Certain activities change us. We emerge from them altered, out of touch, diminished, or even morally corrupted. Whether returning from a warzone or from an experience as common as caring for an aging parent, one might remark, “I’m not the same person I was.” We do not normally take such utterances as expressing anything deep or precise about personal identity; rather, they convey simply that the person has undergone a transformation that has implications for her sense of self. I aim to argue that such transformations are relevant to debates about what morality requires of us. To undergo an alteration in one’s self is to make a special kind of sacrifice, a sacrifice of self. Since projects are more or less likely to be obligatory to the extent that they require more or less sacrifice, we must incorporate these unique sacrifices into any accounting of the contours and limits of moral obligation. But sacrifices of self pose a special difficulty for any such accounting, precisely because of their transformative nature.

There is a longstanding debate among moral philosophers over how much we are required to give up for the sake of morally worthy ends. One way of approaching these questions is to claim that if there are limits to what morality can require of us, they are determined by (among other things) trade-offs between the moral impact of a given action and the sacrifice involved.Footnote 1 In other words, actions with relatively high moral impacts that require relatively low levels of sacrifice are more likely to be obligatory than actions with the opposite profile.Footnote 2 Alternatively, we might claim that certain types or amounts of sacrifice can render an action nonobligatory regardless of the action’s moral value or impact. This could be because there is some absolute amount of sacrifice that it would be unreasonable to require an agent to bear, or because there is some particular sphere of life (such as family or personal projects) that is considered worth protecting from morality’s demands.Footnote 3

Both of these approaches to questions about the demands of morality require a way of understanding what sacrifice is. Elsewhere I have argued that all sacrifices can be understood as gross losses of wellbeing. If our theory of wellbeing is objective, then we can make sense of how personal sacrifices might count as legitimate constraints on moral obligation.Footnote 4 Yet it is not clear that the wellbeing theory of sacrifice can accommodate sacrifices of self. While a sacrifice of health, happiness, or knowledge might best be understood as a loss of wellbeing, sacrifices that involve transformations or diminishments of the “self” are more than just losses of wellbeing. This paper focuses on two types of sacrifice of self, those involving changes in or damage to identity and to moral agency. These sacrifices change the very constitution of the person on whom all other losses and gains are imposed. To make these sacrifices vivid, I examine how they arise amongst undercover law enforcement agents, humanitarian aid workers, and soldiers. Although these professions are unusual, sacrifices of self are a risk of a wide range of moral projects, both ordinary and extraordinary. If we want to understand moral demands as costs-to-the-agent, we must accommodate these unique costs.

The paper has four sections. In Sect. 1, I distinguish “sacrifices of self” from what is often called “self-sacrifice” in the philosophical literature, and suggest that the conversation about morality’s demandingness has been hampered by its overemphasis on material and financial sacrifices. In Sects. 2 and 3, I appeal to real-life examples and empirical research to argue for the existence of two main types of sacrifice of self: changes in or damage to identity and to moral agency, respectively. In Sect. 4 I argue that these sacrifices cannot be reduced to losses of wellbeing, and are extraordinarily difficult to compare with other losses or with moral gains. I consider a proposal for giving sacrifices of self a special trumping value in moral calculations, but show that this proposal is unlikely to succeed, and so a puzzle remains.

Sacrifices of Self, “Self-sacrifice”, and Morality’s Demandingness

Sacrifices of self are a subset of all sacrifices: those involving a significant change in or damage to the self. It is tempting to simply refer to them using the familiar term “self-sacrifice”. Indeed, Connie Rosati (2009: 313) has argued that “the key to understanding self-sacrifice is, so to speak, to put the self back into the sacrifice”. She claims that to count as a self-sacrifice, an act must “involve a sacrifice of one’s interests that is at the same time a sacrifice of self.” (Rosati 2009: 316) Rosati’s paradigm cases of self-sacrifice involve “life, limbs, and loves.” (Rosati 2009: 317) We have strong interests in continued existence [life], in physical integrity and functioning [limbs], and in certain of our “projects and pastimes, undertakings and relationships” [loves], and these interests are all deeply connected to the self. (Rosati 2009: 318)

Following Rosati, I will argue that we should be paying special attention to sacrifices that involve the self, and this paper owes its title to her turn of phrase. However, my view departs from hers and others in this literature in two central respects. First, I argue that we can incur damage to the self not just indirectly, as when we lose a life, limb, or love, but also directly, when our identity or moral agency is threatened. Second, in a departure from Rosati and others who have written on self-sacrifice, my account allows that agents might make morally relevant sacrifices unwittingly. Thus I avoid the term “self-sacrifice” and use the more neutral “sacrifice.” It is not always clear in the philosophical literature whether “sacrifice” and “self-sacrifice” refer to distinct concepts.Footnote 5 To the extent that they are treated as distinct, the central distinction seems to be this: self-sacrifice is an especially praiseworthy type of sacrifice which an agent imposes on herself and in which she knowingly gives up something which is hers, broadly-construed, such as something that she values, some part of her body, or her very life. Some recent accounts of self-sacrifice set an even higher bar, requiring that self-sacrifice meet some or all of the following conditions: it must be voluntary, anticipated, intended to benefit another person, result in a net overall loss to the agent, or involve harm or risk to the agent’s very life or self.Footnote 6

But for the purpose of exploring the role that sacrifice plays in mitigating the demands of morality, a broader and weaker notion of sacrifice is more useful. We must pay attention to even those sacrifices that are not voluntary, not anticipated, that do not result in a net loss, and that do not risk or result in total annihilation of one’s life or self. The agent need not explicitly choose someone else’s good over her own, or even realize her own good is threatened, for the loss to count as a sacrifice. Sometimes we take on admirable moral projects without realizing, or perhaps without even having any reason to suspect, that they will harm us. And sometimes we take on losses that do not result in a net overall reduction in wellbeing, because they make possible other gains. Such losses nevertheless count as sacrifices, because even when the gains are commensurable and comparable with the losses, they often cannot strictly be said to compensate for the losses.Footnote 7 To chart the boundaries of moral obligation, we need to account for all the costs agents face, including costs imposed on the self.

The literature on morality’s demandingness asks what morality requires of us, and thus at least implicitly asks what sacrifices we must make. But the most famous examples in this literature seem to focus on the loss of stuff: money, material possessions, and recreational activities.Footnote 8 More recently, some have drawn our attention to the sacrifices involved in being morally conscientious about the things we eat and purchase and the ways we travel and consume energy.Footnote 9 At the more extreme end of the spectrum are losses of body parts or even one’s life.Footnote 10

Perhaps it makes sense that material possessions have become the classic examples of potential moral sacrifices. When he pointed out that you must muddy your clothes to save a drowning child right in front of you, Peter Singer (1972) established a baseline of trivial sacrifice against which other more demanding losses might be compared. When only material possessions or money are at stake, it is easy to carefully titrate the losses in successive examples in order to seek thresholds or limits. It is also easy to stack the deck in favor of moral obligation: descriptions of material possessions (a new flat-screen TV) and recreational activities (trips to the opera) can trigger stereotypes about extravagance and snobbery independently of any consideration of how else the money or time might be spent. Peter Unger (1996: 135), following in Singer’s footsteps, asks whether you must sacrifice the wooden trim on your prized vintage boat in order to save a stranger’s life, and ultimately whether you must give away “most of [your] financially valuable assets, and much of [your] income”. Vintage boats and “assets” are the trappings of privilege. It is not surprising that we can convince ourselves that we are not entitled to keep them. But there is more to the problem of morality’s demandingness than what is captured by thinking about the purchasing power of surplus wealth. Some argue that we must accept radical changes to our economic circumstances for moral reasons. Must we ever accept radical changes to ourselves?

Threats to Identity

The first and most obvious way in which one might make a sacrifice of self is by undergoing a change in one’s identity in the course of pursuing a morally worthy project. Here I use “identity” not to refer to the philosopher’s narrow concept of personal identity, but to the broader notion that consists of one’s personality, self-conception, self-presentation, and the ways these notions interact with each other and persist or change over time. One’s “identity” in this sense includes both the kind of person one considers oneself to be and the particular being in the world with whom one identifies. Changes in identity can be partial, as when one says “I’m not the same person anymore—I’m angrier,” or complete, as when one loses all ability to distinguish between one’s true self and a false persona adopted for the purposes of undercover work. Changes in identity can also be gradual or sudden, temporary or permanent, noticed or unnoticed, and welcome or unwelcome.

Consider an example taken from the world of undercover law enforcement. William Queen, an agent with the Federal Bureau of Alcohol, Tobacco, and Firearms, signed up to go deep undercover in an attempt to bring down the Mongols outlaw motorcycle club in Southern California in the late 1990s. Reflecting on the operation later, he explicitly described it using terms like “sacrifices” and “demands.” (Queen 2005: 53–54) He had known it would be tough, and volunteered for the job thinking he could be the one who could “hold it together under circumstances of extreme deprivation, isolation, and paranoia.” (Queen 2005: 27) But he was nevertheless surprised by how much it changed him, and he regretted the decision.

Queen was required to adopt an all-consuming false identity as a prospective member of the club. He cultivated a disheveled appearance to blend in at carefully selected motorcycle bars. He took on a fake apartment, fake job, fake criminal record, fake school transcripts and IRS forms, and even a fake family. Law enforcement agents around the country sat poised to answer the phone pretending to be his parents or relatives, if the Mongols’ private investigator called during the club’s vetting process. Despite his prior undercover experience, Queen found maintaining a false identity to be extremely difficult. “It can be emotionally debilitating,” he writes, “to show the gang one side of your personality, while carefully masking who and what you really are, from what you’re feeling to what you actually stand for.” (Queen 2005: 35)

Queen had always thought of himself as one of the “good guys”. He not only had to suppress and hide this aspect of his self-conception, but had to actively pass as one of the “bad guys”. When he came to find that even the bad guys had some sympathetic qualities, including “love, loyalty, [and] respect,” he felt an understandable cognitive dissonance. (Queen 2005: 167) Ultimately, after being accepted into the Mongols and rising to a leadership position, Queen found that his double-life was not just stressful and exhausting, but damaging at a deeper level: with his appearance radically altered, he lost the ability to live as himself, even when off-duty. Not being able to present his true identity to the outside world left him alienated and confused. Here Queen explains this damage to his identity:

There was one thing I didn’t realize when I accepted this undercover assignment: how much it was going to change me. I’d been all fired up to infiltrate the Mongols and do the grueling investigative work; I’d been mentally prepared for the danger, the violence, the guns and drugs, the challenges to my undercover identity.

What I hadn’t fully anticipated was the emotional turmoil. I really had to abandon all semblance of my personal life for the duration of the undercover role. […] Before you can become a full patch, they make copies of your picture and distribute it to every other chapter in the club; every patch studies your face closely and considers whether they have a problem with you or may have known you at some point. Every patch stares hard at you and weighs the possibility that you might be a cop trying to infiltrate the club.

[…] You realize that even in a state the size of California, you’re trapped in a kind of prison of your own making—you can no longer pick up your kids and go to a movie or amusement park; you can’t walk into a restaurant […] Holidays, weekends, birthday parties, baseball games, soccer games, the weekly ‘Dad time’—I had to give it all up.

[…] my hair got longer, my beard got longer, my sleepless nights longer. There was no such thing as downtime or being off-duty. Because of my appearance, I couldn’t go anywhere and be considered a member of mainstream America. In my mind, of course, I remained a federal agent, a decorated military veteran, a clean, caring, conscientious parent. But my offensive appearance announced something quite different to the world. (Queen 2005: 179–180, emphasis added)

What Queen found most challenging about this assignment, it seems, was not the risk or the violence or the long hours, but his inability to reclaim himself and redeem himself, even temporarily, by presenting his true, non-criminal identity in public. What he may not have foreseen was just how much his identity as a “clean, caring, conscientious” parent and federal agent was tied to his physical appearance. With his long, unkempt hair and long beard, even going to a parent-teacher conference was a kind of psychological torture.

Queen is not unique. Indeed, he may have escaped relatively unscathed. There are documented cases of undercover agents who lost the ability to distinguish between their true and false identities. One British Secret Service agent from World War II experienced the “uncontrolled re-appearance” of his undercover identity after he had finished his assignment. (Girodo et al. 2002: 632) Agents in the Israeli Mossad have displayed “role confusion” and “alterations in identity,” with one becoming “unsure of his true name and identity” after years spent pretending to be an Arab merchant. (Girodo et al 2002: 632) One longtime undercover FBI agent took a person hostage and was charged with attempted murder. During the hostage situation he negotiated with police in the persona of his “alter personality,” “Ed”. He claimed to have suffered Dissociative Identity Disorder. (Girodo et al 2002: 633)

In an empirical study of identity disturbances in undercover work, Girodo et al. (2002) put 48 federal police officers through a 3-week undercover training course in which they practiced adopting a false identity. Many subjects reported that their “false identity reappeared in a nonoperational context without their having called upon it.” (Girodo et al. 2002: 638)Footnote 11 Observers substantiated this phenomenon; 66 % of the participants were seen acting in their false personas outside of the training course. (Girodo et al. 2002: 638) In keeping with Queen’s experience with his Mongol outlaw hairstyle, Girodo and colleagues also found that “Making changes to one’s physical appearance seemed to bring about greater self-reported identity reappearances” than did changes to walking style, speech patterns, facial expressions, or clothing. (Girodo et al. 2002: 641) The authors hypothesize that this is due to the fact that changes to physical appearance “accompanied the person both in and out of undercover settings” which may have led to an “identity response generalization.” (Girodo et al. 2002: 641)

Of course, one’s physical appearance is just one part—and some would say a relatively insignificant part—of one’s identity. What was damaging to Queen was not simply that he looked different, but that he looked—and acted—like a criminal. Elizabeth Joh (2009) has argued that the psychological complications of undercover work may be exacerbated by “authorized criminality,” which is the practice of legally permitting undercover agents to do things that would otherwise be criminal offenses. Authorized criminality “adds yet another layer of strain to this tangle of conflicting demands and loyalties by heightening role confusion.” (Joh 2009: 190) Such role confusion can have serious moral consequences:

When an agent is permitted in his official capacity to participate in crime, this may be justifiable and non-criminal as a legal matter, but to the agent, this authorized criminality is, in psychological terms, not a mere simulation. There is camaraderie in a band of thieves; participating in the same crimes as those being investigated increases the risk of over-identification with targets. Authorized criminality may also contribute to the “moral corrosion” of the undercover agent who is immersed in a world where ethics have already been compromised. (Joh 2009: 190–191)

The psychological strain of going undercover is not solely attributable to the fact that one is pretending to be someone else. Joh brings to light the added significance of the fact that the person one is pretending to be is—at least for most law enforcement officers—precisely the kind of person they do not want to be, namely a criminal.Footnote 12 Moreover, in committing (what would otherwise be) criminal acts, the undercover agent may feel that there is no practical difference between pretending to be a criminal and just being a criminal. An undercover agent working alone is unlike an actor in a play in one crucial respect: no one around him knows he is acting. Others take his false identity to be real, so it is understandable that over time he might himself lose track of the line between reality and fiction.

One need not pretend to be someone else to be at risk of a changed or damaged identity. Soldiers, for example, risk injuries and disabilities that involve not only pain and functional impairments, but also disruptions in one’s self-understanding as an able-bodied, athletic, strong, or independent person. Someone who enlists during wartime may understand that one of the many risks of her job is losing a limb; but she may fail to anticipate just how much of her identity is tied up with her ability to navigate the world without being seen by others as different or needing assistance. And of course, one need not have one’s bodily integrity or appearance altered; simply experiencing trauma can itself damage the self. Even those soldiers who recognize the considerable risk of returning from war with Post Traumatic Stress Disorder may not appreciate how it will affect their sense of self.Footnote 13 It is common for trauma survivors—including veterans, Holocaust survivors, and rape survivors—to claim that their trauma rendered them a different person. (Brison 2002: 38) Becoming a different person is, like death, a higher-order loss—a loss of the subject of a life. Unlike in death, though, something is left behind—the new, or at least changed, self. Given this fact, how can we reconcile the risk of this type of loss against the value of morally worthy pursuits? Agents certainly cannot estimate this cost before acting. They are in a position akin to that of prospective parents, who simply cannot know what it will be like to have children until they have them, at which point it is too late to change their minds. (Paul 2015)

What is true of undercover agents and soldiers may also be true of humanitarian aid workers, who risk emerging from their work with the sense of having changed for the worse. As one doctor working with Médecins Sans Frontières (MSF) put it in the documentary Living in Emergency (2008), “Helping to stop the suffering of people is tremendously rewarding, although also fucks you up a bit.”Footnote 14 Another MSF doctor, Tom Krueger, describes his reaction to his first medical mission in the mid-2000s:

I was pretty much in shock when I got there. I mean if you can talk to some of your friends about some of the stuff you saw, and … You can’t describe the smells, the feeling of the heat on your body, the sweat running down your back, the smell of the pus that hits your nose, the unwashed bodies in a closed room, you know, no circulation, the smell of your own panic, you know, when you’re not sure what to do… You can’t share that stuff. (Hopkins 2008)

The shock and exhaustion of confronting insurmountable levels of human suffering can have long-term implications for aid workers. This seems especially likely when the human suffering is caused by the actions and inactions of other human agents. During the 1994 genocide in Rwanda, Dr. James served as the MSF Head of Mission, treating scores of victims as the world around him erupted in chaos and death. Reflecting on that experience and others, he cites not sadness or exhaustion, but anger: “I still have, and I always will, I think, a nearly uncontainable rage about what happened in Rwanda and Somalia and many other parts of the world.”Footnote 15

Although this rage is surely warranted, and may even be the kind of driving force that allows Orbinksi to continue his extensive work on behalf of humanitarian causes, he would be better off—other things being equal—if he were not consumed with rage, especially not “uncontainable” rage. The interesting thing for our purposes is not simply that feeling rage is a negative, aversive experience—though it normally is. It’s that may have become a rageful person. When Orbinksi returned to Canada, he writes, “My mother said something to the effect of ‘You never should have gone to that place. You’re not the same.’”Footnote 16

Longtime OXFAM worker Tony Vaux has claimed that atrocities like Rwanda can lead humanitarian workers to a very dark place—not just a place of sadness, but a place of lost capacity for humor or joy and deep moral confusion.

Many aid workers became deeply troubled by their experiences and some lost faith in their work. Many sought counseling after their involvement and some found it difficult to adapt to normal life. Such experiences undermine the ability to laugh because we can see only the one overwhelming event, and all else loses perspective. We are unable to see the smaller shifts of perception that are the basis of humor. Without the benefit of a fixed morality, including its helpful notions of good and evil and its rituals of absolution, humanitarians can be plunged into despair by the smallest failure in human nature. (Vaux 2001: 186)

The aid workers’ lost faith in humanity is perhaps a trivial complaint compared with what befell the genocide’s real victims. Still, it is interesting that the salient sacrifices from this type of work are not (merely) the long hours, dangerous conditions, and lack of creature comforts. Some aid workers give up parts of their identity itself, including their sense of humor and moral compass.

Even situations far less extreme than Rwanda can leave aid workers changed. Aid workers who are exposed to trauma, either personally or vicariously through contact with victims, are at high risk for developing anxiety, depression, and PTSD. (Eriksson et al. 2001; Holtz et al. 2002) Even workers who are not exposed to trauma are at risk for “burnout”, a condition which may involve “emotional exhaustion,” “depersonalization,” and a diminished sense of “personal achievement” or lowered self-evaluation. Depersonalization, in particular, implies an interesting breakdown of the self—the person feels alienated from her own feelings and behaviors, as though they are happening to someone else, and may as a result treat others in a dehumanizing way. A recent large study of NGO workers from nineteen different organizations found that after a deployment, workers are at increased risk of anxiety, depression, lowered life satisfaction, and burnout. (Cardozo et al. 2012)

Of course, many of us will never need to decide whether to take on an undercover identity in pursuit of a morally worthy goal, to sign up for military service in wartime, or to join a humanitarian relief effort, leaving ourselves at risk of emerging a changed person. These are obviously remarkable cases. My goal is not to make a point that applies only to remarkable cases, but rather to use such cases to make vivid an underlying phenomenon that may be present in lesser degrees in more common circumstances. There are numerous examples of activities or choices that threaten to leave us confused or uncomfortable with respect to some particular aspect of selfhood. For example, reproductive choices—including the choice to have an abortion, to give an infant up for adoption, to act as a gestational surrogate for someone else, or to donate one’s gametes—can leave one confused about one’s identity or status as a parent.Footnote 17 We are socialized to think of parenthood as one of the central organizing features of our identities; ambiguity or angst about one’s status as a parent thus introduces a kind of instability in one’s sense of self. Although these choices may result in a profound loss of personal wellbeing, they also change the very subject on whom losses and gains are imposed. Since reproductive choices are often considered to be moral choices, the resulting sacrifices of self need to be included among the costs associated with complying with morality’s demands. Yet, as I argue in Sect. 4, it is unclear just how we ought to account for these costs.

Threats to Moral Agency

I have suggested that some moral projects might involve compromising an agent’s identity. But sacrifices of self need not involve the whole person, as it were, nor need they challenge her very sense of who she is. Some sacrifices change the person qua moral agent. When we take on new projects or lifestyles, our moral agency itself might be at stake. By “moral agency” here I simply mean those aspects of the self that are involved in moral choice, moral behavior, moral judgment, and the feeling or expressing of associated reactive attitudes.

The most obvious threat to moral agency is anything that compromises a person’s autonomy. Autonomy, or the capacity to govern oneself and one’s actions, is arguably both a component of human wellbeing and a prerequisite for many other sources of wellbeing. Threats to autonomy thus entail threats to wellbeing. And yet losses of autonomy do not merely decrease wellbeing, as do, say, losses of material possessions or recreational opportunities. Autonomy is connected to the self in ways that material possessions and recreation are not. In particular, autonomy is connected to the moral self because it is necessary for moral responsibility.

Susan Brison has articulated the way trauma in particular affects autonomy and thus the self:

[The autonomous self] is considered responsible for its decisions and actions and is an appropriate subject of praise or blame. It is the transformation of the self as autonomous agent that is perhaps most apparent in survivors of trauma. First, the autonomy-undermining symptoms of PTSD reconfigure the survivor’s will, rendering involuntary many responses that were once under voluntary control. Intrusive memories are triggered by things reminiscent of the traumatic event and carry a strong, sometimes overwhelming, emotional charge. Not only is one’s response to items that would startle anyone heightened, but one has an involuntary startle response to things that formerly provoked no reaction or a subtler, still voluntary one. The loss of control evidenced by these and other PTSD symptoms alters who one is, not only in that it limits what one can do (and can refrain from doing), but also in that it changes what one wants to do. (Brison 2002: 59–60)

One implication of Brison’s view is that to take on a pursuit where one might potentially witness or experience trauma is to risk not just one’s future wellbeing, but also one’s very capacity for moral responsibility. Of course, one can experience trauma in even the most mundane contexts—a trip to the store might lead to traumatic accident on a bridge, leaving the person with a paralyzing, autonomy-compromising fear of bridges.Footnote 18 But this phenomenon is most interesting to us in contexts where the trauma is part of an endeavor—like helping distant strangers, or serving one’s country in wartime, or putting one’s life on hold to care for a sick relative—with a controversial moral status. To determine whether such actions are obligatory, supererogatory, or something in between, we need to know what we give up in performing them. If one of the things you risk losing—your autonomy—is central to moral agency, then it seems we are faced with an odd trade-off between exercising moral agency in the present and preserving it for the future.

And there is more than just autonomy at stake. Some activities risk a kind of moral corruption that has significant implications for the self. We saw earlier that Queen had difficulty identifying with elements of his alter ego that he considered to be morally corrupt, and mourned his inability to present himself to the world as the “clean, caring, conscientious parent” he considered himself to be. He was also placed in various situations where he had to choose between blowing his cover and performing actions he considered to be morally blameworthy. These situations threatened not only his identity, but his moral identity in particular. Just imagine: you cultivate an identity as someone whose role and mission is to prevent, resist, discover, denounce, stamp out, and punish criminal behavior. Then you try to maintain that identity while befriending criminals, participating in crime with them, and conspiring with them to avoid detection by the police. This would seem to be especially harmful to one’s moral identity if one believes, as many law enforcement officers surely do, that most or all forms of criminality are not just illegal but immoral. Indeed, even an undercover agent who manages not to commit any crimes will need to lie and dissemble just to maintain the false identity. “Moral corrosion,” as Joh (2009) calls it, seems to be a minimal necessary condition just for showing up to work.

Behaviors like lying about one’s true identity and forming insincere relationships with people could be morally corrosive even when the agent believes them to be morally justified in a particular context. They are corrosive because they can develop into habits or tendencies that then manifest in other contexts where they are not justified. This has been noted in the case of authorized illegal drug use. Agents who use drugs to establish credibility with criminal targets are at risk of continuing to use drugs even when the operational context does not require it or the agency does not authorize it. (Joh 2009: 168) Even if we do not consider drug use itself to be seriously morally problematic, it allows law enforcement officers to practice criminality—to get comfortable with the idea of breaking the moral–legal rules they had internalized, and to break down the barriers between themselves and the criminals they are targeting. When an agent is then faced with an opportunity to commit an unauthorized criminal act—say, a violent assault—he may be less inhibited about doing so, having already crossed a line he formerly regarded as sacrosanct.

What begins as a change in behavior could easily morph into a change in moral judgment or even a change in the principles or theories that (implicitly) underlie one’s judgments. For instance, an agent might engage in bar fights in order to maintain cover. This brings a new comfort level with physical violence. The agent may seek rationalizations or justifications for the behavior, and consequently adjust his underlying (implicit) moral theory. In this case, perhaps, a behavior that was previously regarded as morally forbidden (namely, violence outside a self-defense context) could come to be regarded as justified in light of the overall moral value of a successful long-term undercover operation. Thus an action that was regarded as forbidden via a kind of deontological agent-centered restriction might now be regarded as justified via consequentialist reasoning. This shift from a more deontological to a more consequentialist stance may be prompted not by careful moral reflection, but rather by the desire to avoid dissonant thoughts or feelings.

A related phenomenon—another injury to moral agency—has been discovered in soldiers returning from war. Unlike the undercover agents, though, these soldiers did not necessarily personally commit any actions that they believed to be illegal, immoral, or otherwise against the rules. Instead, they returned from war feeling “tainted” by the morally dubious actions of others, or feeling “guilty” about bad things that happened to comrades through no fault of their own. One Vietnam War veteran tells philosopher Nancy Sherman that his feeling about participating in the war was not guilt or shame, but taint: “‘Taint’ implies pollution, staining, fouling. […] In the case of taint, the complicity is of the mildest kind. […] In [the soldier’s] mind, there is a passive yet pervasive association with a policy that is wrong or unjust. The tinge spreads, irrespective of his moral protest.” (Sherman 2010: 51–53) What phenomena like this show is that the moral self is quite sensitive, and can be damaged even by events for which the agent is not morally responsible. Of course, many Vietnam War era soldiers served in some sense against their wills, which may seem to entail that we should not count their moral wounds as sacrifices of self in the way that we do the wounds of the humanitarian aid workers. However, precisely because they were drafted, we can view their service as a moral obligation—a duty of citizenship, fulfilled in some cases with extreme reluctance. And this raises to salience our central question: how can we determine whether such an obligation is a reasonable one, if one of its associated costs is something so unquantifiable and intractable as moral taint? In the era of all-volunteer armed services, the question becomes even harder. Soldiers in Iraq or Afghanistan may have signed up because they felt called to contribute to what they believed was a just mission. When they later changed their minds about the mission, they felt not only tainted but also “suckered.” (Sherman, 2010: 53–58) For those who signed up out of explicitly moral motives, to come home feeling tainted, suckered, or guilty is made worse by the fact that it is the moral opposite of what was expected.

In addition to moral “taint,” Sherman finds that returning soldiers are plagued by various manifestations of guilt, several of which persist despite the absence of the belief that one has personally done something morally wrong. In one case of what she calls “accident guilt,” an Army Major in Baghdad feels guilty when the gun on a Bradley fighting vehicle malfunctions and gruesomely kills a young private standing just a foot in front of it. The malfunction was caused by a bad replacement battery. The Major, John Prior, was causally responsible for the death only in the most indirect sense: he was in charge of the maintenance of the vehicles, and he signed off on replacing the original battery with a slightly different model, the only one available, “after consulting with his maintenance team and after a thorough reading of the manuals.” (Sherman 2010: 96) Given that he was blamelessly ignorant of the danger, and that he exercised due diligence, it is fair to say that Prior did nothing morally wrong. Yet, Sherman claims,

In a very important way, Prior’s guilt is not irrational. […] The guilt is reparation for harm done. It records something morally significant, something that marks his deep connection to his troops and his moral accountability to them, and to himself. It is not just a response to the fact that he was causally implicated. He feels morally implicated. (Sherman 2010: 96)

It is daunting to think that you can go off to war, take your command very seriously, follow all the rules, and then nevertheless find yourself “morally implicated” in a gruesome accident that you could not have foreseen—one that is not even directly related to the war itself, and is independent of the war’s moral status. Although the accident was horrific and traumatic, Prior evidently experienced a sacrifice that is independent of any PTSD he may suffer. That sacrifice is the injury to his moral self, his feeling of being morally tarnished despite his best efforts. This kind of loss will need to be reconciled against the moral value of an endeavor like volunteering for national service.

In the case of humanitarian aid workers, the moral self can be threatened not only by perceived complicity in harm, but by diminished motivation to help, and a sense of guilt about one’s own inability to live up to an ideal of moral helpfulness. One young MSF doctor, Davinder Gill, found himself assigned to a rural outpost in Liberia on his first mission. At age twenty-six, he was the only doctor managing a small hospital and forty local staff under harsh conditions. He found himself making a psychological adjustment in order to live with the fact that he could only work so hard, could only save so many patients:

… patients would be dying at night and I wouldn’t be there. They’d be dying during the day and I’d have to be eating lunch. I think the first two patients who died were the hardest. Those two cases stressed me a lot at the beginning—stressed me quite a lot. But after that I think appropriately I got more used to it. It just becomes part of the work and you just try your best. I compare myself to others and I wonder whether another doctor in the same setting would have had the energy or whatever to spend longer with that patient, sleep less that night, and get more work done the following day than I got done. (Hopkins 2008)

In Gill’s case, we see that he felt a sense of moral inadequacy, a concern that he was not doing as much as he could or that he did not deserve to take time to himself. At the same time, we see that he found a way to compartmentalize the suffering around him so that he could survive a trying situation. The worry, though, is that the compartmentalization progresses to a kind of compassion fatigue. When Gill’s supervisor visits the hospital, he updates her on the case of a sick baby whose condition he could neither diagnose nor treat with the resources available to him. The baby’s mother decided to take him home unannounced, and Gill says it does not bother him (after all, he had thus far not been able to do much of anything for the child). The visiting supervisor is dismayed: the young doctor is apparently already burnt out and has lost the conviction that his patient is better off in his hands. His idealism weakens and he soon describes his enthusiasm as “spoiled”. After one mission, he leaves MSF and returns to Australia.

The emotional exhaustion that arises in situations like this is unsurprising. What makes it morally relevant is that it can involve moral emotions—compassion, sympathy, guilt—and can impact the strength of one’s moral motivation. Paradoxically, firsthand experience in responding to human suffering could make one less responsive to human suffering. Depending on the circumstances, this “hardening” might either enable or hinder further humanitarian work. But whether the agent’s new disposition has good or ill effects on balance, it still comes at the cost of losing his old disposition, which may have been central to his identity and may have been morally admirable independently of its effects.

Sacrifices of Self and Moral Accounting

So what exactly have the undercover agents, soldiers, and humanitarian aid workers given up, and why does it matter? I want to suggest that these losses cause trouble for the view—which I have previously defended—that all sacrifices are best understood simply as losses of wellbeing. On that view, a sacrifice is a gross loss of well-being incurred while (or after) pursuing an action or project that is, or is believed to be, morally worthy. (Carbonell 2012) That it is a gross loss is important. To be sure, we often make sacrifices that enable other gains. But even when it is possible to compare the losses and gains so as to declare a “net” benefit, there is a sense in which the gains do not directly compensate for the losses. For example, someone who takes a demotion at work in order to have time to care for an ailing relative can still be said to have made a sacrifice even if the caregiving ultimately yields more in wellbeing than the higher ranking position would have.

Sacrifice is here a normative idea, one that acts as a kind of currency in the moral ecosystem: if helping you would require too much of a sacrifice from me, then I may not be required to help, even if I have otherwise good moral reason to. In order to adjudicate these conflicts between the needs of others and the things that matter to us, we need a public, shared understanding of when a loss counts as a sacrifice. The fairest way to do this, I have argued, is to think of sacrifices as losses of wellbeing in an objective, agent-neutral sense: wellbeing consists not simply in whatever a person thinks makes her life go well, but rather in whatever in fact makes her life go well. One such agent-neutral account is Stephen Darwall’s (2002: 4) rational care theory: a person’s well-being (or her welfare, her “good”) is “constituted, not by what that person values, prefers, or wants (or should value), but by what one (perhaps she) should want insofar as one cares about her”. If those who know and care about Orbinksi—his mother, for example—want him not to live with uncontrollable rage, then living with uncontrollable rage is bad for him.

To generate a more specific list of the kinds of things that would, as a substantive matter, contribute to people’s wellbeing, we could pair Darwall’s agent-neutral theory with an “objective list theory”—a theory according to which there are certain things that are good or bad for us, whether or not we desire them. (Parfit 1984) On standard objective list theories, the well-being-generating categories include familiar things like health, knowledge, friendship, love, recreation, and access to beauty. While these lists differ in the details, what we notice is that many moral sacrifices will straightforwardly involve losses of items on the objective list of good things, or impositions of items on the objective list of bad things. Examples include loss of necessities or comforts like hygiene, sleep, nutrition, or recreation; loss of material possessions, wealth, or income; loss of access to art, music, knowledge, nature, or time with loved ones; interaction with aversive or unpleasant things or circumstances; loss of social or relational position in society; imposition of pain, bodily harm, or risk thereof; imposition of emotional or psychological suffering; opportunity costs; and restrictions on freedom of movement, free time, or privacy.

The canonical examples from the literature on moral demands—damage to a prized vintage car or boat, threat of bodily harm, depletion of one’s wealth, relinquishment of vacations—fit this picture nicely. But unlike these canonical cases, the sacrifices I have explored in this essay are not best characterized as losses of the constituents of a person’s wellbeing, because it does not make sense to think of the person herself as being part of her own wellbeing. The agents I have described did not merely give up wellbeing-generating “stuff” (money, possessions, time, opportunities) or mental states (pleasure, comfort, security). Nor did they merely undertake what we might call a “second-order” loss of wellbeing—the loss of the ability to garner future wellbeing, as might happen when one becomes depressed. What they incurred was, rather, an injury to the psychological or moral integrity of the person whose wellbeing is at stake. So while Queen’s confusion about his true self may have caused a decrease in wellbeing, it is not reducible to a decrease in wellbeing.

How could this be? After all, if a sacrifice of self is to count as a bona fide sacrifice, surely it must be in some sense bad for the person. All else being equal, we would not want someone we cared for to undergo a sacrifice of self. And I have claimed—following Darwall—that we should think of a person’s wellbeing as, roughly, whatever it would be rational to want for her insofar as we care for her. Am I not then committed to arguing that a sacrifice of self is just a sacrifice of wellbeing? In other words, do “a stable sense of identity” and such simply belong on the correct list of the substantive constituents of wellbeing, alongside health, recreation, and the rest?

I think the answer is "no," because to put these broadly “self”-related notions on the list of the constituents of wellbeing would be, at best, problematic, and at worst, confused. It would be confused because the list is meant to contain the items whose presence or absence, or relative scarcity or abundance over time, is good or bad for a person. But to measure changes in these items over time for a person requires that we assume some stability in the person herself. To appreciate this, we need not make any deep metaphysical commitments about what “the self” is and how it persists over time, or about when and whether someone becomes “a different person” in some literal rather than metaphorical sense. We simply must acknowledge that what it means to care for someone, and what we will want for her insofar as we care for her, will change dramatically if she changes dramatically. What is good for Queen, the ATF agent, is different from what is good for Billy St. John, his alter ego. As Queen finds himself becoming more like Billy, and losing track of the distinction between his “true” self and his “false” persona, it becomes less clear how we should be measuring increases or decreases in wellbeing: relative to Queen before the assignment, or relative to the “changed” Queen, as he emerges after the assignment? This problem is not necessarily unique from the more general problem of determining whether any long-term project affects a person’s wellbeing, given that all of us change over time. But I have tried to show that certain lifestyles and projects may bring forth this problem in particularly dramatic and interesting ways, which must be accounted for when we evaluate the moral demandingness of those projects.

Perhaps it is splitting hairs to claim that changes in “the self”—changes to self-conception, sense of identity, agency, etc.—cannot be counted as losses of wellbeing because a relatively stable “self” is a prerequisite for, rather than a constituent of, wellbeing. Or perhaps it is unfair to claim that some level of “stability” is a prerequisite for wellbeing, given that we are all unstable over time—the “self” in the broad sense we have been considering is an inherently dynamic and subjective entity. But even if these concerns are valid, and even if we could therefore think of sacrifices of self as simply gross losses of wellbeing, I want to argue that these sacrifices are still quite problematic in the context of determining what morality requires.

What makes sacrifices of self especially problematic is that even if, say, “moral taint” and “identity confusion” were reducible to losses of wellbeing, just as a lost vintage boat or reduced income is, they would still be extraordinarily difficult to compare with other losses of wellbeing and with the moral gains they make possible. Since I do not think sacrifices of self are reducible to losses of wellbeing, I think part of the problem is that sacrifices of self and most moral gains are “incommensurable”—we cannot express them in the same units. My view does not entail that sacrifices of self and moral gains are strictly “incomparable”; that is, I am not making the strong claim that it would be impossible to make meaningful comparisons between them. We can sometimes make such comparisons even without common units. (Chang 1997) The problem, though, is that rough comparability does not guarantee that a particular moral gain compensates for a particular sacrifice, in the sense of meaningfully “making up for it” or somehow filling the hole it creates. Indeed, I think compensation would be a problem even if we could express sacrifices of self in “units” of wellbeing, as it would still be awkward and difficult in many cases to make meaningful trade-offs between such sacrifices and the moral gains they enable.

Queen, for example, lost his ability to live “as himself” for several years, and as a result lost track of who he really was; in return, let us suppose, dangerous criminals were arrested, perhaps resulting in the prevention of some violent crimes. At the time Queen did not know how much of himself he would need to sacrifice. But now he does know. If faced with another similar undercover opportunity, where he is the best qualified for the job, would it make sense to say that he has a moral duty to sign up? If there is a way to answer this question, it is hard to see how it could involve the weighing of things as disparate as Queen’s identity on the one hand, and the value of crime reduction on the other. Indeed, I would argue it is hard to negotiate this trade-off even if we could say that Queen’s damaged identity is worth 10 “units” of wellbeing and the crime reduction is worth 15. For wellbeing no longer seems like the only thing morally relevant in this situation, if the wellbeing in question consists in someone’s sense of self on the one hand, and, say, some prevented theft on the other.

So if we aim to delimit the bounds of moral obligation by examining trade-offs between what moral agents stand to lose and what others stand to gain, sacrifices of self will make those trade-offs especially troublesome. Again, I am not claiming that it is impossible to reconcile sacrifices of self with moral gains, nor claiming that sacrifices of self are the only kinds of sacrifices that would be so difficult. I am simply suggesting that the connection these sacrifices have to the “self”, and the role the self plays as the subject on whom other gains and losses of wellbeing are imposed over time, creates an interesting difficulty.

If we instead aim to delimit moral obligation simply by determining what degree of sacrifice it is “reasonable” to demand that moral agents bear, without regard to the compensating effect of moral gains, then we must ask when it is reasonable to demand that someone risk losing a part of her self for a moral end. Of course, such losses will rarely be complete and guaranteed; instead we will have to evaluate projects that involve some chance of incurring some degree of sacrifice of self. Yet these qualifications make the problem harder, not easier. We may agree, for example, that the complete destruction of someone’s moral agency—her transformation into a selfish, uncaring, unmotivated, or downright dangerous person—is always too high a cost to bear. But what about lesser degrees of moral compromise? The humanitarian aid worker might come home cynical and angry, numb and unsympathizing. But people can acquire these traits through a variety of life experiences, and can even remain quite morally effective despite them. The hardest and most worthwhile moral pursuits may be those most likely to leave participants changed, diminished, or lost. If we set the threshold of unreasonable sacrifice low enough to protect agents from even moderate changes in their moral agency, we may be declaring certain moral goals unreachable except through the good will of the moral overachievers. We would seem to be valuing the intact moral self so much that we fail to make use of it.

Alternatively, rather than seeking some threshold above which sacrifices of self are unreasonable, we might simply declare the substrates of these sacrifices—identity, autonomy, and moral agency—as part of a protected class, simply outside the scope of what we can legitimately be asked to give up for morality’s sake. If personal projects and integrity have been thought to be worthy of protection in that way, then certainly one’s moral agency itself should also be so protected.Footnote 19 And yet some moral projects—humanitarian work, military work, caregiving and parenting work—present agents with opportunities to foster and protect in others the very same morally-relevant psychological capacities they are at risk of sacrificing themselves. A soldier may sign up for duty precisely because she thinks she is defending the ideals of freedom and autonomy. In doing so, she may risk her own future autonomy, which could be compromised by trauma. Could we consistently hold that the autonomy of those she sees herself as protecting is important enough to generate a legitimate claim on her, but that she could not be obligated to sign up to provide the protection if doing so would significantly risk her own future autonomy? This is of course an oversimplification of what would be a far more complex scenario in real life. But the point is simply that, while it may be difficult to commensurate and compare simple losses of wellbeing with sacrifices of self, there may be scenarios in which a “self” is to some extent threatened on both sides of the equation. To insist that sacrifices of self are beyond the scope of morality’s demands would seem to favor protecting the autonomy or moral agency of one group at the expense of the autonomy or moral agency of another group.

Ultimately, I am unable to solve this problem, only to point to it and suggest that we pay it some attention. I do not think an obvious, simple answer is lurking. We might be tempted to consider the extraordinariness of the case studies I have examined—elite undercover agents, all-consuming humanitarian missions, soldiers in wartime—and conclude that such projects are obviously supererogatory, and any case likely to involve similarly spectacular sacrifices would be too. The problem, of course, is that these cases are not obviously supererogatory. Whether one is morally required to sign up for national service, or to take on a dangerous assignment for which one is uniquely qualified, is very much a matter of dispute. We cannot get around this problem by simply declaring that no ordinary person could ever be faced with the possibility of being morally required to give up her very self, in whole or in part. Thousands of confused and broken soldiers returning from war show us that this problem is very real indeed. Moreover, there is much to learn about the limits of moral obligation by examining actions that are not obligatory but rather supererogatory. If some action would be an unreasonable demand because it requires too much sacrifice, there is nevertheless some other action that would not be unreasonable somewhere down the line—how else to find where the boundary lies than by examining the real-life sacrifices people make above and below it, and seeking some way of assessing them?Footnote 20 At the very least, these kinds of cases should be considered alongside the famous thought experiments in which we pit prized possessions against the needs of distant strangers. If Bob and his Bugatti deserves our attention, why not also Ivan and his Identity, or Margaret and her Moral Agency?