It’s obvious that people can communicate with dogs. Throw a ball and tell a dog “fetch,” and the dog retrieves the ball. But do dogs understand the word “fetch,” or do they just understand that you threw a ball and made a sound? They could also be focused on some other cue entirely — the tone of your voice, or the direction of your gaze. Luckily, scientists have looked into this.
What Is “Understanding,” Anyway?
A recent study in Frontiers in Neuroscience is interested in exactly this question. Can dogs understand words? “Understanding” is a big, complex concept — difficult to test in the scope of one study, and difficult to define for a dog. Think of all the things that go on in your mind when you hear a word you “understand”: you remember what it means; if it has multiple meanings, you figure out which one makes the most sense in context; you grasp its tone and level of formality; and so on.
Do we require dogs to do all of that? Just some? It’s an open question, but the researchers decided that one fundamental, cross-species element of understanding was “the discrimination of words from non-words.” In other words, dogs should be able to differentiate familiar commands from nonsense sounds.
They decided to take a new approach to testing dog understanding, though. Historically, researchers have measured dogs’ understanding through action. In other words, if a dog understands the term “fetch,” they’ll fetch a ball. This mixes up understanding with obedience, however, and muddies the results.
So in this study, the researchers relied not on action, but on fMRI brain scans. While dogs were in the scanners, their owners said a variety of familiar and gibberish words to them. However, none of the familiar words were action-oriented — instead, they were the names of two different dog toys (ex: “Piggy” and “Monkey”). Would the dogs exhibit different levels of brain activity when they heard “Piggy” than when they heard a gibberish word? Short answer: yes!
Inside a Dog Brain
Here’s the long answer: This study of 12 dogs began with training. Each dog spent between two and six months learning the names of two different toys. First, the dogs’ owners just repeatedly named the objects. They’d play fetch or tug-of-war with their dogs using, say, Monkey, and repeatedly say “Monkey!” as they played to reinforce the object’s name in the dog’s brain.
Then, the dogs transitioned into identifying the objects. Owners would place Monkey and Piggy several feet from the dog, and several feet apart. The owners would give commands like “Get Monkey!” or “Where is Piggy?” and only give out treats if their dogs picked up the correct toy.
The fMRI scan was the final stage. While dogs held still in the tubular machines, their owners addressed them from outside, saying a mix of known words like “Monkey” and “Piggy” and mechanically generated non-words like “stru,” in various combinations.
It turned out that the dogs did reach the base criteria for understanding — their brains reacted differently to familiar words than they did to unknown words. Interestingly, the dogs’ brains reacted differently than human brains; their brain activity spiked when they heard unfamiliar words, whereas human brain activity spikes when we hear familiar words. It’s unclear why dogs react differently to the unfamiliar, or how deep their understanding of our words goes. But this study at least suggests that our speech is more than background chatter to dogs — they can distinguish one word from another.