If X undertakes to Y that X will do Ax if Y will do Ay, and Y undertakes the converse to X, they do not agree that X will do Ax and Y will do Ay, for both will comply with their undertakings if neither does anything. Nor is an agreement generated by complicating the conditionals, so that, for example, X undertakes that X will do Ax if, if X will do Ax, Y will do Ay, and Y undertakes the converse; for then, depending on the interpretation of “if,” either both parties can still comply by doing nothing or—if the ifs are truth-functional—each party in effect merely undertakes to perform the action in question; for (P⊃Q)⊃P≡P.1
Compare C. Radford, I Will, If You Will, 93 MIND 580–81 (1984). A more elaborate complication of the conditionals is proposed by H. Prichard in an analysis of exchange: H. Pritchard, Exchanging, in MORAL OBLIGATION (1949).
If X merely undertakes to Y that X will do Ax, and Y merely undertakes to X that Y will do Ay, there is just a pair of undertakings, not an agreement.
Can an agreement be modeled in terms of undertakings by the parties in a way that avoids these problems and is otherwise satisfactory? Some philosophers2
Notably Gilbert; see M. Gilbert, Is an Agreement an Exchange of Promises? 90 J. PHIL. (1993), reprinted in M. GILBERT, LIVING TOGETHER (1996); M. Gilbert, Agreements, Coercion, and Obligation, 103 ETHICS (1993), reprinted in M. GILBERT, LIVING TOGETHER (1996).
think not, but the orthodox view, insofar as there is one, appears to be that such a model can be provided.
3 See P. ATIYAH, PROMISES, MORALS, AND LAW 204–205 (1981); K. Bach Terms of Agreement, 105 ETHICS (1995); A. BAIER, POSTURES OF THE MIND 199 (1985); D. LEWIS, CONVENTION: A PHILOSOPHICAL STUDY 34, 35, 84 (1969); H. Prichard, The Obligation to Keep a Promise, in MORAL OBLIGATION (1949); Radford, supra note 1, at 582; J. Raz, On the Nature of Rights, 93 MIND 203 (1984); M. ROBINS, PROMISING, INTENDING, AND MORAL AUTONOMY 105 (1984); D. Velleman, How to Share an Intention, 57 PHIL. & PHENOMENOLOGICAL RES. (1997), reprinted in THE POSSIBILITY OF PRACTICAL REASON (2000); D. Velleman, Deciding How to Decide, in ETHICS AND PRACTICAL REASON app. (G. Cullity & B. Gaut, eds., 1997), reprinted in THE POSSIBILITY OF PRACTICAL REASON (2000). For a similar view of the nature of contracts, see 1 J. CHITTY, CHITTY ON CONTRACTS 1–001 (28th ed., 1999).
Talk of orthodoxy is perhaps out place, for the philosophical literature on the nature of an agreement is relatively sparse; this is surprising, given the number and the importance of agreements. To state the obvious: agreements form one of the main types of social bond; they are central to the law—to contract law,
4 A standard definition of a contract in common law is that it is an agreement giving rise to obligations that are enforced or recognized by law; G. TREITEL, THE LAW OF CONTRACT 1 (9th ed., 1995). Contracts and agreements overlap but do not coincide; for example, unilateral contracts and deeds made by one person are not agreements in any normal sense. Also, contract law has been extended since the late nineteenth century to protect detrimental reliance in the absence of an agreement other than in an unusually broad sense; see P. ATIYAH, THE RISE AND FALL OF FREEDOM OF CONTRACT passim (1979); G. GILMORE, THE DEATH OF CONTRACT passim (1995); and P. COOKE & D. OUGHTON, THE COMMON LAW OF OBLIGATIONS 47 (1993). For a review of objections to the identification of contracts with agreements, see CHITTY, supra note 3, at 1-003-6. Contrast De Moor, who says that “an agreement” adds little to the meaning of “a contract”; A. De Moor, Are Contracts Promises? in OXFORD ESSAYS IN JURISPRUDENCE 115 (J. Eekelaar & J. Bell, eds., 1987).
of course, but also to criminal law (specifically, the rules on conspiracy), to company law (the rules on concert parties), and to competition law (the rules on cartels); and the concept of agreement is prominent in various theoretical areas, such as the theory of cooperative games and social contract theories of political obligation.
5 In these areas the concept of an agreement or of a contract is more often presupposed than analyzed. Game-theoretic accounts of bargaining throw some light on the nature of an agreement, but cooperative solutions such as Nash's assume an enforceable agreement between the parties; J. Nash, The Bargaining Problem, 18 ECONOMETRICA (1950); A. DIXIT & S. SKEATH, GAMES OF STRATEGY ch. 16 (1999). As to the social contract, Hobbes gives a brief account of a contract as “the mutual transferring of right”; T. HOBBES, LEVIATHAN 192 (Penguin Books 1968). Hume distinguishes a strong and a weak sense of “contract” in D. Hume, Of the Original Contract, in ESSAYS, MORAL, POLITICAL AND LITERARY 454–455 (1963). Rawls makes some remarks on the appropriateness of the term to describe his own theory; J. RAWLS, A THEORY OF JUSTICE 16 (1972).
I shall defend the orthodox view by arguing for two models that develop the idea that there is an agreement where one party gives a conditional undertaking and the other responds with an unconditional undertaking. Each model is intended only to represent a central case of a central concept of agreement; there may be more than one concept of agreement (and the boundaries between them may be hard to draw).6
This is true even if only everyday rather than legal agreements are in question. Gilbert talks of “the everyday concept of an agreement,” which implies that there is only one such concept; Gilbert, Is an Agreement an Exchange, supra note 2, at 628. For one distinction between concepts of an agreement, see S. Freeman, Contractarianism, in 2 ROUTLEDGE ENCYCLOPEDIA OF PHILOSOPHY 663 (E. Craig, ed., 1998).
I therefore leave open the possibility that certain agreements are better understood in other terms, for example as “joint decisions” in Margaret Gilbert's sense.
7 M. GILBERT, ON SOCIAL FACTS 380 (1989); Gilbert, Is an Agreement an Exchange, supra note 2, at 636; Gilbert, Agreements, Coercion, supra note 2, at 292–296. I also shall not discuss the relation between agreements and other forms of cooperation, as to which, see D. GAUTHIER, MORALS BY AGREEMENT ch. 5 (1986). For a distinction between agreements and joint action, see O. Black, Concerted Practices, Joint Action and Reliance, 24 EUR. COMPETITION L. REV. (2003).
Moreover, for any given concept, a model of this kind need not purport to specify necessary and sufficient conditions for every one of the concept's applications; accordingly, the model should not be rejected merely if a case can be found that either does not fit certain of its clauses or has certain features relevant to the concept's application that the model fails to capture. It is debatable whether illuminating necessary and sufficient conditions can be given for any concept, but the aspiration to specify them is quixotic in the case of concepts, such as that—or those—of agreement, which have been described and applied in various obscure, confused, and arguably contradictory ways. For such concepts a model may be treated as prescriptive, regimenting our intuitive judgments. Nevertheless it is a virtue of a model that it should fit those judgments, and it must be rejected if it diverges from them too often or too widely. The aim should be a reflective equilibrium of intuition and theory.
THE FIRST MODEL
Consider this first attempt:
“Undertakes” will be left unanalyzed. The obvious case of undertaking is promising; all promises are undertakings, but it seems that some undertakings are weaker than promises—vows, for example, are weaker in that they need not be made to anyone
8 See De Moor, supra note 4, at 109; C. FRIED, CONTRACT AS PROMISE 40–43 (1981).
—and others, such as oaths, are stronger than mere promises. Since many agreements appear to involve nothing so solemn as a promise, the broader concept is preferable.
9 K. BINMORE, PLAYING FAIR 60 (1994) characterizes an agreement in terms of undertakings. Prichard says that “undertake,” “promise,” and “agree” are all “equivalent”; Prichard, The Obligation, supra note 3, at 172. Another candidate for the model's right-hand side is the slightly different concept of commitment; see Bach, supra note 3, at 606–607. The U.S. Restatement (Second) of Contracts § 2 (1981) defines a promise in terms of a commitment; Binmore does the opposite in id. at 161; and in BINMORE, JUST PLAYING 47 (1998). For distinctions between types of commitment, see M. LIEBERMAN, COMMITMENT, VALUE, AND MORAL REALISM 5 (1998).
M1 and its successors are intended only to display the structure of the interaction between X and Y; it is not suggested that X and Y need say “I undertake [etc.].” The exchange is likely to be more informal, for example: X: “Why don't you wash the dishes and I'll make the beds.” Y: “Fine.” Indeed X and Y may say nothing, their undertakings being constituted by other actions.
Pretend, however, that X says “I undertake that I will do Ax if you will undertake that you will do Ay.” Depending on the scope given to “I undertake that” and “if,” the sentence can be interpreted to express an undertaking that is either externally conditional—“If you will undertake that you will do Ay, I undertake that I will do Ax”—or internally conditional “I undertake that, if you will undertake that you will do Ay, I will do Ax.”
10 This generalizes to undertakings the distinction Gilbert draws between conditional promises; Gilbert, Is an Agreement an Exchange, supra note 2, at 632–633.
Each interpretation fits some agreements, but the internal one appears to fit more and will therefore be adopted here. So the model becomes:
M2 has the defect mentioned earlier: a mere pair of undertakings does not amount to an agreement, even where the giver of each undertaking is the addressee of the other. The missing element is the idea (expressed in contract law by the principle that, for a contract to exist, there must be an offer that is accepted)
11 It appears to be a principle of English contract law that an act wholly motivated by factors other than the offer cannot amount to an acceptance; Lark v. Outhwaite (1991) 2 Lloyd's Rep 132, 140. Bach adopts an offer-acceptance model in his account of an agreement as an exchange of promises: the idea seems to be that X and Y exchange promises by X's making an offer and Y's accepting it, the offer and the acceptance themselves being promises; Bach, supra note 3, at 606–607.
that Y gives his undertaking in
response to X's. This can be developed by saying that Y's reason—or one of Y's reasons—for giving his undertaking is that X has given
his undertaking. There are many kinds of reason; here the genus is practical reasons and the species comprises both justification and explanation: That X has given X's undertaking both justifies Y in giving and explains the fact that Y gives Y's undertaking. The nature of the justification will be discussed later, when I consider the forms of practical deliberation that would lead rational parties to give their undertakings.
12 Broome holds that deliberation is not reason-giving, but rather is a matter of “normative requirements”; see J. Broome, Practical Reasoning, in REASON AND NATURE: ESSAYS IN THE THEORY OF RATIONALITY (J. Bermúdez & A. Millar, eds., 2002) and the other papers cited in note 45 below. I here use “reason” in a broad sense that glosses over the distinction. Clause (c) in the models of an agreement below can be expanded to accommodate Broome's point explicitly.
As to explanation, the most promising thought is that it is causal. In more detail, let Ux and Uy be respectively X's and Y's acts of giving their undertakings; then the fact that Y does Uy is explained by the fact that there is an appropriate causal chain from Ux to Uy, where “there is a causal chain from Ux to Uy” means “Ux stands to Uy in the proper ancestral of the relation … causes….”
13 In O. Black, The Infinite Regress of Justification 104–110 (1987) (unpublished Ph.D. thesis, University of London) (on file with the author); and O. Black, Infinite Regresses of Justification, 28 INT'L PHIL. 429 (1988), this relation is used to explicate the notion of one belief's being based on another.
The requirement that the chain be appropriate is intended to exclude deviant cases in which the chain involves grossly irrational mental processes, or even no mental processes at all, on Y's part. A full account of appropriateness will not be attempted here,
14 See C. PEACOCKE, HOLISTIC EXPLANATION ch. 2 (1979).
but a necessary condition is that Y believe that X does Ux.
15 This is a point at which contract law may diverge from the present concept of agreement. There is doubtful authority in English law for the principle that a party may accept an offer of which he is ignorant; Gibbons v. Proctor (1891) 64 LT 594, 55 JP 616; compare Neville v. Kelly (1862) 12 CB (NS) 740; and see Chitty, supra note 3, at 2-038-9. The authority is clear—e.g., in the posting rule—for the converse principle that a contract may arise where the offeror is ignorant of the acceptance; Henthorn v. Fraser (1892) 2 Ch 27, 33; Adams v. Lindsell (1818) 1 B & Ald 681; Potter v. Sanders (1846) 6 Hare 1; Harris' Case (1872) LR 7 Ch App 587. See Chitty, id. at 2-043-57, and, for other situations in which the offeror's ignorance does not preclude the creation of a contract, at 2-042A.
Normally not only will Y believe this, but his belief will be justified or amount to knowledge. Where the belief is justified, the justification will be epistemic rather than the practical kind of justification mentioned above; but it can be argued that, by virtue of the requirement of appropriateness, the explanatory element of the reason-relation here entails the element of practical justification. The model now becomes:
(c) says “X gives,” whereas the previous discussion used “X has given”; but, given that “Y's reason” signifies a causal chain from Ux to Uy, and that effects normally succeed their causes,
16 It may be argued that sustaining, rather than efficient, causes can occur at the same time as their effects; e.g., a weight's lying on a cushion causes the cushion to be indented. See Black, The Infinite Regress, supra note 13, at 117, 183–184.
(c) can be taken to imply that Ux precedes Uy. “Y's reason” is intended to include the case where Y also has other reasons. Note that the relation specified in (c) between the two undertakings is external in a sense stronger than the one discussed earlier, in which X's undertaking is not externally conditional: the latter sense concerns the respective scopes of “I undertake that” and “if”—matters that are internal to the content of X's undertaking—whereas the relation in (c) is external to the contents of Ux and Uy in that it connects the acts themselves. Note also that because this relation is strongly external and because it is not one of simple conditionality, M3 does not have the defect of the reciprocally conditional undertakings mentioned at the start, that both parties will comply if neither does anything.
It may be suggested that the model should mention not only Y's reason but also X's—not for doing Ux but for doing Ax. The thought17
Put forward by a referee for this journal.
is this: Suppose that, first, X knows that if Y does Uy, Z will force X to do Ax, where Ax is something that X hopes not to do and believes it would be bad for him to do; second, X, in order to inform Y of this, tells Y that, if Y does Uy, X will do Ax; and third, for the reason that X tells this to Y, Y vindictively takes advantage of X's predicament by doing Uy. Then it seems that M3 is satisfied, but X and Y surely do not agree that they will respectively do Ax and Ay. One proposal for dealing with this case is to exclude it by rewriting the end of M3a to say that X, for the reason that Y will have done Uy, will do Ax. But the proposal is unnecessary because the case is not a counterexample to M3. Without analyzing the concept of an undertaking, it is clear enough that X's utterance does not amount to an undertaking that if Y does Uy, X will Ax; rather, it is a conditional prediction and perhaps a warning.
As noted earlier, M3c implies that Y believes that X does Ux, and Y's belief will normally be justified or amount to knowledge. It may be suggested that a corresponding condition should be imposed on X—that X believe, or believe with justification, or know, that Y does Uy; for otherwise, it may be said, the model fails to capture the fact that an agreement involves a meeting of minds. This step can lead to an infinite regress: If X must believe that Y does Uy, perhaps Y must believe that X believes this, and perhaps X must believe that Y believes that X does Ux, and so on. Such thinking motivates the addition of a clause attributing to X and Y the common belief, or justified belief, or knowledge, that they respectively do Ux and Uy, and perhaps also that Y's reason for doing Uy is that X does Ux: roughly, X and Y have the common belief that P if and only if X believes that P, Y believes that P, Y believes that X believes that P, X believes that Y believes that P, and so on, and likewise for common justified belief or knowledge.
18 Gilbert's account of an agreement attributes common knowledge to the parties; Gilbert, Agreements, Coercion, supra note 2, at 292–296; see also M. Gilbert, Agreements, Conventions, and Language, 54 SYNTHESE 109 (1983), reprinted in LIVING TOGETHER (1996). On common knowledge generally, see R. Aumann, Agreeing to Disagree, 4 ANNALS OF STATISTICS (1976); K. Binmore & A. Brandenburger, Common Knowledge and Game Theory, in ESSAYS ON FOUNDATIONS OF GAME THEORY (K. Binmore, ed., 1990); J. Geanakoplos, Common Knowledge, 6 J. ECON. PERSP. (1992); J. Heal, Common Knowledge, 28 PHIL. Q. (1978); LEWIS, supra note 3, at 52–60; P. Milgrom, An Axiomatic Characterization of Common Knowledge, 49 ECONOMETRICA (1981); L. Nielsen, Common Knowledge, Communication, and Convergence of Beliefs, 8 MATHEMATICAL SOC. SCI. (1984); R. Parikh & P. Krasucki, Communication, Consensus and Knowledge, 52 J. ECON. THEORY (1990); S. SCHIFFER, MEANING chs. 2, 3 (1972); S. SCHIFFER, REMNANTS OF MEANING 245–246 (1987); MUTUAL KNOWLEDGE (N. Smith, ed., 1982).
How far one should go along this path depends first on whether an agreement involves a meeting of minds
19 The principles in note 15 above support the view that a meeting of minds is not needed for a contract to exist; see, further, Chitty, supra note 3, at 1-004, which cites Byrne v. Van Tienhoven (1880) 5 CPD 344 in support of the principle that a contract can arise where the parties are not “ad idem.” Also it appears to be a principle of English contract law that where the offeree believes the offeror to have the relevant intention—that the offer is to become binding on acceptance—it is not necessary that the offeror be aware of the offeree's belief; The Hannah Blumenthal (1983) 1 AC 854, 924; Chitty, supra note 3, at 2-003. This pulls against a requirement of common belief. A variant is the concept of a meeting of wills, which is more prominent in French than in English contract law; see A. De Moor, Contract and Agreement in English and French Law, 6 OXFORD J. LEGAL STUD. (1986); and supra note 4, at 115–116. But see Lord Scarman's justification for “economic duress” as a defense against an action for breach of contract: “duress … is a coercion of the will so as to vitiate consent”; Pao On v. Lau Yiu Long (1980) AC 614, 635. For a recent example from European Community law, see Bayer v Commission, a competition case in which the court held that an agreement involves a “concurrence of wills”: Case T-41/96 (2001) CMLR 176.
and second on how strong a sense “meeting of minds” has: the answers are partly a matter of stipulation, for the phrase is more venerable
20 See G. GORDLEY, THE PHILOSOPHICAL ORIGINS OF MODERN CONTRACT DOCTRINE 135–136 (1991), which cites Plowden and Comyns.
than clear, and it can be given a meaning weak enough to apply to M3. To add a clause attributing common belief (etc.) to X and Y would capture an unusually strong concept of agreement and would also raise the general difficulties with common propositional attitudes, notably that it seems psychologically unrealistic to hold that people ever have them.
21 See SCHIFFER, supra note 18, at 246. Proposals for psychologically plausible forms of common knowledge or belief are found in J. BENNETT, LINGUISTIC BEHAVIOUR 127 (1976); A. Kemmerling, Utterer's Meaning Revisited, in PHILOSOPHICAL GROUNDS OF RATIONALITY: INTENTIONS, CATEGORIES, ENDS (R. Grandy & R. Warner, eds., 1986); B. LOAR, MIND AND MEANING 250 (1981); R. TUOMELA, THE IMPORTANCE OF US 41–51 (1995). On the broader issue of a person's having an infinity of beliefs, see Black, The Infinite Regress, supra note 13, at ch. 8, and Black, Infinite Regresses, Infinite Beliefs, PROC. KNOWLEDGE AND BELIEF (W. Löffler and P. Weingartner, eds., 2003b).
A more attractive proposal is to add just the clause that X has the justified belief that Y does Uy: intuitively, this captures the idea that not only Y but both parties are aware of what is going on.
22 A strong version of this idea is embodied in Gilbert's “‘knowledge of agreements’ assumption” that, if a person enters into an agreement, he knows that he does; Gilbert, Agreements, Coercion, supra note 2, at 284.
The model becomes:
23 Where the undertakings are promises, Raz appears to hold that (a) and (b) are sufficient for an agreement; Raz, supra note 3, at 203. He appears to take the same view of (a) and (b) in M6 below. But Raz explicitly disregards the distinction between the internally and the externally conditional (which, in the case of duties, he conflates with a distinction between present and future duties); Raz, supra note 3, at 196, n1. Bach likewise appears to think that an agreement may involve promises which respectively have the structures of the undertakings in (a) and (b) in M4; Bach, supra note 3, at 606–607. In Gilbert, Is an Agreement an Exchange, supra note 2, at 642, Case 8, read as involving an internally conditional promise, instantiates (a) and (b) in M4. Surprisingly, Gilbert does not examine a case instantiating the corresponding clauses of M6.
M4 is the first of the two models I propose.
ALLEGED CRITERIA OF ADEQUACY
Four criteria may be proposed for the adequacy of a model of agreement: Symmetry, Obligation, Simultaneity, and Interdependence. M4 breaches all of them, so either M4 is inadequate or the criteria are misconceived. The latter is the case, but certain of the criteria are approximations to more plausible ones which M4 meets.
The idea of Symmetry is that both parties do the same thing.24
Radford, supra note 1, at 581, talks of the “parallel nature” of the things said by the parties, but he in effect rejects Symmetry.
It is hard to make this precise, but there is no need to, as it is clear that the criterion is breached by M4: specifically, the content of Ux is conditional and that of Uy is not, and, whereas Y does Uy for the reason that X does Ux, the converse need not be true. So much the worse for Symmetry. M4 does, and the criterion does not, reflect the fact that normally in the making of an agreement, one party—X in the model—goes first.
25 Compare Bach, supra note 3, at 606–607. Bach describes a counterexample in which a third party suggests a plan and effects its joint acceptance.
Contract law recognizes this in the doctrine of offer and acceptance.
The remaining three criteria have been proposed by Gilbert, as part of an argument against that version of the orthodox view which holds that an agreement is an exchange of promises. Gilbert expresses Obligation by saying that “the agreement directly generates the relevant performance obligation for each of the parties,” a performance obligation being “an obligation to perform the specified act.”
26 Gilbert, Is an Agreement an Exchange, supra note 2, at 629–630.
The specified acts in the model are Ax and Ay. Now the following principles are plausible:
27 Raz, supra note 3, at 202, states a similar principle for promises. Tuomela, supra note 21, at 241, says that it is constitutive of the notion of agreement that an agreement creates an obligation for the parties.
Different views may be taken as to the ground of the obligations. On one view, they arise from the fact that an undertaking causes, or is intended or can be foreseen to cause, reliance; see Atiyah, supra note 4, esp. chs. 1, 6, 22; Atiyah, supra note 3, esp. chs. 3, 7; P. ATIYAH, ESSAYS ON CONTRACT (1986); J. Demarco & R. Fox, Putting Pressure on Promises, 30 S. J. PHIL. 48 (1992); R. Fogelin, Richard Price on Promising: A Limited Defence, 21 J. HIST. PHIL. (1983); C. Knapp, The Promise of the Future—and Vice Versa: Some Reflections on the Metamorphosis of Contract Law, 82 MICH. L. REV. (1984); N. MacCormick, Voluntary Obligations and Normative Powers (I), PROC. ARISTOTELIAN SOC. (Supp. vol. 46, 1972); C. McMahon, Promising and Coordination, 26 AM. PHIL. Q. 239–247 (1989); T. Scanlon, Promises and Practices, 19 PHIL. & PUB. AFF. (1990); SCANLON, WHAT WE OWE TO EACH OTHER ch. 7 (1998); and Scanlon, Promises and Contracts, in The THEORY OF CONTRACT LAW (P. Benson, ed., 2001). For criticism, see J. Altham, Wicked Promises, in EXERCISES IN ANALYSIS (I. Hacking, ed., 1985); P. Ardal, Promises and Reliance, 15 DIALOGUE 54–61 (1976); J. Cartwright, An Evidentiary Theory of Promises, 93 MIND (1984); R. Downie, Three Accounts of Promising, 35 PHIL. Q. (1985); R. Fox & J. Demarco, The Immorality of Promising, 27 J. VALUE INQUIRY 82 (1993); FRIED, supra note 8, at chs. 1, 2; C. Morris, Some Notes on “Reliance,” 75 MINN. L. REV. (1991); D. Patterson, The Value of a Promise, 11 LAW & PHIL. (1992); J. Raz, Voluntary Obligations and Normative Powers (II), PROC. ARISTOTELIAN SOC. (Supp. vol. 46, 1972); Raz, Promises and Obligations, in LAW MORALITY, AND SOCIETY (P. Hacker & J. Raz, eds., 1977); G. WARNOCK, THE OBJECT OF MORALITY ch. 7 (1971); A. Woozley, Promises, Promises, 90 MIND (1981). On the relations between reliance and obligation generally, see O. Black, The Relation between Reliance and Obligation, in APPLIED ETHICS (P. Kampits, et al., eds., 1998); and Black, Reliance and Obligation, RATIO JURIS (forthcoming).
A referee for this journal has objected to O2b as follows. Suppose that I am a dentist and I undertake to you, my patient, that, when you raise your hand, I shall stop drilling. I do not discharge this obligation if, when you raise your hand, I stop for some different reason and while ignoring the fact that you have raised your hand. But I would discharge it if the obligation were merely to stop when you raise your hand. So my obligation is a stronger one, to stop for the reason that you raise your hand. There are two replies to this. First, my having the stronger obligation does not imply that I do not have the weaker one—on the contrary, my having the weaker follows from my having the stronger—so the example does not impugn O2b. Second, it is doubtful that I do have the stronger obligation. An alternative and plausible view is that the weaker obligation is the only one that, in Gilbert's phrase, is directly generated by my undertaking: if I stop only for some different reason, I thereby and luckily fulfill it. It can be conceded that I am morally at fault in failing to pay adequate attention to that obligation. The fault might perhaps be described in terms of a meta-obligation (which might, in order to avoid potential regresses, be taken to apply to itself) to pay attention to my obligations.
In (b) the obligation is externally conditional and in (a), considered by itself, the obligation is internally conditional; but the obligation in (a) is externally conditional with respect to the protasis of O2. The distinction between externally and internally conditional obligations corresponds to that between externally and internally conditional undertakings.
30 The question whether an obligation is conditional or unconditional is distinct from the question whether or not the obligation motivates. Gilbert blurs the distinction in Is an Agreement an Exchange, supra note 2, at 634, 645.
O1 and O2 concern pro tanto rather than on-balance obligations; the question can be left open whether the obligations are moral. Assume that the type of obligation in the principles is the same as in Gilbert's criterion.
31 In Agreements, Coercion, supra note 2, at 299–300, Gilbert allows that the obligations of agreement are pro tanto and doubts that they are moral; see also Gilbert, Is an Agreement an Exchange, supra note 2, at 631–632, n13; Gilbert, Notes on the Concept of a Social Convention, 14 NEW LITERARY HIST. 79 (1983), reprinted in GILBERT, LIVING TOGETHER (1996); Gilbert, Agreements, Coercion, supra note 2, at 108–109; Gilbert, Walking Together: A Paradigmatic Social Phenomenon, in THE PHILOSOPHY OF THE HUMAN SCIENCES 181 (P. French, et al., eds., 1990), reprinted in GILBERT, LIVING TOGETHER (1996). Raz, supra note 28, at 225, denies that promissory obligations as such are moral.
Gilbert does not explain what she means by “directly generates,” but it appears that M4 meets the criterion so far as Y is concerned: By M4b, Y undertakes that he will do Ay; so, by O1, Y has an obligation to do Ay. X, however, gives only the conditional undertaking in M4a; hence by O2a he has the internally conditional obligation to do Ax if Y will undertake to him that Y will do Ay. But this is not X's performance obligation, which is simply to do Ax. So M4 appears to fail Obligation.
But Gilbert gives no reason for imposing such a stringent test. (It is a weakness of her discussion that she says little about the supposed sources of her proposed criteria; she might have argued, for example, that they somehow flow from the purposes served by the concept of an agreement32
Compare the approach taken to knowledge in E. CRAIG, KNOWLEDGE AND THE STATE OF NATURE ch. 1 (1990), and to truth in B. WILLIAMS, TRUTH AND TRUTHFULNESS (2002).
or by agreements themselves.) Granted that an agreement imposes obligations on each party, a more plausible criterion, in the case of the party giving the conditional undertaking, is that he has the corresponding conditional obligation and that, when the other party gives the unconditional undertaking, he has the performance obligation. O2a ensures that M4 meets the first limb, as just noted, and O2b that it meets the second. Gilbert's criterion is an approximation to this more plausible one, and indeed may be interpreted to amount to the same thing if “the agreement directly generates the relevant performance obligation” is construed broadly enough to cover the second limb. Gilbert probably means something narrower.
The next criterion is Simultaneity: the parties to an agreement acquire their obligations under the agreement at the same time.
33 Gilbert, Is an Agreement an Exchange, supra note 2, at 630; Gilbert, Agreements, Coercion, supra note 2, at 290. Gilbert may—her wording is not clear—have in mind a stronger criterion, that the parties both acquire and lose their obligations at the same time; if so, the following discussion is unaffected. The distinction between lasting for the same period and lasting for adjoining or overlapping periods seems to resolve the conflicts between requirements of simultaneity and succession that Benson finds in contract law; P. Benson, The Unity of Contract Law, in THE THEORY OF CONTRACT LAW (P. Benson, ed., 2001). It thus avoids the need for the obscure solution Benson proposes, that “‘Simultaneity’ must … mean something other than ‘at the same time’” (id. at 148) and that “Value is the unity of temporal succession and simultaneity” (id. at 195).
O1 and O2 say nothing about the times at which the obligations they mention arise, but temporal references can plausibly be introduced
34 On the time of an obligation, see H. Prichard, The Time of an Obligation, in MORAL OBLIGATION (1949).
thus:
These principles can be refined (in particular, to specify the time at which the person is to φ) but they are accurate enough for present purposes. The temporal references in M4 can likewise be made more precise. Using names for times that match those in the principles (so that T1=T3):
Ux precedes Uy, as already noted, so T2 precedes T1. By M5a and O4a X gets an obligation at T2, and by M5b and O3 Y gets an obligation at T1. So M4 fails Simultaneity.
But again the test is too stringent. A more plausible criterion, to which Simultaneity approximates, is that the parties get their performance obligations at the same time. M5 meets this criterion: both X and Y get their performance obligations at T1—Y by M5b and O3, X by M5a, M5b, and O4b. In support of Simultaneity, Gilbert says “In many cases, a given party will have no desire to take on some particular obligation before the other party has taken on a corresponding obligation.”35
Gilbert, Is an Agreement an Exchange, supra note 2, at 630.
However, if X wants to make an agreement with Y, X must be willing to take the steps needed to achieve this—specifically, to do Ux and thereby to acquire a conditional obligation. Since the obligation is only conditional, it is not burdensome, for it does not require X to do anything unless Y does Uy. It might be argued that the conditional obligation entails certain unconditional obligations on X—in particular the obligation not to put himself in such a position that, if Y does Uy, X will be unable to do Ax—and that in some circumstances such obligations will be a burden on X; however, even if this is true, it is not a good enough reason to prefer Simultaneity to the weaker criterion.
Gilbert's third criterion is Interdependence: “if one party defaults on his performance obligation, the other ceases to have his original performance obligation.”36
Gilbert, Is an Agreement an Exchange, supra note 2, at 632; see also Gilbert, Agreements, Coercion, supra note 2, at 291; and compare Fried, supra note 8, at 117. Gilbert, Is an Agreement an Exchange, supra note 2, at 631, and Agreements, Coercion, supra note 2, at 291, proposes a stronger version of Interdependence: if either party defaults, he nullifies the obligations of both parties.
Reverting for simplicity to M4, it appears that the model fails the criterion so far as a default by Y is concerned. By M4a, M4b, and O2b, if X does Ux and Y does Uy, X gets the performance obligation to do Ax: nothing in the model or in the normative principles so far proposed implies that this obligation falls away if Y fails to do Ay. It might be replied that, if Y fails to do Ay, Y thereby ceases to undertake to do Ay,
37 See the discussion of the caducity of promises in Black, Independent Promises and the Rescission of Contracts (2003c).
in which case O2b ceases to generate the performance obligation on X. There is no need to consider whether this reply can be sustained—specifically, by reintroducing temporal references and/or by elaborating the concept of an undertaking (e.g., so as to distinguish between the act and the resultant state)—for it is clear that the model breaches Interdependence so far as default by X is concerned: by O1, if Y does Uy, he gets the performance obligation to do Ay whether or not X does Ax. It makes no difference that, as M4c provides, Y does Uy for the reason that X does Ux.
Gilbert's case for Interdependence is based on our “common understanding”38
Gilbert, Is an Agreement an Exchange, supra note 2, at 630.
of an agreement, but it is doubtful whether we do commonly understand that agreements meet the criterion: a plausible view is that, at least in some cases, if one party defaults on his performance obligation, the performance obligation of the other party persists. This view is supported by English contract law, which provides that a breach justifying rescission
39 The use of “rescind” in this context has been criticized (see Chitty, supra note 3, at 25-046; Treitel, supra note 4, at 674–676), but is a common and convenient shorthand for those remedies that, broadly, absolve the injured party from performance of obligations under the contract that have not accrued. See the cases cited in Black, supra note 37, at n.22.
does not automatically determine a contract
40 Cases are cited in Treitel, supra note 4, at 757, n.9. Where the contract comprises promises that depend on each other, rescission is sometimes but not always available, the main deciding factor being the seriousness of the failure to perform; Chitty supra note 37, at 25-036-045; Treitel, supra note 4, at 681–739. In the unusual case where the promises are independent of each other, neither party is entitled to rescind for failure by the other party to perform: the defaulting party can still enforce the injured party's promise. The distinction between dependent and independent promises is analyzed in Black, supra note 37, which cites cases in notes 2 and 4; it is argued there, however, that a contract comprising independent promises does not amount to an agreement.
and moreover that, where a contract is rescinded, the injured party remains liable to perform obligations that accrued before rescission.
41 Chitty, supra note 37, at 25-001, 25-050; Treitel, supra note 4, at 762–763.
Possible counterexamples are agreements of coordination—as where the parties agree to meet at a certain place and time—for there a breach by one party makes the other's compliance pointless; but these are not enough to sustain Interdependence as a general criterion for the adequacy of a model of agreement. The performance obligations of the parties in M4 are interdependent in a looser sense (to which, however, Interdependence cannot be viewed as an approximation): Y's performance obligation arises from an undertaking which he gives for the reason that X gives an undertaking which, together with Y's undertaking, gives rise to X's performance obligation.
The view that where X fails to do Ax Y remains obliged to do Ay strains our intuitions. The strain is partly eased by recognizing that X's default does not free X from all obligations relating to the agreement: in most cases, he is obliged to make some form of reparation to Y (in the context of the law on rescission, this is a secondary obligation to pay damages). But it may now be objected that X may breach this obligation in turn, but by O1 Y remains obliged to do Ay, and that this is just as hard to accept. A better response to the intuitive resistance is to attribute it, at least in part, to a confusion between those obligations Y has and those it is fair for him to have: it can be granted that here it is unfair for him to be obliged to do Ay, but it does not follow that he is absolved from the obligation. Obligations are often unfair: obvious examples can be found among cases (unlike the present one) in which an obligation arises otherwise than through the will of the obligor, such as certain obligations to look after one's parents. If an intuition still supports Interdependence, I think the intuition should be abandoned.
THE SECOND MODEL
In M4 X's undertaking is conditional on Y's undertaking to do Ay. The second proposed model, which fits some cases better, is a simpler variant in which X's undertaking is conditional on Y's doing Ay,
42 X's undertaking in the first sentence of the paper is of this kind.
thus:
43 The conversation in Radford, supra note 1, at 582, has the structure of (a) and (b). There is some similarity between the causal-explanatory component of (c) and Radford's requirement that the party making the unconditional statement have an appropriate motivation. Compare note 11 above.
M6, like M4, fails to meet the four alleged criteria, and again the right response is to reject the criteria. There are some differences of detail: As to Obligation, the revised criterion which M6 meets, in the case of X, is that he has the corresponding conditional obligation and that, if Y fulfills Y's performance obligation, X gets X's performance obligation. As to Simultaneity, X and Y need not get their performance obligations at the same time, so M6 does not even meet the weaker version of that criterion proposed earlier; by O3, Y gets his performance obligation at T1, and by O4b X gets his at T3, but unlike the case of M5 there is no implication that T1=T3. As to Interdependence, if Y defaults on his performance obligation, nothing in the model or the normative principles implies that X ever gets a performance obligation; hence M6 substantially meets the criterion so far as default by Y is concerned—“substantially” because, if X never gets the obligation, it is inaccurate to say that he ceases to have it. The parties' performance obligations are now interdependent in the looser sense that Y's arises from an undertaking which he gives for the reason that X gives an undertaking which, if Y fulfills Y's obligation, gives rise to X's obligation.
JUSTIFICATIONS FOR MAKING AND COMPLYING WITH AGREEMENTS
M4 and M6 accommodate intuitively plausible justifications for making and complying with agreements. The justifications to be discussed are arguments that would lead a rational agent to make or comply with an agreement; they are not justifications on grounds of morality or public policy for the promotion or enforcement of agreements. The basic idea is that X wants, and believes that doing Ux is the best means of getting, Y to do Ay, and correspondingly for Y. I shall develop the idea for M4 only and begin with the parties' arguments to their respective undertakings. Since M4a embeds Y's undertaking in X's, the approach will be to show how Y reasons to his undertaking and then to show how X, in reasoning to X's undertaking, takes Y's reasoning into account. Ux precedes Uy, as noted above, and the undertakings are in the future tense, but for simplicity I shall cast the arguments in the timeless present: this will not affect their substance.
Y may rationally argue to his undertaking as follows:
3 follows from 2 on the assumption that X's undertaking is reliable, where to say that a person's undertaking that if P he will φ is reliable means that if he gives it and it is the case that P then he will φ. That assumption is in turn underpinned by X's argument 24–28, discussed below, to the conclusion that he will do Ax: it is underpinned by the argument in the sense that the assumption follows from the propositions (i) that if X and Y respectively do Ux and Uy then X carries out the argument and (ii) that the intention expressed by the argument's conclusion is reliable in the sense that if X forms it then he will carry it out.
As to the move from 3 to 4, “If I φ then P” does not by itself entail “My φ-ing is a means for me to secure that P”: possible counterexamples to a supposed entailment are the case where “I φ” is substituted for “P” and the case where there is some sufficient cause of both my φ-ing and its being the case that P but there is no other causal relation and no constitutive relation between my φ-ing and its being the case that P. Nevertheless it seems clear that “If I φ then P” can be supplemented to yield an entailment (which is not to say—what is false—that “My φ-ing is a means for me to secure that P” entails “If I φ then P”). What the supplementary assumptions are will depend on the analysis of the means-end relation, which will not be pursued here; the obvious approach is to analyze it in terms of the relation … causes … and perhaps also the relation … constitutes…. In any event, let the appropriate substitution-instances of the supplementary assumptions—whatever exactly they are—be granted to Y.
6 follows from 4 and 5 on the assumption that the other means mentioned in 5 are commensurable, in respect of their goodness, with Y's doing Uy; there is no need here to amplify the relevant concepts of commensurability and goodness. 7 is required by the principle of means-end reasoning used to derive 8:
—where (a) expresses a desire, (b) and (c) beliefs, and (d) an intention. (c) is needed to exclude such cases as the one where I want that P less than I want not to φ. Without an informative description of the types of “thing” that need to be equal, (c)—and hence PR—is fairly trivial, but that does not matter for the present purpose. I shall simply assume that PR is valid without attempting to explain why or in what sense. Clearly it is not valid in deductive logic: following Broome, one might say that PR's validity is a matter of the “formal” rather than logical relations that hold among (a)–(d);
44 J Broome, Normative Practical Reasoning, PROC. ARISTOTELIAN SOC. 182 (Supp. vol. 75, 2001).
but so far that is only a label. It is plausible to say—adopting another idea of Broome's—that reasoning in accordance with PR is correct in that the mental states expressed by (a) to (c) “normatively require” the intention expressed by (d), in the sense that I ought to see to it that I have that intention if I have that desire and those beliefs.
45 J. Broome, Normative Requirements, in NORMATIVITY (J. Dancy, ed., 2000); Broome, supra note 44. PR is similar to the schema of best-means reasoning in Broome, supra note 12; the main differences are that Broome's first premise expresses an intention whereas mine expresses a desire and, consequently, that PR contains the premise that other things are equal. PR is similar in other respects to the basic schema for practical reasoning in R. AUDI, PRACTICAL REASONING 99 (1989), the main difference being that the conclusion in Audi's schema is normative: “I should φ.” For strictures on the detachment of normative conclusions in means-end reasoning, see Broome, Normative Requirements; Broome, supra note 12; Broome, supra note 44; Broome, Are Intentions Reasons? And How Should We Cope With Incommensurable Values? in PRACTICAL RATIONALITY AND PREFERENCE: ESSAYS FOR DAVID GAUTHIER (C. Morris & A. Ripstein, eds., 2001); Broome, Reasons, in REASON AND VALUE: ESSAYS ON THE MORAL PHILOSOPHY OF JOSEPH RAZ (R. J. Wallace, et al., eds., forthcoming).
X may rationally argue to his undertaking thus:
10 assumes that Y is rational enough to carry out the argument at 1 to 8. 11 makes explicit the fact, already noted, that 8 expresses an intention. The import of 12 is that Y's intention is reliable in the sense specified above, and of 13 that Y's undertaking is reliable in a similar sense: that assumption is underpinned by Y's argument 20–23, discussed next, to the conclusion that he will do Ay; the point here is parallel to the one made above about the assumption that X's undertaking is reliable. 14 follows from 10 to 13 given transitivity of the conditional: this assumption may need to be qualified, depending on the interpretation given to “if.”
46 Unqualified transitivity does not hold for counterfactual conditionals: R. Stalnaker, A Theory of Conditionals, in STUDIES IN LOGICAL THEORY, MONOGRAPH 2, AM. PHIL. Q. 38 (1968), reprinted in CONDITIONALS. (F. Jackson, ed., Oxford 1991); D. LEWIS, COUNTERFACTUALS 32–35 (1973).
The rest of the argument is parallel to 4 to 8.
Once X and Y have given their undertakings, they may rationally argue their way to fulfilling them. Y may argue:
21 follows from 20 by O1. 22 is required by the principle of normative reasoning used to derive 23:
The remarks on PR apply mutatis mutandis to NR.47
Broome's schema of normative practical reasoning in Broome, supra note 44, is similar to NR but uses “ought” instead of “obligation” and does not have the provision that other things are equal. Compare Principle 2 in M. SMITH, THE MORAL PROBLEM 12 (1994).
X may argue:
26 follows by O2b, and 28 by NR.
These arguments show that it is rational to make agreements and, having made them, to keep them.
