An argument I don't understand

As Lichtenberg says: "there are very many people who read simply to prevent themselves from thinking."  (Others realize the same end by looking up quotations for the original wording: Es gibt wirklich sehr viele Menschen, die bloß lesen, damit sie nicht denken dürfen; G82.)  I myself just can't get enough of Millgram.  Here's a short paper of his (jstor link) containing the following argument:

Suppose that an ambitious and subject-irrelevant theory of belief T is true, and suppose, for purposes of illustration, that T identifies particular beliefs with specified brain states. We can imagine that T is embodied in a belief oracle: an appliance that sits on my kitchen table and answers questions about my beliefs. One morning, I am sitting in my kitchen, groggy and hung over, and my niece comes up to me and asks me whether there is life on Mars. I have an opinion about this, but it would be too much of an effort just now to collect my wits and figure out what it is. Fortunately, I have prepared myself for occasions like this. I turn to the belief oracle and ask whether I believe there is life on Mars. It scans my brain and tells me that I do. I now turn back to my niece and tell her that there is, indeed, life on Mars. (This is licensed by the coassertability constraint.) My niece goes back to her game of space invaders, and I am struck by the thought that warrant is conserved when I move from one member of a Moore's pair to the other: I have asserted that there is life on Mars on the basis of facts about my brain-facts that (I agree) are irrelevant to the question of whether there is life on Mars. (One would not appeal to the brain states of a Weekly World News reader to determine whether there is life on Mars, and nothing in the story has ruled out my being a Weekly World News reader.)

It cannot be legitimate to make assertions on the basis of facts that are acknowledged to be irrelevant.

What I don't understand is this: given that he's groggy, hung over, etc, why should we assume that he does have the relevant brain state?  The fact that, if he sobered up, etc, he would believe certain things about Mars (or, since I'm skeptical about nonoccurrent beliefs generally, would at least be able to frame an answer to the question about Mars once it's put to him), doesn't seem to be much warrant for the assumption that, while he's still drunk, he is also in this brain state.  Why shouldn't the machine scan his brain and say "As best as can be determined you don't have an opinion"?  This seems like such an obvious response that I am forced to believe that I have badly misunderstood something.

Comments

[dave]

Might this not be merely a disagreement about nonoccurrent beliefs? A disagreement that is WEARING NEUROLOGY'S FACE?

[ben wolfson]

Maybe, but—I disagree with most people about them, too.

[dave]

You do, or you would?

[ben wolfson]

I will disagree with any man or woman! This is my promise!