The idea that morality should be regarded as a science has a long and respectable history. The tradition begins with pre-Socratic philosophers like Antiphon and Xenophanes, and embraces Aristotle and Epicurus on its way to Hobbes and such luminaries of the Scottish Enlightenment as Hume and Adam Smith. Through Adam Smith and Malthus, it exerted a direct influence on Darwin's formulation of the theory of natural selection. Russell and Popper carried the standard of scientific philosophy forward into modern times, but perhaps the most thoughtful recent work by a moral philosopher working in this tradition is J.L. Mackie's Ethics: Inventing Right and Wrong.
Mackie systematically demolishes the standard objections to taking a naturalistic approach to ethics in the early part of his book, but runs dry when it comes to erecting a building on the foundations he laid so firmly. But he points the way by drawing attention to the relevance of game theory to the problem. Twenty years later, Brian Skyrms’ Evolution of the Social Contract picks up the baton where Mackie left it. Philosophers will find that Skyrms’ book provides a magnificent overview of the light that game theory can shed—not just on morality—but on human sociality in general. Other books provide more detailed accounts, notably my own books Game Theory and the Social Contract (I and II) and Sugden's Economics of Rights, Cooperation and Welfare, but Skyrms has a gift for language and an eye for apt examples that are granted to few others.
The core of the naturalistic approach is the claim that morality is a set of norms or conventions that evolved as equilibrium selection devices for the many games of life that we play with each other every day. Everybody knows that the one-shot Prisoners’ Dilemma has only one equilibrium, in which both players defect. Since the one-shot Prisoners’ Dilemma has been widely canvassed as expressing the essence of the problem of human cooperation, philosophers have put much effort over the years into showing that this analysis of the one-shot Prisoners’ Dilemma is mistaken, but I hope I am right in thinking that this enterprise has now largely been abandoned as hopeless. Game theorists never understood this obsession, because we never believed that the one-shot Prisoners’ Dilemma was particularly relevant to the problem of how people cooperate. It seems obvious to us that if the one-shot Prisoners’ Dilemma captured what matters about the human game of life, then we would not have evolved as social animals.
Realistic games, particularly repeated games that are played over and over again, typically have very large numbers of equilibria. The real problem is therefore not how we can cooperate when cooperation cannot be sustained in equilibrium, but how we manage to choose between all the many ways of cooperating, when all of these many ways can be sustained as equilibria in a realistic game of life. This is why naturalists like Skyrms and myself think the human species needs morality and all the other conventions that make up the social contract of a society. We need them as equilibrium selection devices. Critics—of whom there are many—tell us that by reducing ethics to the same level as traffic signals, we degrade the human spirit. But much the same was said of the biologists who figured out the nuts-and-bolts of the supposedly ineffable mystery of life.
Skyrms approaches the problem of justice by considering computer simulations of the workings of evolution on the strategies of people playing simple bargaining games. Following a number of economists (such as myself), he finds that the equilibrium which evolution selects tends to split the available surplus evenly. This is true even for the famous Ultimatum Game, in which one player makes a take-it-or-leave it offer to the other. This does not have only one equilibrium, as critics of game theory usually claim. Any outcome whatever can be supported as a Nash equilibrium of the Ultimatum Game. Which of these equilibria is chosen by an evolutionary process seems to depend on the detailed nature of the small perturbations built into the process, but the robust conclusion is that the player who makes the ultimatum does not get the entire surplus.
In brief, the evolutionary simulations generate outcomes that are not so distant from the outcomes observed in laboratories that the subjects defend as fair. My own view is that such results explain why symmetry is important in locating focal points, but this is at best only half the story. We need to get to grips with the norms that evolve in repeated games, whose study is too complex to be left to computer simulations—as the failure of Axelrod's claims for tit-for-tat makes abundantly clear. We also need to take account of the manner in which people make interpersonal comparisons of welfare. I think that this is where the human capacity for empathy and the Rawlsian original position come onto the scene—but this is another story.
Rather than exploring further in this direction, Skyrms surveys several other applications of game theory to human sociality. Recognizing David Hume as the father of the subject, he offers a translation of Hume's view that language evolved as a coordination device into modern terms. He also discusses the possible evolution of commitment devices, and provides evidence of the importance of correlation in evolutionary processes.
It is only on the latter point that I differ from Skyrms in any serious way. We do not differ on the importance of correlation—which arises when the players can condition their choice of strategy in a game on a signal the players commonly observe before playing. A traffic signal is a typical example. You go when it is green and stop when it is red. But I disagree very much with the idea that the evolution of correlated conventions provides any support for either Kant's categorical imperative, or for Jeffrey's theory of decision. Game theorists think that both are founded on fallacious forms of reasoning (Binmore Reference Binmore1998). It is true that evolution can generate outcomes that are close to what one would get by applying Kant's or Jeffrey's ideas, but no exotic reasoning is necessary to explain the phenomena. All that is happening is that evolution is selecting an equilibrium in a game whose rules particularly favor cooperation.
Moral philosophers with a naturalistic bend tend to spend their time reassuring each other that Mackie was right when he cleared away the philosophical obstacles to treating morality as a science. In Evolution of the Social Contract, Brian Skyrms points a way out of this slough of despond to the little wicket gate that opens on the strait and narrow path of serious scientific enquiry. It will be interesting to see how many philosophers will be ready to follow him on his brave journey.
