A new global study allows dogs to ‘talk’ to their owners by pressing buttons that say human words

Researcher Jack Terwilliger looks at a video feed in his lab — a cozy room with a couple of tables and computers. He’s not just watching animal videos for amusement but examining a unique method to determine language patterns in dogs and other animals.

He’s part of a global study that originated at the University of California San Diego’s Department of Cognitive Science and that has now extended to 47 different countries around the world.

“Berlin, New York, Chicago, Phoenix,” he pointed to the different videos on the screen. “I think that might be Baja [California], Paris, Edmonton, Melbourne, so, all over.”

Since the COVID-19 pandemic began — a time when many crazes emerged out of mass boredom — owners have been training their dogs to “tell” them what they want by mashing buttons with their paws to express words in various human languages.

A vast spreadsheet records all the dogs in the study. Data points show where in the world they are, the date and time when they press a certain button and the word on that button.

Some of Terwilliger’s research shows animals on TikTok hitting buttons that play words like “outside,” “food” and “treats.”

The purpose of the global study is to determine whether the animals are actually communicating or if they’re just forming behaviors to secure treats for themselves.

The talking dog buttons, as they’re called, are scientifically known as augmentative interspecies communication (AIC).

“I think we’ve sold over 2 million of these buttons at this point,” said Leo Trottier, a PhD graduate from UCSD and the founder and CEO of FluentPet, the company that makes the button devices.

“We created a community of people where they could find common ground and they could see that they weren’t the only ones doing this and they could relay their incredible stories — which 99% of their friends didn’t believe [at the time] — but which were very evident to the people who they were meeting with on Zoom,” Trottier said.

One video that went viral online showed Christina Hunger’s dog, Stella, a Blue Heeler-Catahoula mix.

One afternoon, around the time for her daily trip to the beach in San Diego, Stella started mashing several buttons on her device.

She first pawed at the void where the missing button for “beach” would normally be and then went on to push the buttons for “help,” “water” and “outside” in succession — essentially creating a phrase that summarized her wish to go to the beach.

“She was able to generalize the concept of water, from being just, ‘Oh, water in my bowl’ to water as a more general thing that can exist at the beach and be much larger than four inches in diameter,” Trottier explained. 

The excitement this video garnered online was a starting point for gathering more people from around the world to participate in wider scientific research. Trottier said that thousands of people signed up in just a few weeks.

“I was like, well, we’re a small company, a tiny company,” he said, “but I believed that no self-respecting cognitive scientist would turn down this opportunity to understand non-human minds in a way that had never been understood before.”

One such person was Federico Rossano, an associate professor in the cognitive science department at UCSD.

Since 2021, Rossano had been researching non-human communication — primates and dogs — but by answering questions that scientific reasoning sent his way, rather than from viral online trends.

He admitted that he initially didn’t see the value in the button-pressing type of study, thinking the trend would soon pass, until he realized that people were embracing it anyway, and in large numbers.

“It’s now 10,000 participants from 47 countries, currently the largest animal communication studies ever conducted longitudinally, and we have people signing up weekly,” Rossano said.

He explained that, to determine if the dogs were interacting as communicative partners or treating their owners as vending machines, “we tried to create a situation where, basically, you present them with a container that has a treat, and they really want the treat, and they try to open it, but they can’t.”

The owners were instructed not to respond to attempts to seek attention through normal dog behaviors, like barking or scratching their legs.

In one video from his own research study, Rossano saw a dog named Parker give up on trying to get a treat out from under a clear plastic box. After being ignored by its owner, Parker went over to the soundboards and pressed the button sequence, “Help, look, look.”

“At the very least, this tells me that they understand what those buttons mean because they could have pressed anything else and that they are capable of understanding that the human needs to perceive what’s going on in order to respond and to communicate about what they need, which I think is amazing,” Rossano said. “It’s like a child telling you that they’re hungry versus just crying. And I think being able to communicate your needs and what you want, I think it’s very powerful.”

In France, the button systems have words like jouer, which means “play” in French, or dehors, “outside.”

Paris resident Héloïse Chevalier and her dog, Tantrum, have also been part of the UCSD study since she adopted him just over a year ago. Her cat, Flint, has been in a similar study for four years. Chevalier posts animal training videos on her Instagram account, where she has more than 21,000 followers.

“Flint tolerates Tantrum and Tantum thinks he’s his best friend,” she said.

Chevalier explained that she already believes animals can communicate with humans by combining separate concepts to build phrases or sentences, and that she joined the study just to confirm the theory.

Leo Trottier of FluentPet even believes that dogs are capable of putting together more than two concepts — three, four or more.

“Previously, humans have thought language was maybe something akin to an inborn sense or something like an organ, like an eyeball or the nose, we have this language faculty, like we have an opposable thumb,” Trottier said.

In fact, Trottier said, language might be more like a bicycle, a collectively evolved artifact that has been iterated on for many, many, many hundreds of thousands of years.

“And if it’s an artifact that’s external, then the fact that a dog can’t ride a bicycle is actually not that interesting,” he said.

“The fact that it’s been difficult for non-human animals to learn what we call human language, the fact that that hasn’t happened, it might be for very contingent reasons that, with design, could potentially be overcome.”