What’s best

Ian Hacking

  • The Nature of Rationality by Robert Nozick
    Princeton, 226 pp, £19.95, August 1993, ISBN 0 691 07424 0

Robert Nozick has a unique place in the annals of rational choice theory: he refuted it. Or so say I in my role as the last of the true Popperians. That was back in 1969. But now the mature philosopher is out to turn the theory into, not exactly a transcendental reality, but something implanted deep in the minds of some, if not all, human beings who have been sculpted by Darwinian evolution. This is an ideological book, concluding with evolutionary premises implying a complacent vision in which something like our present social order arose out of biological facts. The book begins, innocently enough, with technical questions about making reasonable choices. I’ll follow Nozick up that garden path, which is wonderfully landscaped, fresh and fragrant. But I’m giving warning now that I’m afraid of the ogres at the bottom of his garden.

Two principles form the pillars of the abstract analysis of rational choice theory. The first says that if one action has more desirable consequences than any other, no matter what happens now or in the future, then that’s the right thing to do. It is dominant, as they say in the rationality business. That doesn’t merely seem self-evident, it is self-evident, or so most readers will say to themselves. The second principle needs a little cultivation, for it is best put in terms of probabilities. In one version, when you are uncertain what will happen, you should perform the action that would be most rewarding on average, if you were presented with the same choice on many occasions. There are many ways to state this idea without recourse to averages; all use versions of a technical concept called ‘expected value’. This principle of maximising expectation may need qualification to allow, for example, for people who are averse to taking risks: it may seem better to keep what you have, rather than risk losing some of it even though in the long term you should end up better off. It seemed obvious that these two principles are mutually consistent; Nozick, however, showed that they aren’t when, long ago, he published a counter-example, a case in which the two contradict each other. He says that it was invented by a physicist, William Newcomb, but if one thinks of refutation as a public event, than all honour goes to Nozick. He showed exactly how the two principles come into conflict, and briskly swept away all easy solutions. It would take most of a column to state Newcomb’s paradox properly, so I won’t do it here (anyway it makes me slightly nauseous, as good antinomy should). You’ll have to go to the book, where Nozick once again displays his extraordinary skill at combining precision and a light touch.

I dare say there have been hundreds of published stabs at the problem, a good many of which have increased our understanding of rational decision. That is the point of paradoxes: harsh confrontation, painful rethinking, new speculation, trial and lots more error. No one has offered a solution that satisfies more than a handful of immediate cronies. Part of the trouble is that anyone who has any interest in rational choice theory wants both principles, dominance and maximising expectation. Most theorists want to ignore the issues. There may be an interesting group of examples of Newcombian situations (elegantly developed by Alan Gibbard and William Harper in another classic study), but it is not infectious. Most decision problems don’t catch even the germ of an inconsistency. For the logician, however, such monster-barring (to use a catchy phrase of the late Imre Lakatos) is not just a dishonest ‘don’t look now, avert your eyes’; it also prevents us from learning from conflict.

You are not logged in