In a paper now 60 years old ‘A Theory of Play and Fantasy’ Gregory Bateson the brilliant anthropologist discussed the topics of communication, the logic communication follows, and how communication implicitly works in a psychotherapeutic setting. Bateson drew evidence from work in mathematical logic, observation of animal behaviour, and psychotherapeutic data.  In this blog-post I will discuss Bateson’s interpretation of animal behaviour and how his interpretation of animal behaviour influences his views on psychotherapeutic treatment.

Bateson gives the evolutionary conjecture that our current linguistic capacities may have arisen out of the capacity of non-human animals to engage in play, to issue threats, and deceit. These forms of behaviour indicate that the animal is not merely using a signal to denote internal mood or external object. Rather the behaviour is meta-communicative. If you consider a common form of behaviour in dogs the ‘Play-bow’, where the dog stands in front of the owner and bows the front part of the body as he stares intently on his owner. This form of behaviour is typically engaged in when the dog wants to engage in play. Once the dog has indicated he is playing with the ‘play bow’ the dog will then pretend to fight, or run away to indicate he wants to be chased etc. Here is a short video showing two dogs using the ‘play bow’ https://www.youtube.com/watch?v=Jf3Tzx1ooIM .

The ‘play-bow’ has been well studied scientifically Hare and Woods describe it as follows:

“Dog’s frequently use visual signals to communicate with others during their natural interactions. When dogs are playing, careful observations have shown that gestures such as play bows (dropping the chest to the ground and preparing to spring back at the first sigh of a chase) are used to signal that their behaviour is meant in good fun. Where a quick approach and forceful contact might usually lead to aggression, the same behaviour preceded by a play bow leads to a friendly response. When other dogs see a play bow, they can agree to the game with another visual gesture. They usually do this using a ‘self-handicapping’ gesture that makes them more vulnerable (rolling on their back). This means that dogs might flexibly use visual gestures to communicate, perhaps even more than vocal behaviour. ” (Hare and Woods ‘The Genius of Dogs’ p. 136)

John Bradshaw notes the same structural features of the ‘Play-bow’ as Hare and Woods and notes that dog play differs depending on whether dogs are playing with other dogs or with humans:

 “In short, dogs appear to be in a completely different frame of mind depending on whether they are playing with a person or another dog. When the play-partner is a dog, possession of the toy seems to be most important- and indeed, it is possible that competitive play is one way that dogs assess each other’s strength and character…When the play-partner is a person, however, possession of the toy seems almost irrelevant; the important thing is the social contact that the game produces. This finding is entirely compatible with the observation that dogs are unable to 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.” (John Bradshaw ‘In Defence of Dogs’ p. 204)

We can see from these descriptions that dogs can indicate when they are about to engage in play and can modify their play depending on who they are playing with.

Bateson noted that there is an interesting logic to animal play. Thus suppose a dog nips another dog on the ear while playing with him; Bateson would parse this as follows ‘These actions in which we now engage, do not denote what would be denoted by those actions which these actions denote’ (Bateson: ‘A Theory of Play and Fantasy’ p. 4). In other words the playful nip denotes the bite, but it doesn’t denote what would be denoted by the bite (ibid p. 4). Bateson’s interpretation of the logic of play behaviour is interesting as an analysis of the behaviour, but it becomes less convincing if we view it as a description of the animal’s psychological processes.

When a dog does a ‘play-bow’ is there any good reason to think that the dog is explicitly using it to indicate that all further actions within this frame are taking place on a different logical level? Here it is helpful to draw a distinction that Michael Devitt makes in his (2006) ‘Ignorance of language’:

“In sum, to be a processing rule that governs the activities of a chess player, logic machine or dancing bee is one thing, to be a structure rule governing the outputs of such activities is another. And a structure rule of the outputs may have no place among the processing rules that produce these outputs” (Devitt: ‘Ignorance of Language’ p. 21)

Devitt notes that while scientists were able to discover structural rules in the Bee’s waggle dance; this still left them with a hard job of discovering the processing rules that are used to cause the dance. It isn’t a priori true that the processing rules used by the Bee must represent or even embody the structural rules; rather all that is needed is that the processing rules respect the structural rules. A similar argument holds in the case of animal play behaviour; just because we can analyse the behaviour as having certain structural properties doesn’t mean that we should a priori assume that the animal’s behaviour is caused by processing rules that do anything more than respect the structure rules.

Bateson seems to think that when engaging in play the animal is capable of recognising that this play is similar to non-play behaviour. Thus in Bateson’s example of a dog playfully nipping another dog in play; Bateson thinks that the dog recognises that the nip is similar to a bite done in anger. The dog recognises that the play nip though similar in structure to an angry bite is different. Furthermore the dog knows that the other dog he is playing with can recognise the distinction as well.

Now while it is entirely possible that the Bateson’s hypothesis is the correct one, it should be acknowledged that simpler explanations of the dog’s behaviour are possible which don’t attribute such complex propositional attitude thought processes to the animal. At the very least we need some further experimental data to help us decide between Bateson’s hypotheses and simpler rival explanations.

In his 1991 paper ‘Bateson’s concept of “Metacommunication” in Play’ Robert Mitchell noted that it is unclear whether animals recognise that their play is similar to fighting behaviour:

“Do organisms recognize that their partner simulates other actions? This one question is actually two: Do organisms distinguish between the simulated actions in play and the actions they simulate? Symons (1978 p. 124) answers the first question affirmatively for monkeys: “both the human observer and other monkeys can almost always distinguish playfighting and fighting.” The organism’s differentiation between simulation and that which is simulated (see Fagen, 1981, p. 356) need not be evidence of the organism’s knowledge of the simulation as such, since it fails to tell us whether or not the animal recognizes the similarities amid the differences, that is whether or not the simulation represents or refers to the simulated thing for the organism. If an organism recognizes that s (the simulation) is a simulation of f, then they recognise that s is not f…Thus, the significant question is whether or not organisms ever recognize the similarity or “sameness” between simulation and simulated actions of another while also recognising the difference between them.” (Mitchell, R ‘Bateson’s concept of “Metacommunication” in Play’ p. 79)

Mitchell thinks that evidence of dogs trying to avoid being deceived by their owner in games of “fake out” is good evidence that the dogs engage in dual recognition of both literal and play moves (Mitchell and Thompson, 1991). However the evidence provided by Mitchell is inconclusive at best. So as it stands we don’t have conclusive evidence as to whether animals understand play in the way Bateson suggests. Furthermore it is pretty sloppy to even speak of animal cognitive capacities in relation to play given the different capacities of animals. However, even when we specify and just speak of a particular species such as the domestic dog, we are no closer  to understanding whether Bateson’s conjecture is accurate about the understanding dogs have of the meaning of play.

Obviously a valuable heuristic to work with is to adopt the simplest explanation of the behaviour consistent with our current knowledge. Thus Daniel Dennett in his recent book ‘From Bacteria to Bach and Back’ discusses complex animal behaviour which appear perfectly rational and he shows ways that we can explain this behaviour without implausibly attributing to the animal representations of the reasons for this behaviour. Thus Dennett discusses cases such as Cuckoo’s dropping their eggs in the nest of another bird. When the baby Cuckoo is born he proceeds to push the bird’s eggs out of the nest with the “aim” that the bird will use all its resources to feed the baby Cuckoo. While there is no reason to assume that the baby Cuckoo represents reasons for his behaviour; we can appeal to what Dennett calls ‘Free-floating-Rationales’ to explain his actions. In this sense we can argue that there is a reason for the baby Cuckoo’s behaviour even though the reason wasn’t represented by any subject. These reasons can be discovered by theorists after the fact who use ‘reverse-engineering’ to discover the evolutionary logic that makes sense of the Cuckoo’s behaviour. Dennett uses this logic to explain the behaviour of other animals without attributing complex propositional attitudes to animal; thus he explains antelope stotting by using the logic of natural selection and ‘free-floating-rationales’. Antelope stotting is when Antelopes leap up in the air while being chased. The Antelopes who engage in stotting are less likely to be eaten by the Lion chasing them. Dennett notes that it is possible to tell a story where the Antelope realises that Antelopes who stot are less likely to get eaten and hence the Antelope decides to stot. Likewise it is possible to view the Lion as reasoning that creatures who can jump that high are likely to present more of a struggle than those who don’t stot hence the Lion avoids attacking stotting Antelopes. But Dennett skilfully shows that we can explain the behaviour of the Antelopes and the Lions without attributing these representational states by appealing to the logic of natural selection

(Dennett ‘From Bacteria to Bach and Back pp. 89-90). Dennett goes on to argue that competence without comprehension is nature’s way; beavers can build dams without representing reasons as to how to build them; birds can build nests without representing explicit plans as to how to build them; the same is true of termites building termite castles. So Dennett reasons plausibly enough that we shouldn’t attribute comprehension to creatures unless we have compelling reasons to do so.

Dennett’s logic can be used in the dog play case as well. There may be evolutionary reasons why dogs play together but the dogs don’t have to represent these reasons. Dogs can have competence in playing without necessarily having any comprehension as to why they play. It is easy enough to give an evolutionary explanation as to why dogs need to play interms of social bonding, practicing fighting etc without assuming that the dog ever represents these reasons or understands them at a meta-level like Bateson suggests. Obviously, using Dennett’s evolutionary logic to offer a simpler explanation of Dog play doesn’t settle the issue on whether Bateson is correct or not. To settle this issue detailed experimental data is needed. However despite their being good experimental studies on dog cognition ( see Hare, Tomasello, and Call), there is not as of yet sufficient data to decide the issue.

This issue isn’t just restricted to discussions of animal cognition. The issue actually extends into issues in mental illness. Bateson’s logical analysis of animal play was extended in a sloppy way into understanding children’s play, and even into understanding mental illness in general. Bateson argued that people, like non-human animals, often slipped between different logical levels when in conversation. He believed that some mental illnesses could be better understood if we could understand the different logical levels of communication. So Bateson’s understanding of animal cognition influenced how he understood mental illness in humans as well.

Even today our understanding of mental illness is filtered through our understanding of non-human animals. In his book ‘Anxious’ neuroscientist Joseph Le Doux criticised the Darwinian theory of emotions one of the primary exemplars of which is the late Jaap Panksepp. Panksepp described all mammals as sharing the same basic emotional systems: Seeking, Rage, Fear, Care, Play etc. Le Doux argues that the evidence that Panksepp provides to indicate that all mammals share the same emotional systems, is consistent with these non-human animals engaging in instinctive behaviours without necessarily experiencing anything like human emotions. Le Doux argues that while the evidence clearly indicates that threats do release innate behavioural responses and physiological patterns, but that attributing to animals reacting in these ways feelings of fear is an interpretation that goes well beyond the facts. Le Doux champions the view that fear in humans are psychological constructions that heavily depend on our linguistic abilities. Le Doux describes the difference between him and Panksepp as follows:

 “The main difference between my view and Panksepp’s is therefore, whether subcortical systems are directly responsible for primitive emotional feelings or instead are responsible for nonconscious factors that are integrated with other information in cortical areas to give rise to conscious feelings. What Panksepp calls cognitive feelings, are I maintain, what feelings are. The subcortical states are, as he also says at times, “truly unconscious” and thus not feelings at all. They are, in my view, nonconscious motivational states.” (Anxious p. 150)

“The evolutionary function of this ancient capacity is not to generate emotions like fear or anxiety, but simply to help ensure that the organism’s life continues beyond the present…Survival circuits do not exist to make emotions (feelings). They instead manage interactions with the environment as part of the daily quest to survive” (ibid pp 43-44)

Le Doux argues that medication that works on rats to help them overcome things like freezing when threatened may have some behavioural effects on human behaviour but they may not touch on the actual emotional feelings which are heavily tied up into linguistic and cortical systems. Obviously I will not here be able to decide between whether Panksepp, or Le Doux are right on this complex issue (I have discussed the topic elsewhere: https://www.academia.edu/29575444/Le_Doux_Anxiety_and_Neuropsychoanalysis ). My primary reason for bringing it up is because it further illustrates how our interpretation of the behaviour of non-human animals will influence our understanding of topics like mental illness. At this moment in time we have competing interpretations of animals which differ on the degree to which we are justified in attributing consciousness to them, or the degree to which we are justified in attributing propositional attitude explanations of their behaviour etc. While heuristic advice such as Dennett’s (basically an updated version of Occam’s Razor applied to animal ethnology) is useful it can only take us so far. To decide the issue we need detailed experiments. Hypotheses such as Bateson’s are useful in that they propose models which can then be tested using rigorous experiments. But this experimental probing is in its infancy so progress is likely to be slow. Below is a brilliant example of different theoretical models of cognition (Jerry Fodor vs Donald Davidson) leading to some great experimental work https://www.youtube.com/watch?v=1qdG1HHnmhA . One can only hope that whether right or wrong Bateson’s hypothesis inspires future experimental work.