1 The first is best illustrated by John Rawls’ analogy between the sense of justice and human linguistic abilities, in A Theory of Justice (Cambridge, MA: Harvard University Press 1971), section 9. Brandt, Richard A Theory of the Right and the Good (Oxford: Clarendon Press 1979)Google Scholar is an outstanding recent example of the second.

2 Gauthier, David Morals by Agreement (Oxford: Oxford University Press 1986)Google Scholar. All page and chapter references are to this book. Note that Chap. VI, which is discussed in section III below, is reprinted in Richmond Campbell and Lanning Sowden, ed., Paradoxes of Rationality and Co-operation (Vancouver: University of British Columbia Press 1985). The title of an earlier paper, ‘Three against Justice: The Foole, the Sensible Knave, and the Lydian Shepherd’ in P. French et al., Midwest Studies in Philosophy 7 (Minneapolis: University of Minnesota Press 1982) 10-29, expresses well Gauthier’s interest in the challenges rational scepticism poses for morality.

3 At the Conference on Contemporary Contractarian Thought at London, Ontario, April 1987.

4 We avoid marking this distinction with the common contrast between ‘moral’ and ’political’ because we agree with L. Greene, The Authority of the State (Oxford: Oxford University Press 1987) that political authority must be more than external coercion; authority has a distinctive structure that shares internalist features with morality.

5 Varian, Hal R. ‘Distributive Justice, Welfare Economics, and the Theory of Fairness,’ Philosophy and Public Affairs 4 (1975) 223-47Google Scholar; reprinted in Hahn, F. and Hollis, M. ed., Philosophy and Economic Theory (Oxford: Oxford University Press 1979)Google Scholar. Varian’s criticism that property rights are not necessary to achieving an efficient market outcome is directed against Nozick, Robert Anarchy, State, and Utopia (New York: Basic Books 1974)Google Scholar, chapter 7. It is stronger against Gauthier who does not claim that property rights or the proviso have independent moral status.

6 Cf. the concept of wealth fairness, Varian, 247.

7 In ‘Deterrence, Rationality, and Rationality,’ Gauthier writes, ‘[O]ur analysis of deterrence is intended to apply generally … I am particularly concerned with the rationality of deterrent policies in the context of relations among … nations’ (Ethics 94 [1984], 478).

8 This distinguishes Gauthier’s argument from the ‘basic argument for cooperation’ discussed by Davis, Lawrence H. ‘Prisoners, Paradox, and Rationality’ (American Philosophical Quarterly 14 [1977] 319-27Google Scholar; reprinted in Campbell and Sowden) where interdependence is supposed to follow directly from the symmetry of equal rationality. It still may be the case that Gauthier relies too heavily on indirect appeals to equal rationality, which is a suspiciously moral assumption. We take this up in a sequel to the present paper: P. Danielson, ‘Closing the Compliance Dilemma’ in P. Vallentyne, ed., Contractarianism and Rational Choice: Essays on Gauthier (forthcoming).

9 CC is analogous to one of the three meta-strategies associated with equilibria in Howard’s meta-game analysis; see Steven Brams Paradoxes in Politics (New York: Macmillan 1976), section 4.6 discussing Nigel Howard, Paradoxes of Ratioality: Theory of Metagames and Political Behavior (Cambridge, MA: MIT Press 1971). It is only analogous because, concurring with Jon Elster’s objection to Howard’s asymmetrical procedure, Ulysses and the Sirens (Cambridge: Cambridge University Press 1979), 143, our constructions will be symmetrical. We therefore abandon Howard’s asymmetrical labels, retaining the category of meta-strategy to isolate the component of CM that matches some pattern of the others’ strategy choice, choice rule, or meta-strategy. (See Brams section 8.6 for a more limited choice rule interpretation of CC.) Notwithstanding our differences, we owe the RC meta-strategy to Howard.

10 In an accessible, if uncritically enthusiastic, broad introduction to Howard’s results, Rappoport puts the coherence argument for asymmetry: ‘it is essential, however, that the game be constructed by the stepwise method, assigning four conditional strategies to one player and then 16 to the other. If both players are simply given four conditional strategies apiece, it is impossible for a referee to determine the outcomes of some pairs of choices, for instance the outcome of both players’ choosing strategies such that each selects the same move he expects the other to make’ (Scientific American 217 [1967] 55).

11 In this footnote we formulate the meta-strategy Selfsame Co-operation, (SC more formally than in the text. Using capital letters for variables, we define a predicate, sc(I,J), meaning player I is committed to meta-strategy SC with player J. Consider two players, AI and Betty. Using lower case constants, we can state Al’s commitment to SC: co-operates (al,X) IF sc(X,al). AI will co-operate with anyone committed to Selfsame Co-operation with AI himself. Betty’s commitment is similar: co-operates(Betty,X) IF sc(X,Betty).

Finally, we need to define the principle SC in a way to avoid regress when these symmetrical agents appeal to it. The operation needed to define meta-logical predicates is available in (Edinburg) PROLOG (Programming in Logic) as the predicate, clause(C,A), which is true when C is the consequent of some antecedent A. Thus in something close to PROLOG, we define:

sc(I,J) IF clause(co-operate(I,J),sc(J,I)).

That is, SC is true of I with respect to J if I is committed to co-operate with J in case SC is true of J with respect to I. Mechanically executable formulations of the meta-strategies discussed in this paper prove coherence by simple exemplification. For more details, see my Artificial Morality (in progress). A tutorial with working PROLOG versions is available via electronic mail from the author: DANIELSN@YORKVM1.BITNET

12 David Gauthier, ‘What Do I Think Morals by Agreement is About?’ (Paper read at the Canadian Philosophical Association meeting, Winnipeg 1986), 7f.

13 Cf. Axelrod, Robert The Evolution of Cooperation (New York: Basic Books 1984), 54, 120-3Google Scholar. Indeed, we have argued for a version of CC over RC in Axelrod’s context. Cf Peter Danielson, ‘The Moral & Ethical Significance of TIT FOR TAT,’ Dialogue 25 (1986), section 3. ‘Perhaps’ (in the text) marks our ignorance and suggests we need empirical work, similar to Axelrod’s tournaments, to extend our knowledge here. In drawing an analogy to Axelrod, we are aware that Gauthier distinguishes CM from TIT FOR TAT (169, n. 19). Notwithstanding the differences, the empirical questions raised by partial transparency do seem amenable to tournament methods. Of course, the Prisoner’s Dilemma would not be iterated, and questions of communication would become central. Indeed, our claim that RC solves the Prisoner’s Dilemma has yet to be tested against the unexpected deviousness that a tournament brings out. Those willing to participate in such a tournament should contact the author.

We also should note that Gauthier exaggerates his differences with Axelrod. For example, on p. 207 he claims that ‘in natural interaction it would be irrational for Joanna to run any risk whatsoever to save Jonathan from the sharks in the lagoon … .’ But if Axelrod’s ‘appeal to reciprocity’ has any application, it applies here, in a case of continuing interaction.

14 At least one evolutionary biologist claims SC is the stable meta-strategy: ‘Hence a social contract can ensure stable cooperation only if it reads “I will cooperate; I will join in punishing any defection; I will treat any member who does not join in punishing as a defector’” (Maynard Smith, John ‘The Evolution of Animal Intelligence,’ in Hookway, C. ed., Minds, Machines, and Evolution [Cambridge: Cambridge University Press 1984], 68-9)Google Scholar. Recently Axelrod has developed this idea extensively under the concept of meta-norms; ‘An Evolutionary Approach to Norms,’ American Political Science Review 80 (1986) 1095-111.

15 This can be demonstrated using Axelrod’s method of ecological projections, discussed in Danielson, ‘ … TIT FOR TAT.’

16 See note 8. The parameterized conditional meta-strategies needed to deal with the n-player case are developed in Danielson and Roosen-Runge, ‘How to Strengthen Moral Principles by Weakening Them’ (York University, unpublished manuscript 1986).

17 I would like to thank David Gauthier, Leslie Green, Peter Roosen-Runge, David Copp and Richmond Campbell for their helpful comments.