Evolution in Action


The Nature of Rationality, by Robert Nozick, Princeton, N.J.: Princeton University Press, 226 pages, $19.95

A whimsical genie presents you the following choice. Two boxes are in front of you. The box on the left contains $1,000. The box on the right holds either $1 million or nothing. You may either take both boxes or just the one on the right.

The genie will try to predict your behavior. If he guesses that you will choose to take both boxes, he will put nothing in the right box. But if he predicts that you will choose only the right box, he will stuff it with $1 million. First the genie predicts, then he puts either nothing or $1 million in the right box, then you pick. One more relevant piece of information: This is not your everyday rub-the-lamp-and-make-a-wish genie. This genie is a specialist. He has played the game many, many times with many other people. On every occasion when a player selected both boxes, the right box had been left empty. On every occasion when the player selected only the right box it has contained $1 million. You therefore believe that the genie is a remarkably gifted predictor and will almost surely know in advance what you have decided to do. The choice is yours: Do you select one box or both?

The riddle, known as "Newcomb's Problem" is not merely some child's brain teaser. Rather, it has been a staple of the philosophical literature since 1969, when it was first put on public display by a bright, brash young man named Robert Nozick, the same Nozick who four years later would win a National Book Award and ratchet libertarianism from curio to academic respectability with Anarchy, State, and Utopia.

Newcomb's Problem is redolent with paradox. There seem to be exceedingly strong reasons to choose both boxes, but the reasons for choosing only the right box are also strong. On the one hand, previous experience establishes that one-box selectors invariably end up with much more money than those who choose both. But on the other hand, the genie has already set up the boxes. Nothing you now do can change what has happened in the past. (Don't worry about the genie fiddling with the boxes after you have already made your choice; this is a scrupulously aboveboard genie.) Since you may get more by choosing two boxes and cannot get less, it is irrational to do otherwise than choose both. And yet there is all that past experience…

What Newcomb's Problem reveals is a deep-seated tension within our conception of rationality. The well-known Prisoner's Dilemma similarly puts the concept of rationality under pressure. The latter is an interaction among two people such that each independently deciding player does better for himself by not cooperating with the other than by electing to follow a cooperative strategy. Thus, from the perspective of self-interest the rational course is to decline to cooperate. But the outcome when neither chooses the cooperative course is worse for each than if they had both followed the cooperative strategy. That is, rational individuals collectively do worse than nonrational ones. If this is not a paradox, it is at the very least a challenge to the idea that by being more rather than less rational we thereby improve our prospects for doing well.

The Nature of Rationality is a freewheeling investigation of these and other conundrums surrounding the theory and practice of rationality. For its author it represents a full turning of the circle. Prior even to the publication of the Newcomb's Problem essay, he had written a Princeton University doctoral dissertation titled The Normative Theory of Individual Choice (1963, reprinted in 1990 by Garland Press). That dissertation was an ingenious, engaging, sometimes dazzling analysis of conditions of rational belief and action. So too, in no small measure, is this book. There aren't many places where one can find so many ideas cascading through so few pages.

The standard reviewer's exhortation when confronted with such a bravura performance is "everyone should read this." Almost never is such an exhortation credible. Neither would it be here. In the preface Nozick observes that one unfortunate trend of specialization in intellectual inquiry is that it has become more difficult, sometimes impossible, for the proverbial intelligent layman to be able to read and understand important contemporary results in fields that once were of general humanistic concern. That is true with regard to the study of rationality.

For Socrates and Aristotle the investigation of reason by reason was a necessary activity for the living of a maximally flourishing human life. (Now, I am half-tempted to say, what is required for the good life is examination of REASON by reason.) Contemporary practitioners of rational decision theory rigorously deduce theorems from an axiomatic base. Their pieces should carry a surgeon general's warning: "Can be lethal to those with math anxiety. Stay away!" Topics that everyone should understand have become off-limits to those who don't possess the technical essentials. Alas, such is the case with extended sections of this book. Even many professional philosophers will find them heavy going. Still, it may be of general interest to take a brief overview of Nozick's project.

The Nature of Rationality is primarily an exercise in synthesis. Its pervasive theme is the unsatisfactoriness of monolithic accounts of rationality. Some examples: Nozick combines a Kantian defense of acting on principle with an evolutionary account in which reason emerges in human beings through the same process that gives us the opposable thumb: We judge and deliberate the way we do because those modes of thought and decision tend to pay off in the genetic lottery, not because we have been designed to discern the good and the true. Nonetheless, we do aim for truth and goodness, and Nozick offers an engaging account of why that is so. He also argues that standard decision theory's endorsement of defection as the rational strategy in a Prisoner's Dilemma and the altruist's counsel of cooperation are both mistaken. Rather, sometimes one has reason to snatch the bigger outcome, but sometimes it is rational to accept less though more is available.

And his response to Newcomb's Problem is strikingly eclectic. Precisely because it is such a vexing riddle, Nozick denies that we can have 100-percent confidence in either of the proffered solutions. So we adopt from evidential decision theory a little of the single-box prescription, but also a dollop from causal decision theory's endorsement of the two-box solution.

Both Newcomb's Problem and The Prisoner's Dilemma concern rationality of action. But we invoke reasons not only for acting in the way we do but also for believing this rather than that. Nozick's synthesis at its widest brings together the practical and the theoretical. The first principle of rational acceptance is theoretical: Do not believe any statement less credible than some incompatible alternative. The second principle, though, is practical: Believe a statement only if the expected utility of doing so is greater than that of not believing it.

This formulation conceals one further twist in Nozick's argument. When theorists speak of utility they essentially mean the output of a thoroughly instrumental reason. Nozick, however, insists on bringing into the reckoning the symbolic component of belief and action. We act and believe as we do not only to bring about results but to express our commitments to certain ideals and ways of being: loyalty to friends, respect for the truth even if it hurts, and so on. Accounts that omit these may pride themselves on being hardheaded, but what they really exhibit is tunnel vision.

Indeed, Nozick directs this criticism at himself: "The political philosophy presented in Anarchy, State, and Utopia ignored the importance to us of joint and official serious symbolic statement and expression of our social ties and concern and hence…is inadequate." The language is almost identical to the words employed in his previous book, The Examined Life, where Nozick announced to the world that he has abandoned the libertarianism that had made him both famous and notorious. In that book he left the matter there, not deigning to exhibit precisely what it is about the symbolic component of joint action that undercuts libertarian political philosophy. Here in The Nature of Rationality he is, unfortunately, no more forthcoming.

And that truly is unfortunate, not simply or primarily because we are deprived of juicy biographical tidbits concerning the pilgrim's progress of one R. Nozick, but because it is by no means obvious how or whether recognition of the symbolic/expressive component of action directs one away from a classically liberal conception of politics. I should now confess a bias in this regard. Like Nozick, I believe that political theory needs to acknowledge the crucial role played by symbolic activity. But unlike Nozick, I have concluded that such recognition supports rather than undermines the imposition of tight constraints on the scope of collective choice. That is, an expressive theory of political behavior represents a positive rather than negative for libertarian political philosophy. (Interested readers may care to look at "The Booth and Consequences," in the November 1992 issue of REASON, or Democracy and Decision, published last year by Cambridge University Press.) In this I may be mistaken, but the matter at least requires argument. On almost every other aspect of the nature and status of rationality Nozick offer arguments aplenty; only here are we treated to gnome-like silence.

In a free society individuals are, of course, entitled to decline to dredge up from their past episodes that they now for personal reasons find distasteful to recall. But here Nozick's reticence transcends the personal realm and displays disrespect for his readers. A philosopher who announces ex cathedra the unsatisfactoriness of a viewpoint against which he refuses to provide evidence or argument is guilty of professional malpractice. Any philosopher should know this. All the more so one who, for the most part, writes so insightfully concerning what we have reason to believe and to do.

Contributing Editor Loren E. Lomasky is a professor of philosophy at Bowling Green State University.