¹ William Fitzpatrick notes that the intending-foreseeing distinction may be “indispensable to a proper characterization of terrorism”. See William J. Fitzpatrick, The Intend/Foresee Distinction and the Problem of Closeness, 128 Phil. Stud. 585–617 (2006), at 586.

² For one locus classicus, see Judith J. Thomson, The Trolley Problem, in Rights, Restitution and Risk 176–202 (1986).

³ At least this is the case with the standard Trolley case. Many variants are possible, of course, including variants in which the motivations of the relevant agent are of a much more complicated structure. For discussion, see David Enoch, Ends, Means, Side-Effects, and Beyond: A Comment on the Justification of the Use of Force, 7 Theoretical Inquiries L. 43–57 (2005) and the references there.

⁴ Jonathan Bennett and Frances Kamm do draw this distinction (between the two ways of understanding the intending-foreseeing distinction) explicitly. See Jonathan Bennett, The Act Itself 198 (1995); and Frances M. Kamm, Failures of Just War Theory: Terror, Harm, and Justice, 114 Ethics 650–692 (2004), at 667. If I understand Kamm correctly, she thinks the appropriate moral distinction is closer to the one in terms of causal structures than to the one in terms of mental states, though at one point Kamm seems to be looking for some kind of middle ground between these two ways of understanding the distinction. See Frances M. Kamm, Non-Consequentialism, the Person as an End-In-Itself, and the Significance of Status, 21 Phil. & Pub. Aff. 354–389 (1992), at 376. It is not completely clear whether Warren Quinn's favorite way of understanding the intending-foreseeing distinction—in terms of whether or not someone's involvement is anything to the agent's purpose—falls on the mental-state or the causal-structure side, but it seems to me the former is better in line with what he has to say about the distinction, in particular its rationale. See Warren Quinn, Actions, Intentions and Consequences: The Doctrine of Double Effect, 18 Phil. & Pub. Aff. 334–351 (1989); reprinted in Morality and Action 175–193 (1993).

⁵ Indeed, as one referee insisted, it may even be said that the causal-structure understanding is not an understanding of the intending-foreseeing distinction at all. Of course, nothing hinges on the terminological issue. As long as the two distinctions are clearly distinguished (as they are in the text here and in the arguments below), it does not matter what one calls the causal-structure distinction. I choose the terminology in the text—where the causal-structure distinction is considered one kind of the intending-foreseeing distinction—simply because (as can be seen from the previous note) it is more in line with the literature.

⁶ Discussions of the moral significance of the intending-foreseeing distinction in the literature are often conjoined with discussions of the moral significance of the different but perhaps related doing-allowing distinction. For now let me restrict my discussion to just the intending-foreseeing rather than the doing-allowing distinction.

⁷ Perhaps some examples will help here. It is sometimes suggested that we understand the distinction in dispositional terms; see, e.g., Alison Hills, Intentions, Foreseen Consequences, and the Doctrine of Double Effect, 133 Phil. Stud. 257–283 (2007). But—depending on your view of dispositions and of mental states—such dispositions may be just mental states, and if not, what underlies the dispositions will turn out to be mental states or causal structures, or both. It is also sometimes suggested that the intending-foreseeing distinction is best understood in terms of what reasons the relevant agent acts on. But of course, the notion of acting-for-a-reason is notoriously loaded, and once unpacked it, too, I believe, will come down to mental states, perhaps combined with some features of the relevant causal structures. But I cannot do justice to all this here.

⁸ For the distinction between the conceptual and the normative questions, see also Joseph M. Boyle Jr., Toward Understanding the Principle of Double Effect, 90 Ethics 527–538 (1980); reprinted in The Doctrine of Double Effect 7–20 (2001) (P.A. Woodward ed.), at 13; and Kamm, supra note 4, 114 Ethics 650–692 (2004), at 654.

⁹ See Thomson, supra note 2; see also the references in Frances Kamm, The Doctrine of Triple Effect and Why a Rational Agent Need Not Intend the Means to His End, 74 Proc. AristotelianSoc'y (Supp.) 21–39 (2000).

¹⁰ An anonymous referee suggested that if Loop is symmetrical—if, that is, only if the trolley hits the one will it avoid hitting the five, and also only if it hits the five will it avoid hitting the one, then the permissibility of diverting the trolley in Loop may be rather easily explainable—for in this case the only asymmetrical consideration is that of the number of people on each track. I am not sure this way out actually works, but just to make sure we do not run into problems here, assume that the “extra bit of track” technically allows only one-way trolley traffic, so that unless the trolley is diverted to his track, the one is safe whether or not the trolley hits the five.

¹¹ For a similar point, see Bennett, supra note 4, at 199.

¹² As far as I know, the appropriate-question test does not appear in the literature, but I do not claim originality for the intuition it is meant to capture. For a discussion that forcefully emphasizes intuitions about which differences make a moral difference, see Peter Unger, Living High and Letting Die ch. 4 (1996). For something closely resembling the appropriate-question test, see T.M. Scanlon. Moral Assessment and the Agent's Point of View, 20 Draft for NYU Colloquium on Law, Philosophy, and Political Theory, October 31, 2002, available at http://www.law.nyu.edu/clppt/program2002/readings/scanlon/scanlonnyu02.doc (last visited August 13th 2006).

¹³ There is an important particular instance of causal-structure-questions that may—I am really not sure here—be appropriate. Suppose the agent asks whether other agents will be involved in the bringing about of the bad effect and indeed whether they will be acting impermissibly in being so involved. (I thank Doug Husak for raising this kind of case.) I tend to think that such questions are inappropriate, but I do not find them as inappropriate as the one in the text. Unfortunately, I cannot discuss this kind of case in detail here—doing so would have to involve a discussion of agent-relativity and of the way in which we should take into account in our deliberation the foreseeable immoral behavior of others, both interesting and important topics that deserve more discussion than I can afford here. Let me just note, though, the following three points. First, to my ears such questions sound (at least somewhat) inappropriate. (Consider: “OK, I know how many people will die in each scenario, but how many of them will be killed by me?”) Second, there are many cases where the intending-foreseeing distinction is supposed to apply that do not involve actions by such third parties. Third, such questions sound particularly inappropriate when in the context of state action (which I discuss in the text).

¹⁴ As stated in the text, the appropriate-question test gives reason to doubt the moral significance of the intending-foreseeing distinction when it is understood as a distinction between different causal structures. It is not clear to me whether a version of this test can be applied to the mental-states version of the distinction as well. This is so because it is not clear what sense can be made in this context of the deliberating agent's relevant question about her own mental states (If I foresee that I will intend harm, is it now a case of intention or mere foresight?) and because any answer to this question may very well bring about a change in her relevant mental states. For the somewhat related claim that the deliberating agent has no reason to ask about her own motives, see Robert E. Goodin, Utilitarianism as a PublicPhilosophy 51–57 (1995).

¹⁵ I thank an anonymous referee for pressing me on this and related issues.

¹⁶ For this way of talking, see Larry Alexander, The Jurisdiction of Justice: Two Conceptions of Political Morality, 41 SanDiego L. Rev. 949–966 (2004), at 957.

¹⁷ It may be thought that rights-talk can help here. For perhaps the relevant characterization of the difference between Trolley and Transplant is that in Transplant, saving the five involves violating the right of another person, while in Trolley, no such violation need be involved. (And the question: “But will I be violating any rights?” sounds paradigmatically appropriate.) But this, too, it seems to me, does not help. For now, in the process of filling in the details of such a view, its advocates will have to explain why it is that we have the right relevant to Transplant and not to Trolley, and in this discussion, it seems to me, all the problems for the intending-foreseeing distinctions will resurface.

¹⁸ I thank Larry Alexander for a related point.

¹⁹ For many relevant references here, see Boyle, supra note 8.

²⁰ Note that I do not claim that the Catholic doctrine is unable to cope with purported counterexamples to the moral significance of the intending-foreseeing distinction. Perhaps Catholic doctrine has other resources it can use in order to cope with them; for an argument along these lines, see Joseph Shaw, Intentions and Trolleys, 56 Phil. Q. 63–83 (2006), at 69–70. My topic is not the doctrine of double effect per se but rather the intending-foreseeing distinction. In the text here I put forward a conjecture that relies only on the fact that the two are historically connected.

²¹ To know whether this is so, one would need to know much more about Catholic doctrine than I do. For many more details, see Boyle, supra note 8.

²² See, e.g., Judith J. Thomson, Physician Assisted Suicides: Two Moral Arguments, 109 Ethics 497–518 (1999), at 514–516; and Thomson, Self-Defense, 20 Phil. & Pub. Aff. 283–310 (1991), at 293–294. For similar lines of thought, see James Rachels, More Impertinent Distinctions and a Defense of Active Euthanasia (1994), reprinted in Killing and Letting Die, 139–154 (1994) (B. Steinbock and A. Norcross eds., 2nd ed.); and Scanlon, supra note 12. For the contrary claim, supported by intuitions about specific cases rather than by independent argument, see Shaw, supra note 20, at 81. William Fitzpatrick argues that Thomson in fact argues here against a straw man, as the doctrine of double effect is best understood as claiming that the moral permissibility of an action depends not on the intentions with which it is performed but rather on the intentions with which it can be performed. See William J. Fitzpatrick, Acts, Intentions and Moral Permissibility: In Defense of the Doctrine of Double Effect, 63 Analysis 317–321 (2003). But then it seems Fitzpatrick owes us a story of why it is that some actions can and some cannot be performed with such intentions, and the only available answer seems to me to be one referring to the different causal structures involved. The (possible) intentions thus drop out of the picture of what does the ultimate normative work. So Fitzpatrick really understands the intending-foreseeing distinction in the causal-structure rather than the mental-state way.

²³ I am not sure, but something along these lines may be what Gardner has in mind. See John Gardner, Wrongs and Faults, in Appraising Strict Liability 63–64 (A.P. Simester ed., 2005).

²⁴ For the claim that we should understand intention as including foresight because anything else will be an evasion of responsibility, see Henry Sidgwick, The Methods of Ethics 202 (7th ed., 1981). For a similar point, see Goodin, supra note 14, at 49 n. 5, and the references there.

²⁵ So that if one shoots into a crowd intending just to derive aesthetic pleasure from the sound of the shot but foreseeing (to a practical certainty, almost) that people will be killed, one can be convicted of murder. See Glanville Williams, Textbook of Criminal Law 84–87 (2nd ed., 1983).

²⁶ For discussion of such complex motivational structures, see Kamm, supra note 9; and Enoch, supra note 3. For a somewhat different analysis of such cases, see Shaw, supra note 20.

²⁷ This, I think, is the way to understand Kamm's project, e.g., in her “Doctrine of Triple Effect,” see Kamm, supra note 9.

²⁸ And, of course, nothing like a comprehensive evaluation of this distinction has been supplied here. First, I have not discussed all the versions of this (and related) distinction in the literature; second, I have not discussed attempts at consequentialist proxies of this and related distinctions and the success such attempts can have in somewhat deradicalizing the effect of the moral revolution I mention in the text. I think—but I am not sure—that Alison McIntyre's attempt to rescue the intending-foreseeing distinction (but not the doctrine of double effect) should be understood along such lines. See Alison McIntyre, Doing Away with Double Effect, 111 Ethics 219–255 (2001), at 237–241. And third, I have not discussed any argument for the moral relevance of this and related distinctions. Of course, it is not easy to think of any. The appeal of the distinction relies primarily not on arguments but rather on the intuitive force of specific-case intuitions, and sometimes what is presented as an argument for the distinction is merely—as noted in the text above—a tendentious new name for the problem rather than its solution. Let me very briefly address one recent attempt to give such an argument in Samuel Scheffler, Doing and Allowing, 114 Ethics 215–239 (2004). Scheffler puts forward what may be thought of as a transcendental argument, claiming that a distinction between primary and secondary manifestations of agency (of which the intending-foreseeing distinction may very well be an instance) is needed if we are consistently to hold ourselves and others to be responsible agents. Now, Scheffler fills in this argument with many details which I cannot discuss here, though I should say that I think the argument is flawed in several ways. What I do want to note here is, first, that the distinction Scheffler has in mind is the doing-allowing distinction, and it is not immediately clear that his argument applies (even if it works) to the intending-foreseeing distinction (though putting things in the most generic terms of the distinction between primary and secondary manifestations of agency may include the intending-foreseeing distinction as well). Second, even if the argument works, and furthermore works as an argument for the intending-foreseeing distinction, still it seems to me clear that it does not work in the case of state action, to which I am about to turn (for even if it is in a sense necessary that we hold each other responsible in the sense Scheffler gives these terms, it is not at all necessary that states hold and be held responsible in anything like the same way).

²⁹ In the context of the debate over capital punishment, Cass Sunstein and Adrian Vermeule have recently argued against the application of the act-omission distinction to government action. See Cass R. Sunstein and Adrian Vermeule, The Ethics and Empirics of Capital Punishment: Is Capital Punishment Morally Required? Act, Omissions, and Life-Life Tradeoffs, 58 Stan. L. Rev. 703 (2006). The intuition underlying their account is rather similar to the one I develop and defend in what follows, but it is nevertheless important to note the following differences. First, Sunstein and Vermeule focus on the doing-allowing rather than the intending-foreseeing distinction (though at times they seem to conflate the two; id. at 720–702). As the Trolley and collateral damage examples show, though, the doing-allowing distinction certainly does not capture all we intuitively want here, and something like the intending-foreseeing distinction is needed. Second, the arguments Sunstein and Vermeule present for their claim differ from the ones I am about to develop (though they hint at the points I proceed to make in the text; see, e.g., id. at 707 and 723). And third, they do not seem to be aware of how radical a conclusion it is that they flirt with—unconvincingly, they think that even deontologists with regard to state action can accept their thesis (see, e.g., id. at 707). For the claim that Sunstein and Vermeule miss the intending-foreseeing distinction (referred to as the purposeful-nonpurposeful distinction), see Carol S. Steiker, No, Capital Punishment Is Not Morally Required: Deterrence, Deontology, and the Death Penalty, 58 Stan. L. Rev. 751, 757 (2006). Steiker also complains that against this distinction Sunstein and Vermeule “offer surprisingly little in the way of argument” (id. at 759). For Sunstein's and Vermeule's reply—which more closely resembles some of the points I am about to make—see Cass R. Sunstein and Adrian Vermeule, Deterring Murder: A Reply, 58 Stan. L. Rev. 847, 849–852 (2006). Let me also stress here that nothing in what I say or think commits me to the empirical part of Sunstein's and Vermeule's argument for the death penalty. For a convincing empirical criticism, see John J. Donohue and Justin Wolfers, Uses and Abuses of Empirical Evidence in the Death Penalty Debate, 58 Stan. L. Rev. 791 (2006).

³⁰ For a related point, see Seana V. Shiffrin, Speech, Death, and Double Effect, 78 N.Y.U. L. Rev. 1135–1185 (2003), at 1182. But Shiffrin then proceeds to argue that the doctrine of double effect should be understood as distinguishing between intentions more objectively understood, say in terms of the rationales or reasons for the relevant action, and that thus understood the problem goes away. It is not entirely clear to me what Shiffrin has in mind here when she talks of intentions more objectively understood.

³¹ If, for instance, functionalists are right about mental states, then given some plausible empirical claims about states—that they have decision-making mechanisms, that their decisions can have some coherence over time, etc.—there seems to be no principled difficulty in ascribing mental states to corporations and states. If, on the other hand, having a phenomenology is a necessary condition for having mental states, then assuming, as seems plausible, that there is nothing it is like to be France, France cannot have mental states (unless, that is, phenomenological states are themselves best understood functionally). And there may be other possibilities as well.

³² Immanuel Kant, 4 Groundwork of the Metaphysics of Morals 394 (Cambridge Texts in the History of Philosophy, Mary Gregor ed., 1997), 8 (4:394).

³³ For some discussion along these lines, see, e.g., Philip Pettit, Akrasia, Collective and Individual, in Weakness of Will and Practical Irrationality 68–96 (Sarah Stroud & Christine Tappolet eds., 2003) and the references there. Pettit—here and elsewhere—is strongly sympathetic to the literal attribution of mental states to political bodies.

³⁴ See, e.g., Andrei Marmor, Interpretation and Legal Theory ch. 8 (1992); and Jeremy Waldron, Law and Disagreement ch. 6 (1999).

³⁵ Indeed, it seems to me that often enough when we are willing to attribute an intention to a state, we do this not as a factual matter that can then be used as input in the normative discussion that follows but rather as a conclusion of such a normative debate. In other words, we first decide questions of responsibility and only then proceed to attribute intention. Perhaps this is so—to an extent—even in the case of individuals. But it seems to me to be so to a much greater extent in the case of state intention.

³⁶ I thank Matthew Butterick for pressing me on this point.

³⁷ For a study of the limits and dangers of the personification of government in the context of constitutional law, see Daryl J. Levinson, Personified Government and Constitutional Morality (unpublished manuscript), on file with another (2007).

³⁸ I thank Larry Alexander and Seana Shiffrin for this objection. See also Alexander, supra note 16, at 953; and Elizabeth S. Anderson & Richard H. Pildes, Expressive Theories of Law: A General Restatement, 148 U. Pa. L. Rev. 1503–1575 (1999–2000), at 1519.

³⁹ Anderson & Pildes, supra note 38. I will use this text as my example of an expressive theory here.

⁴⁰ See id. at 1554, 1564.

⁴¹ For a general critique, see Mathew D. Adler, Expressive Theories of Law: A Skeptical Overview, 148 U. Pa. L. Rev. 1363 (1999–2000).

⁴² As Adler repeatedly notes, the understanding of expression Anderson & Pildes work with is amazingly broad, so that “To express an attitude through action is to act on the reasons the attitude gives us”; Anderson & Pildes, supra note 38, at 1510. If this is so, it seems that expression drops out of the picture and everything done with it can be done directly in terms of reasons.

⁴³ This may be true of what Anderson and Pildes have in mind when they say that “expressive norms regulate actions by regulating the acceptable justifications for doing them”; id. at 1511.

⁴⁴ Goodin emphasizes similar points. See. e.g., Goodin, supra note 14, at 10.

⁴⁵ That the committee does not administer the drug itself but rather allows someone else to do so may be of importance here, because it may show that the doing-allowing distinction also has problems in applying to policy decisions and state actions. See Sunstein and Vermeule, supra note 29.

⁴⁶ See Levinson, supra note 37.

⁴⁷ I set aside here questions about the justifiability of a state's special relationship with its own citizens (compared to other individuals).

⁴⁸ It is speculative for two reasons: first, I do not know enough about Mill's and Bentham's thought and writings to be remotely confident in the point I am about to make. And second, even if the historical details in this subsection are right, still alternative explanations of them may be available, explanations that do not support the point I am about to make in the text.

⁴⁹ Bentham and Mill are both well aware, of course, of the conceptual distinction between intending and foreseeing. For Bentham's related distinction between direct and oblique intention, see Jeremy Bentham, The Principles of Morals and Legislation ch. 9 (1988). For Mill's discussion of the distinction, see John Stuart Mill, Comments upon James Mill, Analysis of the Phenomena of the Human Mind, reprinted in 31 Collected Works: John Stuart Mill, 252–253 (John Robson ed., 1981). (I thank Michael Ridge for this reference.) For some relevant discussion, see Michael Ridge, Mill's Intentions and Motives, 14 Utilitas 54–70 (2002).

⁵⁰ Goodin, supra note 14, often makes this point.

⁵¹ Like, for instance, the distinction between the classical examples of a strategic bombing of an ammunition plant during a just war, with the merely foreseen death of some children in a nearby kindergarten, and the terrorist bombing of a kindergarten, with the intended effect of killing some children in order to achieve a demoralizing effect. Another possible example here would be a case precisely analogous to Transplant, only with the relevant agent (the one making the decision) being the state.

⁵² So we should look more diligently for consequentialist proxies and substitutes to this distinction in the case of state action, a point I return to below.

⁵³ It is a central thesis of Goodin, supra note 14, that different moral considerations may guide states and public officials on one side and private individuals on the other. Let me briefly note, though, the following differences between his view and mine. First, Goodin does not explicitly discuss the intending-foreseeing distinction (though he does discuss the role of motives). Second, if I understand him correctly, he assumes a utilitarian first-order morality for all, then applying second-order reasons of a different nature to public and private contexts, so that, for instance, going through the utilitarian calculus is a better idea if you are a public official making a policy-decision than if you are a private person deliberating how to act. I, on the other hand, do not assume a consequentialist (much less utilitarian) underlying morality but rather attempt to evaluate the prospects of a moral theory that attributes to the intending-foreseeing distinction intrinsic moral significance.

⁵⁴ Seana Shiffrin discusses some such interconnections; see Shiffrin, supra note 30.

⁵⁵ In the context of a discussion of some moral issues concerning terrorism, for instance, Frances Kamm devotes a subsection to the possibly different status of states and individuals, but she does not notice that the moral significance of the distinction between intending and foreseeing may depend on the identity of the agent performing the relevant actions. See Frances M. Kamm, Terrorism and Several Moral Distinctions, 12 Legal Theory 19–69 (2006), at 24–25.

⁵⁶ See, e.g., Shiffrin, supra note 30, at 1177–1180 and the references there.

⁵⁷ Shiffrin notices that the distinction between content-based and other restrictions on speech “looks like double effect reasoning”; id. at 1177. For a detailed defense of the claim that “First-Amendment Law is best understood and most readily explained as a kind of motive-hunting,” see Elena Kagan, Private Speech, Public Purpose: The Role of Governmental Motive in First Amendment Doctrine, 63 U. Chi. L. Rev. 413–517 (1996) (the quote is at 414). For the claim that only purposeful restrictions raise First Amendment issues, see Larry Alexander, Is There a Right to Freedom of Speech? ch. 2 (2005). For some related problems, see id., ch. 3. If I understand Alexander correctly, his view is consistent with viewing the purposeful-nonpurposeful distinction here as one that is of only extrinsic significance—see esp. 41. And for a general overview of the question of motive in a constitutional setting, see Lawrence Alexander, Introduction: Motivation and Constitutionality, 15 San Diego L. Rev. 925 (1978).

⁵⁸ See, e.g., Ronald Dworkin, Liberalism, in A Matter of Principle 191 (1985).

⁵⁹ See, e.g., George Sher, Beyond Neutrality: Perfectionism and Politics ch. 2 (1997).

⁶⁰ Steven Wall notices this point; see Steven Wall, Neutrality and Responsibility, 98 J. Phil. 389–410 (2001).

⁶¹ See, e.g., Anthony Duff's critique of Lewis's suggestion of a punishment lottery in Anthony R. Duff, Auctions, Lotteries and the Punishment of Attempts, 9 Law & Phil. 1–37 (1990), at 17–30.

⁶² This is true, I think, already of Coase's presentation of his famous theorem. See Ronald H. Coase, The Problem of Social Cost, 3 J. L. & Econ. 1 (1960).

⁶³ For the claim that the difference between legal luck and plain luck that has legal implications collapses if the intending-foreseeing distinction (and related distinctions) lack intrinsic normative weight when applied to the state, see David Enoch, Luck between Morality, Law and Justice, 9 Theoretical Inquiries L. 23–59 (2007).

⁶⁴ See Kagan, supra note 57, at 507–509.

⁶⁵ This Appendix may be read as fleshing out an intuition briefly expressed by Anscombe, though in the context of discussing absolute deontological constraints. See G.E.M. Anscombe, War and Murder, in Nuclear Weapons: A Catholic Response 45–62 (Walter Stein ed., 1962); reprinted in The Doctrine of Double Effect 247–260 (P.A. Woodward ed.,), supra note 8, at 256.

⁶⁶ I put here to one side deontological options and hybrid views of the kind discussed by Scheffler because they are irrelevant, I think, to our concerns here; see Samuel Scheffler, The Rejection of Consequentialism (1982).

⁶⁷ The discussion here is subject to the point made in note 13: Perhaps the collapse to consequentialism can be blocked if, though causal-structure-facts in general are not morally irrelevant, still causal-structure-facts that involve other agents are. But then, perhaps the argument in the appendix shows a collapse at least to an agent-relative kind of consequentialism. And furthermore, if we add the premise that at least sometimes one should not violate a constraint even in order to minimize violations by one, the argument in the text, I think, stands.

⁶⁸ This result may pose a problem for Scanlon, because he expresses doubts about the intending-foreseeing distinction. See T.M. Scanlon, Intention and Permissibility, 74 Proc. Aristotelian Soc'y (Supp.) 301–317 (2000); and Scanlon, supra note 12.