Can our Hands Stay Clean?

This paper argues that the dirty hands literature has overlooked a crucial distinction in neglecting to discuss explicitly the issue of, what I call, symmetry. This is the question of whether, once we are confronted with a dirty hands situation, we could emerge with our hands clean depending on the action we choose. A position that argues that we can keep our hands clean I call “asymmetrical” and one that says that we will get our hands dirty no matter what we do I call “symmetrical”. Not acknowledging this distinction is a problem because, firstly, it adds to the existing confusions about how best to define what dirty hands are. Secondly, it prevents the concept of dirty hands from being applied properly to other contexts such as, for example, the responsibility and accountability of politicians. I argue that we have good reason to favour a symmetrical understanding because it gives a more convincing account of what makes an action dirty and because it more accurately captures our complex moral decision-making when faced with dirty hands situations. The paper concludes by outlining possible implications that the distinction between the symmetry view and the asymmetry view has on wider debates surrounding the problem of dirty hands.


Introduction
Dirty hands (from here on referred to as "DH") situations are, loosely speaking, situations in which an agent is forced to choose the lesser of two evils. Should you torture a terrorist to find out the location of hidden bombs? Should you agree to free criminals in order to save hostages? Should you silence a journalist in order to prevent a riot? The DH literature is largely in agreement that politicians who choose to torture, circumvent justice, and violate civil liberties get their hands dirty. It is interesting to note, though, that very little has been said in the DH literature about what happens should they choose not to kill, torture, or deceive. Do they still get their hands dirty or can they escape from the situation unstained? On this question, there are two possible stances that could be taken. I will call a view claiming that we can emerge with our hands clean when confronted with a DH situation the "asymmetry view". It is asymmetrical because choosing one side leads to DH while by choosing the other we can keep our hands clean. A position that argues that, in such a scenario, we will inevitably get our hands dirty I will call the "symmetry view". I begin by arguing that the DH literature has overlooked a crucial distinction in neglecting this issue. Many writers seem to have taken an inexplicit or unjustified stance on this issue. Others have expressly bracketed the question and excluded it from their analyses. Others make contradictory claims on the matter without noticing. This oversight is of great significance, because it contributes to existing confusions about how best to define DH. Additionally, it prevents the concept of DH being properly applied to other issues, such as the responsibility and accountability of politicians. The first part of this paper will survey these different claims. I will then argue that we ought to prefer a symmetrical understanding because it gives a more convincing account of what constitutes the dirt in DH and more accurately captures our complex moral decision-making when faced with DH scenarios.
Clearing up the confusion regarding symmetry and asymmetry is a crucial step toward an improved understanding of the phenomenon of DH. The conclusion will have a direct effect on our evaluation of both what it means to have DH and the agent who is responsible for them. Finally, I will consider the implications of my discussion for other questions relating to the wider DH debate.

The Problem of Dirty Hands
There are broadly three ways of explaining the nature of DH problems. The one that I will argue captures it most accurately understands it as a clash of plural and conflicting moral values. There exist a variety of moral values that we reasonably think to be important but that cannot always be aligned or easily traded-off with one another without a significant remainder. For Michael Stocker, DH conflicts arise when we face situations in which people will be "wronged, they and their trust, integrity, and status as ends […] violated, dishonoured, and betrayed" (1990, p.17). Additionally they can also involve the destruction of items of great worth (e.g. art or a holy place) or "the violation of an important principle rather than a person" (1990, p.18) (e.g. circumventing justice). Choosing the lesser evil does not mean that the competing value or principle not acted upon is therefore annulled or cancelled. While it may have been outweighed in this case, its normative force is still intact. This then results in a moral remainder that attaches to us. Not every choice between conflicting values that results in a remainder is a DH situation, though, because not every morally wrong action is also a dirty action. While it is not entirely clear what this distinction amounts to on Stocker's account, he seems to draw some sort of quantitative distinction in which DH are simply particularly grievous attacks on a given value or principle. DH situations are then clashes of plural and conflicting values that result in a particular remainder because they require the agent to violate or betray a person, value, or principle in an especially grave way. This view opposes two potentially more common understandings of the problem.
In Michael Walzer's article "Political Action: The Problem of Dirty Hands" he sums up the issue in the question of "how can we get our hands dirty by doing what we ought to do" (1973, p.164). Walzer illustrates this through the following scenario: A political leader is asked to authorise the torture of a terrorist in custody who is said to know the location of bombs in several apartment buildings across the capital set to go off in less than a day (1973, p.167). 1 Walzer argues that the reason this choice is so difficult is because we are torn between our deontological and consequentialist commitments. On the one hand, we have the obligation not to torture, but on the other, we are faced with the potentially disastrous consequences of not doing so. Torturing "may be exactly the right thing to do in utilitarian terms and yet leave the man who does it guilty of a moral wrong" (1973, p.161). According to Walzer, the politician's consequent guilt is a sign that she takes her deontological commitments seriously and will only override them when absolutely necessary. The problem with this conception is that it unduly limits the kind of situations that could be called DH. Imagine a political leader who is faced with a situation in which she has to curtail the liberty of some people in order to promote equality. She might come to the reasonable judgement that her foremost duty is to protect her citizens' equality and that doing so would also maximise overall happiness or well-being, and yet, she would still rightfully feel guilty about having to violate some people's liberty to do so. In such a situation our deontological commitments and utilitarian reasons point toward the same course of action, and yet we would rightly experience a moral remainder. The underlying conflict in DH is therefore better understood as a clash between plural and conflicting values.
There is yet another common conception of DH. Thomas Nagel conceives of the issue less in terms of conflicting moral theories and more in terms of an agent's obligations qua private individual and qua public office holder. Even if both moral codes should be ultimately derived from the same source, he argues, the principles emphasised in private and public morality are different and can therefore result in a clash leading to DH. Public action involves both "a heightened concern for results and a stricter requirement of impartiality. It warrants methods usually excluded for private individuals, and sometimes it licenses ruthlessness" (1993, p.82). DH conflicts then result from a clash between our private principles and commitments and public standards of impartiality and a focus on the consequences of our actions. If an agent wants to fulfil the requirements of her public role, she is required to choose the latter and forgo the former. We ought to reject this understanding because it does not capture cases that strike us as instances of DH. Imagine, for example, that you are good friends with a couple one of whom you find out is cheating on the other. You are now torn between being truthful and sparing your friend's feelings. While there might be good reasons to think that DH are particularly prevalent in the public sphere because of the heightened concern for impartiality and the consequences of one's actions, there is no reason to limit DH cases to this realm. The conflict inherent in DH is therefore better understood as a particularly grave clash between plural and conflicting values.
Even if DH are not necessarily due to the divide between public and private duties, there is a reason why much of the literature centres on political cases. Walzer supplies us with two reasons for why the political sphere is particularly susceptible to DH cases arising. Firstly, politics is a realm that brings power and glory to leaders and people are willing to go to great lengths to claim these benefits. Many will be ready to use morally dubious means in the process and once they have achieved a position of political leadership the power they have over the lives of others can further corrupt them and their actions. Once some agents start behaving in these ways this will likely force others into situations in which they will have to employ similar means. Even a politician who wants to be morally good will therefore be forced to engage in the morally dubious behaviour of these others and get her hands dirty in order to make it in politics. Secondly, Walzer argues, DH problems in politics are particularly pressing because politicians are able to use "violence and the threat of violence" (1973, p.163). In the pursuit of the interests of her citizens a political leader will often be forced to put the lives of people at risk. Trying to protect and further the interests of her citizens will therefore make the politician face additional, and particularly violent, DH problems. The result is that any political leader, no matter how good their character or intentions, will have to get their hands dirty throughout their career. Walzer therefore agrees with the Machiavellian credo that politicians will have "to learn how not to be good" (1973, p.168). 2 While I think that there are good reasons then to subscribe to an understanding of DH as a clash of plural and conflicting values, this does not preclude an acknowledgement that DH cases will be particularly frequent and pressing in politics.

Symmetry and Asymmetry in the Dirty Hands Literature
This section will survey the different, often implicit, positions various authors have taken on the symmetry of DH problems: whether we can emerge from such situations with clean hands, or whether our hands will be dirty no matter what we do. Let us start with those seemingly endorsing the asymmetrical reading. Walzer repeatedly claims that when a political actor is faced with a ticking bomb scenario he could "refuse to dirty his hands" (1973, p.165) and "[remain] innocent" (1973, p.161). Additionally, Walzer states that "here is the moral politician: it is by his dirty hands that we know him. If he were a moral man and nothing else, his hands would not be dirty; if he were a politician and nothing else, he would pretend that they were clean" (1973, p.168). Someone who acts like a morally good private person would choose the option that was not dirty. A non-dirty option must therefore be available, even if, according to Walzer, a good politician would not choose it. Demetris Tillyris appears to adopt a similar stance when he states that "the costs of refusing to get dirty are likely to increase with time" (2015, p.67) (i.e. the cost of losing office). If we can refuse to get dirty, this clearly leaves open the possibility of emerging untarnished when faced with a DH problem. Richard Bellamy also seemingly embraces the asymmetry view when he notes that politicians could "keep their own hands clean but at the expense of leaving the rest of us in a dirty situation" (2010, p.417). He argues that when one person refuses to get their hands dirty, it is likely that this will force someone else into a DH situation. He objects to this because, given that politicians are supposed to act in accordance with the public good, "preserving the moral integrity of a saint in public life seems not just irresponsible but even immoral" (2010, p.417). While he argues that keeping one's moral integrity would be irresponsible and immoral, he does not state that it would be impossible. Jeremy Waldron also appears to allow for an option whereby an agent could keep her hands clean. When he discusses the case of the Clay Cross Eleven, a group of Labour councillors who refused to adhere to a Conservative government's policy they thought unjust, he says that "they refused to dirty their hands" (2018, p.225). Finally, Suzanne Dovi talks about political actors who "refuse to compromise their moral integrity" (2005, p.129), "live according to moral principles, refusing to compromise them even in the face of great hardships" (2005, p.133), and so "make vivid what political actors with dirty hands should feel guilty about" (2005, p.134). She implies that the agent, when faced with a DH problem, can and sometimes should choose to remain with her hands clean by adhering to her moral principles. 2 I will primarily talk about politicians in this paper because this is what a lot of the DH literature has focussed on. This is not to say, however, that DH cannot arise in other spheres of life. Moral values can clash and cause dirty hands scenarios in private as well as public life. The responsibilities we have to choose one over another will depend, among other things, on the particular roles we occupy that are relevant to the situation at hand and generate particular dutiese.g. as a friend, as an employee, or as a citizen.
On the opposing side, various writers embrace a symmetrical view of DH. Karl Klockars states that in a Dirty Harry scenario 3 "the choice must always be between at least two wrongs and in choosing to do either wrong, the policeman inevitably taints or tarnishes himself" (1980, p.37). Similarly, Martin Hollis argues that "once a dilemma has been posed for a person in office […] it is too late for clean hands, whatever he does" (1982, p.394). He later reiterates this stating the following about a political leader: "his dilemma was that of inescapable responsibility under partial constraint and his hands were dirty before he even began to resolve it" (1982, p.397). As soon as an agent is faced with such a situation his hands will be dirty no matter how he tries to resolve it, because he is constrained to choose between two morally objectionable courses of action. Stuart Hampshire also appears to support this view. He states that politicians "should at all times be prepared for the occurrence of an uncontrollable conflict of duties in situations which seem to exclude the possibility of a decent outcome, and in which all lines of action seem dishonourable or blameworthy" (1991, p.170). On a similar note, Christopher Gowans defends the claim that in life, "moral innocence would be virtually impossible to attain, regardless of how pure of heart we may be" because whenever we encounter a DH situation "we will do something morally wrong no matter what" (1994, p.220). Steve de Wijze also appears to hold a symmetrical understanding when he claims that if an agent "chooses to act in accordance with the obligation and duties required of office-bearers (or indeed refuses to act), then she will get her hands dirty" (2007, p.4). Michael Yeo, similarly, states that "no matter what choice the agent makesincluding not choosingsomeone will be seriously harmed or wronged, or some principle held dear will be otherwise negated. The agent's hands will be dirty whichever course is taken" (2007, p.157). Lastly, Kai Nielsen likewise argues that "to try to wash one's hands, Pontius Pilate-like, of a dirty hands situationto say 'it is none of my business my hands are clean', where some choice on our part might make a differenceis impossible. We do not escape responsibility by so acting. Failing to act in such a circumstance is in itself an action. By so refraining, we dirty our hands just as much as, and perhaps more than, a person who acts resolutely to achieve the lesser evil, though in doing so he does horrible things" (2007, p.21). Nielsen even goes as far as saying that the person who refuses to intervene, e.g. by ordering the torture, may in some circumstances have dirtier hands.
There are also writers whom we earlier placed on the asymmetrical side of the debate who, later on in the same papers, use the language of symmetry. Bellamy claims that "whatever the choice, he cannot wash his hands or keep them clean" (2010, p.419). This appears to contradict his earlier claims on the options available to an agent facing a DH situation. The same problem can be found in the paper by Tillyris. Earlier he implied that agents could, even if they should not, choose to keep their hands clean. While he initially uses language that suggests asymmetry, later in his paper he endorses the above quote from Martin Hollis: "once a dilemma has been posed for a person in office, integrity does not demand that he keep his hands clean by stepping aside. It is too late for clean hands, whatever he does" (Hollis 1982, p.394). This would seem to favour the symmetry view. A closer look at their papers suggests that both authors inadvertently make asymmetrical claims when bringing forward their criticism of the views of others (Bellamy 2010, p.417;Tillyris 2015, p.67) and use the language of symmetry when putting forward their own accounts (Bellamy 2010, p.419;Tillyris 2015, p.68).
Finally, there is a small category of authors who recognise the distinction between symmetry and asymmetry. Michael Stocker, discussing Walzer's ticking bomb scenario, explicitly states that he refuses to tackle the following questions: "is the non-doing of an act of dirty hands the doing of another act of dirty hands?", "what are the relations between dirty hands and the doctrine of double effect?", and "what are the relations between dirty hands and the distinction between doing and not doing?" (1990, p.12). Similarly, Neil Levy hints at the reasons we might have for embracing a symmetrical or asymmetrical understanding of DH but does not engage in any in-depth argument on this matter. Discussing an agent facing a DH scenario, he writes, "unless we are impressed by the doctrine of double-effect, or some act/omission distinction, we can clearly see that she can no longer avoid an action that is wrong, categorically wrong" (2007, p.45). Both Stocker and Levy have identified that there are two potential views of DH that we could take. They then make the further claim that these different conceptions rest on our understanding of the action/omission, doing/allowing, and intending/foreseeing distinctions. While they note the distinction and its potential importance, they do not engage in any in-depth debate about it. The only sustained discussion of the issue of symmetry can be found in Christopher J. Finlay's Humean account of dirty hands (2011, pp.436-439). He argues that we can make two claims about dirty hands scenarios. Firstly, "that where we imagine the politician choosing to uphold the law [i.e. to uphold the ban on torture], it gives rise to a case of 'dirty hands'" and secondly, "that if we imagine the politician authorizing torture […], then his decision may be seen as an attempt to keep his hands clean" (2011, p.436). This is a flipped version of the asymmetry view which holds that we keep our hands clean when we order the torture and dirty them when we refuse to do so. In the following section I will examine the different grounds we might have for supporting either the symmetry or asymmetry view.

The "Dirt" in Dirty Hands
The distinction between symmetry and asymmetry hinges on different understandings of what constitutes the "dirt" in DH. I want to suggest that a symmetrical view is based on a more plausible conception of what it means for an action to be dirty. When an agent is faced with a DH problem, both actions will carry a particular kind of remainder. Following Stocker's understanding, this remainder is best understood as the result of a betrayal or violation of a person, or important principle or value. Whatever option the agent chooses, she will betray or violate something of great importance, and this will inevitably taint her. This is not to say, however, that the symmetry view could not be spelled out in terms suitable to other understandings of DH. If they are understood as a clash between deontology and consequentialism or a clash between private and public ethics, then whichever moral code or source of obligation we forgo, we will inevitably get our hands dirty according to the symmetry view. In the ticking bomb scenario the politician is torn between violating her commitment to not torture and violating her commitment to protect the citizens she represents. Both options create remainders of varying strength, so constitute more or less dirt, but both result in the agent dirtying her hands. 4 It would then be impossible for the agent to avoid getting her hands dirty whichever option she chooses.
Before getting into more detail about how best to characterise the asymmetry view, it will be helpful to clear up two potential sources of confusion. Firstly, one might object to my characterisation of the above accounts as asymmetrical. If an asymmetrical account holds that there is a possibility of keeping one's hands clean, would that not deny the inherent conflictual nature of DH situations? In fact, all of the accounts that I have characterised as asymmetrical do acknowledge this conflict at the heart of the DH problem. Take the case of Walzer for example: if one acts like a good politician one forgoes what a good person would do, and if one acts like a good person one forgoes what a good politician would do. So there is clearly a conflict here. Therefore, these accounts could not be asymmetrical and my characterisation of them would be wrong. This objection, however, misunderstands what is at stake in the symmetry debate. Saying that an account is asymmetrical is not supposed to show that it denies the existence of the conflictual nature of DH situations. Instead, what I am arguing is that the asymmetrical position differs in that it holds that the agent gets her hands dirty only on one horn of the conflict (e.g. choosing to act like a good politician) but keeps them clean on the other (e.g. choosing to act like a good man). Choosing one horn of the conflict therefore does not automatically equate to getting one's hands dirty. The aim of the paper is to critically assess why different authors think that, despite the conflictual nature of DH situations, one horn of the conflict could be said to leave the agent with her hands clean. The existence of a conflict at the root of dirty hands situations is not what is in question, instead it is the correct conceptualisation of both horns of the conflict that is at stake.
Secondly, one might wish to object to the asymmetry view from the outset for the reason that it cannot make sense of cases such as Sophie's Choice. Sophie arrives with her two children at a Nazi concentration camp where an evil doctor presents her with the following choice: she must choose one of her children to be killed in order for the other one to be spared. Should she refuse to choose, both will be killed immediately. In such a situation there may simply be no better or worse option and clean hands would be impossible. Assuming that one of Sophie's core moral commitments is to ensure the welfare of both of her children, there is no way in which this choice will leave her unstained. The asymmetrical view of DH therefore cannot hold that we can always emerge with our hands clean from DH situations. Instead, the argument has to be the weaker claim that sometimes, but not always, clean hands are possible. While this criticism changes the claim the asymmetrical view makes, it does not provide a reason to reject it outright. According to a symmetrical understanding there are no cases of DH in which the agent could keep her hands clean, whereas on an asymmetrical understanding the agent can keep her hands clean in some, even if not in all, DH situations.
So what can the asymmetry view tell us about the dirt in DH? Stocker and Levy think that the different understandings of dirt rely on a distinction between acting and omitting, doing and allowing, or intending and foreseeing. On the asymmetrical reading, once you are faced with a situation in which you are forced to choose, giving the order to torture results in DH and not ordering it leaves one's hands clean, so the thought might go that this difference must be the result of a distinction between what we do and what we omit (or something similar); if we act we dirty our hands, but if we do not act we can keep our hands clean. We might think that such an understanding would make sense in light of how Walzer sets up the ticking bomb scenario. Walzer says that, given the circumstances, ordering the torture might be the lesser of two evils. This means that he implicitly stipulates not ordering the torture to be the greater evil. After all, the politician has a duty to protect her citizens and not doing everything in her power to prevent the death of innocents that it was in her ability to prevent would be a serious moral failure. For Walzer the politician would be responsible whether she orders the tortures or refrains from doing so, and yet, the politician will keep her hands clean should she choose the latter. This leaves us with a situation in which the not-doing of an action is the greater evil and the politician can be responsible for it and yet will not be dirtying her hands if she does so. One understanding of the asymmetry view is therefore that we can only get our hands dirty through our actions and what we do, but not our omissions and what we merely allow.
From Walzer's description of DH as a clash between deontology and consequentialism we could find a second version of what constitutes the dirt in DH situations, namely a violation of a deontological constraint such as, for example, the violation of a person's rights. Some actions violate people's rights while others do not and only the former create the right kind of remainder for an action to be called dirty. Torture involves the violation of a person's rights. Letting people die, however, when the only way to save them would involve the torture of someone else, does not involve the violation of people's rights. Imagine, for example, a doctor who has five patients who all need different organ transplants in order to live, and a sixth who happens to have healthy and matching organs for all of them. Clearly the five do not have a right to be saved if that involves killing the sixth. Another plausible version of the asymmetry view might therefore be that the dirt in DH is the violation of people's rights of interference; but it is not a rights violation to fail to rescue or benefit people when doing so would violate the rights of a third party. In the ticking bomb scenario, not ordering the torture may be the greater evil, but as it cannot be said to violate the rights of citizens, the politician's hands remain clean.
In the language used by both Walzer and Tillyris, we can find yet another version of what constitutes the dirt in DH on the asymmetry view. This conception relies on two premises; firstly, what makes DH dirty is that they constitute an attack on the agent's innocence. Secondly, by not acting in accordance with the demands ofin Stuart Hampshire's wordsexperience, our innocence can stay intact and our hands remain clean. A good starting point to understand what is meant by the agent's innocence is to examine Hampshire's discussion of that term, 5 to which Tillyris makes explicit reference. Hampshire examines the notion of innocence in a political context by discussing Machiavelli's account of politics and morality. He acknowledges that in political life emergency situations will inevitably arise in which a leader has to use means usually thought to be morally reprehensible. Machiavelli opposes the demands on a good political leader to the demands on individuals wishing to pursue a morally admirable private life, where both ways of life are mutually exclusive. The virtues of the latter, such as loyalty, friendship, and fairness stand in contrast to political virtús such as courage, prowess, and the single-minded pursuit of power that lead to glory for oneself and one's city (Machiavelli 2003, pp.23-28;35;50-51). Someone acting in line with the virtues leads, according to Hampshire, a life of innocence, while one who acts in line with Machiavellian virtú leads a life of experience. Political leaders, in order to be successful and effective, need to be able to leave innocence behind and turn to a life of experience. Hampshire goes on to describe innocence by comparing it to the life of early Quakers; "their conception of the good was a vision of simplicity, whiteness, straightness, uprightness, cleanness, of sweeping away anything contaminated or corrupted or squalid" (1991, p.173). From this we could construct a version of the asymmetry view on which our hands become dirty when we abandon the virtues constituting innocence and turn toward the virtús of experience. Should we, however, choose to prioritise the former over the latter, our hands can remain clean. While Tillyris and Hampshire focus on political innocence, we can see in Walzer a way to construct innocence more generally. He writes that, should the politician choose the utilitarian side of the conflict, "the innocent man, afterwards, is no longer innocent" while, on the other side, "he remains innocent" if he were to choose "the absolutist side" (1973, p.161). Innocence for him is a resolute adherence to absolutist moral principles (not unlike Hampshire's vision of the early Quakers), whereas innocence is abandoned and our hands dirtied when we act according to consequentialist reasoning (not unlike Hampshire's or Machiavelli's vision of the moral importance of political effectiveness). On this account, DH would therefore be asymmetrical, because agents can choose innocence and an adherence to absolutism over experience and reasoning in consequentialist terms.
While Walzer and Tillyris talk about the agent's innocence, Bellamy and Dovi use the language of integrity. Waldron does not actively use the word integrity, but his view seems to fall into this camp as well. 6 There is therefore yet another version of what constitutes dirt on the asymmetry view. Similar to the innocence version, it can be constructed as the following two premises; firstly, what makes DH dirty is that they constitute an attack on the agent's personal integrity. Secondly, by not acting in accordance with the demands of impartial morality, our integrity can stay intact and our hands remain clean. We can find this view in Dovi's description of DH as a clash between the agent's moral integrity and some overriding moral end. Bellamy also alludes to such a clash between integrity and morality, and traces this view to Susan Mendus. It will be helpful to take a closer look at Mendus's discussion on integrity in order to understand this version of the asymmetry view. Integrity for her "is a matter of acting on those commitments which are very important to me and which serve, in part, to define who I am" (2009, pp.31-32). This is a distinctly personal notion of integrity that does not require that the commitments one chooses adhere with social or conventional morality. On such a view, an individual can possess integrity and yet be morally bad (e.g. Heinrich Himmler could be an exemplar of someone with personal integrity). The result of defining personal integrity in this way, argues Mendus, is that it can come into conflict with morality. She considers the following example by Bernard Williams: George who just finished a PhD in Chemistry is desperately looking for a job to support his family. A friend of his knows of his problems and offers him a job in a lab that does research into biological and chemical weapons. George is deeply opposed to using such measures in warfare and is now torn about whether to accept the position or not. When he tells his friend about his qualms, his friend says that the reason he wants George to take the job is that it will otherwise be given to another person about whose excessive zeal he feels rather worried (1973, pp.97-98).
Mendus argues that in such a case there exists a tension between integrity "understood as a matter of sticking by what one believes to be ethically necessary" and morality "understood as acting impartially towards all who are affected by one's actions" (2009, p.29). According to her understanding, if George wants to preserve his integrity he will have to refuse to engage in what he perceives to be an evil and immoral pursuit. Morality, conversely, would favour him accepting the job because it would benefit both his family through the financial support this would offer, and society at large because he would approach the job more cautiously than the over-zealous candidate. Mendus argues that a similar analysis can be made in response to 6 To dirty one's hands, according to Waldron, is to commit a wrong action that violates one's own moral standards. In the case of the Clay Cross Eleven, for example, adhering to what they perceived as unjust legislation would have involved getting their hands dirty. By refusing to uphold the law and standing up for their moral convictions, however, the Clay Cross Eleven kept their hands clean.
Walzer's ticking bomb example. She says that "here then is a case in which politics undermines, or at least threatens to undermine integrity" (2009, p.39). To keep her integrity she would have to refuse to order the torture and to engage in what she perceives to be evil. Morality, on the other hand, would urge her to consider the fates of the hundreds of innocent people who could be killed unless she orders the torture. For Mendus, we have a choice between acting in accordance with impartial morality and thus dirtying our hands, and keeping our integrity intact and our hands clean but allowing the greater evil to occur. Here, the "dirt" in DH is a stain on one's integrity, which occurs when we violate our commitments to valued principles such as never to kill, torture, or deceive. The conclusion then is that, if there is an option for personal integrity to stay intact, the agent can emerge from a DH situation with her hands clean.
Finally, as mentioned briefly in the previous section, we can also construct a flipped version of the asymmetry view in which one's hands stay clean by ordering the torture and get dirtied by refusing to order it. Finlay starts from the Humean assumption that, when confronted with a scenario such as the ticking bomb, we start by experiencing certain moral sentiments and try to arrive at a general moral judgement or rule based on those sentiments. Finlay suggests that on such a theory, "natural moral sensibilities dictate approval for stronger feelings of benevolent concern and a correspondingly intensified sense of duty of care" and conversely, "our moral sympathies are diminished for those we see engaged in acts of cruelty or violence" (Finlay 2011, p.425). The result is that we would have heightened compassion for the innocent citizens threatened by the bomb and diminished compassion for the terrorist. We would therefore approve of the agent who orders the torture. For Finlay, "a properly balanced measure of compassion as a motive tending towards the safety of the innocent should outweigh in quantitative valency any sympathetic concern that might remain for a terrorist determined to let the bomb detonate" (2011, p.427). Through her actions the terrorist has forgone her right to our compassion or benevolence. Within the confines of the ticking bomb scenario, Finlay concludes, torture could be seen as a distinctively moral course of action. He goes on to argue that there are good prudential and institutional reasons to uphold a universal ban on torture, though. He is worried that allowing exceptions to this ban would turn into a slippery slope and weaken the overall integrity of democratic values and institutions. Returning to Hume, Finlay goes on to argue that in ticking bomb scenarios the agent is confronted with a choice between natural and artificial virtues. The former represent our moral concern and compassion for the innocent victims while the latter represent the institutional ban on torture. The final step in his argument is to say that we get our hands dirty by violating a natural virtue but can keep them clean if all that we are violating is an artificial virtue. If the politician decides not to torture, "then we may say that he allows artificial virtue to override natural virtue. In doing so, he has to betray the innocent civilians targeted by the terrorist's bomb, suppressing the natural urges of compassion and the moral duties these generate. This is how he dirties his hands" (2011, p.347). If the politician refuses to torture and upholds the legal ban she will get her hands dirty, but if she chooses to torture she can keep her hands clean. For Finlay, the latter situation is more akin to a case of civil disobedience in which a law is violated for moral reasons so that "a Humean view would interpret the choice as being between an action involving dirty hands and one resembling civil disobedience" (2011, p.439).
On the symmetry view the dirt in DH is the violation of a moral value, code or obligation, depending on one's understanding of what is the source of the conflict that creates DH in the first place. On the asymmetry view, however, something can be dirty in virtue of being an action, doing, or intending, or the result of violating a deontological constraint such as committing a rights violation, the result of choosing the demands of impartial morality or a life of experience over integrity or innocence or, on the flipped view, the result of choosing artificial over natural virtue.

Defending Symmetry
So should we adopt a symmetrical or asymmetrical understanding of DH? I will argue that we should choose the former because of the costliness of adopting the latter.
Remember that on the asymmetry view if the agent decides to order the torture she will get her hands dirty, but should she decide not to, she will keep her hands clean (or vice versa on the flipped version). Holding any version of the asymmetry view, however, appears to stand in stark contrast to one of the core insights resulting from discussions of the problem of DH. What the DH literature has highlighted is that many moral theories are exclusively concerned with action-guidance at the expense of making sense of our complex experience of moral decision-making. If, for example, the agent decides not to order the torture, then her regard for the ban on torture is the action-guiding principle, while any value attached to protecting the lives of innocent citizens would be the non-action-guiding principle. The only way to think that by not ordering the torture keeping one's hands clean were possible, is to focus entirely on the action-guiding considerations present in the case. Only if I think that, for example under the integrity view, upholding my core commitments exhausts the morally relevant features of my decision can I think that my hands could be clean. This, however, overlooks what should happen with the non-action-guiding moral considerations, in this case the value of saving innocent lives. We see the value, code, or obligation we have not acted upon as a genuine moral demand and so we "double-count" (Stocker 1990, p.12) it. We weigh up the ban on torture against the potential of innocent lives being saved and decide to act according to the former. The latter value is not thereby cancelled or annulled and should still be given weight, even if it was not acted upon. While we have identified the lesser evil in the situation, we in this way give due weight to something we take to be a genuine moral requirement: we are taking the wrongness of our action seriously and it becomes a remainder that attaches to our evaluation of the action taken. As soon as the agent takes the non-action-guiding features of the situation to be morally relevant she will be forced to acknowledge that holding onto her core commitments came at a considerable moral price. This moral cost will inevitably leave a moral residue and get her hands dirty.
For the agent to think that her hands could remain clean when faced with a DH scenario would require some act of moral compartmentalisation in which she detaches her conscious decision to choose one course of action over another (e.g. not torturing) from the consequences of that decision (e.g. the death of innocents). Such views draw a sharp divide between the agent's actions and adherence to principles, and the outcomes of those actions. Such a divide is both unlikely and undesirable. Take for instance the innocence version of the asymmetry view. While it is true that your choice meant you did not order the torture of anyone, this decision also critically endangered the lives of hundreds of citizens or even failed to prevent their deaths. The knowledge that external circumstances have forced you into a position where you have to make such choices should be enough to attack your innocence. It does not matter how good your intentions are and whatever option you choose will inevitably have grievous consequences. This knowledge is incompatible with innocence as described by Hampshire of "a vision of simplicity, whiteness, straightness, uprightness, cleanness, of sweeping away anything contaminated or corrupted or squalid" (1991, p.173). If the agent is unable to keep her innocence she would also no longer be able to keep her hands clean on this understanding of the asymmetry view.
A similar issue can be found on the integrity version of the asymmetry view when we consider the ticking bomb scenario. While your action certainly adheres to your personal deontological principles, it does so at the cost of ignoring the commitments and obligations you have as a politician, thus drawing an artificial line between you as a private and as a public person. You have many different roles that constantly overlap in your life and it is unclear how you could abstract from that and construct a notion of integrity concerned only with a single subset of these. The idea behind this must be something like the thought that somehow the private individual behind the politician can keep her hands clean. As Bellamy argues, this "position turns on abstracting to a universal moral code behind any special circumstances, and to a universal moral individual behind any social role" (2010, p.418). However, it is not obvious that either abstraction is possible. Imagine a situation in which an agent's roles of being a good politician and being a good friend clash; for example, the agent's work commitments keep interfering with his ability to spend time with his close ones. Bellamy argues that "there is no point beyond the two roles to which he could retreat to ask what is universally required" (2010, p.418). In such a situation it simply makes no sense to ask what a good person would do. We should therefore abandon the idea that integrity can be defined as adherence to the moral commitments of some private or universal individual and instead acknowledge that the constraints of integrity will emerge from the interplay of the social roles we occupy. Integrity, on this broader conception, could not stay intact by adhering to some deontological constraint when this results in the violation of another rule, obligation, or value and therefore our hands could not stay clean on such a version of the asymmetry view.
What about notions of asymmetry that rely on the action/omission distinction or the like? In cases of DH, I think that these distinctions do not accurately capture the way in which the agent becomes implicated in the harmful consequences of a DH scenario, even if they choose not to harm anyone themselves. While it was not your action that caused these negative consequences, your conscious decision not to act meant that you failed to prevent them, even though you were capable of doing so. This makes you complicit in the negative consequences and it is not clear why this should not be enough to get your hands dirty. The rights version of the asymmetry view especially might reply that there still is an important distinction because you did not actually violate anyone's rights in this situation and as such you can keep your hands clean after all. To argue that no one's rights were violated by not ordering the torture implies a view on which negative rights of non-interference always trump positive rights of assistance. It is unclear, however, why this ought to be the case. Imagine that, on a walk through the countryside, you come across someone injured and in desperate need of food and water. The only way to help them is by breaking into a nearby cabin and stealing supplies. In such a case, arguably, a small infraction of one person's rights could be morally justified by the way it would assist someone else. If we concede that negative rights should not always trump positive rights then rights could have been violated by not ordering the torture and the agent could therefore no longer keep her hands clean on this version of the asymmetry view.
Finally, let us consider the flipped version in which we can keep our hands clean by ordering the torture and dirty them by upholding the ban against torture. The former option, according to this view, is an act of civil disobedience in which we violate a law for moral reasons, whereas the latter is an instance of dirty hands in which we violate morality for legal reasons. Finlay's interpretation of Hume implied that only the violation of natural virtues (e.g. benevolence, compassion), as opposed to artificial virtues (e.g. justice, conformity to law), can generate dirty hands. This, however, seems to be an odd conception. Hume distinguishes between natural and artificial virtues to demarcate a difference between those virtues that arise simply in virtue of human nature and those that arise as a result of people living together in society. Every act based on natural virtue will be beneficial, whereas only the overall tendency of actions based on artificial virtue will be beneficial; particular instances of acting based on artificial virtues may actually be harmful. This, however, does not mean that artificial virtues do not put genuine moral demands on us. There are very weighty reasons for us to respect justice and uphold our legal system. They ensure that we can live together in a society and fairly adjudicate conflicts amongst people. We care about bringing justice to people and upholding the rule of law in our country, and when either of these are broken we generally think that something of moral importance has been violated. Finlay himself describes DH, following de Wijze, as a violation of a cherished value. Only by saying that upholding justice or the rule of law does not involve such an important moral value could he, by his own standards, argue that violating them would not be an instance of dirty hands. We do not, however, only care about justice and the law for prudential reasons; we care about them as important moral values. His view therefore would have to rely on an unappealing understanding of why we care about artificial virtues.
The problem overall is not that these different understandings of the asymmetry view are not possible to hold in principle, but rather that they come at a significant cost. If we take into consideration not only the action-guiding value, but also the important moral value that we have forgone in a DH situation, the asymmetry view results in rather unpromising understandings of innocence, integrity, conscious action, rights, and artificial virtues. The onus here is on the defenders of the asymmetry view to show why we should hold any of these conceptions and how they more accurately reflect on the nature of DH situations. Even if any of the asymmetry views could give us a satisfying answer to this, there is yet another problem for these accounts. If we need to take into account the full moral weight of the non-action-guiding principle, how should an agent feel after having chosen, for example in the ticking bomb scenario, not to order the torture?
According to the asymmetry view she might reasonably feel proud for having adhered to her moral principles and not having become complicit in the immoral plans of others. She stood by her commitments when doing so was extremely difficult and she deserves praise for this. She did something laudable in making that choice because she preserved something of distinct and important value. This, however, seems an odd assessment of the situation and her decision. In a situation in which the lives of hundreds of people are at stake that it is her job to protect, her character and her personal convictions do not seem important. In fact, she has misunderstood the situation if she failed to save them. This conflict is reflected by Max Weber's distinction between the "ethic of principled conviction" and the "ethic of responsibility" (2010, 359). The former is an absolutist approach to moral action in which, as long as the agent strictly adheres to her moral rules, she cannot be responsible for any negative consequences that may flow from her actions. The latter ethic, on the other hand, is based on an acknowledgement that "the achievement of 'good' ends is in many cases tied to the necessity of employing morally suspect or at least morally dangerous means" (2010,360). On this view the agent is fully responsible for the negative consequences of her actions, even if she adhered to the moral rules she holds dear. A result of the asymmetry view is that, because there is an option in which the agent would not get her hands dirty, this option would be in some sense praiseworthy because the agent would have kept her hands clean even if to do so she chose the greater evil. On the symmetry view, on the other hand, the agent simply ought to choose the lesser evil. What counts is that the agent has assessed what ought to be done all things considered based on the obligations she has to others and the consequences of her actions, and not based on a concern for the cleanliness of her own hands. An asymmetrical view would encourage a self-centredness incompatible with grasping what is morally at stake in DH situations. We should therefore prefer a symmetrical understanding that captures more accurately our assessment of the agent's actions in a DH situation based on the moral weight attaching to the non-action-guiding principles present.
None of the arguments in this section are intended to constitute a decisive win for the symmetry view. Instead they are supposed to show some of the costs involved in adopting an asymmetrical understanding. Defenders would have to show why their view can make sense of our complex moral decision-making process and the emotions and reactive attitudes that are fitting as a response to a DH situation. They also would have to show that the criteria for cleanliness they use align with how we would normally employ these concepts. I think that these costs are rather unappealing and that we should therefore side with a symmetrical understanding of DH. Ultimately, the debate between symmetry and asymmetry may hinge on some of our most fundamental intuitions about the possibility of leading a moral life. Many things in our lives are beyond our control and we will often face misfortunes. The hope may therefore be that at least the option of leading a morally good life would be immune to bad luck. If only we do the best that we can to hold on to our principles, then we must be able to keep our hands clean. Otherwise we would end up with a rather dreary picture of ourselves and the world around us. The idea of symmetry, however, threatens this hope because it argues that once we are faced with a DH situation there is no way for us to emerge from it with our hands clean. Accepting symmetry therefore entails accepting a tragic view of our lives in which we are not beyond the reach of bad moral luck. It does not matter how hard we try and despite our best efforts we will sometimes have to engage in evil and get our hands dirty. I have given some suggestions as to why we should prefer a symmetrical understanding of DH. What makes something dirty is the remainder caused by a violation of a value, moral code, or obligation, and whatever option the agent chooses she will inevitably become tainted.

Conclusion
My argument in this paper has been twofold. Firstly, I started by outlining my negative thesis that the DH literature has overlooked an important question in the definition of DH; namely whether we dirty our hands no matter what we do, or whether we can possibly keep our hands clean. I showed that a variety of papers on DH have made implicit assumptions about the symmetry question, but have failed to defend them. I then discussed different versions of the asymmetrical understanding that hold that DH can be avoided and that our hands can, at least in principle, stay clean. According to the symmetrical view, however, our hands become dirty when we violate or betray a value, moral code, or obligation and choosing between two evils in a DH situation will always inevitably involve such a violation. Our hands can therefore never stay clean. I have argued that we have good reasons to support a symmetrical reading of DH because adopting an asymmetrical reading comes at a cost that we should try to avoid.
Clearing up this issue has results beyond merely improving the way in which we define DH. The stance we take on symmetry or asymmetry might, for example, affect the level of responsibility or blame we ought to ascribe to the agent for her actions. If whatever she does she will be doing something dirty this might lead us to think, all other things being equal, that her responsibility for the wrongdoing and our ability to blame her is diminished; after all, there is no way she could have avoided getting dirty. 7 Similarly, though, if we subscribe to symmetry this might also lead us to widen our understanding of what the agent can be responsible or blameworthy for. If choosing not to torture does not enable the politician to keep her hands clean, then it would make sense to argue that she is in an important sense responsible and blameworthy for risking the lives of innocent citizens. As the stance we take on the symmetry or asymmetry of DH could have effects on our understanding of the responsibility and blameworthiness of the agents, it could also in turn affect whether or when we think we should hold dirty-handed agents accountable for their decisions. 8 It should therefore be clear not only that the symmetry question has been overlooked in the DH literature, but also that it is important for philosophers in future to be explicit on whether we can emerge with clean hands, or whether we will dirty our hands no matter what.