Skip to main content Accessibility help


  • Emily Sherwin (a1)


One key premise in Shapiro's book Legality is that rationality requires those who have accepted the master plan for a system of law to obey the system's rules. In this paper, I question this premise, arguing instead that although it may be rational for agents to commit to follow the system's rule in all (or most) cases to which they apply, it is not rational for agents to follow the rules in fact when the rules appear to require the wrong outcomes in particular cases. My argument is based largely on epistemic responsibility, which I view as an element of rationality.



Hide All

1. Scott J. Shapiro, Legality (2011). In characterizing laws as plans, Shapiro draws on Michael Bratman's work on the function of planning in practical reason. See Michael E. Bratman, Intention, Plans, and Practical Reason (1987).

2. Shapiro's master plan appears to be a set of “fundamental legal rules” that authorizes some members of the community to plan for others. Id. at 180–183.

3. Id. at 184–188.

4. Id. at 213–217.

5. Id. at 187–188.

6. Id. at 355.

7. Id. at 183.

8. Larry Alexander & Emily Sherwin, The Rule of Rules: Morality, Rules, and the Dilemmas of Law (2001).

9. Id. at 53–95.

10. For a discussion of the relationship between instrumental rationality and epistemic rationality, see, e.g., Kelly, Thomas, Evidence and Normativity: Reply to Leite, 75 Philosophy & Phenomenological Research 165174 (2007). For an argument that instrumental rationality does not exhaust the concept of rationality and indeed is not a distinctive form of rationality at all, see Raz, Joseph, The Myth of Instrumental Rationality, 1 J. Ethics & Soc. Phil. 1 (2005). Raz is not specifically concerned with epistemic standards.

11. Alexander & Sherwin, supra note 8, at 9.

12. Shapiro, supra note 1, at 170. Cf. David Hume, An Enquiry Concerning the Principles of Morals (1751), pt. 1, sec. 3; David Hume, 3 A Treatise of Human Nature (1739–1740), pt. 2, sec. 2.

13. Alexander & Sherwin, supra note 8, at 11–12.

14. Shapiro, supra note 1, at 170.

15. Alexander & Sherwin, supra note 8, at 15–17.

16. Shapiro, supra note 1, at 36–40, 180–181, 213.

17. Alexander & Sherwin, supra note 8, at 14.

18. Shapiro, supra note 1, at 150–153.

19. Continuing the parallel, we title our book “The Rule of Rules;” Shapiro refers in his concluding chapter to “The Rule of Plans.” Alexander & Sherwin, supra note 8; Shapiro, supra note 1, at 392.

20. Alexander & Sherwin, supra note 8, at 32–33.

21. Shapiro, supra note 1, at 275.

22. Alexander & Sherwin, supra note 8, at 32–34.

23. Id. at 17–19, 37–49; Shapiro, supra note 1, at 165–168; 205–211. See H.L.A. Hart, The Concept of Law (1964), at 91–94. Hart suggests that secondary rules of adjudication and change are part of what identifies law as law. For Alexander and me, secondary rules are desirable, but only the rule of recognition is necessary to accomplish settlement.

24. I use the term “error” here rather than “moral error” to avoid questions about the scope of morality. If pressed, I would define morality broadly and describe the effects of discord and controversy about moral questions as forms of moral error.

25. Shapiro, supra note 1, at 178–188. Shapiro does not provide extended discussion of the binding effect of plans in Legality, but the assumption that plans can and do control future actions lies at the heart of his planning theory of law. Earlier work lays out the arguments for this assumption at greater length. See Shapiro, Scott, The Difference That Rules Make, in Analyzing Law: New Essays in Legal Theory (Bix, Brian ed., 1998); McClennen, Edward F. & Shapiro, Scott, Rule-Guided Behavior, in 3 New Palgrave Dictionary of Economics and the Law 363 (Newman, Peter ed., 1998).

26. See Shapiro, Legality, supra note 1, at 331–387. Alexander and I also differ from Shapiro in our approach to interpretation. We believe that if the reason for having rules is to settle controversy, it must follow that the meaning of rules depends on the rule-maker's intent. When members of a community select an authority to settle controversy by means of rules, what they want is the set of rules that person chooses, after deliberation, to promulgate. The task of interpreters, therefore, is to determine what meaning the rule-making authority intended the words of the rule to have—not the rule-maker's purpose, not the standard meaning or best understanding of the rules’ text, but the intended meaning of the rules. This is the only methodology we recognize as “interpretation.” Intended meaning may not always be accessible; in fact, it may not always exist (as in the case of multimember authorities). When intent is not available, those who apply law can proceed in a number of ways, but none of these counts as interpretation. See Alexander & Sherwin, supra note 8, at 97–122; Larry Alexander & Emily Sherwin, Demystifying Legal Reasoning (2008), at 129–232. Given the settlement function Shapiro associates with plans, we would expect him to take a similar approach. Yet Shapiro equivocates. He appears to hold that a variety of so-called “interpretive” methods may be sound and that which of these a particular interpreter ought to use depends on how much trust the community places in the interpreter. But this is a confusion. In a system founded on settlement, the meaning of a determinate rule (or plan) must be the meaning that the person entrusted with settlement intended the rule to have; otherwise settlement does not occur at the level of planning but instead occurs at some point further down the line. If we are correct about this, then the community's trust in the interpreter is irrelevant. Trust is reposed in planners, not interpreters. In some cases, the legal norm may a relatively indeterminate norm—a standard or a plan that lays out standards for subsequent planning. In that case, the community may be trusting the person to whom the norm is addressed, but it is trusting that person to engage in further planning rather than to interpret the norm. In other words, interpretation ends when the addressee concludes that the norm in question is a standard. At that point, the norm is a delegation of authority, and the addressee must engage in first-order practical reasoning within the bounds set by the norm. In Shapiro's terms, the community is trusting the addressee to engage in subplanning. But subplanning and the trust reposed in subplanners have nothing to do with interpretation. Thus interpretation is always an attempt to recover the authorially intended meaning. If the legal planner chooses to settle matters by means of a determinate norm, then the objective is to determine what settlement the planner intended to impose. When the planner chooses a standard, the question is still one of intended meaning. In this case, the planner intends to delegate settlement to others. Once the delegatees discern this intention, they have no further interpretive role; their role is to plan.

27. Frederick Schauer, Playing by the Rules: A Philosophical Examination of Rule-Bound Decision-Making in Law and in Life (1991), at 94–100. McClennen and Shapiro refer to defenses of this type as “compatibilist” defenses: when the value of rules is property taken into account, the agent's reasons for action will be compatible with obedience to the rule. McClennen & Shapiro, supra note 25, at 364–365.

28. Alexander & Sherwin, Rule of Rules, supra note 8, at 68. We also examine and reject a variety of other putative solutions to the problem of the gap, including the possibility of sanctions. Id. at 68–95.

29. See supra note 25 and accompanying text.

30. Shapiro, Difference, supra note 25, at 39, 44.

31. Shapiro, Legality, supra note 1, at 181, 183; Shapiro, Difference, supra note 25, at 39. For a skeptical discussion of instrumental rationality, see Raz, supra note 10. Raz takes the position that rationality means response to reasons. An agent's ends may determine which reasons are reasons for that agent, but if the agent's ends lack value, they are not reasons in themselves. Raz's discussion implies that so-called instrumental rationality is best understood as the process of choosing the means for complying with reasons; as such, it is not an independent, self-justifying form of rationality but only one part of the overall process of deliberating about reasons.

32. Shapiro, Difference, supra note 25, at 39.

33. Id. at 47.

34. Id. at 47–52.

35. The scheme of the argument appears to be this:

  1. (1)

    (1) An action is not rationally feasible unless it is based on a reason.

  2. (2)

    (2) An agent's commitment to a plan of action exhausts the force of the deliberative reasons on which the decision to commit was based and replaces them with an implementation reason to follow the plan.

  3. (3)

    (3) Therefore the deliberative reasons on which the agent's decision to commit was based are no longer available to the agent as reasons for action.

  4. (4)

    (4) Therefore defection from the plan cannot be based on deliberative reasons pertaining to commitment to the plan.

  5. (5)

    (5) Therefore defection is not rationally feasible.

36. The term “causal” is Shapiro's. Id. at 40. In a discussion of the reasons generated by promises, Joseph Raz make the point that the desirability of a normative power to create reasons for action cannot prove the existence of the power. Joseph Raz, Is There a Reason to Keep a Promise? Columbia Public Law Research Paper No.12-320; Oxford Legal Studies Research Paper No. 62 (2012), available at

37. Although Shapiro does not qualify the statements he makes about the binding effect of legal plans in Legality, he suggests in earlier work that psychological mechanisms, such as repression of ordinary emotional responses to error, may be needed to make rule-following possible in fact. See Shapiro, Difference, supra note 25, at 52–54.

38. Shapiro, Legality, supra note 1, at 124.

39. Id. at 202. Shapiro refers here to the necessity defense in criminal law.

40. More accurately, we do not believe in commitment per se. Commitment can have moral implications if it appears that others have relied on the agent's commitment. But in this case reliance, not commitment itself, counts as a reason to carry through on the commitment. See infra notes 49–51 and accompanying text.

41. In previous work, Alexander and I use a similar library example to illustrate points about interpretation. Alexander & Sherwin, Rule of Rules, supra note 8, at 114–116.

42. A further problem is that the overall justification for the rule is probably best determined by officials with authority to overrule the rule rather than by individual agents considering how to respond to particular cases. Agents may be relatively poorly informed and distracted by the facts of the cases they face, and overruling has coordination advantages.

43. The exception for compelling reasons might also be understood along lines suggested by Joseph Raz. Raz depicts legal rules as sources of “exclusionary reasons” for action—second-order reasons that exclude from deliberation the set of first-order reasons on which the rules depend. He indicates, however, that the exclusionary reason provided by an authoritative rule excludes only those first-order rules on which the rule-making authority relied in issuing the rule. In effect, reasons on which the authority relied are merged into the rule, but other reasons may still be in play. By analogy, Shapiro's “compelling reasons” might be those that the lawmaker did not consult in deciding to issue the rule. This, too, seems too narrow an exception for Shapiro's purposes; and in any event, how it would apply in practice is something of a mystery. See Joseph Raz, The Authority of Law (1979), at 16–19, 22–23, 30–33; Joseph Raz, The Morality of Freedom (1986), at 57–62; Raz, Joseph, Authority & Justification, 14 Phil. & Pub. Aff. 3 (1985) (giving the example of a rule based solely on economic considerations).

44. This is the position Alexander and I advocate. See note 26, supra; Alexander & Sherwin, Rule of Rules, supra note 8, at 96–122. What we mean by intent is not the purpose for which the lawmaker issued the rule but the meaning the lawmaker intended the words of the rule to have. What meaning the lawmaker intended is, of course, a complicated question, but in the case of emergencies, the interpretive problem is not too difficult.

45. This roughly tracks the approach Frederick Schauer defends—an approach Schauer refers to as “presumptive positivism.” Schauer, supra note 27, at 202–205.

46. See Greco, supra note 10 (suggesting that the appropriate standard of “epistemic probability” for purposes of rational choice should reflect the evidence actually available to the agent but should exclude idiosyncratic conclusions the agent may draw from that evidence).

47. I am setting aside questions about how reasons for action can actually cause actors to act. See, generally, Wallace, Jay R., Practical Reason, in The Stanford Encyclopedia of Philosophy 8–10 (Zalta, Edward N. ed., 2009), available at

48. Shapiro uses this term in his extended essay on the subject. See Shapiro, Difference, supra note 25, at 39.

49. For discussion of content-independent reasons for action, see H.L.A. Hart, Essays on Bentham (1982); Raz, Is There a Reason, supra note 36.

50. For an argument against the existence of content-independents reasons, see Marwick, P., Law and Content-Independent Reasons, 20 Oxford J. Legal Stud. 579 (2000). Raz makes the point that a reason with no “strength-determining factors” cannot in itself be a reason for action even if the existence of the reason would have significant benefits. Raz, Is There a Reason, supra note 36, at 14, 16. In the case of promising, Raz locates strength determinants in the value of the normative power to make particular binding commitments; id. at 25–27.

51. If the commitment takes form of a promise to a third person, the promise may lead the promisee to take action in reliance or may have other consequences that create reasons to perform. Reasons of this kind, however, depend on the effects of the promise, not on the promise itself. See Raz, supra note 36, at 7.

52. This interpretation avoids any questions about whether standards of rationality should distinguish between good and bad reasons for action. A plan based on bad deliberative reasons generates a bad implementation reason, which is no reason at all.

53. See, generally, Steup, Matthias, Epistemology, in The Stanford Encyclopedia of Philosophy 34, 8–9 (Zalta, Edward N. ed., 2013), available at

54. Kelly, Thomas, The Rationality of Belief and Some Other Propositional Attitudes, 110 Phil. Stud. 163 (2002), at 171–177.

55. Id. at 171.

56. Id. at 165.

57. This is similar to Gregory Kavka's toxin puzzle, in which a billionaire offers to pay you $1 million if you intend tonight to drink a vial of toxin tomorrow, adding that you do not have to drink the toxin in order to win the money. Can you form the necessary intention? Kavka, Gregory, The Toxin Puzzle, 43 Analysis 33 (1983). Bratman, defending a theory of intentions similar to Shapiro's model of constraint, deflects this example by arguing roughly that it is rational to induce oneself to intend tonight to drink tomorrow by some artificial means (such as self-hypnosis). But the ability of an agent to form an intention and resist reconsideration is limited to intentions formed deliberatively; therefore the agent remains free to reconsider an intention formed by nondeliberative means. Therefore Bratman does not have to concede that is rational to drink the toxin. See Bratman, supra note 1, at 101–106. This argument, however, elides the question whether it is possible to form an intention to drink deliberatively (without the aid of hypnosis).

* Thanks to Larry Alexander and to participants in a faculty workshop at Cornell for helpful comments.


Full text views

Total number of HTML views: 0
Total number of PDF views: 0 *
Loading metrics...

Abstract views

Total abstract views: 0 *
Loading metrics...

* Views captured on Cambridge Core between <date>. This data will be updated every 24 hours.

Usage data cannot currently be displayed