Brilliant.
I have a very clear memory from my childhood. I was arguing with my father that Princess only ran to the door because of the excited inflection in “Princess wanna go outside??!!?!!!?”
So we tried in the most mono-tone conversational voice we could: “Princess, do you want to go outside?” My dad was right. The dog ran to the door. Alas, we did not rush to the local scientific research facility to report our findings…
Edit: if i recall correctly, we also tried saying a bunch of jibberish with the same excited inflection, Princess lifted her head and looked at us, but did not run to the door.
This question inadvertently implies one notion that Blake has been suggesting, that humans, but not dogs or cats, have a theory of mind. “How is that different than when you turn and ask, “Me?” when a stranger just says your name?” (from upthread). We relate the sound of our name with a concept - ourself, “Me.” It seems obvious that if you say, “Rover,” to your dog (assuming that’s his name) and he looks up, and you say, “No, Rover, not you. I mean a different Rover,” it’s unlikely that he’d just go back to sleep. He’d keep looking at you. He does not know that Rover is his name. He recognizes that word by sound and association.
By the same token, during the little flurry of who makes the most mistakes, and how important is it that parrots aren’t 100% correct, this point was not raised and I think it’s important. You could no doubt show people colored objects and ask them to identify the color. Most people would be close to 100% no matter how long the trial lasted. But, if they DID make a mistake, they’d correct it immediately. I’m pretty confident that Alex the parrot never miscalled a color and then immediately said, “No, I mean blue.”
Dickens my black lab (15 years gone now) after awhile would not be fooled. There were certain trigger words that would drive him apeshit-if I said “Do you want to go swimming?” he’d immediately go to the back door where the pool was and start barking in a frenzy. If I said “Do you want to go hockey?” in the exact same tone of voice, he’d start to wag his tail for a moment, but then look at me like I was crazy. (“Baseball” would still drive him nuts as he loved to field tennis balls off my bat.)
What this really gets down to is whether animals have signifieds for the signifiers we use with them-in other words mental concepts to go along with the words/sounds-it seems pretty clear to me from my readings that they often do (tho some undoubtedly don’t). Closely related to this is their ability to solve novel problems, and again it is clear to me that some can, some can’t, be they of a linguistic or other nature.
The flip side of this of course is the sin of anthropocentrism-that there is an iron wall separating us from the beasts and our abilities aren’t merely of a different (higher) degree than theirs, but of a completely different kind. Science for good or bad has always biased itself towards avoiding the Type II error (embracing falsehoods), but if you spend all your time trying to ferret those out, you risk the chance of Type I errors (rejecting the truth). Avoiding anthropomorphism while committing anthropocentrism is just par for the course in that sense-I know why they do it that way, and can accept at least part of the rationale involved, but it still bugs me.
Blake mentioned how “100% of professionals agree” with his position, but in what field exactly? Depending on how narrow said field is, this could be a true statement purely within that field, but meanwhile another group of researchers are approaching the problem from a different angle, start with different base assumptions & defining philosophies, and derive different conclusions, and all the while neither group communicates with the other or even acknowledges their existence (never reads the literature in the other field).
I don’t know if this is the case with animal behavior and cognition, but Blake’s position seems to hew very close to classical behaviorism, whose conclusions and assumptions have been sternly questioned in recent years. Wikipedia indicated that ethology arose from classical zoology, but comparative psychology descended from the psychological disciplines (natch)-but maybe they’re more closely related than I’m assuming.
Quoth Minnie Luna:
That was another thing we noticed with Bear: We had to come up with a lot of synonyms for the W word, because he went crazy whenever he heard it, in any context. And then, after a while, he’d learn the synonym, too, and so we’d have to move on to a new one. At the very least, he recognized “walk”, “W-A-L-K”, “W-word”, “stroll”, “hike”, “trek”, and “perambulation”, plus probably some others I’m not remembering. Towards the end, I started just leaving blank spots in sentences: “I’m going to ____ over to the grocery store for milk; do we need anything else?”.
Was this before or after you double punched a raging Gorilla into unconsciousness?
When I say “Jump!” my cat can jump six feet onto the top of a fence from a standing start.
D&R
Blake is always hostile and dismissive and insulting in these threads.
That was my impression. It’s hard to learn from a hostile teacher.
Yup, as much as I like the viewpoints he espouses (and this thread is no exception), it’s unfortunate that his responses always seem to carry an unnecessary element of vitriol with them. If he could tone down his argumentative nature and describe these topics dispassionately, I think more people would actually read and understand what he wrote rather than get rankled by his tone and shut out any chance of hearing him out.
First off it is not well “well-known to anyone who’s ever owned a dog’”. It may be widely believed, but it is not even universally believed.
Secondly it is publishable because it is an observation that contradicts all the published observational evidence of dogs to date and because it immediately and irrefutably resolves one of the major issues in the field animal intelligence.
If all the journals articles ever published stated that it was impossible to determine the direction of sunrise, why do you think this would not get published immediately and cause a sensation?
Your whole argument here seems to be begging the question. You seem to be assuming that this issue is agreed amongst the scientific community, when in fact the agreement is that exactly the opposite is true and that what is described is impossible.
No, because the issue isn’t whether the tone of voice conveys the information. The issue is whether the dog understands the way in which the information is conveyed. That is what understanding a word means. If a reaction is prompted without knowing what the speaker of the word meant, then by definition the meaning of the word is not known, it is simply a learned response.
In English, the meaning of a word is not conveyed in the tone of voice, it is conveyed in syllables. Regardless of whether I say “socks” sarcastically or angrilly or completely neutrally, anybody who understands the word “socks” will know what I mean and they will not associate the word with Volkswagons. That is what understanding a word means. If a dog can not understand the word without the tone then by definition he does not understand that word, he only responds to the sound. And similarly, in Chinese the meaning of a word does depend on tone. Anybody who *understands[/i the Chinese word for socks will only understand when it is said with the correct tone and they will always associated the same sound with Volkswagons if it is said with a different tone. If a dog can not differentiate words based on tone it does not in fact understand any of the words, he only understands the sounds.
While I have no problem with believing that other species might find one language or the other easier to understand, that is not the issue. The issue is that we have no evidence at all that they have any understanding of either language. In stark contrast we have scads of evidence that humans are capable of understanding these languages.
J. Kominski published a study in April of 2004 entitled *Word Learning in a Domestic Dog: Evidence for “Fast Mapping”
That link doesn’t present any such evidence. People have known for millennia that dogs can be trained to fetch your pipe and your slipper sand your newspaper o command. That dog has learned 200 objects rather than 3, but for the most part it isn’t doing anything that hasn’t been well documented before.
The only slightly interesting thing is that it appears to be able to engage in basic elimination. If the command does not correspond to any of the previously conditioned responses its selects the only object that it has not been conditioned on. Mildy interesting, but animals long been shown to have the ability to use deductive elimination. The only difference is that this animal is using elimination in conjunction with conditioned responses.
As the researchers themselves note, there is absolutely no evidence that this animal understands any of these words:
IOW they have absolutely no evidence that this is anything but conditioning. It certainly all fits with classical conditioning.
This animal would actually be a perfect test to falsify the hypothesis that animals understand words. In human studies, one of the time-tested ways of proving that a child has learned a word is through the use of proxies such as dolls or pictures. If a child has actually learned the meaning of the word “Sock” then you can show him a doll wearing socks, ask him to point to the socks, and he always will. If a child can’t do that then they obviously do not understand the meaning of the word sock.
So all that is needed to prove that this animal understands the word is to show him a stuffed gorilla wearing socks and ask him to find the socks. If he truly understands the word he will do so without any training at all. If he has simply been conditioned to find socks in a pile of objects in response to a sound. he will run around looking for the pile of objects.
And this type of thing highlights the gulf between understanding words and merely responding to them. A child who learns the word socks has no dificulty at all pointing tot he socks on a doll or in apicture, because they understand what “socks” means. They have a concept of the word socks that is totally independent of context or action. As such they can never be confused by socks in unusual forms or in unusual places and they can take actions related to socks that they have never taken before. A sock is a sock to child.
To the dog, “sock” is just a sound, and when it hears that sound it performs a certain action. The have no concept of sock and as such they can only respond to the sound “sock” in very limited and very limited contexts.
The researchers state explicitly that they have no reason to believe that the dog understands the words: “it is too early to give up on the view that babies learn words and dogs do not”.
Very true. Dogs are capable of reasoning, theory of mind and limited abstraction. But there is no evidence that they are capable of these things as they apply to words. And that is the issue under discussion here.
I wish you to clarify something Blake. Are you basically saying that dogs (& other animals) have no mental signifier for the signified object? And if the organism lacks the signifier, it also lacks true comprehension of the term in question? If they lack the mental concept, then how can Rico find the objects? He would have to have a mental map of “blue striped sock” that his trainer asks him to find, wouldn’t he? And if he can’t identify the blue striped sock on various stuffed animals, wouldn’t that just merely indicate that dogs don’t have nearly the mental flexibility that humans do (in which case that falls into a “sun rises in the east, news at noon” kind of revelation)? You haven’t convinced me nor many in this thread that the difference in this case (if there is in fact a difference) is a difference in kind, vs. a mere difference in degree. Humans are perfectly capable of acting in mechanistic and rote ways too (Skinner certainly got a lot of mileage out of all that), doesn’t mean that they lack the proper mental maps when they do.
I could open a brand new can of worms and introduce the Chinese Room argument at this point if you like. 
I would argue that that border collie must have at least some grasp of grammar, in that he’s able to break the command up into “go get the…” and the noun. If he just new a bunch of separate commands of the form “go get the ball” and “go get the teddy bear” and “go get the striped sock”, but didn’t actually understand what “go get” by itself meant, then he would have been confused by the new command “go get the doll”. But instead, he started by going to the place where his toys were, and only then became confused: He knew he had to go get something, and just had to figure out what the something was.
Now, obviously, that border collie is smarter than most dogs, and I imagine that most dogs don’t have that rudimentary grasp of grammar. But there’s a jump from there to “no dogs do”.
Well that’s an interesting assertion, but until you can show us show us these different conclusions it really isn’t worth much.
Ultimately the issue is one of falsifiability.
All the anecdotes provided so far could be equally attributable to three hypoteheses:
- Dogs respond to their names purely due to operant conditioning.
- Dogs respond to their names because they understand the word that is their name
- Dogs are telepathic and they respond to their names because they read your mind and know that you expect a response.
Any and each of these hypotheses corresponds perfectly to all the observations. But, and this is a very important but, only one of these hypotheses avoids explains all the observations without introducing additional variables.
Everybody i the world, everybody in this thread, accepts that dogs can and do respond to sounds *solely *due to operant conditioning. Pavlov proved that a century ago. We all accept that dogs can do this. The other hypotheses do not add anything to this universally accepted explanation. We all accept that dogs can respond to sound due to operant conditioning, and there is no reason not to accept that this is why dogs respond to their names.
IOW the conditioning hypothesis has never been succesfully falsified. And this is where it gets really weird.
Everybody agrees that animals *can *respond to a unique sound purely based on operant conditioning.
And everybody agrees that when a name is spoken it *produces *a unique sound.
And there is no example of an animal responding to its name that can’t be explained by operant conditioning.
Yet some people still propose that animals are responding to their name for some reason aside from operant conditioning. 
I honestly don’t understand it. Why do you believe this? Why do you leap to believe the more elaborate hypothesis when all the observations are easily explained by the more simple hypothesis?
It’s completely irrational, and I think that gives us a hint as to what is going on. People are becoming emotionally involved in this. Their pet must be able to understand its name because that is the image they have projected into their pet. Evidence be damned, that is that their heart tells them, so it is true, and to hell with any evidence.
Don’t get me wrong, I love dogs in general and my dogs in particular.I have lived with dogs my entire life, and unlike most Dopers most of that time was pent in rural areas, so I have seen dogs exposed to more situations than most of you can dream of. My current two are a mixed bag, one is frankly not very bright, the other is very smart. But she isn’t the smartest dog I had. The smartest dog I’ve had taught herself how to operate electrical appliances by observing humans.
So yeah, dogs are damn clever animals. I know exactly what dogs are capable of. But, and this is the important point, I know what they are *not *capable of. I know the frustration of watching a dog that has been trained to handle pigs and sheep perfectly totally fail to understand how to handle cattle, and then fail again when working horses. And by failure here I mean watching the dog get kicked and injured because it made errors that are painfully obvious to any human.
I have watched dogs that have hunted rabbits and rat kangaroos their whole life encounter their first hare, and laughed myself silly as the hare literally danced rings around them. And I mean literally. The hare was leaping over their backs and turning pirouettes around them. The hare could have escaped a dozen times, and I swear to this day that at the end the hare ran towards the dogs and goaded them into chasing it again. And I know what embarassed dogs look like. Sheepish doesn’t even begin to describe it.
So yeah, I know dogs as well as anyone on these boards. My first hand experience of the *breadth *of dogs’ capacities is probably as good as anvbody in the world. That is why few of the examples here surprise me. But that experience has also taught me just how damn inflexible the response of a dog is. They learn one response set to any given stimulus, and they apply that to any similar situation. They have to learn an whole new response set to any other stimulus, even if it is only slightly different.
Their ability to adapt or to learn though association is amazingly poor. That is why some people are amazed that a dog can learn to fetch 200 objects on command, when we all know that dogs can fetch two objects on command. It is because we know how inflexible dogs are that his sort of thing amazes us, whereas we expect any human child to be capable of he exact same act.
And that is all consistent with the idea that their learning is primarily based upon operant conditioning. It’s not solely based upon that of course, but everything I have ever seen in personal experience or the literature suggest that this is the most common way that animals learn.
I would dearly love to believe that my dogs are different and that the respond out of some deep seated ability to understand one damn word I say, but there simply isn’t any evidence for that.
That sort of thing teaches you just how inflexible dogs are when it comes to learning
But they are animals. I would love to see some evidence that my dogs
I drank a glass or two more wine tonight than I usually do so perhaps I am missing something but I parse that as, “dogs can learn language akin to human babies.” I am missing the part where dogs can’t acquire language in that sentence.
I am amazed at people’s resistance to intelligence in animals other than humans.
Dogs are mammals. They have mammalian brains. Take a dog brain and it looks a lot like a human brain (albeit smaller but size is not everything…crows have far, far smaller brains than a dog yet exhibit remarkable intelligence…and yes I know crows are not mammals).
Dogs are also social creatures which stems from their ancestry. Wild wolves are social animals as well. Wolf packs display communicative abilities in a variety of ways. Hell, they are cooperative hunters. They display submission or dominance or affection and so on. Why is that not “language”?
Certainly they do not have the nuance and subtlety and range humans do. But they definitely communicate and language is not all verbal (ask a deaf human who uses sign language).
Dogs clearly have a capacity for communication so there is no reason to suppose they cannot plug human words into their framework. A “blue striped sock” may not mean the same thing to them as it does to you but they will know they word represents “that thing and not the other thing”.
I am not entirely clear on what you mean by"mental signifier". If you simply mean that there is some sort of mental place-keeper for an object that meets certain pre-conditioned criteria than yeah, there is a mental signifier. This could be as simple as the type of programming you can do on a 40 year old computer:
- Goto pile.
- Select new object.
- If colour = red then 4.
- Goto 2.
- If texture = fuzzy then 6
- Goto 2
- The same as 3 and 5 for every other criterion
- Return object to owner.
If this is what you mean not “mental signifier” then yeah, that obviously exists.
But that is also radically different to having any concept of what the *word * “sock” means independent of this specific exercise. To the extent that the concept exists at all it exists as a string of variables related specifically to this exercise and in this location. It is intrinsically linked to the situation and response pattern of this trial
I hope we both agree that this signifier program will produce perfect results in the trial described. But the first point to note is that you could produce it on a TRS 80, and I doubt that anybody would consider that a TRS 80 could know what word meant. So clearly understanding a word is much, much more than simply having an event-linked “mental signifier” that is triggered by external stimuli.
The second point to note is that it is situation dependent. While it produces perfect results in this very specific situation, you try using the the same signifier routine elsewhere and you get nothing. For example, you ask this routine to find the pair of socks on the beach where only one person is wearing socks. It fails entirely. And similarly if you ask the dog to perform that trick I will bet you pounds to pence he will fail.
He simply will not know to look for socks on feet. Yet the very definition of socks is that they are footwear. While any any infant who *understands *what the word socks means will be just as capable as the dog of fetching the socks out of the pile of objects, the infant will also be able to find the socks on the beach or on the doll with the same precision with no further conditioning.
And that is the big difference. Understanding of a word produces a concept that is *transferable *and independent of context, whether linguistic or physical. That is one of the standard qualifiers when testing whether human understand words.
As the researchers note, there is no evidence that this dog has any concept of socks independent of the search routine. Therefore it doesn’t understand the word, it is merely responding to the sound.
Once more, the question comes back to falsifiability. Assume that this dog had no understanding at all of the word sock, if it had the same understanding as a TRS 80 and it was simply conditioned to initiate search routine based on the sound. How would that dog differ objectively from the dog in the real world? IOW what observations coudl we collect to differentiate the two?
Because if we could not collect any such observations it becomes nothing more than a philosophical semantic quibble with no objective resolution. ie, not scientific.
But that is simply an argument from ignorance. You are simply arguing that since I can’t prove that dogs aren’t telepathic, so they must be telepathic.
I am not attempting to convince anyone that there is a difference in kind. I am persistently pointing out that there is no evidence that there *is *such a similarity, and it is totally unnecessary to explain the evidence.
Quite simply:
Do you accept that Pavlov was right when he said that dogs can be trained to respond purely do to operant conditioning? That it is indisputable that the response of a dog *can *be due to operant conditioning?
Do you accept that any observation of a dog responding to a word can be explained quite simply as basic operant conditioning? Or to put it the other way, do you accept that all observations can be explained by operant conditioning? I am not asking whether you think that is the explanation, just whether you agree that it can be explained without invoking telepathy or anything else aside from basic operant conditioning?
If you do accept these two points then it becomes an argument form ignorance to argue that I have to convince you that it *is *operant conditioning. We know that operant conditioning exists. We know that it explains the observations. Accepting any other hypothesis defies Okhams razor. i can never prove that dogs are not telepathic, nor would I bother to try. All I can do is point out that there is no evidence that they are and no reason to invoke such an ability
Not in this thread, I beg you.
No, two problems
The first problem is that the phrase “go get the” is only being used in a single context, so it fails a basic test of grammar. To establish that this is some form of grammar you would need to teach the dog to respond to the phrase in unique and combinable contexts. For example, you teach the dog “bite the” with a novel range of objects, then you teach it “bark at” with a novel range of objects. Then you ask it to “bite the sock”, a combination that it has never heard before. If it can do that then you have some evidence of grammar. If it goes and fetches the sock or bites some other object then you have conclusive evidence that it does not understand grammar.
The second problem is that there is only one response in this test. For all we know the dog might have exactly the same response if we said “rape the sock”. It may just be responding to tone of vice or handler body language. None of that invalidates the result presented, but it is not evidence of grammar comprehension nor do the researchers make such a claim.
But as it is all you have is evidence that two separate conditions can be applied in succession. To show you why it makes little to call this grammar, consider the following scenario.
I take a slug, and I give it a mild electric shock to the foot, while flashing red. If I do this enough it will always turn upside down when I flash red lights. I then take the same slug and I flash green lights at it while it is upside down, then I give it a mild electric shock. It turns the right way up. After I do this a few times I can make it turn upside down when I flash red lights and right itself when I flash green. But then I flash blue lights while I shock its tentacles, and after a few times every time I flash blue lights it withdraws its tentacles. So now I can make it turn upside down and withdraw its tentacles by flashing red then blue lights. Or I can make it withdraw its tentacles then turn right side up by flashing green then blue. Then I flash random blue lights followed by yellow light. And the slug withdraws its tentacles and then becomes confused.
So a slug must have at least some grasp of grammar, in that he’s able to break the command up into “withdraw the tentacle.” and the orientation. If he just new a bunch of separate commands of the form “turn upside down and withdraw the tentacle” and “withdraw the tentacle and turn right side up” but didn’t actually understand what “withdraw tentacle” by itself meant, then he would have been confused by the new command". But instead, he started by withdrawing the tentacle, and only then became confused: He knew he had to go withdraw the tentacle and just had to figure out what the next command was.
So that proves that slugs understand grammar.
No it doesn’t. It proves that animals can be conditioned to respond to multiple conditioning at once. Every time the dog hears “go get the” it always goes to the room. The same way as the slug always withdraws its tentacles when it sees blue light. The fact that the rest of the signal is indecipherable is irrelevant because part of the signal is perfectly clear and evokes the correct response.
For about the seventeenth time n this thread, nobody is arguing that dogs can’t acquire language. Anymore than we are agruing that they cna;t be telepathic.
What is quite clear is that there is no evidence for either position. If you want to believe that your dogs are telepathic or that they can understand grammar, great. Just be aware that there is not a shred of evidence for this position
I exlained this at length already. Please read the thread then ask any questions that I have not already answered multiple times.
Cabbages also clearly have a capacity for communication so there must be no reason to suppose they cannot plug human words into their framework too.
Right?
No, there’s no evidence that they know that at all. All the evidence can be adequately explained by a sound triggering a conditioned response. There is no reason at all to believe that the sound represents anything at all.
I have an entirely mechanical punchcard machine.I put in the punchcard with the picture of a sock and it causes the machine to open the slot that leads to a box of socks. A sock drops out of the machine. Do you believe that punchcard *represents *something to the machinery? Or was it entirely a reaction triggered by a stimulus? After all the machine will never drop out a toy giraffe in response to that card, and it will always dropout a toy giraffe in response to the giraffe card.
So why is that not evidence that the each card represents “that thing and not the other thing” to the machinery?