As soon as I got Ellie, my black-eyed, bat-eared German shepherd puppy, I trained her to be a good dog. And so she was. When I spoke, she listened; she kept her barks quiet and her teeth clean. Two years on, I started to think she was a little too obedient. There was something pitiful about the way that, even unleashed in the park, she would stop at a fork in the path and look to me to point the way. In the house, she would pause behind a half-closed door instead of pushing through. She was hesitant, whining when she was unsure of herself, in a way that clashed with her big muscles and pointy canines. One day, when the cat in our local bodega poked its head out of the store, Ellie squealed like a child.
Her training, I saw, had come at the cost of something precious. Her independence, certainly. But also, something more intrinsic — something like her animality.
Ellie’s dad — my partner, Jesse — agreed. “I think she needs more self-confidence,” he said.
The solution, maybe, was buttons. Around this time, I started to see dogs on social media seeming to express their desires by the most absurdly simple, low-tech means possible: stepping on multicolored plastic buttons on the floor, each disc emitting a word when the dog pressed it. I scrolled through videos on my phone of dogs pawing FOOD and MORE and NOW, sometimes in that order.
The most famous was Bunny, a lanky sheepadoodle in Tacoma, Wash., with 8.6 million TikTok followers, a vocabulary of 105 buttons and a diagnosis of generalized anxiety disorder. Bunny’s online store sold a starter pack of six buttons for $65. Through the buttons, Bunny reported her experience of the world. She said LOOK CAT. She was frequently CONCERNED. She seemed to have restless dreams about a STRANGER ANIMAL. A favorite topic of conversation was poop, as in POOP PLAY.
Bunny was advanced in her use of the buttons, a star student. But there was also an endless video stream of other dogs — also cats, pigs, horses and cows — making equally clever and entertaining statements. I would open the apps and see a mini Aussiedoodle with a taste for dirt demanding to go OUTSIDE for a SNACK. A French cat being denied a treat and complaining, with apparent disdain, PAS CONTENT.
These pets weren’t just standing by to serve their human owners. They were companions with voices of their own. I looked over at Ellie, lying in the far corner of the room. Her eyes were dark, her feelings mysterious. What was buried inside that impenetrable dog skull? If I could teach her to use these buttons, she would tell me. Or so I imagined.
Ellie’s first word, I decided, would be OUTSIDE. “Outside” was one of the recommended first words in the informal pedagogy of dog-button learning, created by a speech-language pathologist named Christina Hunger. In 2018, when Hunger was raising her Blue-Heeler Catahoula puppy Stella, she noticed that Stella was progressing through the early stages of communication much as the young toddlers she worked with did. Hunger taught children to speak by tapping icons on a tablet. She wondered: Could Stella learn words through a similar method?
To test this hypothesis, Hunger bought a set of recordable buttons that, when pressed, would play back her voice saying simple words — OUTSIDE, WATER, PLAY — and fastened them to a board on the floor. Every time she talked to Stella, she pressed the corresponding button. About a month later, Stella caught on, pressing OUTSIDE to use the bathroom in the yard, and PLAY to request playtime. After a few months, she pressed WATER when Christina was watering plants. Stella was narrating what she saw, Hunger thought. Hunger expanded Stella’s board, adding emotions like MAD and social words like BYE.
When Stella started putting multiple words together, Hunger wasn’t too surprised. Dogs are about as smart as a 2½-year-old human, and Hunger knew that toddlers of that age typically compose sentences of two or three words. After a year, Stella was saying BED LATER and WANT OUTSIDE NOW. One day, the end of daylight saving time delayed Stella’s mealtime. She asked for food, and Hunger told her to wait. In protest, she stomped over to her buttons and pressed LOVE YOU NO.
Hunger’s blog of Stella’s progress went viral, and other dog owners started experimenting with the buttons. Those accounts themselves went viral. A movement grew. The dogs who used the buttons had some things in common. First, their owners spent a lot of time with them, talking to them, looking at them, pressing buttons with them. Second, these owners were often women, with no children in the house. As for the dogs themselves, many had bossy personalities. They had things they needed you to know. A pit bull named Tilda pressed SOUND twice every time her owner joined a weekly game of Dungeons and Dragons. Should I wear headphones? the owner wondered. A golden retriever named Cache pressed WORRIED when his owner turned on the stove to cook dinner, because he wasn’t a fan of all the sizzling and popping.
“When we actually can hear firsthand how an animal is experiencing the world, it profoundly changes how we treat them and how we view them,” Hunger told me. “Both my husband and I really view Stella as an equal part of our family. Her opinion matters, her feelings matter, her thoughts matter.”
Of all the dogs I saw online, Bunny, the sheepadoodle, had the strongest opinions and the biggest feelings. She hated feet. She hated water. She hated birds. She was known for her outspoken nature and philosophical musings. I DOG, she remarked. DOG WHY. Scrolling through her videos, I paused on an exchange between Bunny and her owner, Alexis Devine. MAD, Bunny said, and looked at Devine. “Why are you mad?” Devine asked. OUCH, Bunny said. “Where is your ouch?” Devine asked. Bunny pressed STRANGER and pawed the side of her head. “In your ear?” Devine asked. “Where stranger?” Bunny appeared to reflect. PAW, she said, and came over to Devine. She held out her shaggy left paw, and the video cuts to Devine holding a foxtail thorn that she explains was stuck between her toes.
Now, people online are unkind. Bunny’s posts incite fervent ridicule as much as they inspire and amaze. Detractors accuse the videos of being faked, selectively edited, cherry-picked. The more scientifically minded skeptics bring up Clever Hans, the horse whose owner toured Germany at the turn of the 20th century to prove he could do mathematical calculations and ended up showing that animals can be very good at picking up on unconscious cues.
Everyone knows that while dogs recognize human words, they are not supposed to be able to use words themselves. But for the pet owners, pressing buttons with dogs is not so different from communicating with a small child. As Devine says, “What’s not to believe?”
Since at least the Victorian age, when we first began inviting birds, cats and dogs into our homes en masse to live alongside us, people have endeavored to talk to their pets. Charles Darwin, a lifelong dog owner, drew on observations of his own dogs to develop some of his ideas about the natural world. Darwin was modern science’s first great believer in animal minds. He thought that animals, especially primates and some other mammals, had many of the same emotions and mental abilities as humans, only to a different degree. Primates, he wrote in “The Descent of Man,” feel “jealousy, suspicion, emulation, gratitude and magnanimity, they practice deceit and are revengeful, they are sometimes susceptible to ridicule, and even have a sense of humor.”
Darwin’s ideas inspired many Victorians to wonder whether their pets might be taught human languages and vice versa, setting off a long period of preposterous at-home experiments. Darwin’s neighbor John Lubbock was among the first Victorian naturalists to perform language experiments on an animal. He taught a terrier puppy named Van how to associate cards imprinted with words like “bone” or “tea” with the objects they named. To ask for a bone, Van would select the card and bring it to Lubbock in his mouth.
Around the time of Lubbock’s experiments with Van, a New York publisher produced a phonetic dictionary of some 600 cat sounds compiled by a Frenchman named Alphonse Leon Grimaldi. The transcriptions included everyday vocabulary — “Lae” for milk, “Ptlee-bl” for mouse meat, “Mieouw” for here — as well as complex concepts like “zuluaim” for millionaire. In India, meanwhile, the explorer and writer Sir Richard Burton brought 40 monkeys of various species into his house and gave each a name, including one he referred to as his wife who wore pearl earrings and dined at his side in a highchair. Burton recorded about 60 monkey sounds that he claimed to be able to understand and speak. In London, a 20-year-old Alexander Graham Bell tried out some of his lifelong language experiments on the family terrier, Trouve. He would shape Trouve’s lips with his hands while prompting her to growl, producing sentences like “How are you, Grandmama?”
By the 1960s, researchers understood how naïvely anthropomorphic those Victorian exercises had been, but communication experiments with animals still promised to settle a great linguistic debate: Did language belong to humans alone or did nonhumans also have some capacity for it? As a challenge to Noam Chomsky’s assertion that such a thing was impossible, a baby chimp was christened “Nim Chimpsky” and sent to live with a human family in Manhattan to learn sign language. Novel and ambitious experiments with primates, dolphins and parrots proliferated, many of them documenting animals following specific cues and using words and phrases to communicate with their human trainers.
Some of these studies had serious flaws in their methodology, however, and in many cases their treatment of the animals would be considered unethical today. Project Nim, the flagship study of this era, enlisted a rotating cast of student volunteers to train Nim in American Sign Language, all while young Nim was disruptively shuttled among homes and caregivers. Eventually the lapses in these studies began to overshadow the findings. In 1980, a New York Academy of Science conference was held with the mocking purpose of declaring animal-language endeavors a failure and the scientists frauds. The excitement, and funding, flamed out as spectacularly as it began. Nim finished his life in a cage at an animal refuge. Animal-language studies acquired an unshakable odor of malfeasance and scientific malpractice.
In 2019, Federico Rossano, an associate professor of cognitive science at the University of California, San Diego, got an email from a colleague suggesting someone they might invite to give a talk to the department. In it was a news article about Christina Hunger and Stella. “Mixed feelings about this,” Rossano replied, attaching a book chapter he used in his classes titled “What did we learn from the ape language studies?” The answer: not a lot.
But by the following summer, the pandemic had halted Rossano’s research on how chimps communicate to share food and how human children make requests. Leo Trottier — a former cognitive-science Ph.D. student at U.C.S.D. who left the program to found a company that makes game consoles for dogs, and recently started the FluentPet recordable-button brand with Devine — reached out to pitch a study about the buttons. He said he knew of more than a thousand pet owners who would be willing to participate. Despite his initial hesitation, Rossano saw that this was a unique opportunity to circumvent Covid restrictions and conduct a citizen-science study from the participants’ homes. Because the study would not rely on a single animal trained by researchers in a lab but rather on thousands of animals living their normal lives, it was less likely to suffer from the problems and ethical lapses that afflicted past animal-language studies.
The field of animal behavior, also known as ethology, had itself transformed since 1980, when the term “animal cognition” was barely used. It is now known that primates, despite being our closest genetic relatives, are not all that good at understanding humans, while dogs (and cats) get us more easily. Research suggests that dogs, which have co-evolved with humans over the past 18,000 to 32,000 years, might in fact be the ideal species with which to study two-way communication. They were the first domesticated animals and very likely the first that listened to us talk. When dogs hear human speech, their brains light up with interest. They look where we point. They recognize our facial expressions. They mirror our anxiety. In the buttons, Rossano saw an opportunity to learn what else they could do. But he was cautious. “I did not want to destroy my academic career, to be very honest,” he told me.
The study that Rossano and his team began in 2020 is the largest animal-communication study ever tried, with 10,000 dogs and cats in almost 50 countries around the world. It has proved hugely popular with the public and much less so with other scientists.
When I talked with Rossano in July, he had a beleaguered air. His second paper on the buttons was about to come out, but the path to publication had been slow and onerous. For the past two years, his papers were consistently rejected or not even considered for review. It felt like gatekeeping. Earlier that day another scientist asked him, “Are you the person responsible for my wife wasting my time with clips of dogs?” He was looking forward to going dark for three weeks in Kenya, where he would follow a troop of about 150 baboons on foot across the savanna to observe their collective decision-making.
Rossano reminded me that he had started at a place of skepticism. “I definitely have a lot of issues with the social media part of this thing,” he said. “But there are, right now, several thousand people who are doing this to their pets, and I think we should try to take it seriously. Is this good? Is it bad? What would it tell us about these animals’ minds?”
The data that Rossano collects is entirely independent of any social media posts. (“I don’t even have TikTok,” he told me.) He screens for the Clever Hans effect by using double-blind testing and controls for unintentional cherry-picking on the part of pet owners by setting up cameras running 24-7 in some homes and sending in his own researchers into other homes a few times a year. He has found that the dogs, on average, recognize certain common words and press certain two-word phrases nonrandomly, without only mimicking their owners’ presses. This indicates that the dogs aren’t just mindlessly slamming buttons. But the tests do not yet show that dogs are doing anything scientists didn’t already know they could do. More than a decade ago, studies with two Border collies named Rico and Chaser demonstrated that some dogs can learn a remarkable number of words — more than 1,000, in the case of Chaser, who also understood semantic instructions like “To ball take Frisbee.” Hungarian studies involving so-called gifted-word-learner dogs have found a few dozen dogs from around the world who can learn the names of hundreds of objects even without training — and can remember these words years later.
These are remarkable feats of cognition, but they mostly rely on the familiar mechanisms of associative learning first established a century ago by scientists like Ivan Pavlov. Press a button; get a reward. The primary claim of the button users — that dogs are producing original combinations of buttons they haven’t been trained to press, to communicate something they couldn’t otherwise say — is entirely different.
When a dog named Parker on TikTok sees an ambulance and presses SQUEAKER then CAR, she appears to be using words in a way that’s closer to how humans do. Multiword phrases like Parker’s are creative, spontaneous and flexible expressions of language, a quality that the linguist Charles Hockett called productivity. Productivity is thought to be unique to human language. It’s what the most famous animal-language studies appeared to show nonhuman minds were capable of: anecdotally documenting, for example, that Washoe, a chimp, signed “water bird” to refer to a swan; that Koko the gorilla signed “finger bracelet” to mean a ring; and that the orangutan Chantek called himself “orangutan person” to differentiate himself from ordinary, nonsigning orangutans.
Herbert Terrace, who led the Nim study, is now a critic of such anecdotes (his latest book is titled “Why Chimpanzees Can’t Learn Language and Only Humans Can”). He maintains that the animals in these studies were merely signing things in random sequence in the hopes of a reward. None of the animals got anywhere close to full grammatical sentences. The longest sequence Nim ever signed, after years of confinement and training, was “Give orange me give eat orange me eat orange give me eat orange give me you.” That, Terrace said, is not language.
But not all scientists share Terrace’s skepticism. “We don’t know if it’s real, we don’t know if it’s stupid, unless we test it,” says Irene Pepperberg, a research professor at Boston University. Pepperberg, now sometimes called the “mother of avian cognition,” is well known in the field for her 30-year-long communication study with Alex, an African gray parrot. Pepperberg demonstrated that Alex asked questions, performed simple addition and, in a few instances, coined neologisms. Partly because of Pepperberg’s work, parrots are now widely recognized to have an intelligence roughly comparable to that of a 5-year-old human child. Of Rossano’s study, she said: “I’d like to see data. I’d like to see what the animals are capable of. All these studies that we’re doing in terms of looking at other animals’ abilities, it’s opening up a world to us.”
Rossano’s next research questions will look at multiple-word phrases, the ability to talk about things that aren’t present and emotions. He plans to take biometric measurements to determine whether a dog is physiologically calm when it presses HAPPY and stressed when it presses WORRIED. When it presses BALL, does it mean one specific ball or any object in the abstract category of spherical toys? In the distant future, he imagines the possibility of testing episodic memory: Does your dog remember the time you took them camping?
Rossano’s most advanced inquiries will rely on the top-performing dogs in his study, the genius dogs: dogs like Bunny and Parker. Most dogs — like most humans — are just mediocre. Out of 10,000 dogs, Rossano estimates, about 65 use more than a hundred buttons. The median is nine.
Ellie was turning out to be an average dog, button-wise. Her technique was basic but effective. In the afternoon she would rouse herself from a nap and smash PLAY over and over until I got up from my computer. She went through a period of insisting on more FOOD after dinner; I assumed she was being greedy until I finally weighed her and found she’d lost six pounds. We moved apartments, and she started pressing OUTSIDE, a button she rarely used before. I realized she wanted to sit on the stoop, a feature of our new place. During a thunderstorm late one night, Ellie got up, nervously whining, and paced the living room. She pressed each button in turn, again and again, unleashing a chaotic babble of tinny button words. She was frantically trying to tell us — what? It was intolerable. Jesse got up and threw the buttons under the table.
Rossano sent me a list of his colleagues in dog-cognition research and where they fell on a spectrum from “thinks we are OK” to “hates buttons.” On the “hates buttons” end is Alexandra Horowitz, the head of a dog-cognition lab at Barnard College and probably the most publicly known scientific voice on the subject of dogs. In the first chaotic weeks after adopting Ellie, I read her best-selling book, “Inside of a Dog,” in which she warns against assumptions like thinking your dog’s kisses mean he loves you.
I called Horowitz and described my experiments with Ellie. “To me, that’s not great,” Horowitz said. “Dogs already do so much to accommodate our lives. They’re on our schedule. They have to ask us if they want to urinate. They socialize on our schedule. They walk where we want to walk on a leash. Ostensibly, the interest in having dogs is that they’re another species. There’s something unknown about them, and that’s wonderful. Why do we lean into forcing them to wear clothes and speak our language?”
I told Horowitz that Cache, the golden retriever, repeatedly pressed BACKYARD WATER until his owner figured out that he might be asking to swim in the river behind the house. When Cache was let outside, he ran right to the water. “It’s not scientific,” Horowitz said. It’s possible, she went on, that the dog was requesting to swim. Equally possible, though, was that the dog pressed some buttons and then, unrelatedly, was happy to swim. Neither possibility could be proved false. “If the buttons said ROCKET MOON and the dog pushed ROCKET MOON, most people would say: ‘No, no, that’s insane. My dog is not wishing they could take a rocket to the moon,’” Horowitz said.
Within the dog-button community, Horowitz is regarded as a killjoy. Another dog researcher in the same camp is Clive Wynne, a behavioral scientist at Arizona State University. Wynne’s theory about why dogs occasionally seem to express complex and creative phrases is that they get the most enthusiastic reception from their owners when they press a lot of different buttons and that the owners pay attention only when these sequences make sense.
As a gesture of good will, Devine sent Wynne a button as a gift for his dog. Wynne recorded “My name is Xephos” into the button, and he taught Xephos to press it. “So she said, with my voice, ‘MY NAME IS XEPHOS,’” Wynne told me. “She enjoyed it, and we enjoyed it. I think it’s all good clean fun, but I don’t think we learn anything about dogs by doing it.”
For Wynne, the buttons are not only entirely superfluous to the communication humans and dogs already have but also potentially a distraction from their natural barks and howls, their tail wags and tucks. “I actually think that having dogs press buttons runs the risk of obscuring dogs’ voices in the sense of dogs’ expressions of themselves,” Wynne said. Rossano disagrees with this notion. “It doesn’t mean that they lose their first language,” he said. “They’re still dogs. They still do their thing, and they still bark at people, and they still scratch the door, and they still try to get toys. It’s just that, in addition, they have another way to communicate, and sometimes they choose that way to communicate, which I think is interesting to study.”
Pet owners will tell you that their dogs love them unrequitedly, that they feel guilty, that they make jokes. According to the science, though, dogs don’t necessarily do any of these things. They may whine when you leave the house, but separation anxiety isn’t the same as missing you. They can tell when a human is sad, but scientifically speaking, that doesn’t mean they care. What’s undeniable, though, is the emotional connection that humans feel with our pets. Maybe our language-loving brain needs that bond to manifest in words in order to feel real. Juliane Kaminski, a comparative psychologist who led the Rico study, shared one of her more recent findings with me: Dogs have a muscle around their eyes for the sole purpose of making big, sad puppy-dog faces. Wolves do not have this muscle. “This eyebrow movement most likely has no meaning for the dogs,” Kaminski told me. But it may have meaning for humans: It makes us want to take care of the poor things. When humans and dogs look each other in the eye, each species produces oxytocin, the same hormone that bonds mothers to their children.
Kaminski believes that dogs have evolved into creatures we treat like our own children and that the buttons feed into this impulse. “It’s this trap we are falling into,” she said. “We evolutionarily created this niche for infants that we now have designed this other species to fit right into. People can’t help themselves. They see the dog pressing the I LOVE YOU button, and they’re like, ‘We’re having a conversation.’”
In August, I drove up an unpaved road about three hours east of Tacoma to visit Alexis Devine and Bunny. Devine and Leo Trottier have sold more than two million buttons. Earlier this year, she moved into a stylish two-story cabin on a 55-acre property to give her dogs “a life with more agency.” It was fire season, and the hills around the cabin were covered in parched yellow brush. An elk moved on the opposite slope. Mount Rainier, wreathed in snow, loomed in the distance. As soon as I parked, Otter and Tenrec, Devine’s second and third dogs, bounded to the car. Devine kept Bunny on a leash, hanging back a safe distance.
Since moving to the cabin, Bunny’s chronic anxiety and gastrointestinal issues had calmed. She was less fearful. Upon seeing wildfire smoke around the cabin earlier that year, she pressed FIRE but not CONCERNED. Her temper was better. She’d been playing with Otter rather than going UGH about him. About Tenrec, a tiny papillon who looked less like a dog than some kind of plushy gremlin, she was mostly confused. “She thinks he’s a cat; she’s said that several times,” Devine said.
Devine got Bunny as an 8-week-old puppy. “I wanted a different complexity of relationship you can only get with a dog,” she said. She chose a sheepadoodle because she wanted a smart and active dog, one that could offer “a real connection.” At the time, she had a successful career as a wearable-art designer. When the pandemic began, her fashion shows were canceled. More time for Bunny. “She was my next big project,” Devine said. “I knew before I brought her home that I wanted to teach her to talk.”
Devine spent those months making jewelry at the coffee table in her living room, Bunny and the buttons spread out right in front of her. She demonstrated the buttons constantly to her new puppy, and Bunny pressed the buttons spontaneously all day long.
“There was this need for her to feel like she had some control over her environment,” Devine said. “And I believe that the buttons were a way for her to do that. She could ask me what was going on outside. She could hear a sound and then express an emotion about it. She could request specific places she wanted to go. This is all hindsight. But she was telling me all along how I could help her be in a better frame of mind.”
The first words — OUTSIDE, BALL, MOM, DAD — were easy. As Bunny’s vocabulary grew, Devine modeled more abstract concepts, like YESTERDAY, by narrating that Bunny had, the previous day, GO SEE FRIEND. A monthslong period when Devine was getting tattoos done was an opportunity for MOM OUCH. Bunny picked it up easily. One day she pressed OUCH OUCH EAR HELP, and Devine understood that she had an earache and flushed her ear with medication.
Through her social accounts, Devine met other dog owners teaching their dogs to use buttons, and a couple dozen now gather in person every year. Bunny had changed Devine’s life entirely. Devine picked up a new career and moved house. She got an autism diagnosis. She started to question the entire logic of pet ownership. “It’s just not cool, man, that we have these captive animals,” Devine said. “Like, it’s weird. Animal ownership is a very strange thing to me now.”
In our conversations, Devine didn’t evangelize for the buttons. She believed that they were a good way to understand her dogs and that they allowed her dogs to advocate for themselves. “My overarching goal is for that dynamic to exist with or without buttons,” she said. In the past several months, Bunny had been using the buttons less, Devine said. She was more relaxed, less needy. She didn’t need to speak out so much.
Whether or not Bunny was a genius dog, she carried a remarkable presence. Her shaggy head followed me around the cabin. Her eyes were a soulful, piercing hazel that seemed to see me exactly. Her enormous paws flopped around when she walked. She sat next to me on the floor of the cabin, watching, while I drank a seltzer.
I asked if I could talk to Bunny with the buttons. We went over to the board, and Alexis stood by the buttons. Bunny sat and looked up at her expectantly. Devine switched into a button voice that I recognized from her videos. “You want talk?” she said to Bunny. “You want talk friend?”
HI BUNNY, I pressed, and waited as Tenrec trampled the mat, releasing a cascade of words.
“Bunny, you want talk new friend?” Devine said. “You want talk?” She pressed WANT, TALK, HMM? in emphatic staccato. Bunny looked at her.
“What do you want to say? Do you want to say hi?” Pause. “New friend love you.” She pressed FRIEND. Pause. “Do you want talk?”
Bunny pawed at Devine. YES, Bunny pressed.
“Yes, what?” Devine said.
HI BUNNY, I pressed. I FRIEND.
“How about small talk?” Devine said. She pressed SMALL TALK. “Remember when I said friend come later? Now friend is here.”
Bunny seemed to recognize something was expected of her. She pawed at Devine. “Buhbuh, I know it’s hard, but could you use your words for a moment, please?” Devine said. Bunny came over and licked my face.
We tried some more, but Bunny didn’t want to talk to me; she just wanted to nap. I didn’t mind. The cabin was peaceful, with the late-afternoon sun streaming in through the windows. It was more pleasant to sit there with a dog in the loose quiet, without buttons going off. I sat with Bunny while she dozed and threw a ball for a bit with Otter and Tenrec. When I got in the car to leave, the three dogs lined up on the porch and watched me drive off.
The next day I got a message from Devine. “This morning we were talking about you,” Devine wrote, “and Bunny said BYE STRANGER FRIEND.”
Camille Bromley is a writer and editor who has worked at Wired, The Believer and Harper’s.
The post Do Our Dogs Have Something to Tell the World? appeared first on New York Times.
