Does a Tilted Head Make a Dog Look More Appealing?
Data shows that dogs who look at you with their head tilted seem "cuter".
Posted Apr 04, 2018
I was sitting on a sofa with my Nova Scotia Duck Tolling Retriever, Ranger, beside me. A photographer was taking some snapshots of us which were going to be used as part of the publicity for a lecture series that I had agreed to give. At one point she said to me, "Is there any way that you could get him to tilt his head while looking at the camera? You know that makes a dog look much cuter."
I assured her that, although my dog knew a lot of commands, "Tilt your head" was not one of them. However her question did start me thinking about the issue of how people respond when they see a dog with its head tilted to one side. A few years ago I did some research which suggested one reason that dogs tilt their heads involves a visual factor, at least in part. The problem is that when we speak to to a dog his muzzle blocks his view of our mouth area so tilting his head gives them a better view of our face. Dogs with less pronounced muzzles are not as likely to tilt their heads when spoken to.
Immediately following the publication of that data a number of people protested saying that they believed that head tilting in dogs was a social signal or a form of communication. The idea is that this is one of those social signals that has evolved because of the association of dogs and people over millennia. In this case the suggestion is that we respond to that particular posture in a positive way (since it certainly can look cute especially in puppies). When we are responding positively to a dog it is more likely to get rewards in the form of social attention, treats, etc.
While I considered this to be an interesting idea, the first step in establishing whether it was valid or not was to determine whether or not people actually considered a dog gazing at them with its head tilted as being more appealing or cuter then seeing that same dog with its head upright. Unfortunately no experimental data existed establishing that fundamental fact.
As I now reconsidered this issue I wondered whether there was some simple method in which I could test the question of whether head tilt affects our emotional response to a dog. I knew that it would be easy to find a huge number of photographs of dogs in all different poses on the Internet, and I have used data like that in other research (click here for more about that). The trick is that simply having the photographs without some indication as to how likable these images appear to the people viewing them is not enough to answer the question. It was then that I remembered that in previous research I had found a way of getting that information by using the huge archive of photographs on Flickr.com. The reason is that Flickr's bank of images includes a way in which viewers can express an opinion about each picture. When you roll your mouse over a photo you can see that in the lower right corner of each image there is a star icon which indicates the number of "faves" that have been posted for that image. A fave is much the same as expressing a "like" on Facebook and it serves as an indication that the person viewing the image likes it, or finds it to be appealing and attractive. Simply clicking on that star allows the viewer to express that they find that particular photo to be pleasing.
Therefore I went to the Flickr site and used the search term "dog" which allowed me to collect 125 images of dogs looking directly at the camera with their head upright, while the search terms "dog tilt" allowed me to collect another 125 images of dogs looking at the camera with their head tilted. I accepted photos which were either black and white or in full color, and dogs of all ages, although I did exclude obviously young puppies since people might find these photos attractive simply because all puppies are considered to be "cute". The images had to include only one dog and that dog had to be facing the camera relatively squarely. I eliminated images where the dog's head appeared to be tilted because of the camera angle, or because the dog had twisted around in some way to look at the camera. Any image which included a person, had the dog costumed or decorated in any way, or had the dog showing an obvious emotion (for example showing teeth, or submissively lowered ears) were not entered into the sample.
The resulting data based on the number of faves for each image were then statistically analyzed (and for those of you who want to see those details they can be found in the statistical appendix at the end of this article).
The results were actually quite clear, with the images of a dog with its head tilted receiving an average of 14.1 faves as opposed to 9.5 faves for dogs with their heads upright. This difference turns out to be statistically significant which means that people viewing a dog with its head tilted are nearly 50 percent more likely to think that it is more appealing then viewing a dog with its head in a normal right position.
To the extent that when a dog is viewed as more attractive it is more likely to receive rewards of both a social and a tangible nature, this is a good thing for dogs. Since puppies also adopt this posture at a very young age it is unlikely that this behavior is learned. Thus it is possible that that as dogs and humans co-evolved the head tilt that some dogs use when humans are speaking to them became encoded in their genes. This might happen because it is adaptive in their interactions with people — after all dogs that are viewed more positively are more likely to receive better care, more protection and so forth. Of course this is a bit of an inferential leap, but based on this data it is something probably worth thinking about.
Still, I don't think that I'm going to try to find a way to teach my dog to respond to the command "Tilt your head" just to make him appear to be more photogenic.
Copyright SC Psychological Enterprises Ltd. May not be reprinted or reposted without permission
A total of 125 pictures of dogs with no head tilt and 125 dogs with head tilt which met the selection criteria (see above) were collected in the order of their appearance on Flickr.com searches,
For reasons that are unclear occasionally some photographs tend to become "viral" and have atypically high fave counts. For example in the "no head tilt" group, which has a median of 9 faves per picture, one image scored 587 faves. A few outliers like this can distort the overall data picture by artifactually inflating the variance. There are a number of ways to "clean" the data set to prevent such distortions, such as rejecting all extreme data points defined as those which are four or five standard deviations away from the mean of the remaining sample. A much simpler and more conservative technique was used in the analysis of this data set, and this involved just deleting the five highest scores in each group, which effectively produced a much more normal distribution of scores and left 120 entries per sample. The statistics below were calculated on the basis of the resulting data entries.
No Head Tilt Head Tilt
N 120 120
Mean 9.46 14.13
sd 13.89 16.88
Because the variances are unequal [F=1.48, df=119, 119] a t test assuming unequal sample variances was used
t = 2.34(df=229.47) p<0.02 [indicating a significant difference between the 2 groups.]