‘Man-sheep-dog': inter-species social skills

Paul Keil and Greg Downey

Paul, the lead author, interviewed sheepdog trialer Damian Wilson about his interactions with his dog, a border collie named Yandarra Whiskey. Damian and Whiskey gave Paul a demonstration of the techniques used in sheepdog competitions as they together tried to move a mob of three sheep. In a competition in New South Wales, a trainer and dog have to move three sheep who have never been herded through a difficult obstacle course, and the trainer loses points if he (or, less frequently, she) breaks from a slow, measured pace walking the course. The rules mean that the dog itself must be trained until it anticipates the sheep’s reactions, and understands, on some level, what dog and trainer, together, are trying to accomplish. Although the trialer may give commands, the dog, too, is a kind of expert.

In the demonstration for Paul, Damian intentionally gave Whiskey a bad command, encouraging the dog to move in a way that was likely to cause the sheep to bolt out of control. After the sheep got loose, Damian described his interaction with Whiskey:

I made the dog come around this way [clockwise around the mob of three sheep]. He said, ‘They’re gonna get away.’ He didn’t want to come. He said, ‘I think it’s a bad call.’ And I argued with him, and I said, ‘No. Come!’ And he said, ‘Nah nah… I tell you, they’re gonna go.’ And then he started to come, and the sheep started to go, and then he went, ‘See, I told ya’…

Of course, at no time did Whiskey actually speak to Damian. And Damian’s signals were whistles, shouts, and gestures, much simpler than the elaborate interpretation that Damian offered in his post-interaction analysis.

Damian was explaining his perceptions of his dog’s thoughts as the two of them, together, interacted with three other animals, the sheep. Five individuals – three species – all attentive, probing, anticipating what the others were doing. In Damian and Whiskey’s brief demonstration of herding technique on their farm for Paul, Damian had hoped that the bad command would highlight Whiskey’s own instincts to herd, the dog’s perceptions of what the sheep were inclined to do, and the dog’s understanding of what he and Damian were trying to accomplish.

While Damian’s recollection was no doubt intentionally anthropomorphised, and possibly better elaborated than usual – giving the dog a voice – because of the sympathetic audience, the interaction that had occurred only minutes earlier was far more complex than a novice like Paul could recognise. Sure, Paul heard Damian give the command and witnessed the sheepdog’s momentary hesitation to follow, but he thought little about it; Damian’s description revealed a reciprocal exchange, a negotiation between human and dog based upon each one’s perceptions of the sheep and their spatial and emotional relations. Paul was simply not privy to a lot of the detail of their communication because he couldn’t see it. Whiskey was a far more sophisticated social agent then Paul could initially grasp.

The dog was not simply a tool, or merely obedient to a guiding human intelligence; on some level, Whiskey grasped what needed to be done, and Damian had come to count on the dog’s ability to herd, including the dog’s perception of how stressed and liable to flight the sheep were. The key to being an expert dog trialer, then, included the ability, not just to train a dog to herd, but to perceive the dog’s intentions and perceptions, and to anticipate the animal’s next move (as well as those of the sheep).

Paul did his Honours thesis, which Greg supervised, in 2010, and is currently preparing to depart for the field to do his PhD research, studying human-animal relations in Assam, India, with mahouts who use trained elephants (kunkis) to work and patrol rural areas where humans and wild elephants come into conflict. (Hopefully, we’ll get some more posts on that project in the future, so make sure to comment to encourage him! [Greg])

Dog experts and expert dogs

The ability to recognise the complex social agency of nonhuman animals may be a product of expertise, although some theorists have begun to argue that the capacity is an at-least-partially inherent legacy of our species’ evolution. A study recently published in PLOS ONE by Kujala, Kujala, Carlson and Hari (2012) reports that dog-experts, compared to non-experts, were significantly more likely to engage their posterior superior temporal sulcus (pSTS) when observing photos of dogs interacting with each other, according to fMRI images of subjects. In addition, experts had richer verbal reports and their visual behaviour, studied using eye tracking, included focusing on a wider range of cues, including bodily posture in the dogs.

Increased activity in the pSTS is generally associated with observing biological movement (Puce et al. 1998) or movement by computer animations, where the ‘protagonist’ images on screen move in a ‘self-propelled or interactive way’ (Schultz et al. 2005). The pSTS has also been found to increase activity when subjects observe goal-directed behaviour and when reasoning about intentional states (Pelphrey Morris and McCarthy 2004; Saxe et al. 2004). The pSTS is involved heavily with social cognition and social perception, ‘the initial stages of evaluating the social communicative intentions of others by analysis of eye-gaze direction, facial expressions, body movements, and other types of biological motion’ (Pelphrey et al. 2004: 1706).

Kajula and colleagues displayed photos of a pair of humans to subjects: in one set of photos, both people walked together and faced each other; in the other, the pair are walking away from each other with their backs turned. For both the control and dog-expert group, metabolic activity in the pSTS was significantly higher for the walking-toward scenario compared to the photo of people walking away. That is, judging from metabolic levels, the ‘walking toward’ scenario seems to imply that interaction is occurring, and observers make inferences about the interactions between the two people in the photo. (Note: It’s a still photo, so there’s no actual interaction occurring.)

With similar images showing dogs instead of humans ‘the results suggest that generally, the social interaction of dogs is processed in the regions that also process human social cues, in both experts and non-experts of dog behavior’ (Kujala et al. 2012: 5, emphasis added). Only the dog-experts, however, showed increased activation of the pSTS when viewing the ‘walking towards’ dog photo, while the control group showed no significant difference in activation. This seemed to support other findings that experts, in general, made more and more detailed mental inferences about dog behaviour than the control group.

(Of course, the fact that these were static photos likely means that active projection was featured more heavily than in normal interpretation, which might have occurred with video, and the perspective was ‘third person’ and non-participating in interaction, which may have also caused some distortion in the normal perceptual and cognitive ecology of human-dog interaction.)

Kajula and colleagues concluded, drawing on eye-tracking evidence as well, that increased activation in the pSTS was evidence of ‘improved visual reading of the dog’s body postures’ and communicative intentions’ (ibid.: 7; see also p. 8). Specifically, and we’ll come back to this,

expertise in dog behavior is not explained by differences in the mentalizing abilities between dog experts and control subjects. Instead, dog expertise seems to be more related to improved visual reading of the dogs’ body postures. This interpretation is supported by the eye-tracking data that showed that the experts, compared with the control group, gazed relatively more the dog bodies than dog heads. (ibid.)

Being an expert at interacting with dogs not only means a brain that’s better attuned to how dogs communicate; in fact, experts and non-experts, in most respects, are quite similar. Expertise means having behaviour patterns that include knowing where to search the animal’s body for information and greater tendency to ‘mentalise’ or impute motives to the animals (whether those projections are accurate is a separate question).

More than just an innate connection with animals

In his essay ‘Why look at animals,’ John Berger (1972) explores humanity’s historical entanglement with animals. He argues that, only in the last few hundred years, the advent of modern urban living has disconnected large groups of people from animals, whereas previously, everyone would have had much more intimate relations with nonhuman animals, even those living in cities. For example, there were so many horses in Victorian London that some commentators of the time feared the city would be buried by the manure before the turn of the 20th Century.

For Berger, humanity’s long fascination for animals still holds for urban dwellers; we see it in the extravagant existence of zoos and high rates of keeping otherwise useless companion animals, but the link is much attenuated. Similarly, the ‘Biophilia Hypothesis,’ a term coined Erich Fromm but popularised by E. O. Wilson, holds that humans have an innate drive to associate with ‘nature’ and living things, which some theorists consider to be a result of our long evolutionary relationship with the ‘natural’ world.

In the 21st century, the majority of the world’s humans have now moved to cities, populated only by a limited variety of nonhuman organisms, most of which we choose to ignore or try to exterminate. At the same time, globally, we face the possibility that we are witnessing the sixth great mass extinction event in the history of life on our planet. For better or worse – for both humans and other animals – we are deeply entangled together, not just in ecological relations, but also in complex cognitive interaction (including ignoring the implications for other species of some of our actions).

A few papers in recent years have focused on alleged ‘hard-wired’ or innate aspects of our cognition, neural mechanisms that make us more responsive to nonhuman animals in the world. Changes in the state of animals – human and nonhuman – were detected by subjects faster than state changes in non-biological objects, for example. Evolutionary psychologists New, Cosmides and Tooby (2007) argued that these responses are primed, not a result of expertise or learning, but because of an inherent and evolutionary-beneficial ‘animacy detecting mechanism.’ Morman and colleagues (2011) reported that the right amygdala responded exclusively to images of nonhuman animals, regardless of species or emotional valence, significantly more than images of human faces, objects or landmarks. The authors believe this result ‘argues in favor of a domain-specific mechanism for processing this biologically important class of stimuli’ (ibid.: 1248).

Pat Shipman (2011b), reporting on these two studies, points out that, ‘By being predators-without-equipment, our ancestors came under evolutionary pressure to pay close attention both to the real carnivores, who competed for prey and wouldn’t balk at eating our ancestors, and their own potential prey. Accumulating visual information about the behavior of both sorts of animals became key to survival’ (Shipman’s post on her blog, The Animal Connection, is here). In her 2011 book, The Animal Connection, Shipman argues that our ability to recognise communicative intentions in animals and form relationships to exploit their species-specific skills (the proto-typical example being a dog) that allowed humans to gain evolutionary advantages in a number of domains, including hunting (2011a).

Elsewhere, second author Downey (2010) has argued that interacting with animals also means using their senses as surrogates, noticing when other prey animals flee before a more dangerous predator, realising when animal behaviour suggests coming changes in weather, recognising that one’s own sensory hallmarks need to be hidden from the animals we hunt. All predators learn to hunt, even without great cognitive endowments, but humans learned to hunt many different kinds of animals, including how to study new prey and learn their patterns as we moved into new ecological niches.

Whether or not these theories are accurate, whether we have an inherent affection for other animals or an innate tendency to pay attention to animacy, the fact is that some of us are really terrible at actual anticipating what animals will do. Ask the TV anchor who, while reporting a feel-good story about a rescued dog, got her lip nearly bitten off when she thought that the dog was having a similar ‘feel-good’ moment with her reporting (Rescued dog bites NBC anchor in the face during feel-good segment gone wrong. ‘Feelin’ good?’ ‘Nup, feelin’ bitey!’). Or this guy, who thought an elephant was cute and chilled out until the animal swatted him away after he got a little too close to the elephant’s food.

Unfortunately for the victims, the ‘instinctual attraction’ to animals does not guarantee mutual affection, nor does it mean that we’re necessarily very good at reading their emotional or intentional states. The idea that we have an ‘evolved capacity’ to understand animals can lead us to ignore several of the important implications of the paper by Kujala and colleagues:

1). Whether or not there’s an ‘evolved capacity,’ experts are more likely to engage the brain more broadly and to perform the kinds of behaviour that likely need to happen before the ‘capacity’ becomes realised ability. However ‘instinctual’ our biophilia, as humans, our distinctive cognitive architecture virtually always needs training up to reach the perceptual potential of highly skilled individuals.

2). The fact that so many of us no longer have close relationships with animals probably means that this cognitive capacity, in more and more individuals, is not being fully realised, as experience does not allow nascent capacities to be fully recruited and the neurological architecture for animal perception generated. Just ask that news anchor about her confidence in her ‘innate’ ability to read animals’ behavioural or intentional states. If anything, underlying biophilia might even be dangerous if it is coupled with perceptual ignorance or blunted skill in interacting (that is, being attracted to something we aren’t trained to handle).

3). If, as some of the data suggests, the neurological and cognitive resources used to perceive other humans’ and animals’ intentions overlap in significant ways, the overlap might lead to predictable errors of anthropomorphising.

4). In addition, if the areas of the brain involved in mentalizing about animals overlap significantly with that used to accomplish the same tasks as in humans, and if we routinely make anthropomorphising errors, perhaps the ‘human-animal connection’ is not a domain specific capacity, but rather an extension of our social capacity into other domains, such as interacting with other species.

The overlap in the neural mechanisms employed to interpret human and nonhuman actions points to the complexity of biological behaviour in general and the insight that our relationships with animals are, as anthropologist Tim Ingold (2000: 72) has pointed out, fundamentally social. Even the subjects in the expertise experiments who had little first-hand experience with dogs had the same areas activated by watching humans interacting become more engaged by the sight of dogs engaging.

Yet our brains are not hard-wired to immediately perceive the variation of nonhuman communicative gestures or to interpret their signals accurately. The ability to ‘read’ the behaviour of nonhuman animals requires repeated interaction and focused attention on behaviours and parts of the dogs’ bodies that reveal intent and response within social interaction. And it helps to have an experienced coach.

‘Man-sheep-dog’ interacting

The goal of sheepdog trialing, the subject of Paul’s extensive Honours thesis, is to encourage a mob of sheep through a number of obstacles distributed throughout a course, by commanding the dog from a distance. While all trialers are no doubt dog experts, the best trialers he met were those who paid constant attention to the sheep – not their dogs – visually monitoring the sheep’s behaviour closely and their unfolding interaction with the sheepdog. Trialers spent a great deal of their time trying to ‘to work out what the sheep is going to do before he does it.’

When on the obstacle course during a trial, the dog almost always moves opposite to the handler, across the herd, so that the sheep are ‘balanced’ between the two presences. By putting ‘pressure’ on the sheep, the dog can move the sheep into a more open space, away from both the human trialer and the dog him- or herself. Trialers refer to this arrangement as ‘man-sheep-dog’ because this is the general arrangement of the three species in space.

Ironically, the handler’s interaction with the dog, even though it is intense and fine-tuned because they train together, usually takes a perceptual back-seat to studying the sheep, as both sheep-moving experts (human and dog) focus their attention on their shared task. Handlers’ relationships with their dogs were more intimate, based upon long study of their animals and carefully screening and training the dogs to have predictable responses and sheep-herding instincts. In some ways, the dogs become better co-workers than other humans, as they do things humans are incapable of doing – such as indefatigably, almost obsessively focusing on the sheep and responding at lightning speed ­– and complement their human handler’s own cognitive and physical abilities (Keil, under review).

If we were to sample their foveal fixations using eye tracking during a sheepdog trial, we would likely find that the trialers focus minimally on their dogs, unless something goes wrong, and the trialer has to find where the dog has gone. The more intense and seamless the interaction, the more the trialer can read the position of the dog from the beahviour of the sheep. In this case, expertise in interaction would correlate with decreasing direct sensory perception.

In addition, rather than perceiving sheep as simple creatures motivated by force or threat, handlers treat them as adversaries to be outsmarted with the aid of their dogs’ athletic and perceptual abilities. Handlers spend their time trying to predict what the sheep will do next and how the sheep’s behavioural state and intentions are emerging throughout the course of the trial. Handlers take on the sheep’s frame of reference and assess where the animals are looking by observing where their heads are pointing or whether their ears are moving in a particular way (which indicates where the sheep is looking within its peripheral vision).

By studying the sheep’s vision, the handlers also grasp the sheep’s intentional states, and, indirectly, where their dogs, who perceive the same animals through their own cognitive capacities, are also likely to be. Why are the sheep looking around and what do they intends to do? How nervous or calm are the sheep (perceivable through their body posture)? Is the ‘big horny fella,’ a dominant whether (castrated male), likely to cause trouble or lead the other sheep in an attempt to bolt? Are the sheep growing in ‘confidence’ or coming to ‘believe’ that the dog does not have enough force to combat them? If sheep raise their heads too high and are looking in a particular direction, for example, they are likely getting anxious and are going to try and escape in the direction their heads are pointing.  A handler may have to instruct his dog to back off or hunker down until the sheep calm down.

While discussing strategy with one trialer, the trialer told Paul, ‘I try to fool them [the sheep] into thinking that my dog’s a bit weak.’ The man paused for a second, and then began to laugh, ‘I could be fooling myself though!’ If so, it’s a well-adapted sort of self-fooling.

Trialers’ perceptions may, in fact, be elaborate self-deception, as they are an attribution of cognitive states to animals that may be incapable of having these sophisticated states. But the ‘inaccurate’ attributions serve an important purpose: helping the human to anticipate potential action and scan a few steps ahead at possible scenarios. That is, anthropomorophism may not be neurologically or biologically accurate, but it may be cognitively and practically useful, helping handlers to projectively scenario-build as they interact with their sheep and dogs (who are also engaged in their own cognitive forms of anticipation and negotiation with each other). Part of ‘thinking like a dog’ or a sheep may be inaccurately assuming that dogs and sheep are ‘thinking like an (admittedly odd or not terribly bright) four-legged human.’

Theory of (animals’) mind

Trialers would be the first to admit that the minds of sheep and dog are not the same as that of humans. They believe anthropomorphising the animals is a grave mistake. One criticism veteran trainers, often retired sheep farmers or handlers with long histories working with dogs, offer of urban dog owners and sheep trialing hobbyists is that novice dog handlers overly pamper their animals, failing to recognise the yawning gulf between dog and human. Some are mortified, for example, that hobbyists allow the dogs that they are trying to train for three-sheep trials into their homes, or don’t keep them tied up when they are not working sheep. Most veteran trialers believe that the boundary between dogs and humans is wide and should not be blurred, or the dogs’ performance will suffer.

A dog cannot be trained properly, however, if you are interpreting, judging and responding to their actions in ways that make it impossible to cooperate. In order to properly train a dog, you have to grasp how the dog is perceiving you in ways that help you to give better commands, even if it’s a misattribution of cognitive states to the animal. A good handler, for example, disciplines his or her communication and bearing so that it is unequivocal and easily read by the animal, which requires the human to control his or her own emotional state and unintentional, nonverbal communication through channels like posture and the volume of the voice.

The same holds true for human-horse relations. A successful partnership requires the human handler to recognise the way that his or her emotions, postures, and actions are read by the horse. Likewise, the human partner has to learn to ‘see’ or ‘think’ like a horse, to notice things in the environment that may spook the animal or cause the horse to startle; a plastic bag caught in a fence that might move and appear animated, an unusual surface to walk over, or a feral peacock that likes to wander down to steal horse feed (yes, we’re speaking from experience…).

Still, even sheepdog trialers who are acutely aware of their dogs’ limits attribute internal mental states and dispositions such as ‘confidence,’ ‘beliefs,’ and ‘thoughts,’ not hesitating to project human-like cognitive events to their animals. If humans recruit the same socio-cognitive, neurological mechanisms they use with fellow humans to engage in the same kinds of interactions with other organisms, the overlap between folk psychological language for animals and humans should not be surprising. That is, if we’re using the same equipment to perceive a dog’s or sheep’s intentions as that we use to figure out what each other are thinking, anthropomorphizing is likely, even in old-school dog trialers who have strictly instrumental relations with their dogs and sheep, much moreso than most people with their companion animals.

Anthropomorphic ‘overshoot’: a built-in error?

Of course, if folk psychological language suggests we’re using the same cognitive or neurological mechanisms we use for reading humans, we would expect a degree of anthropomorphic ‘overshoot.’ Cognitive overlap in the ability to perceive animal intentionality would likely lead us to over-estimate the nonhuman animal’s capacities, over-anthropomorphising their cognition. The paper by Kajula and colleagues suggests that, if the projecting of intentionality to animals is innate, it has been extinguished in adults who don’t have expertise with dogs, if higher levels of activity in the pSTS are an accurate indication (experiments with children would better illuminate whether the tendency to project exists early in developmental processes).

For example, the tendency to project can arise with animals that have even more alien minds than dogs, highlight the degree of anthropomorphic overshoot. Take for instance entomologist Jeff Lockwood, who worked in the lab for long periods of time with Gryllacrididae, a fierce cricket-like insect who builds nests. The insects have the ability to recognise individuals and their own scents, and display unusual aggressive responses to strangers. Lockwood began to develop the notion that the insect had a sense of self, he describes in an interview. That was, until he witnessed the same creature eat its own entrails after the accidental splitting of its abdomen (the interview on Radiolab, ‘Killer Empathy,’ is here). Perhaps recruiting those socio-cognitive skills we use for understanding each other to anticipate how animals will act can bring along with it a bunch of anthropo-baggage that is not simply linguistic.

Still, even if there is some anthropomorphising going on, the ability of these trialers to manage sheep and form coordinated working teams with dogs, would seem to imply that conceptualising animals as agents is not a wrong-headed projection, or the activation of the pSTS to interpret animal behaviour is a evolutionary bias. Engaging animals socially, on a continuum with humans, is a successful approach that may prevent (some) humans from being bit, or hit with trunks.

The fact that we will also do it with animated geometric objects in computer-based psychology experiments suggests that we’re likely to project animacy to even novel stimuli. The history of anthropomorphising across a wide range of cultures – for example, in religious animacy or the belief that objects have ‘souls’ – even to Gryllacrididae by entomologists who should know better, suggested that the strategy is not tightly restricted to animals like us or finely-tuned to dogs and large predatory animals.

While our cognitive faculties and large social brains are no doubt a result of our complex primate social systems, we can also think about how our socio-cognitive hardware was more broadly applied and developed in our deeply entangled relationships with a variety of other species. Pat Shipman has rightly pointed out that the ability to have ‘theory of mind’ is essential to working and negotiating with animals throughout evolutionary history (for more on Shipman’s ideas, see this earlier post on Neuroanthropology on the ‘dog-human connection’). To be able to hunt animals and domesticate them, our ancestors must have learned to perceive what they are thinking, or at least to predict their responses, even if this was based on anthropomorphic overshoot. The development of our exceptional socio-cognitive skills, and the plasticity and inclination to extend those skills promiscuously, might help explain why people are so good at living with, working with and understanding such a large variety of animals, more than any other species. But it’s not just humans who seem to do this…

How do dogs do it?

Sheepdog trials and that ‘man-sheep-dog’ configuration remind us  that humans do not face other animals alone. Let’s not forget the dogs. As Shipman has pointed out, dogs are special. When our ancestors and the ancestors of dogs came into contact, two species with extraordinary social skills, and surprisingly similar pack hunting strategies, began a long relationship that arguably shaped the evolution of both over the next thousands of years.

The dog’s evolutionary cousin, the wolf, is able to form complex, coordinated social groups in order to hunt and track down prey. In the last fifteen years, a number of excellent papers have highlighted dogs’ socio-cognitive skills and ability to read human communicative gestures, including following eye gaze, pointing, and mimicry of certain actions, amongst others (anything by Miklosi from 1999 onwards). This versatile social intelligence is often argued to be a result of canine evolution being closely coupled with that of humans, breeding alongside us for generations, but it is just as likely to be an extension of the dog’s own intra-species skills. Dogs may be ‘overshooting’ their attribution of dog-like responses when they interact with us.

And we should not forget that dogs are also very good at building social relationships with a wide variety of other species. For example, a while back, a colleague shared this amazing footage of polar bears and dogs playing (Polar Bears and Dogs Playing (YouTube)).

Or see Tarra and Bella: A Match Made in Ele-Heaven (YouTube) or A dog and a donkey (YouTube) for other examples of the ways that dogs, descended from already-social wolves but bred for increased social skills and blunted aggressive traits, can spontaneously take up relationships with other animals, domestic, tame or wild.

In fact, the very tendency to herd, including to protect a wide variety of animals, is a result of the dog’s ability to extend its social relations to other species, not just humans. In Australia, we even have the example of Maremma dogs being used to protect chickens and a penguin colony (see Guardian dogs look after penguins), showing that they can extend social relations even to animals very un-like themselves.

The tendency of border collies to collect up sheep and ‘balance’ them against a human trialer, ironically, is probably a remnant hunting instinct. Herding, then, paradoxically makes use of both hunting instincts and social solidarity, a mixture of predatory and cooperative processes, just like humans’ own relationships with so many species.

While we might focus on humanity’s skills, dogs as well must have a pretty interesting socio-cognitive architecture. Do dogs bring their own canine-morphistic tendencies to relationships with humans and other animals? Do dogs project their own intentional states onto humans, sheep, donkeys, elephants, penguins? When my border collie, Louis, sits staring at his ball, does he assume that I share his fascination with it? And when our lab, Roxy, rolls in horse manure so that she can share with us what she’s found, does she assume that we like the smell as much as she does?

Conclusion: Training as a skill

Not all people who work alongside dogs are able to read them. The sheepdog trialers Paul interviewed were quick to criticise those rural property owners who would put down a trained working animal, who they had bought, because of the ‘dog’s’ failure to do what the human commanded. Trialers asserted that the owners had failed to learn how to communicate or ‘bond’ with the dog. They pointed out that dog-human communication required both to be trained, to come to an understanding of each other’s capacities. That team could break down from either end.

Donna Haraway (2008) argues that, in our interspecies relationships, we are trained into each other. Certainly, individual dog-human, elephant-human, and horse-human partnerships are the result of reciprocating interactions, observations and experience over time. But the ability to recognise animal states is not just an idiosyncratic skill arising from that specific relationships; social understanding of other species can be cultural transmitted, taught and learned.

Relationships with animals are structured, and the capacity to recognise animal agency is a skill that is usually learned, in what Lave and Wenger (1991) call a ‘community of practice.’ We don’t have to face an undomesticated dog or a wild elephant and invent wholecloth a way to interact with it; we will usually join a social field where animals and humans are already interacting. Complicated relationships actually precede, for each team that eventually becomes skilful, simple one-on-one (or one-on-three-on-one in the case of man-sheep-dog) relationships. Working pairs emerge out of complex social interactions between novice and experienced humans, and novice and experienced animals, rather than the other way around.

Anthropologist Tim Ingold argues that aspects of the world are revealed to the novice by a more experienced teacher who directs and trains the junior partner’s attention; so too is an animal’s agency revealed to a human. A large number of human groups still live in intimate communities with animals, whether they be Mongolian horse or Saami reindeer herders, sheepdog enthusiasts or mahouts who work with elephants, primate ethologists or lab technicians studying rats. Members of these communities are trained to engage with, observe and interpret behaviour, and respond appropriately according to their group’s cultural norms. Whether their ‘readings’ of the animals’ states are accurate is less important than the fact that both species communicate in predictable, useful ways so that they can work together, anticipate each others’ actions, and live in stable inter-species communities.

Whether we might see variation in brain activation for how these people approach animals would be a different story. Would a behaviourist reveal decreased pSTS activation when interpreting animal behaviour? Do individuals who anthropomorphise their domestic dog to a high degree but have less expertise in reading body postures, engage different cognitive functions than a veteran sheepdog trialer who has fewer sentimental attachments to animals? As Kujala and colleagues show, expertise matters, but we suspect that the type of expertise matters, too.




Berger, John. 1972. Why Look at Animals. About Looking. New York: Pantheon Books.

Haraway, Donna. 2008. When Species Meet. University of Minnesota Press.

Ingold, Tim 2000. From Trust to Domination. An alternative history of human animal relations. The Perception of the Environment. Essays of Livelihood, Dwelling and Skill. London: Routledge.

Keil, Paul. Under review. Interspecies Cognitive Scaffolding. Human-dog distributed cognitive systems at competitive sheepdog trials.

Kujala MV, Kujala J, Carlson S, & Hari R (2012). Dog experts’ brains distinguish socially relevant body postures similarly in dogs and humans. PloS one, 7 (6) PMID: 22720054

Mormann F, Dubois J, Kornblith S, Milosavljevic M, Cerf M, Ison M, Tsuchiya N, Kraskov A, Quiroga RQ, Adolphs R, Fried I, & Koch C (2011). A category-specific response to animals in the right human amygdala. Nature neuroscience, 14 (10), 1247-9 PMID: 21874014

Lave, Jean, and Etienne Wenger. 1991. Situated Practice: Legitimate Peripheral Participation.Cambridge: Cambridge University Press

New J, Cosmides L, & Tooby J (2007). Category-specific attention for animals reflects ancestral priorities, not expertise. Proceedings of the National Academy of Sciences of the United States of America, 104 (42), 16598-603 PMID: 17909181

Pelphrey KA, Morris JP, & McCarthy G (2004). Grasping the intentions of others: the perceived intentionality of an action influences activity in the superior temporal sulcus during social perception. Journal of cognitive neuroscience, 16 (10), 1706-16 PMID: 15701223

Puce A, Allison T, Bentin S, Gore JC, & McCarthy G (1998). Temporal cortex activation in humans viewing eye and mouth movements. The Journal of neuroscience : the official journal of the Society for Neuroscience, 18 (6), 2188-99 PMID: 9482803

Saxe R, Xiao DK, Kovacs G, Perrett DI, & Kanwisher N (2004). A region of right posterior superior temporal sulcus responds to observed intentional actions. Neuropsychologia, 42 (11), 1435-46 PMID: 15246282

Schultz J, Friston KJ, O’Doherty J, Wolpert DM, & Frith CD (2005). Activation in posterior superior temporal sulcus parallels parameter inducing the percept of animacy. Neuron, 45 (4), 625-35 PMID: 15721247

Shipman, Pat. 2011a. The Animal Connection. W. W. Norton & Co.

_____. 2011b. Hard- Wired for Animals: Have humans got animals on the brain? The Animal Connection weblog at Psychology Today.com.  (15 September 2011) http://my.psychologytoday.com/blog/the-animal-connection/201109/hard-wired-animals. Accessed on 19 June 2012.

First author
Paul Keil is interested in the strange social couplings between humans and other animals and how human bodies and cognition are shaped by the process, particularly within interspecies working relationships. He is also interested in strange human-human couplings as well, although mainly of a distributed cognitive nature. Paul is a PhD candidate in Anthropology at Macquarie University, supervised by Greg Downey, conducting his research in North East India on mahout-elephant working units mediating human-elephant conflict. If you like his work, make sure to post a comment so we can talk him into doing this again!

Related Posts Plugin for WordPress, Blogger...
This entry was posted in Evolution, Learning, Skill and tagged , , , . Bookmark the permalink.

7 Responses to ‘Man-sheep-dog': inter-species social skills

  1. Daisy says:

    Superb article! This reminds me of Latour’s Actor Network Theory which can present a particularly useful model for anthropologists researching human and non human relations; it recognizes the transient and flexible nature of relationships between all actants in a network and it accords nonhuman animals agency. This is very useful in reconsidering the primacy of human agency and imbues a reflexive awareness of how the intersubjective nature of any social network influences the outcomes of those interactions.

  2. Pat Shipman says:

    GREAT post! I’m going to have to read it about eight times to get a grip on all of the insights and observations, not because it is badly written but because it is rich.

    • gregdowney says:

      Thanks very much, Pat –
      Coming from you, that’s high praise. Paul’s project was fantastic, awarded best Honours thesis in anthropology in Australia for the year. I’m encouraging him to seek publication for parts of it.

  3. daniel.lende says:


    I really enjoyed reading this piece! As Pat says, very rich! I do hope you post more in the future. And don’t think you have to get into the Greg “I do mega posts” Downey scenario! Smaller pieces might be more digestible and fun. I loved the polar bears playing with dogs video, and thought that had some interesting implications on how other species can “understand” and interact with each other.

    So a few things – first, definitely try to publish this! Posts are great for getting information out there, but this sort of work also needs to become a part of the peer-reviewed scholarly record. I’ve made this mistake – writing up stuff and then not doing all the crap work to get a publication. Getting that peer-reviewed acknowledgment matters for an academic career, and is also another way to get your ideas out in front of a specific reading public.

    And now for a couple idea things. First, I’m really intrigued by your and Greg’s points about “theory of mind.” You write at one point, “anthropomorophism may not be neurologically or biologically accurate, but it may be cognitively and practically useful.” And near the end, “Whether their ‘readings’ of the animals’ states are accurate is less important than the fact that both species communicate in predictable, useful ways.”

    I think this way of thinking of theory of mind – less mental, more about being able to make useful, predictable anticipations and interactions with others – opens up new vistas. It’s reminiscent of something I was just reading, Roy D’Andrade’s (1987) “A Folk Model of the Mind,” where being scientifically accurate is not as important as being able to reason about, anticipate, and interact around other people’s intentions and feelings and motivations. It would be really interesting to develop this approach more to “theory of mind”; certainly I’d like to hear more about it!

    The other thing that I’d like to see even more of is your point that successful interaction relies on bringing together different behavioral skills, for example, predatory and social interaction approaches. The playing of the dogs and polar bears had this sense of things – play where the biting/nipping and the grasping by the bear are more predatory, but done in a way that is about social interaction.

    My broader point is that this mix-and-match approach to how different neural systems/behavioral dispositions can be brought together in specific situations is a key way to do some neuroanthropological analysis. And I think your area of work is a great place to bring that out.

    To take my work, I’m struck by the thought that addiction is actually a problem where this ability – the mix-and-match, the successful integration of different brain systems in specific social and cultural environments – falls apart. Rather, specific systems, like behavioral engagement, are revved up, whereas social cognition and behavioral oversight become less a part of the types of actions and interactions done by people with serious drug use problems.

    So I’d like to see you develop that set of ideas even more.

    Looking forward to it,

    • Paul Keil says:

      Hi Daniel,

      Thank you for the comments and the opportunity to post on Neuroanth! This post, while partially drawn from my honours, was also developed out of some hunches I’ve had sitting in the back of mind since I completed the thesis. The Kajula paper, as Greg pointed out, gave a perfect opportunity to play and express them.

      I agree that applying the concept of Theory Of Mind towards human animal interactions can give new insights into what exactly we are doing when socializing or perceiving intentional actions of others. It is an avenue I really hope to explore during my PhD research. Watching people train with animals is fascinating because not only does it give an excellent opportunity to watch as a “theory” develops, but also that the slow realization of the animal agent is one that is strictly interpreted through non-verbal cues. Of course the examples we gave of Theory of (animal) mind in action were lucidly expressed in linguistic form by Damian, but there are also other sheepdog trialists who cannot express themselves in this manner, attributing a gut feeling or an intuitive ‘just knowing’ that enables them to accurately respond to what the animal’s behavioral state is, or what the nonhuman-animal is going to do. That cognizing about complex social behavior can operate on an embodied, immediate level and does not need to appeal to interpretation via internal states is interesting to me as well.

      The mix of predatory and cooperative processes was Greg’s insight, and like you I find the idea intriguing. It’s definitely something I need to think about further, particularly how the domestic context / human niche might play a factor in that. Personally, videos like the dog-polar bear blow me away because you see not only the reciprocating responses to social cues (a shared mammalian socio-cognitive architecture? I dont know), but also because those inter-actions of each individual are adjusted, softened or strengthened depending on size and strength of the partner who is being played with. This moderation of power when interacting with different beings (human and elephant are good examples) are learnt by both partners during the process of the relationship. Baby captive elephants apparently are notoriously rough!

      Again, thanks for the comments and advice Daniel, and I do hope to contribute more to Neuroanthropology in the future!



  4. Pingback: The Scienceblogging Weekly (June 29th, 2012) | Prutic Networks

  5. Pingback: thoughts on thoughts » Blog Archive » Those doggy-people

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>