Bering in Mind

Bering in Mind

A research psychologist's curious look at human behavior

Cur cognition: Do stray dogs have qualitatively different kinds of canine minds?


In previous posts, I’ve discussed my fascination with dogs, such as this recent controversial piece mentioning those good-natured pit bulls whose unearned reputations often precede them because of a few maladjusted, vicious outliers. Yet I’ve never seen anything quite like the canines of Sofia, Bulgaria, from where I’ve just returned after a week of teaching at a cognitive science summer school and from listening to a surfeit of long-forgotten, uplifting '80s pop music, which the weary and unshaven Bulgarian taxi drivers seem to adore to no end. Some recent work by University of Florida psychologist Monique Udell and her colleagues suggests that it’s not just my imagination that stray dogs are special—rather, strays in general may be vastly more different from our pet dogs than we assumed, particularly in their social cognitive functioning.

Now, the stray dog situation in Sofia is notoriously problematic. You know you’ve got a problem when a pack of strays breaks into the deer exhibit at your local zoo and “ruthlessly dismembers” almost the entire collection, as happened earlier this year. And given the general sentiment that an organized roundup and euthanasia is out of the question for moral reasons it’s also a very prickly issue among the people living there. (Stray cats are a problem too, but they appear to be kept in check by a lot of very hungry dogs.) Stray dogs are about common as squirrels there, and from what I saw, most of Sofia’s human population also has about as much interest in strays as Americans in the suburbs have in the squirrels living in their backyards. A June, 2010 estimate placed the figures at around 9,500 dogs running loose in the Sofia confines. So when you’ve got that many animals—even man’s best friend—in a relatively small, concentrated city (not to mention one with its own human homeless problems in the form of the ostracized Romanian Roma, or “gypsies”), mass desensitization is sadly inevitable. But this human-dog indifference is even more striking because it appears to be mutual.

I watched incredulously as the Sofia strays ambled casually down the sidewalks like proper Bulgarian citizens. They stepped aside politely for human pedestrians before continuing on their way, stopped patiently to look both ways before loping across frantically organized, crowded freeways, mingled with one another at storefronts and had their mangy coats tousled by the whooshing tires of passing commercial trucks while in the midday heat they slept quietly in tree-shaded gutters mere inches from the road. Most of these animals are multigenerational strays, which means that they are the offspring of strays who were the offspring of strays and so on, and on, for many breeding generations. Natural selection must work quickly indeed under such conditions: these are the descendents of the craftiest ancestral dogs of yesteryear Sofia, those who survived puppyhood without being crushed by some juggernaut and who managed to live long enough to pass on their wily natures to their offspring. Too much reliance on humans or interest in human behavior may well be maladaptive to these dogs’ overall genetic interests within this selective context, given the situation. I love my dogs, Gulliver and Uma, and they’re pretty smart as far as dogs go. But they wouldn’t last two minutes on the streets of Sofia.

The Sofia strays may be mongrels, and indeed many looked to me more like dingoes or coyotes than they did domestic dogs, but technically they’re every bit as much members of Canis lupus familiaris as your own pampered Lhasa apso or Chocolate Labrador. But perhaps genes alone do not a dog make. In an intriguing study published in March of this year in the journal Animal Behavior , comparative psychologists Monique Udell, Nicole Dorey and Clive Wynn demonstrated that stray dogs reason about human social behavior (and in particular, our intentions) in a fundamentally different way from pet dogs.

The importance of Udell’s work can only be understood in relation to previous experimental findings demonstrating that domestic dogs (but not other canine species) comprehend referential meaning and cooperative intent in human behavior. That is to say, in controlled laboratory research, domestic dogs can do things such as understand that a human experimenter is intentionally trying to share his or her knowledge about the location of a hidden food reward by pointing deliberately at one of two containers. Inferring that an indexical point is symbolically about a referential target—something “out there”—instead of just an erect index finger sticking out from a closed fist requires more cognitive sophistication than it may seem at first blush. This is because comprehending such a gesture demands human-like social cognition, namely a precursor to “theory of mind,” which enables us to reason about the abstract, unobservable psychological states driving overt behavior. My cat, Tommy, is great, but when I tried to point out something in the corner of the room to him just now while sitting in my chair, he stood up on his hind legs to smell my fingertip.

For human beings, pointing is so much a part of our nature as to seem unremarkable. But if you think about it, it’s not even self-evident why our species points most commonly with its index finger rather than, say, its thumb or pinkie finger. According to a 1994 Journal of Comparative of Psychology article by University of Louisiana at Lafayette psychologist Daniel Povinelli and Louisiana State University anesthesiologist Richard Davis, there is an important anatomical difference between the relaxed postures of the human hand and that of our closet living relative, the chimpanzee. Go ahead, let your wrist go limp and look at your hand from the side, or if you’re too insecure in your own sexuality, just picture Adam’s limp wrist at the moment of creation in Michelangelo’s masterpiece on the Sistine Chapel’s ceiling. See how even in this relaxed state the index finger is slightly extended? By contrast, when chimps do this (hence the anesthesiologist coauthor), their index finger falls naturally in line with their other fingers. Povinelli and Davis reason that this subtle evolutionary change in the morphology of our hands, which occurred after humans and chimpanzees last shared a common ancestor five million to seven million years ago, is at least partially responsible for the fact that human pointing with the index finger is so culturally ubiquitous today.

The argument goes something like this. When young infants begin reaching for objects just out of their range, adults are most likely to respond to those reaching attempts and to retrieve the item for the baby when the latter’s index finger is more prominently extended. That is to say, initially, the adult mistakenly reads into the child’s reaching attempt as a communicative gesture on the part of the child. Over time, this dynamic between the child and adult serves to further “pull out” the index finger because the child implicitly learns the behavioral association, so that it slowly becomes a genuine pointing gesture. There are several varieties of pointing, and this is important for understanding why dogs’ comprehension of the gesture matters. Typically, before 18 to 24 months of age, pointing is used to manipulate others’ behaviors only, just like we’ve seen in the description above. If a baby drops her toy on the ground and points to it while looking at you, she’s basically saying, “Well, what are you waiting for, give it to me!” In developmental science jargon, this is called imperative pointing because it’s more or less a demand. But as children’s brains continue to develop, and once they’re able to conceptualize others as conversational partners who have minds that hold information, pointing becomes declarative . The two-year-old now uses pointing to say, “Hey, look over there!” because she recognizes that you’re not aware of whatever it is (the neighbor’s cat in the bushes, the exorbitantly large man eating a snow cone at the park) and wants to share this fascinating information with you. In fact, difficulties with comprehension and production of declarative pointing are often used as diagnostic indicators of autism in young children, a disorder characterized by social cognitive impairments.

So this background is important for understanding why the question of whether or not dogs possess the capacity to reason correctly about human pointing is interesting. Several years ago, Duke University psychologist Brian Hare and his colleagues reported some striking evidence showing that domestic dogs performed above chance on a variety of human-guided selection tasks—including studies in which human experimenters pointed to different objects in the room. At the time, these data were interpreted as showing that dogs have human-like social cognition allowing them to understand cooperative intent in humans. In fact, whereas tame wolves fail to score above chance in such studies, domestic dogs even outperform chimpanzees on similar pointing tests, suggesting that we may have more in common psychologically with dogs than with species for which we’re taxonomically (much) more closely related. This prompted Hare to argue that the co-evolution of humans and the domestic dog had created in the latter a genuine ability to reason about human mental states. Anybody who has ever owned or interacted with a dog can probably think up ample anecdotes here, but if not, just think of Lassie’s proverbial message-sharing abilities.

Udell and her group in Florida, however, say that these impressive social cognitive abilities in dogs may not represent the “default” canine cognitive system. In their review of this literature on dog social cognition, the authors point out that:

The currently available data suggest that populations of dogs differing in [breeding] and in environmental and lifetime pressures might display different behavioral responses to the actions of humans. Despite this fact, the great majority of subjects in studies of the origins of domestic dogs’ human-compatible social cognition have been pet dogs living in human homes, with human-oriented working dogs representing the remainder of the subject pool.

In other words, Udell and her coauthors’ contention is similar to arguments made by many researchers studying human psychological evolution—that our ability to make claims about “human nature” are seriously limited by the fact that the data upon which such claims are made are derived almost entirely from middleclass American undergraduate students between 18-22 years of age and recruited from a psychology department subject pool. She’s basically arguing that existing social cognition research on Canis lupus familiaris has largely neglected large demographic swells of the species and therefore does not necessarily paint an entirely accurate portrait of this species’ natural (default) psychological stance. Bulgaria notwithstanding, consider that within the U.S. alone six to eight million dogs can be found living in shelters each year, which is roughly 11 percent of the entire American domestic dog population.

Thus, to remedy this theoretical oversight and to address the issue of whether dogs simply inherit the cognitive capacity to think about human behaviors in intentional terms without any developmental experiences in interacting socially with human beings—that is to say, whether or not it’s something “innate” to dog psychology resulting from dogs’ brain evolution coinciding significantly with our own—Udell went to an animal shelter and administered two pointing comprehension studies with a range of dogs that were collected from the streets and classified as strays. Obviously, if these strays performed as well on such tasks as the pet dogs from earlier studies, the innateness claim would be justified. But if not, then developmental experiences with humans—and probably very particular types of experience at that—were contributing to their performing above chance in those prior studies.

After a number of training trials to ensure that the subjects understood the basic task demands, the experimenter pointed to one of two cans and then gave the dog the opportunity to respond to this helpful gesture. A “correct” response to the pointing gesture was operationalized as the dog touching or coming within 10cm of the indicated can, and for doing so the animal received a coveted food reinforcer such as a jerky treat. To prevent the dog from simply using its nose to sniff out the reward, both cans were empty—rather, the experimenter simply dropped the food reward on top of the can whenever the subject chose correctly (i.e., selected the pointed-to can).

The most significant findings from Udell’s studies were these. Although the strays performed above chance when the experimenter was kneeling on the floor and the tip of the experimenter’s finger was rigidly held 10 cm from the target can, unlike the domestic dogs in prior studies these strays failed to respond correctly to the pointing gesture when such an obvious physical cue was removed. On pointing trials in which the experimenter’s finger was 50 cm from the closest edge of the target container at full extension and then her arm was retracted back to a neutral position before the subject was allowed to make a choice, the strays’ performance fell to chance levels. This distinction is critical for the debate over whether domestic dogs have some semblance of theory of mind, because in the first instance at least, dogs may be using a simple behavioral heuristic such as “pick-the-box-closest-to-the-hand” that does not require human-like social cognition in which they are inferring cooperative intent.

In a follow-up study, Udell found some evidence that strays who were afforded additional experience with humans in an exposure condition, which involved play, petting and free exploration of the target area, showed better performance on the pointing task, but still they were less impressive in their comprehension of human pointing than the pet dogs in previous studies. Although the authors acknowledge the limitations of this work (it cannot directly address the evolution of domestic dog social behavior), Udell and her colleagues argue that these findings reveal how:

Evidence of differences in the behavior of populations within a subspecies of animals is important when attempting to draw comparative conclusions between species and subspecies of animals …

[and so] comparisons between domestic dog populations should consider different forms of response in the presence of different human stimuli instead of searching for evidence of a universal inborn capacity to respond to human gestures.

The debate reminds me of some work I was involved with back in graduate school, in which there was some evidence that great apes that had been raised by human beings as though they were human beings actually demonstrated species atypical human-like social cognition (namely, a theory of mind). The core question in all of this is what is the ‘default’ type of social cognition of a dog, of a chimp … of a human being?

In any event, now that I’m back to the home-sweet-hominess of Belfast and its ongoing riots, I can’t help but to think back to all of those motley, feral dogs of Sofia. I was ambivalent to learn about the Bulgarian government’s plans to (finally) construct a series of dog kennels to house strays by the end of 2011. On the one hand, getting these animals off the streets is clearly a good thing in many ways, both for their own wellbeing and that of, for example, the zoo animals. And there’s also, of course, the safety and interest of people—there have been several highly publicized attacks by wild dog packs, including fatalities—on children and the elderly in Sofia. On the other hand, most of these dogs are very clever, harmless, disinterested in human beings and surviving in some very complex and novel ecological conditions. Surely these stray dogs, as well as those eking out a living in other major cities in which strays are a notorious problem, present extraordinary scientific opportunities.

And from an animal wellbeing perspective, removing these stray dogs from these environmentally rich, if difficult, industrial surroundings and isolating them to physically restricted concrete-and-steel living conditions strikes me as a rather sad prospect, too. I say this having worked at the Broward County Animal Shelter in South Florida the summer after I graduated from high school, where I saw scenes of dogs in distress that haunt me to this day. (Seriously, what kind of person would discard their happy pair of 12-year-old, sweet-faced golden retrievers because their new girlfriend doesn’t “like dogs”?) In fact, embarrassingly, I was fired from that job for spending too much time petting, playing with and otherwise comforting animals that were clearly suffering. True, I should probably have been doing something more productive and useful like cleaning cages or unloading food bags, but I couldn’t help it: the sight of a suffering dog makes me human—oddly, peculiarly, probably more so than the sight of another suffering human being. But I’ll save that curiosity for another post.  

I found some solace from my poor employee behavior from this 2006 Physiology & Behavior study by Crista Coppola (coauthored by the autistic animal behavior scientist Temple Grandin). The authors point out that when dogs are housed in an animal shelter, they usually experience a severe form of psychological stress caused by exposure to novel or threatening surroundings, separation from attachment objects, unpredictability of external events, lack or loss of control over the environment, and so on. This stress activates their hypothalamic-pituitary-adrenal (HPA) axis and pumps out explosive levels of cortisol, which is the major hormonal indicator of response to stress. Coppola and her colleagues found that, regardless of breed, age of dog and sex, those shelter dogs that received a pleasant “human interaction session” on Day 2 of their incarceration had significantly lower cortisol levels on Day 9—that is to say, the benefits of this simple pet and play session were found a week later, even without any subsequent interaction with human beings during the intervening days. Honestly, this finding brings tears to my eyes: simple human affection is that long-lasting and important for dogs. And Coppola and her coauthors rightly lament that it’s a pity that the vast majority of dog shelters have not instituted routine human interaction sessions with their new, stressed-out arrivals.

But the dogs in Coppolo’s study were largely owner-surrendered ex-pets, those who’d learned to be dependent on the kindness of human beings in their lives. Whether these cortisol-relieving effects of human interaction would be replicated in, say, those stray dogs of Sofia is unclear. It’s possible that the effects would be reversed, given their developmental histories on the streets.

And so it is that those stray dogs in Bulgaria broke my heart and intrigued me by their adaptedness to their unusual surroundings. Like the Bulgarians, I was left without any easy answers to the problem. It’s a real mess indeed, with people bringing puppies home, failing to neuter them and so on. But, for science’s sake, and particularly for our understanding of the coevolution of human and canine social cognition, I’d personally love to see someone do a research documentary on the feral dogs of Sofia before they’re off the streets and locked up for good behind those new kennel walls.  


In this column presented by Scientific American Mind magazine, research psychologist Jesse Bering of Queen's University Belfast ponders some of the more obscure aspects of everyday human behavior. Sign up for the RSS feed, visit, friend Dr. Bering on Facebook or follow @JesseBering on Twitter and never miss an installment again. For articles published prior to September 29, 2009, click here: older Bering in Mind columns. Jesse's first book, The Belief Instinct (Norton) [The God Instinct (Nicholas Brealey) in the U.K.], will be published early February, 2011.

Image ©

The views expressed are those of the author and are not necessarily those of Scientific American.

Share this Article:


You must sign in or register as a member to submit a comment.

Email this Article