CHAPTER 1
Introduction
1. Radical scepticism
Much of current epistemology (the theory of knowledge) in philosophy is best seen as a response to the thesis that we never have the slightest reason to believe anything. This radical skepticism must be distinguished from the more commonsensical idea that nothing can be known for certain and that we can never be absolutely sure of anything. Common sense assumes that practical certainty is possible even if absolute certainty is not. Radical skepticism departs from common sense and denies that even practical certainty is ever attainable. Indeed, it denies that anything is ever even the slightest bit more likely to be true than anything else.
The problem is not that there are radical skeptics who need to be convinced that they are wrong. The problem is that an extremely natural line of argument seems to lead inevitably to radical skepticism. Common sense keeps us from accepting such a conclusion; but that leaves the philosophical problem of saying what goes wrong with the reasoning that seems to lead there. To repeat, the problem has not been to find an argument against skepticism, it has been to find out what is wrong with an argument for skepticism.
One such argument begins by asking how you know that the color red looks to someone else as it looks to you. Perhaps things that look red to you look green to him, things that look blue to you look orange to him, and similarly for other colors. If his spectrum is inverted compared with yours, there may be no way to discover it, since what is for you the experience of red he will call the experience of green. Even if he describes colors exactly as you do, there could still be a systematic difference between his and your experience of color. This suggests that you can have no reason to suppose that others see the world as you do rather than as does the man with an inverted spectrum.
Further reflection suggests that there is no reason to suppose that visual perception gives other people experiences that are anything like your visual experiences. Perhaps someone else has what would be for you auditory experiences. When he looks at the blue sky it is like what hearing middle C on the piano is for you. There seems to be no way to tell, since he would have been brought up to call that sort of experience the experience of blue. Indeed it is not clear that you have the slightest reason to suppose that others have anything you could recognize as experience. When others see things, their visual experience may be something you could not even imagine.
It might even be suggested that there is no reason to suppose that others have any experience at all. The suggestion is that, even if you could know that the people around you were made of flesh and blood, born of women, and nourished by food, they might, for all you know, be automatons in the sense that behind their elaborate reactions to the environment there is no experience. But the suggestion is not merely that you do not know whether other people have any experience but also that you haven't the slightest reason to suppose they do.
Similarly, it might be suggested that you have not the slightest reason to believe that you are in the surroundings you suppose you are in, reading a book called Thought. It may look and feel to you as it would look and feel if you were in those surroundings, reading such a book. But various hypotheses could explain how things look and feel. You might be sound asleep and dreaming — or a playful brain surgeon might be giving you these experiences by stimulating your cortex in a special way. You might really be stretched out on a table in his laboratory with wires running into your head from a large computer. Perhaps you have always been on that table. Perhaps you are quite a different person from what you seem: you are a volunteer for a psychology experiment that involves having the experiences of someone of the opposite sex reading a book written in English, a language which in real life you do not understand. Or perhaps you do not even have a body. Maybe you were in an accident and all that could be saved was your brain, which is kept alive in the laboratory. For your amusement you are being fed a tape from the twentieth century. Of course, that is to assume that you have to have a brain in order to have experience; and that might be just part of the myth you are being given.
You might suppose that there are actually little differences among these hypotheses. After all, you have the same experiences on all of them. So you may feel that, no matter what is actually the case, everything will work out fairly well if you continue to act on the assumption that things are roughly what they seem. But that seems to go beyond your evidence. It may be that up until now everything has worked out well on the assumption that things are roughly what they seem to be. But what reason is there to suppose that things will continue to work out well on that assumption? An inductive inference is needed here: "Things have worked out well in the past; so they can be expected to work out well in the future." But how could you justify this use of induction? You might argue that, since inductions have generally worked out well in the past, you are entitled to expect them to work out well in the future. But that would be circular. You would be giving an inductive argument in order to defend induction.
Furthermore, do you really have any reason to suppose that things have worked out well in the past? You may seem to remember that they have. But how can you justify relying on your memory? Surely not on the grounds that your memory has been reliable in the past, for what reason do you have for thinking that it has been? The fact that you seem to remember that memory has been reliable in the past is irrelevant. You cannot legitimately appeal to memory to justify memory. You seem to have no reason to suppose that things have worked out well if and when in the past you have acted as if things are much as they seem.
Such reflections lead to the philosophical problems of other minds, the external world, induction, and memory. It is important that these problems are not simply whether we could ever know for sure or be absolutely certain of various matters. They are more radical: How is it possible that we should have the slightest reason to continue to believe as we do? How can there be more reason to believe what we do about the world than to believe anything else, no matter how incredible it may seem?
Recall that our interest in these problems is not that we want to be able to refute the radical skeptic. It is rather that we suppose that we often do have reasons to believe one thing rather than another and can sometimes even be practically certain, even if not theoretically certain, that we are right. Common sense tells us that something must be wrong with the arguments for radical skepticism. The philosophical problem is to spell out what is wrong.
Showing where the arguments for skepticism go wrong is philosophically interesting only to the extent that it tells us something interesting about reasons, inference, knowledge, and related matters. This book presents an account of the nature of thought that derives from an answer to skepticism. If the account is interesting, it is not because of light shed on skepticism but because of what is learned about these other things.
But before I present my own theory, I want to discuss a popular response to skepticism which ultimately leads nowhere.
2. The appeal to meaning
Some philosophical theories of knowledge appeal to meaning in the attempt to answer skepticism. The skeptical claim is said to be meaningless or false by definition. For example, philosophical behaviorism holds that statements about experience have the same meaning as do statements about behavior. On this view, the problem of the inverted spectrum is meaningless, since what it means for someone to see the world in the way that you see the world is that you and he are disposed to behave in the same way in the same perceptual situations. The supposition that you and he have roughly the same responses to color but that his spectrum is inverted with respect to yours would have no meaning.
According to philosophical behaviorism you can inductively discover someone's behavioral dispositions on the basis of purely behavioral evidence. You can know about his experience because you know that having certain experience is equivalent to having certain behavioral dispositions. This equivalence holds by virtue of the very definitions of words like "pain," "anger," "hunger," and so on. You know that the equivalence holds because you know what these words mean.
Similarly, a phenomenalist claims that statements about physical objects are equivalent in meaning to complex statements about possible sensory experience. According to a phenomenalist, to say that there is a typewriter on your desk is to say that under certain conditions it would look as if there were a typewriter on your desk, that certain experiences of seeming to type can be followed by experiences of seeming to read something freshly typed, and so forth. Phenomenalists argue that you know about these equivalences simply by virtue of knowing the meaning of statements about the external world. You use induction to justify statements asserting the possibility of sensory experience. Then you use the meaning equivalences to justify your views about the world.
It is true that philosophical behaviorism and phenomenalism appeal to induction as well as to meaning; but some philosophers have argued that inductive conclusions are themselves justified simply by virtue of the meaning of "justified." To reach a conclusion by induction is said to be almost by definition one of the ways you can justify a conclusion. Similarly, to say that something is probable is held to be to say that there are good reasons for it; in this context "good reasons" is supposed to mean "good inductive reasons." It is said to be "analytic" (i.e., true solely by virtue of meaning) that inductive reasons are good reasons, analytic that inductive conclusions are probably true, and analytic that inductive conclusions can be used to justify belief.
Some philosophers, not wishing to accept strict philosophical behaviorism, make a similar but weaker claim. These philosophers say that to understand the meaning of statements about someone else's experience you must realize that certain kinds of behavior will provide good, but not absolutely conclusive, evidence of that experience. According to these philosophers, it is part of the meaning of the word "pain" that certain behavior, such as moaning, is evidence that someone is in pain. Such evidence is said to be a "criterion" of pain. You are supposed to be able to know that others are in various psychological states because you can observe their behavior and know criteria for these states. You are supposed to know the criteria, not by virtue of any empirical investigation, but simply because you know the meaning of your words. So, for these philosophers, certain statements relating behavioral evidence and psychological states are analytic, true by virtue of meaning and known to be true by virtue of your knowledge of meaning.
Similarly, a philosopher who does not wish to accept phenomenalism can still argue that there are purely sensory criteria for statements about physical objects. Experience is evidence for claims about the external world, where the evidential relationship can be said to hold by virtue of meaning. It is said to be analytic that, if you seem to be seeing a blue book, that is adequate reason to believe that you are seeing a blue book, other things being equal. You are said to know an analytic truth like that because you know the meaning of the claim that you are seeing a blue book. You are said to be able to use experience as evidence about the external world because you know the meanings of your words.
The reliability of memory can also be defended by an appeal to meaning. The more radical view, corresponding to philosophical behaviorism and phenomenalism, is that statements about the past are equivalent in meaning to statements about present evidence. The weaker version has it that it is analytic of the notion of an apparent memory that apparent memories are good but not absolutely conclusive evidence for what is apparently remembered.
3. The revolt against meaning
According to philosophical behaviorism, statements about another person's experiences can be translated, without loss of meaning, into statements about his behavioral dispositions. Similarly, according to phenomenalism, statements about objects in the external world can be translated without loss of meaning into statements about the possibility of one's own experience. In either case, knowledge of truths about the experiences of others or about objects in the world is supposed to be made possible by knowledge of these translations, and that knowledge is supposed to be available simply by virtue of the fact that one knows the meanings of one's words. Now, it was always embarrassing to philosophical behaviorism and to phenomenalism that no one was ever able to give a single example of such a translation. Lately it has become clear that the required sort of translation is impossible. No single simple psychological statement is equivalent in meaning to any nonpsychological statement that is purely about possible behavior.
A belief is not a simple behavioral disposition. At best, it is a disposition to behave in certain ways given certain desires. On the other hand, a desire is at best a disposition to act in certain ways given certain beliefs. This means that there can be no noncircular way to give a purely behavioristic analysis of belief or desire. There is no way to translate a simple statement about belief or desire, without loss of meaning, into a statement that speaks only of purely behavioral dispositions. The same point holds for simple statements about other psychological states and about experiences.
Similarly, to say that there is a typewriter on your desk is not to say, among other things, that under certain purely experiential conditions it would look to you as if there were a typewriter on your desk. For these conditions must include such things as that your eyes are open, that nothing opaque intervenes between you and the typewriter, that you have not just taken a hallucinogenic drug, and so on. A statement of relevant conditions must speak not only of possible experience but also of things in the external world. There is no way to translate simple statements about objects in the external world, without loss of meaning, into statements that are solely about possible experience.
Psychological statements are to statements about behavioral dispositions, and statements about the external world are to statements about possible experience, as statements of theory are to statements of possible evidence. The failure of philosophical behaviorism and phenomenalism illustrates the fact that individual theoretical statements cannot be translated without loss of meaning into purely evidential statements. The point goes back to Duhem, who noted that an individual theoretical statement does not have evidential consequences all by itself but only in conjunction with other theoretical assumptions.
A more basic objection can be raised against any philosophical theories that attempt to answer skepticism by appeal to meaning. Skepticism maintains that one never has the slightest reason to believe one thing rather than another. In attempting to meet skeptical arguments, one eventually reaches a point at which (a) one wishes to claim that certain evidence provides a good reason to believe a particular conclusion but (b) one cannot offer further grounds for this claim in response to any further skeptical challenge. In another age, we might have said that one can intuit the connection between evidence and conclusion, but it is no longer felt that an appeal to intuition can explain one's knowledge of such a connection any better than an appeal to magic would. In this context analyticity seems to provide just what we are looking for. If the connection between evidence and conclusion holds by virtue of meaning, it would seem that one can know that the connection holds simply by virtue of knowing what one means. However, two questions immediately arise: (1) How does one know what one means? (2) How does knowledge of that meaning give one knowledge of the connection between evidence and conclusion? Unless these questions can be satisfactorily answered, the appeal to analyticity is as empty as an appeal to magic or to intuition.
