Understanding Canine Communication
Ever gotten the feeling that your dog is listening not just to what you say, but how you say it? You’re not alone among pet owners — and a new study in Science suggests that you’re not wrong, either.
Using fMRI machines, researchers measured the brain activity of dogs as they were given commands. Attila Andics, a neuroscientist at Eötvös Loránd University in Budapest and one of the paper’s authors, says the team found that dogs process vocabulary and intonation in different parts of their brains, similar to the way humans do.
“We said ‘Good boy! Good boy!’ [to the dogs],” Andics says. “But we also said ‘Good boy, good boy’ without the praising intonation. And … we used meaningless conjuncture words like ‘however’ or nevertheless’ in a praising intonation — and also some of these meaningless words in a normal intonation. And so we basically we could test for the differences in the brain.”
When the dogs heard praising words in praising intonations, the reward centers of their brain were active, indicating pleasure. This wasn’t the case when praise words were spoken neutrally, Andics says, or when meaningless words were spoken in a praising voice. The findings indicate that dogs can take communication cues straight from our speech.
“It’s as if you are calling your dog on the phone,” Andics says. “[In the study,] they only have the speech information, and they don’t see you or they don’t see your body gestures. They don’t have all the context. This is only word meaning and intonation which is at play here. So we were really surprised to see that they can, in this setting, actually use both of these types of cues.”
Alexandra Horowitz, a Barnard College professor and author of the forthcoming book “Being a Dog: Following the Dog Into a World of Smell,” is careful to add that while the study shows dogs can be trained to react to our speech, we still can’t say for certain that dogs understand words in the same way that we do.
“There is no question that they’re able to extract something,” Horowitz says. “Maybe the meaning, maybe they recognized the familiar word. These meaningful words were all familiar words that [researchers] saw the left hemispheric activation in. So maybe they recognize that those are familiar, but that doesn’t mean that they necessarily have the same kind of understanding as we would say a human would.”
Nevertheless, Andics and Horowitz agree that the study has cracked open new clues in our understanding of canine-human relationships. For instance, the research strongly suggests that your dog won’t fall for the old “Let’s go to the vet!” tricks — dogs have been studying us for too long.
“Dogs learn these associations much better than we think,” Horowitz says. “They see where we’re going before we’re going there. As long as they live with us and follow our habits, I think they’re kind of anthropologists in our homes, watching us. So we can’t be surprised when they know that they’re actually going to the vet.”
If dogs and humans have similar mechanisms for processing speech, what’s keeping us from communicating better with other mammals? Andics says that the brain mechanism observed in this study probably dates back some 100 million years, to the last common ancestor of dogs and humans. It may be present in other mammals, but other mammals — your insolent pet cat, for example — just may not care as much about communicating with you.
“This is what’s special about dogs,” Andics says. “I think that they care about what we communicate, what we want to communicate. They want to get something from our speech. And this is what makes them able to learn from our words. So they might not have a very different brain from other mammals, but they just live in the same environment. They got used to paying attention to us humans, so they have the opportunity to actually get familiarized with our words. They have the opportunity to learn much more from our speech than other animals who just don’t care.”
Watch Science Friday’s video from the Westminster dog show — and learn more about the science of dogs — here: www.sciencefriday.com/dogs.
Attila Andics is a neuroscientist in the Department of Ethology at Eötvös Loránd University in Budapest, Hungary.
Alexandra Horowitz is author of Being a Dog: Following the Dog Into a World of Smell (Scribner, October 2016). She’s also a professor of psychology at Barnard College in New York, New York.
JOHN DANKOSKY: This is Science Friday. I’m John Dankosky. If you’re a dog person, you know your dog isn’t just hearing your commands. They’re listening to what you say and how you say it. There’s a little side-to-side head movements they make when you’re talk.
You’ve probably tried to trick them by changing the tone of your voice now and again. A group of researchers tested this pet owner’s hunch. They measured the brain activity of dogs and confirmed that you should watch what you say and how you say it around your dog.
The study was published yesterday in the Journal of Science. Speech isn’t the only kind of human canine communication, of course. Dogs can clue into our gestures and even our moods. Communication is a two-way street. How do dogs communicate with us? How do we communicate with them? Here to discuss all this is Attila Andics, who’s one of the authors of this paper in Science. He’s a neuroscientist at Eotvos Lorand University in Budapest, Hungary. Welcome to Science Friday.
ATTILA ANDICS: Good afternoon.
JOHN DANKOSKY: And I also want to welcome in Alexandra Horowitz, who’s a psychology professor, leads the Dog Cognition Lab at Barnard College here in New York. She’s also the author of the forthcoming book Being a Dog– Following the Dog Into a World of Smell. She joins us here in our studios in New York. Welcome to the show.
ALEXANDRA HOROWITZ: Thanks for having me.
JOHN DANKOSKY: So how do you figure out what your dog is trying to communicate? You can call us at 844-724-8255. That’s 844-SCI-TALK. Or you can tweet us @scifri. I’m sure a lot of dog lovers will be calling us in just a moment with their questions.
Attila, though, I have to ask you first before anything else– I know you’ve been asked this. One of the first questions that came from us is, how do you train a dog to lay still in an fMRI machine? To actually be part of this test, you put headphones on them. Explain how you got them to do this.
ATTILA ANDICS: Yeah, it’s an amazing process, actually. I think what we have to [INAUDIBLE], [? what we ?] have to explain is that we want them to be happy volunteers in this [INAUDIBLE] and nothing else. It’s a very easy task they have to do. They just have to lie completely motionless and do nothing. And when they understand that, OK, it’s nothing else, just this, it’s working.
Basically we do a step-by-step positive reinforcement procedure with social learning technique, meaning that it’s not only one dog but also another observing dog who is there seeing the one on the scanner bed being praised and so and feeling, oh, I’m so jealous. I want to be the next one there at a place of happiness. And it seems to work. And these dogs just want to be really nice with us. They want to please us. And as soon as they figure out what’s the way to please us, they do.
JOHN DANKOSKY: OK, so let’s get into what you learned here. When you say, good boy, good boy, to your dog, what exactly is going on in his brain? What did you find out?
ATTILA ANDICS: Yeah, so we said, good boy, good boy. But we also said, good boy, good boy, without the praising intonation. And to [? trick ?] them a bit more, we also used meaningless conjunction words like “however,” “nevertheless,” praising intonation, and also some of these meaningless words in a normal intonation.
And so basically we could test for the differences in the brain. So what are the brain regions which respond differently to praise words, meaningful words, and meaningless sound sequences independently of intonation? And then what are those brain regions which respond differently to praising intonation and on praising neutral intonation [INAUDIBLE] of word meaning.
And what we found is that the brains of dogs separately process word meaning and intonational information. So the left hemisphere is more involved in processing meaningful words. So when they hear a meaningful word, the left hemisphere is more active than the right hemisphere. And when they hear meaningless sound sequences, meaningless words, then there is no difference between hemispheres.
But for intonation, we found the right hemisphere auditory region which responds differently to praising intonation and non-praising intonations. So basically, this showed that they can separate these two types of information from the speech signal. And what we also saw is that they can combine these two types of information in the so-called reward center.
So dogs, just like humans, have a reward center in their brain. And this is a very cool place, because you can just look at it, see when it’s active. And then you can guess that the dog feels rewarded. So this is reward center is active for food, just general– food reward or be petted or for a hug, for a nice music instruments.
And what we saw for praise words in praising intonation, this reward center was active. But for praise words in neutral intonation or for neutral words in praising intonation, we saw no higher activation than for neutral words and neutral intonation.
JOHN DANKOSKY: And is this pretty much what you expected to see?
ATTILA ANDICS: Well, it’s an exciting question, because in a way we were really surprised that we have a brain pattern in dogs which is similar to that found in humans. On the other hand, we all have these antidotes and these memories that we can sometimes use, only our intonation or only some word meanings [INAUDIBLE] dogs, and it works.
But what happens here is, as if you were calling your dog on the phone, they only have the speech information. They don’t see you. They don’t see their body gestures. They don’t have all the context. This is only word meaning at intonation, which is at play here. So we were really surprised to see that they can, in this setting, actually use both of these types of cues.
JOHN DANKOSKY: So Alexandra, tell me, first of all, as you listened to what Attila’s saying, what do you think of a finding like this? And how does it jive with some of the things that you’ve learned about how dogs communicate with us and what they hear from us when we talk to them?
ALEXANDRA HOROWITZ: This is really fun research, because it extends the boundaries of what’s been happening in, I would say, the testing of social cognition in dogs. We’ve always known that dogs are very good at reading signals from us. As you said, the gestures, pointing– they’re very sensitive to that, perhaps more sensitive than we sometimes are to dogs. Communicative signals and the technique of using fMRI to give some evidence that there is a differentiation in their brain between the meaning and the tone of the word is surprising and new and really exciting.
JOHN DANKOSKY: So how do we know, though, Alexandra, that we’re not just talking about animals that are very good at being trained to react to something we do versus comprehending in some way what we do. What’s the difference?
ALEXANDRA HOROWITZ: Right. The question of whether they’re understanding the words is, I think, still an open question. So there is no question that they’re able to extract something, maybe the meaning. Maybe they recognize the familiar word. These meaningful words were all familiar words that they saw the left hemisphere activation in. So maybe they recognize that those are familiar, but that doesn’t mean that they necessarily have the same kind of understanding as we would say a human would understand each other’s words.
JOHN DANKOSKY: Hmm. Attila’s speech is something that’s unique to humans. It doesn’t exist in a dog’s natural environment. Do you think that dogs can process human vocalizations?
ATTILA ANDICS: Well, it seems they can. They definitely process them in some way. And I very much agree with Alexandra that talking about understanding is a very interesting issue. And I never dare to use the word “understanding,” actually, because what we say is that they make differentiation between different types of stimuli.
And OK, what we see here now is that dog brains process words in a way similar to how human brains do. And this is very interesting, also because of humans. If you think about what humans do and how special our neural mechanisms for speech processing are, if you look at it from that perspective, then I think it’s really exciting. But I wouldn’t dare to say that this is a proof that dogs really understand everything we say or even whatever we say.
JOHN DANKOSKY: A lot of people have questions and thoughts about their dogs. Let’s go to the phones. And we’ve got Max, I believe it is, in Athens, Georgia. Hi, Max.
MAX: Hey, how’s it going?
JOHN DANKOSKY: Good. What’s up?
MAX: Well, I have a question about my dog. She’s a puppy. She’s about five months old. And she knows her name. She knows how to sit, shake hands, stuff like that. She’s overall doing really well. And she does respond really well to praise. And she’s very responsive to my voice.
And my question, though, is that, when she’s doing something bad, like maybe she snatches something off the table that she’s not supposed to have and then bolts across the room with it, it doesn’t matter what inflection of my voice that I used. She doesn’t respond. She ignores me.
I could understand her not wanting to come if I was scolding her. But even if say her name in a really, really sweet voice, it’s like she knows. And so I didn’t know if there was some tip or trick that I could use to get her more responsive to when she’s not doing well but is being bad.
JOHN DANKOSKY: Well, let’s turn to Alexandra about this, because it’s a great question. Thank you so much for that, Max.
ALEXANDRA HOROWITZ: I’ve done a lot of work on that moment when the dog is doing something wrong and the question of whether the dog is understanding that they’ve done something wrong. And in fact, it looks like what happens in these situations is dogs really learn about you.
What they learn is your body posture and your tone and how you’re approaching them. And your words are undermined by what your behavior is. So the dog is probably, in that case, looking at your behavior, even subtle behaviors you don’t think indicate that you’re angry. And therefore, even if the dog is maybe processing the word of their name or “come,” they’re not going to come to you. I mean, basically we have to have a richer vocabulary, I think, and use it consistently with dogs in different contexts if we want them to learn the specific meanings of words.
JOHN DANKOSKY: I’ve even read that that whole trick that we try to do where we go, come on, it’ll be fun. Let’s go to the vet. Come on. The dog can tell, look, it’s not time to go to the vet. I don’t really want to go to the vet right now, right?
ALEXANDRA HOROWITZ: Absolutely. Dogs learn these associations much better than we think. They see where we’re going before we’re going there as long as they live with us and follow our habits. I think they’re kind of anthropologists in our homes, watching us. So we can’t be surprised when they know that they’re actually going to the vet.
JOHN DANKOSKY: Yeah, and you–
ATTILA ANDICS: Yeah, and I–
JOHN DANKOSKY: Go ahead, Attila, yeah.
ATTILA ANDICS: And [? I always ?] get this question, what’s special about dogs? And this is what’s special about dogs, I think, that they care about what we communicate, what we want to communicate. They want to get something from our speech, and this is what makes them able to learn from our words.
So they might have a very different brains from other mammals, but they just live in the same environment. They got used to paying attention to us humans. So they have the opportunity to actually get familiarized with our words. They have the opportunity to learn much more from our speech than other animals who just don’t care.
JOHN DANKOSKY: And they want to, Attila, get something from us. And they want to get those praising words. They want to get praising human behavior, sometimes even more than food.
ATTILA ANDICS: Yeah, we are relevant to them. Our reactions are relevant to dogs.
ALEXANDRA HOROWITZ: There’s good research that shows that often phrasing can be as rewarding as food. So we should keep that in mind.
JOHN DANKOSKY: I want to go to Tim, who’s calling from Mason City, Iowa. Let’s go to Tim. Hi there. You’re on Science Friday.
TIM: Hello. Thank you.
JOHN DANKOSKY: Yeah.
TIM: I had a golden retriever named Rusty. And I did a lot of complex cooking in my kitchen. So sometimes, I would say, no dogs in the kitchen, very calmly. If I was kind of exasperated, I might say more forcefully.
But every time I said it, he would go– there was linoleum in the kitchen and then carpet in the dining living area with just an aluminum tack strip. And he would go and just put his front paws right on that aluminum tack strip. So he was out of the kitchen but just barely, just by an inch or so.
And also, when I had guests over and he started sniffing around and stuff like that, I’d say, Rusty, why don’t you go to your room? And he would. He had his own room in my house. And also, when I would say, do you want to go for a walk, he would go grab the leash and bring it over.
JOHN DANKOSKY: Tim, thank you for sharing those stories too. And Attila, these are all things that dog owners, I think, because we all love our dogs so much, there are things that we assume dogs are hearing. And then there may be things that we can know that dogs are hearing.
When you listen to Tim’s story, what do you think Tim’s dog was actually understanding or hearing that he was saying in the words “kitchen” or “let’s go for a walk”?
ATTILA ANDICS: Yeah, I just guess that word was heard by the dog many times. So if a dog gets familiar with a word or an expression, it can associate something with that. And yeah, why not use all the information which is in that speech signal? It’s not only intonation but also the same word again and again. Why not use that if they have the brain capacity for that?
ALEXANDRA HOROWITZ: And let’s remember that this particular dog also got rewarded by not having anyone scold at them anymore for doing just barely enough of what was asked. So what was really being asked is, just barely go out of the kitchen. Just barely go into the room. That’s what the dog learned.
JOHN DANKOSKY: It sounds like an awful lot of humans I know. I’m John Dankosky. This is Science Friday from PRI, Public Radio International. And let’s go back to the phones. Mickey is in Portland, Oregon. Hi there, Mickey.
MICKEY: Hi. Alexandra, I love your book Inside of a Dog. Thank you for writing it.
ALEXANDRA HOROWITZ: Thank you.
MICKEY: It was wonderful. Are you there?
ALEXANDRA HOROWITZ: Yes, thank you very much.
MICKEY: OK, so my question is this. I’ve heard that dogs can tell which eye to look at when they’re looking at us, that they look at our left eye first and then our right. And this is how humans communicate with each other. And even chimps and other primates don’t do this. I’m wondering why dogs do that and what their motivation is for doing that. I mean, I think it’s to bond with us and communicate with us, but I’m really not sure. So I’d like to hear your views.
JOHN DANKOSKY: Thanks very much.
ALEXANDRA HOROWITZ: You’re right that dogs are supremely sensitive to gaze and face of humans more than any other non-human animal. And that’s probably what [? subserves ?] and makes available their learning from our face and our gaze.
And there has been this interesting lateralization work. In fact, this latest research is similar to some of it, where you look at, do dogs do something more with one side of their brain or the other– one nostril or through the other? And in this case, they do look first at the left part of a face, research has shown, just like humans do.
So that might be a coincidence. In other words, other animals might have similar lateralization patterns if they’re social animals, something that we both have evolved separately. And it might be that it turns out that, in living with us for thousands of years, dogs have evolved a kind of special sensitivity to the components of our face that are meaningful. And that’s why they look there.
ATTILA ANDICS: And I think it’s actually something new, because when we did research with hand-raised wolves– or my colleagues, earlier some years ago– they found that those wolves, even if they learn a lot of things about human family, they don’t look back at humans. So this is a big difference between dogs and wolves, that wolves will not look back at you, and dogs keep the eye contact much longer, much easier.
JOHN DANKOSKY: And Attila, that was something. I did want to ask you about your study. I mean, you’ve got these dogs in the MRI machine who are very, very well trained, clearly are very responsive to humans. I mean, if you were able to get a wolf in a machine like that, do you think you’d see something quite a bit different?
ATTILA ANDICS: No, I don’t think I would very much difference, because the fact that we find very similar brain mechanisms for speech processing in dogs and humans, that it’s probably a brain mechanism which was there in the last common ancestor of dogs and humans some 100 million years ago. So I would guess that we could find it in other mammals, yeah, once anyone is able to train them to stay [INAUDIBLE].
JOHN DANKOSKY: Very quickly, Alexandra. How much do we know about how dogs communicate through their barks? I mean, we hear a bark, and we don’t know what’s in that bark. What do we know about the bark?
ALEXANDRA HOROWITZ: Right, I’d say that a bark is actually a minority of their communication to us. But it’s one that maybe even evolved as a kind of response to the barking, the speech that we do. And there has been studies that show that even people who are not owners of dogs can distinguish between many handfuls, six or eight types of barks in used in different contexts.
JOHN DANKOSKY: Alexandra Horowitz, a psychology professor at the Dog Cognition Lab. Thank you so much for joining us. Attila Andics, joined us from Budapest, Hungary. Thank you so much. You can read more about dogs and watch our video from the Westminster Dog Show on our website at sciencefriday.com/dogs.