Rational Choice Theory

Updated About encyclopedia.com content Print Article Share Article
views updated


Rational Choice theory is typically seen as the use of economic reasoning in contexts that were traditionally the concern of disciplines other than economics, especially of political science, sociology, and anthropology. If we take a more nuanced historical view, however, we might as soon see mainstream economics as the stepchild of the kind of reasoning about larger social institutions, norms, behaviors, and so forth that was central to the Scottish Enlightenment in the works of David Hume, Adam Smith, and many others. The genius of these thinkers was to make sense of such institutions, norms, and so forth as the products of individuals acting from their own private incentives. Their concern was that of James Coleman (1990), to explain macrophenomena from microchoices. Through most of the past two centuries, economists increasingly focused such reasoning on explaining the nature and working of the market, for example on prices and conditions for an equilibrium of supply and demand, and the earlier concern with broader sociological issues faded. The great classical economists, such as Smith, Alfred Marshall, and Vilfredo Pareto, were interested not only in the market but also in broader social institutions and practices. Much of their work can readily be counted among the great contributions to sociology in their eras.

Contemporary rational choice theory represents a resurgence of such earlier efforts. The efflorescence of such theory in our time has followed on the development of game theory by John von Neumann and Oskar Morgenstern, Kenneth Arrow's demonstration that individual preferences do not aggregate into analogous collective preferences, and Anthony Downs's analysis of democratic participation. The largest bodies of contemporary work are those on the study of group behavior, under the rubric of Mancur Olson's logic of collective action, and on political participation, which in part can be seen as merely a special case of collective action. Perhaps the fastest-growing area of inquiry today is in the analysis of institutions, much of it focused on historical institutions, as in the work of Douglass North.

A cognate area is social exchange theory, which had its contemporary origins well before the works of von Neumann, Arrow, and Downs. Its origins were also distinctively sociological, as in the work of George Homans, and anthropological.

A seemingly cognate area is economic sociology, although much of economic sociology is like the bulk of voting studies in that it takes the form of simple correlations of behaviors with personal characteristics—gender, age, ethnicity, education, occupation, nationality, religion, and so on. Much of economic sociology might therefore be counted as behavioralist. Rational choice analysis at least implicitly assumes intentionality. Hence, in effect, rational choice theory is contrary to the behavioralism of much of sociology, especially political sociology, in the mid-twentieth century. The point of behavioralism is to avoid the use of mental or intentional explanations of behavior, to treat the mind as virtually a black box. Rational choice theory imputes preferences and intentions to actors. In part, of course, these might derive from or be explained by such sociological characteristics as race, gender, or religion. Although there is a large field of behavioral economics, economics in general was not centrally influenced by the behavioral movement just because it is largely about preferences and intentions. Indeed, much of the work on behavioral economics is directed primarily at establishing the nature and content of preferences and preference functions. It gives measures of the preferences that might go into intentionalist accounts of behavior.

Although much of rational choice theory explains behavior as a response to interests, such theory need not be so narrowly restricted. In its more general form, it explains behavior as the product of preferences, which can cover virtually anything from values to interests. For example, holding all else constant, you might prefer a higher to a lower income. But you might prefer a lower income with peace to a higher income in a state of war, even if your safety and livelihood are not at risk in the war. In many contexts, however, interests seem to be adequate to explain behavior, often because other values are not at stake in the behavior to be explained or because they are substantially less important than interests. Rational choice theory is commonly most compelling in contexts in which interests are predominant. In part, this is because interests can often be more systematically imputed to relevant actors than can other values, although this is not always true. For example, in a given population, particular religious values might regularly trump concern with interests to some substantial degree in some aspects of life. Even when we might suppose other values are very important, however, we might also suppose that analyzing the force of interests gives us a clear baseline for then coming to understand the import or weight of these other values in explaining behavior.

The value theory of rational choice theory is essentially the utility theory that has been developed over the past few centuries by economists and others. It is sometimes asserted that this is an empty value theory and that we can put almost anything we wish into utility functions. For example, I can put your pleasures or various normative commitments in my own utility function. While this is technically correct, most of the major results of rational choice analysis turn on the use of utility functions that are about as spare as we can imagine. They include nothing more than interests, which are conceived to be resources, such as money and time. The results surveyed here virtually all depend only on such simple utility functions. Or occasionally, in a somewhat fuller version, they require inclusion of some of the pleasure one gets from various consumptions, as in the account below of the norm of conformity to neighborhood tastes and manners.

Rational choice theory has been applied to so many diverse issues that a full survey of its characteristic results would be exhausting. I will therefore take up several applications, some of them especially important both in establishing rational choice theory and in recasting the nature of major problems that had already long been the focus of much research. I will first discuss the main methodological or fundamental theories behind rational choice theory. In applying such theory, I will begin with the grandest of sociological issues: the problem of social order. Then I will take up two major areas of research that got the contemporary field of rational choice theory under way: the study of group action and the corollary study of political participation. The first of these is historically a major focus of sociological research, while the second has naturally been the special domain of political science. Then I will take up three efforts that show the breadth of the approach. These are the analyses of institutions, norms, and functional explanation.


The methods of rational choice theory are essentially the methods of economics, including standard equilibrium analysis, price theory, econometrics, and game theory. Game theory is less a theory

Table 1
game theory
game 1. pure conflict
I 1,22,1
II 2,11,2
game 2. pure coordination
I 1,12,2
II 3,31,1
game 3. prisoner's dilemma or exchange
Cooperate 2,24,1
Defect 1,43,3

than a format for presenting the array of choices and outcomes that face two or more actors whose outcomes are determined by the joint choices or actions of all of them. Games can be represented in many forms. In the matrix form, each player has a set of choices or strategies, and outcomes are determined by the intersection of the strategy choices of all players. Games in which two players have two strategies each are called two-by-two games. Such games can be pure coordination, pure conflict, or a mixture of these two, as represented in Games 1–3, in Table 1. In each of these games, Row has two strategy choices available and Column also has two choices. When both have chosen their strategies, an outcome is determined. The payoffs in each outcome are given in ordinal terms. 1 is the best payoff, 2 next best, and so forth. The first payoff in each outcome is to Row and the second payoff is to Column.

In the pure conflict game, one player can be better off only if the other is worse off. In the pure coordination game, both players achieve their best payoff together. Mixed games are commonly called mixed-motive games. There are many different types. The one represented here is the prisoner's dilemma, which is surely the most studied of all simple games, probably because it represents ordinary exchange and is therefore ubiquitous and central in social interaction. In the prisoner's dilemma both players can be made better off together in the move from (3,3) to (2,2), so that the game has a strong element of coordination. But Row is made better off while Column is made worse off in the move to (1,4) from any other outcome, so that the game also has a strong element of conflict.

Unfortunately, the (3,3) outcome of a prisoner's dilemma is an equilibrium in the sense that we cannot move to the Pareto superior (2,2) outcome through individually beneficial or neutral moves, because my change of strategy from defecting to cooperating while you continue with your strategy of defecting makes me worse off. To move to the (2,2) outcome requires joint action. The prisoner's dilemma is the only one of the seventy-eight ordinally distinct two-by-two games that has a Pareto-dominated equilibrium. Its solution therefore commonly requires incentives from outside a single play of the game. The incentive can be from the benefits of cooperative play in an iterated series of plays of the game or from external enforcement by other parties, as under a legally binding contract.


There are three grand schools on social order. One of these is the conflict school associated with Plato's Thrasymachus, Karl Marx's class theory, Ralf Dahrendorf, and many others. Another is the shared-values school of John Locke, Émile Durkheim, and Talcott Parsons. A third is the school of exchange theory associated with Adam Smith, George Homans, and others. These can be characterized by the three classes of games represented above. Because there can be all these—and many other—classes of interactions in society, all these theories are partially right about social order. There is a fourth plausible account of much of the order we see, an account that fits the coordination game. We do not coordinate only because we share values; we can coordinate merely to stay out of each other's way while we pursue our different values. (Indeed, in some sense we can share values—we both want the same thing—in such a way as to have severe conflict.)

One of the simplest of social coordinations is the coordination on driving that makes traffic flow much faster and with less mishap. In North America, we all drive to the right. That is merely one of two possible conventions that could work equally well. The other convention is that we all drive to the left; this convention is followed in the United Kingdom and many nations of the British Commonwealth. It is merely coordination that resolves the traffic problem. But when you and I coordinate on this convention, we do so not because we share any substantial values. We might each be utterly self-serving. The only value we share is to keep others out of our way as much as possible. This is characteristic of much of social order in contemporary liberal societies, in which rampant individualism and great diversity of values might seem to lead to great and disruptive conflict. Instead, it commonly gets channeled in ways that avoid conflict but that could hardly be called cooperative in the sense in which you and I might cooperate in building a house.

Coordination without confluence of values makes the problem of social order seem relatively simple, as it must be for many activities in which we spontaneously achieve order without either the imposition of power by authorities, as is required for the conflict school, or the relatively deliberate cooperation of the exchange school. While such coordination is not the original discovery of rational choice theory or game theory, it is made far more perspicuous by these because these give it a structure of motivation that makes sense of it in many contexts, just as they make the other major schools of social order clearer. Hume presented a clear account of coordination and convention in many social contexts in which stabilizing expectations is fundamentally important to social order. Game theory provides a framework for characterizing the interactions that we must govern if we are to achieve order. Most forcefully, perhaps, game theory suggests why we cannot ground an account of social order in merely one of the traditional schools, because it displays the greater complexity of the forms of interaction that must be ordered.


There is a tradition, still alive today, in which it is supposed that if the members of a group share an interest in some result, they will act to provide themselves that result. For example, Karl Marx's theory of revolution requires this simplistic assumption coupled with his account of class consciousness. Against this tradition, Olson (1965) argued that, if our group interest requires for its achievement that each (or at least many) of us make a personally costly contribution to its achievement, then we commonly do not have individual incentive to act for the collective good. Each of us bears the full cost of our own contribution but receives only a minuscule part of the small piece of the larger collective good provided by that contribution. Typically, therefore, the collective benefit to me of my contribution to our collective provision will be less than the cost to me of that contribution. Hence, if I am narrowly self-interested, I will not make a contribution to the collective good but will hope merely to free ride on the provision that results from the contributions of others. If all other members of my group have my structure of interests, none of us will contribute and our collective good will not be provided. If we could vote to have ourselves compelled to contribute, we might all vote to do so. But if we must voluntarily contribute, none of us might do so. This is the logic of collective action.

Olson used standard micro-economic analysis to demonstrate this logic. He modeled the problem as an instance of Paul Samuelson's theory of public goods, in which the efficient price of access to the public good would be zero even though, at that price, there would be no voluntary supply of the good. Olson's logic can as well be demonstrated game theoretically as an instance of a large-number prisoner's dilemma (Hardin 1982). As in the discussion of game theory above, individually motivated action would not lead us out of the dismal equilibrium of no cooperation, even though the Pareto superior outcome in which all would contribute might be enormously superior to the status quo equilibrium in which all defect.

A very large literature has been directed at explaining the collective action and the achievement of collective benefits that we see despite this logic. For example, we apparently see a great deal of collective action in the form of social movements that sometimes entail great individual costs and even severe individual risks. Much of this literature supposes that people are motivated by commitments beyond self-interest, such as social and moral commitments, but much of it supposes that there are incentives apart from the direct logic of collective action. For example, there may be specific personal benefits corollary to the collective benefit. Alternatively, our group might be provided its collective good but not through spontaneous individual actions. Rather, an entrepreneur might see the possibility of making a career out of leadership in providing our group its collective benefit. Such an entrepreneur might especially arise if our group's good could be provided by government without requiring our voluntary cooperation.


There are two main lines of theory on political participation in a democratic system. One of these began with Kenneth Arrow's ([1951] 1963) impossibility theorem. That theorem essentially says that collective preferences cannot be modeled simply on individual preferences. We might each individually have well-ordered preferences over all the choices we face, and yet we might collectively not have such well-ordered preferences. Indeed, we can generally expect not to have such collective preferences over any complex realm of choice, such as we often face in normal politics over the large array of policies at issue. This has far broader implications than merely that democratic choice may have problems. It is an important and broadly interesting instance of the fact that the imputation of various characteristics of individuals, such as their pattern of preferences, to groups composed of individuals is a fallacy of composition.

The typical implication of Arrow's theorem goes back to the Marquis de Condorcet and to Lewis Carroll. It is that our collective preferences may cycle through some set of possibilities. For example, in majority votes we may collectively prefer A to B, B to C, and C to A. If our collective preferences were as well behaved as our individual preferences, the fact that we prefer A to B and B to C would entail that we prefer A to C. In majority votes, the majority who prefer A to B can be different from that who prefers C to A. For example, my preferences may be A > B > C; yours may be B > C > A; and a third person's may be C > A > B. These preferences yield the cycle above if the three of us vote by manjority. Many institutional devices, such as legislative practices of opposing new laws against each other before opposing the winner of such a series against the status quo, tend to block any evidence of cyclic preferences; but such a device gives a strong conservative bias to collective choice.

The other main line of rational choice theory of democracy began with Joseph Schumpeter and was developed extensively by Downs (1957). There are two major classes of claims. First, Downs supposed that two parties or candidates in an election face an electorate that is divided along a left-right dimension. If the voters have a normal distribution about some central tendency on this dimension, the two candidates will want to place themselves at the peak of that normal distribution. Hence, the two candidates will tend to have quite similar positions. Second, he supposed each voter faces what is de facto a logic of collective action on whether to vote. Suppose there are some costs involved in casting a vote—waiting on line, traveling in foul weather, and paying a fee for registration. It follows that individual voters should see the election of candidate A over candidate B as essentially a collective good to be provided at individual cost to themselves. It is therefore subject to the perverse logic of collective action, and we should expect that many voters would not vote unless they have motivations that go beyond their own interests. Furthermore, if it is not in my interest even to vote, it is unlikely to be in my interest to learn enough about the issues to vote intelligently.


The rational choice analysis of institutions has roots in ancient accounts of the rise of civilization. Among many such accounts in the era of the Scottish Enlightenment is Smith's theory (1978) of the stages of development of society from very primitive, to pastoral, to more nearly modern society in his Lectures on Jurisprudence (these lectures were not published in Smith's time and have played little role in the development of such analyses since then). Smith's account turns very clearly on the incentives that individuals have to submit to various forms of social order and, eventually, government. His later analysis of the wealth of nations is itself a theory of one of the grandest of all institutions: the economy of a modern commercial society. Smith argued that the wealth of the nation is a function of the efforts of individuals to do well for themselves.

More recent work has gone in diverse directions. The two main directions are the microanalysis of institutions and the behavior of individuals within them and the more nearly macroanalysis of why certain institutional forms arise and prevail. The microanalysis is applied to a wide variety of institutions, most of them of relatively small scale, such as committee structures and formal organizations of many kinds. The macroanalysis is often broadly historical and is directed at accounting for the rise of economic and other institutions. For example, there is extensive work on the rise of devices for handling trade across a broad array of cultures in the absence of any centralized governmental authorities.

Much of the institutional analysis builds on accounts of transaction costs. According to the socalled Coase theorem, due to Ronald Coase, if there were no transaction costs, property rights assignments would have no effect on overall production. Introducing transaction costs can distort production substantially. Firms sometimes internalize functions for which transaction costs would be high if they had to deal with outside suppliers for those functions, and they externalize functions for which markets work well to reduce transaction costs so that competitive suppliers can drive down production costs. Attention to the structure of transaction costs therefore can explain much of an economic organization's structure. Attention to changes in markets over time can also explain the evolution of such organizations' structures. While most transaction cost analysis has so far been done by economists, including economic historians, it is also increasingly done by sociologists and others.

North (1990, p. 131) argues that the use of standard neoclassical economic methodologies exacts a heavy price in our effort to understand institutions. Because such understanding must of necessity be developmental, it must include stories of how the institutions came to be what they are. Neoclassical price theory is concerned with allocations at a specific moment in time under particular institutional arrangements. Game theory lends itself more readily to developmental stories, but to some extent we lack the methodology for putting such stories into order.

A clear implication of the transaction cost analysis of institutions is that, once in place, institutions influence incentives and interests, so that one cannot simply take institutions as dependent on rational choice. They are, additionally, shapers of rational choice. This is conspicuously true, of course, for such institutions as those of government and law, part of whose function is specifically to give people incentive to behave in some ways rather than others. It is true far more generally of essentially all institutions that have significant value to us, even institutions whose purpose might be seen as merely to produce certain goods or services. This means that the actual set of institutions we have and the set of individual behaviors we see are partly determined by the order and the era in which institutions have developed. Some part of what is commonly referred to as culture is merely the happenstance dependency of such historical developmental patterns.


Much of the study of norms has been psychological or even psychoanalytic, and perhaps most of it has assumed that the motive for following norms is essentially normative or otherwise not self-interested. Efforts to explain the rise of norms, however, are often forced to take account of how the interests of at least some people are to act on and to enforce various norms. One way to characterize the problem of creating and maintaining a norm is as a problem of collective action. It would seemingly be in the interest of almost all of us if a certain norm prevailed, but it is in the interest of almost none of us actually to abide by the norm unless there is some sanctioning system to keep us in line. In some theories, norms are morally or psychologically internalized, so that the sanctioning system is internal to the actor. Such theories require a substantial account of just how the internalization works. No doubt, there is some internalization of norms, but many norms must still depend heavily on external sanctioning, either for them to work at all or for them to work very well. If that were not so, we could dispense with institutionalized law.

Against accounts that require external sanctions, it is sometimes supposed that sanctioning has costs, so that sanctioning a violator of a norm runs against the interests of those who would like to see the norm upheld. For many norms, this is apparently not true. For norms of exclusion, I may actually prefer to shun you if you violate our norm. Hence, I sanction you and benefit from doing so. For example, if you do not follow neighborhood norms of using a relevant slang or dressing in certain ways, I might actively prefer not to associate with you, because your behavior makes me uncomfortable. My reticence and that of others in our neighborhood affects you and damages the pleasures you might get from associating with us or even shuts you out of such association.

For universalistic norms the problem is more complex. Some of these, such as the norm of promise keeping or truth telling, are enforced between dyads or small numbers of participants. In these cases, it may commonly be my interest to sanction you by not cooperating with you on other matters if you break your promise to me on some current matter. Hence, these norms are like norms of exclusion in that they can also be backed by sanctions that serve the interest of the sanctioners. For universalistic norms that govern large-number interactions that are essentially instances of collective action, there may be no sanctioning device that serves the sanctioner's interest, and such norms are, not surprisingly, relatively weak in comparison to dyadic universalistic norms and norms of exclusion (Hardin 1995).


An example of the ways in which rational choice theory is applied to apparently contrary approaches is recent work on functional explanation. This work does not contribute to the older school of functionalism or structural functionalism, as represented in much of twentieth-century anthropology or in the sociology of Parsons and many others. Rather, it reconstructs functional accounts in terms of individual incentives, as did Robert Merton in his effort to be very clear about the logic of functional argument. Oddly, the most important contribution to this new work was intended as a dismissal of functional analysis. Jon Elster (1979) argued that, if the form of functional explanation is properly spelled out, then very few supposedly functional accounts fit that form.

Pared down to its essentials, Elster's account is as follows:

An institution or a behavioral pattern X is explained by its function F for group G if and only if:

  1. F is an effect of X;
  2. F is beneficial for G;
  3. F maintains X by a causal feedback loop passing through G.

Many groups that benefit from some behavior on the part of their members induce that behavior through incentives that they give to their members.

As an example of a functional account of a major institution, return to the problem of social order, which is typically governed in substantial part by a legal system. A common view of much of law is that its function is to coordinate us or to facilitate our interactions. Hence in a functional explanation of law, F is coordination, X is law, and G is our law-governed populace. The feedback loop passing through the populace is that our coordination by law enables us to coordinate to produce still further law to coordinate us still further. Hence, law is functional for us. But we would coordinate in such ways only because it is in our interest to coordinate. Hence, we can explain a major, pervasive, and seemingly all important social institution in functional terms as grounded in the rational choices of the actors.


Perhaps the easiest assessment of where the field of rational choice will go and where it should go is to extrapolate from current trends. Clearly, institutional work looms large for the near term and normative work seems likely to become more important. In both of these developments, one might hope and even urge that research in rational choice take the findings of other approaches seriously. Doing this would entail two quite different programs. The first and simpler program is merely to make extensive use of findings from other approaches. The second and intellectually more challenging program is to attempt to show the complementary relation of various other approaches to rational choice theory—or, alternatively, to demonstrate their incompatibility, which is often asserted but seldom shown. Sometimes, this might even entail showing that some other approach is, at least in some applications, equivalent to rational choice analysis. For example, work that has unpacked the logic of functional explanation often reveals the rationality of actors involved in replicating some supposedly irrational or extrarational behavioral pattern.

Perhaps the greatest challenge to rational choice theory is to fit it to vast bodies of behavioral research that does not focus on individual incentives and intentions. Part of the task here would be to impute incentives and intentions to relevant actors, perhaps by analogy from other studies and contexts. Another part of the task, as in the rational choice analysis of institutions, functional explanation, and norms, is to restructure the problems in ways that make their choice structures clear. Unfortunately, rational choice theory is less well developed in sociology than in economics and even political science, in part because it is embattled. Despite the heyday of exchange theory in anthropology earlier in the twentieth century, rational choice theory is almost entirely absent from that discipline.

Finally, just because rational choice theory focuses on the incentives for microchoices that produce macroeffects, it is particularly suited to policy analysis. Empirical work on incentive systems and how they work can be put to use in designing policies to change behavior in productive ways. This is little more than common sense in many contexts, but a resolute focus on the relation of microincentives to macroresults is an especially natural part of rational choice analyses.

(see also: Social Exchange Theory)


Arrow, Kenneth J. (1951) 1963 Social Choice and Individual Values, 2nd ed. New Haven, Conn.: Yale University Press.

Coleman, James S. 1990 Foundations of Social Theory. Cambridge, Mass.: Harvard University Press.

Cook, Karen S., and Richard Emerson 1978 "Power, Equity, and Commitment in Exchange Networks." American Sociological Review 43:721–739.

Downs, Anthony 1957 An Economic Theory of Democracy. New York: Harper.

Elster, Jon 1979 Ulysses and the Sirens. Cambridge: Cambridge University Press.

Hardin, Russell 1982 Collective Action. Johns Hopkins University Press.

—— 1995 One for All: The Logic of Group Conflict. Princeton, N.J.: Princeton University Press.

Hirschman, Albert O. 1970 Exit, Voice and Loyalty: Responses to Decline in Firms, Organizations, and States. Cambridge, Mass.: Harvard University Press.

Homans, George Caspar (1961) 1974 Social Behavior: ItsElementary Forms. New York: Harcourt Brace Jovanovich.

Mueller, Dennis C. (ed.) 1997 Perspectives on Public Choice:A Handbook. New York: Cambridge University Press.

North, Douglass C. 1990 Institutions, Institutional Changeand Economic Performance. Cambridge: Cambridge University Press.

Olson, Mancur, Jr. 1965 The Logic of Collective Action. Cambridge, Mass.: Harvard University Press.

Smith, Adam 1978 Lectures on Jurisprudence R. L. Meek, D. D. Raphael, and P. G. Stein, eds. Oxford: Oxford University Press. (From notes of Smith's lectures in 1762–1964.)

Ullmann-Margalit, Edna 1977 The Emergence of Norms. Oxford: Oxford University Press.

Russell Hardin