¹ Deontic constraints are rules that prohibit performing an action even when doing so is the only way to prevent more instances of the action being performed by others. Special duties are rules requiring agents to favour those they have a special relationship with even when not favouring them would result in more special relationship favouring by others. Setiya calls the former ‘general restrictions’ and the latter ‘special restrictions’. However, we prefer to use the more common terms ‘deontic constraints’ and ‘special duties’.

² Setiya, Kieran, Must Consequentialists Kill?, The Journal of Philosophy, 115 (2018), 92–105CrossRefGoogle Scholar.

³ Parfit, Derek, Reasons and Persons (Oxford: Clarendon Press, 1984)Google Scholar.

⁴ We prefer to draw the distinction this way not only because it is easier to grasp than talk of ineliminable indexicals but also because, as McNaughton and Rawling have shown, the indexical approach is flawed (McNaughton, David and Rawling, Piers, Agent-Relativity and the Doing-Happening Distinction, Philosophical Studies, 63 (1991), 167–85CrossRefGoogle Scholar). In certain cases it wrongly classifies agent-neutral rules as agent-relative (whereas Parfit's account gives the correct classification in these cases). Although Setiya's argument does not include any cases that the indexical account misclassifies, we nonetheless prefer to avoid using an account that has this well-known problem.

⁵ Actually, we find something puzzling in Future Neglect. If your neglect is independent of mine then shouldn't it also be possible for neither of us to neglect our child? Yet it is crucial to Setiya's example that this is impossible. Ultimately, we don't think this matters because there is a modified version of this case (a version that has the same structure as Indeterminate Trolley bellow) that isn't puzzling in this way and thus is sufficient to establish Setiya's point. On this basis we say we agree with Setiya here.

⁶ See Setiya, footnote 14.

⁷ For influential accounts of these two kinds of explanations, see Nagel, Thomas, The View from Nowhere (Oxford: Oxford University Press, 1986), pp. 175–85Google Scholar; Kamm, F. M., Harming Some to Save Others, Philosophical Studies, 57 (1989), 227–60CrossRefGoogle Scholar; and Kamm, F. M., Morality, Mortality: Volume II: Rights, Duties, and Status (Oxford: Oxford University Press, 1996)Google Scholar.

⁸ As we are about to argue, the agent-neutral reason the agent has in this case derives from the agent-relative reason they have. If so, the case should not be understood as one where an agent's reason for preferring something is overdetermined.

⁹ A possible exception to this rule is the supererogation cases discussed in Pummer, Theron, Whether and Where to Give, Philosophy and Public Affairs, 44 (2016), 77–95CrossRefGoogle Scholar, and Horton, Joe, The All or Nothing Problem, Journal of Philosophy, 114 (2017), 94–107CrossRefGoogle Scholar. However, the cases that we discuss here (and those discussed by Setiya) are not cases of supererogation with the relevant structure. Indeed, they are all cases where the link between what an agent is required to do and what others should morally prefer that she does is highly plausible.

¹⁰ We would like to thank Andrew Forcehimes and an anonymous referee from this journal for helpful comments on an earlier version of this article.