Like the Star Wars movies, the saga of the use of the “dominance” concept in dog training and behaviour modification has recently entered its third manifestation in the academic literature. The trilogy began with a series of papers put together by Steve Zawistowski and Gary Patronek for the Journal of Applied Animal Welfare Science in 2004, entitled “A dog in wolf’s clothing”. For the second phase, Rachel Casey, Emily-Jayne Blackwell and I wrote a paper for the Journal of Veterinary Behavior , published in 2009, exploring alternatives to “dominance” for conceptualising how dogs think about social relationships. This prompted attacks in the “dog press”, especially from organisations keen to promote forceful dog training, but it was not until 2014 that any academics joined in (at least in print), when Matthijs Schilder and Claudia Vinke from Utrecht University, and Joanne van der Borg from Wageningen University, both in the Netherlands, published a commentary “Dominance in domestic dogs revisited: Useful habit and useful construct?”
The points on which we disagree might be dismissed as a typical spat between rival academics, except that the “Utrecht School” version can be and has been seized upon as legitimising the infliction of pain as a way of training dogs. This may be particularly dangerous for dog and owner alike because it appears to be a self-fulfilling theory: in the words of Peter Sandøe and co-authors on page 138 of their excellent new book Companion Animal Ethics “So it is suggested that Dominance Theory, when applied to dog training, may serve as a self-reinforcing hypothesis: by using physical force, the owner elicits an aggressive response from the dog, which in turn is interpreted as a sign of dominance; alpha-rolls and other forms of physical confrontation may actually increase the risk of an aggressive responses from the dog”.
Accordingly, our response, published last month in the Journal of Veterinary Behavior, seeks both to clarify our original position and to comment on some of the issues raised by Schilder and his colleagues, especially those that relate to the welfare of pet dogs.
First, it is clear that our original criticism of the application of the “dominance” concept to Canis lupus familiaris as a species has been interpreted as applying much more widely than we had intended, and we admit that we may not have made this sufficiently clear in our 2009 paper. Our focus then – as now - was intended to be restricted to “the concept of ‘dominance’ in the diagnosis and treatment of dogs that have displayed aggression” – but some seem to have taken our paper as an attack on the concept of dominance in describing the behaviour of free ranging (“feral”) dogs. Not so: in our new paper we reaffirm that dominance is, of course, a well-established concept in academic ethology for extracting underlying social structure from observations of the interactions between the members of any group of animals. We see no reason whatsoever why it should not be used to probe how packs of free-ranging dogs are organised.
However, we and many authors before us have counselled against the presumption that simply because a hierarchical structure can be measured in a group of animals, that the animals themselves are aware of that structure, or are striving to achieve “dominance” within it. Mindless robots with slightly different software or physical characteristics will form measureable “hierarchies” if allowed to interact repeatedly (see this earlier blog post).
So, do dogs “think about” dominance? Are they even capable of “thinking about” their position in the hierarchy? The past two decades have witnessed an explosion of research into the dog’s mind, but have failed to demonstrate that dogs possess “theory of mind” – they seem to have little concept that other dogs – or humans – are capable of independent thought. Rather, the consensus is emerging that if dogs are capable of “thinking about thinking”, they don’t do it in the same way that we do. They are, however, adept at fooling us that they think more than they actually do, because they are such exquisite readers of human behaviour
It is indeed possible that the carnivore brain is constructed in such a way as to preclude any appreciation of intentionality. Kay Holekamp’s laboratory at Michigan State University has concluded that spotted hyenas, the most socially complex of all the Carnivora (far more adept than the wolf), construct their outwardly sophisticated cultures by means of simple associative learning.
Matthijs Schilder and colleagues cite “the logic of the Utrecht School of former professor Jan van Hooff and his former pupil Frans de Waal” as the theoretical basis for their interpretation of dominance behaviour in dogs – but this model is based largely on the behaviour of chimpanzees, which are known to possess a fairly sophisticated (second-order) theory-of-mind, surpassed only by that of humans. We are concerned that this model predicts capabilities for dogs that they may very well not possess, leading to their being punished for “crimes” they can have no comprehension of.
It is easy to imagine that because we humans know that dogs have minds, so must they – a simple anthropomorphism of the kind that drives all dog-human relationships – but so far there is no evidence to support this. If dogs don’t realise that other dogs – or their owners – are capable of thinking about them, how could they be planning to “climb the hierarchy”? It is more parsimonious to interpret dogs’ behaviour as if they were simply trying to maintain access to essential resources, perhaps the most important being, uniquely for this species, access to one or more human attachment figures.
From the perspective of their welfare, we are most concerned that Schilder and colleagues promote the idea that dogs need to learn to “accept a submissive status”. Because they are not specific as to how this might be achieved, enthusiasts for punishment-based training methods appear to have been given their tacit approval. Both for their own safety and to be acceptable to society, companion dogs need to be kept under control, but that can be achieved by reward-based training, without reference to their position in some illusory “hierarchy”.