¹ Any of these claims regarding the content of common-sense morality may be challenged, of course, but most of the authors participating in the debate over the concept of a morally relevant harm seem willing to concede claims of the sort made here.

² Scanlon, T. M., What We Owe to Each Other (Cambridge, MA, 1998), pp. 239–40Google Scholar.

³ Norcross, Alastair, ‘Contractualism and Aggregation’, Social Theory and Practice 28.2 (2002), pp. 303–14Google Scholar. Other philosophers who have commented on Scanlon's remark concerning morally relevant harms, and the potential problems such a concept gives rise to, include Kamm, F. M., ‘Owing, Justifying, and Rejecting’, Mind 111 (2002) pp. 351–3Google Scholar; Parfit, Derek, ‘Justifiability to Each Person’, Ratio 16 (December 2003), pp. 384–8Google Scholar; Suikkanen, Jussi, ‘What We Owe to Many’, Social Theory and Practice 30.4 (2004), pp. 499–501Google Scholar; and Sophia Reibetanz, ‘Contractualism and Aggregation’, Ethics 108 (January 1998), pp. 309–11.

⁴ Scanlon, What We Owe, p. 240. For a more nuanced view of when, and in what ways, a less serious harm may still be morally relevant to a greater one, see Kamm, F. M., Morality, Mortality, vol. 1 (New York, 1993), p. 171Google Scholar.

⁵ Norcorss, ‘Contractualism and Aggregation’, p. 307.

⁶ Of course, there might be more than two categories of harm, so that M and Z would be in different categories, and not relevant to one another. But the kind of problem noted in the text will arise in any case where one category contains three or more harms. So for example, even if category 2 contained only the harms M, N and O, M and O would be relevant to one another, while M and L would not be, despite the fact that there is less difference in harm between M and L than there is between M and O.

⁷ This sketch of a two-level conception of practical reason obviously owes a great deal to the work of Joseph Raz; see, for instance, Raz, Practical Reason and Norms, 2nd edn. (Princeton, 1990).

⁸ As Norcross remarks, ‘contractualism departs from consequentialism in its account of precisely which reasons can balance which other reasons. For the consequentialist, any kind of morally relevant reasons can, in principle, balance any other kind. [Note: This should read “any kind of moral reason”, not “any kind of morally relevant reason.”] Scanlon, on the other hand, claims that some moral reasons are simply not “relevant” to other moral reasons’ (Norcross, ‘Contractualism and Aggregation’, p. 305). The two accounts of reasons and of practical reasoning described in the text reflect this difference between consequentialism and contractualism (and many other non-consequentialist theories).

⁹ Scanlon, What We Owe, p. 51.

¹⁰ For example, it might be that creatures that employ the two-level theory of practical reasoning do better at maximizing utility in the long term than they would if they always took into account all of the harms they could prevent.

¹¹ Rawls, John, Political Liberalism (New York, 1993), pp. 54–6Google Scholar. For discussion, see Lefkowitz, David, ‘A Contractualist Defense of Democratic Authority’, Ratio Juris 18.3 (2005), pp. 349–50Google Scholar.

¹² Lefkowitz, ‘Contractualist Defense’.

¹³ For a defense of this claim, see Lefkowitz, David, ‘On a Moral Right to Civil Disobedience’, Ethics 117 (2007), pp. 228–32CrossRefGoogle Scholar.

¹⁴ At this point, the dedicated utilitarian may claim that it is at least reasonable to hold that a minor, temporary, headache falls within the orbit of death. Not surprisingly, I disagree. Nevertheless, I do not intend the point about reasonable disagreement raised in the text to rule out the utilitarian's claim by stipulation. Rather, I seek only to add a bit more detail to the conception of a relevant reason defended in this article, and in particular to show that it can accommodate different judgments regarding the relevance of two harms. That it can do so contributes to its plausibility. Whether a non-consequentialist moral theory that includes the orbital conception of a relevant harm ultimately proves to be superior to a consequentialist one that does not (or that treats headaches and death as relevant to one another) requires a holistic assessment of the two competing theories, which I do not attempt to offer here.

¹⁵ Similarly, there may be clear cases of being bald and clear cases of being not bald, and cases in which there simply is no answer to the question ‘is this person bald?’.

¹⁶ Like Scanlon, Norcross and the other discussants listed in n. 3, I focus here on rescue cases, in which the probability of successful rescue is the same for all parties. That the possibility of preventing a death excludes preventing any number of headaches in this case need not entail that just institutions for allocating resources for the prevention of harm will focus solely on the prevention of death caused in various ways, regardless of the likelihood of someone dying in these ways, or the likelihood of successfully preventing it, with no attention paid to preventing headaches (or broken arms).

¹⁷ This conclusion oversimplifies Scanlon's position, in that it suggests a straightforward preference for saving the many over the few when choosing amongst relevant harms, whereas Scanlon makes only the weaker, vague, claim that in some cases of choosing amongst relevant harms, it is permissible (or perhaps even obligatory) to save the many rather than the few. Exactly what principle we ought to employ in determining which of two relevant harms we ought to prevent, when we cannot prevent both, is a crucial question, but one I cannot address here.

¹⁸ Norcross does not note this difficulty, instead simply stipulating that one of these options is obligatory. Other commentators on Scanlon's discussion of aggregation have done so, however. See, for example, Pogge, Thomas W., ‘What We Can Reasonably Reject’, Philosophical Issues 11 (2001), pp. 118–47CrossRefGoogle Scholar; Parfit, ‘Justifiability’, p. 384.

¹⁹ Norcross, ‘Contractualism and Aggregation’, p. 309. Norcross does not discuss any particular basis, such as the orbital conception of morally relevant harm, for treating one of the three options (1, 2 or 3) as morally obligatory, but instead simply stipulates that one of three options is required. If successful, however, his objection holds regardless of the justification offered for treating one of these three options as obligatory.

²⁰ Norcross, ‘Contractualism and Aggregation’, p. 309.

²¹ Indeed, Joseph Raz claims that ‘Scanlon implicitly acknowledges that contractualism, while presupposing such a classification [of the moral importance (i.e. relevance) of different harms], contributes nothing to establish it’ (Raz, ‘Numbers, With and Without Contractualism’, Ratio 16 (December 2003), pp. 364–5). Scanlon disputes this claim, but also notes that ‘the challenge for contractualism is to overcome the objection that it rules out plausible aggregative arguments’, and that the arguments he offers to meet this challenge need not be uniquely contractualist (Scanlon, ‘Replies’, Ratio 16 (December 2003), p. 431).

²² Scanlon, What We Owe, p. 234.

²³ I wish to thank Scott James for helpful discussion of the ideas in this article, and the University of North Carolina, Greensboro, for the award of a Summer Excellence Grant, which was instrumental to the completion of this article.