When It’s Rational to be Irrational

We know, according to Pascal, that the heart has its reasons that reason knows nothing about. Poets have sided with the theologians, agreeing that God speaks to the humble heart, and not to the arrogant intellect. Lately, the intellectuals have been stealing their critics’ clothing. Economists, philosophers, logicians, and theorists of “rational choice”—of whom Jon Elster is one of the most prominent—have begun to tell us that it is more rational to be irrational than rational. Their grounds are hardly traditional ones. It is not the voice of God but the calculations of reason that tell us not to listen to reason. It is not a high-minded alternative to economic reasoning but economic reasoning itself that tells persons who want to “maximize their utilities” that they will succeed only if they do not try.¹

Elster’s Solomonic Judgments, for example, argues that judges in child custody cases who try to do the best for all parties would often do better to toss a coin instead. Elster knows that this sounds paradoxical. For he is not suggesting that the judge who follows his advice will be guided by the hand of God; quite the contrary. The result will indeed be accidental. The point is rather that there are situations in which the conscious attempt to achieve the outcome we want goes so thoroughly wrong that in those situations, it is better to toss a coin instead. The paradox vanishes when we recognize that child custody cases are miserable for everyone concerned, especially for the luckless child at the center of the case, and that the damage worsens as the case goes on.

Often, Elster argues, the damage done by prolonging the legal process will be greater than any damage done by a slightly worse choice between the warring parents. Tossing a coin shortens the process, reduces the damage, and produces as good a result as we can hope for. There is no guarantee that it will produce the best possible outcome—but the whole point of Elster’s argument is that the search for the best possible outcome is self-defeating. If the best possible outcome had been obvious, the judge would have been derelict not to choose it at once. Once it is sufficiently unobvious, tossing a coin cuts everyone’s losses.

Jon Elster has made a considerable reputation out of exploring the paradoxical results of apparently rational behavior. As a multidisciplinary cosmopolitan, he has been equally at home in Oslo, Paris, Oxford, and Chicago, and among logicians, economists, political theorists, and applied social scientists. He is a Norwegian of social-democratic inclinations who pursued his graduate study in Paris with Raymond Aron and subsequently became a professor in the ultra-Marxist stronghold of the University of Vincennes. There his closest colleague was Nicos Poulantzas, Louis Althusser’s best known and most original disciple.

Since then, Elster has divided his time between the political science department at the University of Chicago, the Collège de France, and the University of Oslo, and has poured out a stream of essays on the limits of rational behavior, and the problems these pose for the social scientist and the policy maker. His earliest work was on Leibniz and the origins of capitalism. Logic and Society (1978) and Ulysses and the Sirens (1979) were the first books of his to appear in English, and were immediately recognized as something out of the ordinary. They were followed by Sour Grapes (1983), and then by the comprehensive demolition of the best-loved parts of Marx’s oeuvre in Making Sense of Marx (1985). It might more accurately have been titled “Making Mincemeat of Marx.”² Somehow Elster has found time to serve on the commission into the Swedish trade union movement that provided him with the empirical evidence on which The Cement of Society is based, as well as to edit several volumes of essays, and to lecture, apparently simultaneously, to audiences half a globe apart.

Elster’s work is very highly regarded among philosophers and social scientists, but it is not easy to explain its importance. Unlike many social and political theorists who are briefly fashionable and are later seen to be all hot air and bluster, Elster is not given to large, programmatic utterances. Both his prose and his intellectual style are essentially atomic, carefully delineated problems are tackled paragraph by paragraph, and his books are compendia of such puzzles and their solutions—or increasingly, as in his recent books, the acknowledgment that there isn’t a solution.

One thing that excites his colleagues’ admiration is the deft and imaginative way he deals with the intellectual puzzles in which economists and logicians delight. What sets him apart from other practitioners in the field, however, is his eye for their larger implications. Sour Grapes, for instance, is about what its title suggests. The fox in Aesop’s fable tried to seize an attractive bunch of grapes, failed, and then announced that he had not really wanted them in the first place because they were sour. Like the fox, many of us decide that what we cannot have we didn’t really want in the first place, and so we reduce the pangs of disappointment. On the other hand, “sour grapes” is no more common than the opposite syndrome: “The grass is greener on the other side of the fence.” All too frequently we want something passionately only when we cannot have it.

These perverse desires—perverse because they are not formed on the basis of the merits of what we desire—have interesting properties. The attitude associated with “sour grapes” may be perverse, but, Elster observes, it can be useful to the extent that it reduces the pain of failing to get what we once wanted. Its opposite is simply self-defeating: what we have we do not want, what we want we cannot have. On the other hand, “sour grapes” may stop us trying to get the grapes when they are within reach, while the view that the “grass is greener” may spur us to effort and (if the trait is general) make for a sort of social progress. Still, the familiarity of such thinking raises an obvious question—does it matter?

It matters for many reasons. One is that it undermines a great deal of economics. Economics is, at its widest, the study of how people can most efficiently satisfy their desires. It is all about instrumental rationality. Economists usually assert and their critics usually complain that “economics takes desires as given,” that is, economists do not moralize about what people want, they only assess how efficiently people set about getting it. What links traditional economics and modern accounts of “rational choice” is the thought that rational behavior is entirely a matter of choosing the most efficient means to a given goal. Such attitudes as “sour grapes” and “greener grass” cast doubt on that starting point. If our desires are perverse, gratifying them may very well not increase our welfare.

Elster never exaggerates the importance of such desires, but it is worth observing how far their implications stretch. Advertising, for instance, is largely devoted to encouraging the feeling that what we don’t have is vastly more desirable than what we do have. You need not be a Marxist to think that there is something wrong with an economic system that relies as heavily as does our own on inducing perverse desires. If you are a Marxist, on the other hand, you will think “sour grapes” is important for a further reason. An inegalitarian society would seem to be vulnerable to the discontent of the poor and powerless; why do they not rise up and demand more food, better clothes, larger houses, and the boss’s job? A famous answer is that they are systematically taught that they wouldn’t enjoy them, or wouldn’t like the extra responsibility, or wouldn’t care for middle-class culture. They are disarmed before they have armed themselves, because their desires are shaped to suit “the system.” The system induces the “sour grapes” syndrome in the disadvantaged. Michel Foucault, as much as Karl Marx, argued that capitalism protects itself in just this way.

Elster is quite clear that this is not the way to explain what he labels “perverse preference formation.” He rejects any suggestion that societies or economic systems “act” in the way individuals do. Making Sense of Marx was deeply hostile to Marx’s “functionalism”—to the way Marx explained economic activity by claiming that it served the needs of capital. Sour Grapes and Nuts and Bolts are equally skeptical of such explanations. Any explanation of perverse preference formation must invoke plausible psychological mechanisms, not social functions.

Elster sets great store by the truism that a society is not something over and above its members. As Nuts and Bolts reiterates, what the social sciences explain is the doings of individuals and the consequences of those doings. Certainly many—perhaps most—of those consequences are unforeseen and unintended: Adam Smith was right to say it was an invisible hand that ensured that when we all compete in the marketplace we inadvertently maximize national output and minimize prices. We intend our actions, but we do not always intend their consequences. The model of social scientific virtue is economics, which explains collective or aggregate effects according to the interaction of individuals.

In using the economic model, however, many writers have been tempted to stand common sense on its head, or, more elegantly, to pursue “counterintuitive” hypotheses wherever they may lead. In The Economic Approach to Human Behavior³ the Chicago economist Gary Becker tried to explain all social life as a series of economic transactions. Becker’s tactics are, one might say, the reverse of Thorstein Veblen’s. In The Theory of the Leisure Class, Veblen claimed that people who thought they were behaving like rational economic men were really driven by the desire for reputation and glory that drove their barbarian forbears. Becker takes an institution like the family, which we think is sustained by love and altruistic emotions, and analyzes it as a series of bargains for sex, subsistence, and security.

Elster doesn’t flout common sense in that way—whatever the arguments for and against such flouting. He doesn’t suggest that the altruists who make sacrifices for their children and wives, or to help people badly off, are “really” out to gratify themselves. He thinks that the view that self-interested behavior is basic is true in only a restricted sense—namely, that altruism could not define itself as such, and would have nothing to work against, if people did not have some self-interested wishes. The attraction of economics is not its connection with self-interest, but its concern with rationality.

What makes rational action special? Many things, but the first is that rational behavior is interesting to analyze because rationality is essentially flexible. It is flexible because rational people calculate the best means to whatever ends they have in view, and those calculations are of great intellectual interest. To be rational, in Elster’s sense, is simply to try to bring about the best results you can. To do this means choosing the most efficient means to that end. Driving through Manhattan, I adjust my route to the prevailing traffic when I need to. A drug company tries to minimize the mileage its sales people travel in the course of visiting all its customers. Working out what that requires is often very complicated. Companies can now use computer programs to solve such puzzles and save a great deal of money. But what fascinates Elster—and other theorists in this tradition—is the paradoxical cases where rationality demands inflexibility.

Think of Ulysses and the sirens—the legend that gave Elster’s second and best known book its title.⁴ Ulysses knew that if his men heard the sirens song they would row onto the rocks and be devoured; he knew that if he heard the sirens’ song and was in command of his ship, he would steer them onto the rocks. But he did most definitely want to know what song the sirens sang. So he blocked the rowers’ ears with wax, and had himself tied to the mast of his ship so that he could not seize the helm. The rowers could hear neither the sirens nor their captain begging to be released. By literally binding himself so that he cannot change his mind, Ulysses hears the sirens and survives.

Everyday economic life is full of less colorful illustrations of the point. Two firms know they would do well to divide up their local market and not compete. The trouble is that a division is unstable: each would do even better by taking advantage of the other firm’s forbearance in order to steal its customers, while relying on the other firm not to steal its own. The result is predictable; they will rapidly return to cutthroat competition—worse for both than a successful division of the market, but not unstable in the same way as an unenforceable agreement to restrain themselves.

Like Ulysses, they’d do better to be tied to their agreement. The same is true of agreements not to compete in the arms race, agreements between major powers to respect each other’s spheres of influence, and wherever doing what we want here and now will stop us from getting what we want in the long run. Practically, such examples show that rational self-interest can be an unreliable basis of agreements; intellectually, they complicate the question of what rationality is. Do we need to be irrationally committed to keeping our agreements if we are to put an end to the arms race; or is it more rational to side with our long-term interest in survival than with our short-term interest in gaining an advantage in the arms race? The brief answer is that there is no answer.

It is not only the conflict between present and future that subverts rationality. If we are unsure what we want, we may not be able to be rational. If we are unsure how things will turn out, we may not be able to be rational. In Nuts and Bolts Elster admits that much of our life must in that sense be irrational. Worse yet, many of the things we are sure we want cannot be pursued by directly aiming at them. A tennis player may want the glory of winning Wimbledon; but thinking of the glory of winning is the thing most likely to wreck his game. We may long to experience a grand passion, but to fall madly in love with the girl of our dreams we must think of the girl, not of being in love.

Insomnia, impotence and stuttering get worse if one tries to do something about them. They are more likely to go away if one ceases to think about them—but that is not something that can be the outcome of an action.

What he calls “essentially indirect goods”—desirable states of affairs which can only be had as the byproducts of what we can aim at directly—occupied his attention in Sour Grapes, and get more in Nuts and Bolts.

The reason why these difficulties with rationality matter is that we cannot do without rational choice theories. Theories of rational choice are, in Elster’s view, and probably in the view of most social scientists, the best theories we have. They underpin economics and much of contemporary sociology; and they seem to be solidly based in common sense. It is impossible to imagine ourselves doing without them. Why do we lure professors from other universities with higher salaries? They want to maximize their pay or minimize their efforts and will take steps to do either or both; so it makes sense to offer them more dollars, or fewer classes. Nor is rational choice applicable only to selfish choices. If I am driving a van delivering “Meals on Wheels” to the housebound elderly, I ought rationally to minimize the gas I use, the time I take, and the distance I drive, so that I can do more good rather than less.

Situations involving many different people yield peculiar results of course. The situation of the patrons in a blazing movie theater is a familiar example, in which the result of each person doing what is best for himself or herself is likely to be the deaths of almost everyone. If each of us rushes directly to the nearest exit, a few will get out; but most of will get trapped at the door and die. But rational choice theory is not committed to the (obviously false) claim that the uncoordinated efforts of each of us to do the best for himself or herself will produce the best possible outcome for all of us. It is only committed to the thought that we best explain human behavior by assuming that individuals are apt to behave rationally in this sense.

In Nuts and Bolts Elster observes that the assumption of rationality is “like logic.” If someone was wholly illogical, we’d have no idea what he really thought, and if someone’s actions were wholly irrational, we’d have no idea what he was doing—what he wanted, why he wanted it, how his behavior was connected to what he wanted. It is a matter of logic that people should (rationally, not morally) do what in the light of their beliefs is most likely to achieve what they wish to achieve. Why they want and believe what they want and believe can’t, of course, be explained or justified as a matter of logic.

The difference between the logic of rational choice and the facts about our desires and beliefs is extremely important. The apparent rigor of economics is entirely a matter of the way it uses what Karl Popper christened “situational logic”; but this rigor is bought at a price. Economics is rigorous to the extent that it is a branch of logic, and invulnerable to empirical evidence. Reattached to empirical reality, economics ceases to be logic and becomes sociology, no more and no less persuasive than much else in sociology. Anyone who doubts this should reflect on how rarely economists agree on the explanation of growth rates, inflation, and almost any other issue of public interest.

In his three recent books Jon Elster tackles several different tasks. Nuts and Bolts is an introductory essay in the philosophy of social science, though one that will mean more to the already fluent than the real beginner. This is not because it is difficult. It is limpidly clear and easy to follow; but it is full of hints and allusions—quite often to Elster’s earlier work—that will mean more to readers with some economics and philosophy already under their belts. Elster thinks that the social sciences possess no laws, few substantial generalizations, and no real theories; what they do have is a box of “mechanisms” that they can employ to explain why people act as they do. What he means by mechanisms is not elaborate. It embraces all the habits of behavior we invoke when we explain why people act as they do. We might appeal to “sour grapes,” for instance, or to “the grass is greener on the other side” to explain why someone did or did not mind that he couldn’t get what he wanted. What we cannot usually do is know ahead of time which mechanism will affect a person’s behavior. The social sciences are essentially retrospective, not predictive.

We know that either mechanism might come into play; but not which one will be used, or when. We do not know whether a friend who is refused promotion will think the job was not worth having or whether he will be inconsolable. Social science mostly consists of appealing to one of a range of mechanisms after the event. That the social sciences can be persuasive after the event but are hopelessly weak on prediction, as Elster argues, is surely undeniable. Witness the total failure of any political scientist to foresee the upheavals of 1989 in Eastern Europe, and the equally complete, though less remarked, failure of economists to predict the behavior of the economy any more accurately than the man in the street.

If Nuts and Bolts is a compendium of mechanisms and puzzles, Solomonic Judgments presents a larger view of the human condition. It starts by considering the usual premise of such discussions: rational behavior demands the pursuit of coherent goals by the most effective means available. Elster quickly turns to the difficulties of this view. Consider the mushroom gatherer. She would be wise to seek out a good place to gather mushrooms, but would be foolish to take so much time comparing different places that night fell and mushroom gathering became impossible. So how much information is it worth collecting? We could give a rational answer to that question if we knew how likely it was that the crucial information would turn up in a given amount of time—but we don’t. We don’t know how rational it is rational to be.

While we gather information, the world changes. We evaluate the investment proposition we have been offered only to find we have missed the bus. Might we do better to toss a coin? Solomonic Judgments takes seriously the thought that we might. Before considering the judgment of Solomon, Elster takes us on an inquiry into the history of lotteries and other random procedures. In spite of the example of the Athenians and the Florentines, he takes less seriously than I would the thought that we might choose members of Congress by lot. Aristotle famously observed that the lot is the democratic method of selection; it allows everyone an exactly equal chance of success, which in a supposedly democratic country ought to count in its favor. The Athenians selected their rulers by lot because they feared the corrupting effects of power. We have every reason to share their fears, and as a device for ensuring turnover and eliminating the advantages of incumbents, the lottery can hardly be beaten.

I have already mentioned Elster’s argument that child custody cases should often be settled by chance—whence his title. He begins from the fact that in custody cases judges look for the right answer; they try to do the best for everyone, especially the child, but if possible the contending parents, too. Elster thinks it is a feature of our irrational makeup that we try so hard to get things right in cases where the one thing we can be sure of is that trying to get it right will make things worse.

The longer the case lasts, the greater the temptation to leave the child with whoever has temporary custody, simply because the longer the case goes on, the worse the effect of further upheaval. This is inescapable but unjust. When, as in some cases,⁵ the judge is acting in a custody contest between an estranged lesbian couple, and has no legal precedents to go on either, the difficulty of seeking the “right solution” is patent. When no solution is manifestly better than any other, we can minimize the costs of getting into the courts in the first place by tossing a coin. Moreover, such a procedure would, in the long run, be a powerful deterrent to fights over custody; people who knew they had a fifty-fifty chance of losing would have a strong incentive to look for a different solution. The fact that Elster’s argument is both convincing and unnerving is testimony to the insight that lies behind it—we are indeed irrationally convinced that there ought to be a rational solution to every problem.

The impetus to write The Cement of Society came from Elster’s encounter with Swedish collective bargaining. He began by expecting to see rational self-interested negotiations from union and management providing an illustration of the power of rational choice theory. But it emerged that “the complexity of these bargaining problems defied explicit modelling…. In the Swedish system of collective bargaining…everything is up for grabs: the identity of the actors, the rules of the game, the set of payoffs, the range of acceptable arguments.” To do justice to this untidy reality, Elster admits, we need more than the theories we can borrow from the economists. Economics assumes that the identity of the actors and the rules of the game are fixed; if everything is up for grabs, it cannot find a basis from which to analyze economic behavior.

After fighting his way through familiar puzzles about when rationally selfish people would cooperate to achieve some common good—less often than they actually do—Elster throws in the towel. We cannot explain everything by referring to homo economicus, the flexible, rational, self-interested hero of economic theory. We have to recognize the existence of homo sociologicus, the inflexible follower of social norms who has been the hero of sociological theory ever since Emile Durkheim.

The trouble with flexible rational man is that he often does not know what to do because he cannot guess what other rational men will do. Once I have a good idea of the limits of your willingness to sell your goods, I can begin to bargain; but you cannot set those limits until you know the limits of my willingness to buy them. We often cannot predict each other’s reactions until one of us makes a move, but neither of us knows which move to make until we can predict the other’s reactions. No wonder we prefer fixed prices for the goods on the supermarket shelves to having to haggle at the checkout counter. No wonder we rely on habit rather than calculation for most of our decisions.

Where rationality produces unpredictability, obedience to norms produces predictability. “Thou shalt not kill” yields more predictable results than “Thou shalt kill only when the payoff to those whose well-being you mind about outweighs the foreseeable costs to them of such effects as revenge killings and lessened security.” Elster observes that in the negotiations he observed between unions and management what mattered was norms such as maintaining wages above a certain point or insisting that “gains from luck,” e.g., from improved market conditions, should be “shared between capital and labor.” Rational analysis of the facts might have proved that these norms were not helping one side or another get as much as it could have; but adhering to such standards kept the parties to the negotiation united among themselves and the results of the negotiations reflected the standards each side insisted on. The results produced when people are impelled by such norms tend to be messy and inconclusive, a point that is hardly surprising when Elster identifies over a hundred different norms involved in the bargaining process. The tidiness of theorizing about rational men has gone out of the window, and what is left is full of the buzzing confusion of what such theories futilely tried to describe.

It is somewhat disappointing that Elster seems uninterested in the question of how we switch so readily between rational calculation and adherence to norms and then back to calculation. He sees that many of us say to ourselves, “I will follow the rules of fair play so long as everyone else does, and if not, I will do whatever seems best.” But he is uninterested in explaining or justifying these mixed motivations. Nor does he say anything about why we recognize the norms we do, whether the rules of fair play or prohibitions such as “thou shalt not kill.” His attachment to rational choice theories is reflected in the way he thinks of norms. Their interest is that they demand inflexible, nonrational behavior. He rejects any explanation of their origin as “functional” for social order, and seems uninterested in pursuing the question whether an evolutionary process of the kind described in Richard Dawkins’s The Selfish Gene could combine the attractions of rational choice theory with acknowledgment of our nature as norm-following creatures. This is surprising, since Elster acknowledges that we often do better by following norms than by calculation; and the obvious thought is that a “rational gene” would, during the course of evolution, rationally choose to be embedded in an irrational keeper of promises and pursuer of justice rather than in homo economicus.

The lucidity, energy, and intellectual high spirits of Elster’s work are so enjoyable that I hesitate to end on a downbeat note. But the three books under review leave it quite unclear where Elster thinks the social sciences ought to be headed. It appears that the sophisticated work done by the mathematically proficient rational choice theorists, with whom even Elster says he can’t keep up, is of interest for its own sake, but doesn’t advance our understanding of why people act as they do. Elster glumly describes himself as suffering “disillusionment with the power of reason, be it at the level of social actors, or at the level of the social scientist who is observing them.”

He ends The Cement of Society with a commonsensical list of factors that assist (and sometimes threaten) social cooperation and stability—“altruism, envy, social norms, and self-interest all contribute, in complex, interacting ways to order, stability and cooperation.” To know how they interact, we must analyze particular cases, just as he has analyzed the parental custody cases. Well might he observe that if the social sciences are sciences, they are more like chemistry than physics. When Macaulay savagely attacked James Mill’s Essay on Government a hundred and sixty years ago, he complained that the simple picture of rational man that James Mill had relied on just wouldn’t do. Patient, cautious description, and a careful attention to the varied historical experience and the multiple psychological motivations of mankind were all we could rely on. Jon Elster has traced a path from Mill to Macaulay.