What should you do when you’re uncertain about your options’ outcomes? Many moral theories struggle to answer this question in a way that’s compatible with two compelling principles.

Objective-Subjective Link If you know that an option is objectively permissible, then it’s subjectively permissible.

Ex Ante Pareto If an option gives everyone affected a worse prospect than an available alternative, and all else is equal, then it’s subjectively impermissible.

Why accept Objective-Subjective Link? To a first approximation, an option is objectively permissible in a given state of the world when it’s permissible given the facts in that state; an option is subjectively permissible when it’s permissible given your uncertainty. So, if you know that an option is objectively permissible, then—even if you’re uncertain about the state of the world—you know that it’s permissible given the actual facts, whatever they turn out to be. Surely, it seems, such an option is subjectively permissible.

Why accept Ex Ante Pareto? A prospect is a set of probability-outcome pairs: for example, a half chance of a particular benefit and a half chance of a particular harm. Assuming the relevant probabilities represent your uncertainty, if an option gives someone a worse prospect, it seems to you worse for that person. So, if an option gives everyone affected a worse prospect, it seems to you worse for everyone affected. Surely, it seems, when all else is equal—that is, at least, when the option affects only people who don’t deserve to be worse-off—such an option is subjectively impermissible.

A simple case illustrates the appeal of both principles. Suppose that a stranger has a painful medical condition, and you have an extra dose of medicine: if you give it to the stranger, there’s a half chance that they’ll be cured and a half chance that the medicine will be ineffective but harmless. When all else is equal, giving the medicine to the stranger seems subjectively required. Together, Objective-Subjective Link and Ex Ante Pareto explain why. Assuming you know that giving the medicine is objectively permissible regardless of whether it cures the stranger or is ineffective but harmless, you know that giving the medicine is objectively permissible whatever the actual facts turn out to be. So, by Objective-Subjective Link, giving the medicine is subjectively permissible. In addition, not giving the medicine gives the stranger a worse prospect than giving the medicine, since suffering a painful condition with certainty is surely worse than a half chance of being cured. So, by Ex Ante Pareto, not giving the medicine is subjectively impermissible. So, giving the medicine is uniquely subjectively permissible and so subjectively required.

Unfortunately, however, although Objective-Subjective Link and Ex Ante Pareto are compatible in this case, many moral theories generate apparent conflicts between these principles in other cases. As others have shown, such conflicts arise for any moral theory that includes commitments to (inter alia) prioritarianism, contractualism, constraints, limited aggregation, or risk-aversion.1 While some maintain that these apparent conflicts aren’t as problematic as they seem, others conclude that we should reject these commitments altogether. Indeed, since Harsanyi, principles like Objective-Subjective Link and Ex Ante Pareto have played a central role in influential arguments for simple consequentialist theories like utilitarianism.2

In this paper, I show that yet another type of moral theory generates apparent conflicts between Objective-Subjective Link and Ex Ante Pareto: namely, moderate moral theories—that is, those that include agent-centered prerogatives and so imply that it’s sometimes permissible, but not required, to favor or sacrifice one’s own interests even when doing so is impartially worse.3 This is significant for several reasons.

First, prerogatives are central to commonsense morality. We all sometimes give more or less weight to our own interests than to others’. This basic aspect of moral thought and practice regulates our willingness to sacrifice for others, the forms of behavior that we regard as supererogatory, and the expectations of beneficence, courage, and mercy that we regard as reasonable. It’s striking, then, that such a central aspect of commonsense morality seems to force such a difficult choice between such compelling principles. Indeed, given just how compelling these principles are, it may seem that applying prerogatives under risk places prerogatives themselves—and so commonsense morality itself—under risk. Second, prerogatives not only seem to generate particularly clear conflicts between Objective-Subjective Link and Ex Ante Pareto, but they also do so independently of any similar conflicts generated by other commitments. Since it’s an open question whether all of these conflicts should be addressed similarly, it’s worth considering how prerogatives’ distinctive structure and motivations bear on the distinctive conflicts that they seem to generate. In this way, prerogatives provide an overlooked test case that stands to shed new light on the general tension between principles like Objective-Subjective Link and Ex Ante Pareto. Finally, examining the apparent conflicts that prerogatives generate between these principles is a key step in developing our understanding of how prerogatives apply under risk—an issue that moral moderates have largely neglected.4

After I show that prerogatives generate apparent conflicts between Objective-Subjective Link and Ex Ante Pareto, my central aim is to determine whether these conflicts pose a serious problem for moderate theories. My central claim is that these conflicts don’t pose a serious problem because they’re merely apparent: the most promising moderate theories, I’ll argue, understand one’s prerogatives to depend in a particular way on one’s uncertainty such that they’re compatible with both principles. This has two important upshots, one negative and one positive. The negative upshot is that principles like Objective-Subjective Link and Ex Ante Pareto don’t support rejecting moderate theories in favor of immoderate alternatives like utilitarianism. The positive upshot is a better understanding of how prerogatives apply under risk—in particular, of the relevant aspects of prerogatives’ structure, extent, and foundational motivations.

Here’s the plan. In Section II, I explain three assumptions. In Section III, I explain why prerogatives generate apparent conflicts between Objective-Subjective Link and Ex Ante Pareto. In Sections IV and V, I develop the two most promising moderate responses to these conflicts, one of which reflects what I call “ex ante” moderate theories while the other reflects what I call “ex post” moderate theories. In Sections VI and VII, I consider several arguments against each of these responses, finding them unconvincing. In Section VIII, I use the foregoing to provide an abductive argument for my favored ex ante moderate theories that satisfy both Objective-Subjective Link and Ex Ante Pareto, and I consider the implications that this has for the overall tenability of a moderate view of ethics.

In what follows, I rely on three assumptions that I believe moral moderates would accept.

First, I assume that it’s sometimes permissible, but not required, to benefit from a scarce resource when one could instead give that resource to a stranger who would benefit from it more. To illustrate, consider:

Medicine 1 You’re about to receive a dose of medicine. Without it, you’ll suffer a mildly painful medical condition. A stranger needs a dose of the same medicine to prevent a more severe condition, and yours is the only dose available. You have a button: pressing it gives the dose to the stranger; not pressing it allows the dose to be administered to you. You know everything just described.

Table 1 summarizes this case. “0” represents receiving the dose and so being cured, “–1” represents suffering the mild condition, and “–3” represents suffering the severe condition. I explain this choice of numbers in a moment.

I assume that moderates would agree that, in some version of this case, each option is permissible. You have a self-favoring prerogative to allow yourself to receive the dose, even though doing so makes things go impartially worse.

Second, I assume that it’s sometimes permissible, but not required, to use a scarce resource to benefit oneself when one could instead allow a stranger to use that same resource to provide an equal benefit to themself. To illustrate, consider:

Medicine 2 Again, you and a stranger each need a dose of medicine, and only one dose is available. This time, however, you both have a moderately painful medical condition, and the dose isn’t about to be given to either of you (or anyone else). You have a button: pressing it claims the dose for yourself; not pressing it allows the stranger to claim the dose for themself. You know everything just described.

Table 2 summarizes this case. “0” again represents being cured while “–2” represents suffering the moderately painful condition. Accordingly, the numbers that I use are assumed to represent (only) ordinal comparisons: being cured (0) is better than suffering the mild condition (–1), which is better than suffering the moderate condition (–2), which is better than suffering the severe condition (–3).

I assume that moderates would agree that, in some version of this case, each option is permissible. Some would say that each option is permissible simply because their outcomes are equally impersonally good. Others would say that a self-favoring prerogative is needed to justify pressing because it prevents the stranger from being cured, rather than merely allowing them to suffer. Still others would say that pressing is permissible only if you own the dose. I remain neutral between these views. Notice, though, that because pressing doesn’t involve doing harm to the stranger, I’m not assuming that there are prerogatives to do harm.

Medicine 1 and Medicine 2 differ with respect to the severity of your and the stranger’s conditions, and with respect to which option benefits you and which benefits the stranger. These differences help to construct a case, described in the next section, in which prerogatives generate an apparent conflict between Objective-Subjective Link and Ex Ante Pareto. Of course, though, I focus on these cases merely for the sake of illustration; other cases would work just as well.5

Third, I assume that moderate theories must provide an account of subjective permissibility, not just an account of objective permissibility. To illustrate, consider a “Jackson” case: the stranger has a severe medical condition; you don’t know whether medicine A will cure them fully while medicine B will kill them, or vice versa, but you know that medicine C is guaranteed to cure them almost-fully.6 It’s natural to say that giving C is subjectively required, even though you know that it’s objectively impermissible, since you know that either giving A or giving B is objectively required. Admittedly, it’s controversial how exactly to characterize this distinction. But in the simple cases with which I’m concerned, it’s uncontroversial that objective permissibility in a given state depends only on the options’ outcomes in that state, while subjective permissibility depends on the options’ outcomes in all epistemically possible states.7

There’s one subtlety to flag. What’s objectively permissible isn’t always what would be permissible under full information because one’s uncertainty sometimes affects the morally-relevant features of one’s options’ outcomes—and so affects which outcomes one’s options might have, assuming that outcomes are individuated by their morally-relevant features. This happens, for example, when unexpected pleasures or pains are more intense. In general, as Broome says, some outcomes’ morally-relevant features include “a property with modal elements.”8 In such cases, outcomes in different states “interact”: for example, the pleasure in a low-probability state is made more intense by the boredom in a high-probability state. However, when one’s options would have the same outcomes as the options in a case of full information, the morally-relevant features of one’s options’ outcomes are the same as they would be under full information, and so objective permissibility indeed coincides with permissibility under full information.

Consider:

Medicine 3 Either you’ve just been told that you’re in a situation like Medicine 1 (your condition is mild, the stranger’s condition is severe, pressing gives the dose to the stranger—call this possibility “M1”), or you’ve just been told that you’re in a situation like Medicine 2 (you and the stranger have the moderate condition, pressing claims the dose for yourself—call this possibility “M2”). Unfortunately, due to your medical condition, you can’t remember which. Each possibility is equally likely. You know everything just described.9

Table 3 summarizes this case.

What should moderates think is subjectively permissible in this case?

Either you’re in a situation like Medicine 1 or Medicine 2. In each of those cases, moderates think that each option is permissible. The only difference in Medicine 3 is that you’re uncertain about the actual facts, but that doesn’t seem to affect the options’ outcomes. For example, it doesn’t—let’s assume—make your or the stranger’s suffering more or less intense. So, it seems that moderates should think that each option in Medicine 3 is objectively permissible in each possible state, M1 and M2. (Remember: objective permissibility coincides with permissibility under full information when the relevant outcomes are the same.) But since you know that you’re either in M1 or M2, presumably you know that not pressing is objectively permissible. So, by Objective-Subjective Link, each option is subjectively permissible. Call this the argument from Objective-Subjective Link.

The catch, of course, is that not pressing gives both you and the stranger a worse prospect than pressing. Each option gives you and the stranger a half chance of being cured, but not pressing gives both of you a half chance of suffering a more painful condition: it gives you a half chance of suffering the moderate rather than the mild condition; it gives the stranger a half chance of suffering the severe rather than the moderate condition. Surely, a half chance of a more painful medical condition makes for a worse prospect. Since all else is equal, Ex Ante Pareto implies that not pressing is subjectively impermissible. Call this the argument from Ex Ante Pareto.10

These arguments cannot both be sound. But which is unsound, and why?

Really imagine being in Medicine 3. Imagine noticing that not pressing gives both yourself and the stranger a half chance of suffering more painful conditions. Having noticed this, would you even seriously consider not pressing? I wouldn’t. I would regard not pressing as morally outrageous. I think, then, that the argument from Objective-Subjective Link must be going wrong somewhere.

Of course, some—call them immoderates—will say that the argument from Objective-Subjective Link goes wrong in assuming that there are prerogatives. More carefully, they’ll argue that since Objective-Subjective Link is so compelling, it must be that you don’t know that not pressing is objectively permissible, and so it must be that not pressing isn’t objectively permissible in both M1 and M2. But if the outcomes in M1 and M2 are the same as those in Medicine 1 and Medicine 2, respectively, moderates must be mistaken in claiming that not pressing is permissible in Medicine 1 and Medicine 2. And since not pressing is surely permissible in Medicine 2—after all, it just allows the stranger to claim the dose for themself—you must not have a self-favoring prerogative not to press in Medicine 1.

There are two main ways in which moderates could try to respond. First, they could try to accept the argument from Ex Ante Pareto and reject the argument from Objective-Subjective Link without rejecting prerogatives altogether. Call proponents of this response ex ante moderates. Second, moderates could—contra intuition—try to accept the argument from Objective-Subjective Link and reject the argument from Ex Ante Pareto. Call proponents of this response ex post moderates.11

There are two natural questions to ask about these two responses. First, can immoderates show that both responses are implausible? If so, Objective-Subjective Link and Ex Ante Pareto would together support rejecting moderate theories. Second, which response is more plausible? Although ex ante moderates’ response is more intuitively plausible, it seems committed to rejecting Objective-Subjective Link, which, again, is highly compelling. At the very least, then, moderates should try to settle this question more decisively. Accordingly, in the next two sections, I’ll develop what I think are the most promising versions of ex ante and ex post moderates’ responses, and I’ll consider how immoderates and other moderates might object to each. To do this, it will help to have in mind the types of moderate theories that ex ante and ex post moderates would each naturally defend.

Ex ante moderates would naturally defend moderate theories—what I’ll call ex ante moderate theories—that determine subjective permissibility by (i) assessing the prospect that each option gives to each person; (ii) using persons’ prospects to determine the subjective reasons for each option, allowing one to give more or less weight to one’s own prospects than to others’; then (iii) using the subjective reasons to determine what’s subjectively permissible.12 Given this “prospects-first” structure, Ex Ante Pareto reflects a natural constraint on how prospects combine to determine the subjective reasons: if an option gives everyone affected a worse prospect, and all else is equal, these prospects must combine in such a way that there’s decisive subjective reason to choose an alternative.13

Ex post moderates would naturally defend moderate theories—what I’ll call ex post moderate theories—that determine subjective permissibility by (i) assessing the objective reasons for each option in each possible state, where these allow giving more or less weight to one’s interests in each state; (ii) using the objective reasons in each state and the probability of each state to determine the subjective reasons; then (iii) using the subjective reasons to determine what’s subjectively permissible.14 Given this “objective-reasons-first” structure, Objective-Subjective Link reflects a natural constraint on how objective reasons combine to determine the subjective reasons: if an option is objectively permissible in each state, these objective reasons must combine in such a way that there isn’t decisive subjective reason to choose an alternative.

I turn now to considering how ex ante moderates could reject the argument from Objective-Subjective Link (Section IV), and then how ex post moderates could reject the argument from Ex Ante Pareto (Section V).

To reject the argument from Objective-Subjective Link, ex ante moderates must deny either Objective-Subjective Link or the claim that you know that not pressing in Medicine 3 is objectively permissible. It might seem that they could reject Objective-Subjective Link simply by insisting that subjective permissibility is determined by persons’ prospects, not by objective reasons. However, this view is difficult to accept. To see why, it’s worth reflecting further on the motivation for Objective-Subjective Link.

The central motivation for Objective-Subjective Link, I think, is the actual reasons conception of risk, according to which the mere fact that one doesn’t know which possible state of the world is actual isn’t independently significant. On this view, one’s epistemic position is significant merely as a guide to the objective reasons in the actual state of the world, whatever it turns out to be—that is, to the actual reasons, whatever they turn out to be. An option is subjectively permissible if, and only if, it’s the best available means, relative to one’s epistemic position, to fulling the actual reasons. Often, one doesn’t know which options are objectively permissible because one doesn’t know which state of the world is actual, and so one must somehow use the relevant probabilities to trade-off the objective reasons in each state. But in the lucky few cases in which one knows that an option is objectively permissible, one knows that it adequately fulfills the objective reasons in the actual state, whatever it turns out to be—that is, the actual reasons, whatever they turn out to be. Surely, then, such an option is among the best available means, relative to one’s epistemic position, to fulfilling the actual reasons, and so is subjectively permissible. Viewing such an option as subjectively impermissible would reflect a concern for something other than the actual reasons.

A testament to its appeal, the actual reasons conception plays a significant role in the motivations for a wide range of decision theories. Of course, decision theories are usually formulated in terms of value rather than reasons, but this difference doesn’t matter here. As Andrew Bacon has recently argued, one central motivation for expected utility theory is—and originally was, at least for Pascal and Fermat—that one should maximize the expectation of actual value, where an option’s actual value is the value that would be realized given the actual state of the world, if one were to choose it.15 Here, actual value is an objective quantity in the sense that it doesn’t depend on one’s epistemic position, except insofar as one’s epistemic position is relevant to the value of what would in fact happen (as when unexpected pleasures or pains would be more intense). On this view, then, actual value is what ultimately matters, and options are merely instrumentally valuable as a means to bringing about actual value. The subjectively best options are the best means, relative to one’s epistemic position, to promoting actual value. Bacon argues that this is precisely what underlies many contemporary causal decision theories.

In contrast, noncausal decision theories, like evidential decision theories, can’t be understood as instructing one to maximize the expectation of any objective quantity—actual value or otherwise.16 According to evidential decision theories, the mere fact that one’s options affect multiple possible states itself affects what’s valuable or what reasons there are. On these views, there is no well-founded notion of “objective value,” “objective reasons,” or “objective permissibility.” The only relevant notions of decision-theoretic value, reasons, and permissibility aren’t in any interesting sense objective; they essentially depend on the agent’s epistemic position such that one’s epistemic position cannot be thought of merely as a guide to the actual facts. All decision-theoretic value is epistemic, “shot through” with the agent’s mind; the map is valuable in itself, not merely as a means of navigating the territory.

It might seem, then, that to reject Objective-Subjective Link, ex ante moderates could endorse a noncausal decision theory. However, even those sympathetic to such theories might allow that there’s something that plays the role of objective permissibility in simple cases like Medicine 3, and so a plausible restriction of Objective-Subjective Link holds in such cases. The cases that make noncausal decision theorists skeptical of the general applicability of objective notions are cases in which you gain information about your options’ outcomes upon learning which act you choose, in which you face severe forms of decision instability, or in which you face bets on the laws of nature.17 Clearly, cases like Medicine 3 needn’t involve any such complications. So, to reject Objective-Subjective Link, ex ante moderates would have to maintain not only that objective permissibility in general is ill-founded, but also that there’s no well-founded restriction of this notion to simple cases. Suffice it to say that this would be highly controversial.

If ex ante moderates can plausibly reject the argument from Objective-Subjective Link, then, it will be by denying that in Medicine 3 you know that not pressing is objectively permissible—that is, by denying that not pressing is objectively permissible in each possible state. This requires explaining why your uncertainty in Medicine 3 affects the morally-relevant features of your options’ outcomes—or, as it’s sometimes said, why the outcomes should be “reindividuated” (or “redescribed”)—such that not pressing isn’t objectively permissible in each state. Is such an explanation available?18

The obvious difference between the outcomes of not pressing in Medicine 3 and the outcomes of not pressing in Medicine 1 and Medicine 2 is that the former, but not the latter, results from an option that gives you and the stranger a worse prospect. Perhaps, then, ex ante moderates could simply maintain that this makes not pressing objectively impermissible in each state.19 Generalizing beyond Medicine 3, this amounts to accepting a version of Ex Ante Pareto that concerns objective permissibility:

Objective Ex Ante Pareto If an option gives each person affected a worse prospect, and all else is equal, then it’s objectively impermissible.

Unfortunately, this principle is incompatible with the standard view of objective and subjective permissibility. Recall that in the Jackson case described in Section II, it’s natural to say that giving medicine C is subjectively required, even though you know that it’s objectively impermissible, because you know that one of giving medicine A or B is objectively required. However, each of giving A and giving B surely gives the stranger a worse prospect than giving C: a half chance of death and a half chance of being cured is surely worse than being guaranteed to be almost-fully cured. So, Objective Ex Ante Pareto implies that each of giving A and giving B is objectively impermissible. While some ex ante moderates may entertain such a revisionary view of the objective-subjective distinction, this seems to me sufficiently radical that it’s unclear whether it provides ex ante moderates a plausible way of rejecting the argument from Objective-Subjective Link. Here, then, I’ll assume that ex ante moderates should accept the standard view of the objective-subjective distinction, if only to see whether there is a less controversial response available.

Indeed, I think there is. Many moderates defend a version of the claim that one has a self-favoring prerogative not to choose options that are costly to oneself because requirements to impose costs on oneself are in some way distinctively demanding. For example, in Medicine 1, you’re permitted not to press because pressing would involve imposing costs on yourself by preventing yourself from being cured, and so requiring you to press seems distinctively demanding. Plausibly, however, in cases of risk, whether a requirement to choose an option is distinctively demanding in the relevant way depends not on whether the option is in fact costly but rather on whether it appears costly—that is, whether it gives you a worse prospect. If so, then since pressing gives you a better prospect, you don’t have a self-favoring prerogative not to press in M1.20 But if you lack a self-favoring prerogative not to press in M1, then since not pressing is impartially worse in M1 and all else is equal, not pressing isn’t objectively impermissible in M1, and so the argument from Objective-Subjective Link is unsound. Call this the apparent-cost explanation.21

If the apparent-cost explanation is correct, Tables 1 and 3 are misleading because they suggest that the outcomes of your options when M1 is the actual state in Medicine 3 are the same as the outcomes of your options in Medicine 1. A more helpful representation indicates which benefits generate self-favoring prerogatives. In Table 1*, for example, the parenthetical “P” indicates that you have a self-favoring prerogative not to press, while the crossed-out “P” in Table 3* indicates that you lack this prerogative due to your uncertainty.

The apparent-cost explanation has several strengths. First, it allows ex ante moderates to accept both Ex Ante Pareto and Objective-Subjective Link. In cases in which one’s options affect yourself and others, and so self-favoring prerogatives are potentially relevant, any option that Ex Ante Pareto implies is subjectively impermissible gives you a worse prospect, and so the apparent-cost explanation implies that there’s no state in which you have a self-favoring prerogative to choose that option. Since options that give everyone affected a worse prospect must be in some state impartially worse, and since all else must be equal for Ex Ante Pareto to apply, such options will in some state be objectively impermissible, thereby preventing conflicts with Objective-Subjective Link.22

Second, the apparent-cost explanation is compatible with the standard view of objective and subjective permissibility. In the Jackson case, you (the agent) are unaffected, and so the apparent-cost explanation is silent. But consider a variant of this case in which both you and the stranger are in the same situation, and you must decide whether to give both yourself and the stranger doses of medicine A, B, or C. Presumably, giving C is still subjectively required, even though you know that it’s objectively impermissible. The apparent-cost explanation is perfectly compatible with this view. Since giving C gives you a better prospect, the apparent-cost explanation merely implies that you don’t have a self-favoring prerogative to give A or give B, and so the relevant objective reasons in each state are simply the objective reasons to do what’s best for yourself and the stranger (giving A or giving B).

Finally, the apparent-cost explanation extends to cases of risk natural views of the motivation for self-favoring prerogatives. It implies that not pressing is objectively impermissible in M1 for the same basic reason that it’s impermissible under full information not to do what’s impartially best when one lacks a self-favoring prerogative and all else is equal. Moderates will disagree about exactly what this reason is, specifying and explaining in different ways the sense in which requirements to do what’s costly to oneself are distinctively demanding. I’ll say more below about the foundational motivations for ex ante moderate theories (Section VII). For now, let me just say that ex ante moderates could claim that doing what’s impartially worse when one lacks a self-favoring prerogative is objectively impermissible because it fails to promote the good, it fails to satisfy others’ claims to beneficence, it fails to treat others as moral equals, or it’s permitted only by principles that others could reasonably reject.

Immoderates and ex post moderates might object that the apparent-cost explanation renders objective permissibility overly sensitive to the agent’s uncertainty. However, it’s difficult to identify a plausible and relevant restriction on reasonable accounts of objective permissibility that supports this objection. As I said in Section II, all sides should agree that the morally-relevant features of one’s options’ outcomes, and so what’s objectively permissible, sometimes depend on one’s uncertainty—that is, that outcomes in different states sometimes interact, as when unexpected pleasures and pains are more intense. The apparent-cost explanation just implies that outcomes in different states interact in a further way such that one lacks self-favoring prerogatives in each state to choose options that give one a worse prospect. Of course, there must be some limits on how to individuate outcomes, but natural limits don’t threaten the apparent-cost explanation. In the context of rational choice, for example, Broome maintains that “Outcomes should be distinguished as different if and only if they differ in a way that makes it rational to have a preference between them.”23 One might similarly suggest that in ethics, outcomes should be distinguished as different if, and only if, they differ in a way that makes it rational for an agent who is concerned only with acting morally to have a preference between them. But ex ante moderates should insist that it’s perfectly rational for such an agent to prefer that they don’t impose costs on others when they lack a prerogative to do so.

Immoderates and ex post moderates might instead object that although the apparent-cost explanation isn’t ruled out on structural grounds, it nevertheless provides a substantively implausible account of the objective reasons in Medicine 3. However, it’s difficult to settle decisively this type of disagreement about which considerations ultimately morally matter. Immoderates and ex post moderates might insist that objective permissibility in a given state is determined by your and others’ interests in that state, not by your prospects.24 But ex ante moderates will insist that the apparent-cost explanation explains why the objective reasons are indeed sensitive to your prospects. This disagreement is uncomfortably close to normative bedrock. Crucially, moreover, the dialectically relevant question here is whether immoderates or ex post moderates can show that ex ante moderates’ response to the argument from Objective-Subjective Link is implausible, not whether ex ante moderates can convince their opponents.25

I think, then, that if immoderates or ex post moderates are going to show that ex ante moderates cannot plausibly reject the argument from Objective-Subjective Link, they’ll need a further argument. Below, I’ll consider some further arguments to which they could appeal. First, though, I want to consider how ex post moderates could reject the argument from Ex Ante Pareto. For it turns out that ex post moderates’ response is more difficult to reject than it might seem.

To reject the argument from Ex Ante Pareto, ex post moderates must deny Ex Ante Pareto or the claim that not pressing in Medicine 3 gives each person a worse prospect. Again, I don’t see a plausible way of denying the latter. Surely, a half chance of a more painful medical condition makes for a worse prospect. So, ex post moderates must deny Ex Ante Pareto. To do so, they might insist that subjective permissibility is determined not by persons’ prospects, but by the objective reasons in each state. To see whether and how this might support rejecting Ex Ante Pareto, it’s worth reflecting further on this principle’s motivations.

The central motivation for Ex Ante Pareto, I think, is the individual reasons conception of social choice, according to which the mere fact that one’s options affect multiple people isn’t independently morally significant.26 On this view, one’s moral reasons—as far as people go—concern only how one’s options affect each person individually. Often, the reasons concerning each person conflict, and so trade-offs are unavoidable. But in the lucky few cases in which one’s reasons concerning each person support the same option, it seems clear that one should choose that option. Doing anything else would reflect a concern for something other than the ways in which one affects each person, even when all else is equal.

A testament to its appeal, the individual reasons conception plays a significant role in the motivations for many moral theories. Simple consequentialist theories are supported by this conception because options supported by the reasons concerning each individual are options that maximize the good, as these theories understand it. But many sophisticated consequentialist and non-consequentialist theories also at least purport to enjoy support from the individual reasons conception—for example, by including duties of beneficence that apply at least when persons’ interests align.

Nevertheless, some theories are incompatible with the individual reasons conception. According to these theories, the mere fact that one’s options affect multiple people itself affects the relevant reasons: the reasons concerning each person “interact” such that one isn’t required to do what’s in each person’s interests. For example, some egalitarian theories embrace “levelling-down,” implying that even under full information, one sometimes should do what’s worse for everyone because it reduces inequality.27 Ex post moderates needn’t accept anything so strong. A more helpful model for ex post moderates is found in certain prioritarian theories, according to which there’s stronger reason to benefit the worse-off.28 These theories avoid leveling-down, but they still generate apparent conflicts between Ex Ante Pareto and Objective-Subjective Link, to which some prioritarians respond by rejecting Ex Ante Pareto.29 To illustrate, consider

Medicine 4 You and the stranger each have moderately painful conditions. Again, only one dose of medicine is available. This time, however, the dose is about to be split in half, with a half-dose being administered to each of you. With only a half-dose, you’ll each suffer a condition that’s more painful than the mild condition but less painful than the moderate condition—a mild* condition, I’ll say. You have a button: pressing it prevents the dose from being split and gives the dose to you or the stranger, depending on the outcome of a fair coin-flip; not pressing it allows each of you to receive a half-dose.30

Table 4 summarizes this case, where “–1–ε” represents suffering the mild* condition.

For some specification of the mild* condition, prioritarians believe that both of you suffering the mild* condition is at least as good as one of you being fully cured and the other suffering the moderate condition. When all else is equal, they’ll say that it’s permissible under full information to bring about the former rather than the latter. It seems, then, that in Medicine 4, prioritarians should think that not pressing is objectively permissible in each state. So, by Objective-Subjective Link, not pressing is subjectively permissible. But, plausibly, suffering the mild* condition with certainty is a worse prospect than a half chance of suffering the moderate condition and a half chance of being cured. If so, then by Ex Ante Pareto, not pressing is subjectively impermissible.

Some ex post prioritarians would deny that not pressing gives a worse prospect by defending certain forms of risk-aversion. This doesn’t provide a useful model for ex post moderates because, again, not pressing in Medicine 3 surely gives you and the stranger worse prospects. Other ex post prioritarians, though, reject Ex Ante Pareto. What ultimately matters, they say, isn’t giving persons better prospects but promoting objective value. Cases like Medicine 4 show that given a commitment to a prioritarian theory of objective value, persons’ interests sometimes determine objective value in such a way that taking the best means, relative to one’s epistemic position, to promoting objective value involves giving worse prospects. In a sense, persons’ interests interact to determine objective value such that the best means, relative to one’s epistemic position, to promoting objective value isn’t always the best means, relative to one’s epistemic position, to promoting each person’s interests. In these cases, they argue, it isn’t implausible to give worse prospects because doing so is a way of promoting objective value.31

Ex post moderates could reject Ex Ante Pareto in this same general way. What ultimately matters, they could say, isn’t giving persons better prospects but fulfilling the objective reasons. Cases like Medicine 4 show that, given a commitment to prerogatives, your and others’ interests sometimes determine the objective reasons in such a way that the best means, relative to your epistemic position, to fulfilling these reasons involves giving worse prospects. In a sense, your and others’ interests interact to determine the objective reasons such that the best means, relative to your epistemic position, to fulfilling the objective reasons isn’t always the best means, relative to your epistemic position, to promoting each person’s interests. In these cases, ex post moderates could argue, it isn’t implausible to give worse prospects because doing so is a way of fulfilling the objective reasons.

Immoderates and ex ante moderates might insist again that the subjective permission not to press in Medicine 3 is simply too intuitively implausible to accept. But, again, it’s difficult to regard such intuitive judgments alone as decisive, at least until the possibility of a more decisive argument is ruled out. Immoderates and ex post moderates might instead try to show that ex post moderates’ account of the relevant objective reasons is implausible. But, again, disagreements of this kind are uncomfortably close to normative bedrock. If anything, moreover, ex post moderates’ account of the objective reasons seems more straightforward than ex ante moderates’, since ex post moderates maintain that costs and benefits to the agent generate self-favoring prerogatives in each possible state in cases of risk just as they would in cases of full information. Of course, immoderates may think that they have other arguments against moderate theories, but the question here is whether principles like Objective-Subjective Link and Ex Ante Pareto in particular support rejecting prerogatives.

I think, then, that if immoderates or ex ante moderates are going to show that ex post moderates cannot plausibly reject the argument from Ex Ante Pareto, they’ll need a further argument. Here, however, it might seem that a further argument is indeed available. In discussions of other parts of morality, so-called “composition arguments” have been claimed to show that ex post theories are untenable.32 These arguments work by “decomposing” cases in which one’s options affect multiple people into multiple cases in which one’s options affect only a single person; then they claim that it’s subjectively impermissible in these single-person cases to give a worse prospect; then they infer that it’s also subjectively impermissible to give everyone a worse prospect when these cases are “composed” so that one’s options again affect multiple people. Given how influential composition arguments have been in other contexts, it’s natural to wonder whether there’s a convincing composition argument against ex post moderates. To anticipate, it takes some work to spell out such an argument, but the work is worthwhile: once the details are clear, we can see why this type of argument isn’t convincing for general reasons that cast doubt on whether composition arguments against ex post theories are ever convincing.

What I think is the strongest composition argument against ex post moderates has four steps.33

Step one: consider the following single-person variants of Medicine 3.

Medicine 5 Just like Medicine 3, but you know that the stranger doesn’t need the dose. So, either you have the mild condition, pressing gives the dose to the stranger for no benefit, and not pressing allows yourself to receive it (call this possibility “M1*”); or, you have the moderate condition, pressing claims the dose for yourself, and not pressing allows it go unclaimed (call this possibility “M2*”).

Table 5 summarizes this case.

Medicine 6 Just like Medicine 3, but you know that you don’t need the dose. So, either the stranger has the severe condition, pressing gives the dose to them, and not pressing allows yourself to receive it for no benefit (call this possibility “M1*”); or, the stranger has the moderate condition, pressing claims the dose for yourself for no benefit, and not pressing allows the stranger to claim it (call this possibility “M2*”).

Table 6 summarizes this case.

Some moderates will think that, in Medicine 5, each option is subjectively permissible because there are self-sacrificing prerogatives to make oneself worse-off when it affects no one else. Call these strong self-sacrificing prerogatives. Others won’t go this far, maintaining that pressing is subjectively required, and so accepting only weak self-sacrificing prerogatives. I’ll assume that there are strong self-sacrificing prerogatives, explaining in footnotes how to modify the argument given only weak self-sacrificing prerogatives.34

In Medicine 6, your options affect only the stranger, and pressing gives them a better prospect. Because all else is equal, pressing seems subjectively required. This follows from:

Restricted One-Person Ex Ante Pareto If an option affects only a single person compared to an available alternative, the option gives that person a worse prospect, the single person affected isn’t the agent, and all else is equal, then it’s subjectively impermissible.35

Why accept this principle? It’s strictly weaker than Ex Ante Pareto, and its motivation is stronger. As we’ve seen, the main way to resist Ex Ante Pareto is to maintain that the ways in which one affects multiple persons’ interests can interact such that a concern for fulfilling one’s objective reasons sometimes comes apart from a concern for each person’s individual interests. But if an option affects only a single person, it’s unclear how such interaction could occur.

Step two: all and only the options that are permissible in Medicine 5 and Medicine 6 remain permissible if you face both cases simultaneously. Cases like this are somewhat complicated, but they are indeed possible. To illustrate, consider:

Medicine 7 Again, you and a stranger each need a dose of medicine. This time, however, you have different conditions and so need different medicines. You have condition 1 and so need a dose of medicine 1; the stranger has condition 2 and so needs a dose of medicine 2. You have two buttons: button 1 controls the dose of medicine 1; button 2 controls the dose of medicine 2. You know everything just described. Unfortunately, however, you don’t know whether you have a mild version of condition 1, the stranger has a severe version of condition 2, and both doses are about to be administered to you, in which case pressing each button gives the corresponding dose to the stranger (call this possibility “M1**”); or whether you have a moderate version of condition 1, the stranger has a moderate version of condition 2, and both doses are unclaimed, in which case pressing each button claims the corresponding dose for yourself (call this possibility “M2**”). Each possibility is equally likely.

Table 7 summarizes this case.

Compared to pressing both, pressing only button 1 doesn’t affect you and gives the stranger a worse prospect. So, by Restricted One-Person Ex Ante Pareto, pressing only button 1 is subjectively impermissible. Similarly, compared to pressing only button 2, pressing neither doesn’t affect you and gives the stranger a worse prospect. So, by Restricted One-Person Ex Ante Pareto, pressing neither is subjectively impermissible. Finally, because pressing both and pressing only button 2 differ only in how they affect you, each seems subjectively permissible given strong self-sacrificing prerogatives.36

Step three: pressing both is subjectively required in the variant of Medicine 7 in which pressing only button 1 and pressing only button 2 are rendered unavailable. To illustrate, consider:

Medicine 8 Like Medicine 7, except the buttons are connected such that pressing only one button has the same effect on who gets which dose as pressing neither.

Table 8 summarizes this case.

Options are standardly individuated by the outcomes that they stand to bring about. So, Medicine 8 seems just like Medicine 7, except the distinct options of pressing only button 1 and pressing only button 2 are unavailable. But removing these options doesn’t seem to change what’s subjectively permissible, except by removing a subjectively permissible option (pressing only button 2). So, in Medicine 8, pressing both seems subjectively required. This follows from:

Restricted Contraction Consistency for Requirement Suppose that it’s subjectively required to choose one of x₁,…, x_n (each of which is subjectively permissible) from a set of options X, and that x₁,…, x_n differ only in how they affect the agent. Then for any subset X’ of X that includes some but not all of x₁,…, x_n, it’s subjectively required to choose one of these options from X’ when all else is equal.37

This principle says that if it’s required to choose one of several options that differ only in how they affect the agent, then removing some but not all of these options (perhaps along with some impermissible options) makes it subjectively required to choose one of the several options that wasn’t removed. While it may sound complicated, this principle has a natural rationale. For when it’s subjectively required to choose between options that differ only in how they affect the agent, none of these options is so costly to the agent that a less costly alternative is subjectively permissible. But this isn’t affected, it seems, by merely removing some of these options (and perhaps some impermissible options).

Step four: Medicine 3 isn’t morally different from Medicine 8. The only descriptive difference is that the former involves one button while the latter involves two. This isn’t morally significant. So, since pressing both in Medicine 8 is subjectively required, pressing in Medicine 3 is subjectively required—contra ex post moderates.

It might seem, then, that ex post moderates must reject not only Ex Ante Pareto but also Restricted One-Person Ex Ante Pareto. Because it’s unclear how ex post moderates could reject the latter, this might seem to provide precisely the further argument that immoderates and ex ante moderates need to raise a decisive problem for ex post moderates. Is this argument convincing?

I think not. Ex post moderates have principled reason to reject Restricted Contraction Consistency for Requirement by maintaining that pressing neither is subjectively impermissible in Medicine 7 but subjectively permissible in Medicine 8. In particular, they have principled reason to maintain that pressing neither is subjectively impermissible in Medicine 7 only because pressing only button 2 is available. For in Medicine 7, compared to pressing only button 2, pressing neither affects only the stranger and is worse for them.38 However, in Medicine 8, when pressing only button 2 is unavailable, there’s no alternative that affects only the stranger and is worse for them, and so pressing neither (equivalently, at most one) is subjectively permissible. In ex post moderates’ reasons-talk: in Medicine 7, there’s decisive objective reason to press only button 2 over pressing neither in M1**, and there’s decisive objective reason to press neither over pressing only 2 in M2**. But the former are stronger than the latter, since the stranger is worse off in M1** if you press neither than in M2** if you press only button 2. So, since the states are equiprobable, pressing neither is subjectively impermissible. However, in Medicine 8, there isn’t decisive objective reason to press both in either state, since the options affect both you and the stranger just as they do in Medicine 3. So, pressing neither (equivalently, at most one) is subjectively permissible.

The general point is that removing options that affect only a single person can make it such that the remaining options affect multiple people. So, because ex post moderates should maintain that, when options affect multiple people, persons’ interests sometimes interact to determine the objective reasons, they should also maintain that removing options sometimes affects the objective reasons, thereby affecting the subjective reasons and so what’s subjectively permissible, contra Restricted Contraction Consistency for Requirement. This point may generalize to the influential composition arguments against ex post theories of other parts of morality. Although other parts of morality may raise distinctive complications, proponents of ex post theories may also be able to explain why persons’ interests interact to determine the objective reasons. If so, they may have the resources to respond similarly to analogous composition arguments, by accepting the relevant weakening of Ex Ante Pareto, like Restricted One-Person Ex Ante Pareto, but rejecting the key “composition” premise, like Restricted Contraction Consistency for Requirement.

Although the composition argument against ex post moderates is unconvincing, one might now wonder whether an analogous argument works against ex ante moderates. Surprisingly, this type of argument has been largely overlooked in discussions of other parts of morality. Such an argument would work by “decomposing” cases in which one’s options affect multiple possible states into multiple cases in which one’s options affect only a single state; then they would claim that it’s subjectively permissible in these cases to choose certain options, using a weakening of Objective-Subjective Link restricted to options that affect only a single state; then they would infer that it’s also subjectively permissible to choose these options when these cases are “composed,” even though the resultant options together give everyone affected a worse prospect. Accordingly, this type of argument would purport to show that ex ante moderates must reject the weakening of Objective-Subjective Link restricted to options that affect only a single state. However, I won’t work through the details of this argument here. Now that we’ve seen why the above personwise composition argument against ex post moderates isn’t convincing, it’s straightforward to see why the analogous statewise composition argument against ex ante moderates isn’t convincing. Just as ex post moderates have the resources to resist the personwise composition argument by maintaining that persons’ interests sometimes interact to determine the objective reasons such that removing an option sometimes affects what’s subjectively permissible, so too ex ante moderates have the resources to resist the analogous statewise composition argument by maintaining that—given the apparent-cost explanation—outcomes in different states sometimes interact to determine the objective reasons such that removing an option sometimes affects what’s objectively permissible. Again, then, to raise a decisive problem for either ex ante or ex post moderates, a further argument is needed.

Rather than appealing to principles like Ex Ante Pareto or Objective-Subjective Link, one might argue that ex ante or ex post moderate theories conflict with, or aren’t as well-supported by, the foundational motivations for prerogatives. While it’s controversial what the foundational motivations for prerogatives are, considering the most influential proposals will go a long way toward evaluating arguments of this kind.

Consider first Scanlon’s contractualist defense of prerogatives, according to which we could reasonably reject principles for the general regulation of behavior that prohibit impartially suboptimal forms of self-favoring and self-sacrifice.39 There is ongoing controversy among contractualists about whether reasonable rejectability is more plausibly extended to cases of risk in a way that is sensitive to persons’ ex ante or ex post claims. While this controversy has focused on other aspects of contractualist theory—especially the limits it places on aggregation—it applies just as well to prerogatives. If an ex ante theory proves the most plausible development of the contractualist view of morality, it will straightforwardly support an ex ante view of prerogatives. If an ex post theory proves the most plausible development of the contractualist view of morality, it will straightforwardly support an ex post view of prerogatives. But while it may be illuminating to consider prerogatives as yet another frontier in the ongoing debate between ex ante and ex post contractualists—a task I leave to future work—it will remain unclear which type of moderate theory contractualism favors until this broader debate is settled more conclusively.40

Although the structure of contractualist justification doesn’t clearly favor ex ante or ex post moderate theories, contractualists and non-contractualists alike might think that the substantive considerations that motivate prerogatives favor one type of theory over the other. Consider, then, the view that including prerogatives allows a moral theory to respect agents’ moral autonomy,41 and the related view that including prerogatives allows a theory to reflect agents’ moral status as ends-in-themselves.42 If moderate theories better protect agents’ moral autonomy or reflect agents’ moral status by being more permissive than immoderate alternatives, one might think that ex post moderate theories better protect agents’ autonomy or reflect agents’ status by being more permissive than ex ante moderate theories. But this is too quick. It’s plausible that for any ex ante moderate theory, there’s a more permissive ex post moderate theory. For example, for any ex ante moderate theory that subjectively requires not pressing in Medicine 3, there’s a more permissive ex post moderate theory that subjectively permits each of pressing and not pressing. However, it’s not plausible that every ex post moderate theory is more permissive than every ex ante moderate theory. An ex ante moderate theory will be more permissive than an ex post moderate theory if the former extends a moderate theory that includes much stronger prerogatives under full information.

One might instead suggest that appeals to autonomy or moral status more strongly support ex ante moderate theories because agents in cases of risk are typically in fact concerned with each person’s prospects, and properly respecting the agent’s autonomy or status involves making reasonable room for them to act on what they care about. But, again, this is too quick. These views imply that morality makes reasonable room for the agent’s evaluative commitments, where these commitments reflect the agent’s views of what has final, non-instrumental value or of the objective reasons. Yet an agent’s concern with persons’ prospects is not typically a concern for what has final, non-instrumental value or for the objective reasons; it’s a concern for the best means to achieving such value or fulfilling such reasons. Moreover, even if some motivations for prerogatives in effect grant the agent a protected sphere within which they can select not only ends but also means, and even if most agents typically assess their options in accordance with ex ante moderate theories, some agents could as a matter of contingent psychological fact assess their options in accordance with ex post moderate theories. And just as there have to be reasonable limits on the choice of ends, so too there would have to be reasonable limits on the choice of means. The question would then arise whether these limits are more plausibly characterized by an ex ante or ex post moderate theory. In general, as far as I can tell, autonomy- and status-based views are compatible with ex ante and ex post moderate theories because nothing internal to these views bears on the question of how to extend them to cases of risk.43

Finally, consider Scheffler’s view, according to which moderate theories provide one acceptable way of reflecting “the independence of the personal point of view”—that is, the fact that agents standardly, although not necessarily, give disproportionate significance to their own projects relative to those projects’ impersonal value.44 One might again be tempted to think that if the content of one’s prerogatives depends on one’s evaluative commitments, and if one is centrally concerned with persons’ prospects in cases of risk, then one’s prerogatives depend on persons’ prospects in cases of risk, and so ex ante moderate theories are well-motivated while ex post moderate theories are not. But, as above, this would be too quick. Insofar as people do standardly value their prospects, they value their prospects merely instrumentally and so not in the way that on Scheffler’s view provides the basis for a prerogative. Moreover, even if Scheffler’s view did regard instrumental valuing as providing such a basis, the question would again arise whether the reasonable limits on this type of valuing are better characterized by an ex ante or ex post moderate theory.

However, unlike autonomy- and status-based views, which seem neutral with respect to the choice between ex ante and ex post moderate theories because nothing internal to them bears on the question of how to extend prerogatives to cases of risk, there seem to be aspects of Scheffler’s view that do indeed bear on this question. Perhaps, then, once we understand how best to extend Scheffler’s view to risk, we’ll find that it more strongly supports one of ex ante or ex post moderate theories. Alternatively, and to anticipate, even if this suggestion proves difficult to defend, the extension of Scheffler’s view to cases of risk may nevertheless provide an illuminating way of characterizing the conceptions of morality that underlie ex ante and ex post moderate theories. Let me explain further, then, how Scheffler’s view works and how I think it is most naturally extended to cases of risk.

Motivating Scheffler’s view is a conception of the moral domain as centrally involving, perhaps deriving from, the relationship between the personal and impersonal points of view. Applied to a choice situation, each point of view provides an assessment of the reasons for choosing each option. Assessment from the impersonal point of view is impartial: the effects of each option on each person are given equal and positive weight. Assessment from the personal point of view reflects the agent’s evaluative position: the fact that they act from a position characterized by deep and forceful evaluative commitments, the significance of which standardly isn’t weighed by the agent in direct proportion to their impersonal value. This, again, is the independence of the personal point of view.

Moral theories are then distinguished in part by the way in which they reflect the independence of the personal point of view. Impersonal theories defer to the impersonal point of view, reflecting the independence of the personal point of view only by allowing that impersonal value depends on the fulfillment of each person’s evaluative commitments. Egoist theories defer to the personal point of view, reflecting its independence by giving it carte blanche. Moderate theories reflect the independence of the personal point of view by integrating the personal and impersonal assessments to arrive at an overall assessment of one’s reasons that permits both what is impersonally best, and what is, within reason, impersonally worse. Such integration can be thought of as involving two stages: first, an option’s morally relevant features are integrated into impersonal and personal assessments of the reasons for choosing it; second, the personal and impersonal assessments are integrated to determine the strength of one’s overall reasons and so what’s permissible. Scheffler claims not that a moderate form of integration provides the uniquely acceptable way of reflecting the independence of the personal point of view, but that “it gives rational expression to some plausible attitude towards persons.”45

To extend Scheffler’s view to cases of risk, I want to make two main suggestions. First, the agent’s epistemic position is another aspect of their personal point of view, alongside their evaluative commitments. My main reason for thinking of one’s epistemic position in this way is that one’s agency is both evaluatively and epistemically situated: like one’s evaluative commitments, one’s epistemic position partly constitutes the distinctive position from which one exercises one’s agency. Whatever the prospects for a belief-desire model of action, it’s in some sense one’s evaluative commitments and epistemic position that together explain why one pursues some alternatives and not others. And just as the evaluative aspect of one’s personal point of view is independent in the sense that it standardly diverges from the evaluative content of the impersonal point of view, so too the epistemic aspect is independent in the sense that it standardly diverges from the “fully informed” point of view—that is, from the actual facts.46 Second, moral theories can be distinguished in part not only by the way in which they reflect the independence of the agent’s evaluative point of view, but also by the way in which they reflect the independence of the epistemic aspect of the personal point of view. It is here that the choice between ex ante and ex post moderate theories arises within Scheffler’s view.

On an ex ante version of Scheffler’s view, the agent’s epistemic position is centrally relevant to the first stage of integration, in which the personal and impersonal assessments are generated by integrating options’ morally relevant features. The agent’s epistemic position is used to assess persons’ prospects, which function as the “inputs” to the personal and impersonal assessments. Then, in the second stage of integration, in which the personal and impersonal assessments are integrated to determine what’s subjectively permissible, the agent’s epistemic position is irrelevant.

On an ex post version of Scheffler’s view, the agent’s epistemic position plays a more minor role in the first stage of integration. The agent’s epistemic position is relevant to the generation of the impersonal and personal assessments only insofar as it affects how the options would affect each person in accordance with ex post moderates’ view of the objective reasons, as when unexpected pleasures and pains are more intense. Then, in the second stage of integration, in which the personal and impersonal assessments are integrated, the agent’s epistemic position plays a central role. In this stage, the subjective reasons are determined by the objective reasons given by the first stage together with the probabilities of each state as given by the agent’s epistemic position. On this view, the agent’s epistemic position is used primarily to assess the “outputs” of the first stage of integration so as to determine the subjective reasons and so what’s subjectively permissible.

While my aim here isn’t to defend Scheffler’s view, it seems to me a potential strength of this view that it makes possible this characterization of the disagreement between ex ante and ex post moderate theories in terms of the two stages of integration and so in terms of the way in which each type of theory reflects the independence of the personal and impersonal points of view—or, in more dramatic terms, the evaluative and epistemic situatedness of agency. Nevertheless, the point to emphasize is that Scheffler’s view seems capable of motivating both ex ante and ex post moderate theories, and it doesn’t seem to motivate one more strongly than the other. It would motivate one more strongly than the other if the ex ante or ex post view of the two stages of integration better reflects a morally significant fact about persons. As far as I can tell, however, each provides a natural way of reflecting the fact of epistemic independence and in turn provides a natural characterization of the relationship between personal and impersonal points of view. If one worries that the ex post version alienates us from the moral significance of persons’ prospects (as understood by ex ante moderates), one might just as well worry that the ex ante version alienates us from the moral significance of the objective reasons (as understood by ex post moderates). So, as Scheffler says of moderate and impersonal theories, I say of ex ante and ex post moderate theories: each gives rational expression to some plausible attitude towards persons.

Absent a more decisive argument, the best we can do may be to use the foregoing to evaluate the overall strengths and weaknesses of the rival theories. When we do this, I think, there’s a compelling case to be made for ex ante moderate theories.

Consider first how ex ante and ex post moderate theories compare in accommodating our intuitive judgments about cases. Any ex post moderate theory sometimes permits giving everyone a worse prospect, even when all else is equal, as in Medicine 3. As I have said, this seems implausible. When I imagine being in your position in Medicine 3 and noticing that not pressing gives both the stranger and myself a worse prospect, not pressing seems outrageous. Moreover, I don’t know of any cases in which the implications of ex post moderate theories seem more plausible than those of ex ante moderate theories. In general, then, it seems that ex ante moderate theories better accommodate our intuitive judgments about cases.47

Consider next the motivations for ex ante and ex post moderate theories. I’ve just argued that neither is more strongly supported by the foundational motivations for prerogatives. In addition, ex ante moderate theories are compatible with both Ex Ante Pareto and Objective-Subjective Link and so with both the individual and actual reasons conceptions. In contrast, ex post moderate theories are compatible only with Objective-Subjective Link and the actual reasons conception, although they will insist that their more straightforward account of objective reasons provides principled reason to reject Ex Ante Pareto and so also the individual reasons conception. However, ex post moderates’ more straightforward account of objective reasons doesn’t seem to diminish the force of the intuitive judgment that pressing is subjectively required in Medicine 3. This seems to me particularly significant. As we’ve seen, ex post moderates and some ex post prioritarians reject Ex Ante Pareto in analogous ways in Medicine 3 and Medicine 4, respectively. Yet when I keep in mind ex post prioritarians’ account of objective value, not pressing in Medicine 4 doesn’t seem to me particularly implausible. In contrast, even when I keep in mind ex post moderates’ account of the objective reasons, not pressing in Medicine 3 still seems outrageous. This suggests an underlying difference between prerogatives and prioritarianism. Ex post prioritarians’ violations of Ex Ante Pareto may be more reasonable because the moral significance of giving priority to the worse-off is plausibly understood as primarily determining the objective value of outcomes independently of the agent’s uncertainty, whereas ex post moderates’ violations of Ex Ante Pareto seem far less reasonable because prerogatives aren’t as plausibly understood as primarily determining the objective reasons independently of the agent’s uncertainty; rather, they’re more plausibly understood as primarily determined by the agent’s epistemic position, as ex ante moderates claim. More generally, this suggests that ex post moderate theories’ simpler account of objective reasons isn’t sufficiently compelling to make up for ex post moderate theories’ implausible implications.

I conclude, then, that ex ante moderate theories are more plausible than ex post alternatives. When competing moral theories have similarly compelling foundational motivations, but one better accommodates our intuitive judgments—at least our sufficiently clear judgments that persist upon reflection and for which we lack a compelling theory of error—it’s reasonable to conclude that the more intuitively compelling theory is the more plausible. Of course, it would be preferable to have a more decisive argument, but often in ethics this is the best we can do.

How do ex ante moderate theories compare with immoderate alternatives? Of course, my aim here isn’t to give a general defense of prerogatives. The point to emphasize for present purposes is that the apparent conflicts that prerogatives generate between Objective-Subjective Link and Ex Ante Pareto don’t support rejecting ex ante moderate theories in favor of immoderate alternatives because ex ante moderate theories can satisfy both of these principles by appealing to the apparent-cost explanation. This provides new reasons for skepticism about Harsanyi-style arguments from principles like Objective-Subjective Link and Ex Ante Pareto for simple consequentialist theories like utilitarianism.

Admittedly, some moderates may be skeptical of whether principles like Objective-Subjective Link and Ex Ante Pareto ever could have raised a serious challenge for the general tenability of a moderate view of ethics. So far, I have ignored this response, but it’s sufficiently common among those sympathetic to prerogatives that it deserves acknowledgment. In particular, some moderates might suggest that the problem on which I’ve focused is merely symptomatic of the attempt to subject our moral thought to a level of precision inappropriate for the subject matter of ethics. Insofar as our moral thought includes elements of the individual and actual reasons conceptions, they might claim, the relevance of these conceptions to human conduct is not well-characterized by principles like Ex Ante Pareto and Objective-Subjective Link or well-illustrated by highly-stylized cases like Medicine 3. This view might seem especially appealing to moderates who take prerogatives to be motivated by the search for a moral theory compatible with a suitably realistic conception of human psychology.

However, a response along these lines depends for its force on an account of the subject matter of ethics that clarifies and supports the sense in which that subject matter isn’t amenable to the broad methodological approach I’ve employed here. I don’t know whether such an account can be given. Reasonable expectations about the level of precision appropriate in ethics are largely set by the extent to which attempts at greater precision succeed in delivering an acceptable theory. The possibility of failure in this regard is all the more reason to try.

If anything, the foregoing suggests that prerogatives allow for more systematic theorizing than it might have seemed. Such theorizing may also have significant practical implications. According to ex ante moderate theories, we can assess the content of our prerogatives simply by giving our own prospects more or less weight than others’; we needn’t keep track of the objective reasons in each possible state of the world. This may simplify moral deliberation, for example, in cases of charitable donation when it’s clear whether there’s a reasonable weighting of one’s prospects that justifies non-donation. Of course, even seemingly simple cases may raise significant challenges. It’s particularly natural to wonder how prerogatives interact in cases of risk with other parts of morality—not least aggregation, constraints, priority, and equality. This, though, is precisely the type of question that motivates the further systematic development of moderate moral theories.