1 Hobbes, Thomas Leviathan, Tuck, Richard ed. (Cambridge: Cambridge University Press 1991), 117Google Scholar

2 See Parsons, Talcott The Structure of Social Action, 2 vols. (New York: Free Press 1968)Google Scholar.

3 For a model that gets around this problem, see Fudenberg, Drew and Maskin, Eric ‘The Folk Theorem in Repeated Games with Discounting or with Incomplete Information,’ Econometrica 54 (1986) 533–54CrossRefGoogle Scholar. However, it is doubtful that any existing sanctioning systems have such a structure.

4 Parsons, Talcott The Social System (New York: The Free Press 1951), 210–18Google Scholar

5 Some of the most insightful analysis of the way that everyday social interaction is supported by sanctions can be found in Garfinkel, Harold Studies in Ethnomethodology (Cambridge: Polity Press 1984)Google Scholar.

6 See Gintis, Herbert ‘Strong Reciprocity and Human Sociality,’ Journal of Theoretical Biology 206 (2000) 169–79CrossRefGoogle ScholarPubMed.

7 See Savage, Leonard J. The Foundations of Statistics (New York: Dover Publications 1972)Google Scholar; and Jeffrey, Richard The Logic of Decision (Chicago: University of Chicago Press 1983), 1Google Scholar. Jeffrey differs from Savage in that he allows the probability of states to be affected by the actions chosen. I follow Savage here, primarily because his framework is the one adopted by most game theorists. See Myerson, Roger Game Theory (Cambridge, MA: Harvard University Press 1991), 5–12Google Scholar.

8 For the full argument to this effect, see Heath, Joseph ‘Foundationalism and Practical Reason,’ Mind 106 (1997) 451–73CrossRefGoogle Scholar. The term ‘myth of the given’ is from Sellars, Wilfrid Empiricism and the Philosophy of Mind (Cambridge, MA: Harvard University Press 1998)Google Scholar. On the same point, see Smith, Michael The Moral Problem (Oxford: Blackwell 1994), 107Google Scholar.

9 Wright, Georg Henrik von An Essay on Deontic Logic and the General Theory of Action (Amsterdam: North-Holland Publishing 1968), 58–68Google Scholar. ‘Impossible’ is being used here in the sense of ‘false at all possible worlds physically accessible to our own, i.e. having the same laws of nature.’

10 One might be tempted to reserve the term ‘principle’ for the subjective intentional state, and use the term ‘norm’ only for reference to shared rules. My reasons for not wanting to draw such a distinction stem from a set of background commitments that cannot be fully elaborated here. See Heath, Joseph Communicative Action and Rational Choice (Cambridge, MA: The MIT Press 2001)CrossRefGoogle Scholar; and Brandom, Robert Making it Explicit (Cambridge, MA: Harvard University Press 1994)Google Scholar. Readers uncomfortable with this usage should feel free to substitute ‘principle’ for ‘norm’ throughout the remainder of this section.

11 As has been widely noted, the primary doxastic and deontic modalities are interdefinable in exactly the same way: Mp =_df ¬N¬p =_df ¬Ip, and Pp =_df ¬0¬p =_df ¬Fp. In words: it is possible that p means that it is not necessary that not p, and that it is not impossible that p. In the same way, it is permissible to do p means that it is not obligatory to not do p, and that it is not forbidden to do p.

12 The precise scale that is used makes no difference. One might be tempted to represent forbidden actions as having negative appropriateness. Similarly, one often represents outcomes as having negative utility. All of these functions remain unique through any positive linear transformation. The only thing that matters is that the scale used to represent the desirability of outcomes be correctly calibrated to the scale used to represent the appropriateness of actions. More on this below.

13 This of course raises a number of technical and methodological questions that form the natural subject of future research. Here I deal only with the most abstract theoretical features of the model.

14 It is perhaps worth noting that, whatever the merits of addition, multiplication is completely implausible so long as the endpoints of the scale used to measure appropriateness are chosen arbitrarily. If one were to multiply appropriateness by desirability, the difference between assigning an action an appropriateness of 0 and assigning it an appropriateness of 1 would be enormous.

15 For overview of different proposals, see Myerson, Game Theory, 310-16Google Scholar.

16 The exposition most people look to is Luce, R. Duncan and Raiffa, Howard Games and Decisions (New York: Wiley 1957)Google Scholar.

17 Jeffrey, The Logic of Decision, 82–4Google Scholar

18 In particular, it is able to provide deliberative microfoundations for a Schelling-style ‘focal point’ solution to the equilibrium-selection problem. For further discussion, see Heath, Joseph ‘The Structure of Normative Control,’ Law and Philosophy 17 (1998) 419–42Google Scholar.

19 See Binmore, Ken Playing Fair: Game Theory and the Social Contract Vol. 1. (Cambridge, MA: The MIT Press 1994), 104–7.Google Scholar

20 Davidson, Donald ‘The Structure and Content of Truth,’ Journal of Philosophy 87 (1990) 279–328CrossRefGoogle Scholar. See also Hausman, Daniel M. ‘Revealed Preference, Belief, and Game Theory,’ Economics and Philosophy 16 (2000) 99–115CrossRefGoogle Scholar.

21 An example of such a theory for deontic constraints would be the conception of ‘practical discourse’ developed by Habermas, Jürgen Moral Consciousness and Communicative Action (Cambridge, MA: The MIT Press 1991)Google Scholar.

22 Here I have in mind so-called ‘cultural dope’ theories of social integration. The discussion in Heritage, John Garkinkel and Ethnomethodology (Cambridge: Polity 1983)Google Scholar provides a useful introduction. See also Wrong, Dennis ‘The Oversocialized Conception of Man in Modem Sociology,’ American Sociological Review 26 (1961) 183–93CrossRefGoogle Scholar. Much of this discussion is unfortunately marred by a deeply uncharitable reading of Talcott Parsons.

23 This proposal bears a certain resemblance to a proposal made by Nozick, Robert in The Nature of Rationality (Princeton: Princeton University Press 1993), 56Google Scholar. However, Nozick retains an essentially consequentialist framework, according to which actions can acquire value either through probabilistic or symbolic connections to other events. He rejects the idea that an action might have ‘the usual kind of utility’ directly associated with it. In my view, this privileging of symbolic associations is unmotivated. Actions are just as likely to have ‘symbolic value’ as outcomes- all of this will be reflected in the agent's preferences. Thus the entire question of symbolic value belongs in the portion of the theory that explains where norms and desires come from, and thus not in the theory of practical rationality strictly construed.

24 I am borrowing this term from Robert Sugden, who uses it in a somewhat different sense.

25 Rasmusen, Eric Games and Information, 2nd. ed. (Oxford: Blackwell 1989), 48–57Google Scholar.

26 Here I am using the term trust in the very general sense in which it is used in sociological theory (and especially the literature on social capital). Consider Fukuyama, Francis: ‘Trust is the expectation that arises within a community of regular, honest, and cooperative behavior based on commonly shared norms, on the part of other members of the community’ (Trust [London: Penguin 1995], 26)Google Scholar. Naturally, more specific trust relationships can develop as agents acquire greater knowledge of the particular norms that individuals accept.

27 Compare Gauthier, David Morals by Agreement (Oxford: Clarendon Press 1986)Google Scholar with Gauthier, David ‘Assure and Threaten,’ Ethics 104 (1994) 690–721CrossRefGoogle Scholar.