I’d say we’ve always known there’s more to “intelligence” than playing chess. Mir Sultan Khan, for instance, achieved chess Grandmaster status without being able to read or write. I’d also say that a computer that could explain why it chose a particular chess move more completely than “I ran all the possibilities from the starting position through my evaluation program and this move got the highest rating” might indeed show intelligence.
I would propose a friendly amendment: delete the word “substantially.” I would agree with Charles Darwin, who held that the differences between human and nonhuman animal intelligence were differences of degree, not of kind.
Someone trained me to use the words I used in the way I used them, or to use tools that explain to me the way to use the words I used. Someone trained me that the darker portions of this screen represented symbols that represented sounds that communicated meanings, etc.
I would find it hard to believe that your son suddenly came up with “Daddy, don’t throw the ball to me, throw it to my sister” without having been trained at some point on what that sequence of words meant, how he could put words together to communicate meaning and so on.
My argument at this point would be that, whether or not there is or is not a “deep structure” somewhere for language, humans are not likely to be able to use such a structure in novel ways without first having been trained to use it or having come to some mutual understanding about how to use it. Ditto for nonhuman animals.
I agree that you have to know what all the words mean and how they are allowed to be strung together to make a valid sentence, but your command of language allows you to string them together in complex and novel ways to convey meaning to me. This is [substantially] different than stringing together a couple of hand signals or sounds, without regard to grammar, syntax, etc.
When a dog is hurt, it may whine; when it is angry, it may growl or bark. That’s not really language, although it does communicate. I’m not saying that dogs and monkeys cannot communicate, I’m just not convinced that they can use language.
Communicating by pointing to a banana and then pointing to yourself can get the idea across that you want the banana. However, it’s no substitute for “Please pass me that banana, since I haven’t eaten for a couple of hours and I’m getting peckish.” That sentence, which can be arbitrarily long and complex, shows (IMHO) language being used, rather than just simple communication.
Even something as simple as JW Kennedy’s ball example would show a better understanding of language than (I believe) has been accomplished by dogs or monkeys. I don’t need to train my son to “find the ball” specifically – he knows what “find” means and what “ball” means, and that when they are strung together that way, it means what it does. I could then ask him to “find the red ball” or to “find the red ball that is on top of the table, not on the floor”, and he could parse that sentence and get to it, even if he’s never heard that particular command before.
I probably should confess that I’ve read a few of Pinker’s books, including The Language Instinct and I really liked them. As ejh mentioned in some other thread, Pinker is a follower of at least some of Chomski’s ideas.
It seems that people both over- and under-estimate animal abilities. People act like it’s amazing that a dog or monkey do something that demonstraights nothing more comples than eyesight, and then are willing to give our fuzzy friends credit for enjoying the humor of Ambrose Bierce.
As far as I can tell, Alex does nothing but follow a simple set of rules. It’s a large set of rules, but still very simple. This seems to be the case with marine mammals, as well.
The apes are interesting. The kind of babble they make… well, it does tempt me to transcribe a conversation with a three-year old. I believe that most of the apes who’ve had this training do, to some extent, “get it” that symbols stand for things, but I’ve can’t work out an arguement for this, and nothing I’ve read proves it. That’s not language as we know it, but it’s special.
Actually, I would be easy to have a dog respond this way. It’s not hard for a dog to learn a word like “ball.” You could argue that the dog was doing whatever fancy mental work and get a paper published and go on David Letterman, but you’d be either a fraud or a fool.
I’ve known horses who knew 100+ voice commands and, because they were agreeable horses, it often looked like they “understood” new commands. The reality, IMHO, is that they were following a simple set of rules. For instance, a horse who was trained to make an eager face in response to the phrase “Do you want alfalfa” and a mean face to the phrase “Do you want worming paste” would understand the tone of voice in “do you want” and “worming paste” well enough to respond properly to new nouns. If you wanted to do a sideshow act, you could easily (not quickly, but easily) teach a horse to respond with a silly face when you asked a question with your right foot forward, and a mean face if you left foot was forward. Then you could delight the rubes by standing in front of the horse offering wild variety of things, and watching his response to your questions.
This is a true story: Hope the service dog was outside, with her person, when the person had an accident. Hope jumped through a glass window to get inside the alarm, pushed the alarm button, and then went back outside, laid down with her person, and waited for help to come.
It’s an amazing story; not a display of fabulous intelligence. Hope had been selected for training because she was smart, aggressive (as in hard working and determined) and even tempered. She was trained to push the alarm when certain things happened - and she had innate abilities and fostered confidence that made if possible for her to overcome obstacles to achieve that goal. She was trained to stay with her person - it was perfectly natural for her to go back outside after pushing the alarm. The magical part is the accidents of evolution responsible for the canine/human bonding that motivated Hope during her training and her life as a service dog.
If you want to look at animal intelligence, you should look at how quickly they catch onto the idea of training, and to some extent how willing they are to work to learn new things.
(A horse, for instance, really needs a good break after as little as 10 minutes of work. If you ask a horse to make a mental effort for more than 20 minutes, he will sour quickly. An animal that’s had a lot of training will develop trust and can work longer.)
But you couldn’t get him to do the latter without first training him as to what “red,” “on top of,” “table,” “on” and “floor” all mean. I’d argue – and if I were a betting man, and I am, I’d bet – that a dog that “knew” all of those concepts could perform the new task “find the red ball that is on top of the table, not on the floor” (or a similar task adapted to canine abilities, i.e. one that might not involve distinguishing red from green and/or following multi-part instructions) at least as well as your son could when both were presented with it for the first time.
A different approach might be to ask your son and a dog to perform the new task using language neither has been previously trained to understand; would either be able to “traiga la bola roja que esta’ encima de la mesa, no la que esta’ en la pared”?
We might then argue over whether this ability would demonstrate “language use/understanding” or merely “intelligence,” but I believe I’d find that kind of argument fairly uninteresting.
Hope’s story is amazing, but Coren has one to top it. During a house fire, a companion dog woke a young girl, led her through the house to the latched back door, unlatched it with his nose and led the girl to safety in the back yard. The rest of the family had made it to safety in the front yard. That dog had not been trained for service nor had it ever been in a fire. It may have learned to unlatch the back door with its nose in order to chase squirrels. Amazing story or canine intelligence; you be the judge.
But, following the multi-part instructions is exactly what I mean by understanding language. The ability to string together known words to form new concepts according to a grammar or syntax (I’m not sure which term is correct here) is precisely what I would consider language ability.
Your example using a foreign language is a good one. My son would have no idea what you were talking about because he wouldn’t understand the words. A dog (who was previously trained to respond to the exact instruction) may actually be more successful in that test – he (the dog) may respond to your tone of voice, cadence, and expression and go and get the correct ball. If you told him to “bind da fred call cat biz cron da fable”, he would probably go and get the ball he was trained to get upon hearing that sort of phrase. My son, trying to understand what you said (rather than responding to cadence, expression, etc.) would be baffled. That’s because my son wasn’t trained on that particular command, rather he understands language.
I would also be willing to bet that a dog that understands all those words would NOT be able to distinguish between:
Find the red ball that is on top of the table, not on the floor
Find the red ball that is not on top of the table, and not on the floor
and would be baffled by:
3. On top of the table, there is a red ball. Find that ball.
4. Find the ball – the ball that is not on floor, the one on the table.
Mmmmmm. Maybe. But Coren’s pretty convincing on the issue of whether dogs respond to actual words (his and, I think, my contention) or only or primarily to extra-verbal and/or non-verbal cues.
I do find my (obedience- and agility-trained) dog responds in different ways to different tones of my voice, more readily to a sort of low growl for example. But she never, for example, lies down when I say, “sit.” “Sit” always (eventually) results in the same behavior on her part regardless of my tone or cadence. And as was noted, dogs are demonstrably better than nonhuman primates at responding to human non-verbal cues.
I readily grant the point that dogs can’t make head or tail (pun intended) of human syntax and/or grammar. It takes no little study before humans can understand the canine equivalents of syntax and grammar, as far as that goes. I’d still maintain that the difference in their “language” ability and our “language” ability is one of degree, not of kind. As I said, making invidious distinctions between “language ability” and “intelligence” seems pointless to me.
Tom – once again, your example is quite good to prove my point. I remember seeing a demo of a computer program quite a few years ago that could parse commands and perform the action requested. The universe was quite small – just colored blocks of different shapes. But you could give it pretty complex commands, like “place the red cube on top of the yellow square”, and so on.
This didn’t illustrate any real intelligence, rather it showed that a computer could be programmed to parse sentences in a limited environment. However, it had more “language ability” than a dog.
However, your story illustrates that dogs are really quite intelligent, even if they aren’t good with language. There has never been a computer smart enough to find its way out of a burning building, let alone the infinite other situations that a dog can figure out. They can barely make a robot with enough pattern recognition to roll down the sidewalk without going off the edge. Dogs have no such problems.
So, while you could program a computer to have some language ability, even arguably better than a dog, computers (today) are nowhere near as intelligent as dogs (or fish, for that matter).
That’s how I would make a distinction between intelligence and language ability.
Not necessarily. I have an aunt and uncle who, when raising their kids, would always spell out i-c-e-c-r-e-a-m so the kids wouldn’t know what they were talking out. Except that the children quickly figured out that if their parents were spelling something out, they were talking about ice cream, and started asking for z-x-c-s-f-w-g, or other random combinations of letters, when they wanted ice cream. So they recognized the pattern of “spelling out a sequence of letters” without actually recognizing the particular letters.
That gets us down to the $64 question, which seems to be whether we can talk about something called “intelligence” that is more than the sum of its parts; that is, whether we would or should grant “intelligence” to a machine that could parse sentences, explain games strategy, discuss philosophy, compose music and poetry, show emotions and in every observable way act like “one of us” even though we knew it was merely carrying out instructions, however complicated.
The flip side of that question is whether we humans and, by extension, any other entity to which we extend the courtesy of terming “intelligent,” are actually doing anything more than carrying out particularly complicated instruction sets.
I don’t feel enough is known to arrive at a definitive answer. But I do feel that if one is to be consistent with a Darwinian, evolutionary view of nature – human and otherwise – one has to lean toward the conclusion that “intelligence” is simply the product of many small, simple decisions and hundreds of thousands of years of evolutionary pressure. And I do feel that a growing body of evidence is pointing in at least the general direction of that conclusion.
The Chomskian camp apparently claims there is some “deep structure for language” that humans, but not nonhuman animals, possess. That kind of argument seems to owe much more to Descartes than to Darwin.
Chronos – I’m not saying that humans can’t be trained, as in your example where the kids were trained that to utter a sequence of apparently random letters could bring ice cream. Dogs, monkeys, birds, and humans can all be trained that way. Still different than using language, though, I would argue. I have wondered, though, if you brought up a kid and only spelled out everything you say, never uttering the words, would the kid learn to speak that way? I’m not willing to try it on my own kids.
Tom – this is getting pretty off-topic, but I agree that intelligence is probably the product of small, simple decisions. One of the points of Cecil’s responses is that teaching animals to use limited language in limited ways is more along the lines of a neat trick and any demonstration that animals can use language in any really substantial way. Animals may be intelligent (they hunt, they avoid being hunted, they solve problems, they recognize patterns, they learn tricks, they communicate, etc.), but that is different than a language ability. Language is a really, really useful tool, not necessarily a sign of intelligence. Humans have limitations in areas where animals excel (I can’t manipulate objects with my nose the way an elephant can and I can’t do the amazing acrobatics on tree limbs the way monkeys can).
Again, we’re back to what we mean when we say, “language.” I grant your points about nonhuman animal language use if we mean “human, verbal communication including complicated grammar and syntax.” If we can broaden the definition to include extra-verbal and non-verbal modes of communication – what I’d like to think of as some of the building blocks from which human language evolved – then I think we have to give animals a little more credit.
First, a working definition of language for the purpose of this post: a system of communication composed of arbitrary signs which are combined into utterances based on a syntax. Users of the system must be able to communicate about to abstractions, absent objects, and past, present, and future events (and distinguish which of these it is they’re communicating about). They also must be able to make statements, ask questions, and issue commands, and to differentiate between these types of utterances.
I used to work with Alex the parrot in Dr. Pepperburg’s lab. While Alex is initially impressive, I don’t feel that he has any abilities that deserve the term language. Some reasons:
Alex has never asked a question.
While Alex can correctly associate sounds with objects, he has no ability to communicate whether he is identifying the object, requesting the object, inquiring as to the location of the object, etc.
Alex cannot talk about events.
Alex has no syntax.
I also disagree strongly with Dr. Pepperburg’s claim that Alex “bosses around lab assistants in order to modify his environment!” Alex often repeats stock phrases like “Wanna grape” but when given a grape in response to this, throws it on the floor as often as he eats it. Of course the true believers say this is a game he likes to play, because he thinks it’s funny that us poor lab assistants have to pick the grape up and put it back on his table no matter how many times he throws it down. Like fontor said, “when there’s a miss, treat it like a hit, and keep going.”
In response to Tom Arctus’ comment that, “The Chomskian camp apparently claims there is some ‘deep structure for language’ that humans, but not nonhuman animals, possess. That kind of argument seems to owe much more to Descartes than to Darwin.” I would point out that there are differentiated brain structures known to be involved in human language processing (i.e. Broca and Wernicke’s areas) that other primates lack. If I claim that fish posses a “deep structure for breathing under water” that mammals lack, no one bats an eye. Why are people so hesitant to accept that language is an ability humans have based on unique brain structures that developed in our evolution and not in the evolution of other animals?
J.C. wrote, “. . . it does tempt me to transcribe a conversation with a three-year old.” I’ve done this. It’s fascinating. Kids that young use an amazing range of grammatical constructions. The abilities of a three-year-old make any animal’s abilities look like a joke. What’s more, kids gain these amazing abilities without any explicit training.
As a footnote, because this debate is sharply divided along disciplinary lines, I should mention that I did an undergraduate degree in linguistics. I think the reason you find this debate divided along disciplinary lines is that people coming from disciplines other than linguistics tend to confuse animal intelligence with animal language (as can be seen to some degree in this thread). People who study language, and realize how amazingly complex it is, tend to see pretty quickly what the animals lack.
Zach – thank you, you have put those points together far more succinctly than I’ve been able to. I have no formal training in linguistics. Regarding a question I half-raised in an earlier post – if a child were brought up in an environment where all words were spelled out instead of spoken, would he be able to speak that “language”? I’m sure he couldn’t say the words, but would he be able to spell out whole sentences, group the spoken letters into units and use those unit when making new sentences? I was wondering whether there were any linguistic theories on that.
> As a footnote, because this debate is sharply divided along
> disciplinary lines, I should mention that I did an undergraduate
> degree in linguistics.
I have a master’s degree in linguistics, and I’m not utterly convinced that the dividing line between animal and human language is as sharp as you think it is. What humans do is certainly much greater than what animals do, but I think that we can just call the abilities of humans and animals a spectrum on which the humans are much farther along. I’m suspicious of Chomsky’s claims to have found a set of constraints on language because I’ve actually read his various claims. He does a lot of the following: He claims that some particular constraint holds in all language. He derives this constraint by looking at a bunch of English sentences and claiming that this constraint holds there. When I look at these sentences myself, I’m not even sure that the constraint holds in English. He tells his followers to show that these constraints hold in all human languages. When, after a few years, it becomes clear that the constraint is badly formulated or just doesn’t work, he simply forgets that constraint and moves on to a new one. It’s not that everyone who’s looked at human language carefully hasn’t been convinced that there are constraints on it that definitely look like they’re inborn, it’s just that it’s hard to state those constraints with any precision.
Many species of animals especially ones that hunt in groups or packs (like chimpanzees) have amazingly subtle systems of communication. They are able to do many communal taskswhich involve the transfer of information.
However the use of recursive sentences and conditional tenses (for instance) make it possible for humans to discuss theology, quantum mechanics and linguistics-
it strikes me that this is a evolutionary adaptation that is unnecessarily powerful for the required task (of hunting and gathering), like the antlers of the giant elk.
It is also very specific to humans, and a quantum leap away from the functional communications of pack hunters.