Do Dogs Understand Pointing Gestures Better Than Young Kids?

6 | Posted:

By Sophia Yin, DVM, MS

R.I.P. 1966-2014

Homer and Jonesy score below the mean.Have you ever thought you were pretty brilliant, or at least above average at some common skill only to find out, after taking a test, that you were practically incompetent at that task? Well, that’s what I just found out about my three year old Jack Russell Terrier, Jonesy and his foster dog side-kick, Homer, a couple of years ago when I tested them on a two-choice task to see if they could understand pointing.

According to a research study by Gabriella Lakatos and her colleagues at Eötvös L. University (Budapest, Hungary), dogs are able to recognize the significance of pointing and have the ability to do so similar to a 2-year old child. In this study, the researchers took 15 dogs, 13 two-year-old kids, and 11 three-year-old children and put them through a similar test.

Pre-training the dogs and kids to find a reward in one of two bowls

For dogs, two bowls (brown plastic flower pots: 13 cm in diameter, 13 cm in height) were placed 1.3-1.6m apart, in front of a person on the floor. For kids, the bowls were placed on chairs. The experimenter, in the presence of the dog or kid, placed a reward in one of the bowls: food for the dogs and a favorite toy for the children. The subjects could witness this waiting from a distance of 2-2.5 m with their owner/parent standing behind them. After having the experimenter put the food/toy into the bowl, the owner/parent allowed the subject to take the reward out from the bowl. One trial lasted 30 seconds, and the procedure was repeated twice for each bowl to ensure that the subject knew that the bowls might contain some reward.

The Basic Experiment

Once this was performed they went immediately to testing where the set-up was the same, but the subject didn’t get to see which bowl had the food. The experimenter stood 0.5 meters back from the middle line between the two bowls facing the subject at a distance of 2-2.5 meters. The owner/parent held the child or dog back gently until the experimenter gave the cue. The experimenter got the child’s or dog’s attention, and when the child or dog was looking at the experimenter’s face, she pointed towards the bowl with the food. The child or dog was then allowed to choose only one of the bowls. The other was removed if the dog or child made the wrong choice

The researchers performed two experiments.

Experiment 1: Pointing with the Arms

In the first one they compared four types of pointing.

  • Distal pointing (Figure A): the experimenter pointed by extending both the arm and finger away from her body and in the direction of the bowl for 1 seconds and then putting her arm back down by her side. Theoretically, this should be the easiest to interpret because the arm is clearly extended away from the body.
  • Long cross-pointing (Figure B): This is the same as forward cross-pointing except that her index finger does protrude from her silhouettes.
  • Forward cross-pointing (Figure C): Again some part of her arm is crossing her body. This time her arm is straight but point across her body to the bowl with the food. Her hand does not extend past her silhouette.
  • Elbow cross-pointing (Figure D): in this case, the elbow is bent such that the experimenter’s elbow is facing the bowl that has no treat but her hand and index finger are pointing towards the bowl with the food. Her elbow extends outside her body but her pointing hand does not extend past her midline.


What were the results? Well, in all cases, the 3-year-old children performed better than the 2-year-olds, who performed better than the dogs. But, at least, dogs were able to interpret pointing correctly more often than not when the pointer’s index finger extended past the body—e.g. distal pointing and long cross-pointing. And in the case of the elbow cross-pointing, they tended to go to the side the elbow was pointing (although barely).

Overall the dogs’ performance was not stellar

  •  About 80% of the time they went to the correct bowl if there was distal pointing,
  •  Around 70% of the time they went to the correct bowl with long-cross-pointing.
  •  About 65% of the time they went to the bowl where the elbow pointed (rather than where the index finger pointed) with elbow cross-pointing. That’s not necessarily that much more than the 50% correct answer they would have with chance. That is, if you were taking a true-false test and just guessed, you would be likely to get 50% correct by chance.

Performance of children

  • 3-year-old children got the pointing answer correct about 100% of the time.
  • 2-year-old children went to the right bowl about 90% of the time in all cases where the pointers arms were straight.
  • 2-year-old children had problems knowing where to go when the elbow pointed one way and the index finger pointed another though and performed similarly to dogs here.

Experiment 2: Pointing with the Legs

In Experiment Two, the researchers wanted to see if the children and dogs could generalize to a different type of pointing cue—pointing with the legs.

  • Pointing with the leg that was closer to the bowl (Figure E)
  • Leg cross-pointing with the leg further from the bowl (Figure F)
  • Pointing with her knee (Figure G)
  • As a control they also used distal hand pointing


Again, 3-year-old children performed the task correctly nearly 100% of the time. Everyone performed the distal pointing, pointing with the leg, and cross-pointing with the leg better than chance, but only at 65-70% correct for dogs (e.g., barely more than compared to chance). Two-year-olds consistently performed better than dogs but, neither dogs or 2-year-olds understood the knee-pointing.

Overall study conclusion:

Overall, the researchers concluded that 3-year old children are more advanced than 2-year-olds and dogs in terms of their ability to interpret pointing with the hands as well as gesturing with other body parts to locate a hidden object. And they understood to focus on the hand and index finger in particular such that if the elbow pointed opposite the hand, they still went the direction the finger pointed.

Dogs and 2-year-old children tended to need to have the arm and hand protruding from the body to infer the direction of the hidden prize. Additionally, dogs and 2-year-olds were somewhat able understand that respond to pointing of the foot, but not as well as 3-year-olds. Finally, dogs consistently performed worse than 2-year-old kids, but in many cases, performed better than chance.

Well, that’s what the researchers concluded. But like all researchers I was skeptical. Seems like most of the time when I point to something the dog’s first response is to sniff my hand. So to see for myself, I decided to repeat the experiment on my own dogs. Surely two Jack Russells could perform as well as the average study dog.

To see the results, stay tuned for part 2 to appear in a few days.

Tags: , , , , , , , , , ,

6 responses to “Do Dogs Understand Pointing Gestures Better Than Young Kids?

  1. I can’t wait to see the results. I am terrible about assuming my guys know what I am trying to communicate to them. I have gotten as far as realizing they really don’t understand my constant prattling and will use hand gestures more and more. If I had to concentrate and think about the success of the hand gestures, I would probably hazard that I am not much more successful than chance. Maybe we should try the experiment ourselves.

  2. I’m anxious to find out how Homer and Jonesy did!

    I wonder how much sense of smell might come into play with a dog who has been trained to find things with his nose… Seems like our dog might follow his nose to some degree once he got closer to the bowls. Maybe I will get a chance to test him.

  3. Natalie,
    That’s definitely something that researchers have had to prove is not a factor when they publish their papers.

    In these studies they typically only put treats in one of the sides. But if they were concerned they could put treats on both sides but only allow access to the one. Pretty sure MY dogs were not using their nose–especially since they kept picking the wrong answer in spite of their good sense of smell:-).

    When you watch the dogs they make a beeline towards one or the other rather than sniffing or slowing down and sniffing as they get closer.

  4. I have a dog that participated in similar experiments at a canine cognition laboratory at a major university. On our first visit, with a female tester (I am female), she did exceptionally well on the tests and was probably better at it than the dogs in the study that you mention. However, at our second session, the tester used a very loud bell to signal ready (she is not scared of banging pans, etc. – she’s a therapy dog and goes into nursing homes, day cares, parking garages, and tv studios with no problem) and was male, plus there was a scary image on a large screen in front of her. She handled the male and the screen ok, but when the bell rang (it was a very loud one) she backed up under my chair. After that, she could as well do the same tasks that had been asked of her at the first session. So, experimenters need to take those kinds of possibilities into consideration. Her results were skewed at that visit in my opinion, and though I communicated that, I wonder if I was just viewed as a disgruntled dog owner. Had the same young lady who conducted the first experiment conducted the second, I suspect she would have recognized that there may be a problem with the data comparison due to the difference in the way the session was conducted. I was told that the bell had *nothing* to do with the study, and they just thought it would be a good signal – the first experimenter just said “ready” in a quiet voice. I’d be interested in your thoughts on this.

  5. Hi Anne:

    You make a really good point. The study is only as good as the researchers who design it and testers who carry it out. And sometimes the researchers are crossing over from a different field and are not that familiar with dogs. Or the testers are of variable experience. They really need to work with someone who is already an expert in dog behavior. When I was teaching a research course to graduating seniors in Animal Science at UC Davis, the students were required to work in groups designing and running an animal behavior experiment and writing it up. It was basically like doing a miniature Master’s. One of the major factors that I stressed until I was blue in the face was that the biggest hurdle they would face would be training the animals to feel comfortable in the training environment. It sounds like not everyone on this project had learned this yet.

    I would write to them so that they are aware of what you observed
    1) that whether or not they felt the bell meant something or nothing, it meant something to your dog as shown by her body posture. you can send them to the chapter 1 and point out the signs of anxiety and fear that they may have missed.

    In real life, it doesn’t matter what humans think, it matters what the dogs think in these experiments!

    I would also make it clear that you were not invested in the results… unless you were.

  6. This is really interesting to think about. I am sure that it matters on the age of the of the kid as well as the dog. Did you calculate the age of the dog in your studies?

Leave a Reply

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