Introduction

In countless movies, computers play a starring role. Some talk in synthesized voices; others write a stream of words on a screen. Some manage spaceships; others, the “brains” of robots, manage their own “bodies.” People converse with them, are understood by them, exchange information and greetings with them.

Still, a suspicion remains. We human beings have always thought of ourselves as special. We all assume some contrast between the world of material things and the world of spiritual things. If the computer really is a “material mind,” then not only must we rethink this distinction, but we have broken it with our own creations. We should be careful to avoid such an important conclusion until we have really thought it through. However natural it seems to take it for granted that computers can think and act, then, we shouldn't just assume it. In philosophy we often find that what we normally take for granted—the “commonsense” point of view—gets in the way of a proper understanding of the issues. So let's see if the way I spoke about computers in the first paragraph is accurate.

I said that they talk. But do they really talk in the sense that people do? It isn't enough to say that they produce something that sounds like speech. Tape recorders do that, but they don't talk. When people talk they mean something by what they say. To mean something, they need to be able to understand sentences.

Well, could a machine understand? There are two obvious responses to this question. The first response I'll call mentalist, for the sake of a label. It's the response you make if you think that understanding what people say involves having a mind. The mental- ist says:

Computers can't really understand anything. To understand they would have to have conscious minds. But we made them from silicon chips and we pro­grammed them. We didn't give them conscious minds. So we know they don't have them.

At the other extreme is the response I'll call behaviorist. The behaviorist says:

Naturally, everyone should agree that some computers don't understand. But there's no reason why a computer couldn't be made that does understand. If a machine responds in the same ways to speech as a person who understands speech, then we have just as much reason to say that the machine understands as we have to say that the person does. A machine that behaves in every way as if it understands is indistinguishable from a machine that understands. If it behaved in the right way, that would show that it had a mind.

It is clear why I call this response “behaviorist.” For the behaviorist says that to understand is to behaυe as if you understand.

What we have here is a situation that is quite familiar in philoso­phy. There are two opposing views—mentalist and behaviorist, in this case—each of which seems to have something in its favor, but neither of which looks completely right.

In fact, if you hold either of these views you can face difficult intellectual choices. Let's start with a problem you get into if you are a mentalist. Suppose the computer in question is in a robot, which, like androids in science fiction, looks exactly like a person. It's a very smart computer, so that its “body” responds exactly like a particular person: your mother, for example. For that reason I'll call the robot “M.” Wouldn't you have as much reason for thinking that M had a mind as you have for thinking that your mother does? You might say, “Not if I know that it's got silicon chips in its head.” But did you ever check that your mother has got brain tissue in her head? You didn't, of course, because it wouldn't prove anything if you did. Your belief that your mother has a mind is based on what she says and does. What's in her head may be an interesting question, the behaviorist will say, but it isn't relevant to deciding whether she has thoughts. And if it doesn't matter what is in your mother's head, why should it matter what's in M's?

That's a major problem if you're a mentalist: how to explain why you wouldn't say an android had a mind, even if you had the same evidence that it had a mind as you have that your mother does. Surely it would be absurd to believe your mother has a mind on the basis of what she does and says, yet refuse to believe M has a mind on the very same evidence. If it's the evidence of what your mother does that entitles you to believe she has a mind (and not, say, an innate prejudice), then the very same evidence about something else would entitle you to believe that it had a mind.

But if you decide to be a behaviorist, you have problems too. You and I both know, after all, since we both do have minds, what it is like to have a mind. So you and I both know there's a difference between us and a machine that behaves exactly like us but doesn't have any experiences. Unless M has experiences, it hasn't got a mind. The difference between having a mind and operating as if you’ve got one seems as clear as the difference between being con­scious and being unconscious.

The upshot is this: If you look at the question from the outside, comparing M with other people, behaviorism looks tempting. From the point of view of the evidence you have, M and your mother are the same. Looked at from the inside, however, there is all the dif­ference in the world. You know you have a mind because you have conscious experiences, an “inner life.” M may have experiences, for all we know. But if it doesn’t, no amount of faking is going to make it true that it has a mind.

We started with a familiar fact: computers are everywhere and they’re getting smarter. It looks as though there will soon be intelli­gent machines, machines that will understand what we say to them. But when we look a little closer, things are not so simple. On the one hand, there is reason to doubt that behaving like a person with a mind and having a mind are the same thing. On the other, once we start asking what and how we know about the minds of other peo­ple, it seems that our conviction that people have minds is no better based than the belief that there could be understanding computers. We call someone who asks philosophical questions about what and how we know an epistemologist. And if we ask how we know about the minds of other people it seems plain that it is from what they say and do. We simply have no direct way of knowing what—if any­thing—is going on in other people’s minds.

When we look at the question from the inside, as we have seen, the picture looks different. Someone who looks from the inside we can call a phenomenologist. “Phenomenology” is the philosopher’s word for reflecting on the nature of our conscious mental life. From the phenomenologist’s point of view, M, and all machines, however good they are at behaving like people, may well turn out not to have minds.

From thinking about computers in science fiction we have found our way to the center of the maze of problems that philosophers call the philosophy of mind or philosophical psychology.

As I said in the introduction, philosophical perplexity is a little like being lost in an old city. It is time now to find our way up that tower to have a look around. We have already been forced back to two of the most fundamental philosophical questions, “What is it to have a mind?” and “How do we know that other people have minds?” So let us put aside the question about M and take up these more fundamental questions directly. At the end of the chapter I'll get back to M, and we'll see then if our trip up the tower has indeed helped us to find our way about.

1.2