A guilty dog. Can Owners really tell when a dog is guilty?

guilty dogHave you ever thought that your dog 'looked' guilty  after you returned home and found something broken or eaten that your dog should have known better not to do?

Did you think that you could tell that something was up with your dogs behaviour? That it looked guilty?  These two experiments (2009, 2012) both look at proving one way or another if our interpretations relate to the reality of an event.

OK, the first thing that people need to understand is a dogs intelligence and its ability to make human emotions. That is really what these experiments are trying to find out. Just because we treat dogs like humans, does not mean they have all human emotions.

Yet many people swear that they can tell if their dog has done something wrong. Secondly and I think just as important, dogs live in the moment. Yes they have memories and can learn, but invariably they live in the moment.

That is why they say you cant punish a dog once you have arrived home, because it wont associate the punishment with the crime they have committed. But is this so?

The first experiment by Alexandra Horowitz, in 2009 looked at finding out how well dog owners knew there dogs. In his experiment the owners were either informed correctly or incorrectly about an offence their dog committed, that their dog was trained not to do.


Horowitz found that there was a human tendency to attribute a “guilty look” to a dog not based on hard evidence that the dog did something but GUILT was assumed from the dogs body language. Unfortunately people's inability to correctly determine body language related to a specific event often meant that dogs were wrongly accused.


In the experiment, owners were asked to leave the room after ordering their dogs not to eat a tasty treat. This whole deal smells of entrapment, but it was figured that many dogs sense of smell and scavenger desire for food would overcome a command particularly in the absence of the owner.

With the owners missing, Horowitz provided ONLY some of the dogs a treat before recalling the owners.

In some experiments owners where told their dog ate the "forbidden treat". In other experiements the owners were told their dog had behaved properly and did not eat the treat.

The genius of this experiement is that the owners were often told a lie.


"Whether the dogs' demeanor included elements of the "guilty look" had little to do with whether the dogs had actually eaten the forbidden treat or not. Dogs looked most “guilty” if they were admonished by their owners for eating the treat. In fact, dogs that had been obedient and had not eaten the treat, but were scolded by their (misinformed) owners, looked more “guilty” than those that had, in fact, eaten the treat."

The tests showed that the dog GUILTY look was a response to the owners behaviour and not a result of whether or not it had misbehaved.

The point here should also included that these dogs were fairly well behaved and knew right from wrong, and it looks like they accepted their owner as the alpha dog. It would be interesting to see how they behave after being given the command by a stranger, or if their owner is not the dominant one in the relationship.


I find it slightly amazing that these kinds of research have only occurred recently but I guess funding and practical applications that can make money are all the rage.

But surely knowing more about our dogs from a training aspect, as well as for those service dogs out there is pretty important?

Anyway, this research was basially commissioned to check on the previous reseach and expand on the concept. While the previous research was still a paid research paper, this one was 'free' and so there are more statistics on it that I can site to satisfy your interest.


"The experiment explored (1) whether dogs that were disobedient in owners’ absences show ABs upon owners’ return to a room and (2) whether owners can determine if dogs were disobedient based on dog greeting behavior."

So far this is pretty similar to the previous resarch, however this one used more extensive questionairres and experiments. It was also close enough to the previous research to either verify or dismiss the previous results.

How many subjects? (Canis familiaris; N=64), That is: 52 pure bred bred dogs and 12 of mixed breed origin. Of these 37 were females and 27 males with an average age of 3.62 years.

" Under experimental conditions, owners enforced a social rule that food placed on a table was not for dogs. Dogs were then left alone and had the opportunity to disobey. Owners returned and examined dog greeting behavior to determine whether the dog ate. The experiment also ascertained how dogs greeted owners when no rule was in place. "

Dog were required to fit into the following:

  • (1) Aged at least eight months;
  • (2) dogs had lived with the owner for at least 6 months;
  • (3) Dog could remain calm if left alone in an unfamiliar room for 3 minutes;
  • (4) they had not participated in food-reward studies previously
  • (5) They might or might not eat food after being prohibited from eating and being left alone with the food.

The dog experiment was as follows:

"dog greeted owner before the establishment of a social rule; owner and experimenter established and enforced the social rule that a piece of hotdog on a table was for humans only; owner disallowed dog from eating and dog was left alone with the food; owner returned and dog greeted owner after having the opportunity to break the social rule in the owner’s absence and dog greeted owner when the social rule was no longer in place."

The interesting thing about this experiment is that the scientist identified SEVEN body language 'tells' that the dogs exhibited that could conceiveably 'give away' tyo their owner that they were guilty. While humans typically look at a human face, or a dog face to assess guilt, it was found that dog body language could play just as bit a part of assessement.

The seven 'tells' that were coded and assessed by scientists (to be used in conjunction with the owner questinairres were: the dog:

  • lowering body, tail down, moving away from owner,
  • freezing, lack of jumping,
  • turning head from owner and lowering head.


The scientits " found no significant difference in dog presentation of GUILT between dogs who ate in owners’ absences (N=32) and dogs who did not (N=26).

That said, they also found that " owners appeared able to determine if their dog ate in their absence."


Their test "did not substantiate the owner-reported anecdote that a dog who has transgressed displays GULIT to a non-scolding owner. However, many dogs that did not transgress where shown to show guilt to a scalding owner.

This does suggest a fairly high level of intelligence. That is, if the owner is upset at them for some reason, they will show signs of guilt or submission?

MORE IMPORTANTLY their " findings also suggest dog “guilty” displays could lead owners to scold dogs less."

That seems to suggest that whether or not a dog was guilty and displayed guilt, that if it was scalded, it didn't have to figure out why it was in trouble, but they knew enough to show submissive behaviours that would reduce the scalding. They could use this behaviour to manipulate their owners.

Just like the Horowitz (2009) study it was seen that dogs ONLY (statistically speaking) display GUILT to a scolding or even subtly displeased owner. Sounds like if they can get away with it, they will, if they are being told off, they act accordingly to reduce the negative impact on them.

So while the owners ' appeared successful' (better than even chance) in determining whether dogs transgressed (ate the food when told not to), the scientists could not confirm that the owners’ assessments (higher succes rate) was based on actual dog in-test greeting behavior. It is thought that the owners relied on a more personalized or holistic assessment of their dogs, in guessing that they had done wrong.

Article by Bruce Dwyer. If you wish to use any of this information please use a LINK reference to http://www.dogwalkersmelbourne.com.au

For Daily dog pictures and healthy dog treat info, Join us on http://www.facebook.com/HealthyDogTreatsShop


1 What Really Prompts The Dog's 'Guilty Look' Horowitz (2009).

2 Owner Perceptions and Behavioral Assessment of the “Guilty Look” In Dogs J. Hecht* M. Gácsi Eötvös Loránd University, Ethology Department, Budapest, Hungary


Dog Walkers Melbourne

Dog Walkers Melbourne is a professional dog walking and pet sitting service. We also provide 24 Hour Dog Day Care Services in Melbourne and surrounds. We love dogs as much as you do, and that’s why we are here.

Why not call us and try our service and see for yourself. Call 0402 262 875 today.

Contact Us

sample0402 262 875

sample bruce@dogwalkersmelbourne.com.au

             facebook  twitter  youtube

Find us on Google Map

Custom Services

Like Us on Facebook