The New Science of Canine Cognition

Publishing Author :
Date Published : 10/07/2015


In the last two decades scientists have become more and more curious about the dog’s (Canis familiaris) seemingly innate ability to understand human cues. Historically this ability has been overlooked or dismissed in any non-primate species; dogs in particular were dismissed as an “‘artificial’ species.” (Mikolsi, Topal, & Csanyi, 2004) Then, just before the turn of the century, a trend in research began to emerge: dogs were showing ability near or above that of primates in human gaze driven studies. Following an initial study by Brian Hare, Josep Call and Michael Tomasello in 1998, researchers began to attempt to answer the questions: “How much do dogs understand about human cues?”, “What is responsible for the dog’s ability to do this (age, breed, training, domestication)”, and “If this is a learned ability, when do dogs learn this ability?”

Although researchers agree that the domestic dog has an uncanny ability to both interpret and utilize human-given cues, there continues to be debate regarding the origin of this ability. One school of thought posits that the ability to read human cues is a result of the dog’s domestication over thousands of years by humans. This is known as the “Domestication Hypothesis”. Others posit that this is not enough to explain the dog’s ability. This second group favors the “Two-Stage Hypothesis”. This hypothesis suggests that, along with domestication, dogs must have been given opportunities to enjoy positive interactions with humans, and their age and level of training must be taken into account and their lifelong experiences.

This review will cover research in the area of canine cognition from 1998 up to the present year, 2015. Generally each study reviewed was performed with the object choice task and human given cues. An object choice task is one in which an informant indicates the location of the hidden item to the subject (canid) by looking toward or pointing and looking toward one of two opaque containers (Hare & Tomasello, Domestic Dogs (Canis familiaris) Use Human and Conspecific Social Cues to Locate Hidden Food, 1999). Scoring was done based on the number of times the dog followed human cues (such as pointing or gazing, or a combination of both) and correctly chose the baited object. Before trials, dogs underwent training (commonly referred to as “pretraining”) with the experimenters in order to learn that the objects placed before them held food (small treats used to reinforce the dogs’ correct behavior; commonly called “reinforcers”). Some studies, where applicable, also included temperament testing in the form of questionnaires filled out by their owners. Most studies done have focused on human-given cues (a point, a gaze, a physical marker or any combination or variation thereof). However, some studies have incorporated conspecifics (others of the same species) and tested dogs’ ability to follow their cues in order to find hidden food.

A New Area of Research Emerges

In 1998, Dr. Brian Hare, Associate Professor of evolutionary anthropology at Duke University in North Carolina, along with Michael Tomasello and Josep Call published “Communication of Food Location between Humans and Dog (Canis familiaris)” in which the authors note the lack of “experimental studies of the process of communication between dogs and humans.” (Hare, Call, & Tomasello, Communication of Food Location Between Human and Dog (Canis Familiaris), 1998). This study, marks the beginning of dozens of studies done in the pursuit of our increased understanding of the domestic dog’s ability to correctly interpret, respond to and, in some cases mimic human cues.

The study done by Hare et al. included 4 object choice tasks. In this study (which focused on the ability of non-humans (dogs) to follow human cues in order to find food) the object choice tasks were conducted with 2 or 3 containers and 2 pet dogs. One dog had past training as a hunting dog; the other dog had only minimal, basic training and was raised by researcher Brian Hare (Hare & Tomasello, Domestic Dogs (Canis familiaris) Use Human and Conspecific Social Cues to Locate Hidden Food, 1999). The study also looked at the dog’s ability to lead a naïve human (one without prior knowledge of the food’s location) to the correct location. As a result of these studies, Hare et al came to several interesting conclusions. These conclusions would inspire other researchers to delve deeper into this new information.

Hare et al. found that dogs ignored cues that they determined to be irrelevant. Specifically, when faced with an object choice task comprised of three identical opaque cups (rather than 2), one of which contained food (the “correct” cup) and given a gaze cue from a human for the middle cup, dogs failed the trial. Researchers concluded that the gaze cue for the middle cup kept the experimenter’s head in a “neutral” position and was therefore not enough of a visual cue for the dogs to follow; the dogs assumed the cue was not for them and ignored it. (Hare, Call, & Tomasello, Communication of Food Location Between Human and Dog (Canis Familiaris), 1998)

Hare et al. also found that at least some dogs have a rudimentary understanding of the role humans play in the communication process. During trials, the trained dog attempted to initiate communication with the naïve human by “barking and orienting his body to the food’s location.” (Hare, Call, & Tomasello, Communication of Food Location Between Human and Dog (Canis Familiaris), 1998) He did this only when humans were present.

Finally, Hare et al. found that dogs have great understanding of our state of attentiveness. In the fourth trial, both dogs played fetch with the experimenter in separate trials. When the experimenter’s back was turned, the dogs went around and returned the ball to the experimenter’s front side. However, on a few occasions, the trained dog did not go around the experimenter. In these cases, he demonstrated his understanding of the human’s state of attentiveness by making communicative gestures. He barked and nudged the human in order to get his attention. (Hare, Call, & Tomasello, Communication of Food Location Between Human and Dog (Canis Familiaris), 1998)

In concluding their publication, Hare et al. suggested that future research be conducted on the communicative abilities of dogs with conspecifics, their understanding of the human’s role in communication, and that studies should be done on dogs’ communication abilities with respect to breed. (Hare, Call, & Tomasello, Communication of Food Location Between Human and Dog (Canis Familiaris), 1998)

Battle of the Hypotheses

In 2002, five years after their groundbreaking study in canine cognition, Hare and Tomasello once again studied the ability of canines to follow pointing gestures given by humans in order to find food and compared the results to the same object choice test done with chimpanzees and wolves. The researchers performed four experiments: 1. dogs and chimpanzees were tested to locate food after being given a social cue (such as a gaze or a visible mark), 2. Dogs and wolves were tested for object choice (the experimenter indicated the correct “choice” through various gestures), 3. Dogs and wolves were tested in a food-finding game, and 4. Studies 1 and 2 were repeated with canine puppies from 9 to 26 weeks of age. Dogs are known for their ability to read human body language; as our closest DNA relative, it could be assumed that chimpanzees would perform at least as well, if not better than dogs in these studies. However, researchers found that when the experimenter indicated the location of the hidden objects to the subject (by pointing, gazing or marking the barrier), primates performed on a “chance” scale of accuracy. Dogs performed much more successfully. In control situations, wolves found more food than dogs but, in human cue trials, dogs found more food. In the puppy trials, no significant difference in performance was found with regards to age or rearing (some puppies had been living with humans while others had only lived with littermates in a kennel). They also found that dogs are able to interpret a human’s cues in many different settings including when the experimenter is moving away from the object’s location but is giving an indication of the correct location. Hare and Tomasello concluded that dogs’ ability to correctly respond to human gestures originates from years of domestication, close proximity to humans, and their subsequent time spent with humans and posited the “Domestication Hypothesis”. (Hare, Brown, Williamson, & Tomasello, 2002).

Then, in 2005 and 2006 two additional hypotheses were posited as refinements of the Domestication Hypothesis. First, in 2005, Hare and Tomasello clarified their hypothesis with the addition of the “Emotional Reactivity Hypothesis” which stated that dogs were selected for their nonaggressive temperament and lack of fear towards humans as part of their domestication. In 2006, Juliana Brauer further refined the Domestication Hypothesis with her suggestion of the “Social Dog, Causal Ape Hypothesis”. This latest hypothesis stated that while dogs have evolved through domestication to respond to social human cues, apes have not.

The Emotional Reactivity Hypothesis (2005) is an extension of the Domestication Hypothesis. Hare and Tomasello posited that as part of domestication, dogs were selected for less fear and aggression towards humans (Hare & Tomasello, Human-like social skills in dogs?, 2005).  In a follow-up article, Hare and Tomasello stated that the “…‘emotional reactivity hypothesis which suggests that selection on social-emotional systems could have provided an initial catalyst for wider social cognitive evolution in dogs, other non-human species and perhaps even in human evolution.” (Hare & Tomasello, The emotional reactivity hypothesis and cognitive evolution, 2005) The idea is that dogs who displayed less fear and aggression would have been easier to domesticate; those that continued to display fear and aggression would not have been cooperative in human-dog activities and relationships and therefore would have been selected against.

In a 2006 study led by Dr. Juliane Brauer, researcher at the Center for Emotional History of the Max Planck Institute for Human Development, researchers coined the “Social Dog, Causal Ape Hypothesis”. Their hypothesis is based on results that showed that while apes excelled at correctly identifying the baited cup in an object choice task when a cause-and-effect cue was presented (such as shaking the baited cup to make noise), they did not perform well in the same test when the cue was of a social nature, such as pointing or gazing (Brauer, Kaminski, Riedel, Call, & Tomasello, 2006). Dogs however did excel at the latter and not the former test. Researchers concluded that apes are able to infer the human’s meaning when their cues are “goal-directed” rather than communicative-directed. Dogs on the other hand are masters of human body language and can reliably follow human social cues. In support of the Domestication Hypothesis, Brauer et al. noted that “These facts suggest that dogs’ ability to read human communicative cues is independent of their individual history, and it is very likely that this ability evolved in the context of the domestication process.” (Brauer, Kaminski, Riedel, Call, & Tomasello, 2006)

In a review of Hare and Tomasello’s Emotional Reactivity Hypothesis (2005), Dr. Adam Miklosi, founder of the Family Dog Project in Hungary and Head of Ethology at Eötvös Loránd University and Dr. Jozsef Topal, researcher at the Comparative Ethology Research Group, Lorand Eotvos University refute the idea that domestication and selection for less aggression is enough to explain the dog’s ability to interact with humans and understand their cues since research has shown no correlation “between aggression and performance in the cueing-test” (Miklosi & Topal, 2005). They postulate two problems with the hypothesis: 1. the process of domestication was likely a result of “the type of interaction between humans and [dogs]” (Miklosi & Topal, 2005) and 2. “Any emergent social skill towards humans in [dogs] is probably a function of the social behavior exhibited by the wild ancestor” (Miklosi & Topal, 2005). In other words, as opposed to selection for nonaggression and lack of fear, the process of domestication was likely a result of wolves (dogs’ wild ancestor) approaching humans for food, showing ability and willingness to follow human cues and instructions in exchange for food and willingness to stay close by and hunt alongside humans (as they hunt cooperatively in their own packs) thereby providing reciprocal benefit to humans. Miklosi and Topal suggest that a different approach should be employed to determine the origin of dogs’ social skills: changes in behavior related to social interaction. For instance, dogs bark whereas wolves do not. It seems that “dogs ‘invented’ barking in fearful situations and, unlike wolves, they seem to be able to modify the frequency and pulsing of barking” (Miklosi & Topal, 2005).

Other researchers have posited conflicting hypotheses. In 2008 Dr. Monique Udell Assistant Professor at Animal and Rangeland Sciences of Oregon State University, Dr. Clive Wynne, professor of psychology at Arizona State University, and Dr. Nicole Dorey, lecturer at the University of Florida and researcher at the Canine Cognition Lab began testing the ability of wolves to follow human cues. They also tested the differences between a pet dog’s and a shelter dog’s (one found as a stray and living in a kennel) ability to follow human cues. Then in 2010, they posited the “Two Stage Hypothesis” which challenges the Domestication Hypothesis in that it states that domestication is not enough to explain the dog’s ability to follow human cues; dogs must also have been exposed to positive and recurring experiences with humans (Udell, Dorey, & Wynne, What did domestication do to dogs? A new account of dogs’ sensitivity to human actions, 2010).

The two groups of researchers championing the Domestication Hypothesis and the Two Stage Hypothesis have gone back and forth reviewing and refuting each other’s hypotheses. In 2009, in response to Udell et al.’s (2008) study comparing domestic dogs, human-raised wolves and shelter dogs in their ability to follow human cues, Hare and Tomasello dismissed the study as “invalid” since their methodologies differed greatly from previous studies and since their scoring was inconsistent with other, similar studies (Hare, et al., 2010). Udell and Wynne fired back with their own response stating that they do not deny that domestication has played a part in dogs’ social behavior but, that it is not enough of an explanation. Udell and Wynne challenged Hare et al.’s assertion in their studies on the Domestication Hypothesis that “control conditions ruled out the possibility that the dogs were using other kinds of cues, such as olfactory cues, to find the food” (Hare & Tomasello, Domestic Dogs (Canis familiaris) Use Human and Conspecific Social Cues to Locate Hidden Food, 1999) (Hare, Brown, Williamson, & Tomasello, 2002) stating that “none of the three studies …have reported controls for olfactory cues.” (Udell & Wynne, 2010) Additionally, Udell and Wynne noted that in these same studies, Hare and Tomasello used young dog puppies as subjects (between 9 and 26 weeks of age) claiming that some of the pups had had little exposure to humans (Hare, Brown, Williamson, & Tomasello, 2002). Based on this, they interpreted their findings to mean that dogs had an innate ability to follow human cues, regardless of age or training (Hare, et al., 2010). Udell and Wynne pointed out that these dogs with “minimal exposure to humans” were actually from a breeding facility dedicated to placing dogs into family homes and that, as part of their service, the dogs were socialized with humans quite extensively.

And So The Debate Continues…

With researchers still trying to pin down the origin of the dog’s uncanny ability to follow human cues, there are many more questions that need answering. If this ability can be attributed to domestication alone, then the age and level of training of the dog should not make a difference when tested in an object choice by human cue study. Alternatively, if dogs have learned this skill through experience with humans, this would be evident in studies regarding ontogeny (lifespan), circumstances and socialization.

The final part of this review is organized chronologically from 2007 to the present as scientists attempt to answer these questions in pursuit of this knowledge:

  • “What role does training play in the dog’s ability to interpret human cues?”
  • “At what age is a dog able to correctly follow a human’s cues?”
  • “Is there a difference between shelter dogs and pet dogs when following human cues?”
  • “Can dogs differentiate between cues meant for them and those not?”

Studies Done From 2007-2010

In 2007, five years after Hare et al.’s original study regarding the origin of dogs’ social skills and two years after their refinement, the Emotional Reactivity Hypothesis, researchers led by Julia Riedel, researcher specializing in Primatology at the Max Planck Institute for Evolutionary Anthropology in Germany, studied the ability of young puppies (6, 8, 16 and 24 weeks old) to follow human cues in order to find food. Their test was designed to determine whether human exposure to puppies influences their ability to follow social cues (Riedel, Schumann, Kaminski, Call , & Tomasello, 2008). In preparation for the study, the puppies had learned that a “cup” (as used in the experiment) may contain food. For each age group one experimenter hid a piece of food under one cup with another identical, but empty, cup present. Four variations of the experiment were performed: Control (no cues), Dynamic Cross Point (experimenter pointed to and alternated gaze between correct cup and puppy), Dynamic Cross Point Move (pointed repeatedly at the correct cup and alternated gaze between correct cup and puppy), and Marker (a physical marker was placed on the correct cup). The results of this experiment indicated that puppies as young as 6 weeks old (before leaving their mother and before the “socialization period” is complete) are capable of understanding human cues. These results would seem to support the “Domestication Hypothesis” which states that the ability to understand human cues is not learned but is ingrained in dogs as part of the domestication process. However, as Udell et al. pointed out in their review (2010), even dogs with minimal human exposure have had some exposure, especially those being bred and raised to be family pets. As it’s the responsibility of the breeder to properly socialize even very young puppies to humans, it is difficult to measure how much this exposure has influenced the puppies’ ability to follow human cues.

Two years before publishing their Two Stage Hypothesis, Dr.s Udell, Dorey, and Wynne began building a case against the Domestication Hypothesis with their study on wolves’ ability to respond to human social cues. Udell et al. successfully showed that, given enough positive exposure to humans and reinforcement for correct responses, wolves could and did pass the object choice by human cue task (Udell, Dorey , & Wynne, Wolves outperform dogs in following human social cues, 2008). The wolves in this study were raised by humans and actually outperformed dogs in the cueing task. As part of this study, Udell et al. also compared domestic pet dogs to those living in a shelter. Although the shelter dogs were willing to interact with experimenters and took food from them, they failed every trial which included the pointing cue given by a human (Udell, Dorey , & Wynne, Wolves outperform dogs in following human social cues, 2008).  Results support their idea that domestication alone cannot account for dogs’ social cognitive skills since the wolves had not undergone generations of domestication; they were “tame” and nonaggressive, but not domesticated (Udell, Dorey , & Wynne, Wolves outperform dogs in following human social cues, 2008).  It is also notable that Udell et al. corrected a significant flaw occurring in previous studies: the use of dogs’ most profound sense, olfactory. Rather than baiting the cups and asking the canids to choose, the experimenter indicated the “correct” cup with a point and rewarded the canids for the correct choice with a click (from a common training clicker) and a food reinforcer. This method eliminates the possibility that canids were making their choice based on smell. Hare and Tomasello would later dismiss this study as “invalid” (Hare, et al., 2010).

A year later and one year before formulating the “Two Stage Hypothesis” Dr. Nicole Dorey and colleagues studied whether young puppies are able to follow the physical gesture of humans (pointing) and, if not, at what age this ability manifests. Puppies ranging from 9 to 24 weeks were divided into four groups of 3 weeks each. As in similar studies (Riedel, Schumann, Kaminski, Call , & Tomasello, 2008), the puppies had already learned that a “cup” (as used in the experiment) may contain food. Cups were baited with food, outside of the puppies’ view. The puppies were asked to identify the cup containing food (by following the experimenter’s gesture) within 3 minutes.  As they hypothesized, researchers found that the puppies’ ability to follow the cues improved with age. “The performance of puppies on the object choice task, requiring the use of a momentary proximal point, improved as a function of age.” (Dorey, Udell, & Wynne, 2009) These results refute the Domestication Hypothesis’ (Hare et al, 2002) claim that age was not a factor in the dogs’ ability to interpret human cues. The results of this experiment indicate that puppies do learn human cues and improve their interpretation of human cues as they grow. It could be argued, though, that dogs are able to follow human cues due to the longevity of time they’ve spent with humans. The latter interpretation would support the Domestication Hypothesis.

Studies Done From 2011-2015

Two years after the postulation of the Two Stage Hypothesis, researchers found more evidence for the significant effect training has on dogs’ communicative skills with humans. Led by Adriana Jakovcevic, researcher at the Institute of Medical Research in Argentina a group of researchers studied the relationship between a dog’s ability to communicate by gazing when food is present but not in sight and their temperament (personality/disposition) (Jakovcevic, 2012). Test subjects first underwent a temperament test (owners completed a questionnaire regarding their dog’s reaction to strangers). Dogs then underwent “acquisition” trials in which reinforcers (treats) were delivered for gazing at the experimenter’s face. Researchers found that the dogs identified by the questionnaire as less sociable took longer to gaze at the experimenter’s face. Dogs that were more sociable and open to receiving physical contact from the experimenter (a stranger) gazed longer at her face during the last phase of the trials, “extinction” (a process by which the dog “unlearns” a learned behavior when reinforcers are no longer given). Both of these results indicate that the original hypothesis was correct: more sociable dogs do gaze longer to receive a reward. These results favor the Two Stage Hypothesis since socialization is a form of training; the dogs with more training and thereby more exposure to humans and their cues performed better in this study. If the Domestication Hypothesis is to be believed then the dogs with less training should have performed equally as well since the hypothesis claims that training and age do not affect dogs’ ability to respond to human cues. Of particular interest for future research is the latter result which indicates that the more sociable dog adapted his learned behavior (short gaze) by gazing longer at the experimenter’s face when that behavior was no longer being reinforced. This implies problem solving. In this case dogs changed their behavior after a short amount of training. They implemented their own version of a human cue (eye contact) in order to get what they wanted (a treat).

In a very recent study, researchers led by PhD candidate Lisa J. Wallis, a student at the Messerli Research Institute in Vienna found evidence for the first hypothesis: Domestication. In this 2015 study, Wallis et al. tested whether dogs’ ability to follow a human’s gaze is affected by age and/or training (Wallis , et al., 2015). During trials, the experimenter first gained the dog’s attention and conditioned the dog to watch her face (not to follow her gaze; dogs either did or did not follow of their own accord, without reinforcers), once the dog’s attention was constant (in the presence of distractions in the testing room) the experimenter gave a surprised expression and then either gazed at the door (in the test group) or down at her feet (in the control group) (Wallis , et al., 2015). Results showed that dogs have a strong propensity for following the gaze of a human and that age had no effect on this behavior. In fact, researchers found “no evidence” that dogs learned to display this behavior through ontogeny (Wallis , et al., 2015). In other words, regardless of age, dogs showed consistent behavior. The only “learning that dogs displayed was in response to the reinforcers; results showed that dogs were less likely to follow the human’s gaze, but rather to keep their attention on the experimenter’s face, after receiving positive reinforcement for the latter behavior. This doesn’t mean that dogs “forgot” how or when to follow a human’s gaze but that this common behavior can be modified with reinforcement for another behavior. These results are fascinating since the finding that age was irrelevant is support for the Domestication Hypothesis and in direct contrast to other, similar studies.


Over the last two decades scientists have been debating the origin of the domestic dog’s social skills with regards to their human companions. Researchers agree that the domestic dog has an uncanny ability to both interpret and utilize human-given cues and they agree that domestication has played a part in the evolution of this skill. Unfortunately, though, that’s about all they agree on; the issue is still debatable and requires more research. However, evidence appears to be skewed in favor of the Two Stage Hypothesis over the Domestication Hypothesis. Evidence that human-raised wolves, with no history of domestication can outperform domestic, pet dogs in the ubiquitous object choice by human cue task cannot and should not be overlooked.








Works Cited

Brauer, J., Kaminski, J., Riedel, J., Call, J., & Tomasello, M. (2006). Making Inferences About the Location of Hidden Food: Social Dog, Causal Ape. Journal of Comparative Psychology, 38-47.

Dorey, N., Udell, M., & Wynne, C. (2009). When do domestic dogs, Canis familiaris, start to understand human pointing? The role of ontogeny in the development of interspecies communication. Animal Behaviour, 37-41.

Hare, B., & Tomasello, M. (1999). Domestic Dogs (Canis familiaris) Use Human and Conspecific Social Cues to Locate Hidden Food. Journal of Comparitive Psychology, 173-177.

Hare, B., & Tomasello, M. (2005). Human-like social skills in dogs? Trends in Cognitive Science, 439-444.

Hare, B., & Tomasello, M. (2005). The emotional reactivity hypothesis and cognitive evolution. Trends in Cognitive Sciences, 464-465.

Hare, B., Brown, M., Williamson, C., & Tomasello, M. (2002). The Domestication of Social Cognition in Dogs. Science Magazine, 1634-6.

Hare, B., Call, J., & Tomasello, M. (1998). Communication of Food Location Between Human and Dog (Canis Familiaris). Evolution of Communication, 137-159.

Hare, B., Rosati, A., Kaminski, J., Brauer, J., Call, J., & Tomasello, M. (2010). The domestication hypothesis for dogs’ skills with human communication: a response to Udell et al. (2008) and Wynne et al. (2008). Animal Behaviour, e1-e6.

Jakovcevic, A. M. (2012). Do more sociable dogs gaze longer to the human face than less sociable ones? Behavioural Processes, 90(2), 217-222.

Miklosi, A., & Topal, J. (2005). Is there a simple recipe for how to make friends? Trends in Cognitive Sciences, 463-464.

Mikolsi, A., Topal, J., & Csanyi, V. (2004). Comparative social cognition: what can dogs teach us? Animal Behaviour, 995-1004.

Riedel, J., Schumann, K., Kaminski, J., Call , J., & Tomasello, M. (2008). The early ontogeny of human-dog communication. Animal Behaviour, 1003-1014.

Udell, M., & Wynne, C. (2010). Ontogeny and phylogeny: both are essential to human-sensitive behaviour in the genus Canis. Animal Behaviour, e9-e14.

Udell, M., Dorey , N., & Wynne, C. (2008). Wolves outperform dogs in following human social cues. Animal Behaviour, 1767-1773.

Udell, M., Dorey, N., & Wynne, C. (2010). What did domestication do to dogs? A new account of dogs’ sensitivity to human actions. Biological Reviews, 327-345.

Wallis , L., Range , F., Müller , C., Serisier, S., Huber , L., & Viranyi, Z. (2015). Training for eye contact modulates gaze following in dogs. Animal Behaviour, 27-35.

Leave a Reply

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