Dog Trained on 200-Word Vocabulary

An anonymous reader writes "The Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany is reporting in Science Magazine today on an example of successful human to non-human communication: Rico, a collie trained on a vocabulary of 200 words. Their conclusion is that 'brain structures that support this kind of learning are not unique to humans...[Rico has a] retrieval rate comparable to the performance of three-year-old toddlers'. In case you ever wondered if your dog understands what you are saying, Rico 'can learn the names of unfamiliar toys after just one exposure to the new word-toy combination.'"

Comments

Does the language matter?

[geek]

For instance it's no break through that dogs understand commands, seeing eye dogs have been doing this for decades, but does the language used make a difference? For instance I assume these dogs were trained in German, would French, Spanish or something like Arabic work better? Can a "dog langauge" be made that works better for them, perhaps allowing a 400 word vocab or more?

Last I heard the average human had a vocab of around 2500 words or less. Raising an animals higher could lead to full fledged conversations rather than just an instructional command oriented relationship.

Re:Does the language matter?

[DwarfGoanna]

I'm speculating here, but I would think that the reason human/animal communication has mostly been limited to commands has a lot to do with the traditional and historic relationships between animals and humans. What if we had spent the last few thousand years trying to communicate with domestic animals in other capacities, and bred them accordingly?

Re:Does the language matter?

[teutonic_leech]

You know, I think you might be on to something here. What if we would have bred them according to intelligence as opposed to preferences for 'shiny fur' or 'pointy ears'? Maybe I'd be playing chess with wuffi by now - and he'd probably beat my butt badly - LOL Seriously - this opens a whole new can of worms! Considering genetic engineering and future advances in related fields - would it be possible to develop super-animals with a certain amount of intelligence and self awareness? some might laugh now, but many of the things we take for granted (like writing this email right now) would have been deemed ridiculous 100 years from now...

Re:Does the language matter?

[MochaMan]

The age at which they learned English varies from early childhood to adolescence, but one thing they have in common is that their vocabulary in either language is not as good as a native speaker's.

Apparently you've never been to Montreal. I'm not a Montrealer, but I do speak both official languages essentially flawlessly -- or as flawlessly as a "native" speaker could be expected to speak. Although I grew up in western Canada, which is primarily anglophone, I've spoken both English and French my entire life, and to this day, I turn my radio dial to both English and French stations, watch both English and French TV, and read novels and websites in both languages.

A number of people in this country worried about the same old-wives tale you've just inferred -- that programmes like French Immersion would cause students' English to suffer. In fact, the opposite turns out the be the case; French Immersion students have tended to do better in their English courses than non-French Immersion students.

Of course, an interesting point is that English and French share some common linguistic history, hence an understanding of French can be very helpful in understanding the etymology of English words and so on. I wonder if the same stats would hold true for students who were bilingual in English-Japanese, for example. That said, I'll admit that I actually spent this afternoon talking to a Japanese guy who immigrated to Canada at age 18 and spoke English fluently enough that I had thought he'd been born here until he told me where he was from.

While I was able to go from zero Spanish to fluent conversational Spanish in about a year of living in Mexico (mostly due to its common roots with French), it has taken six years of study for me to attain the same level of confidence with my Japanese. I also speak a little Slovak. I have less trouble with Slovak than Japanese, but certainly more than with the Romance languages.

I would argue that if a bilingual (or trilingual) speaker does not have an average vocabulary in each language he speaks, it is not through some inate limitation of the human mind, but due solely to his environment. Very few of us live in an environment where we *need* to be bilingual, and hence we tend to favour one language over the others. In cities such as Montreal, where you essentially have a 50/50 split of anglophones and francophones, you'll find a great many people who are fluent in both languages at a level where you'd be hard pressed to determine which language was their "native" language.

(border) collies are _way_ too smart

[BlueLines]

my ex-gf and i had a border collie for over a year. by the end, she (the collie) had a vocabulary of well over 100 words. she knew the difference between the ocean, the lake, and the river. she knew what the "purple squeaky ball" was. her favorite word though was "treat".

a current friend of mine also has a border collie. he is trained to turn off the tv, shut the tv cabinet door, and turn the lights off when his owner falls asleep at night.

i think most border collies are smarter than a lot of people i deal with on a daily basis at work.

Re:Bzzt. Try again

[cos(x)]

This dog actually seems to be understanding quite a bit of what he is going on. It's not just a matter of finding an object he has learned to associate with a particular sound. There was a show on tele earlier today about this dog and they showed an experiment that went something like this:

The dog has a collection of roughly 200 toys, each of which he knows by name. When told a toy's name, he'll go and fetch the toy. That's not really impressive, that's what most dogs do. Now comes the cool part though. They added a new toy - one the dog had never seen before. The toy was added to the collection while the dog wasn't in the room, so he didn't see the toy being added. Then they told him to get this new toy. Simply by telling him the new toy's name, which he had never heard before of course. Now, the dog went to his toy room. He found all the old toys and the new one. Since none of his old toys matched the name he had been told, he figured that they what they meant must have been this new toy he just discovered.

This is really the reasoning part. You don't need to tell the dog what the toy's name is - the dog will figure it out himself. If you tell him to look for something he's never heard of, he will have a look around and if there's something new and unusual, he will guess that's what you meant. Isn't that sort of the way humans learn? At least it's certainly not the way dogs are normally trained.

Re:What about grammar?

[wintermind]

Your assertion does not hold water. Vocabulary size is important because it tells us important things about the cognitive ability of dogs. Dogs that are able to learn a large vocabulary are able to keep track of a large number of objects and distinguish between them. No is claiming that Rico has the same capacity for abstract reasoning as even a three year-old child, but this is an impressive accomplishment.

As an aside, I consult with producers and trainers of working dogs (guide dogs, sleddogs, etc.), Two pertinent things that we have learned is that there is a genetic component to trainability (~20%); and that dogs do not always work the way that we think they do -- when a drug detector dog indicates on cocaine they are actually indicating on a byproduct of cocaine manuafacture.

Please do not sell the dogs short, even if they are not yet our new canine overlords.

Re:Max plank?

[vijaya_chandra]

There's actually no rule that you have got to name an institute after a person only if it's field is related to the person.

It's generally out of respect i suppose
like we have a Mahatma Gandhi institute of technology/medical sciences/business administration/.* in every city of India

good to know that even a scientist receives such a respect in germany

(Karma be damned; I am no better than an AC anyway)

Re:Bzzt. Try again

[NoMoreNicksLeft]

Chimps and other apes do sometimes fashion tools, which is to say they do more than pick up a rock... they actively shape a twig or branch to do what they want.

But the most impressive, was a crow that bent wire into a hook, to form a tool. Weird, eh?

As for being self-ware, recognizing yourself in a mirror... how useful is that as an indicator? I mean, pick some insect with compound eyes, a bee perhaps. Magically make it intelligent, could it recognize itself? What about some species that is naturally blind?

Also, I believe you people want to use the word sapient, not sentient. I would guess that there isn't a mammal in existence, that isn't sentient to an extent. Forgive me from borrowing from scifi, but Data (Star Trek) was argued to not be sentient, even though he was clearly intelligent. Sentience would be the ability to love, empathize, and lots of other things that are difficult to define.

Another interesting note on animal intelligence... anyone ever bothered to read up on octopi? These things can also solve problems if the reward is food, and they can learn to do so, simply by watching another octopus solve it. What's more, they have been known to climb out of aquariums entirely, across a floor, and into another to eat fish that they see.

Mostly, various religions have ingrained (maybe reinforced) the human tendency to discount any "lesser" animal as worthwhile. In modern times, that tends to amount to discounting their intelligence. I'm not about to stake my life on my cat scoring 190 on an IQ test, but it just seems right to think of her as a person. That tends to be difficult for those who can only assign value to an animal.

And lastly, in this mostly random rant of mine, I pose this question. If human intelligence can vary so greatly, from the barely more than vegetable, up to the ubergenius... why is it so hard to believe the same might be true of animals. And if they were already close to the lowest end of human intelligence, might not the occassional animal ubergenius be comparable to an average person? We might very well stumble across some dolphin that tells us to go fuck ourselves. (though how it will flip us the accompaning finger will frustrate it to no end).

Re:3 yr old toddlers?

[Canberra Bob]

My, arent we feeling superior today?

It is also possible that the dogs you have come across are stupid (yes, it may come as a shock to you but animals intelligence varies too).

OK, lets look at it another way. How many 3 year olds would you trust to lead a blind person around safely and successfully, day after day? And that is not something that is based on instinct either. Personally if I was blind I would chose the dog any day.

I would talk to my last dog in basic natural speech when I wanted him to do something and the vast majority of the time he would understand. Now if I said something like "We are going home" in the back paddock he would start heading home, if I said the same while out somewhere he would head for the car. Are you saying that is not reasoning? If I said "go to the car" he would head to the car irrespective of where he was.

Generally you will find the dog will be as intelligent as you treat it (sort of like people really). If you treat your dog as a "dumb dog" then all he will do is bark all day and dig up the yard. If you talk to him in natural language, and treat him as if he has some intelligence, the vast majority of the time he will respond by acting more intelligent.

Mirrors and being self aware.

[arevos]

Prove it. When's the last time an ape told you he wasn't sentient? There are many ways to determine if an animal is intelligent. One is being self aware. Only larger primates and dolphins can recognize themselves in a mirror. Another aspect is knowing of ones' lifespan. Only humans and a few primates are aware of our own demise.

I don't think the mirror-test is an accurate refleciton (no pun intended) of whether an animal is self aware. All a mirror shows is that the animal is aware of its body. And it wouldn't really be too hard to program a robot that could recognise itself in a mirror. Would that make it self aware? Nope. Because your self, your id, is considerably different to your body.

There is no current test for self-awareness. Now, I can tell that I am self aware, because I have a distinct concept of "self". I really can't be sure of anyone else, but I can assume that since others of my species exibit similar behaviour to me, I can reasonably assume that they possess the same trait of self-awareness that I do.

Dolphins and gorillas... Well, I'm not too convinced. They're intelligent, but I don't quite think that they're quite there; the evidence availiable doesn't make a good case, in my opinion. Though I'll admit that this is mainly due to no-one having inventing a convincing self-awareness test, yet.

Problem solving doesn't show an animal is self-aware. Recognising physical objects does not, either. I'm not entirely sure what does, however. Speech helps, of course. It could be that certain language patterns can only arise with self-awareness. It could be that a self-awareness is related to some effect on the quantum level, that cannot be replicated by a Turing Machine. There is some evidence to believe that a Turing Machine cannot represent a self-aware entity.

To be honest, we have such a crude definition of "self", that we'd need to figure out precisely what we mean when we talk about sentience, before we can start to think up tests for it.

Perhaps that will prove to be the greatest scientific challenge of our race.

Pullum is being too harsh.

[Estanislao Mart�nez]

The study is clearly aimed at arguing that the process psycholinguists call lexical access (which laypeople would probably call "remembering words") does not require innate structure specific to the human species.

In fact, if you know just a bit about contemporary research in child language you can pick up the hints in the AP article Pullum links about how it ties in:

The dog seemingly understood that because he knew the names of all the other toys, the new one must be the one with the unfamiliar name. "Apparently he was able to link the novel word to the novel item based on exclusion learning, either because he knew that the familiar items already had names or because they were not novel," said the researchers, led by Julia Fischer of the Max Planck Institute for Evolutionary Anthropology in Leipzig.

This is reminiscent of some of the work of Eve Clark-- which Geoff can't be excused not to know.

Re:Bzzt. Try again

[TheMCP]

Another aspect is knowing of ones' lifespan. Only humans and a few primates are aware of our own demise.

My first border collie knew she was old and sick. She selected her own replacement, taught him some tricks, and then once she saw he was doing okay with me, she died.

He lived until he was old and sick, and then I went off to college and he chose to commit suicide.

They understand that they can die, and they can choose when they're ready to go.

Look, I lived with border collies for 18 years. They weren't my pets, they were family. After 18 years of watching them, I believe they're not only as smart as people, but that part of the reason some people have problems with their border collies is that the dog is smarter than they are.

The thing is, there are two factors which prevent most people from understanding how smart they really are: one is that they can't talk (although mine tried and startled a few people by croaking out a kind of "hello" they don't really have the right vocal equipment) and the other is that they don't have the same priorities as people do: people worry about going to school and earning money and paying for the next vacation... border collies worry about making sure their family is happy and well, and they see you as their family.