Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Think your dog can understand words? This scientist says you might be right

cbc.ca

Post reply on HN
Print view View original

61–67 of 67 posts

Re: Think your dog can understand words? This scientist says you might be right

#61

Post by sonofhans »

observationist wrote: post #54
inglor_cz wrote: post #37

Earlier quoted context omitted.

"What blew my mind was when one of the first dogs to use buttons was confronted with her "beach" button being broken. She started using the phrase "outside-water" to compensate for being unable to say "beach"." This is actually something that the skeptics should address when they claim that all that button pushing is just simple conditioning. Spontaneously creating a logically sound approximation for a broken button…

My own dog has used "outside water good before" to tell me about going potty outside, but over time, outside water has come to mean rain. There needs to be an acknowledgement that conditioning does happen, and that animals who don't learn language can still recognize words as signals without having any deeper understanding. With buttons, animals are brought to understand the words through contextual use and modeling…

Horses are brilliant, and it’s good to hear someone say so. A friend says it would take at most 8 vampire horses to wipe out all of humanity.

Re: Think your dog can understand words? This scientist says you might be right

#62

Post by codr7 »

bityard wrote: post #17
codr7 wrote: post #10

Hmm, I'm not that sure. I did a month of quiet Yoga (no talking whatsoever) and brought my dog with me for the ride. I came to realize that the words are mostly there to help me, she doesn't care. If I switch the word and say it with the same energy/body language, the result is the same. I don't even need words anymore, body language is more effective with my dog.

Dogs can absolutely learn sign language for basic commands. There are books about it. We had some success with ours. The main difficulty is you do have to use some kind of sound to get their attention if they're not looking at you. (Perhaps a clap would do.)

I started snapping my fingers to get her attention.

Re: Think your dog can understand words? This scientist says you might be right

#63

Post by gspencley »

ajuc wrote: post #43
gspencley wrote: post #34

Earlier quoted context omitted.

This speaks to an interesting question that is extremely difficult to answer: "Can pets think conceptually?" A word like "table", to a human, does not represent a concrete. It represents an abstraction, ALL tables, all things that have a flat surface and four legs. When a dog hears "table", or "walk", do they understand the abstract concept of walking, or tables in general? Or do they interpret it as a very specific,…

What would it even mean in practice for them to think about concrete "walking" instead of the abstract concept? Verbs are inherently abstract. When I was with my dog in my parents' house and said "spacer" (walk in Polish) - he was happy and jumped around in anticipation, bringing the leash. He did the same in my grandparents house even though we weren't going to the same place - he just wanted to go on a walk doesn't…

> What would it even mean in practice for them to think about concrete "walking" instead of the abstract concept? Verbs are inherently abstract.

Your example of the leash and the concept of "walking" is an excellent point.

To answer your question, all proper nouns are not abstract by definition. Your name refers to you, specifically and concretely, it does not refer to the concept of "person" or "people."

But to take this further, imagine the following thought experiment:

You have a newborn infant, and you isolate them from the rest of the world (remember this is a thought experiment, so we are not bound by ethical considerations for this purpose). You show them 3 identical staplers. You call the first one "Foo", the second one "Bar" and the 3rd one "Baz."

An infant might, on their own as they grow and mature, start to abstract and conceptualize a concept of a logical grouping here. They might observe that Foo, Bar and Baz have similarities and, being humans, they might even come up with some kind of verbal or non-verbal cue to reference this abstract concept.

But for a while at least, these are individual concretes. If all they ever see is one single concrete "stapler", they may learn to associate the word with the concrete and never even think about abstracting it. But if they do start to abstract it upon observation of similar "things" ... to what degree are non-humans capable of doing that?

You provide a lot of good evidence that dogs can to some degree. But to what extent? What are the limitations?

These are unanswered questions in both science and philosophy.

I hope I answered your question, as it was a valid one. I'm not giving answers to these hypotheses and philosophical conjectures. I'm saying that this is something that is still not quite figured out yet.

Re: Think your dog can understand words? This scientist says you might be right

#64

Post by observationist »

brudgers wrote: post #57
observationist wrote: post #21

Earlier quoted context omitted.

Words are a human construct - language. Dogs don't naturally have words or language - there's no bark that conveys a static meaning. They do correlate noises, like whines, barks, growls, with things they have experience with, and are clever enough to string complex and abstract concepts together. That level of reasoning and communication is not language, however. What the buttons demonstrate is that dogs are capable…

Dogs don't naturally have words or language How do you know that?

You'd need differentiation between coherently patterned noises and a mechanism to pass them between generations. We've been watching dogs for more than 50,000 years, and nobody has ever been able to identify words, or even word-like meanings in the noises dogs make. Dogs don't have language - coherent patterned noises - or culture, outside of what humans carry forward through generations on dogs' behalf.

This isn't to say their vocalizations and body language don't have meaning - they have rich and varied meanings, and that's how experienced trainers and dog owners can observe and interpret dog behavior at a high level of competence. They're just not using language - meaningfully coherent words and grammars.

That's the whole point of the buttons - you can teach them to use words and simple rules of grammar, and get amazing interactions like Bunny talking about "night talk" dreams, or laughing at "play poop" fart situations, dogs planning their eat, walk, settle schedules, and so on. Buttons elevate the level of communication and relationship with the animal, and grants them agency that they otherwise wouldn't have.

As far as non-human language goes, there may be languages developed and learned by pods of whales, orcas, and dolphins. There could even be really old languages, passed down through oral traditions, and scientists are currently researching with AI to see if we can crack cetacean or delphinidae communication. It could also be they have sophisticated "vibes" based communication without highly specific assigned meanings to every vocalization like humans, or some mysterious mixture of unexpected features that's a little alien to the way we think. We know they have culture, and appear to have discrete words and grammar, so this research is really exciting.

Re: Think your dog can understand words? This scientist says you might be right

#65

Post by brudgers »

observationist wrote: post #64
brudgers wrote: post #57

Earlier quoted context omitted.

Dogs don't naturally have words or language How do you know that?

You'd need differentiation between coherently patterned noises and a mechanism to pass them between generations. We've been watching dogs for more than 50,000 years, and nobody has ever been able to identify words, or even word-like meanings in the noises dogs make. Dogs don't have language - coherent patterned noises - or culture, outside of what humans carry forward through generations on dogs' behalf. This isn't t…

nobody has ever been able to identify words

That is a statement about humans.

Not dogs.

Re: Think your dog can understand words? This scientist says you might be right

#66

Post by ajuc »

gspencley wrote: post #63
ajuc wrote: post #43

Earlier quoted context omitted.

What would it even mean in practice for them to think about concrete "walking" instead of the abstract concept? Verbs are inherently abstract. When I was with my dog in my parents' house and said "spacer" (walk in Polish) - he was happy and jumped around in anticipation, bringing the leash. He did the same in my grandparents house even though we weren't going to the same place - he just wanted to go on a walk doesn't…

> What would it even mean in practice for them to think about concrete "walking" instead of the abstract concept? Verbs are inherently abstract. Your example of the leash and the concept of "walking" is an excellent point. To answer your question, all proper nouns are not abstract by definition. Your name refers to you, specifically and concretely, it does not refer to the concept of "person" or "people." But to take…

> I hope I answered your question

It was mostly rethorical, but thanks :)

BTW I think the concrete proper nouns we use (like "Eve") are actually abstractions too and prove some reasoning capability.

At first infants don't realize objects exist. They just see shapes and react to them. When something disappears from their view - it's the same as if it stopped existing. That's why you can play peekaboo with them and not bore them.

To abstract from "a shape that sometimes appears, moves and disappears" to "an object that exists over time even when it's not visible" is a pretty impressive generalization. If AI was able to do it on its own I would be impressed at least. I'd even say it's more impressive than the next step (generalizing from "mom' and "papa" to "person").

Of course we have hardware in our brains dedicated for this stuff and it's probably not conscious reasoning that gets us the object permanence. And we don't remember it in any case. So we take it for granted.

But when you think about it the generalization from "sensations in my eyes" to "objects in the world" is a pretty big one.

Re: Think your dog can understand words? This scientist says you might be right

#67

Post by gspencley »

ajuc wrote: post #66
gspencley wrote: post #63

Earlier quoted context omitted.

> What would it even mean in practice for them to think about concrete "walking" instead of the abstract concept? Verbs are inherently abstract. Your example of the leash and the concept of "walking" is an excellent point. To answer your question, all proper nouns are not abstract by definition. Your name refers to you, specifically and concretely, it does not refer to the concept of "person" or "people." But to take…

> I hope I answered your question It was mostly rethorical, but thanks :) BTW I think the concrete proper nouns we use (like "Eve") are actually abstractions too and prove some reasoning capability. At first infants don't realize objects exist. They just see shapes and react to them. When something disappears from their view - it's the same as if it stopped existing. That's why you can play peekaboo with them and not…

It gets even wilder than that.

There is a psychological concept called "anchoring" which associates two different things perceptually. For example, it's what happens when you listen to a song during a particular time in your life, and then many years later you hear the song and feel all the things you felt in that moment with all of the memories triggered etc.

You make a good point about language itself being some form of abstraction in reference to the objective world. But this conversation reminds me of a famous artwork called "The Treachery of Images." It is a painting of a tobacco pipe with the caption "Ceci n'est pas une pipe" ("This is not a pipe" in English).

Through the painting, we can appreciate that a picture of a pipe is an abstraction that represents a pipe, and is not the same thing as a pipe itself. It is a painting that appeals directly to reason.

The philosophical and scientific question is: when a dog hears the word "walk", or sees a leash, is it conceptual abstraction at play; the understanding of a logical category... or is it emotional anchoring that came about through Pavlovian conditioning and having that word associated with a memory of a concrete?

Post reply on HN