Cunning canines…

Long scorned by psychologists in favour of the chimpanzee or rat, the dog has come into the spotlight in recent years. This is because dogs are unusually skilled at reading human social and communicative behaviour: the product of a long history of cohabitation.

Rico the Border collie knows all his toys by name. He has a remarkably extensive vocabulary and is able to establish a link between a word and an object with astounding speed and recall. © Courtesy Julia Fischer
Rico the Border collie knows all his toys by name. He has a remarkably extensive vocabulary and is able to establish a link between a word and an object with astounding speed and recall. © Courtesy Julia Fischer

"The only thing he can't do is speak!" Is there anyone who hasn't heard a poodle or German Shepherd owner boast how very clever their companion is? Without going as far as to endow them with human intelligence, nobody would deny that dogs have a special relationship with humans.

Despite this, researchers have long preferred to work with great apes, which are philogenetically more closely related to humans, or rats, which are easier to breed in the laboratory. Early this century, though, a number of research results reawakened scientists' interest in canines, which have proven to be astonishingly adept at deciphering the codes of human social and communicative behaviour (much better than primates, in fact). "For psychologists, dogs may be the new chimpanzees," announced American, Paul Bloom, in 2004 (1).

Rico, 200 words

This year, a research team from the Max Planck Institute for Evolutionary Anthropology in Leipzig (DE), headed by Julia Fischer (2), went to study Rico, a border collie that its owners claim "understands more than 200 words." These are names of toys or small objects that the dog fetches on demand, for which he receives a reward. Rico began his apprenticeship at the age of 10 months. First the researchers checked his abilities using controlled experiments in which his mistress was not allowed to give him any clues - even unintentionally. The border collie never once chose the wrong object. To test his skills further, the scientists then placed an "unknown" toy among a few familiar toys. When Rico's mistress called out its name, which Rico had never heard before, the dog immediately guessed that the word denoted the unknown object and brought it back. Four weeks later he could still remember the new name.

Perhaps Rico is an exceptional dog? Undoubtedly he is, judging by the extent of his vocabulary, which Max Planck Institute researchers deem to be "comparable with that of trained great apes, dolphins, parrots or sea lions." Without going so far as to equate the dog's performance to language learning by human babies, they believe that Rico is able to establish a link between a word and an object. Paul Bloom is more sceptical, pointing out that Rico ‘learns' only in a play situation and then only the names of objects he is able to fetch. In his view, Rico the border collie does not assimilate the name of categories of objects but simply associates the word with the act of fetching. Another serious limitation is that the test works only with the dog's mistress. Nonetheless, the speed with which Rico learns vocabulary shows that, like rats and chimpanzees, dogs are capable of inference, a process of logic that leads to a conclusion.

Power of signs

At the same time, more general experiments were conducted on untrained animals with no special emotional ties with experi-menters. It soon became apparent that it was not just a question of one exceptional individual but that the entire species had the same abilities. All the studies were identical in nature: first of all, researchers would set out several identical boxes after hiding food or another attractive object in one of them. They would then let the animal into the room and use a variety of cues to indicate the correct box, such as pointing with a finger, gazing at the box, nodding their heads or placing a coloured cube on top of the box. In other words, by using cues typical of non-verbal human communication. By the age of 14 months, puppies had no difficulty in understanding this type of cue. By contrast, chimpanzees failed dismally. It took them dozens of attempts before they learned to use the information given by the experimenter and proved unable to generalise these skills when novel cues were given that closely resembled the one previously learned (for example, if experimenters turned their heads around instead of nodding). Dogs are able to resolve this problem with the greatest of ease from the first attempt.

Other experiments have shown that the canine species understands the fact that humans see with their eyes, and acts accordingly. For example, if a trainer throws a ball for a dog to fetch and then turns their back, the dog will bring the ball back around the trainer's body to drop it in front of them. A dog will choose to beg food from a human whose eyes are visible rather than from a nearby person who is blindfolded (something that chimpanzees do not do). Similarly, a dog will approach a forbidden object only when experimenters have their eyes closed or if there is a windowless wall between dog and experimenter.

These results are all the more surprising since dogs have real trouble in deciphering non-social cues and are barely able to fathom the physical world. By contrast, a chimpanzee will understand immediately that when it sees two boards, one lying flat on the ground and the other tilted up, it means that the food is hidden under the raised board. Dogs are totally incapable of this.

Wolves and foxes

Why is it that even though great apes have superior cognitive intelligence, dogs can outperform them when it comes to communicating with humans or with their conspecifics? German psychologist Michael Tomasello (3) recently published a research review on the subject and put forward an explanation. First he ruled out the training hypothesis, as it turned out that litter-reared puppies (with relatively little exposure to humans) had the same ability to understand cues as adult dogs that had been raised by humans. The fact that the dog's ancestors, wolves, are social pack hunters and need to read the social behaviour of their fellow hunters does not go far in explaining the dog's ability to read human social and communicative behaviour either. Even where wolves have been reared by humans, they are unable to understand human cues, although they do as well as, if not better than, dogs in non-social problem solving or memory tasks.

Michael Tomasello believes that the answer lies in the dog's specific evolutionary history. "This leaves the possibility that dogs' social skills evolved during the process of domestication; that is, during the tens of thousands of years that our two species have lived together," he writes. Even though it is impossible to date accurately, thousands of years ago humans began to domesticate the wolves that prowled around their settlements in search of scraps. Over time, humans gradually selected those wolves that were not afraid of or aggressive towards them. Was this behaviour selection sufficient to endow these animals with the skills for reading human social-communicative behaviour? Amazingly enough, it seems that it was. Tomasello cites as evidence long-term domestication experiments conducted on foxes in Siberia. Over a 40-year period, an experimental population was selectively bred according to a single criterion - whether they would fearlessly and non-aggressively approach a human. The second population was maintained as a control and had been bred randomly with respect to their behaviour towards humans. When fox kits from this domesticated population were compared with age-matched dog puppies on the basic fingerpointing and gaze-following tests, the untrained foxes were as skilled as the dogs in using the human social cues. By contrast, the domesticated foxes were no more skilled than their wild counterparts in performing non-social tasks. "Perhaps most surprisingly, research with domesticated foxes suggests that a dog's skills for reading human social-communicative behaviour might have initially evolved as an incidental by-product of selection for tame behaviour," adds the researcher.

Inevitably this raises another question: how is it that chimpanzees have failed to develop social and communicative abilities whereas humans have, despite being so closely related philogenetically? After all, the chimpanzee, and still more the bonobo, seems to have many of the non-social problem-solving skills. Also, they are capable of assessing what another individual can see, attributing intentions to others and can draw inferences from the goaldirected actions of experimenters and conspecifics. In short, chimpanzees have all the requisite cognitive skills for understanding human social-communicative behaviours. Michael Tomasello believes that the answer lies in the natural tendency of chimpanzees to compete. Experiments have shown that a chimpanzee will cooperate with a conspecific only when there is no possibility of being attacked (owing to a physical barrier between them) and when there is a prospect of reward. Otherwise, dominance relationships inhibit any form of cooperation. There is little use in developing sophisticated communication skills when individuals are unable to share the rewards of joint effort. Following this line of reasoning, Tomasello advances the hypothesis that an important first step in the evolution of modern human societies was a period of selfdomestication during which a human-like temperament was selected ("Individuals within a social group either killed or ostracised those who were over-aggressive or despotic"). Thus, like domestic dogs, this selection for tamer emotional reactivity may have put our hominid ancestors in a new adaptive space within which modern human-like forms of social interaction and communication could be selected. So, although dogs do not have the power of speech themselves, they may have helped us to understand how we humans have acquired it.

Patrick Philipon

  1. Paul Bloom, Can a dog learn a word? Science 304, 1605, 2004.
  2. Juliane Kaminski, Josep Call, Julia Fischer, Word learning in a domestic gog: evidence for «fast mapping», Science 304, 1682, 2004.
  3. Brian Hare & Michael Tomasello, Human-like socials skills in dogs?, Trends in cognitive sciences, 9(9), 439, 2005