Future Tense

Can That Dog on Instagram Really “Talk”?

Stella’s owner says she can push buttons to say words like play, outside, and eat. But what’s really going on?

A dog with a speech bubble that says "Hey girl!"
Photo illustration by Slate. Photo by Getty Images Plus.

Any pet owner knows that her pet is the best one. There’s real joy in bonding with an animal, a being that loves you unconditionally and is unburdened by bills and career woes and impeachment hearings. Sure, you know that when the dog looks at your steak and lifts a paw, he wants a piece, but you want a deeper way to connect. If only there were some way to really talk to our pets, to know what they really want, and for them to know just how much you love them.

Enter Stella, an adorable, chocolate-colored catahoula and blue heeler mix. Her person, Christina Hunger, is a speech language pathologist “who discovered how dogs can talk,” according to the Instagram page she’s made for Stella. Her followers more than doubled in the week after People magazine ran a Nov. 4 profile on Stella. Nearly 500,000 people are watching Hunger’s posts, which show short clips of Stella pushing buttons that play recordings of words like play, outside, and eat. In each post, Hunger translates Stella’s button pushing: In one of her earliest posts, made in July, Stella pushes outside and Stella, which Hunger says in her post is an example of “how Stella combines words to tell us what she wants!” In more recent posts, Hunger is excited by Stella’s progress; she says Stella has combined good and bye to say goodbye and that Later Jake (Jake is Hunger’s partner), in response to him doing a chore, meant do that later.

I’m no dog expert, so all I can say is that Stella is a very good girl and she deserves all the treats, belly rubs, and outside time she desires. As a former cognitive scientist, though, I wondered what’s actually happening in these videos. There’s a long and storied history of animal enthusiasts and researchers trying to teach language to animals—in particular, to our cousins in the primate family, who, scientists reasoned, are most likely to share our language abilities. People have spent decades training chimpanzees, bonobos, and gorillas to “speak” using sign language or soundboards similar to Stella’s buttons, and while primates have demonstrated an impressive ability to sign words or press soundboard buttons, researchers have pointed out flaws in those animals’ trainers’ extraordinary claims. Penny Patterson, handler of famed gorilla Koko, would often “interpret” for Koko. For instance, she once explained that Koko didn’t really mean to sign nipple during a public event, but instead signed it to represent something that sounds similar, like the word people.

On the one hand, yes, any person who spends a lot of time around an animal learns to pick up on their signals. My dog has learned to paw at blankets to tell me she wants to get underneath them and to offer a high-five in exchange for whatever delicious thing I’m eating. But we necessarily project a bit of ourselves into interpreting what another being wants, especially when we don’t speak the same language. (Even when talking to other humans, we often hear what we want to instead of what the person is really saying.)

I’m impressed that Stella’s learned to press buttons, and I’d guess that, more than likely, Stella even knows what some of the words mean, especially the ones that might result in an exciting reward, like ball or walk. But Hunger makes some bigger claims worth digging into because they would be, in clickbait parlance, BIG IF TRUE. Dog cognition as an area of study has steadily grown over the past decade, probing dogs’ self-awareness, problem-solving abilities, and attention to humans’ communicative cues, like pointing or eye contact. We’re still learning about what dogs know and what they’re capable of, and it would be news if, as Hunger claims in one post, a dog could plan future events and express those desires, like wanting first to eat and then play afterward. I wanted to learn more about how Hunger trained the dog and how she selects videos to post, but she did not respond to my emails. Instead, I enlisted a couple of dog cognition researchers to help me unpack the themes in Stella’s posts.

Brian Hare, an evolutionary anthropologist at Duke University and founder of dog intelligence company Dognition, agrees with me that the Stella project is endearing and fun. “I think it’s wonderful [Hunger] is excited and she’s trained her dog, and that she wants to communicate with Stella better,” he says. Alexandra Horowitz, a psychology professor at Barnard College and author of several bestselling dog cognition books, echoes that sentiment: “What I see first is a person want to understand her dog’s intents and wishes, which I wholeheartedly endorse,” she says.*

Hunger’s observations of Stella are interesting in their own right, and if some of her claims were to pan out, it wouldn’t be the first time an enthusiastic dog owner uncovered something new about dog cognition. Hare points out that Rico, a border collie who was famous for learning more than 200 words, was discovered by researchers after his owner called them with his findings. “It’s not unprecedented that people with their pets at home contribute in real, important ways to our understanding of animals,” he says.

But caution is warranted, too. While Hunger often describes her interactions with Stella as “talking” or “chatting”—in one post, Hunger even claims that “I just had a conversation with my dog about having an açaí bowl”—there’s an important distinction between Stella’s button pushing and actually understanding language in the way that humans do. “The tricky bit is that we humans are easily fooled by the sound of words, and we think that by using a word, one must know the meaning behind it, and intend to use that word, just in the way humans do,” says Horowitz. “That’s definitely not always the case. A lyre bird mimicking someone yelling ‘timber’ because they’ve heard someone yelling ‘timber’ is not themselves meaning to say ‘timber!’ ” Stella pressing the bye and outside buttons in response to Hunger’s question about going out to get açaí bowls doesn’t mean she actually understood the full gist of that sentence.

We read into what she means by pushing buttons, and her “vocabulary”—and hence, our interpretation—is limited by what buttons are available. Horowitz gave an example of this: “What if the button said on Mars instead of later. Would we think that Stella meant it’ll happen on Mars?”

When evaluating animals’ behavior, Hare says it’s important to think about the myriad ways it can be explained. British psychologist Lloyd Morgan came up with a rule that came to be known as Morgan’s canon. As Hare explains it, it’s “Occam’s razor, but for behavior—any time you see a behavior, it doesn’t mean it’s underlain by the same psychology when it’s performed by two different organisms. You prefer the simplest explanation until you can rule it out.” In Stella’s case, a simpler explanation might be that she’s pressing buttons because she knows something good might happen, or even that she’s reading cues from her people and pressing the “right” buttons. (Think Clever Hans, the horse that could “do math” but was really just very skilled at reading its owner’s facial expressions.) It’s “likely [that] the dog is intentionally communicating and it knows if it interacts with these buttons then good things happen. Maybe it even has some idea about what type of good thing might happen,” says Hare. But we need more rigorous evidence to determine what, exactly, Stella knows.

It’s impossible to draw any conclusions from Hunger’s videos, though the captions certainly nudge us toward what Hunger believes is happening. In one recent video, Stella presses Jake (the name of Hunger’s partner) and outside—Hunger says this occurred after Stella heard Jake’s car beep. My dog, too, frequently stands on the couch and keeps watch out the window when she hears my partner coming home, so I want to believe this one. But there are many things to consider that we can’t see in this video, and without the full context of their interactions, it’s hard to gauge what Stella “means” here, if she means anything at all. Did Stella press other buttons around the same time, and if so, what were they? In other words, is there any reason these two button presses were particularly meaningful or indicative of her desires? It seems like Stella spends a lot of time saying outside. (In fact, in another post, Hunger talks about the many different words Stella combines with outside.) What, if any, implicit or explicit cues might she be reading for guidance on which buttons to press?

And in the bigger picture, Horowitz points out, the videos don’t give us context about Stella’s usual behavior that could reveal patterns in her button pushing. How might her behavior change if provided with a different set of buttons? What buttons is she pressing that don’t make it into videos? (I noticed a few buttons in Stella’s soundboard that I haven’t seen used in any videos: kennel, mad, and water.)

Hare suggests a few ways to test this more rigorously. One technique psychology researchers frequently use to gauge consistency in studies is called interrater reliability, in which multiple people review the same evidence and their ratings are compared. If raters were to evaluate Stella’s videos for an actual study, they might note things like whether Stella pressed a button, what order she pressed them (and how many times), and, perhaps, most importantly, if, given the context, they would interpret her meaning in the same way. And in looking through Stella’s feed, there are a lot of things that would be opaque to an outsider: Stella Stella and Help water outside are pretty inscrutable without Hunger’s explanations—and those explanations are exactly what makes the account compelling. But as lovely as Hunger’s stories are, they’re not scientific evidence.

In the end, Stella’s videos are less about her than they are about us. “A lot of people strongly believe or want to believe that dogs are highly intelligent, and they are, but not necessarily for the reasons that Stella seems to be exhibiting,” says Hare. People love science-y “discoveries” as long as you confirm what they believe, he says, but once you present evidence that challenges that? “You’re absolutely the turd in the punch bowl,” he says.

Plus, Horowitz points out, we want them to communicate with us in the way we know best. “It is an ongoing fantasy that they a) understand us, and b) could reply back or initiate conversations,” she says. “Though dogs are talking to us all the time—with their body postures, their expressions, their behaviors—we humans are still most compelled by the sound of our own voices.”

Correction, Nov. 25, 2019: This article originally misidentified Barnard College as Barnard University.

Future Tense is a partnership of Slate, New America, and Arizona State University that examines emerging technologies, public policy, and society.