In Defence of Dogs

by John Bradshaw

  Captive wolf-pack hierarchy

  Wolf-pack family structure

  In the wild, then, a dominant, or ‘alpha’, wolf is just the wolf that leads the pack, usually because it is the mother or father of most or all of the other members of the pack. The normal way for a wolf to become a ‘dominant’ animal is simply to become sufficiently mature and experienced to be able to find a mate, and then to breed. The term ‘dominant’ thus becomes synonymous with ‘mother’ or ‘father’, or occasionally ‘stepmother’ or ‘stepfather’, if one of the breeding pair dies and is replaced before the pack breaks up.

  Since it is clear that modern wolves are almost certain to be much more wary of man than were the domestic dog’s wild ancestors, we should logically consider whether the social grouping preferred by modern wolves – the family-based pack – is likely to have also been the preferred social structure of the wolves that became dogs, tens of thousands of years ago. It is, of course, impossible to be sure about how these ancient wolves organized their societies, because social structures do not fossilize. However, we can be confident that wolves lived in packs then as now, because the family unit is the basis for packs wherever they occur among the canids; family groups are inherently more stable than groups formed by unrelated individuals.

  Since it was probably shared with ancient wolves, the canids’ intrinsically family-based structure is almost certainly the raw material that domestication had to work with. In that case, the most likely explanation for how we have been able to domesticate dogs is that we have been able to insert ourselves into that family structure. By this, I do not mean ‘family’ in the genetic sense; that is clearly impossible. I mean ‘family’ in the sense of a group whose members live together and are able to co-operate because they know one another well. Family members do not have to be genetically related to one another in order to function effectively; for instance, as an adopted human baby grows up, it behaves towards its adoptive parents as if they were its genetic parents. Although such babies may eventually come to know that they have other, biological parents, this knowledge does not displace the bond they have with the adults who have cared for them. The reason is that, in our species, recognition of immediate family members is largely based upon familiarity, even though we come to overlay this with more conscious knowledge of our actual genetic relationships. Biologically, as a species with a long period of childhood dependency, we are set up to base our family bonds on who looks after us and is affectionate towards us; our instinctive ability to distinguish who is actually related to us genetically is comparatively weak (much weaker than in species with little or no parental care).

  Much as human children bond to their carers, regardless of their biological connection, as puppies grow they come to regard both their canine and human carers as ‘family’. This capacity must be a product of domestication, for while dogs can bond to humans in this way it is difficult to get wolves to do the same, even if they are separated from their mother and siblings and hand-reared by people. Domestication must have introduced a much greater fluidity into the process whereby ‘family’ allegiances are learned, such that humans can be incorporated as well as other dogs, whether or not those dogs are actually genetic relatives.

  Since wolves and other canids share a fundamentally co-operative social structure – and since this social system was probably critical to our domestication of dogs – it is illogical to suggest that the process of domestication replaced this co-operative system with a ‘dominance’ structure based on self-interest and aggression. If anything, domestication, by requiring dogs to live in close proximity to one another, should have made dogs more tolerant of one another, and certainly not less so. For this reason alone, concepts of ‘dominance’ and ‘hierarchy’, so beloved among much of the dog-training community, are fundamentally implausible. Moreover, such hierarchical behaviour is not readily apparent in natural wolf packs, and studies of feral dogs have also failed to find hierarchical structures – most of feral dogs’ competitive behaviour is seasonal, and seen in the context of mating and breeding, not in establishing permanent ‘dominance’ relationships. As a whole, the terms ‘dominance’ and ‘hierarchy’ can, it seems, no longer be justified to account for the behaviour of pet dogs.

  Rejecting the idea of dominance as a natural driver of dog behaviour is not the same as saying that dogs are never competitive – of course they are, when they have to be. Put several un-neutered dogs of the same sex that do not know one another into a small space, and they are likely to set up a temporary ‘hierarchy’ based on threats, or even fighting, especially if they can sense that there is a member of the opposite sex nearby. This outcome would occur with almost any species, having nothing to do with the dog being descended from the wolf (although it does resonate with the conflict seen in some artificially constructed captive wolf packs). However, such situations are very unnatural, in that they seem almost designed to provoke conflict, telling us nothing about what goes on in dogs’ heads when they live in a multi-dog household, or when they meet another dog in the park – or, for that matter, how they perceive their human owners.

  The old ‘dominance’ model of dog behaviour is based on three concepts, each of which is now known to be false. First, it is derived from how wolves behave when they are living in unnaturally constituted groups in captivity, and not from the natural behaviour of wolves living in wild packs. Second, ferals or ‘village dogs’, when allowed to establish family groups, do not behave like wolves at all, neither captive nor wild. These feral dogs, which are much closer to the ancestors of our pet dogs than any wolf, are much more tolerant of one another than any other modern canid would be if it lived at such high density; initially, this tolerance was a consequence of domestication, and it has been maintained by the necessity of remaining tolerant in order to survive around people. Third, although dominance based on competition and aggression does occur among wolves in captivity, dogs kept under similar conditions do not establish hierarchies.

  Ultimately, the behaviour of captive wolves is of little or no relevance to understanding dog behaviour. Not only is the behaviour of captive wolves now known to be unnatural, but it is also clear that domestication has radically altered the social repertoire of the wild species. Dogs have taken an evolutionary path quite different from that of the wolf, so, if we are to understand dog behaviour, we shall need to do so on fundamentally different terms.

  Since the practice of explaining the behaviour of dogs by referring back to one of their ancestors, the wolf, has been discredited, some alternative is needed if we are to understand not only why dogs behave the way they do, but also how they interpret and respond to our behaviour. In recent years, much of our understanding of how to relate to dogs has revolved around the notion of hierarchy. Indeed, many animals, not just wolves, appear to construct their groups in a hierarchical way. But often those hierarchies are constructed by scientists for their own purposes, for example, to test whether the animals that fight the most are also the most successful at breeding (and, more often than you might expect, they are not). The fact that a hierarchy can be observed by scientists does not mean that any of the animals involved have any awareness of that hierarchy. In short, the use of any hierarchical model, wolf or otherwise, presumes that the animals behave in the way they do because they are responding to their own conception of that hierarchy.

  The question of whether or not dogs understand the notion of hierarchy – and their own place within it – has a very profound implication for the way that we relate to them. If dogs do understand this concept, then training methods based on concepts like ‘status reduction’ and ‘putting the dog in its place’ have a logical foundation, and thus should be effective. But if dogs have no concept of their own status, then such methods are likely to convey a different message to them than that which was intended. Many of these methods are based on inflicting punishment on the dog. The question of whether or not dogs understand hierarchies is not just an academic exercise, ther
efore; it may have real consequences for their welfare.

  To ascertain whether dogs understand the concept of hierarchy, we need to refine our analysis. Rather than trying to second-guess what dogs are thinking, based on an outmoded concept of what they might have inherited from their wild ancestors, it is perhaps more useful to go back to first principles: dogs’ fundamental objectives in life. We know what dogs need. At the most basic level, they need food and water and opportunities to mate and raise offspring, thus ensuring the survival of their genes. In domestic situations, all of these needs are taken care of by humans. So the next question is, what do dogs want? These desires will generally connect back to something dogs needed during their evolutionary history (but not necessarily currently), and will involve both the things they required for survival and the means of achieving them. For example, a hungry carnivore wants (and needs) to eat, but even if it is not hungry, it will often want to look for food, in case its luck is about to run out and it does not find anything to eat for a while. On a longer timescale, it may also want to maintain a foraging territory where it has exclusive rights to hunt. As a youngster, it may want to play with objects, which may ultimately improve its hunting skills, even though it is almost certainly never aware of this connection: the ‘want’ is enough to promote its survival. Some of these ‘wants’ can be satisfied without disadvantaging any other member of the species. For example, play-fighting between littermates is mutually beneficial; both participants improve their competitive skills, and, because they are closely related, the survival of their genes improves twofold.

  Many of dogs’ ‘wants’ bring individuals into conflict with one another. Two animals cannot satisfy their hunger if there is only enough food for one; if all the available space is already occupied, a newcomer cannot gain an exclusive territory without displacing an existing territory-holder. The dominance framework describes one set of ways in which such conflict is avoided; some individuals, by being able to impose themselves on others, guarantee themselves priority of access to all resources – their ‘wants’ are met first. However, the dominance concept is unnecessarily restrictive in the context of thinking about how conflict can be resolved without coming to blows.

  One alternative to the dominance model can better explain how dogs avoid potential disputes: the Resource Holding Potential (RHP) model. According to this model, whenever a conflict of interests arises, each dog is thought to make its decision based on the answers to two questions: how much do I want this resource (food, toy, etc.), and how likely is it that the other dog is going to beat me if we fight for it? One factor that the individual dog should take into account when answering the latter question is, how much does it look as though my opponent wants the resource? This opens the door to the possibility of cheating; animals that threaten first and most emphatically may win even when they are no bigger than their opponent. If two dogs know each other, their memories of previous disputes are also available to be taken into account. If not, they may use information they have gathered during encounters with similar-looking dogs, perhaps even during disputes between other dogs they have watched. So, prior acquaintance is not absolutely necessary: the RHP model can be used just as well when the two dogs are meeting for the first time.

  The RHP model can therefore be applied to many more situations than the dominance model can. Studies done in my research group5 have shown that dogs do indeed take account of how much another dog wants (or at least appears to want) a particular thing. Within groups of dogs that live together, it is male dogs especially that exemplify this principle. For example, in one group of French bulldogs, one of the four bitches appeared to have priority of access to food, even though she was not the oldest or the mother of the most recent litter (a finding that also contradicts the wolf model). The single male deferred to her over food, but usually had priority of access to toys, and to the first sniff of any unfamiliar male dog. Among five Shetland sheepdogs, both of the males would allow any of the three females to feed first, but not give priority of access in any other context, including access to people and other dogs, male or female. Observations like these provide further evidence that pure ‘status’ cannot be the guiding principle for dogs’ competition over resources; if it were, one of the dogs in these groups should have enforced the right to prior access to everything, for fear that if it did not, it would be showing weakness, inviting a challenge from one of the others. In this respect, dogs seem to be obeying one main ‘rule’ of the RHP model: if another dog seems to want something badly, and you do not, then it is not worth getting into an argument over it.

  The basic RHP model has one significant drawback when applied to dogs. Dogs seem to try diligently to assess others’ intentions, but many appear not to have heard about the other main ‘rule’ of RHP: do not pick a fight if your opponent is a lot bigger than you are. Most animals are extremely good at gauging the size and strength of their opponent, even when the difference is quite small: picking unnecessary fights with those that are bigger than you are is, of course, likely to result in injury. One might expect this rule to apply even more so to dogs, given the enormous range of sizes between the biggest and the smallest breeds, greater than in any other species of animal – but it does not. At some time or other, everyone will have witnessed a small feisty dog meeting a larger dog in the park, and forcing it to retreat. More systematic studies that I have contributed to have borne this out:6 neither the way that two dogs first approach one another nor the outcome (which one retreats first) depends upon which is the larger or heavier of the two. Nor did size appear to have had any influence on the relationships between the dogs that we observed in the Wiltshire sanctuary. Unusually among the whole of the animal kingdom, dogs seem to pay little attention to size in considering whether or not to start a challenge with another dog that they do not know. Instead, individual dogs vary in terms of how bold they are in such situations, as if it were a part of their ‘personality’. The RHP model therefore has to be modified if it is to be applied to dogs: when one dog is deciding how to proceed in an encounter, the apparent motivation of the other dog seems to be given greater weight than its actual size and hence likely strength.

  Dogs’ lack of attention to size when competing over resources may be a product of domestication. In the wild, aggression emerges when much is at stake, that is, when an animal is in danger of losing its territory, when it is very hungry or thirsty and another is preventing it from eating or drinking, and when it is competing with others for the opportunity for access to a receptive member of the opposite sex. Today’s dogs encounter these problems far less frequently, if at all; access to all these key resources is controlled by people, and so the requirement to be very careful before starting an unwinnable fight, imposed on wild animals by natural selection, is no longer as important for dogs. Nevertheless, starting fights under any circumstances is risky for dogs, for three reasons. First, because they may get seriously injured, which could, before the advent of modern veterinary surgery, have had life-threatening consequences. Second, because their human masters, for most breeds at least, value tolerance of other dogs, and tend to breed from animals that were tolerant of the proximity of other dogs. Third, because dogs with high general thresholds for aggression pose the least risk to humans, especially to children.

  Because of the various incentives not to start fights, aggression is unlikely to be the default strategy for any normal domestic dog when it meets another of its own kind. Most owners are able to exercise their dogs in public areas without fear that they will try to ‘dominate’ every unfamiliar dog they come across. Even at first encounters, most dogs prefer to use signalling to defuse any potential aggression: hackles may be raised briefly, but are then lowered again as the two participants decide either not to get too close to one another, or to approach one another to sniff since the associated risk is low.

  As two dogs get more familiar with one another, they will learn to look for specific signs that predict how the other is going to rea
ct to any given action. For example, if a puppy is brought into a household that already contains an older dog that likes its food but has little interest in toys, the puppy will learn that trying to steal food gets a warning growl, but that grabbing a toy from under the other dog’s nose does not. A harmonious relationship should build up between the two, in which each knows and respects the other’s preferences and quirks. The way this happens need not, and probably does not, involve highly sophisticated intelligence on the part of either dog; simple learning is probably sufficient. If there are three or more dogs in the household, however, it is likely that each dog can rely on its observations of the other two interacting, as well as its own interactions with them, as part of the information that it can use to predict how they will behave in a particular situation.

  Importantly, when dogs are learning about other dogs, they do not do so in the same way we humans do when we are establishing our own relationships. Dogs are unlikely to be able to associate events (unless they occur within a few seconds of each other) or to perform mental time-travel. We have no evidence that dogs can actually think ahead in order to predict what another dog will do next, or even recall specific incidents from the past that involved that dog. Rather, they seem continually to refine simple ‘rules of thumb’ that enable them to get along with one another – ‘avoid that other dog when he’s eating’, or ‘playing a tugging game with this dog is fun because he lets me win sometimes, but it’s not fun with that one because he always wants to take the toy away’.