Behavioral Game Theory and Smith's Moral Philosophy

Adam Smith is famous for his remark in The Wealth of Nations about how "it is not from the benevolence of the butcher, the brewer, or the baker that we expect our dinner, but from a regard to their interest." This suggests that Smith would support the idea common in neoclassical economics that human beings are predominantly rational egoists who maximize their selfish interests. But then Smith is also famous for his teaching in The Theory of Moral Sentiments that no matter how selfish human beings might be, they are still moved by some fellow-feeling or sympathy with other human beings, so that they share in one another's pleasures and pains, and this is the ground of morality as rooted in moral sentiments that show a concern for others and for how we appear to others.

A lot of the scholarly commentary on Smith turns on the debate as to whether these two positions of Smith are contradictory or compatible. I agree with those who conclude that they are compatible because they manifest a complex view of human nature as both self-regarding and other-regarding. We are selfish animals who care for ourselves. But we are also social animals who care for others. Smith's social science is an attempt to account for that complex human nature that is both selfish and social. As selfish animals, we are inclined to exploit others for our own selfish advancement. But as social animals, we are capable of cooperating for the good of the group.

How do we decide the truth or falsity of Smith's position? It often seems that in moral and political philosophy, we must try to judge the relative plausibility of various positions without any possibility of experimental testing such as we find in the natural sciences. But there are ways of introducing experimental research into moral and political philosophy, and one is to employ behavioral game theory. In fact, the general movement in behavioral game theory research today is to confirm Smith's complex conception of human nature.

Almost 50 years of game theory experimentation has produced results that contradict the narrow conception of human beings as purely or predominantly rational egoists. For example, in the "ultimatum game," we can give some amount of money--say, $100--to a "proposer" who is asked to propose a split with a "responder" who is free to accept or reject the proposed split. If the responder accepts the offer, the split occurs. But if the responder rejects the offer, neither one gets any of the money. If the participants were rational egoists, then the proposer would offer $1 and keep $99 for himself, while the responder would accept this, because $1 is better than nothing. But this is not exactly what happens. Most proposers will offer something close to a 50/50 split, which is accepted. Those 15-25% of the proposers who offer an unfair split will usually have their offer rejected. This pattern of behavior has been replicated consistently, even across diverse societies around the world (with some cultural variation). It suggests that a substantial minority of people really are rational egoists. But a majority of people are "strong reciprocators" who are conditional cooperators (cooperating with others as long as they think others are cooperating) and altruistic punishers (punishing those who violate the norms of cooperation). Those who reject unfair proposals in the ultimatum game are willing to bear a cost to punish unfairness.

Other kinds of games--"dictator games," "public goods" games, "trust" games, and "charity" games--show a similar pattern of behavior. Much of this research is surveyed in Herbert Gintis, et al., Moral Sentiments and Material Interests (MIT Press, 2005). The authors in that book conclude that this experimental research provides scientific support for Smith's understanding of human beings as other-regarding animals whose selfishness is checked by their concern for others.

As the authors in the Gintis book indicate, experimental game theory research needs to be combined with other kinds of research. We might wonder about "experimenter effects" in game-theory experiments: Are these rather contrived games creating behavior that we would not see in "real-world" behavior. But when this experimentation is combined with evolutionary theories of cooperation supported by historical, ethnographic, statistical, and other kinds of empirical research, we can gain growing confidence in our conclusion that human beings are by nature "strong reciprocators" whose motivations conform to Smith's account of human beings as guided both by the moral sentiments and by material interests.

This kind of work would support the sort of interdisciplinary research required for Darwinian political science. We need to explain the political universals of human behavior as reflecting our ambivalent nature as both self-regarding and other-regarding animals. We then need to explain how this ambivalent human nature is expressed in the cultural history of political institutions. And, finally, we need to explain how individual political judgments of political actors show the contingency and uncertainty of political history as constrained by both the genetic evolution of political universals and the social evolution of political cultures. So, for example, we might explain the history of American constitutionalism as a search for institutional structures to channel the ambivalent motivations of human beings so that they can cooperate for the common good, but without expecting them to become perfectly altruistic. And that constitutional history will reflect both the cultural history of constitutional republicanism and the individual history of political actors who must judge what should be done in particular circumstances.

This debate over whether human beings are predominantly rational egoists or strong reciprocators has implications for the possibility of liberty, and thus it's a crucial issue for Darwinian conservatives. Sometimes, it is said that conservative political thought is Hobbesian in its pessimistic assessment of human nature. But this is wrong. Conservatives are not utopians, because they recognize the imperfectibility of human nature, and part of that imperfectibility is the power of self-interest in motivating human behavior. But to assume--as Hobbes does--that human beings are by nature rational egoists and nothing more is contrary to conservative thought, because this Hobbesian view of human nature would say that there is no natural basis for morality. In this respect, Hobbes prepares the way for Kant, for whom morality requires a transcendence of Hobbesian human nature. The Hobbesian conception of human beings as rational egoists would also subvert the case for liberty, because it suggests that the only way for individuals to solve collective action problems is coercive central planning by a Leviathan state. By contrast, conservatives assume that human beings are naturally social animals who are capable of solving collective action problems by establishing norms of trust and reciprocity, and punishing those rational egoists who violate such norms.

Some good illustrations of this point come out in the chapter by Elinor Ostrom in the Gintis et al. book. It has been common for public policy analysts to assume that the management of common pool resources--for example, fisheries, grazing lands, and irrigation systems--requires a national bureaucracy to enforce a top-down plan for the efficient management of resources. But Ostrom's research has shown that local communities can develop their own arrangements for collective management of common resources that are far more effective than any bureaucratically devised and enforced plan.

Here we see the superiority of Hayekian spontaneous order over bureaucratic central planning. The possibility of such spontaneous order assumes a natural human disposition to cooperation based on reciprocity and trust, as well as moralistic punishment of cheaters. This rests on a view of human nature that is neither cynical nor utopian but Darwinian in its realism.

Some of these issues will be discussed in a few days at a Liberty Fund conference in Charleston, South Carolina, that I am directing. We will be reading selections from Smith's Theory of Moral Sentiments along with the book edited by Gintis, Moral Sentiments and Material Interests. From my experience, Liberty Fund conferences are a wonderful way to explore such topics with a great collection of people.  This group includes a Nobel-Prize-winning economist (Vernon Smith), a distinguished evolutionary biologist (David Sloan Wilson), a leading scholar of Adam Smith (Ryan Hanley), and a prominent libertarian journalist (Ron Bailey).   It also helps to have lots of good food and drink in a beautiful location.