Follow-up on “Can you identify dogs’ emotions from their facial expressions?”

Last April, I published a post about dog’s facial expressions and how humans tend to be able to correctly classify the emotions they convey.

I included 2 photographs of my own dog, Charlie, each of them portraying him in a different condition. I used procedures to induce emotional states that were either positive or negative. Then I asked readers to rate the pictures in terms of the emotions they perceived were expressed.

Here are the method I used and the results of the ratings.

charlie1charlie2Charlie was asked to sit and stay put. He was then presented with an object – a treat (dog biscuit) to induce positive emotions, like happiness and interest, or a pair of tweezers (that I use to pull hair out from between his teeth and gum) to induce negative ones, like fear or sadness.

The photos were taken almost immediately after object presentation. The first on the left corresponds to the positive condition, whereas the second on the right corresponds to the negative condition.

Now, what did the respondents think?

First of all, I’d like to say that I’m amazed at how many responses I got: 94 in total!

Figure 1. Number of responses per emotion rating scale point

Figure 1. Number of responses per emotion rating scale point

Figure 1 demonstrates an overall floor effect – most responses were situated at the lower extreme of the rating scales. I seems like people found it easier to declare that a given emotion was absent rather than say they perceived one. This makes sense since I was not looking to elicit strong emotions. Had I done that, the responses might have been more equally distributed.

Figure 2. Basic emotion ratings in each of the conditions.

Figure 2. Basic emotion ratings in each of the conditions (sorry for the missing error bars – I could not add them without not messing up the graph).

The results show that, in both cases, people perceived Charlie to feel happy, sad, surprised, and fearful (Figure 2). This is interesting because it suggests that his face displays a mix of those emotions regardless of how he feels, as if his idiosyncratic facial features are always evocative of them like a human’s can be. Of course, a simplest explanation would be that the mildness of his affective state made it difficult to identify the emotion.

The major difference between the 2 conditions seems to lie in the 5th basic emotion – no one in the positive condition thought that Charlie felt angry.

Perhaps an element of his facial expression differed distinctly between the photos. If so, I’d say it was his ears. They either face forward or pulled back depending on the valence of the presented stimulus, i.e., positive or negative.

For a methodologically flawed little questionnaire, I think it yielded pretty interesting results. Thank you to all who rated!


2 thoughts on “Follow-up on “Can you identify dogs’ emotions from their facial expressions?”

  1. Mados

    The type of dog and the expressivenes of the individual dog also matters. A boxer’s face is pretty much always the same in my view – a “flat affect” caused by its face being sort of sqeezed together (same with pugs and bulldogs) whereas with some other dogs, like labradors and german shepherd with more distance in their faces, it is much easier to tell face expressions apart. Plus, individual dogs can have very expressive faces VS sfinx-like pokerfaces, just like humans differ wildly in expressiveness. (maybe owners of breeds with flat faces can better tell the expressions apart though)

    I would also like to point out the obvious…. That an Internet survey has so strong built-in selection bias that you can’t draw general conclusions from the results (the subpopulation you are surveying is the segment of your readers that chose to participate, which is extremely unlikely to represent a ramdom sample of people in general)… Still a great way to present/discuss survey design though.

    1. Ria Pi Post author

      Thank you for your comment!

      I imagine there would indeed be differences between different breeds and between participants, according to the ‘kind’ of face they are used to seeing. It’s interesting because I always reflected that flat faces like boxers’ are ‘easier’ to interpret/read (or we think it’s easier) because they are morphologically closer to humans’. Then again, it’s the breed that I have always had…

      You are definitely right about the participant sampling bias of my small survey. I thought I mentioned its limitations but maybe not this one. Not only were the people who responded non-randomly sampled, they were also very few as compared to what I’d need to actually publish the results. So of course, I wouldn’t draw any general conclusions from it at all. It was more a way for me to interact with readers by asking their opinion and see if some interesting results might come out, even if they are heavily biased.


Share your thoughts

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s