I. Introduction

Commonsense morality involves a number of seemingly conflicting features. On the one hand, all persons have equal moral worth, such that all agents ought to have the same moral reasons to act and all patients ought to be treated impartially. On the other hand, commonsense morality tells us that we are permitted, if not required, to be partial to certain individuals: for example, one's family and friends. In addition, there are certain optimific actions we ought not perform: for example, pushing a hiker off a footbridge to stop a trolley from killing five people. These insights often pit two approaches to normative ethical theorizing against each other: agent-neutral views tend to focus on the shared and impartial features of commonsense morality, while agent-relative views capture the exceptions, that is, partiality and constraints.Footnote ¹ Many have seen this tension as the core debate between consequentialists and deontologists. However, with the introduction of the consequentializing project and agent-relative value, the debate seems to have shifted.

There are two ways to frame the issue. On the one hand, some consequentializers, most notably Doug Portmore (Reference Portmore2007, Reference Portmore2011), have conceded the need for agent-relativity, while attempting to remain consequentialists, shifting the debate away from the importance of neutrality to the importance of a fully axiological view, relative or not. I think this is a mistake. Agent-neutral views allow for crucial features in a normative ethical view; ones that I argue are left behind in consequentializing.

In response, other consequentializers, most notably Jamie Dreier (Reference Dreier and Timmons2011), argue that the consequentializing project allows for the distinction between consequentialists and nonconsequentialists to be set aside altogether. The debate does not become whether or not to hold an axiological view; rather, by placing all views into one (albeit axiological) framework, we can now look carefully at more important distinctions, such as the distinction between agent-relativity and agent-neutrality. If this is right, then consequentializing does not obscure the importance of neutrality; rather, it allows us to judge its importance more clearly. Granting this for the moment, my concern remains. On this way of framing the worry, the issue becomes whether agent-relativity and agent-neutrality function differently when placed in an axiological or non-axiological framework. I will argue that they do. On my view, agent-relative non-axiological views are better suited to capture the important features that motivate agent-neutral views than their consequentialist counterparts.

I begin in Section II with an assessment of the motivation for an agent-neutral view in ethics. In particular, I will argue for two desiderata of a moral theory. A normative theory must reflect morality's shared nature, as well as a certain kind of impartiality. In Sections III and IV, I turn to a particular deontological view, one that upholds agent-relative constraints (ARCs). Section III argues that when ARCs are captured in terms of agent-relative reasons, this particular deontological view is able to maintain the features that motivate agent-neutral views as laid out in the preceding section. Section IV then considers a consequentialized counterpart view, in which the deontic results of ARCs are captured by agent-relative value. Here, my conclusion is a dilemma for the consequentialist. Either the relativization of the good fails to meet the desiderata laid out in Section II or else the consequentializer must abandon what makes the view distinct and attractive. Section V concludes.

Before moving to the main argument, it is worth quickly noting two issues that I will not be tackling in this article. First, I will not take a firm position concerning how to define the distinction between agent-relativity and agent-neutrality.Footnote ² There is a lengthy debate in the literature as to how to best understand the distinction, which I will set aside. What I say about agent-relativity in this article is not affected by where one lands on the technical details of the distinction. For my purposes, agent-relativity, whether applied to reasons, constraints, or value, involves an essential reference to a particular agent, whereas agent-neutrality has no such reference. For a simple example of the distinction, consider two constraints borrowed from Tom Dougherty (Reference Dougherty2013):

The second issue concerns whether we should include agent-relativity in our normative ethical view at all. As mentioned above, there is at least an apparent need for something like agent-relativity in order to capture many commonsense ethical judgments. One might of course respond by arguing either that those judgments can be captured without resorting to agent-relativity or that such judgments need not be captured. For my purposes, I want to grant that agent-relativity is needed, or at least explore what particular views look like when they incorporate agent-relativity. Thus, the goal is not to argue for the inclusion of agent-relativity; rather, my aim is to show that if one wishes to incorporate agent-relativity, it is best to do so at the level of reasons, not at the level of the good or value. For doing so, I argue, better maintains what is important about neutrality in ethics.

II. What agent-neutral views do well

Agent-neutral views have been defended on a number of grounds. Yet, my focus here is not on agent-neutrality itself. Rather, it is on the features of morality that agent-neutral views are best suited to capture – those which agent-relative views often leave behind, or so opponents may argue. In particular, I want to focus on the idea that we are all moral persons. As such, morally, we are all deserving of the same treatment. This includes both how others are morally permitted to treat us, as well as the moral expectations and requirements placed on us as moral agents. I take it that prima facie, these are the features of morality that agent-relativity most clearly threatens.Footnote ³ Agent-relative views present agents with different moral reasons, which leads to individuals being treated differently by different agents. Notice that this will be true regardless of whether those reasons are agent-relative or patient-relative.Footnote ⁴ On agent-relative views, moral reasons will differ between agents insofar as there is an essential reference to each particular agent. Since those reasons differ, patients will receive different treatment based on which agent is acting. On patient-relative views, the essential reference to the patient requires that agents treat each patient according to the particular reasons on hand. Again, moral reasons will differ between agents, in this case based on which patient is featured in their reasons. Here again, patients will be treated differently.Footnote ⁵ Along these lines, it will be helpful to group the important features associated with neutral views into two categories: agent-concerns and patient-concerns.

Consider first agential concerns. Christine Korsgaard (Reference Korsgaard1993) argues that morality is importantly a shared endeavor. Morality springs from the recognition of our common humanity. I treat your ends as my ends when I recognize that those ends stem from our shared personhood. Agent-neutrality captures this feature of morality quite simply. When reasons and values are neutral, we are able to share those reasons and values. However, if reasons or values are relative to a particular agent, then your reasons and values are necessarily different from mine. In turn, those reasons require that I be concerned with my own wrongdoing, and you with yours, or so the argument goes.Footnote ⁶ As Korsgaard puts it in later work,

Moreover, this has further-reaching consequences for patients. Korsgaard again:

If morality is agent-relative, so the argument goes, any sense of morality as a shared enterprise is lost. We each have our own moral concerns, and not even the victims or beneficiaries of our actions have a voice.

A related agential concern arises out of this first worry. If Korsgaard is right, then neutrality is likewise crucial for moral development and education. If morality is indeed a shared endeavor, it seems required that we educate and motivate one another to succeed morally.Footnote ⁷ It is just as much my concern as it is yours whether you are able to reason well morally and act on the results of your moral deliberations. This is easily captured with an agent-neutral view. However, agent-relativity seems to lack any reason for one to aid others in their moral development. As argued above, agent-relativity calls for me to be concerned with my own wrongdoings. There seems to be no reason for me to be concerned with your wrongdoings.Footnote ⁸ As such, an agent-relative view seems to provide no motivation to provide moral education or encouragement to others. Moreover, even if I were so motivated, our reasons differ. Thus, I would need to be able to understand and embrace your reasons in order to even be in a position to provide guidance. An important desideratum of a moral theory, then, is that it provides us with reasons that we can share, along with the motivation for and path to moral education. Agent-neutral views clearly capture this desideratum. The question addressed below will be whether agent-relative views can do the same.

Consider now the patient concerns. As a moral patient, I am concerned not with how I act, but how I am treated. As Thomas Nagel (Reference Nagel1986) points out, it is a basic moral insight that I am no more important than anyone else. Likewise, no one else is any more important than I am. As such, all persons should be treated in some sense impartially. Here again, at least prima facie, neutral views are well suited. When we are all given the same reasons, no agent can be said to be more important than any other. Yet, to capture Nagel's insight, it is not enough simply to maintain agent-neutrality. After all, we can have agent-neutral reasons that are patient-relative. For instance, consider the following agent-neutral moral constraint:

Here, all agents have the same reasons to act, but the constraint only protects those who are mothers. If I am a mother, this constraint seems to conflict with the idea that I am no more important than anyone else. After all, this reason to not disrespect does not extend to non-mothers. This might lead someone to argue that true impartiality is only achieved when moral reasons and values are completely patient-neutral, in addition to agent-neutral.

While a fully neutral view would indeed be impartial, I suspect few would want to go as far as to argue that all persons ought to be treated exactly the same at all times. It seems clear, for instance, that children ought to be treated differently from adults: for example, a softer touch when a mistake is made. Similarly, those with disabilities might require different treatment in certain circumstances: for example, granting a request for an ASL interpreter at a talk. Thus, we need to get clear on what the requirement of impartiality amounts to for a moral theory.Footnote ¹⁰

Begin with Nagel's insight that no one person is more morally important than anyone else. At this point, it will be helpful to point to a distinction between “equal treatment” and “treatment as equals”.Footnote ¹¹ Equal treatment implies that everyone ought to receive the same moral consideration no matter the circumstances. On this view, if I can only save one or two persons from drowning, it might be best to let both drown, as that is the only way to ensure equal treatment. Likewise, “to inflict a one-year jail sentence on all accused persons, regardless of whether they are guilty or innocent, is to provide equal treatment to members of that group” (Jollimore Reference Jollimore2017). However, treatment as equals requires that persons be treated “in accordance with what rights they possess, what legitimate claims they put forward, and, in general, with what they deserve” (ibid.).

I contend that it is important for moral theories to treat persons as equals, not to treat persons equally. It matters whether you or I have committed a crime or not. It matters whether you are a child or I am disabled. Thus, room is made for a bit of patient-relativity with respect to the rights and relationships that a particular moral patient possesses. In order to embrace the idea that no one person is any more morally important than another we need not deny that our duties with respect to mothers may differ from our duties with respect to siblings. We simply must embrace the idea that there is nothing special about my being a mother that would justify a difference in treatment between myself and another mother.Footnote ¹² I take it to be an open question, one I will address further below, how far this relativity may go while remaining consistent with this idea of treatment as equals. For now, the crucial point is that with respect to patients, impartiality in an ethical view is important to the extent that it ensures patients are treated as equals, not that they receive equal treatment. This leaves us with two desiderata normally associated with agent-neutral views: a shared sense of morality that allows for mutual concern and impartiality in terms of treatment as equals. In what comes next, I will argue that when agent-relativity is cashed out in terms of reasons, these desiderata can be captured, whereas a value-based view of agent-relativity ultimately fails to meet the challenges raised for agent-relative views here.

III. Deontology and agent-relative reasons

The above discussion worked to show what is sometimes thought to be left behind when we shift away from agent-neutrality to an agent-relative view of normative ethics. However, as already mentioned, many theorists concede that some sense of relativity is important in an ethical view. Although an agent-neutral view may be able to capture the intuition that I may not push a hiker off a bridge to save five others, such a view has a much harder time explaining why I may not kill one to prevent five other killings. That is because an agent-neutral constraint against killing calls for me to prevent killings no matter who is doing the killing. Agent-relativity, on the other hand, allows that I myself am not permitted to kill. Thus, if faced with killing one to stop five others from killing, I ought to refrain from killing. There are of course a number of issues that arise in the literature for such a view.Footnote ¹³ However, for the purposes of this article, grant that such relativity is desired in a normative view. The challenge I want to home in on is whether and how such a view might be able to accommodate features of morality normally associated with agent-neutral views as laid out above, while allowing for the relativity required to prevent my killing one to stop five other killings.

Consider first a deontological view that accepts agent-relative constraints, cashed out in terms of agent-relative reasons. On this view, an agent may not push the hiker, since there is a constraint against killing that generates an agent-relative reason that cannot be outweighed by the prospect of either saving five lives or preventing five other killings.Footnote ¹⁴ Constraints, on this view, are restrictions on the actions of particular agents, not a call for agents to see to it that there are fewer killings in the world by other agents. That is, a reason is not generated neutrally for me with respect to every possible killing.Footnote ¹⁵ Rather, a constraint against killing captures the idea that I have strong moral reasons to not myself kill.

Turn now to the target desiderata laid out above. Consider first the idea that morality is in some sense a shared endeavor, one that allows for and motivates moral education. To see the concern for a relative view more clearly, consider an agent-relative version of a constraint laid out above:

On such a view, I am directed to be concerned with my own behavior, not yours. In this way, morality hardly seems a shared endeavor. Moreover, if I am concerned with my own behavior, I have no reason to educate others. Indeed, one might argue that I should mind my own business.

Despite these initial concerns, there are two ways in which I suggest that agent-relative constraints cashed out in terms of agent-relative reasons might ultimately capture the shared sense of morality that Korsgaard calls for. First, consider an indexically formatted version of constraints:

ARC applies to all agents in the same way across all situations. What differs is simply what is substituted into each variable in particular situations. While agents may be given different reasons in different situations, what remains consistent is that every individual has the same reasons to act when they are in the same situation.Footnote ¹⁶

Just as with moral patients, what might ultimately be important is that agents are treated alike in like situations.Footnote ¹⁷ The point of laying out constraints in indexical form is to elucidate this very idea. We can see how our reasons are in some sense the same reasons. What changes is simply the situation that each of us is in. We all share in the agent-relative reason to not ourselves kill. What changes from person to person is simply which of us is substituted in for “A”. In this way, we can again see morality as a shared endeavor. Moreover, once I take myself out of the equation, there is a shared framework that can be taught across persons.

There are two concerns one might raise at this stage. First, one might point out that the indexical form of a reason is not a reason at all. Until an agent is substituted into the indexical form, there simply is no reason. As such, agents do not share reasons simply by identifying the indexical form from which their reasons are generated.Footnote ¹⁸ I take the point, here. The indexical form is not itself a reason. Perhaps my point is better made with a sports analogy.Footnote ¹⁹ Consider a game of baseball. There is a sense in which the two teams do not share in an aim: each team's goal is to win, which requires the other team to lose. However, there is another sense in which the game is a shared endeavor. The teams play the game together by agreeing on rules and hold each other and themselves accountable for respecting the shared end of playing a game of baseball. Morality is like this. Morality is a shared game, with rules that we hope all will abide and individual aims that we are given, which may conflict. We can discuss and share in these rules and aims in the abstract, while acknowledging that in practice, they may apply to each individual in ways that may be incompatible. The indexicality of agent-relative reasons, I suggest, allows for this kind of sharing.

Granting the above, one might still, worry that the incompatibility of our individual aims makes morality a personal, not a shared endeavor. We have the same reasons in some sense. Yet, those reasons are still, for each of us, rather isolating. It remains the case that I, myself, ought not to kill, while you, yourself, also ought not to kill. We can have a conversation about how those reasons stem from the same framework, but we still lack a motive with which to aid each other's moral development. After all, it is rare, indeed, that the runner on the basepath stops to let the opponent run down a misjudged baseball. Here, I agree with my opponent that agent-relativity may not get us fully there. However, this is only a problem for a view that holds that our moral reasons are exclusively agent-relative. I simply deny that this is the case. A moral theory requires both agent-relative and agent-neutral elements in order to capture all of the features of commonsense morality. If we are sticking with the sports metaphor, the reasons that drive players are not only those that aim at victory for their team. They also have reasons to support the sport as a whole.Footnote ²⁰

Consider ARC above. On this view, the very strong reason that I have to not, for instance, kill an innocent person is captured in this constraint that I not, myself, perform this action. This reason stands, but it is not the only reason at issue. I contend that the same motivation that generates these agent-relative reasons, for instance concern for persons, also generates other agent-neutral reasons, reasons we all have to minimize infringements in general.Footnote ²¹ Of course, for the constraint to hold, I cannot kill another in order to prevent others from killing. Yet, it remains the case that I have reason to prevent other killings. If I can stop you from killing by giving you a moral pep talk, then I certainly have reason to do so. How these reasons all weigh together is a topic for another day. What matters for our purposes is that a view can embrace both agent-relative and agent-neutral reasons.Footnote ²² In this way, the agent-relative reasons need not provide all of the benefits; they simply cannot interfere.

Turn now to the second desideratum. Recall that a kind of impartiality is required insofar as no one individual is more important than any other. What is required, then, is that we are in some sense treated equally, that is, treatment as equals, as opposed to equal treatment. Consider a patient-relative version of the duty we have been discussing:

Is this consistent with being treated as equals? This response is likely more controversial, but I think it is. After all, if you were the mother to my children, then you would justify the same treatment by my children as I do, and likewise if I were the mother to your children. To see this, consider the indexical form:

If different patients are in different situations, they may be treated differently. However, any patient placed in the same situation will have the same rights as any other. Relativity, on this view, allows that my relationship to an agent may be different from yours, and so we may be treated differently. However, a sense of impartiality is maintained insofar as it is not you or I who makes the difference. If our positions were switched, so too would be our rights.

Although this patient-relative move is parallel to the agent-relative move above, one additional concern may be raised. I will admit at the start that I do not have a wholly satisfactory response to the following concern, except to suggest that it is a shared issue. Consider again PATIENT-RELATIVE MOTHER CONSTRAINT. I argued that insofar as you would justify the same treatment by my children as I do, if you were the mother to my children, and likewise if I were the mother to your children, there is a very real sense in which we are treated equally by the patient-relative constraint. But what if I am not the mother to any children? Of course, there are other duties and constraints out there that I might find protection under, but what if I lack family ties, friendships, and community altogether? While it is true that if I had those relationships, I could be plugged into a patient-relative duty or constraint of some kind, the fact is that I just do not get plugged in. In this case, it seems that I am not treated as well as others who do get plugged into patient-relative equations. Those others are treated as if they are in some sense more important than I, or so the worry goes.

I do worry about this objection. Patient-relativity affords everyone treatment as equals according to the rights they possess and the claims they can legitimately put forward. What this allows is that those with fewer ties have fewer rights and fewer claims they can make against others.

As a first pass to a response, note that not all moral reasons are patient-relative. Indeed, most are not. Even if I do not have any friends and family left in the world, I retain my humanity. As such, I am afforded a great deal of rights that allow for a host of claims against how others treat me.

A second response is more of a promissory note. This is an issue that must be dealt with by any view that maintains there are morally important patient-relative components to a normative view. I think it is fair to lean heavily on the rights afforded to everyone regardless of whatever special relationships they might have. However, the above concern must be at the forefront of our discussion when we develop our views of rights and duties, both what they are and how they weigh. If I have a duty to my community that can outweigh my general duty to humanity, then we need to be careful concerning what constitutes a community. It seems morally permissible to favor my family over strangers, but certainly not my race, ethnicity, gender, or sexuality.

Where does this leave us? I have argued that ARCs, cashed out in terms of agent-relative reasons, can, for the most part, capture the desiderata normally associated with agent-neutral views. By invoking indexicality, and allowing for agent-relative reasons to coexist alongside agent-neutral reasons, a deontological view is able to provide a truly shared endeavor as well as a strong motivation for moral education. On the patient-relative side, I argued that the sense of impartiality defended above is maintained insofar as all individuals ought to be treated the same in like situations.

IV. Consequentialists and agent-relative-value

Turn now to the consequentialized counterpart of the deontological view described above. On this view, agent-relativity is approached in terms of agent-relative value, rather than reasons. The question is whether such a view faces additional or alternate challenges with respect to capturing the desiderata at issue.

Traditional consequentialist views maintain that there is an agent-neutral value ranking of states of affairs. Agents are directed to perform whichever action in their power will lead to the state of affairs ranked highest in terms of value. In order to consequentialize agent-relative constraints, it is required that each individual agent is presented with a separate ranking of outcomes, in which goodness is relative to that particular individual. What is “good-relative-to-me” is different from what is “good-relative-to-you” such that an outcome in which I violate a constraint is worse-relative-to-me than one in which you do so, while the opposite is true relative-to-you. In this way, it is worse-relative-to-me, and so impermissible, for me to kill one to stop five others from killing, thus capturing the same results as agent-relative constraints.

As an agent-relative view, this brand of consequentialism will face the same concerns that agent-relative deontology faced above. If every individual has a separate ranking of states of affairs to promote, then morality no longer seems a shared endeavor. Moreover, simply by working to capture agent-relative constraints, moral patients will not be treated in a clearly impartial manner. In response, of course, this view can make the same moves that I made on behalf of the deontologist above. The consequentializer can make use of indexicals: for example, agents ought to promote that which is good-relative-to-x. She can also incorporate both agent-relative and agent-neutral conceptions of value, in order to foster a shared vision of morality more generally and provide a stronger motivation for moral education. Finally, how a moral patient would affect the relative rankings of states of affairs will likewise model treatment as equals. In these ways, the consequentialized counterpart, prima facie, is able to deliver the same results I argued deontology delivers above.

To see whether these moves succeed for agent-relative value, consider first the move to indexicals. On the agent-relative reasons view above, agents share reasons insofar as they are faced with the same set of indexical reasons that they substitute into depending on their situation. When faced with whether to kill one in order to stop five other killings, each of us can look directly to the relevant constraint, plug ourselves into the equation, and figure that reason into our decision-making process. Agent-relative value, even with indexicals, works differently. Importantly, agent-relative consequentialism requires that there is an entire set of rankings of states of affairs for each individual in every situation. That is, what is good-relative-to-me does not simply home in on whether I am permitted to kill one to stop five other killings. The rankings must consider every morally relevant factor in my environment, and provide a ranking of every possible outcome. Why does that matter? By requiring rankings of entire states of affairs it is hard to see how we can share in anything. It is true that if I am in your exact circumstances, my ranking would match yours. However, how often are we in identical circumstances? Even if we each are faced with killing one to stop five other killings, other factors in our individual situations will affect the overall rankings. Granted, it is still likely to be the case that the worlds in which we kill the one rank lower than worlds in which we do not. But, by providing full rankings, individual moral considerations are much more difficult to abstract and share across persons and situations.

Of course, one might flip this around and press a similar concern with respect to agent-relative reasons. It remains the case that there are morally relevant features of individual circumstances that will differ. Those morally relevant features will likewise need to be considered by the deontologist. So, why is this only a concern for the consequentialist? In response, return to the sports analogy. While both teams share the goal of playing the game together, individuals have their own lives and face all sorts of concerns outside the game at issue. Yet, when it comes to stealing second base, each of the players can home in on the specific game-related reasons for that individual to steal, and cooperate in that aim, or at least accept the outcome. In a game debrief, both sides can come together and focus on that play. I suggest that framing things in terms of reasons allows for this breakdown. Consequentialism, with its full rankings of states of affairs, plays out more like a single-player video game. Someone can look over your shoulder and try to help guide you on your personal quest, but she has to immerse herself in your entire world. Meanwhile, she also has her own console back home, with its own complete world that she must play.

The main thrust of this concern is that reasons allow, and rankings of states of affairs do not allow, agents to simply abstract away and consider each morally relevant consideration on its own. I may never be faced with all the same reasons that you are faced with at any given moment. But when we deal in reasons, we can better see the overlapping concerns. We can then share in those parts of our moral deliberations. When we deal in rankings of states of affairs, all of the ways in which our situations differ must be built right into each state of affairs that is ranked. Thus, why I am faced with a particular ranking will not overlap with yours in the same way that my reasons might. In this way, the good-relative-to is much more isolating.Footnote ²³

A similar concern arises when we look at the second move made: incorporating both agent-neutral and agent-relative elements. As suggested above, noting that an important facet of morality involved agent-relativity need not require a complete rejection of agent-neutral reasons. Indeed, I argued that the motivations for agent-relative constraints likewise provide motivation for agent-neutral reasons to minimize violations of those constraints.

At the start of this section, I suggested that the consequentializer could make a similar move by including agent-neutral value. After all, if the deontologist can ask agents to weigh two kinds of reasons, it seems that the consequentializer should be able to ask agents to weigh two kinds of value. However, this move is problematic when applied to the consequentialist methodology. The agent-relative consequentialist calls for agents to perform whichever act maximizes agent-relative value. It is unclear, then, how agent-neutral value would fit into an agent's deliberations.

Suppose I am faced with the prospect of killing one person to stop five others from killing individuals. As a deontologist considering agent-relative reasons, I recognize that I have a strong agent-relative reason to not kill the one (see ARC). I also recognize that I have ordinary agent-neutral reasons to save the five, as well as reason to prevent killings. Indeed, I recognize that there is a very real sense in which it would be a better result, there would be more (agent-neutral) good in the world, if I killed the one. However, for ARC to function as a constraint, I ultimately understand that my agent-relative reason trumps the agent-neutral reasons in this case. If the case were modified, and I could prevent five killings without killing anyone, then I would simply be faced with the agent-neutral reasons to save lives and prevent killings, and would proceed accordingly. In this latter case, no agent-relative reasons obtain.

Now consider the agent-relative consequentialist in these two situations. According to the view, I am to perform whichever action leads to the state of affairs ranked highest with respect to value-relative-to-me. Thus, I simply consider how much good-relative-to-me each option would generate, and perform the action accordingly. I am not faced with any agent-neutral consideration whatsoever, just whatever is agent-relatively best. Granted, I could note that the world in which I kill the one is ranked higher with respect to agent-neutral value than the world in which I refrain from killing the one, but that's of no action-guiding moral concern to me. Interestingly, the same is true in the second case. When faced with the prospect of saving five lives without having to kill anyone, the deontologist is faced with nothing but agent-neutral reasons. However, for the agent-relative consequentialist, it is still a matter of what produces the best result with respect to what is good-relative-to-me.

In sum, I can note that the agent-neutral value of a given state of affairs is greater than the agent-relative value, but my action must be guided by the agent-relative value regardless. There is no real weighing of the two. Of course, agent-neutral considerations can be figured into one's agent-relative rankings, but insofar as the two collapse, we are left with simply agent-relative value. Perhaps we can share in an acknowledgement of what is agent-neutrally best. However, insofar as that does not directly affect our deliberations, it is unclear what role that really plays.

Recall the desiderata that inclusion of agent-neutral considerations was meant to capture. With only agent-relativity on the table, the concern was that even if indexicality allowed for meaningful conversation about shared reasons or rankings of value, there would be no real motivation to have those conversations. It remains the case that my reasons are about my actions, and the evaluation of states of affairs that ought to guide my actions is a ranking that is relative to me. Why, then, should I concern myself with your actions? The agent-relative consequentialist must still answer this question. Even if we can agree on what is agent-neutrally best, my actions are not to be in any way guided by that knowledge. As such, what is good-relative-to-me remains my sole focus, and morality continues, on this view, to be isolating.

At this stage, one might recognize that both challenges for the agent-relative consequentialist are really issues with the methodology of ranking states of affairs. If we can approach agent-relative value differently, perhaps we can save the consequentialist.Footnote ²⁴ The idea is that instead of placing all of the relevant considerations into one ranking, the agent-relative consequentialist considers “valuable features” of outcomes. On this view, an agent could recognize that her own killing is of agent-relative-disvalue, while saving lives is of agent-neutral value. Valuable features, then, could be indexicalized. Moreover, when weighing the valuable features of outcomes, agents could consider both the agent-relative and agent-neutral features of the respective possible outcomes and act accordingly.

On the whole, I concede that such a view could handle the agent-relative and agent-neutral features of commonsense morality in much the same way that I argue the agent-relative reasons wielding deontologist can. However, I have concerns. To start: what remains of the simple “do the most good” adage of consequentialism? Indeed, what is an agent to do in these situations? One of the benefits of ranking states of affairs is that the decision-making procedure is simplified. Once ranked, the agent simply performs the action that leads to the top-ranked state of affairs. With the ranking gone, what is the consequentialist to do? If only agent-relative value were at issue, the directive could remain, “do the most good”. However, we are now considering a view that takes both agent-relative and agent-neutral goodness seriously in moral deliberations. How are those considerations of goodness to be weighed? In what way could the consequentialist continue to be a “maximizer” of value?

Granted, the deontologist must answer the same questions with respect to reasons. Yet this is familiar ground for the deontologist. After all, the deontologist never directed agents to maximize anything. Her directives were always a matter of weighing different considerations of neutral value with relative-reasons stemming from considerations of dignity and the like. The new “valuable features” consequentialist could, of course, co-opt the deontologist's weighing mechanisms, but then it is unclear what consequentialism is bringing to the table. Not only does it no longer resemble the consequentialism we once knew, but it also seems to have lost one of its main advantages: maximizing.

Of course, the consequentialist is still bringing one major feature to the table: value. Perhaps maximizing is lost, but on this view, what one ought to do will still be cashed out in terms of value explanations.Footnote ²⁵ What one ought to do is a matter of (somehow) weighing the valuable features in different possible outcomes. I wonder, though, how we are to know what the valuable features of outcomes are. When it comes to the agent-neutrally valuable features, things are simple enough. Candidates are likely to be features such as the amount of total happiness in the world or the extent to which that happiness is distributed equally. But what of the agent-relatively valuable features? What makes a particular feature of an outcome contain goodness-relative-to-me? Recent work on agent-relative value has seemed to favor buck-passing views in which a state of affairs, or for our purposes a feature of an outcome, has agent-relative value when it would be fitting for me to prefer it, or when I have reason to desire it.Footnote ²⁶ The worry is that while these strategies provide a fine mechanics for the view, they do not provide content. To say that a state of affairs is good-relative-to-me is to say that it is fitting for me to desire that that state of affairs obtains. But when and why is it fitting for me to desire that such a state obtain?

To answer this question, we will ultimately need to look to the underlying deontological explanations. Why does the deontologist argue that it is wrong to kill one to save five? As an example, say that it is because one's dignity is inviolable, such that we cannot use an individual to further one's ends, however noble they may be. From there, I suppose the agent-relative consequentialist can argue that the inviolability of one's dignity is such that it would be unfitting for me to desire that an outcome have the feature of my killing one, even to save five. This seems to work. However, it is unclear to me that we are left with anything but a terminological dispute. Once we disaggregate the considerations of value, and generate which features of outcomes are valuable by looking at the underlying deontological explanations, consequentialism has lost its attractive methodology and value has lost its independent footing. Whether the underlying explanations generate agent-relative reasons or agent-relative values at this stage seems to make no discernible difference.Footnote ²⁷

V. Leaving agent-relative value behind

Where does all of this leave us? I began with the thought that the move to agent-relativity in consequentialist ethics seemed to leave important desiderata of normative ethical theories behind. Agent-neutral views are known for easily capturing particular features of morality, such as its shared nature and its impartiality. The question, then, was whether an agent-relative view, deontological or consequentialist, could capture these desiderata. With respect to impartiality, agent-relative views of both sorts can capture a sense of impartiality once we recognize that commonsense morality requires treatment as equals, not equal treatment. As for morality's shared nature, through recognizing both the indexicalized form of agent-relative reasons, as well as the importance of agent-neutral reasons, a deontologist can capture the shared nature of morality.

As for agent-relative consequentialism, we are left with two possibilities. On the one hand, the consequentialist can hold on to the methodology of choosing whichever action leads to the state of affairs ranked highest with respect to agent-relative value. In this case, a shared sense of morality is lost and the view is unable to incorporate necessary agent-neutral features of commonsense morality. On the other hand, the consequentialist can give up on this aggregative methodology and simply look to particular features of outcomes. Here, the consequentialist will ultimately need to look to the deontologist both to settle the weighing of different kinds of value, as well as to discern what counts as agent-relatively valuable. In so doing, the consequentialist has given up on the features that made it an attractive view, and we are left with a mere terminological dispute over whether underlying moral explanations generate “reasons” or “value”. That is, this new consequentialist ultimately wields value in the same way that the deontologist wields reasons. As such, it is unclear what is left of the “value” in her value-explanations. Either way, it seems, agent-relative value is left behind.Footnote ²⁸