Perhaps like no other animal, dogs have been offering companionship to us humans for millennia. Indeed, it has been suggested that dogs have been partners in our evolutionary journey. Dogs split from grey wolves more than 30,000 years ago. Since then, the brains and some other organs of domestic dogs have evolved in ways that are similar to our own evolutionary history.

It is not surprising that we have considered the behavior of dogs several times on this blog recently. For example, we reported that at least one dog, Batman, can learn the concept of oddity. This was a remarkable finding because the detection of oddity requires recognition of the relation among objects—a cognitively highly complex task. We also reported how canine vision differs from human vision and learned that dogs may be able to sense UV light as they have lenses that transmit UV rays. Additionally, dogs may sense the Earth’s magnetic field, judging by the presence of a flavoprotein that is sensitive to light-dependent magnetic orientation.

A recent article in the Psychonomic Society’s journal Learning & Behavior examined one particular perceptual ability of dogs, namely their skill at differentiating between different human emotions. After living in close contact with humans for so long, dogs have honed their skills in interpreting various human gestures and facial expressions. Recent evidence has identified differences in canine brain activity associated with exposure to human faces that display different emotions.

Building on this background knowledge, researchers Marcello Siniscalchi, Serenella d’Ingeo, and Angelo Quaranta focused on three aspects of canine responses to emotion depicted in human faces: the valence response, physiological activity, and their arousal-related behaviour.

Siniscalchi and colleagues created stimuli for their study by taking photos of human volunteers who were photographed while displaying the 6 emotions that have been identified as being “universal”: fear, anger, happiness, surprise, sadness, and disgust. In addition, a picture of a neutral expression was also produced. The original photos were then converted into a “mirrored chimeric picture” by taking each half of a photo (left or right) and pairing it with a mirror-reversed copy of itself. The figure below shows those stimuli consisting either of left-left or right-right composites for the two volunteer actors. (The reasons for this manipulation are not relevant here.)

The participants were 26 family dogs who were placed into a laboratory with a food bowl and two monitors for displaying the facial stimuli. The figure below shows the experimental setup:

At the beginning of a session the dog—who had been fasting for 8 hours—was placed in front of the food bowl and the experiment commenced once the subject started feeding. After some time, two faces appeared on the monitors to either side of the food bowl and remained there for 4 seconds. There were multiple such trials, and the identity of the stimulus face and the emotion being displayed was counterbalanced between trials and subjects.

The experimenters recorded various aspects of the behavioural response to the emotional stimuli: Did the dog interrupt feeding? If so, which way would the head turn? What kind of behaviour did the dog engage in? (The experimenters catalogued 26 different behaviors from the stress category, ranging from braced legs, tail down-tucked, and panting to flattened ears, whining, shaking of the body, and running away.) Finally, the experimenters also recorded the heart rate response to the stimuli using a monitor that the subject had grown accustomed to previously.

Because of the large number of measures, the results were quite rich and complex. The most readily-interpretable aspect of the data involves the time it took for the dog to react to the stimuli (called “reactivity”), and the time it took for the participant to resume feeding. Those results are shown in the two panels of the figure below:

Considering reactivity first (bottom panel B in the above figure), it is clear that the participants were quickest to respond to expressions of fear and anger, and took longest to take any notice of the neutral expressions. This pattern of results was not affected by the identity of the stimulus face (e.g., gender and laterality of the composite).

Now consider the time it took for dogs to resume feeding (top panel A in the above figure). The most striking aspect of these data is how much longer it took dogs to resume feeding after viewing an angry face than any other emotion. (There were other small differences but they pale by comparison.)

This result was mirrored in the behavioural scores involving the stress-related behaviors: Again exposure to anger maximized the participants’ subsequent stress responses, and neutral expressions elicited minimal stress. Somewhat surprisingly, happiness gave rise to the second-largest stress response.

The cardiac data largely mirrored the behavioural results: Stimuli that displayed arousing emotions (i.e., fear, anger, and happiness) led to greater increases in heart-rate than the less arousing stimuli (i.e., sadness, disgust, surprise, and neutral).

The difference between arousing emotions and their less arousing counterparts was also observed in the head-turning data. Arousing emotions tended to result in head movements that were biased to the left, whereas the remaining emotions tended to result in movements to the right. Because the direction of movement is contralateral to the hemisphere responsible for processing the stimuli, the behavioural data point to a right-hemisphere involvement in processing clearly arousing stimuli (negative emotions) and a left-hemisphere dominant processing of positive emotions. Siniscalchi and colleagues considered this result to be consonant with prior work on canine neuroscience.

Perhaps the most intriguing aspect of the results of Siniscalchi and colleagues revolves around the faces displaying happiness. Why did happiness elicit a behavioural stress response? And why did heart rate increase in response to happy faces?

Siniscalchi and colleagues argued that this occurred because of the absence of vocalizations: dogs are known to be responsive to happy faces accompanied by people’s “happy” vocalizations, but in the absence of vocalizations, dogs might perceive a person who is baring their teeth as threatening—even if that person is smiling rather than snarling.

So dogs can read your facial expressions. Perhaps it would only be fair if we learned to read theirs? The online world is full of tutorials that teach you how to interpret your dog’s body language or expressions. Here is one that you may want to try for starters:

Psychonomic Society article focused on in this post:

Siniscalchi, M., d’Ingeo, S., & Quaranta, A. (2018). Orienting asymmetries and physiological reactivity in dogs’ response to human emotional faces. Learning & Behavior. DOI: 10.3758/s13420-018-0325-2.