the second toy and going off to play with it on its own. Here, the difference between the way dogs played with humans and the way they played with other dogs was dramatic. When two dogs were play-partners, they would often each take a toy and play with it for a while on their own before coming back to play together again. But when the play-partner was a person, the fact that there was a second toy available seemed almost irrelevant: The dog kept on bringing one toy back to the person and inviting her to keep on tugging.

  Dogs playing with a tug-toy are actually competing

  In short, dogs appear to be in a completely different frame of mind depending on whether they are playing with a person or with another dog. When the play-partner is a dog, possession of the toy seems to be most important—and, indeed, it’s possible that competitive play is one way that dogs assess each others’ strength and character. (For dogs, these games are primarily a way of assessing resource holding potential.) When the play-partner is a person, however, possession of the toy seems irrelevant; the important thing is the social contact that the game produces. This finding is entirely compatible with the observation that dogs can’t calm one another down but can be calmed by their owners. It also indicates that dogs put humans in a completely different mental category from other dogs.

  This distinction between play with people and play with dogs is not confined to Labradors but, rather, appears to be a universal attribute of domestic dogs. As part of the same overall study, we also surveyed dog owners—and observed them with their dogs—to determine whether play with people interferes with play with other dogs. We hypothesized that if the two potential playmates were interchangeable as far as the dogs were concerned, playing with a person should diminish their appetite for playing with other dogs and vice versa. But we saw no evidence for this at all: When watching owners playing with their dogs off-leash in parks, we found the quality of play to be the same whether the owner had one dog or more than one. In a related survey, we asked 2,007 owners with only one dog and 578 owners with more than one dog how often they played with their dogs, and we discovered that the owners with several dogs actually played slightly more with each of their dogs than did those with only one. Although it was impossible to record the quality of the play, we concluded that most dogs are very happy to play with their owners—whether or not they have a canine alternative. Again, this suggests that dogs’ minds have separate categories for “people” and “dogs.”

  Play between dog and owner is such an everyday occurrence that it’s easy to lose sight of the fact that inter-species play is otherwise very rare (indeed, virtually unknown outside the realm of domestic pets). To be successful, play requires well-synchronized communication; both partners must be able to convey their intentions precisely while at the same time convincing each other that they’re not using the game as a prelude to something more serious, such as an actual attack. The rarity of such play is probably accounted for by the limitations of communication between members of different species.

  The capacity of dogs to engage in play with humans is particularly surprising given the sophistication of dog’s communication with other dogs during play. For instance, when two dogs are playing together, play-bows are much more likely to occur when they are facing one another than when one is facing away, indicating that dogs are sensitive to whether or not their play-partner is paying attention to what they’re trying to convey. 19 Dogs that want to perform play-bows but are being ignored have a variety of ways of getting another dog’s attention, including nipping, pawing, barking, nosing, and bumping. Humans are much less clever at this than they are, so the boundless appetite that most dogs seem to have for games with their owners, and even with people they don’t know so well, must be due to the strength of their attachment to mankind in general.

  Play-bow

  However, our knowledge of dogs’ cognitive abilities gives us no basis for thinking that dogs are aware of what they are doing when they are playing in the same way that we are aware of (and so can talk or write about) how we play with a dog. Dogs may appear to “deceive” their human play-partners—for example, by dropping a ball and then grabbing it again before the person has time to pick it up. But there is a much more straightforward explanation for this behavior: We know that dogs find play rewarding—“fun”—and since it takes two to make a game, any sequence of actions that happens to stimulate play in others should quickly become part of the ritual that that dog and person engage in whenever they play, by simple association. Thus in this instance the dog must have accidentally dropped a ball when near a person in the past, and then quickly grabbed it again (as a wolf would do if it accidentally dropped a piece of food). And with its priorities focused on human reaction, it must have noted the person’s excited reaction to this apparent “deception.” Therefore, it will repeat the sequence of actions in the hope of getting the same reaction again—which of course is likely to happen.

  Indeed, although we humans are undoubtedly less good at interpreting dog behavior than another dog would be, dogs are uniquely so focused on the reactions they get from people that they are able to adapt their behavior to fit our own. I don’t mean that they do this consciously; rather, my point is that our behavior comprises the most salient cues available to them, such that, without having to think about it, they can adjust their reactions to us using quite simple associative learning.

  Many dogs are friendly toward people in general, but it’s obvious that they all know the difference between strangers and familiar people and between individual people that they know well. So far, science has only just begun to investigate how dogs tell people apart from one another. Evidence suggests that they build up a single, multisensory “picture” of people they know: In one study, researchers played for dogs a recording of the voice of one person they knew but then showed the dogs a picture of someone else they knew. The dogs gave a look of surprise, as if the voice had already conjured up the face that should go with it. (In addition, there is probably an olfactory dimension to the picture that dogs create of us—a dimension about which we ourselves are largely unaware.)

  Dogs are also very sensitive to what goes on within relationships—not just those in which they’re directly involved but also those they observe between people. In one recent study, a dog was allowed to watch three people performing a scrip
ted transaction.20 In this exchange, one person acted as the “beggar” and each of the others either gave him the money he was asking for (this was the “generous” person) or did not do so (this was the “selfish” person). Once the beggar had left the room, the dog was released and allowed to interact with the other two people. The dogs preferred to interact with the “generous” person; most went to her first and chose to spend more time interacting with her than with the “selfish” person. It seemed to be the actual act of handing over the money that was important to the dog, because when the whole scenario was repeated but with no beggar present (meaning that the transaction had to be mimed), the dogs showed no preference for the “generous” person.

  Dogs also demonstrate some understanding of relationships between people and other dogs in the household. In an experiment designed to verify this, a Labrador was placed on one side of a transparent gate and allowed to watch tug-of-war games between a person and another Labrador.21 Unknown to either of the dogs, the games were manipulated so that during some trials the person always won possession of the tug and during others the dog was allowed to win. Furthermore, some of the games were made “playful” (the person performed play-signals) while others were made “serious” (the person did not perform such signals). After the game, the spectators were let out from behind the gate where they had been watching. After “playful” games, the spectators preferred to interact with the apparent “winner,” whether dog or person; after “serious” games, the spectators were reluctant to approach either dog or person. Thus dogs don’t just react to dogs or people as individuals; they also react to what they’ve seen go on between them. However, this is not to say that they necessarily understand what the concept of a “relationship” is. More likely, they simply modify their behavior toward each of the participants depending upon what they’ve just seen them do.