A new take from Nozick on Newcomb's Problem and Prisoners' Dilemma

Abstract

In a salient part of his recent book The Nature of Rationality [4], Robert Nozick offers a new argument about Newcomb's Problem and Prisoners' Dilemma. This new argument is salient in its own right, but also in part because Nozick himself was responsible for the first article on Newcomb's Problem, in 1969 ([31), and this is his first return to the problem since then. It's illuminating to compare some of his earlier remarks with his present view. In Newcomb's Problem, I'm offered the choice of taking either just the opaque box in front of me, or both it and another transparent box, in which I can see $1000. I am told, and believe, that a predictor of human behaviour has already put either nothing or $1,000,000 into the opaque box. He has done so on the basis of whether he predicts that I will take both boxes, or only one: he has put $1,000,000 into the opaque box if and only if he has predicted I will take just it, and not the other transparent box as well. Moreover, I know that in 99% (or some other very high percentage) of the many other cases of choices people have made in the same situation, the predictor has predicted correctly. Newcomb's Problem has been of great interest to decision theorists and more generally to those interested in rationality because it seems to pit two principles against one another and force us to choose between them. On the causal view, you may as well take both boxes, whatever the predictor has done. If he's put a million in, you may as well get a thousand as well, and if he hasn't, you may as well get at least a thousand. Either way, you are better off taking both. By refraining from taking the transparent box, you do nothing to bring it about that you get the million, since the predictor has already acted. But the evidential rebuttal is, in effect: 'Well, if you're so clever, why aren't you rich?' People who take just the one box in this game tend to be rich, while people who take both tend not to be. It would be good news to learn that you are the kind of person who takes just one box, because that would mean the predictor has probably put the million