Verified by 
Does your dog feel guilty when he's spilled out the innards of the trash can, gnawed your penny loafers into sandals, or de-potted your favorite houseplant after a high-speed chase through the living room? Does he come to comfort you when you're upset, laying his head on your lap empathetically? Does he flaunt a tennis ball found in the bushes, tail high and wagging, obviously proud?
The ease with which we can picture these situations gives away the probable answer to my questions: it sure seems like dogs are expressing these emotions. Though these "secondary" emotions of guilt, pride, and empathy are slower to develop in human children than the so-called primary emotions such as happiness, sadness, and anger, we often find it easy to see them in our dogs from a very early age.
As both a dog owner and a researcher of dog behavior, I am curious how much of my natural tendency to anthropomorphize my pup by attributing guilt, empathy, or pride to him is actually correct. "Anthropomorphisms" are often taken to be untrue attributions that we make to animals. In reality, they are simply unproven attributions, usually of some known human ability or feeling, to animals. We use them to explain behavior, when the behaver—the dog, the cat, the chimp—can't tell us verbally what they are doing or feeling.
To test whether my attribution of emotions to my dog is correct, then, I've begun to empirically test our use of anthropomorphism. I began with "guilt," which one research study found 74% of dog owners claimed their dogs felt. Their confidence about this comes from the easily identifiable "guilty look" of a dog: he may put his ears back, duck his head or avert his eyes, drop his tail and wag low between his rear legs, or slink away from his increasingly-suspicious owner.
I designed an experiment to test what prompted that guilty look: was it in fact "guilt," or was it a response to owner behavior? We went to owners' homes, and videotaped various scenarios with these variables. Each trial began with an owner showing his dog a desirable treat, instructing him not to eat it, and then leaving the room.
While the owner was gone, the dog was given a chance to eat the treat (disobeying the owner) or we took the treat away (thereby enforcing obedience). When the owner came back, we told them that their dog had disobeyed or obeyed; they scolded the dogs for disobedience and greeted the dogs happily when they obeyed. The twist was that we sometimes misled the owners, telling them that the dog had disobeyed when he had not, or obeyed when he had not.
Later, we reviewed the dog's behavior in each case, looking at the intensity of the "guilty look," if any, on the dog. What we found was that there was no more guilty look when the dog disobeyed than when he obeyed: dogs didn't just put on the look when guilty.
By contrast, there was much more guilty look when the owner believed that the dog had disobeyed—and thus scolded him—than when the owner believed the dog had been obedient. And the dogs showed the most thoroughly guilty look when the owner scolded an innocent dog.
What this shows is that the guilty look, our index for assuming a dog knows when he's done something wrong, is actually prompted most as a response to the behavior of the human. Whether actually scolding a dog—using a deep, chastising tone of voice, approaching with a wagging finger and an angry face—or in preparation for scolding, the owners' behavior is recognizable to dogs. To avoid being punished, they act submissively, non-threateningly, as they might with a larger, more forceful playmate.
Besides satisfying my curiosity about whether we are warranted in making a quick judgment about what our dogs feel, based on their expression and behavior, there is another reason this result is relevant. Our relationships with dogs are founded on the feeling of mutual understanding: that dogs understand our rules and return our love. I am not out to disabuse us of the notion that the bond between humans and dogs is real, and is strong. What I do think can be improved is our sense of how much dogs understand what we expect of them. Our tendency to assume that dogs get it—and thus should be held responsible when they disobey—has far-reaching consequences. A few hundred years ago, this sense that animals should be held responsible for their actions—and even for the actions of other members of their species—led to what now look like preposterous punishments of animals. According to E.P. Evans' fascinating The criminal prosecution and capital punishment of animals, in the late Middle Ages, pigs were regularly arrested and convicted of various crimes, from eating a consecrated wafer (punishment: public hanging), to killing a child (punishment: hanging by hind feet; the sow's sucklings were indicted as accomplices). In 1519, a whole population of field mice was tried in court for tunneling through a field, thereby damaging a farmer's crops. They were given fourteen days to vacate the premises before being extinguished. And for years a dog was annually crucified in Rome as a lesson to the species, one member of which failed to bark and sending warning when the Gauls stealthily attacked the city at night.
These cases make it easy to see that our perception of the animal's "guilt" is a misapplication of the word to animals who didn't assent to the terms of behavior we held them to. Did they feel guilty? We can't be sure yet. So far, the animals themselves are quiet on the issue. It's up to us to look more carefully and form a judgment based on observation, not on assumption.
