Epistemic status: highly suggestive.

[EDIT: Added more info on research methods. Addressed some common criticism. Added titles for video links and a few new vids. Prevented revolution with a military coup d'état]

A combination of surveys and bayesian estimates[1] leads me to believe this community is interested in autism, cats, cognition, philosophy, and moral valence of animals. What I’m going to show you checks every box, so it boggles my mind that I don't see anyone talk about it. It has been bothering me so much that I decided to create an account and write this article.

I have two theories.

  1. The community will ignore fascinating insight just because its normie coded. Cute tiktok-famous poodle doesn't pattern match to "this contains real insight into animal cognition".

  2. Nobody tried to sell this well enough.

I personally believe in the second one[2] and I'll try to sell it to you.

Stella

There’s an intervention to help non-verbal autistic kids communicate using “communication boards” (not to be confused with facilitated communication which has a bad reputation). It can be a paper board with pictures or it can be a board with buttons that say a word when pressed. In 2018 Christina Hunger (hungerforwords.com) - a speech pathologist working with autistic children using such boards - started to wonder if her dog was in fact autistic. Just kidding, she saw similarities in patterns of behavior between young kids she was working with ("learner" seems to be the term of art) and her dog. So she gave it a button that says “Outside” and expanded from there.

Now teaching a dog to press a button that says “outside” is not impressive or interesting to me. But then she kept adding buttons and her dog started to display capabilities for rudimentary syntax.

Stella the talking dog compilation - Stella answers whether she wants to play or eat, asks for help when one of her buttons breaks, alerts owner to possible "danger" outside.

Stella tells us she is all done with her bed!

Stella the Talking Dog says "Help Good Want Eat"

(most of the good videos are on her Instagram @hunger4words, not much is on YouTube)

Reaction from serious animal language researchers and animal cognition hobbyists was muted to non-existent, but dog moms ate this stuff up. One of them was Alexis.

Bunny

Most useful research is impractical to do within academia

The Importance of Methodology and Practical Matters

Ethology has some really interesting lessons about how important various practical matters and methodology can be when it comes to what your field can (and can't) produce. For example, it turns out that a surprising amount of useful data about animal cognition comes from experiments with dogs. […] The main reason is because they will sit still for an fMRI to be the goodest boy (and to get hot dogs). […] On the other side of that coin, elephants are clearly very smart, but we've done surprisingly little controlled experiments or close observation with them. Why? […] They're damn inconvenient to keep in the basement of the biology building, they mess up the trees on alumni drive, and undergrads kept complaining about elephant-patty injuries while playing ultimate on the quad.

https://astralcodexten.substack.com/p/your-book-review-are-we-smart-enough

A lot of useful research isn't done because it's too inconvenient, too expensive or otherwise impractical to execute within confines of academia. This is a massive shaping force. Existence of ImageNet and its quirks is a stronger shaping force on AI research than all AI ethics committees combined.

Nobody had done this before because it takes months of everyday training to get interesting results. Once your dog gets the hang of it, you’re able to add more buttons faster, but it’s never quick. Dogs take a while to come up with a response (they’re bright, but they’re not humans), and you can’t force your dog to learn, so you have to work together and find motivation (for the dog and for yourself!). And not every pet has a strong desire to communicate.

But it may be practical to do for a layperson

Lucky for us, Alexis has many more commendable qualities besides willing to spend time and effort on her dog. She maintains healthy skepticism, she's well aware of confirmation bias and "Clever Hans" effects and of the danger of over-interpreting the dog's output. She has partnered with researches from University of California, San Diego to have several cameras looking at the button pad running 24/7, for them to do more rigorous study.

Please watch this vid first where she gives a brief explanation of what she's doing, and importantly shows her attitude and skepticism towards her dog's "talking". She even namedrops Chomsky & Skinner 😄

Study by Comparative Cognition Lab at the University of California, San Diego

I encourage you to read their research methodology page yourself. Here's a summary:

This is an ongoing study with hundreds[3] of participants and a mission to "use a rigorous scientific approach to determine whether, and if so, how and how much non-humans are able to express themselves in language-like ways". It's headed by Prof. Federico Rossano, Director of Comparative Cognition Lab at UC San Diego and Leo Trottier, PhD Candidate, UC San Diego. The latter has websites selling dog soundboards and interactive pet toys, indicating a potential conflict of interest, but I also don't want to knock him for simply being enterpreneural.

They mention previous research with a dog named Rico and a border collie named Chaser (video of Chaser) that had rigorous experiments performed with an opaque barrier between the dog and the experimenter to preclude possibility of unwittingly influencing the dog's behavior. In those studies dogs were able to recognize 200+ toys by spoken name and perform fast-mapping (better known in some circles as one-shot learning). Encouraged by the studies they aim to ask:

Is what we're seeing clever dogs or merely Clever Hans? Can we explain the surprising button pressing behavior we're seeing using a simple first-order associative learning model, or will we have to reconsider the idea that language is an ability that is 'uniquely human'? And do we see any change in the type and complexity of communications that non-human animals (and dogs in particular) generate once they are able to use concepts that have been associated with buttons?

They're splitting the study into 3 phases:

  1. Initial data collection, where they gather information about learners, their owners, methods of training and have participants log reports about when a specific word was first introduced, used in an appropriate context and used as part of a multi-button expression.
  2. Video collection and analysis, where they get participants to install at least one video camera pointing at the soundboard that records every interaction. That allows them to see how button usage changes over time and to measure behavior more reliably and precisely.
  3. Interactive studies: "Based heavily on the insights gained in phases 1 and 2, we will be piloting direct, controlled tests of learner sound button use and understanding that aim to determine how language-like learners' sound button use is. We anticipate that these will be done with a smaller number of participants."

Things I've seen the dog (appear to) do that surprised me

There's a long-running debate about whether human brains possess a special ability for language.  Although "feral" human children who are raised with no language lose the ability to pick it up later in life. Maybe they could learn button talk, I don't think anyone tried for lack of steady supply of feral children.

But what I see is strongly suggestive that language facilities are not unique to us and a dog that is given ability to produce words and is taught from puppyhood with the same massive amount of effort that we put into human children will be able to talk. Don't get me wrong, I don't expect dogs to start writing poetry and doing particle physics. But I expect them to produce something that can undoubtedly be called language[4].

It's all "Clever Hans"

I don't think this can be explained by classic Clever Hans where the owner tells the dog what to do with subconscious cues. You see lots of interactions where the dog is supposed to make a decision and the owner doesn't know the right answer, or the dog alerts the owner to something they're unaware of.

When Bunny used "stranger paw" to indicate a splinter in her paw, how was the owner supposed to influence the answer without even being aware of the splinter?

It's all operant conditioning

First, it can't be all just conditioning. By induction: pets already communicate with owners to request things, teaching your dog to press "Food" instead of barking or "Outside" instead of scratching at the door simply changes the modality. Usage of simple buttons like that doesn't require clever hans or conditioning as an explanation.

Second, look at this video of Billi the cat. Assuming it's just conditioning, we'd expect the cat to always go "yes food" when asked about food, because it doesn't understand "yes" or "no". But here we see it getting practically railroaded by the owner and still refusing food, and in several different ways (first "no", then "all done later"). How could this happen if it was just simple conditioning?

The owners over-interpret and anthropomorphize the button "speech"

The is the biggest danger in my opinion. Hopefully with rigorous analysis during the study and specifically set up experiments we'll be able to understand better at what level of communication the dogs actually are.

Interestingly, at least for humans, misinterpretation may be a necessary requirement for language acquisition, as mentioned in this comment! The short summary of the mechanism:

  1. Toddler raises arms up randomly with no intention.
  2. Mother thinks he wants to be held, so picks him up.
  3. Toddler learns the association and the next time he raises his arms, it's an intentional attempt at communication. Similarly for words, say "maa" randomly, mother comes and smiles excitedly, the association is built.

I think the videos are fake

I wrote a comment on that and I think the videos are done in good faith. Of course this doesn't preclude other problems, Clever Hans was in good faith too, after all.

This doesn't look like real science, it's just "dog moms" enjoying a fun hobby, YouTube videos aren't evidence of anything.

Early stage science often looks like "messing around", before theory and rigor is built. Telling what "messing around" is likely to be fruitful is a meta-rational skill, and it can be done, somewhat.

YouTube (and especially Instagram and TikTok) videos go against usual aesthetic sensibilities of what evidence looks like, but that's not a reason to discount them completely. They still constitute a lot of evidence, albeit the kind that is weak and easy to misinterpret. If we're to be good rationalists, we can't exclude the messy parts of the world and then expect to arrive at a useful worldview.

The work on the ground being done by laymen may be a blessing in disguise - I won't be surprised if taking a formal and procedural approach to training would "ruin the magic" and lead to poor results. Especially if mutual misinterpretation is an important part of the mechanism. I hope that given the number of participants in the study and the amount of data (every interaction recorded) and well-designed experiments will together let us separate signal from the noise.

Koko

So what, you ask, some apes have been taught sign language and they produced rudimentary syntax as well.

For starters you wouldn't predict dogs to be capable of the same, and it's significant to see that ability given their evolutionary distance from us (even with selective breeding pressure from domestication).

Most of the ape research was done in the 70s, and it is, well, very 70s. Those things aren't known for being well-run or replicating well. And it was done with sign language, perhaps buttons are much more conductive to language acquisition. Since the 70s craze, it apparently became unfashionable, and nothing new happened for decades. To this day any conversation on animal language is about Koko (who died in 2018) and the parrot who said "love you, bye" before dying in 2007. Utter stagnation.

What we have is something new, orders of magnitude easier to study and reproduce (how many of you have gorillas at home?), massive PR potential, modern tech that allows you to have cameras running 24/7 to preclude criticism. It started with an outsider to the field, who wasn't conceptionally limited by prior art. And it's accessible to regular people, potentially revolutionizing our relationship with our pets.

This raises the natural question: what if you gave an ape the buttons, and taught it from childhood, and put parent-level effort into it, not "70s research”-level effort? Perhaps the answer would surprise us.

[EDIT: Exactly this has been attempted with bonobos, but unfortunately little data is available and the experiment disintegrated over human drama. Read the linked comment for details and a few existing videos]

Honorable Mentions

Billi

A cat who initially became famous for pressing "MAD MAD MAD" at a slightest inconvenience, but she has meollowed out a bit.

"Mad" A Short Film Starring Billi the Cat

Mom’s Choice - appears to ask "mom" what she wants to play. Normally mom is the one asking her what toy she wants to play.

Imposter - an important video. In it, Billi repeatedly refuses food, despite the owner practically railroading her. With simple conditioning you could expect "want food hm?" -> "yes food" from the cat, without understanding of what "yes" or "no" is. But if all of this is just clever conditioning, why did this video happen?

Others

Talking dog Tiktok compilation | Flambo the dog

Dog Communicates with Human by Talking Buttons PT2

Izzy the talking dog uses recordable buttons to help her sister Luna

What did we do with the word buttons? 😯 | Pharaby the Talking Dog

HowTheyTalk.org

Community of people trying to replicate this, ran by the people running the UCSD study.

Maybe later

“Maybe later” at substack for trying in vain to tell others about this, only to be summarily ignored. I got your back, buddy.


  1. aka blindly trusted stereotypes ↩︎

  2. Normie blindspot does exist in the community, but that's kind of obvious and expected, and should be a separate article. ↩︎

  3. Possibly, they're not clear on the number of participants. I personally find it hard to believe that there would be more than about a hundred. ↩︎

  4. In the layman sense of "tool for communication". Philosophical discussions about the exact border between non-language communication and "true language" aren't really interesting to me. Duck typing, etc. ↩︎

New Comment
110 comments, sorted by Click to highlight new comments since:
Some comments are truncated due to high volume. (⌘F to expand all)Change truncation settings

Another honorable mention is Chaser, a border collie who was trained and tested by John Pilley et al with a language design and testing regime that was specifically aimed at making Clever Hans criticisms impossible, and also to make it very clear that certain grammar recognition tasks could be statistically detected VS alternative ways to solve the linguistic challenge that "don't seem like they are doing language learning right"... What if a dog learns "fetchblue" as a single sound that lacks a breakdown into a coherent verb about an action and a coherent object named blue?  What if, in the dog's head, "fetchblue" is just like a name for a scene that includes the object and the actions normally done with the object, in a giant swirl? Well...

They taught Chaser more than a 1000 proper names, and at least 3 verbs, and did tests with her in front of audiences using objects that (so far as they could tell from their notes?) hadn't been paired with the tested verb before.

Here's the kind of performance they could do, for video cameras, as supplementary material for the 2010 paper:
 

In the paper, they also claim something I'd previously thought impossible for dogs, which was to b... (read more)

5StyleOfDog
Thank you for the info, I was not aware of Chaser! By the way, how did you do the YouTube embeds? I couldn't make them work in my article.
6JenniferRM
I put a raw URL in the raw text, and javascript rewrite magic happened to it and made it an embedded video.   I was strongly tempted to try and hit ^z on the rewrite because in my experience embedded stuff changes over times and thus makes the writing "not able to persist in archives for the ages", but... :shrugs: I'm not surprised that the article didn't have it. LessWrong has had the issue that "comment markdown stuff and article markdown stuff work differently" essentially forever.  I guess another possibility: maybe the LW forum software devs changed the WYSIWYG javascript editor(s) to work the same, but you have a different browser than me, or that they tested on? If you respond with a raw youtube URL, and don't get an embedded video based on a javascript rewrite of the comment, that would help clarify what might be going on :-)
6habryka
Yep, sorry, we don't currently support Youtube embeds for the markdown editor. Just turned out to be much easier to implement for the WYSIWYG editor, since it came out of the box with the framework we are using.

I find it quite fascinating that in the videos I've seen with Bunny, she tends to use SOV word-order (Subject Object Verb) even though her owner always (that I've seen) uses the SVO order of English. Most The majority of human languages use SOV, about 75% 45% of the ones that care about word order. It seems to be mathematically convenient as well.

If the "75%" stat is coming from Wikipedia, that's for SOV and SVO combined: https://en.wikipedia.org/wiki/Subject%E2%80%93object%E2%80%93verb#Incidence .

2Adele Lopez
Ah, good catch thanks!

This sort of thing seems to suggest that EY's claims in this post about the scale of the relative intelligence differences between chimps, a village idiot, and Einstein is incorrect. The difference in intelligence between village idiot and Einstein may be comparable to the difference in intelligence between some nonhuman animals and a human village idiot. Which is a priori surprising, given that human brains are very structurally similar to each other in comparison to nonhuman animal brains.

3[anonymous]
The wiring in the 'village idiots' brain may be faulty, such that most of the available circuitry isn't usable.
[-]Raemon32-1

Curated.

This was one of the most offbeat-but-surprisingly-on-point LW posts I've read in awhile. I quite appreciated the epistemic status woven throughout the post (i.e. concerns about Clever Hans, the steps attempted at addressing it, an the current status of how the jury is still out on some studies)

[edit: I thought the post's title and opening epistemic status line were too strong given the level of evidence displayed in the rest of the post. However, I appreciated the rest of the post for the way it summarized the evidence]

I had recently looked into "how robust is chimpanzee/gorilla sign language, or sound-button language." I was disappointed to find AFAICT a quite small number of case studies. So this article was timely for my interest.

I recall a line in HPMOR when Harry notes that scientists hadn't even bothered checking what a 4 year old human could understand until relatively recently, despite that not require any advanced modern technology. My takeaway from this article was "man, it is sad that we haven't had, like, 100 dog case studies, 100 pigs,100 chimpanzees, 100 gorillas, etc" all with concerted attempts to teach language with varying methodologies. C'mon civilization... (read more)

This is an interesting response; mine is of the opposite valence. To me, this doesn't feel too dissimilar from something my cousin-who-is-into-pyramid-schemes would send me. I believe that this post has:

  • Large claims that are not evidence supported
  • Mirages of evidence that do not meaningfully constitute such
  • Cursory dismissal of potential concerns

Claims that set off alarm bells to me in this post include:

  • Your Dog is Even Smarter Than You Think
  • Epistemic status: highly suggestive.
  • There's a revolution going on and you're sleeping on it.
  • her dog started to display capabilities for rudimentary syntax
  • Once your dog gets the hang of it, you’re able to add more buttons faster, but it’s never quick. Dogs take a while to come up with a response (they’re bright, but they’re not humans), and you can’t force your dog to learn, so you have to work together and find motivation (for the dog and for yourself!). And not every pet has a strong desire to communicate.
  • Bunny is creative with the limited button vocabulary available to her and tries to use words in novel ways to communicate: "stranger paw" for splinter in her paw, "sound settle" for shut up, "poop play" for fart, "paw" to refer to owner's hand.
... (read more)

I'm not sure that we disagree much about how likely Bunny is to be doing complex language. My primary takeaway was not "this dog can talk", it's "man, we really should be checking more comprehensively whether dogs can talk." I tried to be pretty clear about that in the curation notice. 

I think early stage science looks more like messing around than like rigorous studies. I think you need to do a lot of messing around before you get to a point where you have something to rigorously study. My curation of this is a celebration of checking things and being curious, not an endorsement of the theory.

I think I do agree that the post's title and opening epistemic status are too strong (and yeah I think it was a mistake not to include that in the curation notice. I'll edit it to do so).

I don't think it's fair to say my dismissal of concerns is "cursory" if you include my comments under the post. Maybe the article itself didn't go deep enough, partly I wanted it to scan well, partly I wanted to see good criticism so I could update/come up with good responses, because it's not easy to preempt every criticism.

As for cursory evidence, yes it's mostly that, but cursory evidence can still be good Bayesian evidence. I think there's enough to conclude there's something interesting going on.

Are the vids even real?

For starters, all of this hinges on videos being done in good faith. If it's all creative editing of pets' random walks (heh) over the board, then of course we should dismiss everything out of the gate.

  • For Stella IIRC all of the interesting stuff is on Instagram @hunger4words, so I only had those two YouTube vids. I agree they're not the best for leading evidence.
  • Please watch this video even if you have time constraints (it works fine at 1.5x speed).
  • She shows (excessive IMO) humility and defers to who she considers experts.
  • Considers herself a "hopeful skeptic", when Bunny does something unexpectedly smart, she still wonders if it's just coincidence (at 4:0
... (read more)

I appreciate your response, and my apologies that for time-efficiency reasons I'm only going to respond briefly and to some parts of it.

I don't think it's fair to say my dismissal of concerns is "cursory" if you include my comments under the post. Maybe the article itself didn't go deep enough, partly I wanted it to scan well, partly I wanted to see good criticism so I could update/come up with good responses, because it's not easy to preempt every criticism.

I'm somewhat sympathetic to this. I do feel as though given large claims e.g. "revolutionary" and the definite rather than the hedge in the title, it was worth doing more than the cursory in the article itself. I haven't read your comments nor looked at the timing of them, but I imagine some to most readers read the article without seeing these comments. I'm saddened that those readers likely had much too strong a takeaway and upvoted this post.

As for cursory evidence, yes it's mostly that, but cursory evidence can still be good Bayesian evidence. I think there's enough to conclude there's something interesting going on.

This stuff is highly suggestive,

I agree with the first and not with the second. I think this is lightly sugge... (read more)

4weft
I am super-duper surprised she says it took a few weeks to teach the Outside button! It took about... 15 minutes to teach my dog to use her Food bell. And then the Outside bell and Treat bells were similarly fast. I don't think button pressing is inherently harder than bell ringing, so that shouldn't make a difference.  I guess if the dog was starting at zero training it would take two weeks. (Robin already knew how to Target an item, which she learned after learning hand Touch, which she learned as part of the process of teaching how clicker-like training with positive reinforcers works in the first place. ) I can imagine abstract words like "Tomorrow" and "Where" taking a whole lot longer, but the words that are just ways to obtain concrete things are extremely easy to teach. Outside bells are a very well-known and frequently-done thing. Look them up on Amazon and you'll see about 20 options for sale. 

I like the idea of interacting with the dog normally as you would a 2-year-old human, while having the cameras running 24/7 so less biased people can look over the data.

Yeah it's an important point that some phenomena (perhaps most phenomena) are impractical to recreate under a strict research protocol. If you tried to teach your dog with a very formal approach, you'd probably "lose the magic" that makes it happen. Kaj Sotala posted a comment that suggests that "incorrect" overinterpretation of babies' behavior is actually an important mechanism by which the learning happens! It's a slow, messy, iterative process to get that "meeting of minds".

I also really like the research setup, and I'm glad they're sourcing from several pet households. Most of the attention is on Bunny, because she's the most coherent and is actively posted on YouTube, but I believe there are quite a few more people participating, they just don't post it publicly.

And even though the learning process isn't under strict protocol, you can still design more rigorous experiments. AFAIK the mirror near Bunny's buttons was placed at the suggestion of the researchers specifically to see if she'll recognize herself in the mirror.

I watched the first two videos and was kinda shocked. I could see the dog thinking. Like, it stopped, was slow, then made a deliberate choice to click certain words. It was not something I believed a dog could do. I feel way more like I could chat with a dog. I have a bunch more empathy for dogs too now.

I think there’s a fair chance that somehow this is a mistake / a fluke, but have now visualized the world where it’s not, and it’s really cool.

(Thanks for the post! You may be interested in this other post on dogs that also increased my respect for them.)

1Gunnar_Zarncke
Well, not every dog. There must be a billion dogs on this planet and this is probably one in the top 99.99 percentile. That said I think it could still be a fluke. 

I've been watching this with interest when youtube randomly serves me videos (I sometimes search youtube for things like "crows playing", so I think the algorithm figured out that I like animal intelligence videos), but I wasn't aware of everything you mentioned.

imagines a future where people would be surprised to learn that humans thought animals couldn't talk

Thanks for this! This definitely does intersect with my interests; it's relevant to artificial intelligence and to ethics. It does mostly just confirm what I already thought though, so my reaction is mostly just to pay attention to this sort of thing going forward.

From reading the discussion, I get the impression that some of the commenters are writing from a position of "the prior is against significant dog intelligence, and the evidence here could have alternative explanations, so I'm skeptical". That is, the people feel that it would be quite surprising if dogs really were this intelligent, so establishing it requires some pretty compelling evidence.

At the same time, my own feeling is more like "there's no strong prior against significant dog intelligence, and the evidence here could have alternative explanations, so I'm being tentatively open to this being real". As in, even if you hadn't shown me these videos, dogs being approximately as intelligent as described would have seemed like an entirely plausible possibility to me.

If there are any people who feel like my first paragraph does describe them, I'd be curious to hear why they're coming into this discussion with such a strong prior against dog intelligence.

If I had to articulate the reasons for my own "seems plausible" prior, they'd be something like:

  • a general vague sense of animal research tending to generally show that animals are smarter than people often think
  • some of the animals
... (read more)
7StyleOfDog
I think you hit the nail on the head here. When I was writing the article I definitely had someone with a high prior in mind, to the point where I expected people to say "so what, why wouldn't dogs do that if you trained them". Sometimes people seem to put dogs closer to reflexive automatons like insects than to fellow mammals. My prior is the base affects that we feel aren't fundamentally different between us and dogs (and most higher mammals). I'm talking about stuff like fear, excitement, generalized negative or positive affect, tiredness, sexual arousal. Even something like craving a specific food, I don't see why it should be unique to us, given that dogs are often picky eaters and have favorite foods. People with strong priors against dog intelligence seem to ascribe everything to anthropomorphism, and there's often an undertone of "these people are too soft and weak, they call themselves ridiculous things like 'dog parents', they'll project human baby characteristics onto a Furby if you gave them the chance". FWIW I don't have a dog and don't plan to, and in my experience most dogs are fairly dumb. But to me they're clearly a bit more than simple automatons blindly reacting to stimuli.
7Archimedes
My priors include the idea that both animal intelligence is not that different from humans and also that humans tend to overly anthropomorphize animal cognition. The biggest misunderstandings of animal cognition are much like misunderstandings humans have of foreign cultures, often involving forms of mind projection fallacies where we assume other's values, motivations, priorities, and perceptions are more similar (or more different) than is justified.
3qbolec
I'm a bit confused by people in the comments entertaining the idea that priors should influence how we interpret the magnitude of the evidence, even though when I look at the Bayes' rule it seems to say that the magnitude of the update (how much you have to multiply the odds) is independent of what your prior was. I know it's not that simple because sometimes the evidence itself is noisy and needs interpretation "pre-processing" before plugging it to the equation, but this "pre-processing" step should use a different prior then the one we try to update. I'm not sure how exactly this "multi-stage Bayesian update rule" should work, and I was trying to describe my struggle in my https://www.lesswrong.com/posts/JtEBjbEZidruMBKc3/are-dogs-bad

I found the first conversations(?) here very interesting:

It seems like a lot of the words could be swapped for each other or reordered and still look meaningful, so you could get a lot of the way to these conversations(?) just by learning what general categories of words tend to produce surprise/confusion (seemingly irrelevant words like 'stranger', 'friend', 'oops', 'dog', 'hi') versus happiness/engagement ('walk', 'outside', 'beach', 'now', 'soon', 'please', 'want', 'go', 'why', 'when') when you want to go for a walk. I'll certainly be interested to see what falls out of a larger-scale analysis of videos/transcripts.

The fact that this post was curated, and raised so little concern in the comments is worrying.

The author expresses himself maniacally in a mix of meme-speak and LW-lingo. The main claim is never really clarified*, but is presented as if it were, and in an outrageous way. There is no evidence besides YouTube videos**. I have no idea what is going on, and don’t even know where to start.

Comparative cognition is a real scientific field. If you’re interested in the cognitive abilities of non-human animals, and how they compare with the abilities of other species, then there’s a whole scientific literature you can take a look at. Ironically, this post sends us back to the lowest of this field, more than a century ago, when researchers made abundant use of observational anecdotes and interpreted them as they wished.

 

*None of the videos linked show a use of language at the level of even a child, so it can’t be “Dogs can speak”. Incidentally, nobody ever doubted that dogs can communicate with humans, especially regarding things as food, the presence of other animals or affection.

**That is, cherry-picked, sometimes even edited, videos where the owners of the dog superimposes what they think the dogs are saying.

Really? The main claim is presented "in an outrageous way"? 

I can imagine reading the post and being unconvinced by the evidence presented. In fact, that was my reaction (although I haven't watched the videos yet). But... being outraged

Posts should not make large, unsupported claims, and criticism should not be hyperbolic. Here is what I have learned from your critique:

  • You think that StyleOfDog writes "maniacally" and uses "meme-speak" too much.
    • Unclear why I should care. Not my favorite style of writing, but I understood what they were trying to say, and so communication occurred. It's not obvious to me that marginally more "meme-speak" would be bad for the site. It just read informally to me, which I don't mind for this kind of "fun / curiosity" post.
    • Ironically, your comment is hard to parse. The structure is scattered, which disorients me as I read your critique. In particular, you include two footnotes in quick succession, destroying flow. You could have just left them in the body of the comment.
  • You found the main claim unclear.
    • At least you formulated a hypothesis? I'm unsure why you found it unclear.
  • You are outraged.
    • Since you made a new account, I can't even Bayes-u
... (read more)
9subconvergence
Sorry for over-reacting to what I perceived as essentially a curated list of youtube videos with no real context. I made a probably more substantial comment as an answer to the OP.
0anon_standards
FWIW, I do not think you over-reacted, nor do I think I agree with any of the criticisms of the comment above.

It sounds to me like you're thinking of curated as 'vetted' or 'confident', while Ray is thinking of it more as 'representing a direction we want to see LW move in more' (including 'this topic seems neglected here, I'd like to see it get more attention').

It's the difference between 'curation is like publishing a non-replication-crisis-y journal article or encyclopedia article, a durable summary of humanity's knowledge we can call back to and base further conclusions on', versus 'curation is like sending out an email to a private researcher mailing list you run "hey, I'd love to see more discussion here in the vein of X"'.

5subconvergence
Thanks for the precision, I was unaware of that. I still think this post is of very limited value nonetheless.
4anon_standards
To me, and I expect to a group of other readers as well given upvotes to my comments and those of @subconvergence, this is a direction we really think LessWrong should not go in. I think Raemon is able to see some appeals of this post that are something like: * Boggling at the world is good, and we don't get as much of that as we'd like * People reading this post may be able to form their own takeaway, which is something like "man, we really should be checking more comprehensively whether dogs can talk.", and ideas like that are worth sharing. I think there are some significant problems with this, however: 1. This post doesn't really advocate boggling at the world, instead it makes specific, strongly worded claims that there is already decently meaningful evidence of something. 2. This post does not really advocate for that takeaway, and to the extent someone thinks it does, it's certainly far from the most prominent message. 3. If LWers took-this-to-heart, there are likely between 10,000-1,000,000 things of similar potential interest and evidence base that could be shared in this way, the vast majority of which are highly unlikely to be proven out. Furthermore, in addition to thinking that if I've guessed at Raemon's motivations correctly, they're based on flawed reasoning for the above 3 reasons, I think there are other factors that make this actively damaging: 1. This makes strongly worded claims that are either untestable (assumptions on LW readership's priors in title), or not nearly supported by the evidence. 2. It employs a number of techniques that, to some, are seemingly effective at actively misleading them (see bullets 2 and 3 of my previous comment, I also find a number more employed in the author's comments). 3. This, to me, pattern matches with LWers having much less critical readership than expected, and than should be warranted when confronting a post of this type, which I've observed in multiple prominent instances over the past ~3 mon

Plenty of concern was raised in the comments, have you gone through all of them and all the replies?

I'm aware of comparative cognition, the people posting the pet videos are participating in ongoing research at the Comparative Cognition Lab at the University of California, San Diego. They give a description of their methodology, but the status updates appear hidden to ensure integrity of the data.

Short recap of the comments: This is a very new thing, early-stage science often looks like messing around, so don't expect lots of rigor so early. If they had a paper, I would post that. On the balance of evidence, the videos seem to be made in good faith, I don't think it's some staged viral crap. Don't discount evidence just because it's normie YouTube vids. The main claim is that there's something interesting going on that makes me suspect dogs could possibly produce something that looks like language. I'm not claiming certainty on that or on the level of dogs' supposed language ability, it's research in progress, but I think it's exciting and worth studying.

Thanks for responding, and also for illustrating all the issues I have in your post in a compressed way. Basically, what you're saying is:

  1. Something new and exciting is happening
  2. There's not a lot of evidence BUT

I think this community should be able to see the issue there. (To also be polemical, occultism was also something that was new and exciting in the 19th century, with many intellectual of their time spending their evening around a turntable, most of them also in good faith when they reported paranormal activity.)

1. is being conditioned on something really happening in 2.

But 2. has a lot of issues and you know it, and I think you do a bad job at convincing me. Your justifications are:

  • "There's a study ongoing": good, I haven't read the methods and it's probably interesting, but there are no results yet. Maybe you could summarize it to people then if you want to introduce us to this work?
  • "early-stage science often looks like messing around": yes, but many other things also looks like messing around, and really are messing around in the end
  • "videos are made in good faith": Clever Hans, the most infamous exemple in comparative cognition, was literally also in good faith
  • "don't discou
... (read more)
  • I agree I should've summarized the study methodology in the article. For some reason I expected people to click the links and actually read and watch everything (this is not a knock on anyone, one shouldn't expect that when writing articles).
  • There is a lot of evidence, it's just weak and easy to misinterpret, and it's in the form of youtube vids, which goes against aesthetic sensibilities of what "evidence" looks like. If you want to have a holistic picture, you'll have to actually watch a lot of them, I'm sorry.
  • I think it's quite obvious that the evidence here has a rather different shape from 19th-century medium reports or grainy VHS tapes with a blob that is claimed to be the Loch-Ness monster.
  • Being able to tell which "messing around" is likely to be fruitful is a meta-rational skill, but it can be done. Somewhat. To me this has a "shape" of something that could be fruitful, but I can't transfer the pattern matching inside my brain to yours.
  • Vids being done in good faith doesn't preclude clever hans or overinterpretation. It precludes the vids being fake or staged.
  • I and others addressed multiple times the dangers of seeing phantoms in noise and operant conditioning. I don't
... (read more)

I think both of us made our arguments clear, so instead of answering point by point, let me give a quick holistic response that should summarize what I think, and provide a general interesting point of view on animal cognition.

(Maybe you know about the following, but I think it is interesting enough by itself to be presented here to other people)

Corvidae are very intelligent birds. There's ton of evidence of that. You can read studies that test how they can solve problems, you can watch tons of youtube videos showing them interact with their settings and with other animals. These videos are all made in good faith, showing birds evolving in ecological or lab settings, and demonstrating their intelligence.

As you put it, you can build an "holistic picture" of them as very strong problem solvers.

Then comes this observation. You see crows dropping nuts on the road. Cars go over the nuts, crushing them. The crows delight themselves with the opened nuts.

What do you conclude? That the crows are using the passage of cars as a way to break the nuts open? Considering the abilities demonstrated by these birds, it seems like the logical hypothesis to me.

And so this was that a lot of people thou... (read more)

2StyleOfDog
I think we're mostly in agreement, and I'm not disputing that it pays to be careful when it comes to animal cognition. I'd say again that I think it's a meta-rational skill to see the patterns of what is likely to work and what isn't, and this kind of stuff is near-impossible to communicate well. I've read about the car-nutracker thing somewhere, but without the null result from research. If you had asked me to bet I'd say it would be unlikely to work. But it's illustrative that we both still agree that corvids are smart and there's a ton of evidence for it. We just don't know the exact ways and forms, and that's how I feel about the dog thing. There's something there but we need to actually study it to know the exact shape and form. I don't think it will be niche because it's already not niche, considering the massive viewership. But your 1% figure sounds about right as a higher bound, given the sheer number of dog owners, the amount of work required and people's low desire to train their pets. A cursory google search says 4% of US dog owners take a training class, so serious button use will have to be a fraction of that.
1Slider
Being interested in the lethality of cigarettes or the likelyhood of climate change would also deal in being interedsted in issues where high-scrutinity or high social status stances are not available. I do think that people have an easier time reading a bit too much into it because the buttons are so legible for humans. The buttons were developed for communication needds for deaf people. Would not pressing buttons in that cotext qualify as peech? What is the relevant difference to make that not apply there? I am wondering whether this is a too short conversation. This would probably be of good length. To me it feels there is a discernible difference in the level of vividity that the dog follows on the stories. Also On taht linked place the dog is "confused" by a new word which to me reads as a behaviour about actively trying to keep on board with the story and adaptation to expand vocabulary by talking (if not by definition which the human tries to hand over). To my underdstanding the core vocabulary needs to be bootstrapped by reinforcdement anyway so that doesn't make it a disqualifying attribute. YOu can't learn chinese with only the help of a dictionary in chinese. But until the dogs have a primary language they can not keep developing their language skills by simply talking, it needs to link to their "actual" life. The interesting thing about K'eyush being commanded around is that it is more like a negotiation. For a command the dog can react by complying, saying "no" or saying "why". And when they say "why" then giving an explanation can give compliance without repeating the command (but then the dog can also think the reasoning is a rubbish one): I do wonder on what basis judging whether other humans are talking a language the observer doesn't understand vs speaking gibberish. And when tourists are aboard don't they use "bag-of-word" kind of communication a lot? But also doesn't that actually serve for the genuine thing? 

I wonder if this could be made more scalable by having a dog who has learned this teaçh its puppies. 🤔

8StyleOfDog
I've had that thought as well. So far it doesn't look like dogs spontaneously use the buttons to communicate among themselves (they mostly don't have the opportunity though). Some of Bunny's neighbor dogs' owners got on the button craze as well, so maybe we'll see some of that. I think the mama dog would still likely teach it to puppies as a way to communicate with humans, even if it's not used as a way to talk between them.
2ejacob
I'm not aware of any existing "dog culture" that is passed between generations, but I'm not aware of any reason it's impossible either. I suppose wolf packs might teach the young adaptations relevant to local conditions, but it's not clear to me how to use that hypothetical evidence.
[-]weft130

My dog does this unusual roundhouse butt attack when she's playing with other dogs. It's unusual enough that people comment on it.

I've definitely noticed other dogs start doing it too after playing with her a bunch.

There also SEEMS to be a thing where in e.g. Berkeley the dogs at the park play quietly. I wondered how they taught their dogs not to bark while playing, because this is NOT the case in midwest dog parks. But apparently it's "cultural". If the dogs don't bark at the dog park you frequent, your dog will also not bark.

3Viliam
Imitation seems like an important part of teaching/learning, and apes have famously high ability to imitate, and humans are the best imitators among the apes. This may put a limit on how difficult concepts a species can transfer to the next generation. Humans love copying what others do, and that may be useful at learning habits that are not immediately useful, so the learning at the beginning is not rewarded by anything other than being intrinsically desirable. Also, training animals is difficult; I probably couldn't teach a dog to communicate. How much would belonging to the same species help the dog to teach its puppies?

An interesting data point for putting a ceiling on an animal's ability to produce language is this lecture discussing various animals (mostly apes) learning American Sign Language (starting at ~1h18m).

My take-away is that some chimps may be capable of remembering and reproducing words, but not string them together meaningfully: word order is effectively random, length of sentence does not correlate with information content. The lecture discusses (at about 1:34:56) the turning point of the field when H. Terrace was unable to reproduce the generally optimistic findings, e.g. about Koko using "syntax", when teaching his own chimp American Sign Language and then published Why Koko Can't Talk.

If the above is correct, my prior would be for dogs to be at or below the level of chimps - which would still be an interesting finding. What the history of the field highlights, however, is that many humans really want animals to be able to speak AND humans are great at pattern matching meaning into arbitrarily strung together words, which decreases my confidence in anecdotal evidence without more rigorous studies.

4StyleOfDog
That seems to be the case with dogs, and it won't surprise me if they never progress much further than that.

I've updated the article to include a more in-depth explanation of the study design and philosophy instead of just two links (I suspect almost nobody clicked them). Also added responses to common criticism and titles and short explanations for video links (I suspect a lot of people didn't click on most of the videos). Also removed the revolution part.

If you've already read the article, I suggest you read the research and criticism parts under Bunny and watch the new Stella video I added, which is more representative of the kinds of videos that led me to watch the dog space more closely. All of the good Stella stuff is on Instagram, but not on YouTube.

If the folk wisdom of "pigs are smarter than dogs" is true, this suggests it should be possible to get a pig perspective on the pork industry. Could be a big deal for animal rights (but it's also highly likely that a pig/cow/etc taught how to say "please don't eat me" will be ignored or rationalized away like everything else).

It is commendable that OP put a lot of work into this post, but tbh it does seem like many claims are way too overconfident given the evidence. I fear the "specialists in field X are grossly incompetent" is a frequent bias on lw, which is why not many people have pointed out the problems with this post. 

1) Animal researchers have engaged with these type of videos; that they are not in awe about it, could also mean that they do not find it impressive or novel. Here is a good summary. It did not take me long to find this, and this link (or similar ones) should not be the 81st comment. 

2)Yes, doing research on elephants is impractical, but that has nothing to do with doing research on dogs.Many animal cognition researchers have dogs and are totally happy and willing to try to teach their dogs language in their free time.  

3)There are lots of studies with insane amount of resources poured into them with the goal of teaching animals language. Take this study in the 60s where they tried to teach a dolphin language by filling an apartment  with water, having the handler live with the dolphin, and giving him an occasional hand*** . (yes, you read that right)

4) Bunny app... (read more)

2Ben Pace
At what betting odds?
1MichaelLowe
I am offering 2:1 odds in favour of the other side.
4MichaelStJules
(...) Are you still offering this bet? I'm interested. To clarify, you mean not just that the consensus will be that such studies find no (strong) evidence for episodic memory, but that dogs (in such studies) do not have an episodic memory that they can communicate like claimed in the post at all? And, can you clarify what you mean by "like claimed in this post"?
1MichaelStJules
Have you seen unexpected question tests of episodic memory in animals? Some examples: 1. Mental representation and episodic-like memory of own actions in dogs 2. Animal models of episodic memory (see the section "Incidental Encoding and Unexpected Questions") 3. Episodic-like memory of rats as retrospective retrieval of incidentally encoded locations and involvement of the retrosplenial cortex
1StyleOfDog
Fully agree on the bias part, although specialists being incompetent isn't a thread in my article? There's an entire aside about why some research doesn't get done, and incompetence isn't among the reasons. 1. I've read the Slate article you linked, and I think it's good. I don't see anything in there that I disagree with. The article is from 2019 when the amount of evidence (and importantly the number of people who successfully replicated it) was just one Instagram dog. Even back then in the article scientists are cautious but lukewarm and want a more rigorous study. Now we have a more rigorous study running. All this stuff has been addressed in the comments and in the updated article. I'm quite adamant that misinterpreting dog output is the primary danger and I don't claim confidence in specific abilities, precisely because we need more study to determine what's real and what's confirmation bias/misinterpretation. 2. That wasn't a point about dog research, it was a point about dynamics of what kind of discoveries and research gets made more often. 3. "in the 60s" for social/cognitive/psychology-adjacent research has to be a bit like "in mice" for medicine. Either way, people try to do something and fail, 60 years later someone comes up with an approach that works. That's a completely normal story. I thought about taking you up on the bet at 3:1 but I don't like the "vast majority" part. I think it's too much work to specify the rules precisely enough and I've spent enough time on this already.
0Slider
The anxiety reduction was more improvised to surprising needs rather than thought to have a cognitive payload by itself and thought off beforehand. It is a notable thing about it but not excatly a resource that measures effort extent.

From my experience, dog owners and people in general tend to see human-like patterns in dog behavior even when more simple explanations ("looking at the world through dog's eyes") have better predictive power.

As many dog owners would tell, dogs learn how to show their needs (staring at a door to be let outside, bringing a leash when they want to go for a walk, or a toy when they want to play, etc.). I find it pretty impressive that Bunny learned to communicate her needs through the board, but from what I've seen, I don't really think there's much evidence for some of the complex behaviour this post and the owner suggested there is.

I think that she learned which words or parts of the board to use in specific contexts (questions about time and place, mirror) but I don't it shows she can tell time or recognize herself in the mirror (AFAIK, no dog can do that). On the other hand, I think that she can somewhat describe the world around her and I'm really curious what is the extent of this ability.

A lot of the behavior from the videos (eg. the negotiations) reminds me of a dog training technique when you let the dog try things and reward them when they do the thing you want. Dogs can ge... (read more)

This raises the natural question: what if you gave an ape the buttons, and taught it from childhood, and put parent-level effort into it, not "70s research”-level effort? Perhaps the answer would surprise us.

The bonobo Kanzi had something very similar ("lexigrams"). And his sister Panbanisha was born in the research center and grew up with the lexigrams. As far as I'm aware, the research never seemed to generate extreme attention, so probably the learnings remained somewhat limited?

Bunny is quite obsessed over her bowel movements (how Freudian) and about her owners' poop cycle.

As a youtube comment on the video points out, maybe the dog is just trying to be polite by imitating the conversation topic of its family. People probably ask their dog all the time about whether the dog needs or wants to go potty. 


 

I wasn't aware of this and it indeed looks very similar! Sue Savage-Rumbaugh who started the work did appear on Oprah and gave a TED talk, so she generated some attention. And in the Oprah clip she says she's living and sleeping with the bonobos 95% of the time, and is raising a small one (Teco) being exposed to lexigrams from birth. That's about as close to "parent-level effort" as one can get!

Unfortunately in 2012 she had a spat with her bonobo center over ethical concerns, which included a shadowy group of whistleblowers named "Bonobo 12" among other drama. According to her wiki she has left the center and is "embroiled in several legal battles" with her former employer. Their current research page provides scant information on what they're doing these days. They're pretty active on Facebook though, but it's mostly begging for donations and asking people to buy stuff of their amazon wishlist, with no mention of any research. I think they're struggling to stay afloat.

I can't find any videos of Teco using the lexigrams, but here are some of the other apes:

First False -Belief Test Passed by Ape (Liz from the vid apparently was kicked off the project in 2014 and her dying wish was t... (read more)

Continuing the previous examples, I think there are old experiments to get animals to talk which are maybe why research into this area has been less than one might expect for a while (which are different from the examples given in the OP).

I have a dog and was aware of these people. My lack of reaction was due to a default assumption that this will turn out to be Clever Hansian once science brings its customary rigor to bear.

If not, I wonder if I will conclude that it's unethical not to teach my dog how to communicate.

By induction: it's obvious that you can teach most dogs to press a button for "Food", "Outside", "Pets", "Play", and they won't need to rely on clever hansian subtle cues from you to express desire for food, they're already doing it just with a different modality.

For more abstract concepts like mad, happy or concerned or the ever-popular "love you", you're supposed to model those buttons when the dog is feeling these things, so you need to understand your pet very well, and it's easy to delude yourself into thinking you understand when you don't.

So the danger is over-interpreting their output. Was this sentence intended or just random babbling? Does the dog understand the word differently from what it means in English? E.g. "bye" seems to become a verb meaning "leaving", "love you" is used for affection but obviously doesn't reflect deep understanding of the human concept of love. And every pet is going to develop its own idiolect with the owner, further complicating things. Alexis seems to be very aware of those issues and UCSD study claims (or hopes?) to have hundreds of participants, the large scale should help separate signal from noise.

There's also some conflict of interest with Leo Trottier who has sites selling the buttons, pet games and learning material while being part of the research team.

There are some real concerns, but I still think there's something real here.

So the danger is over-interpreting their output. Was this sentence intended or just random babbling? Does the dog understand the word differently from what it means in English? E.g. "bye" seems to become a verb meaning "leaving", "love you" is used for affection but obviously doesn't reflect deep understanding of the human concept of love. 

Interestingly, there's an argument that human infants also learn language by their parents over-interpreting their input, with the infants then adopting those interpretations as true. So one could argue that even if over-interpretation happens with dogs, that only makes it a process similar to human language learning, with the parent/child and owner/dog creating a shared language game.

Our first type of example comes from our own data concerning Zulu infants of between three and four months of age interacting with their mothers, and suggests an answer to this question. [...]

As noted above, there are times when a caregiver will want an infant to fall silent, or in isiZulu to 'thula'. Zulu children are traditionally expected to be less socially active than contemporary Western children, to initiate fewer interactions, and, crucially, to show a

... (read more)
4Josh Smith-Brennan
I tend towards the 'Clever Hans' hypothesis on this one too. Human language skills are much more complex due to our neurology, as are our cognitive abilities. That doesn't mean dogs aren't communicating at some rudimentary level, as they are at least able to recognize patterns and determine likely outcomes, even if they are at a more reflexive instead of reflective level. Hot dogs for example satisfy reflexive hunger based desires, instead of reflective philosophical desires to be seen as intelligent. Of course one reason more of this type of research isn't done in academia anymore is that no one wants to shell out tons of money to disprove a lot of 'crazy' hypothesis. Proving things tends to be seen as more productive, especially in the short term, although the long term implications of disproving multiple hypothesis is that it tends to point one in a better direction for particular areas of potential for truth.  At any level, money is always a polarizing factor in the consideration of 'scientific validity', whether it's selling buttons to 'Dog Parents' or selling data processing clout to Corporations and their beneficiaries. Just making something more convenient to study though doesn't necessarily make the overall scientific community better I think, as it will just provide more resources to easier topics to research, not necessarily more important topics. In this case though, I think animal cognition is important to study. In a Cohumane sense, it can provide clarity on issues like the ethical use of animals for food sources, experimental subjects, and issues regarding the destruction of the natural environment and extinction of animal and plant life. Is animal suffering even a thing first of all (of course, but does that extend further to insects or plant life?) and if it is a thing, what kinds of considerations are there for our past and present treatment of these life forms as 'resources' akin to minerals we mine out of the ground ( or for that matter, for '

When you say "Clever Hans" are you talking specifically about the handler's subconscious cues determining what the dog does? I think that's very unlikely, in a lot of interactions you can see an exchange where the pet is supposed to make a decision - the owner doesn't know the right answer! When Bunny presses "ouch stranger paw" to indicate a splinter in her paw, how was the owner supposed to "influence" that, without even being aware of the splinter? Some interactions are owner asking a question with a defined right answer, but there's clearly much more than that happening.

they are at least able to recognize patterns and determine likely outcomes, even if they are at a more reflexive instead of reflective level. Hot dogs for example satisfy reflexive hunger based desires, instead of reflective philosophical desires to be seen as intelligent.

I agree that on the continuum animals are much further towards reflexive, but I just want to point out that most people inhabit reflexive states very often. Maybe it's normie bias cropping up :D But most people aren't obsessively reflective. A lot of self-reflective smart people have trouble understanding what it's like to be someone far ou... (read more)

I also mentioned Clever Hans, and you made a good point in response. Rather than sound like I am motte-and-baileying you, I will say that I was using "Clever Hans" irresponsibly imprecisely as a stand-in for more issues than were present in the Clever Hans case.

I've updated in the direction of "I'll eventually need to reconsider my relationship with my dog" but still expect a lot of these research threads to come apart through a combination of 

  • Subconscious cues from trainers - true Clever Hans effects (dogs are super clued in to us thanks to selection pressure, in ways we don't naturally detect)
  • Experiment design that has obvious holes in it (at first)
  • Experiment design that has subtle holes in it (once the easy problems are dealt with)
  • Alternative explanations, of experimentally established hole-free results, from professional scientists (once the field becomes large enough to attract widespread academic attention). Like, yes, you unambiguously showed experimental result x, which you attributed to p, which would indeed explain x, but q is an equally plausible explanation which your experiment does not differentiate against.

This is based on a model of lay science that tends to sh... (read more)

2Dustin
While I haven't done a rigorous study of the effect, my gut feeling is that the vast majority of suggestive and interesting phenomena eventually fall apart due to these exact reasons. This is why I do not give much thought to this sort of stuff.  Ideally, I'd do a rigorous study of initially-interesting-but-later-fell-apart-studies or find someone else who has and then maybe I'd be better able to spend my cognitive resources...
1Josh Smith-Brennan
I'm thinking of it more as a variation on that idea. I think it's possible that in the button case, the buttons could be stand ins for cues from an owner. Simply training with the buttons over time would modify a dogs behavior based solely on the presence or absence of the button. Work the toys in, and you're simply training a dog to respond 'correctly' or 'incorrectly' in the presence or absence of a 'thing' associated with a particular sound and visual cue. I think of lion tamers and other animals who have been trained to do tricks, like unicycling bears, and dolphins and orcas that are trained to jump through hoops when I see these types of things. Circuses things. It would be cool if dogs really could understand human language to the point where they could communicate back, but I just don't believe it's the case. Our brains developed complex areas devoted specifically to making and decoding speech, which depend on specific structures of our throats. These are all things all animals except for humans lack. I sometimes think modern life is just learning to inhibit our reflexive actions, which are based on our reflexive states. We do this inhibiting not only because of social training (like dog training or obedience school, what to do, not why to do it), but also because of our reflective ability (why or why not to do it), which allows us to (theoretically) do things like putting off getting rewards now for more rewards in the future, or to develop interpersonal relationship skills to try to work with people we reflexively dislike. These are things dogs can't do, as they involve ability to think abstractly about concepts like time and etiquette. Dogs can be trained to inhibit their reflexive behavior, and 'act' a particular way (don't bite, don't bark, come to me when I make particular sound or give a particular gesture, or press this button when they hear a particular sound) but not to reflect on why it's important to do so.  If they can be taught to reflect l

I found an interesting video of a cat using the similar buttons:

The cat insisting to go outside looks like a lot of intent of the side of the cat. Pressing the "Love you button" while pleading is also funny. 

5ChristianKl
Another great video that's also has no room for Clever Hans. The cat owner repeatidly asks the cat whether she wants food play and the cat says first "no" then "all done" and then "later" as the owner repeats the question with the idea that the cat first pressed "no" when the cat meant "yes".
1Slider
I thought that "outside catnip" was an alternate activity so that would be a no to a different question, so that is only 2 nos to "food play". The human also seems insistent and for some reason I attribute uncomfortableness to the cat, so I see "later" as a "yes but later" compromise on being directed to food play.
2ChristianKl
catnip is something that the cat can eat so it might be food play that happens outside, but I'm not sure about that.  My main point is that it's very hard to imagine this interaction happening if the cat presses buttons randomly or presses them to fulfill expectations of the human. 
1Slider
I think the cat is reacting to the opinion of the human. You might be mostly concerned that the main driving force would be "which button I am supposed to press?". But on the repeat the cat is much slower to say no. It feels like saying "no" is sligthly punished. And the human is not exactly hiding their enthusiasm about directing food play activity. I read into that a balancing between internal desires vs social expectations which would be even more complex than the "simpler effects can explain nothing interedsting going on" approach but this is not strongly screening off non-concious suggestions from the human. The human is very visble to the cat, the cat is very interested in the physicality of the human and we don't see footage of the human to evaluate what signals they might be beaming or not. Sure I think the humanis not comanding the presses of buttons but advdertisers target kids because they can induce money spending on toys and they don't command their parents wallets either. It's like a kid saying "Are we going to McDonalds?" (cat feeling something "but I am feeling so greasy already, let's not"), its not disambiguation, its negotation/influence. Another imagination pattern would be when police ask you 10x times whether you did the crime that it might elict a false confession. Having a pattern of "Do you want to do it?", "No", "Wrong. Do you want to do it?","no", "Wrong. Do you want to do it", "Maybe", is not an expression of enthusiastic consent. I felt I coudl pretty rapid fire come up with these scenarios, they are fairly different and their applicability is not clearly absurd so "very hard to imagine" is not how I experience the interpretation challenge level.
2ChristianKl
To me it's very hard to imagine such an exchange without the participants knowing the words that are spoken.
1Slider
The analog starts to get a little far but I can imagine atleast two scenarios where atleast half of the interaction doesn't consist of words. First one. Somebody hitting their TV until it starts working again. The TV sure as hell doesn't know what the interaction is about. As hitter projecting anger into the situation is a common cognitive fallacy, but low percentage of the time is is a way to wiggle out of the error state of not having a working TV. Another would be an exchange like "that *itch", "ermh...","witch", "ermh","girl","erhm", "woman". Is the one doing discouragement on the connotation requesdting a reformulation or neutrally rejecting a wrong answer? It doesn't involve an established word and it can involve stuff like eye rolling and stuff that is harder to delineate into deiscrete chuncks. This would be a instance of clever hans as the formulator doesn't need to know what PC-filter is appropriate for the situation.
2ChristianKl
The analog starts to get a little far but I can imagine atleast two scenarios where atleast half of the interaction doesn't consist of words. Of course if the interaction is not about words then it's not important that the participants of the inteactions understand words. 
1Slider
I interpret it not being part of the pleading but gratitude for a concession or like gratitude for willing to withstand danger for their amusement.

In the following video the dog owner tells Bunny that another dog is "home". Given our human language it's clear the dog owner meant that it's at the home of the other dog owner but given that home has always meant the dog owners home the dog seems to understand that the other dog is around.

You wouldn't get such misunderstanding with the clever hans effect.

4ESRogs
You could get it with the seeing-shapes-in-clouds effect though.
1Slider
I am interested in from where such impression are formed from. To my eye the dog doing the looking and looking back is a form pointing at the house and the dog is expecting to walk around the house. The human is not having any of the walking. This combined with expressing verbal uncertainty makes the dog think that a miscommunication took place. Why is the dog expecting a a walk? Because the dog knows that the human knows that selena is exciting. The human being disintresed to walk around is in conflict with the theory of exciting things to be found around. To me it is not super clear what makes a look "pointy" and which is only to have a better visual access to something. The video being sped up at the portion suggests that to the editors eye "nothing is happening" communication wise. The clarificatino of "no selena to be found" suggests that outside and home mean more things like "elsewhere" and "here". And this is an example where the dog is setting up the definition goal posts and the human is following that lead. Here the interaction pattern is "express confusion/doubt","provide suggestion","get on board suggestion". "home", "outside" could have aspects of "inside" vs "outside" and it would be curious whether "elsewhere inside" is ever relevant to the dogs life.
2ChristianKl
One interesting feature of the interaction is that the dog is first going to look for Selina and only confused when Bunny didn't find her. That suggests that their mental model of what's going on around him isn't pretty good. To the extend that elsewhere inside is relevant I expect it's relevant when the dog isn't at their home and thus doesn't have any access to any buttons to express themselves.

I found this one particularly impressive: https://m.youtube.com/watch?v=AHiu-EDJUx0

The use of "oops" at the end is spot on.

4ChristianKl
Yes, that video seems very unlikely to happen without the dog having some idea of what the buttons it presses mean.

So this is hearsay, but a zoologist friend of mine has a zoologist friend of hers :) and that person has spent a while observing cows. She says cows are "smart" but kind of live in a different timescale than we do, a "longer" one. I have no idea what this means for any particular cow, yet what I find interesting is the notion of different timescales and how it affects ethology research.

OTOH, I have geese, and I do wonder about that Hans thing. Geese are 1) super fun, 2) variable in aggression, maternal instinct strength, shyness etc. Sure they can't comm... (read more)

Different timescale is an important part of it. In a lot of those pet buttons videos you can see that the pet takes an uncomfortably long time to respond. They would often go away from the board to "pace" and then after a solid minute come back to finally respond. In Bunny videos this part gets fast-forwarded to make it watchable, in Billi vids you can see the cat go away and sit with her tail swishing, you can practically see the gears turning in her head.

It's not really suprising to me that dogs are able to have thoughts of the complexity of "I was outside this afternoon". 

Given what Neurolink is doing with the chimpanzees it will be interesting how that tool could be used to let animals express themselves. 

I still appreciate this post for getting me to think about the question "how much language can dogs learn?". I also still find the evidence pretty sus, and mostly tantalizing in the form of "man I wish there were more/better experiments like this."

BUT, what feels (probably?) less sus to me is JenniferRM's comment about the dog Chaser, who learned explicit nouns and verbs. This is more believable to me, and seems to have had more of a scientific setup. (Ideally I'd like to spend this review-time spot-checking that the paper seems reasonable, alas, in the gr... (read more)

This post has successfully stuck around in my mind for two years now! In particular, it's made me explicitly aware of the possibility of flinching away from observations because they're normie-tribe-coded.

I think I deny the evidence on most of the cases of dogs generating complex English claims. But it was epistemically healthy for that model anomaly to be rubbed in my face, rather than filter-bubbled away plus flinched away from and ignored.

The owners over-interpret and anthropomorphize the button "speech"

The is the biggest danger in my opinion. Hopefully with rigorous analysis during the study and specifically set up experiments we'll be able to understand better at what level of communication the dogs actually are.

I think this is most of what's going on here. I'd guess that the owners have in fact taught their animals some new words and associations, but that they're way over-interpreting what the dogs are "saying".

I don't think this question has been asked before here but I'm wondering about the effect of the age of the dog at the start of the training.

  1. IIRC feral children that learn to communicate as adults have missed their critical period for grammatical expression. In effect there oral skills are akin to "dog speak". Maybe if the dog starts at age 0 it can lead to more complex word associations. If indeed there is a "grammatical critical period" for dogs even though dogs don't ever have to use grammars in their natural state that would be a breakthrough in my

... (read more)

great post, two points of disagreement that are worth mentioning

  1. Exploring the full ability of dogs and cats to communicate isn't so much impractical to do in academia; it just isn't very theoretically interesting. We know animals can do operant conditioning (we've known for over 100 years probably), but we also know they struggle with complex syntax. I guess there's a lot of uncertainty in the middle, so I'm low confidence about this. But generally to publish a high impact paper about dog or cat communication you'd have to show they can do more than "condi
... (read more)

Anyone have any particular intuition for whether a dog would be likely to be able to learn to use a device that combined phonemes as its basic constituent parts, rather than entire words?

I'm working on a device that affords this possibility. I have a functional prototype that is mouth-held and -operated; it uses a pressure sensor (measuring bite) and gyro/accelerometer to produce (at the moment) a limited selection of vowels and consonants; they can be varied by moving the head around and biting down.

My inspiration is these button boards together with the ... (read more)

3ChristianKl
When it comes to teaching the meaning of the words with the buttons it seems much easier to introduce new words and the dog can easily see it when the button is pushed. It's going to be much harder for the dog to learn sounds on their own.  It might be worth writing out a plan of how you think all the individual subskills can be taught. 
2Raemon
How easy is it to ship a prototype to LW readers who have dogs?
1_nuchi
(new account because I couldn't log in again with github) Good question — difficult at the moment because I only have the one prototype, and I'm still updating it. Though I'm open to show-and-tell and/or short-term loans for readers in San Francisco.

Note: if you convert this post from markdown to LessWrong Docs (available at the top of the edit-post-page), and that you paste the youtube-links into the editor, they will automatically turn into embedded links, which you might find nicer. (You can experiment with it and see if you like it without committing to re-publish the post)

1StyleOfDog
I think I'm gonna keep the links, YT embeds are pretty large and break the flow of text.

If this turns out to be basically true, then what about wild wolves? I think there is a strong case that the capacity for this sort of communication to have been bred into domestic dogs as a result of humans selecting for e.g. better overall intelligence and ability to understand human commands.

Another option is that wild wolf packs have the capacity for this sort of communication but don't (unless we've simply not noticed it) and this seems much less likely to me, for the sole reason that being able to communicate in this way would give a very large advan... (read more)

5ChristianKl
Humans who learn to drive cars have amazing cognitive machinery for it despite not being evolved for it. The human brain being able to edit the sense of time when driving cars seems pretty amazing. Similar things go for reading.  The ability of brains to learn is very general.  Given that the buttons work for autistic humans who are not capable of learning sign language learning to use them is likely less cognitively demanding then learning to use sign language.
1J Bostock
I more meant "keeping around cognitive machinery which is capable of this" without making use of it. Given that wild wolves use (relatively) simple hunting strategies which do not seem to rely on much communication, there doesn't seem to be much need to have a brain capable of communicating relatively abstract thoughts. That doesn't seem to affect your core argument though Good point about autistic humans who can't learn sign language though, I hadn't considered that. I guess my model of autism was more like: "Autism affects the brain in lots of different ways which is able to knock out specific abilities (like speech) without knocking out other abilities (like the capability to have and communicate complex thoughts, which would not have evolved in an animal without speech)" than drawing on some amount of general purpose computing behind each one. I haven't studied autism enough to know if this is correct.
2ChristianKl
I would expect that moving fast running fast through a dense forest does take a lot of cognition to know where to go effectively. Terrain knowledge gets likely also used. 
1Dirichlet-to-Neumann
If that were the case, don't you think that animals that actually run faster than humans in dense forest would be more intelligent ?
2ChristianKl
Which animals are you thinking about?

So we could create another intelligent species on Earth by combining selection and designed culture. Any risks?

1dr_s
"Where are my testicles, Summer?"

Other non-primate animal doing similar: Alex the gray parrot https://en.wikipedia.org/wiki/Alex_(parrot) He also made his own terms as "cork nut" for almond.

I was surprised not to find K'eyush as an example. For anybody wondering about interprtations one can compare with his buddy Sherpa. Sherpa is clearly  alot behind K'eyush but does have a notion on when it is her turn to "speak" althought with sherpa it doesn't seem to have specified content.

The way the owner subtitles the videos also is a interesting reference point. They are half-littered with jokes which are no an honest interpretation form the handler but also do have genuine disambiguation. Perhaps importantly by following the videos one develops... (read more)

5StyleOfDog
I've checked him out, this kind of animal "speaking" always seemed like just a fun party trick, you can read anything into dog vocalizations. I expected nothing there, but in this clip (and only at 1:15) I can see him actually trying to mimic speech. But the rest is just reading on tea leaves. Like, there's no chance in hell that the dog would know the word "werewolf" and know to use it in context as happens later in the video. https://youtu.be/dxBpESjiefo?t=75 Siri does understand his "hello", but "I know he can" a few seconds later is a ridiculous reach. The dog was just repeating his "hello" vocalization. https://youtu.be/yqCMA2Pm-IY?t=85 I can totally buy that the owner and the dog developed a private language of barks and whines loosely based on English. And yet dogs' vocal cords and mouth are not made for producing human speech, while their paws are quite good at pressing buttons. If you spend all your time trying to make your dog actually vocalize English, you'll never progress to the interesting stuff, and you'll delude yourself into reading whatever you want into the sounds. I think that's a dead-end as far as animal language research is concerned. The human equivalent of not being included in the culture is not homeschooling, it's growing feral being raised by wolves. And pets are surrounded by plenty of language. It only starts looking more interesting than blind pattern matching when you give them tools to actually use language instead of listening passively.
1Slider
I understand that when a clear interpretaion is not super handy it is tempting to give up on it. However human mother will babble with their babies, they don't deem their children incapable of speech if they don't speak like they do. Siris tea leaf reading is not partial to dogs. There is also the mirror effect that because the buttons give them "perfect" vocalization there can be temptation that they mean more stuff with them. Making a dog pronounce English at a human level seems impossible but that is why the barks and whines. There are regional human accent with humans with native languages have certain dialects in english. These can already hinder comprehension so atleast similar if not greater patience should be extended to "dog english". If it could be clearly established to which word each dog sound corresponds then the vocals would be as good as buttons. It is proper for a trainer to treat the distinguishable sounds as proper dialogue lines. Part of the buttons is that it makes the human confident that word was meant (human has less chance to think it is a meaningless wiff and has more pressure to treat it as intentional communication) I did mean that homeschooling would be an adequate substitute but there is a diffrence of keeping your children sensory depirvation chamber in a cellar and giving a competent homeschooling. The expectations on what dogs can do and I tried to get afforded to do is very limited. We don't diagnose people with dyslexia if they have never been introduced to the alphabeth (I guess it has forms that manifest that don't depend on symbols, point is shortcomings need a baseline to stand out). The pets are around language ubt are they actually participating in the language games, is management of things that matter to them done throught language? People do not babble with their dogs and in that they are on uneven ground with babies. Sure the dog understand some words, better but hr can sometimes hold coversation to the point that h