Bunny the dog and her button soundboard became a viral hit on TikTok amid 2020 pandemic lockdowns. In videos, the adorable, if neurotic, sheepadoodle taps buttons related to play, outside, or poop, and her enthusiastic human owner and trainer fulfills requests while engaging in something that resembles back and forth conversation. Sometimes Bunny’s button taps are straightforward: demands for attention or a walk. Yet other times, they seem to veer towards the abstract and unsettling, as in one instance where the dog appears to interrogate her own reflection.
It’s undoubtedly entertaining, engaging content. But is it legit? Are Bunny and dogs like her pawing at buttons to communicate with their humans? Or is it another example of the stomping horse at the center of the Clever Hans fallacy–where people project what they want to be true over an animal’s actions?
@whataboutbunny Bunny: Bringing you existential content since dogs could talk #bunnythedog #talkingdog #fypシ #aac #WeekendVibes #doggos ♬ original sound – I am Bunny
Early scientific assessment may disappoint the skeptics. Dogs are using button boards deliberately and in ways that are distinct from their owners, according to a study published December 9 in the journal Scientific Reports. The research further finds that certain two-word button combinations can’t be explained by chance, and that dogs may be willfully stringing together short phrases.
“They ask for things that are reasonable for dogs, they’re not purely matching training, and combinations are non-random,” says Frederico Rossano, senior study author and an associate professor of cognitive science at the University of California, San Diego, where he directs the Comparative Cognition Lab.
The research does not prove that dogs understand abstract concepts like time or verbs (whether Bunny knows what she is asking when she presses “who this” remains unresolved). Nor does it demonstrate that dogs can ‘speak human language,’ with anything akin to the structure and comprehension that people have. But it does put scientists on track to try to address those, and other, questions in future work.
“The way I think of my studies is kind of like traffic lights [on the road] to deciding whether to continue [a line of inquiry] or not,” says Rossano. So far, this is the second peer-reviewed, empirical study of the soundboard dogs that Rossano’s lab has released, and the second green light. The first, published in August, established that the dogs do seem to understand and respond to the button board sound cues, signaling comprehension. The new study builds on that, demonstrating that dogs have flexible control over pushing the buttons. “The fact that [multibutton sequences] are non-random, it invites further research,” he says.
There’s a vast collection of prior published science indicating that dogs are well-attuned to people and can learn to understand some human words and hand signals. If you’ve ever trained a pet dog, your personal experience likely backs this up. But the new study is somewhat unique for showing that, not only can dogs understand us, they can also use our own communicative signals back at us.
@whataboutbunny Anything for you Buns🥺 #scritches #cutedoggy #sosweet #dogstory ♬ original sound – I am Bunny
“The novelty of this specific paper is that it’s kind of a production study. Everybody else does the comprehension studies. We are the first ones looking at how [dogs are] trying to combine signals actively to communicate with humans,” Rossano notes.
The button board approach with dogs is similar to experimental methods that have been used with chimpanzees and bonobos in the past. For instance, in the well-known (if controversial) work of Sue Savage-Rumbaugh. But in the new dog research, there’s the added benefit of studying domestic dogs in their comfortable home environments as opposed to a contrived and potentially stressful lab setting.
To do this, Rossano’s lab turned to community science, relying on dog owners who’ve trained their pet to use buttons to track activity. They recruited participants online and had people report every time they or their dog pushed a button. The study analysis included data from 152 dogs, each of whom interacted with their soundboard at least 200 times over a 21-month period. In total, 194,901 instances of dogs pressing buttons were assessed, of which 56,676 were multi-button combos. Additionally, 65,682 instances of humans pushing buttons in modeling events or training sessions were considered.
Button labels were standardized into categories like “food” to account for specific sound bites such as “kibble” or “dinner.” And the researchers statistically compared the ways the dogs used the buttons with the ways the humans did.
They found that the amount of button presses varied a lot dog-to-dog. The median number of presses was 10.9 per day, but the highest individual average was 90 daily button presses and some dogs pressed buttons once a day or less. Yet despite the variation, there were clear trends.
Most dogs pressed buttons related to their routine activities and needs like those in the “food”, “play”, “go outside”, or “potty” categories. Common dog-composed combinations that occurred more included variations on “food”+”treat” and “own name”+“want.” In contrast, humans tended to push button combos indicating “love” or “later” more than their canine counterparts.
@hunger4words Note to self: bring Stella a puppaccino next time! When I saw Stella licking inside my Starbucks cup, I picked up my drink, said it wasn’t for her, then hid it behind my back. Stella immediately argued, “Yes where? Mad no.” Conversational interactions like these REALLY highlight Stella’s skills. Knowing what to say during routine activities that are practiced every day is one thing, but generalizing word meanings to respond in new situations is a true demonstration of language proficiency & intelligence. 🐶🧠 ✨ #starbucks #puppachino #puppaccino #pupcupstarbucks #pupcup ♬ original sound – Hunger for Words
“It confirms what I had kind of already thought to be true,” says Amritha Mallikarjun, a canine cognitive science researcher at the University of Pennsylvania’s Penn Vet Working Dog Center who was not involved in the new study. “It makes a lot of sense that, if a dog hits a button and then I do something the dog likes, that they would hit that button again.”
Mallikarjun emphasizes that the findings don’t indicate dogs have formal language ability. “There are a lot of components of human language,” she says, like the properties of semanticity, discrete infinity ordered utterances, and displacement. “The dogs are not sufficiently meeting any of those components,” she says. And dogs likely aren’t interpreting the buttons the same way people do. To a dog, for instance, the “ball” button may be linked with whatever their human does in response to that button being pushed–not necessarily with the ball itself. Through reinforcement and response, people are training their pets to associate certain buttons and sounds with certain desired outcomes.
But she agrees that the study robustly demonstrates the dogs are pushing buttons non-randomly and with intent. “We and dogs have evolved together over thousands of years to be communicative with each other, so the idea that dogs are using whatever tools we give them to communicate with us, it makes perfect sense,” she adds.
In many ways, dogs already communicate their needs to people, Mallikarjun points out. Think of your pet standing by the door when it wants to go on a walk, bringing you a favorite toy, or barking by a food bowl for a meal. “The buttons aren’t necessarily needed to get these pieces of information,” she says. However, studying their use among willing dogs and owners is an opportunity to do “a lot of cool research.”
Already, Rossano and his colleagues have lots of additional work underway. In this most recent study, there are notable limitations. Because they relied on dog owners self-reporting button presses, it’s possible there were biases in what humans documented and shared. Plus, the data collection was conducted early on in the soundboard trend, meaning it’s a snapshot of initial outcomes and a small subset of pets, not necessarily representative of the limits of what’s possible, says Rossano.
@whataboutbunny As you wish. #whataboutbunny #sheepadoodle #hunger4words #BestFriendDay #mydogtalks #petroutine #talkingdog #cute #smartdog #tiktokdog #cutepup #aac ♬ original sound – I am Bunny
Through follow-up research, he plans to address these limitations and beyond. The lab now has participants rely on an app that auto-tracks button presses to reduce the potential for unintended bias and many subjects are now recording standardized videos for evidence. Plus, the project is currently at 10,000 self-enrolled participants, reducing the 152 dogs assessed in this paper to “a tiny fraction of the data we’re collecting,” Rossano says.
With this larger dataset and more fine-tuned protocol the scientists are pursuing follow-up questions. They have a study in progress assessing how dogs use the “help” button and engage with less concrete concepts. They’re more closely examining the outliers in their dataset, to determine why some dogs engage with the buttons more than others. Early evidence suggests it’s not breed, extraversion, or cognitive ability, but rather anxiety level, that might motivate a dog to use the soundboard. “The ones who tend to be a little anxious might find this device a way of getting more control over their environment and getting some reassurance from humans about what’s going on,” Rossano posits. (Bunny is reportedly on anti-anxiety medication, according to a 2022 post on her owner’s Instagram.)
They’re aiming to assess longer combinations of words, if and how dogs communicate about objects they can’t see, how the pets interpret buttons related to time, and if dogs can combine different buttons to convey new meanings. This last point is a classic facet of human language ability. Confronted with a forgotten or unknown word, humans will remix phrases to get a point across. It’s possible, based on anecdotal evidence, that some dogs do the same. Rossano says he’s seen footage of a dog pressing the buttons for “squeaker” and “car” in response to an ambulance passing by outside. Another dog pressed “water” and “outside” when the “beach” button that it was used to was taken away. Yet another dog seems to have coined the phrase “water”+“bone,” which the owner interpreted to mean ice. “The next step is to design studies that can creatively capture this behavior in a more systematic way,” he says.
Assuming the green lights keep coming, all these studies together will hopefully allow us to get inside the minds of our canine companions in new ways. “This device can help us scientists find evidence that dogs can think about things we didn’t know they were capable of,” Rossano says. Maybe it “can allow us to see abilities that we didn’t know they have.”
Win the Holidays with PopSci’s Gift Guides
