|
| |
Language, Action and Nonconscious Knowledge
From: Cambridge University Press
| By:
Roger C. Schank |
EDITOR'S INTRODUCTION |
As humans, we're aware that we know that we know. But the knowledge we utilise and of which we are not aware is more significant and interesting by far. Here, Roger C. Schank offers some insights into the surprising workings of our minds in the most ordinary situations, from our use of language to the nonconscious knowledge employed while driving a car. |
 hen we understand a sentence--when you read this one for example--we are dealing with nonconscious knowledge. We know that we don't know how we go about attempting to understand a sentence, we just understand it. There is, of course, a big distinction, here as always, between that which we are taught about language (our conscious knowledge of language) and that which we remain blissfully unaware of but use all the time (our nonconscious knowledge of language). So, we can say things about nouns and verbs and parsing and concepts, things we were taught in school, but we really have no idea how we do what we do when we understand; we just do it. |
I often use the sentence "John prevented Mary from leaving the room" when I teach about how language understanding works. I ask students to tell me what this sentence means, using only concepts about actions that were specifically referred to. That is, when I ask what John did, and a student says "prevent," I do not allow that answer. "Preventing" is not something we can do. Why not? |
One way to answer this is to try to imagine the action or actions the sentence is attempting to depict. So, we can ask what John actually did. Students speculate that maybe John hit Mary, or barred the door, or yelled at her. He might have done any of these things, or he might have done none of them. The sentence simply doesn't say. All it says is "prevent"--and "prevent," I claim, is not an action anyone can directly do. |
 | |
| Car's-eye view: when you drive, you are using nonconscious knowledge. | |
|
At this point, students get frustrated. They cannot determine what prevent could mean at all. I ask them to tell me what they "know" happened, not what they can guess might have happened. At this point someone usually says that all that is known is that John did something and that, whatever it was, that action had the effect of causing Mary to not be able to leave the room. This is, of course, the real meaning of prevent. Further, it is the meaning that everyone "knows" prevent to have. But what kind of "know" is this? None of the students knew this meaning of prevent when I asked about it. It is always quite difficult to draw it out. Yet, as soon as one student recognizes what's going on, every other student in the class agrees, and they all claim to have known it all the time. |
The mechanics of language
What can this mean? Can these both be right? It is important to point out that there is no doubt the students knew it all the time. They are college students who can and do use the word prevent quite frequently in their everyday language. Nevertheless, they did not consciously know its meaning. Our conscious can become aware of knowledge that is nonconscious, by means that I have just demonstrated, but there is still a big difference between these types of knowledge. To see this in another way, consider the meaning of the word by. |
If you ask these same students what the word by means, they are dumbfounded. They cannot articulate anything about by. They can only cite examples of its use. But, when I further inquire about trying to understand what John did in the example sentence, after they have all concluded that they don't know what he did, they can suddenly recognize that if there were a by in the sentence, then they would indeed have known. Thus, if the sentence were "John prevented Mary from leaving the room by offering her ice cream," students would know what John actually did. |
Students insist that they of course "knew" that the action John did would be found after the word by, and they readily understand that the meaning of by is functional. By tells us how to properly decompose a sentence into its component parts. The rule is that after prevent, by indicates the action inherent in prevent that is the cause of whatever action is being prevented, which itself is to be found after the word from. |
This is parsing knowledge. Any speaker of a language has such knowledge. All my students "knew" this stuff, but none knew what they knew. It is all nonconscious. Further, it is all incredibly complex and detailed. The amount of such knowledge we all must have in order to communicate is unknown. However, we do know, from attempts to build computer programs that need such knowledge in order to analyze English sentences, that the amount of such knowledge needed for comprehension is huge (see R. C. Schank, Conceptual Information Processing, 1975; L. Birnbaum, "Lexical Ambiguity as a touchstone for theories of language analysis," Proceedings of the Ninth IJCAI, 1985, pp. 815-20). |
Much of what we are conscious of about language comes from what we have been explicitly taught. So we are conscious of nouns and verbs, and of poetic ways of talking, or of having heard eloquent speech. But, the mechanics of language, how it works, is unknown to us. One reason we don't consciously know about this is that cognitive scientists haven't officially discovered it; otherwise, it might well have become part of the curriculum. Prior to the advent of computers and the attempt to get computers to use language, this was not an area of science many people concerned themselves with. Thus, students have been spared having to memorize how by works and what prevent means. Such memorization would be worth no more than the attempts to make everyone parse sentences and understand gerunds, as is done in schools today. Unfortunately, once nonconscious knowledge is made conscious, it often finds its way into what we make children learn. |
What comes through loud and clear here is that nonconscious knowledge functions perfectly well without consciousness of it. We don't know what we know and that's okay. There is probably some value in knowing some of this stuff explicitly, at least for communication purposes, but not much. Knowing such things consciously can be useful for some people, but is not typically useful for the students who are taught it. |
The <I>but</I> test
As an example, consider the meaning of the word but. When I ask students what but means, I get the same blank stares I get when I ask about the real meaning of any other word. Of course, students can use the word but properly, but they cannot articulate what it means. This does not mean they don't know it, of course; it only means they aren't conscious of what they know. However, the nonconscious knowledge of the word but that we do have, when made conscious, can be quite useful if we are trying to get computers to understand English. |
But means, in actual nonconscious use, that whatever proposition follows but is intended to negate the normal implications of the proposition that preceded it. So, for example, if I say, "It was cloudy all day but it didn't rain," I am saying that normally you would have expected rain but it didn't happen. If I say that John hit Mary but she didn't cry, I am implying that I expected that she would cry after being hit. A student of language needs to know consciously the meaning of the word but so that he can use this knowledge in attempting to understand what a sentence means. For example, the sentence "John hit Mary but it didn't rain" would be seen as silly. Hitting doesn't imply rain. It is important to know this if we are going to analyze what a word (like hit) does mean. For example, if I say that John hit Mary but his hand didn't touch her, we can understand that this might be true, but it seems odd. If I say that John hit Mary but nothing at all ever came in contact with her, then this seems not only odd, but unlikely to be true. |
By looking at sentences containing but in this way, we can come to know what we know about words. For example, we can see from these sentences that the meaning of hit certainly includes physical contact between the object and something, that we can assume it means contact between the hand of the actor and the object, but that this might not be true, and that it implies nothing at all about rain. Becoming conscious of what we know about the meaning of words is very important in studying the mind and how it works. But as we can see, it is unimportant to teach a student any of this, because he already knows it. We know what hit means--the but test shows us that we do. This knowledge is nonconscious and is critical to the basis of what it means to be an intelligent human being. Our conscious knowledge is useful, but being conscious of how language works is not at all important for the user of language. |
The conundrum is this: As we become more knowledgeable as a society about the nature of the universe, we believe it is this knowledge that makes us human. We feel, as a society, that imparting this knowledge is what the educational system should be all about (see S. Strauss and T. Shilony, "Teachers' models of childrens' minds," in S. Gelman and L. Hirschfeld, Mapping the Mind, 1994). If we understand the facts about how language works, we feel we should teach that knowledge to children. But, there is no reason at all to do this. Knowing how things work doesn't make us better at doing them. The person who understands why a curve ball curves will not necessarily know how to pitch, and a pitcher who can throw a great curve ball is more than likely clueless about why the ball does what it does when he throws it. A person who is articulate in his use of language does not possess this ability because he is conscious of how by and from work. He is articulate because he practices language as a user, not as a scientist. From this it follows that as long as schooling concentrates on conscious knowledge, it is doomed to have little real effect. |
Of course, I am not talking only about language here. Everything we know about everything is important to our ability to comprehend the world around us, but we very rarely know that we know any of it. For example, when we enter a restaurant and look for a table and decide what we want to eat, we haven't a clue as to how this all works. We can't say how we recognize restaurants, or how we know who the waitress is, or what is and is not a table. We can move our bodies in such a way that they get to the table and seat themselves, but if anyone ever asked us to describe how to walk or sit down, we would become tongue-tied. Perhaps we can give a rational explanation of why we have chosen what to order, but when we simply feel like having a hamburger, we cannot really say why. |
Significant thinking processes
All of what I have described here belongs to the realm of nonconscious knowledge; we use it in thinking, but we cannot say how. We can say why we decided to take an important decision like accepting a new job, or buying a house or a car. We like to imagine that this, at least, is the stuff of conscious thought, but there is plenty of evidence to show that the unconscious influences in such decisions are profound indeed (see A. Tversky and D. Kahneman, "The framing of decisions and the psychology of choice," Science 221, 1981, pp. 453-458; E. Langer, Mindfulness, 1990; and S. Freud, The Psychopathology of Everyday Life, trans. A. Tyson, 1965). |
Our racing minds that keep us awake when we want to be asleep must also be working with nonconscious knowledge. We are sometimes aware of the thoughts we have in these episodes, but often we are not aware of the knowledge upon which those thoughts are based. But, the nonconscious knowledge used in such thinking, such as our views of people or remembrance of events, seems quite different than the nonconscious knowledge we use when we read a sentence or decide how to sit down. We don't know (consciously) that we have to adjust our bodies in a certain way to get them into a sitting position--we just do it (using the unconscious knowledge we have about how to do this). Surely this is something that animals do. Is this really knowledge? |
To see what kinds of knowledge we are dealing with, consider driving a car. Animals don't drive cars, and although they might presumably be taught to do so, it is unlikely that they could learn to avoid accidents. (The reason is simple. Accidents damage cars, and damage to a car is bad only if you value that car.) When we teach someone to drive a car, we attempt to make explicit to them in words what we do not really have explicitly represented in our minds. We try to articulate how pressure should be applied to the brake, how the wheel should be turned back after turning, and that the correct way to stop at a stop sign is by inching forward. Although I am sure that professional driving instructors know exactly how to say all this to their charges, I am equally sure that most everybody else does not. I can easily recall, while attempting to teach my son to drive on the streets of Chicago, that I could not properly explain how to go through a stop sign on a crowded city street. I had to watch myself do it later, when he was not in the car, and then remember what I had done, so that I could describe it to him. He was quite annoyed with me when I said that stop signs meant to stop twice, not once, but I was certain we were going to get hit when he insisted on stopping at a stop sign once and then starting back up again. |
Am I conscious when I drive? In any standard definition of consciousness, I am. Yet I don't know how I do what I do. I am using nonconscious knowledge. In fact, sometimes I drive where I wasn't supposed to be going because I am not thinking about where I am going. When I do this, I am surely conscious, I am surely thinking, but when I am lost in thought in this way, and a companion asks what I was thinking about, I don't always know. How can this be? |
My argument is that what passes as consciousness--our awareness that we have had a thought and can sometimes say what that thought was about--is actually a rather uninteresting part of the human thinking process. The more significant human thinking process, or processes, are nonconscious. We have no control of them and most of the time we are totally unaware of them. But, these processes make up the bulk of what we are talking about when we talk about thinking. Similarly, what passes for knowledge is conscious knowledge, when the really interesting and important knowledge we have is unknown to our conscious selves. |
|
| |
