Appraising facial expressions of dogs

Kujala, M. V., Somppi, S., Jokela, M., Vainio, O., & Parkkonen, L. (2017). Human empathy, personality and experience affect the emotion ratings of dog and human facial expressions. PloS one, 12(1), e0170730.

Despite the long history of domestication and associated proximity of dogs and humans, human perception of canine facial expressions has not been widely researched.  This study investigated the ways in which humans rated a variety of dog facial expressions, and the extent to which psychological traits of the observers affected these emotion ratings. 

In the study, 19 female and 15 male human subjects between 25 and 46 years of age reviewed and rated 80 images.  Images included threatening, neutral and pleasant faces of a variety of unfamiliar adult dogs and humans, as well as control images of household and pixelated objects.  Subjects were instructed to rate the images subjectively in their own time, selecting responses from pre-determined numerical scales to reflect the extent to which they perceived a particular emotion to be depicted in the image.  The emotional states included were: happiness, sadness, disgust, fear, surprise, aggression/anger (all scored 1-7 according to extent of presence), arousal (scored 1-7 according to extent of arousal generated by the image), and valence (scored 1-7 from ‘very negative’ to ‘very positive’).  Subjects also completed questionnaires about their own psychological characteristics to assess personality, human-directed empathy, animal-directed empathy, and prior canine experience.

Variation in human personality traits affects individual perception of emotions in dogs 

For both dogs and humans, happiness was rated higher than other emotions for pleasant faces; aggression/anger higher than other emotions for threatening dogs; and sadness higher than other emotions for neutral dogs.  Pleasant humans were rated as happier than pleasant dogs, and neutral humans sadder than neutral dogs. 

Highest arousal was recorded for threatening faces of both species.  Threatening dogs generated more arousal than neutral and pleasant dogs, and pleasant dogs more than neutral.  Pleasant humans generated more arousal than pleasant dogs.  For valence, with the exception of neutral faces, human faces were typically rated more positively than dog faces, and threatening dogs more negatively than threatening humans.  These results suggested conspecific bias for pleasant faces, and non-conspecific bias for threatening faces, which may be related to natural sensitivity to threat perception. 

Of subject personality traits, emotional empathy had the greatest effect on ratings.  Subjects with greater emotional empathy also recorded faster response times; similar results have been found in studies investigating response to human facial expressions.  Subjects with greater prior canine experience recorded higher arousal for pleasant dogs, and rated neutral dogs with higher happiness and positivity; however, no differences were found for canine aggression based on prior experience. This and other studies have indicated that canine-experienced observers react more positively to dogs than inexperienced observers, but canine experience is not necessary to identify basic facial expressions.

Overall, results indicated that subjects assessed both human and canine facial expressions similarly, suggesting common mechanisms of appraisal.  Assessment of emotions was affected by observers’ psychological traits, and this was similarly applicable to both human conspecifics and dogs. 

World Animal Protection’s view

Understanding the emotional lives of animals is of particular importance for domesticated species with high levels of human interaction, such as dogs.  Therefore, studies investigating factors affecting human ability to detect canine emotion can help inform positive management strategies and increase human recognition of positive and negative affective states.