some Rico correspondence

[PennyT]

Rico is that German border collie

studied by Julia Fischer (that might be Fisher, I forget) and others. Rico

knows about 200 words and has shown as well some evidence of fast-mapping (that seems to be learning a new word by process of elimination).

 

In a comment in the Perspectives section of the June 11th Science issue,

Paul Bloom, a scientist and philosopher at Yale, wrote: "These experiments

are carefully designed, and so there is no worry about problems akin to

those of Clever Hans (a horse that seemed to have mastered arithmetic but

was actually responding to subtle cues by its owner). Yet, if Rico is really

learning sound-meaning relations, as Kaminski et al. maintain, it should not

matter who the speaker is."

 

On reading that, my reaction was if a dog has to respond regardless of who

the speaker is, then most very intelligent dogs would not do very well at

anything (even things some dogs are quite good at like tracking or working

livestock) because for most dogs who the speaker is usually matters a lot.

But possibly that is not the case.

 

I wonder if anyone else had the same reaction.

 

I was also hoping that a few people in the sciences could comment on Paul Bloom's criterion above for

demonstrating the understanding of words.

 

I think I should I add that I Paul Bloom is a nice fellow and not at all out to show that the experiments with Rico are bunkum.

 

Paul Bloom sent me a pleasant note explaining his stance on Rico and does seem to believe that if Rico will not (in some way, not necessarily by fetching as in the experiment) demonstrate that he knows words regardless of who the speaker is, then Rico has not learned words in any way that we might regard as normal language acquisition.

 

Thanks.

 

Penny

[Eileen Stein]

>

 

When I read those words, before I got as far as your comments, I thought to myself (as I have many times while reading Rico-related articles) that it's amazing how little these scientists know about dogs. I know I live a particularly dog-saturated life, but it seems so strange to me that there are people in the world who obviously know dogs only as lab animals or experimental subjects, and speculate tentatively about things any dog owner knows from experience.

 

Anyway, I think his position is downright weird. Can he not conceive that dogs could be simply uninterested in what people with whom they have no relationship are saying? If I react when my husband says something complimentary to me, and I don't react when a construction worker on a sidewalk says something complimentary to me, does that mean I don't understand what "Hey babe, lookin' good!" means? I guess an anthropologist from Mars might conclude that, but he would be wrong.

[Denise Wall]

The first thing I thought was, tell him to try closing his eyes when people talk to him and see how much more goes into communication than just understanding what is verbally spoken.

 

Edited to say, "him" meaning Paul Bloom.

[Tuck's BCBuddy]

Hmmm, well all of my experiences with scientific testing and experiments, tells me that the scientist wants to control all outside parameters, i.e. introducing variables will only serve to skew the results. For example, this was why, so often, in our dynamics calculations, one of the basic assumptions was to neglect the effect of wind resistence (obviously, a variable that could not be accounted for, in the perfect theoretical world). And so, a different voice, in my opinion, would be a variable to the experiment. Different pitch, different intonation, etc., etc.

 

Now, if Rico learned the various words, could he learn to respond to another voice on these words? I'd say probably so, but I don't think it unreasonable to think this dog may not respond immediately.

 

Take, for instance, the country of Germany itself, or any of a number of countries, for that matter. Herein lies a huge variable. Different people, from different areas, speaking the exact same language, sound quite different. I work for a German company, and upon hearing the same sentence from two different people, I've already found myself replying, "The two of you could not possibly have just said the same thing", and yet, on further examination, they have proven that they did. And yanno, this doesn't even require a large geographical difference. In the area where I live, I've seen differences in accents of people from one town to the next.

 

One last thing, I had read that Rico had response similar to approximately a 3 year old, with regard to word association. They had not made the claim, this dog has an IQ of 136, a vocabulary of 6000 words, and the education/mental capacity of a Harvard Law graduate (although, Rico might be able to successfully debate a couple of presidential candidates in the fall, LOL). So, I guess I really don't see how someone could expect the dog to immediately respond exactly the same to all voices, eh. I know my 2-1/2 year old son doesn't.

[Bill Fosher]

I get your point, Eileen. If someone was to say "Hey good lookin'" in my presence, I would presume they were talking to someone else and ignore them. That wouldn't mean I didn't know what the words meant.

 

That's where defining language as being willing to engage in certain behaviors based on verbal cues breaks down. The fact that you don't giggle and offer your phone number in response to the wolf whistle doesn't mean that you don't know what was meant by the whistler.

 

I've also been disturbed by the imprecise use of the word "language." It may only be the popular press that's throwing around the idea that this dog has language capacity, but we should all bear in mind that there's a big difference between a dog that knows the difference between a sock and duck and a primate who can form and understand novel sentences using sign language.

 

Trainer: "Put the duck in the sock."

 

Ape: "You put the duck in the sock. I'm watching Jerry Springer."

 

Some might infer from this that the Ape knows what the duck and the sock are, but it's not necessarily so. All we can really tell is that he'd rather watch Jerry Springer than obey the commands of his handler.

 

It's sort of like calling confusing prey drive with herding ability. Vocabulary is undoubtedly part of language, but it is not the same thing.

[Pearse]

Originally posted by Eileen Stein:

<< Yet, if Rico is really learning sound-meaning relations, as Kaminski et al. maintain, it should not matter who the speaker is. >>

Anyway, I think his position is downright weird. Can he not conceive that dogs could be simply uninterested in what people with whom they have no relationship are saying?

I think that the original quote doesn't accurately reflect Paul bloom's position. what I took from reading his article was that he was proposing two explanations for the way in which "Rico" was learning to associate words with objects. One was similar to the way human children learn, in which case the context would be irrelevant as would the identity of the person giving the command. The second was by some other method of learning word association that was "doglike" not human.

 

He wasn't making a value judgement about whether one was better than the other merely observing that humans and dogs might not learn word association in the same manner and proposing possible further tests to discriminate between the various possibilities.

 

It may have nothing to do with understanding as Eileen suggests. Take working dogs. Three dogs may both understand that "Come Bye" means "go out to your left around yonder sheep until you are on balance or until your handler screams 'lie down you ornery little sob several times'". Dog one will take that command from anyone. Dog two will only work for one person. Dog three will take the command but only if issued by a male in a tweed cap and a Highland Scots accent.

 

My take home message from the Perspectives article and the original paper by Kaminski et al., is that dogs can learn to associate words with objects, and develop a bigger vocabulary than most people thought, but not necessarily in the same way as humans do. Not suprising and doesn't lessen their accomplishments. After all, there are many other things that dogs do so much better than humans do. It all comes out as a wash.

 

Pearse

[Deanna in OR]

One of the key things I have learned about training dogs is how much the context matters.

 

They usually have to learn that "Sit!" means to do the same behavior in the house, in class, in the park, in the obedience ring. They usually learn the behavior faster in the new situation, but it is still situational at first.

 

Some dogs are able to generalize to many situations faster than others, learning that "Sit" means plop your butt down, no matter where I say it.

 

And dogs pay attention to different clues than just "a word"--the tone of voice, accent, body language associate with the command (very subtle clues that humans often don't pick up on), and yes, the person speaking the command.

 

Dogs aren't computerized robots that always respond to an input in exactly the same way. They can choose to obey a command, or not. If they don't think (in their canine logic, which isn't human logic) that there is a good reason to perform a behavior upon command, they won't do it. Doesn't mean they don't "know" it. Just because a 2-year-old human won't eat their green beans at dinner, doesn't mean they don't "understand" Mom telling them to do so...the kid just doesn't WANT to.

 

All experimenters need to be clear about the assumptions that go into the experimental design, with some common aggreement on those assumptions, if they want to have common acceptance of the results. But this is the area that too many scientists brush over, not even recognizing their own unspoken assumptions. OK, I'll stop here on this topic--this is a pet peeve of mine at work! (I'm an engineer...)

 

But Paul Bloom is making some assumptions about the meaning of "understanding language", that those of us who live with dogs may not agree with. He needs to get those assumptions out in the open and validate them to have a meaningful argument.

 

Deanna in OR

[Tuck's BCBuddy]

Deanna,

That is sooo cool, you get to drive a train for a living !!!!

Seriously though, I know of what you speak, being a BSME myself.

[border_collie_crazy]

all I gotta say is why does a dog have to understand human language to be smart, yet humans dont have to understand dog language to be smart lol I mean humans have to do all kinds of studys to figer it out, I dont see the dogs testing the humans lol now really who is the smarter?

[PennyT]

If I sounded as if I thought Paul Bloom were making an objectionable value

judgment, I conveyed an unintended message. The meaning of the sentence

"fetch

or find the

sock" is, indeed, different from the meaning of the noun "sock."

 

What struck me was the suggestion toward the end of the short commentary in

Science that to demonstrate understanding of sound-meaning relationships

Rico needs to show he understands the words for objects without having it

matter to him who the speaker is. My reaction,

which was how little someone evaluating that particular

study knew about getting dogs to do something they know how to do, was much

like Eileen's. Naturally, I said nothing of the kind when writing to Dr.

Bloom. I simply wrote about how for most dogs who the speaker is matters a

lot.

 

In his short and very prompt and polite note in reply, Dr. Bloom said "the

point in my article concerned

word learning--if you

know a word, then you can understand it regardless of which person

says it. The person doesn't need to have a good relationship

with you, or even know you, because that's not how

language works. Once you know that "milk" refers to milk,

you understand the word even if its spoken by a stranger.

This is true of any normal 2-year-old child. If it is not true

for Rico, then he is not learning words in any normal sense of the

term."

 

For all I know, there may be ways to find that out which would satisfy Dr.

Bloom. In looking at information on him on the net, I stumbled across a

piece he wrote on www.edge.org

 

I noticed he accepts that babies

can add and subtract in some fashion. He stated: "Karen Wynn showed that

babies can even do addition and subtraction, of a rudimentary sort. You put

an object down, a screen rises to hide it, and then you put another object

behind the screen. Then the screen drops, and there is one object, two

objects, or three objects. If there's one object there, babies are

surprised. If there are three objects, babies are surprised. They know one

plus one equals two."

 

Obviously, no one was asking for a verbal response, and researchers were

accepting some kind of visual evidence of surprise. I think maybe it was

looking longer at the unexpected numbers but I could be wrong. In the same

piece, Dr. Bloom says "I have done some other research-in collaboration with

Valerie Kuhlmeier and Karen Wynn-that suggests that 5-months-olds see bodies

and souls as entirely distinct." (He acknowledges that is controversial,

btw.)

 

Clearly a man who can devise (or help devise) an experimental protocol for

infants about dualism and reason from those results to the idea that infants

have a sense of bodies and souls as distinct is not someone who needs to

see an elderly border collie play competitive Scrabble in order to be

convinced

the dog understands a couple of hundred sound-meaning relationships.

 

However, my point was that most dogs won't give their best performance for

people the dogs don't have a working relationship with. Few trainers have

the knack of getting anything close to that right off the bat at any canine

endeavour. For example, if someone

gets a good, trained dog without working it some first, takes it home, and

then tries to gather a large field the next day, no one would the least

surprised if the results were less than ideal or even if the dog didn't go

out at all. The dog's skill would not be called into question, at least not

at that

time.

 

Understanding that a working relationship matters to intelligent dogs is

important if someone is going

to comment on a study of one.

 

I can't stress enough, though, how thoughtful it was of this brilliant,

young

academic

to have answered my note, which he did from where he is teaching for the

summer in Budapest.

 

Pearse said: "My take home message from the Perspectives article and the

original paper by Kaminski et al., is that dogs can learn to associate words

with objects, and develop a bigger vocabulary than most people thought, but

not necessarily in the same way as humans do. Not surprising and doesn't

lessen their accomplishments. After all, there are many other things that

dogs do so much better than humans do. It all comes out as a wash."

 

I don't disagree with anything in the paragraph above. I doubt anyone who

has responded to my post does.

 

I want to edit in this last: another thing that amazed me about the Rico article and the commentary in Perspectives was how little interest it generated in people who own border collies.

 

Were we saying ho-hum, what's new?

 

Maybe.

 

I have two other reasons the research should be discussed: first, researchers and commentators need to educate themselves about humans and dogs together outside of a lab animal environment; second (and on a lighter note), our dogs may be sandbagging us. Did the verbal bar just go up? Did Rico let the cat out of the bag?

 

Penny

[Northof49]

In his short and very prompt and polite note in reply, Dr. Bloom said "the

point in my article concerned

word learning--if you

know a word, then you can understand it regardless of which person

says it.

__________________________________________________

 

Is he kidding or what?

 

I had my dogs going to a school for an 8 week dog program, where some of the kids are FAA and have slight speech impediments and on top of that a Cree accent. I had a hard time understanding some of the commands they were giving the dogs myself, and so were the dogs. I had to spend some time with the kids to get them to the point where their "down" or "wait" sounded close enough to the average person's "down" or "wait" in order for the dogs to be able to respond. When I couldn't get around that I would stand behind the kids and give the dogs hand signals, so the kids thought the dogs were responding to them.

[Deanna in OR]

In his short and very prompt and polite note in reply, Dr. Bloom said "the point in my article concerned word learning--if you know a word, then you can understand it regardless of which person says it. The person doesn't need to have a good relationship with you, or even know you, because that's not how language works. Once you know that "milk" refers to milk, you understand the word even if its spoken by a stranger. This is true of any normal 2-year-old child. If it is not true for Rico, then he is not learning words in any normal sense of the term."

My point about assumptions is very clear, here.

 

His statement that

Once you know that "milk" refers to milk, you understand the word even if its spoken by a stranger. This is true of any normal 2-year-old child.

is an assumption that I do not agree with, and thus I cannot agree with his related conclusion--not for dogs nor even for humans.

 

Some dogs can generalize well to different situations, especially with practice--for others, especially the "analytical" ones, the situation (including the relationship with the human) is an important and critical element of the communication process, including both understanding a word AND the related, subsequent behavior. Similarly, for humans--If a 2-yo child is told "eat your green beans" by Mom, they may or will understand AND comply. If they are told these same words by a big, bearded, scary stranger with a deep gruff voice, they may be so terrified that the words have no meaning to them--the fear takes over their mental processing. If they are told the same words by their babysitter, they may understand them quite well but choose to do nothing or to do some other behavior altogether (like stick out their tongue!).

 

Why should it be any different for dogs?

 

While it is certainly a more challenging experimental design to account for the context in a study of communication and language, the simpler "only the words carry meaning" approach doesn't really tell us anything, whether studying language and communication in chimps, dogs, snails or humans.

 

On an interesting note--I recently took a class on improving communication at work. One of the starting premises was based on studies that indicated that, for humans in general, body language carries 55% of the information we "hear", voice tone carriers 38% of that information, and the content (the WORDS) only carry a scant 7% of the information! If this is true for humans, how much more true it must be for dogs! (an assumption I am making, of course).

 

And this is why written communications such as this can be so challenging at times, even for us humans!

 

Deanna

[Doug Boyder]

Dogs are HEURISTIC learners. This means that dogs will remember consistent negative or positive experiences. Anyone who has done Obedience Training will know that re-enforced habits can be difficult to re-train.

 

Border Collies also seem to have early LINGUISTIC ability. Rico and Benji could associate hundreds of different words with objects, demonstrating Early Childhood-level language ability. By Middle Childhood, the abstract thinking and symbolic language required of a human would probably exceed the abilities of a dog, even Rico and Benji.

 

"Rico probably has the general ability to connect things - not a language ability" (Alex Kacelnik, Oxford University, UK). Border Collies have inbuilt ABSTRACT ability, which could explain why they are so good at herding - organizing patterns of sheep into groups (I know you?ll correct me if you think I?m wrong).

 

Interesting study though...

[Eileen Stein]

Been thinking about this. Seems to me it's possible that

 

(a) Rico might be able to do the same thing in response to commands from someone else. After all, many border collies will work for anyone, and will take herding commands from a new owner the first time it's taken to sheep by that person.

 

( Rico might not be able to do the same thing in response to commands from someone else, because he cannot understand that person. I've found that imported sheepdogs seem to have a lot of trouble with verbal commands for quite a while after they arrive. Even if you try to give the commands with the accent and intonation of their country of origin, and think you're succeeding, they often look blank.

 

© Rico might not choose to do the same thing in response to commands from someone else, because he has no relationship with that person, and therefore doesn't recognize that person's right to command him or isn't motivated to work with him. The problem here is that such a lack of performance could easily be interpreted as a lack of understanding. I don't know why a scientist would jump to the conclusion that it was the latter, though, when the evidence is equally consistent with the former. There are certainly parallel situations with humans learning language -- the 2-year-old who is asked to point to the milk and hides her face behind her mother's leg, or who says "Watch me jump down the step!" or utters the 2-year-old's mantra "No!" I don't believe a scientist would conclude from such evidence that the kid didn't understand "point" or "milk," so I don't know why he would draw that conclusion with Rico.

 

I think this is probably just a reformulation of what Pearse and Deanna were saying. The bottom line is that I think it would be bad science to conclude that dogs don't learn words the same way humans do if Rico won't perform for a stranger.

 

You know, what I think ought to be really interesting to these scientists (though I don't know if they are flexible enough to see it as a language issue) is the way border collies respond to variations in whistle commands. "Putting whistles on a dog" is usually a pretty straightforward exercise -- using a whistle with a verbal command the dog already knows, until the dog grasps that they mean the same thing. The interesting part comes after that, though. One of the most dazzling things I ever saw occurred when I showed Kent Kuykendall (who is one of the true whistle virtuosos) a bitch I had recently imported who was very nice and generally took all her commands but seemed to have only one gear -- very, very fast. Kent asked me what her whistles were, I told him, and there then ensued what I can only regard as a conversation between him and the dog. It was a dialogue, because part of it was his giving variations of those whistles and indicating to the dog what he meant by those variations, but part of it was his ascertaining what the dog understood him to mean by different variations, and then his modifying what he "said" to her accordingly. Within perhaps ten minutes of this, with him just whistling to the dog as she moved the sheep around, he and she had worked out a vocabulary with which they were in perfect accord, and she would speed up or slow down, go wide or go tight, turn in or turn out, and other elaborations of those basic whistle commands.

 

I wish those language boffins could have seen that. To me it is much more interesting than just multiplying the number of objects a dog can recognize by name.

[SoloRiver]

I think this is probably just a reformulation of what Pearse and Deanna were saying. The bottom line is that I think it would be bad science to conclude that dogs don't learn words the same way humans do if Rico won't perform for a stranger.

 

I think the point isn't that scientists concluded that the dog's CAN'T, but that given the evidence they cannot eliminate that hypothesis and therefore cannot say that the dogs CAN. When the hypothesis of interest is "do dogs learn and understand words the same way humans do?" then the alternative explanations for observed behaviors have to be eliminated first. It's not scientists being obtuse, it's just the way science works.

[kevinb]

My only thoughts when I first read the article were "Why are they excited about 200 words?"

[Eileen Stein]

Melanie, when you say "CAN" and "CAN'T," what do you mean? Do you mean can or can't respond the same no matter who the speaker is? Or do you mean do or don't learn the same way humans do, or do or don't learn sound-meaning relations?

 

I understand the scientific method and the null hypothesis. But IS the hypothesis here "do dogs learn and understand words the same way humans do?"? Or is it "do dogs learn sound-meaning relations"? (What Bloom said was "Yet, if Rico is really learning sound-meaning relations, as Kaminski et al. maintain, it should not matter who the speaker is.") I realize I let myself be distracted by his comment about kids and milk; I should have said, "I think it would be bad science to deny that Rico's demonstration shows he has learned sound-meaning relations if Rico won't perform for a stranger." What are the alternative explanations for the observed behavior?

[Bill Fosher]

I'm still scratching my head over why anyone would think that a dog would learn like a human. I would presume that their cognitive evolution would have shaped the development of their understanding of the world in more adaptive ways than the ones we have adopted.

 

Ever seen a dog in divorce court?

[Pearse]

Originally posted by SoloRiver:

I think this is probably just a reformulation of what Pearse and Deanna were saying. The bottom line is that I think it would be bad science to conclude that dogs don't learn words the same way humans do if Rico won't perform for a stranger.

 

I think the point isn't that scientists concluded that the dog's CAN'T, but that given the evidence they cannot eliminate that hypothesis and therefore cannot say that the dogs CAN. When the hypothesis of interest is "do dogs learn and understand words the same way humans do?" then the alternative explanations for observed behaviors have to be eliminated first. It's not scientists being obtuse, it's just the way science works.

Exactly. Poor Dr. Bloom is getting slammed here for something he neither said nor implied. He never said that dogs CAN'T learn word association indpendent of handler or context. He said IF they can't THEN one would need to design additional experiments to determine in what ways their learning prrocess is different. And once again, I will make the point that different does not equate to inferior.

 

Pearse

[amc]

I'm reminded of the old saying, "next time you're feeling important, try ordering someone else's dog around"!

[Tuck's BCBuddy]

Pearse,

I fail to see any "slam" occurring here, but instead, simply a discussion. Whether we didn't have Dr. Bloom's full premise, or there was some misunderstanding of the premise, there still was no slam. Conversely, even if there was complete understanding, etc., you cannot possibly call discussing the points and counterpoints a slam.

 

Also, I don't believe this....

 

Yet, if Rico is really

learning sound-meaning relations, as Kaminski et al. maintain, it should not

matter who the speaker is."

 

is the same as this....

 

IF they can't THEN one would need to design additional experiments to determine in what ways their learning prrocess is different.

 

Note that the first quotation was taken referenced from Perspectives section of the June 11th Science issue, and in essence, is what started some of this discussion.

 

Was this a misquote, or is the quote being misinterpreted? And I am sincerely asking the question, no hidden agenda, and no slams or wisenheimer tact intended.

[Eileen Stein]

>

 

Who said it does? I didn't understand Bloom to be making any kind of value judgment. That's not an issue here, as far as I can see.

 

I understood Kaminski to be asserting that Rico's performance shows that dogs can learn sound-meaning relations, i.e., they can relate a word to the object it designates, and that they can learn many more of these relations than scientists (as opposed to border collie owners) had previously thought. I know that dogs don't understand words in the same way humans come to understand them (e.g., knowing that the green stuffed dinosaur represents a large extinct reptile), and I don't see how you could ever demonstrate--with our present state of knowledge--whether or not the process by which they learn these sound-meaning relations is identical to the process by which humans learn them.

 

When Bloom said, "Yet, if Rico is really learning sound-meaning relations, as Kaminski et al. maintain, it should not matter who the speaker is," I took his implication to be "It should not matter to his performance who the speaker is, and if it does affect his performance, then that would be some evidence that he is not really learning sound-meaning relations." I think that was a reasonable interpretation on my part, and if that was what he meant, I disagree with him for the reasons I've stated and I conclude it indicates he doesn't know very much about dogs.

 

I trust scientists don't feel they're being slammed if someone disagrees with them.

[Eileen Stein]

>

 

So true! There are young kids around here who sometimes latch onto me and follow me around while I'm doing chores. They will sometimes chant "Sit! Sit! Sit!" at my dog, who of course ignores them. They are aggrieved; they ask me why he won't sit when they tell him to. But it hasn't even occurred to them to conclude that he might not know what "sit" means, because they've seen him sit when I say "sit," so they know he does.

[Pearse]

Originally posted by Eileen Stein:

<< I'm reminded of the old saying, "next time you're feeling important, try ordering someone else's dog around"! >>

 

So true! There are young kids around here who sometimes latch onto me and follow me around while I'm doing chores. They will sometimes chant "Sit! Sit! Sit!" at my dog, who of course ignores them. They are aggrieved; they ask me why he won't sit when they tell him to. But it hasn't even occurred to them to conclude that he might not know what "sit" means, because they've seen him sit when I say "sit," so they know he does.

Of course the correct response to that is; "same reason you don't always clean up your room, put away your toys, and wash your hands before dinner when I tell you to"

 

To answer your previous questions Eileen; if scientsists thought they were being "slammed" every time that someone questioned their assumptions, they'd be best to find some other line of work. Questioning and testing other people's assumptions, methods, and conclusions is what it's all about.

 

By rights I should have put "slammed" in quotations marks to indicate some degree of irony/humor/(insert rhetorical device of choice here) but it was my impression that he was being criticised for things that he hadn't actually said rather than people disagreeing with what he had actually said. Perhaps that was me misunderstanding other people's points.

 

Either way, I have no strong opinion either way in this debate being neither a behavioral scientist, nor an experienced trainer/handler so I'll return to the sideline and watch the match from the comfort of my easy chair.

 

Cheers,

 

Pearse

[Eileen Stein]

>

 

Oh please don't do that -- you sound like someone we'd like to be hearing from.