Dog Cognition and the Domestication Hypothesis

Written by Brett Weiss

August 2019

Image by Gorkhs from Pixabay

Previous studies have shown that domestic dogs, Canis familiaris, respond to pointing gestures from humans to find hidden food.  Previous studies have also shown that hand-reared wolves, Canis lupus, do not have this ability, failing to find hidden food given human pointing gestures.  Moreover, studies have shown also that dogs have more skill than chimpanzees with using human communication to search for food (Hare et al., 2002).  This has led to the proposal that domestic dogs have abilities to understand intentions of others’, an ability not present in the ancestral population of wolves (Udell et al., 2008).  These observations in domestic dogs have led to the ‘domestication hypothesis,’ which posits that domestic dogs have special skills in comprehending the communication of humans (Hare et al., 2010).  The ‘domestication hypothesis’ also lends itself to dogs having a generalized ‘theory of mind’ in that they can infer mental states of humans with their special abilities to comprehend communication.  The following will examine studies done comparing dogs to wolves in their abilities to understand human communication.  Differing perspectives on interpretation of the results will be presented as well.

Wolves can learn human communicative cues but do not show this skill when they are young and must have training before doing so (Hare et al., 2010).  Dogs, on the other hand, can use pointing or gaze cues from humans even as puppies (6-9 weeks of age).  Puppies have this ability even when they are unadopted or still living with littermates, having very little exposure to human care (Hare et al, 2002; Hare et al., 2010).  Accordingly, this suggests that the special abilities that dogs have in understanding human communication does not require much exposure to humans if any at all (Hare et al., 2002).  A study that Udell et al. (2008) performed runs counter to this point, though.

In their study, Udell et al. (2008) used five experimental groups of canids: a group of wolves and four groups of dogs.  Caregivers tested the wolves in familiar outdoor enclosures.  Three of the dog groups consisted of pets which families raised.  These three groups were tested in different combinations with familiar or unfamiliar experimenters.  Dogs in the ‘home unfamiliar’ group had an experimenter test them in-home, dogs in the ‘outdoor familiar’ group had a familiar human test them in a less familiar outdoor environment, and dogs in the ‘outdoor unfamiliar’ group had an experimenter test them in an unfamiliar outdoor location.  The fourth group of dogs consisted of dogs from a shelter that had an experimenter test them in an unfamiliar room.

The canids of each group went through warm-up trials to test for motivation for food.  When canids from each group displayed motivation to eat food placed on top of ‘hiding locations’, the canids were included for 10 experimental test trials.  In the test trials, no food was hidden.  Instead, the experimenter pointed to a container.  When a subject touched or stayed in the vicinity of a correct container, indicating a correct choice, the experimenter clicked a counting device and dropped food on the chosen container.

After the Udell et al. (2008) group performed statistical analyses on results, the findings showed that wolves have better abilities to follow human pointing gestures in comparison to dogs.  Thus, the findings of the Udell et al. (2008) study refuted the ‘domestication hypothesis’ and that dogs have any special theory-of-mind capabilities in comparison to their ancestral population (wolves).

Another group critiqued this study and performed another experiment on shelter dogs, which showed that shelter dogs with minimal exposure to human care understand human communicative cues.  Hare et al. (2010) said that two factors confounded the results of the Udell et al. (2008) study.  First, canids that made ‘no choice’ in the Udell et al. (2008) study were counted as ‘incorrects.’  Following conventional statistical analyses that Hare et al. (2010) performed and taking into account whether the canids participated, virtually no differences were found between these groups; and all groups performed well-above chance in making ‘correct’ choices.  Thus, all groups did well in responding to human pointing cues.  The second issue that Hare et al. (2010) pointed out was that the wolves used for the Udell et al. (2008) study were familiarized previously with human pointing cues.  Some of these wolves had even been included in programs where trainers presented wolves to the public.  Hence, these wolves were not naïve to human care and human communication, which may have confounded the results.  These issues make results from the Udell et al. (2008) study difficult to interpret and detract from credibility in refuting the ‘domestication hypothesis.’

In order to find whether shelter dogs with very little exposure to human care respond to human communicative cues, Hare et al. (2010) performed another experiment.  This study used 23 domestic shelter dogs, 11 of which were thought to be feral with very little exposure to humans.  The experimenters tested the dogs in an unfamiliar room.  Two unfamiliar experimenters tested each dog, also.  The experimental test of each dog involved placing food in a container and having another container with no food.  The dogs had no knowledge of which container held the food.  The experimenter gave cues as to which container contained the food.  The first cue was ‘pointing’ to the container with food, the second cue was ‘marking’ where a colored piece of wood was placed over the container containing the food, the third cue was ‘shaking’ the correct container, and the fourth cue was ‘shaking’ the empty container.  Each subject received eight trials with each cue, with each cue given in random order.  The results showed that the dogs used all cues at above-chance levels, even the feral dogs.  The results provide evidence that shelter dogs naïve to human care have skills using a variety of human cues.  These findings also give support to the consensus in literature that dogs use human communicative cues to find hidden food, although dogs vary in degree to which they use these cues.  Previous research has shown that wolves with no prior exposure to human communicative cues do not perform this well in using human cues to find hidden food.  Hence, this study also supports the ‘domestication hypothesis’ of dogs.

A lot of evidence shows that the ‘domestication hypothesis’ of dogs has credulity.  Dogs seem to have an unusual ability to use human communication as a result of domestication.  Although some studies, such as Udell et al. (2008) seem to refute the ‘domestication hypothesis,’ experimental methods and statistical analysis methods may call the results of these studies into question.  Overall, more studies are needed to draw a firm conclusion regarding the ‘domestication hypothesis’ of dogs.


Hare B, Brown M, Williamson C, & Tomasello M (2002).  “The domestication of social cognition in dogs.”  Science.  298: 1634-1636.

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.  79: e1-e6.

Udell MAR, Dorey NR, & Wynne CDL (2008).  “Wolves outperform dogs in following human social cues.”  Animal Behaviour.  76: 1767-1773.

Leave a comment

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: