M again

lang=EN-US style='font-size:10.0pt;line-height: 112%'>M was a machine that would behave in every situation exactly like your mother. A machine that is made to have internal states that function like a human mind we can call functionally equivalent to a person.

How do I know whether M knows, as your mother does, what it feels like to see red? Your mother, I believe, does know, because she, like me, is a human being. I have reason to think that human beings with normal vision know what seeing red feels like. For I know what it is like, and I believe that other human beings are like me.

The functionalist replies:

All the evidence you have that your mother knows what it is like to see red is from what she says and does. Since M does the same, it is unreasonable to believe that your mother has a mind and M does not.

Notice, first, that we cannot appeal to any evidence to settle the dis­pute. Even if we were discussing an actual machine instead of a hypothetical one, it wouldn't help, for example, to ask it if it knew what it felt like to see red. For any machine functionally equivalent to your mother would say “Yes” if you asked it if it knew what it felt like to see red, because that is what your mother would say. If you didn't believe that what the machine said was true, you might try to test it, just as you might try to test your mother, if you suspected that she was colorblind. But whatever she would do in the test the machine would do also. So no amount of such testing is going to give you a reason to say something about the machine that you wouldn't say about your mother.

This is already a rather strange situation, since we normally think we can tell whether people know what it feels like, for example, to see red by testing their responses to red things. Nevertheless, despite the fact that no amount of evidence could settle the issue, the conviction that there is a real doubt about whether such a machine would have a mind is very widespread, including among philosophers. In the next chapter I shall be looking at arguments for the view that if no amount of evidence could decide an issue, there is no real issue. Someone who believes this is called a Verification- ist. And if verificationism is correct, then the phenomenologist must be wrong.

But even if the phenomenologist is right in thinking that some states, such as seeing that something is red, can be had only by someone with an inner life, there are other mental states for which this does not seem to be true.

Take beliefs once more. We do not normally talk of “knowing what it feels like to have a belief.” Indeed, we can have beliefs— unconscious ones—that we are unaware of altogether, and even our conscious beliefs do not have a special “feel” to them. What does it feel like to believe consciously that the president is in Washington, or that the rain in Spain stays mainly on the plain?

If this is so, then, even if the phenomenologist was right to be suspicious about the claim that M knows what it feels like to see red, that would not give you a reason to doubt that it had beliefs. And, as the functionalist will insist, you would have all the same reasons for thinking that M did have beliefs as you have for thinking that your mother has them. But beliefs are a pretty important feature of peo­ple's minds, and if having beliefs is enough to have a mind, then, as I said, we might end up holding that machines could have minds, even if they don't yet.

1.11