In Defence of Dogs 

defense dogs

 Click on the cover above to go to this book at

John Bradshaw is a biologist who heads the Anthrozoology Institute at Bristol University. His 'In Defence of Dogs' (called 'Dog Sense' in the US) has become a best-seller, partly because it presents some interesting research in a way that's easy to read.;

The book certainly has some strengths, for example an account of feral dogs in West Bengal, where dogs show pack-like behaviour, with males sometimes helping to feed their young (p73). Bradshaw contrasts this with the less pack-like behaviour of feral dogs in Italy. So, while he argues that dogs are not pack animals in the way wolves are, he does recognise that environment plays a part in how canids organize their lives.

Similarly, he's uneasy about the idea of hierarchies among dogs, but he does note that some groups of dogs, especially sled and hunting dogs, tend to show clear hierarchies (p128). The issues of whether or not groups of dogs tend to 'pack', and the extent to which they form hierarchies are complex. The evidence cited by Bradshaw reflects this complexity, even if his conclusions may be a little simplistic.

Bradshaw also explains the potential of dogs for bonding with humans and other species, which sets them apart from wolves. Wolves can be tamed, but a tame wolf isn't truly domesticated, geared for life with humans. His account of the risks of breeding dogs for appearance is a useful reminder to would-be owners who may be seduced by the looks of the dog. Lastly, Bradshaw emphasises that dog-owners need to be more considerate of non-dog owners, and use discipline, or control, to prevent their pets from annoying people.

Actually, These Issues Are Complicated

So, Bradshaw makes some important points in a persuasive way. What are the book's weaknesses? First, Bradshaw's throw-away remarks on supermarket pet food as safe and nutritious are misleading. Canine nutrition is a complex topic. Bradshaw is aware of this, since he has worked in the pet food industry. He could either explain the issue in some depth or simply omit the topic.

Secondly, his treatment of aggression and its links to breed is simplistic. He illustrates the topic of by showing a pit bull baring its teeth (p269). This contributes to popular fears, especially in the UK, where most people have never knowingly met a pit bull. Bradshaw quotes statistics which appear to show that a higher proportion of pit bulls are involved in reported bites, compared with other breeds. They're dubious figures, given that less is known about the total population of pit bulls than of the other breeds quoted, such as German shepherds and Rottweilers. Furthermore, as Bradshaw himself notes, dogs from similar breeds are often described as pit bulls. So why include these statistics, if he knows they are misleading?

A definition of 'aggression' would be useful, and more recognition that dogs using their teeth as weapons isn't a one-dimensional issue. A dog may, for example, be very tolerant of small children, and extremely intolerant of other dogs, or easily spooked by intrusive owner handling, but too fearful of strange humans to be a danger to the public unless cornered. It's also worth researching the circumstances of bites, which often point to the value of public education, especially for children. Adults are not always sensible either. Some people in the UK are prone to petting strange dogs, especially those which look cuddly and 'safe', like golden and labrador retrievers, but temperament can vary enormously within 'cuddly' looking breeds.

Thirdly, Bradshaw's book is geared to pet dog owners. His throw-away remarks at the end, that owners of gundogs, sheepdogs and guard dogs are more likely to see their dogs as tools rather than companions (p277) do beg questions. Pet dog owners often choose dogs unwisely, following fashions rather than understanding the nature of what they are taking on. The dogs may be left alone for much of the day. A shepherd has a good idea of the dog he wants, and spends more time working with the dog as a partner. Which is the tool, the fashion accessory, or the working partner? As a shepherd friend said when his dog was stolen 'They not only stole a valuable dog with useful skills, they also stole a companion'.

Do Bradshaw's Conclusions Fit the Facts?

The biggest flaw of this book, however, is that Bradshaw's treatment of training fits uneasily with his evidence. He argues that 'traditional' trainers are misled by studies of captive wolves into thinking that dogs want to dominate humans. Recent research has shown wild wolves to be friendlier than previously thought, and dogs as more sociable than wolves. So the advice that owners should assert dominance by status reduction efforts, like eating or going through doors before the dog, is suspect. So far so good. Interestingly, Cesar Millan, Bradshaw's bugbear, has rethought his ideas on dominance, some of which were drawn from Bruce Fogle's 'The Dog's Mind', published in 1990 (1).

Bradshaw goes on to argue that, because dogs aren't wolves, we should use training techniques developed by wild animal trainers such as Karen Pryor. These techniques are based on operant conditioning, a method of learning developed by B.F. Skinner in the 1930s. He was a founding figure in a school of psychology known as behaviourism. Pryor emphasises the reward element of operant conditioning.

A key figure in British training, John Fisher, doesn't fit Bradshaw's model of traditional vs modern. In 'Think Dog', which came out in 1990, Fisher argued both for status reduction efforts, and for Karen Pryor's version of Skinner's model, with her emphasis on rewards (2). Fisher's position was actually more coherent than Bradshaw's - Fisher saw dogs as similar to wild animals, and advocated methods that were used with wild animals. Later he changed his mind about status reduction efforts, but sadly died before he had a chance to develop his ideas in depth.

In contrast to Fisher, Bradshaw has had the benefit of recent research on the nature of dogs, so could weave this research more firmly into his ideas on training, rather than opting for Skinner's somewhat impoverished view. Skinner saw punishment as a stimulus that discouraged a behaviour, and a reward as a stimulus that encouraged a behaviour. The animal's desires, its feelings, and its relationship with its handler were irrelevant, only observable behaviour was important. Ian Dunbar, whom Bradshaw praises as a 'modern' trainer, stresses that punishment 'doesn't have to be painful or scary', and, though Dunbar emphasises rewards, he argues that 'you will not get reliability unless you punish the dog when he gets it wrong' (3).

Bradshaw, however, defines 'positive punishment' as 'learning that results from physical pain or discomfort' (p113). This confuses the technical notion of punishment as a stimulus that discourages a behaviour with its everyday meaning of an unpleasant stimulus which may or may not discourage the behaviour. Bradshaw's definition also doesn't fit the facts. There are many stimuli that can discourage unwanted behaviour, and not all involve physical pain or discomfort. Dogs can learn not to do something simply because it annoys or upsets their owner, which in turn may upset the dog - mental rather than physical discomfort. Furthermore, some stimuli may just convey the message 'I don't want you to do that', which a dog may accept as useful information rather than something upsetting. Skinner, who focused on observable behaviour, wasn't interested in explaining that sort of communication.

Neither was Skinner especially interested in individual personalities, and there's a great variety in canine personality. Some dogs can be very upset by a slightly cross tone, while others respond to an angry bellow with a look that says 'Did you speak?' The best guide to whether or not you're seriously upsetting a dog is what that dog is telling you. It makes sense to read the dog in front of you rather than to read a book that makes assumptions about your dog.

Now, under what circumstances might we accept someone telling us 'I'd rather you didn't do that', or even 'You really mustn't do that' without feeling attacked? When we need information, and we trust the person who is giving us this information. In general, dogs are constantly checking us out for cues as to what's happening, and when something interesting, like a walk, might be on the cards. Some dogs, especially those designed to work closely with humans, have a very strong need for information, and can suffer when we don't give it to them. Steven Lindsay, who has written a science-based work on behaviour and training, places a lot of emphasis on building up trust (4). When dog and owner have learnt to trust one another, the meaning of 'Don't do that' changes.

Overreacting, for example, bellowing at a sensitive dog, is obviously unhelpful. Millan, mentioned by Bradshaw as an example of a 'traditional trainer', says 'it's never, ever right to punish a dog out of anger'(5). Millan also argues that 'Punishing a dog does not cure aggression - usually if a dog is in the red zone, it exacerbates it' (5). He uses 'punishment' in an everyday sense, and prefers the term 'correction' as a way to set limits. This is actually closer to the behaviourist concept of 'punishment' than the everyday meaning of the word (6).

Clearly, the word 'punishment' arouses our emotions, and we're not comfortable with it. It's also clear that relying too much on punishment, in either the everyday or the technical sense, doesn't help foster a good relationship. Dogs need to know what they are meant to be doing, as well as what they are not meant to be doing. However, the opposite extreme is also unhelpful. If you're afraid of ever saying 'don't do that' in case it causes upset, then you're failing to provide useful information. One difference between dogs and wolves is that it's easier prevent a dog doing something it enjoys. Dogs care more about what we want, and are better able to learn to trust us.

Bradshaw is well placed to explain how to develop a relationship of trust, but the sense of a developing relationship is missing in his chapter on 'The Science of Dog Training'. Furthermore, it's not until the end of the book that Bradshaw mentions one reason for training, and recognizes that some suffering may be necessary if dogs are to be discouraged from behaviour that could endanger them - or others, especially children. And, as Lindsay notes, training can also greatly improve the quality of a dog's life, by allowing owners to trust their dogs more and give them more freedom.

Bringing Up A Dog

Bradshaw doesn't actually define 'training', though his chapter on the subject covers teaching commands, eliminating unwanted behaviour, and teaching skills like retrieving. In a broad sense, basic training goes beyond this. It provides dogs with the skills they need to cope with everyday life. This includes coping with the physical world, as well as learning the social rules of the species the dog interacts with. We try to bring up dogs and kids to be the kind of individuals people want to be with, and to have strengths such as impulse control and resilience. This task is easier if we have a relationship of trust.

Bradshaw quite rightly stresses that we can have a richer relationship with a dog than with a wild animal, because a dog is capable of bonding and co-operation. First dogs need to be socialized. Bradshaw argues that the term 'socialization' 'ought to be reserved for what happens during the sensitive period' (p137), not during the juvenile period. Such a narrow definition is perhaps unhelpful. Socialization during the sensitive period is important, but isn't enough to allow dogs to develop their potential, especially if they are then kept isolated as juveniles.

Furthermore, what Bradshaw doesn't explain is that the ultimate goal of socialization is learning the social rules which allow the dog to get on with humans, other dogs and other species such as cats (7). Bradshaw could spell out that socialization goes beyond just meeting individuals from another species. Youngsters can learn good habits from sensible children, calm, well-mannered dogs, and cats that stand their ground. Unfortunately, they can also learn bad habits from unruly children and dogs, and cats that offer exciting chase games!

The Good Guys vs the Bad Guys

A major problem with this book is that Bradshaw perpetuates stereotypes, not just relating to pit bulls, also to polarization within the world of trainers. He presents the 'good guys', such as Karen Pryor, Jean Donaldson, and Ian Dunbar, vs the 'bad guys', notably Cesar Millan. John Fisher, who doesn't fit into this picture, doesn't get a mention, nor do the significant differences between some of the 'good guys' such as Pryor and Dunbar.

So Bradshaw sets up a simplistic division of dog people into 'two camps' in his chapter on training. He later complains, at the end of the book, about the problem of polarization within the world of training! What do Bradshaw's 'good guys' have in common? A belief that Skinner provides the best model for understanding how dogs learn, so the best way to train dogs. Is this a 'scientific' approach to bringing up a dog? Given the limitations of Skinner's model, perhaps not. Take the process of becoming a sheepdog. A pup starts to learn the ways of sheep long before the shepherd allows the dog to actually work them. Inexperienced youngsters may learn from more experienced dogs. Shepherds and their dogs can communicate in ways that are mysterious to onlookers. Sometimes shepherd directs the dog, other times the dog uses its initiative. Behaviourism focuses on what can be understood from observing single interactions, but what we have here is a long-term process of development, both of the skills of the dog, and the relationship between dog and shepherd.

Most dog owners don't have sheepdogs, we have pets. One major drawback of Bradshaw's 'modern methods' is that they can turn bringing up a pet dog into a very disjointed process. Responsible owners who rely on professionals may first 'socialize' the pup, then take the youngster to obedience training classes, teaching commands with 'hands off' methods. Then they teach the dog to accept being handled in separate lessons. However dogs are learning 24/7, not just in the few hours a week they may spend with professionals.

The issues that Bradshaw tackles after his chapter on training, like socialization, bonding, canine cognition and emotions, are all relevant to understanding how dogs learn. A chapter looking at training in a broad sense would fit at the end of the book, drawing on previous chapters. Bradshaw, for example, complains about the behaviour of dogs in his local park. Here's an opportunity to spell out that dogs are capable of learning social rules, like 'be gentle with children', or 'a polite dog does not chase cyclists', that dogs are learning all the time, so can learn good or bad habits from walks in the park, and that it's easier to teach polite behaviour right from the start than to eradicate bad habits later on. Socialization goes beyond just meeting another dog or human, dogs need good teachers to help them learn how to behave.

There's actually a fair amount of consensus among trainers on some issues, for example that if we want mannerly dogs, we need to be mannerly ourselves, and be considerate of cyclists and small children. Trainers also agree that right from the start, owners need to be able to handle pups in ways that calm them, rather than overexcite them, and that raising a dog is a long-term process, requiring structured lessons pitched to the youngster's level of development, gradually becoming more challenging as the dog develops its abilities.

There's little sense of this long term development from puppyhood to maturity in Bradshaw's book, partly because of his narrow approach to training. Perhaps also, Bradshaw's vision of a dog doesn't include maturity. He argues that canine 'behaviour becomes arrested at the (wolf's) juvenile stage' (p58), a vision that to some extent infantilizes adult dogs. People sometimes expect too much of pups. Ian Dunbar believes that an 8-week-old pup should come housetrained, and be able to 'come, sit, lie down and roll-over on cue'(8). Conversely, we may expect too little of adult dogs. A very important role for canine professionals is providing owners with long-term goals, and a vision of what their youngster may be capable of, if they put in the effort.

Is it worth reading this book? It's easy to read, though it might make you sputter and fume at some of the more simplistic assertions. However, people who are seriously interested in dog behaviour, and in particular, cognition, will find Miklosi more rewarding, while those who want to link behaviour and training can find a wealth of resources in Steven Lindsay's Applied Dog Behaviour and Training, especially vol 3. Neither work is perfect, both require more commitment and concentration than Bradshaw's book, but both authors are more thorough, take more pains to ensure their conclusions fit the facts, and their arguments are more coherent. In short, both take a more scientific approach.

Review by Alison Lever, 2014


1) Fogle, B. (1990) The Dog's Mind. Pelham, London

2) Fisher, J. (1990) Think Dog. Blandford, London

3) quoted in Millan, C. (2011) Cesar's Rules, Ch 6 Dr Ian Dunbar and Hands-Off Dog Training. Hodder and Stoughton, London, p101.

4) Lindsay, S. (2005) Applied Dog Behaviour and Training vol 3, Procedures and Protocols. Blackwell, London

5) Millan, C. (2006) Cesar's Way. Hodder and Stoughton, London, p194

6) ibid, p217

7) Miklosi, A. (2007) Dog: Behaviour, Evolution and Cognition. Oxford University Press, Oxford, p22)

8) quoted in Millan, C. (2011) Cesar's Rules, Ch 6 Dr Ian Dunbar and Hands-Off Dog Training. Hodder and Stoughton, London, p191