Weekend Reads | Dogs are 'Man's Best Friend,' but Can We Ever Truly Know How They Feel?

I realize that in a dog-loving city like Seattle, this column could get me into trouble … but I'm going to plow ahead into the danger zone anyway. This weekend's read is a new report arguing that to the extent that dogs have emotions, we humans are pretty terrible at reading them.

People generally believe dogs feel the same range of emotions as people. But since a dog can't tell us what it's feeling, it's sort of an act of faith on our part. And to the extent that it's been tested, those studies test how well humans agree on what emotion a dog might be feeling (assuming it is indeed feeling one), rather than an objective measure of what the dog is actually feeling. At the same time, other studies show that people tend to anthropomorphize dog behaviors: They equate them to human behaviors. Humans, in general, have a strong tendency to assign human qualities to both living creatures and inanimate objects, and pets are no exception. That doesn't mean they're wrong — dogs may indeed have the same range of feelings as humans, but even if they do, that doesn't mean we're reading them accurately.

A pair of researchers at Arizona State University decided to look at what factors might influence how people perceive dog emotions. Rather than trying to mix-and-match a list of possible emotions they might be feeling, the researchers decided to limit the measurement of emotion to two metrics: valence, the extent to which a feeling is good or bad; and arousal, how excited the feeling makes us. It turns out that you can map a wide range of emotions onto different places on the spectrum of valence and arousal.

The researchers showed test subjects videos of dogs on a blank background with all context removed, and asked them to rate the dog on valence and arousal. They then repeated it with the same video of the dog, but this time with context included: room background; the dog's owner present and interacting with the dog; and any other kind of stimulus, such as food rewards, a leash, or even a vacuum cleaner. They found that the context was very important: It caused test subjects to rate valence differently, and a lack of context caused subjects to rank arousal differently. There were some demographic influences too: Younger people tended to give higher valence and arousal scores, and people who rated themselves as more familiar with dogs also tended to give higher valence scores (but not higher arousal). 

The researchers then created a second experiment to dive deeper into how context affects perception of dog emotions. This time, they showed the dog videos (both positive and negative situations) with their original context. Then, they made two additional sets of videos: one in which they swapped out positive context for different positive context and negative context for different negative context; and one in which they swapped out the context for an opposite one (positive for negative, negative for positive). And the results were striking: across the board, test subjects' ratings for valence consistently matched the context shown, not the dog emotion.

The clear conclusion: Our perception of a dog's emotional state is heavily biased by the context in which we are experiencing it. In total, they found that 90% of the test subjects' responses were anthropomorphized. The researchers include in their report some interesting discussion of various theories of emotional states, including whether emotions are universal or are constructed from our context, culture, and past experience.

I'm sure many dog owners will disagree with these results based upon their beliefs about their own relationship with their dog, and it may be that over time we get better at reading our own dogs' experience. The researchers also point out that ethical constraints prevented them from subjecting dogs to extreme stimuli (particularly negative ones), and humans may be better at reading dogs' reactions when the context is more extreme. Nevertheless, this should cause us to question whether we truly understand our dogs' emotions, or we are simply applying our own human biases to them.

Barking Up the Wrong Tree: Human Perception of Dog Emotions Is Influenced by Extraneous Factors