All Episodes
Previous Next
Plain Text
Oct. 8, 2024 - Skeptoid
18:15
Skeptoid #957: Dogs Who Type

Can dogs be taught to speak intelligently using floor buttons that represent words? Learn about your ad choices: dovetail.prx.org/ad-choices

Transcriber: nvidia/parakeet-tdt-0.6b-v2, sat-12l-sm, and large-v3-turbo
|

Time Text
Can Dogs Really Talk 00:01:30
Maybe you've seen YouTube or TikTok videos where dogs are communicating fluently with their owners using buttons on the floor to speak words.
Are these amazing communication abilities truly possible for our furry friends simply by buying an electronic device and teaching our dog to use it?
Well, we're going to find out today on Skeptoid.
A quick reminder for everyone, you're listening to Skeptoid, revealing the true science and true history behind urban legends every week since 2006.
With over a thousand episodes, we're celebrating 20 years of keeping it focused and keeping it brief.
And we couldn't have done it without your curiosity leading the way.
And now we're even offering a little bit more.
If you become a premium member, supporting the show with a monthly micropayment of as little as $5, you get more Skeptoid.
The premium version of the show is not only ad-free, it has extended content.
These episodes are a few minutes longer.
We get rid of the ads and replace them with more Skeptoid.
The Extended Premium Show available now.
Come to Skeptoid.com and click Go Premium.
You're listening to Skeptoid.
I'm Brian Dunning from Skeptoid.com.
The Soundboard Myth Explained 00:13:47
Dogs who type.
Welcome to the show that separates fact from fiction, science from pseudoscience, real history from fake history, and helps us all make better life decisions by knowing what's real and what's not.
If you're a dog owner, you may have seen these soundboard products, buttons that your dog can push to speak words and communicate with you.
You may have even seen videos on social media where a dog communicated an abstract concept, apparently understanding the idea of language.
But is this really language?
Or is the dog simply making associations between the button and a reward, a treat or a walk or going outside?
Or is it even less than that, the dog responding to cues from the owner and stepping on buttons while approaching their owner who's standing next to those buttons?
Today we're going to evaluate the evidence and thus the validity of these products for dogs being able to usefully communicate.
Most research into this question, which has been going on for nearly 100 years, has been done using devices called AICs for augmentative interspecies communication.
Typically these consist of several buttons that you place on the floor big enough to be easily pressed with a paw, often in different bright colors and or with some visual icon printed on them.
Pressing the button plays an audible sound, which is usually a recording of you, the dog's owner, saying a familiar word like food or outside or walk.
These devices are commonly called soundboards.
Some may have just a few buttons.
Others may have several dozen.
A variation on these has an unfortunately similar name, ACIs, for animal computer interfaces.
The difference between an AIC soundboard and an ACI is that the soundboard just plays a sound and requires a human to respond to the dog's request.
An ACI, on the other hand, automatically performs a function, such as dispensing food or opening a door, without any human interaction.
ACIs are sometimes used in zoos or other animal captivity situations.
They're frequently used in animal communication research in many species, not just dogs.
So we turn to the research being done in the field today.
Are dogs able to communicate to their owners what they want via associative learning with these soundboard devices?
As always, when I look at a research paper, I look at the author's conflict of interest statement.
Here's one example from a 2024 paper published in PLOS One that found positive results for dogs using a commercially available soundboard.
From the discussion of their results.
Our findings provide the first evidence of button word comprehension by owner-trained soundboard using dogs and demonstrate that dogs' contextually appropriate responses to button presses were comparable regardless of the identity of the person using the soundboard and the absence of other environmental cues related to that word.
Our findings also suggest that dogs attend to the sounds recorded on the buttons, given that they responded equivalently to words when they were produced by button presses and when they were spoken by their owners.
To be clear, the conflict of interest statement disclosed that six of the authors had previously consulted for the manufacturer of that soundboard, and two more of the authors were current employees of that manufacturer.
Normally, this would make the article a hard pass for me, and I'd strike it from my sources, but I mention it here because its results were consistent with other papers I found.
I did look carefully at another paper, which is probably the most cited in the field these days, and it has a similar problem.
Four of the five authors disclosed that they were directly paid by another one of these product manufacturers to perform this study.
It was published in 2023 in WIRE's Cognitive Science, and it's a survey of lots of studies, though it focuses on the ACI computer devices among all animal species, not just dogs on soundboards.
The reason I point this out is that, despite the conflict of interest, the paper's findings are generally negative.
The authors point to the history of research in this field and note that over the past 50 years, funding and interest have both almost completely dried up due to the lack of positive results.
It points to specific cases like Coco, the gorilla who allegedly learned sign language, but it was found that this illusion was nearly always the dubious real-time interpretation provided by her handler.
There's a complete skeptoid episode on Coco and other signing apes, episode number 630, if you want to know more.
The paper finds the decline of the field to be due to three areas of controversy.
Number one, use of these ACI devices don't require that the animal has learned language, or indeed any form of communication.
Pushing a button and having food come out is simply learned behavior, more formally called associative learning.
Pavlov's dog didn't salivate because it understood language, but because it had learned to associate the sound of the bell with food.
So this isn't really the interspecies communication the researchers are hoping for.
Number two, the systems are often found to be ecologically irrelevant, meaning researchers have observed animal behavior in the wild that can be described as communication, and a button to push is not a valid proxy for that.
So as far as researching animal communication goes, an ACI experiment is kind of upside down and backwards.
And number three, these studies are all too often rendered invalid by researchers cueing the animals, either consciously or unconsciously, and a lack of systematic approaches for recording data in such an abstract field.
It's very hard to do animal language study that meets high methodological standards.
But at the very end, the paper makes what I think is a very salient comment.
It mentions that we're in a, quote, renaissance in the field of AIC device-facilitated interspecies communication.
To me, this is a very careful and qualified statement, as if they wish to meet the goals of the consumer product manufacturer who paid for the study, but also don't want to make any unscientific declarations.
The fact is that products like these dog soundboards are inexpensive and widely available today, which wasn't the case 25 to 50 years ago.
Any renaissance is probably due simply to the fact that we can now all have these things if we want them.
In a world that can feel overwhelming, spreading thoughtful, evidence-based content is one of the best ways to make a positive impact.
Ask your local public radio station to air the Skeptoid files, a 30-minute radio-friendly version of Skeptoid that pairs two related episodes promoting real science, true history, and critical thinking.
And in these challenging times for public media, we're offering these broadcasts for free to radio stations, available on the PRX Exchange or directly from Skeptoid Media.
It's an easy ask.
Just send a quick message to your station's programming director.
By helping to bring the Skeptoid files to the airwaves, you'll help promote the essential skills we all need to tell fact from fiction.
Just go to your local station's website, find the programming director's email address, or just their general email address.
You can even use the telephone.
I know that might sound crazy.
It's an old legacy device that allows real-time voice communication.
I know that's weird, but hey, it's an option.
The world can feel chaotic, but you're not powerless.
When you promote critical thinking, you can help your community tell fact from fiction.
And that's how we shape a better future.
In uncertain times, spreading good ideas can make you feel helpful, not helpless.
Let's stand up for reason, truth, and understanding together.
Get them to air the Skeptoid files from Skeptoid Media, available on the PRX Exchange, and they'll know what that is.
Is there any science that remains to be done?
Well, at least as far as these dog soundboards go, there may not be.
The decline of research over the past 50 years has shown us that true language skills are not necessary for simple associative learning devices like these soundboards to be useful.
Dogs easily associate the word walk with going for a walk, and most can also associate the button that pronounces walk with going for a walk.
Dogs do associate devices like these soundboards with a few basic functions that are most important for us as dog owners to know.
They're hungry.
They need to go outside to pee.
They want to go for a walk.
The pitfall lies in assuming this means the dog has learned language or might be able to expand their communication skills if given a soundboard with more abstract concepts, let alone an alphabetic keyboard.
Yet this is where all the attention is being focused these days.
The idea that a dog with a soundboard can construct sentences of several words and communicate abstract concepts that require a comprehension of language.
Google this today and you'll get promotional articles for a lab at UC San Diego working in partnership with one of the manufacturers of these soundboards.
Anecdotes run strong through this study and it's over 10,000 participants worldwide.
People who believe their dogs are constructing abstract sentences by pushing word buttons, meaning things like, hey mom, my paw hurts.
Can you look at it please?
Or let's go for a walk so I can see my friend.
The question this leaves us with is whether these soundboards have truly given dogs the tool they need to construct grammar, to communicate abstractly, or whether hopeful dog owners are simply combining confirmation bias and apophenia to perceive complex sentences in data that is actually just random.
There are a lot of videos on social media with optimistic titles like The Dog Who Can Talk and so forth.
I watched many of them, and few are very impressive once you realize that the owner is cueing the dog in almost all cases.
Something I saw a lot of was the owner doing this while standing at the soundboard, and the dog was looking up at the owner and just stepping on buttons without even paying attention, their attention being focused on the owner.
There are also videos where the soundboards are being used as intended.
The owner videos it from across the room without cueing, and the dog goes over, looks at the buttons, and pushes one.
Cases like these represent deliberate use of associated learning by the dog.
I never saw an unqueued video where the dog, when left alone, employed multiple buttons in what some of these owners described as sentences.
The San Diego lab, run by principal investigator Federico Rossano, only just this year published its first academic paper.
It was extraordinarily unimpressive.
A fact Rossano acknowledged, calling it a, quote, necessary first step.
The finding was that dogs do respond to soundboard cues the same as they respond to verbal cues from their owner.
There's no surprise that dogs respond to verbal cues.
You can say walk and your dog will practically jump out of its skin in excitement over the prospect of a walk.
All this study found is the dog will respond the same way whether that command is spoken or is triggered by someone pushing the button on a soundboard and regardless of whether it's the owner or a random research assistant pushing the button.
Dogs respond to cues.
Nothing earth shaking there.
Anything beyond that, which you may read about the San Diego study, is so far unproven.
So here's the bottom line.
We have solid evidence plus thousands of years of anecdotal personal experience that dogs understand basic commands and will respond to them.
Many dogs, but not all, can learn to associate a limited number of buttons on commercially available soundboards with basic actions.
Food, outside, walk, potty, and possibly a few others.
These soundboards do appear to be effective tools for many dogs to be able to make requests of their owners.
But these communications should not be expected to exceed the normal body language and behavioral requests that dogs already make.
The idea that dogs can make more complex or abstract requests using these soundboards is not supported by evidence, despite lots of cherry-picked videos appearing to support this on social media channels.
So if you want to try one of these devices, you can reasonably expect your dog to be able to learn to make the same requests he already makes of you now, perhaps with a bit less ambiguity.
And that's a good thing.
But if your expectations exceed that, you're likely to be disappointed.
And who wants to be disappointed with their best friend?
Why Evidence Falls Short 00:02:53
We continue with more on how dogs may or may not be physically able to watch television with you in the ad-free and extended premium feed.
To access it, become a supporter at skeptoid.com slash go premium.
A great big Skeptoid shout out to our premium supporters, including Rutger, Philip Cannon from Brisbane, Australia, Keith and Ada Baston, and my old friends from Principles of Curiosity, Phil, Lisa, and Tyler Williams.
This month I'm giving my newest live in-person talk, Balderdash and Poppycock, at two locations right here in Central Oregon.
For details, come to skeptoid.com/slash events.
Or to book your own gig for your company, conference, meetup group, or pub night, come to skeptoid.com slash speaking.
If you liked this episode and you're a Discord user, come to skeptoid.com slash Discord to access our listener-exclusive Discord channel, where you can discuss past episodes, suggest new ones, or talk about our upcoming Skeptoid Adventures events.
Hope to see you there.
Skeptoid is a production of Skeptoid Media.
Director of Operations and Tinfoil Hat Counter is Kathy Reitmeyer.
Marketing guru and Illuminati liaison is Jake Young.
Production management and all things audio by Will McCandless.
Music is by Lee Sanders.
Researched and written by me, Brian Dunning.
You're listening to Skeptoid listener-supported program.
I'm Brian Dunning from Skeptoid.com.
Hello, everyone.
This is Adrian Hill from Skookum Studios in Calgary, Canada, the land of maple syrup and moose.
And I'm here to ask you to consider becoming a premium member of Skeptoid for as little as $5 per month.
And that's only the cost of a couple of Tim Horton's double doubles.
And that's Canadian for coffee with double cream and sugar.
Why support Skeptoid?
If you are like me and don't like ads, but like extended versions of each episode, Premium is for you.
If you want to support a worthwhile nonprofit that combats pseudoscience, promotes critical thinking, and provides free access to teachers to use the podcast in the classroom via the teacher's toolkit, then sign up today.
Remember that skepticism is the best medicine.
Next to giggling, of course.
Until next time, this is Adrienne Hill.
From PRX.
Export Selection