11/30/2015 05:14 pm ET | Updated Nov 30, 2016

How and Why Dogs Play Revisited: Who's Confused?

Stefan Cioata via Getty Images

Watching dogs play is very exciting, and there has been a lot empirical research on how and why dogs (and other animals) engage in this activity with boundless zeal. A number of people have asked me to comment about dog play after reading this section in a new book by Raymond Coppinger and Mark Feinstein called How Dogs Work. So, I decided to do so.

The authors begin their chapter 9 on play by claiming, "Hundreds of scientific papers have been written on the subject of 'play' behavior - an activity for which dogs are, of course, famous." (p. 159) Recognizing that there is a solid and growing literature on play - there's really no reason to put the word play in square quotes - I assumed that what followed would be a detailed review of this research, but rather, what I discovered was a disjointed discussion of play and not an in-depth review of the scientific literature. Instead, the authors offer their own unpublished observations and the results of unpublished student projects, all of which are impossible to assess.

Do dogs and other animals actually play? Coppinger and Feinstein write that they put the word play in scare quotes because "in spite of the fact that people feel like they know it when they see it, it's not at all obvious that play is a unitary 'thing-in-itself' that can easily be characterized, let alone explained in evolutionary terms." (p. 161) No one I know who has spent years studying play would argue that play is a "unitary 'thing-in-itself'," nor would they agree that play cannot be explained "in evolutionary terms." Indeed, some of the references the authors include show there are a number of highly plausible evolutionary explanations (and the University of Tennessee's Gordon Burghardt, who has studied comparative aspects of play for many years and wrote The Genesis of Animal Play, provided the Foreword for Coppinger and Feinstein's book).

Why do animals play? Briefly, various theories have been offered about why animals play, and there's no one explanation that fits all examples of animal play. Detailed comparative data show play is important in social development, physical development, and cognitive development. And, neurobiological research strongly suggests play can be pleasurable and fun and animals may simply play because it feels good, "for the hell of it." Indeed, many researchers are taking fun seriously, and the 25th anniversary issue of the journal Current Biology was devoted to the biology of fun with many play researchers weighing in on the topic. Coppinger and Feinstein write, "We agree that there is good reason to believe that animals derive pleasure from play - indeed they do from all of their motor activities." (my emphasis) While animals might derive pleasure from play, eating, and sex, it's difficult to argue they feel good running from competitors or predators, but the necessary research has not been done.

Based on an extensive review of available literature, my colleagues Marek Spinka, Ruth Newberry, and I proposed that that play functions as training for the unexpected by increasing the versatility of movements and the ability to recover from sudden shocks, such as the loss of balance and falling over, and to enhance the ability of animals to cope emotionally with unexpected stressful situations. To obtain this training, we suggested that animals actively seek and create unexpected situations in play and actively put themselves into disadvantageous positions and situations.

Comparative data from a wide range of species support this hypothesis. And, while it is difficult to test these ideas in the field, a study of mountain goat kids by Rachel Théoret-Gosselin, Sandra Hamel, and Steeve D. Côté called "The role of maternal behavior and offspring development in the survival of mountain goat kids" showed that "play behaviors could enhance the emotional resilience to stress not only for unpredicted events but also in stressful group situations because play could reduce aggressiveness in gregarious species." (p. 183) More field data are needed and this study is an excellent example of what needs to be done.

The play bow: Are dogs really confused when they play and what does this mean?

The authors also dismiss the detailed work that has been conducted on the play bow, a highly ritualized and stereotyped action by which animals signal their intention to play (please see accompanying image). When dogs and other animals bow they crouch on their forelimbs, raise their hind end, and occasionally wag their tail and bark. Coppinger and Feinstein write, "But we wonder if the so-called play bow in fact really has any adaptive, let alone cognitive, significance." (p. 168) A good deal of very detailed research has been conducted on the bow by my research group and also by Barbara Smuts and her students that clearly supports the claim that bows are adaptive and have cognitive significance (please also see along with Mechtild Käufer's excellent book called Canine Play Behavior: The Science of Dogs at Play and a comprehensive review essay by Elisabetta Palagi and eight other play experts called "Rough-and-tumble play as a window on animal communication"). The abstract for this excellent evidence-based and extremely significant up-to-date essay reads: Rough-and-tumble play (RT) is a widespread phenomenon in mammals. Since it involves competition, whereby one animal attempts to gain advantage over another, RT runs the risk of escalation to serious fighting. Competition is typically curtailed by some degree of cooperation and different signals help negotiate potential mishaps during RT. This review provides a framework for such signals, showing that they range along two dimensions: one from signals borrowed from other functional contexts to those that are unique to play, and the other from purely emotional expressions to highly cognitive (intentional) constructions. Some animal taxa have exaggerated the emotional and cognitive interplay aspects of play signals, yielding admixtures of communication that have led to complex forms of RT. This complexity has been further exaggerated in some lineages by the development of specific novel gestures that can be used to negotiate playful mood and entice reluctant partners. Play-derived gestures may provide new mechanisms by which more sophisticated communication forms can evolve. Therefore, RT and playful communication provide a window into the study of social cognition, emotional regulation and the evolution of communication systems.


The so-called play bow. Based on an unpublished student project in which "Border collies were confronted with normal and drugged roosters" (p.168), Coppinger and Feinstein believe that the "so-called play bow" is a posture assumed by an animal confused by its next move. They write, "... the play bow occurs when an animal is in a temporarily indeterminate state ... In short, the 'playing' animal is in conflict about its next move - and the play bow in fact looks just like a combination of multiple conflicting behavioral shapes." (p. 170) The authors ignore detailed research that shows how play bows are extremely stereotyped (they are what ethologists call a modal action pattern), they vary in shape and duration depending on where they are performed in a play bout, and they allow a dog to perform a wide variety of movements from this posture. There are no data that support their belief and the student's data are impossible to assess. And, it's not clear at all why they refer to the "so-called play bow," when many researchers have studied it and concluded, based on substantial data, that it is indeed used almost exclusively in the context of play both as a play invitation signal and also to maintain the play mood.

Let's briefly think about what it means when a dog or other animal is confused, because every definition I can find indicates that there have to be cognitive and emotional underpinnings. In the case of dog-dog play, a simple view would be that Harry (a dog) wants to play with Mary (another dog) and that Harry isn't sure what to do so he carefully pays attention to what Mary has done and is doing, and tries to factor this information into what she is likely to do in the future. In essence, Harry is pondering if he chooses to do "X" or "Y," what will Mary do (and, of course, vice versa). Because play is indeed a hodge-podge of various actions, a kaleidoscopic behavior, on the authors' view, Harry is confused, and to overcome his confusion he performs play bows.

There are no data that support the belief that dogs are confused when they play, however, there are data that show that there is a good deal of rapid of thinking and feeling on-the-run based on what Harry thinks and feels Mary is likely to do during the on-going interaction (and vice versa). These sorts of interactions make it clear that play is also a good place to observe and to study what researchers call a "theory of mind," because Harry and Mary need to pay very close attention to what each has done and is doing, and how that will influence what she or he is likely to do in the future (for further discussion please see Alexandra Horowitz's essay called "Attention to attention in domestic dog (Canis familiaris) dyadic play"). There is a good deal of mind-reading going on here as Harry and Mary make careful and rapid assessments and predictions of what their play partner is likely to do.

The cognitive and emotional underpinnings of "being confused" are rather rich, and do not lend themselves to simple mechanistic explanations that are favored by the authors. Available and ample data for a number of different species show there are predictable rules of play that cross species lines, namely, ask first, be honest, follow the rules, and admit you're wrong. This is why play is so exciting to engage in and also so much fun to watch and to study. And, this is also why play among young and old dogs only rarely escalates into injurious aggression, although the authors recall an instance when play among four-week-old Border collie littermates was fatal and use this observation to claim that play "can itself cause significant harm" (p. 165). Indeed, Shyan, Fortune, and King (2003) reported that fewer than 0.5% of play fights in dogs developed into conflict, and only half of these were clearly aggressive encounters. Their data agree with our own observations on wild coyotes and free-running dogs at play.

Behavioral variability. Another example of a claim that is countered by available data centers on behavioral variability in young dogs and wolves. Coppinger and Feinstein write, "When we observe wolves, we see a similar picture. Wolf puppies are often noticeably more robust and varied in their play routines than dogs of the same size and age. This means, according to our hypothesis, that they should have more available motor patterns than the dogs do. That is in fact the case." (p. 178; my emphasis) However, they offer no data.

Along these lines, years ago Robert Fagen, another play expert and author of the classic book Animal Play Behavior, analyzed the sequential variability of play and aggression in young dogs (beagles), wolves, and coyotes using data my students and I collected, and discovered that social play in the beagles was more variable than social play in wolves and coyotes of the same age (and coyote play was more variable than wolf play). These data were published in an essay I wrote with John Byers ("A critical reanalysis of the ontogeny of mammalian social and locomotor play: An ethological hornet's nest," in K. Immelmann, G. W. Barlow, L. Petrinovich, and M. Main, Eds., Behavioral Development, The Bielefeld Interdisciplinary Project. New York: Cambridge University Press, pp. 296-337, 1981) that the authors list in their reference section. And, we also found that young beagles and wolves shared the same basic ethogram and number of motor patterns. Perhaps working dogs such as Border collies are different from beagles and other dogs, but we really don't know if this is the case.

The way in which the authors routinely dismiss a wealth of detailed research on animal play is characteristic of much of their book, that is essentially a tapestry of criticism using stories and unpublished projects in lieu of published data. It's easy to see how one might walk away feeling that just about everyone else is wrong about dog behavior, cognition, emotions, and consciousness, and much of the research that has been done can be tossed into the garbage because it's merely debris.

All in all, the one-sided assault on the ever-growing literature in the growing field called cognitive ethology (the study of animal minds) fails. How Dogs Work does not really tell us how dogs work, but rather provides an extremely narrow view of mainly working dogs as machines. I find the topic to be of great interest and am always eager to learn more about why some people favor reductionist and mechanistic accounts to explain the behavior of complex sentient beings (see, for example, Sara Shettleworth's book, Fundamentals of Comparative Cognition). However, How Dogs Work doesn't convince me that the authors' over-arching views are tenable. Beliefs don't substitute for data that have been reviewed by peers, and there are plenty of data that are readily available.

All in all, we really know far more than the authors offer, and for numerous wide-ranging and critical discussions of many different aspects of dog behavior I suggest Domestic Dog Cognition and Behavior: The Scientific Study of Canis familiaris edited by Alexandra Horowitz, Adam Miklosi's Dog Behaviour, Evolution, and Cognition, The Social Dog: Behavior and Cognition edited by Juliane Kaminski and Sarah Marshall-Pescini, and Mechtild Käufer's Canine Play Behavior: The Science of Dogs at Play. For more on play I highly recommend the excellent and comprehensive review article by Elisabetta Palagi and her colleagues called "Rough-and-tumble play as a window on animal communication" and (in addition to the references above) Sergio Pellis and Vivien Pellis' The Playful Brain: Venturing to the Limits of Neuroscience.

What's so incredibly exciting about the study of play behavior and the cognitive and emotional lives of dogs and other animals is how much we're learning about how individuals negotiate challenging and complex social and non-social situations by carefully analyzing what's happening and by using hard-wired actions when needed (for example, when they need to do the right thing instantaneously or the first time they are faced with a specific situation and there's no room for error), along with behavior patterns that require careful thought and flexibility motivated by what individuals are feeling about the situation in which they find themselves.

Please stay tuned for more on dog behavior, cognition, and emotions, because there is a lot of research being done by research groups around the world, and we still have a lot to learn. Dogs are amazing sentient beings who challenge us in many different ways.

I thank a number of people for help with this essay.

Note: In an email message I was asked if I knew what happened to the 1000's of sled dogs dogs for whom Dr. Coppinger was responsible. On page 25 we're told, "Some four thousand dogs 'went through the yard'" when "Ray spent fifteen years breeding and training dogs that pull sleds." I have no idea, but according to some people I consulted, this is an incredibly large number of dogs, an average of around 267 a year.