Nowadays, so many people’s social media algorithms are filled with animal videos. One particular type of video — dogs pushing word buttons — has really gone viral. In these videos, the dogs appear to have meaningful conversations with their owners using word buttons to ask for things like treats, walks, or even to snuggle. But is it real language, real communication, or just a very clever dog trick? NOVA’s new special, “Can Dogs Talk?” follows scientists who set out to find out what’s really happening. Olly Bootle, director of “Can Dogs Talk?” joined GBH’s All Things Considered’s Arun Rath to talk about these dogs and the film. What follows is a lightly edited transcript of their conversation.
Arun Rath: We’re very excited to talk with you because we were all talking about this. Everybody who has a dog and has watched these videos has been dying for the NOVA documentary, where you explain to us, 'Is this really real?’ But, for the people who haven’t watched tons of these videos, first off, explain what these word buttons are and what’s happening in these videos.
Olly Bootle: Sure. So, there are plastic buttons. You put them on the floor around your house, and your dog can step on them with a paw to say a word that you’ve recorded on the button. So, you might record “food” on one button, “outside” on another button, and “play” on another button. You might recall all the different kinds of words you think your dog might want to say, might want to learn, so that your dog can communicate with you. Some dogs have up to about 150 buttons. Most have three or four. Most don’t really get beyond the handful, to be honest, because it’s a new form of communication for them to learn.
Rath: I want to play a cut from Federico Rossano, a cognitive scientist who appeared in the film.
Dr. Federico Rossano excerpt: What we want is to look at these anecdotes and think, is this even possible? And if it has not been trained extensively, if it seems to be spontaneous, then the next question is, can we elicit it in a more systematic way?
Rath: Can you talk about the kind of test researchers did to tell if the dogs were actually communicating ideas, like real sentences, or if it was just a trained behavior, as they said?
Bootle: Yeah, absolutely. So this gets to the heart of it. It’s so hard for scientists and for us viewers when we see these clips online to understand what’s going on. It could be entirely random. If you cover a floor with these buttons, dogs are going to walk over them by accident sometimes, or maybe just out of curiosity, they’ll press some buttons. That doesn’t really mean anything interesting. It doesn’t mean they know what they’re saying or what they are doing. So, it’s incredibly hard for scientists to really understand what’s going on in a dog’s mind. So there are all sorts of studies that Dr. Rossano and his colleagues are undertaking to try to figure out what all this means and what’s going on. Some of the first studies they’ve done are looking at whether the dogs even hear the buttons in the same way we think they hear them. So, if I say the word “play” to a dog, that dog may understand what I’m saying. If, however, I record the word “play” onto a button and then press the button, does the dog understand it in the same way, or is that a different thing? Maybe the digitalization is causing it not to fully understand, so that was one of the first studies they did. And, yes, the dogs do understand the buttons in a way that’s not particularly exciting. It doesn’t tell you anything about whether they understand them when they’re using them, or whether they mean to use them in the same way. But it’s nonetheless essential to ... start with the basics, otherwise, you can’t understand the more interesting things. They’ve also looked at whether dogs’ button use is random or not. So, of course, if it was random, if dogs are just haphazardly pressing any buttons that they’ve got without ever knowing what they’re asking for. Then, Dr. Rossano hypothesized that all the buttons that the dog had would have an equal probability of being pressed. There wouldn’t be any skew towards any one button or over another, because it’s all just random. But he found that certain words are more popular than others, and that indicates that dogs are deliberately pressing particular buttons.
Rath: It’s fun seeing that world cloud in the film of the dog’s favorite words.
Bootle: Yeah, exactly, and reassuring to know that their favorite words do make sense. There are things like, you know, “food” and “outside” and “play,” rather than some of the more random buttons that some people seem to have chosen for their dogs, like “friend” and, you know, almost more philosophical concepts.
Rath: There’s this wonderful moment in the film which kind of shows how hard things are for the researchers, where one of these kind of genius dogs puts together “water” and “outside,” and they take that as “beach.” But, then the researcher explains that, once you take the dog to the beach after that, you’ve reinforced it in a training way, and you can never actually know then if it was random or not the first time.
Bootle: Exactly. That just shows why it is so hard, because ultimately, whatever kind of studies you try to carry out, there’s no way to then ask the dog, “So when I did that, when you press those buttons, just what exactly did you mean? Did you mean this?” You can’t analyze, you can’t ask the dogs, obviously. They’re sadly not good enough at using buttons yet for that. But yeah, like you say, as soon as the dogs have pressed any set of buttons that you may think is interesting, you can’t then go back and study their use of those buttons. Because, by saying something, by rewarding a dog for pressing particular buttons, you’re persuading that dog that now those buttons mean something. So, it is incredibly hard to study, but they are devising experiments to find out as much as they can. There was the impossible task, which Federico’s colleague, Amalia Bastos, ran, which I think is really interesting. Again, it’s not fully conclusive, but it’s really interesting when you see dogs using buttons in ways that they can’t have been trained to use them.
