Self-Other Asymmetry

In this paper, I present a non standard objection to moral impartialism. My idea is that moral impartialism is questionable when it is committed to a principle we have reasons to reject: the principle of self-other symmetry. According to the utilitarian version of the principle, the benefits and harms to the agent are exactly as relevant to the global evaluation of the goodness of his action as the benefits and harms to any other agent. But this view sits badly with the “Harm principle” which stresses the difference between harm to others and harm to the self. According to the deontological version, we have moral duties to ourselves which are exactly symmetrical to our duties to others. But there are reasons to believe that the idea of a duty to the self is not coherent.

Many philosophers deny that impartiality could be all there is to ethics. According to them, a morality limited to impartiality would be unrealistic, globally irrelevant to our lives and even repugnant in some cases.
I present another kind of objection to moral impartiality, less melodramatic if I may say. My idea is that moral impartiality is questionable when it is committed to a principle we have reasons to reject: the principle of self-other symmetry.
But what is self-other symmetry?

SELF-OTHER SYMMETRY
Self-other symmetry is a basic commitment in many moral theories, but it takes different forms depending on the global structure of the theory. One could say, for example, that utilitarian theories are based on self-other symmetry because, according to theses theories, the benefits and harms to the agent are exactly as relevant to the global evaluation of the goodness of his action as the benefits and harms to any other agent 1 . And one could say that deontological theories of Kantian flavour are based on self-other symmetry because, according to them, we have moral duties to ourselves, which are exactly symmetrical to our duties to others. The famous second main formulation of the categorical imperative, in the Groundwork of the Metaphysics of Morals, also called the "Formula of Humanity", stresses this symmetry as explicitly as possible: Act in such a way that you always treat humanity, whether in your own person or in the person of any other, never simply as a means but always at the same time as an end 2 .
The clause "whether in your own person or in the person of any other" is a non-equivocal affirmation of self-other moral symmetry. Suicide or masturbation are "moral crimes" according to Kant, partly because of their supposed moral symmetry with killing and sexual abuse. Finally, one could even say that virtue ethics is based on selfother symmetry because it values equally care to others and self-care. Actually, this is Michael Slote's master argument in favour of virtue ethics 3 , and it raises a perplexity concerning the scope of my criticism of self-other symmetry.

SOME PERPLEXITIES
If virtue ethics is committed to self-other symmetry, as Michael Slote claims, and if virtue ethics does not belong to the class of impartialist moral theories, as some moral philosophers would probably say, then by objecting to self-other symmetry, the target could be larger than moral impartialism. It could include virtue ethics as well, or some versions of it at least. It could make my argument less limited than I have suggested. But there are other perplexities.
Self-other symmetry seems to be a very important feature of many moral theories, but, at the same time, one can find elements of self-other asymmetry in these theories. Think of the "Harm principle" put forward by John Stuart Mill. According to Mill: "The only part of the conduct of anyone which he is amenable to society is that which concerns others. In the part which merely concerns himself, his independence is of right absolute. Over himself, over his own body and mind, the individual is sovereign" 4 . In more concrete words, according to the Harm principle, one can be morally or legally permitted to do to oneself what one is not permitted morally or legally to do to others, the most striking example being, again, suicide as opposed to killing. The "Harm principle", as a kind of self-other asymmetry, goes against the utilitarian general commitment to self-other symmetry. One could have expected that Mill, being a prominent representative of utilitarianism, would also be a prominent supporter of self-other symmetry. But for him, the harms to the agent are not as relevant to the global evaluation of the goodness of his action as the harms to any other agent. So, he is not a supporter of self-other symmetry after all.
According to Michael Slote, Kant is also guilty of inconsistency in his treatment of self-other symmetry 5 . On the one hand, Kant claims that we should apply to ourselves exactly the same moral rules we apply to any other person. He argues not only for the wrongness or impermissibility of killing others but for the wrongness or impermissibility of suicide as well. And this is clearly a commitment to self-other symmetry. But for Michael Slote, this view sits badly with what Kant says about the absence of duties to pursue one's own happiness 6 . For Kant, the concept of duty applies only in case we are reluctant to do something. It does not apply when we do something inevitably and spontaneously. Now, if it is true that we tend naturally to care about our own interests and that we tend, as naturally, to neglect the interests of other people, then, in Kant's perspective, there should be no duties to further one's own interests but only duties to further the interests of other people. The fact that some of us may have a tendency to ruin themselves through stupidity, carelessness, laziness does not weaken the argument. Personal imprudence leads to its own natural punishment, while ill treatment of others does not come with its own natural punishment, except in fairy tales. If ill treatment of others it is to be controlled, it can only be through a sort of socialised system of duties and punishment.
According to Michael Slote, there is a contradiction in Kant's views about morality between his claim that we should apply to ourselves exactly the same rules we apply to any other person, and his other claim that there are no duties to further one's own interests but only duties to further the interests of other people. In other words, he supports both self-other symmetry and self-other asymmetry, and this is not coherent.
One could add that the disharmony is even bigger when we introduce degrees of closeness to ourselves in the picture. We tend naturally to care not only about our own interests, but about the interests of those who are dear and near to us. And if we have a natural tendency to further the interests of those who are near or dear to us, then strictly speaking, there could be no duty to further their interests.
If Kant's logic in this domain were followed, we could have moral duties only towards those we dislike or those who are personally the most distant from us. So there is an element of moral asymmetry in Kant's moral system after all.
Actually, I will not enter into these difficulties. I am not especially interested in testing the coherence of Mill's utilitarianism, of Kant's moral system or of Slote's version of virtue ethics. I just want to present reasons to reject some forms of self-other symmetry and other reasons to endorse some forms of self-other asymmetry.

FORMS OF SELF-OTHER ASYMMETRY
Self-other asymmetry expresses itself in different ways. The most well-known are 1. selfishness, which gives priority to the agent's own interests over the interests of other people; 2. selective altruism, which gives priority to the interests of those who are near and dear to the agent over the interests of the agent himself and the interests of those who are not near and dear to him; 3. radical altruism, which gives priority to the interests of every other person over the interests of the agent himself. But I will only defend self-other asymmetry when it takes the form of the harm principle according to which harm to self is morally indifferent. I will try to show that there is no good reason to turn away from common-sense moral thinking on this specific point. Whenever philosophers have tried to abandon this form of commonsense self-other asymmetry, whenever they have tried to line up the harm we do to ourselves with the harm we do to others, they ended up with what I take to be unsound concepts like "duties to oneself " or "self-regarding moral virtues", if not with absurd questions like "Is it possible for me to rescue myself?" or "Is it possible for me to compensate myself?" or "Is it possible for me to be grateful to myself ?" 7 . At least, this is what I will try to show.
My plan will be the following. First I will present the Harm principle as a form of self-other asymmetry grounded in common-sense moral thinking. Then I will examine and reject an argument to the effect that self-other asymmetry is not grounded in common-sense moral thinking but in the theoretical importance of consent. Finally, I will try to show how wrong one can go when one departs from common-sense views about selfother asymmetry, by presenting Kant's arguments regarding "duties to the self " as they are called.

WHAT IS "COMMON-SENSE MORAL THINKING"?
A few words should be said at this point about what I mean by "common-sense moral thinking". I am not referring to empirical sociolo-gical or psychological data about everyday moral judgements of different people from different countries and different backgrounds. Actually, I could have referred to data of this kind, because we have now an interesting body of knowledge about everyday moral judgments, thanks to the works of psychologists Jean Piaget, Lawrence Kohlberg, Carol Gilligan, Jonathan Haidt, Larry Nucci or Elliott Turiel among many others 8 .
But when I refer to "common-sense moral thinking" I am pointing to something less empirical. I am concerned with formal or substantial norms of moral thinking which are not always openly expressed in everyday judgements. What I have in mind are norms like "You should treat like cases alike", "You should not sacrifice innocents", etc. These norms are formulated by philosophers and presented as central or basic "moral intuitions". They are sometimes rhetorically attributed to everyone, whether real people have them or not.
What is philosophically special about them is that they are supposed to be moral assertions which need no further justification and to be valid objections to some big theories in ethics. Think of our supposed "intuition" that we should not sacrifice one innocent person even in order to save the lives of thousands of innocents, which is so often directed against some versions of utilitarianism.

IMPARTIALITY AND PARTIALITY IN ETHICS
The debate around impartiality and partiality in ethics has many different aspects and meanings. It may oppose philosophers who believe that, as far as morality is concerned, we should, a priori, have the same benevolent attitude toward every human being, from the nearest and dearest to the distant other, and philosophers who believe that it is morally permitted or even admirable to give some priority, systemically or in certain specific circumstances, to yourself, your friends, relatives, compatriots, etc., out of love, friendship, loyalty, quest for self perfection. In this context, we are offered two possibilities.
1. Impartiality I should have the same benevolent attitude toward every human being, myself included.
2. Partiality I am permitted to give some priority, systemically or in certain specific circumstances, to myself, my friends, relatives, fellow countrymen. But if we take into account the Harm principle as well, moral par-tiality itself should be divided in two classes: positive partiality based on benefits to ourselves and those who are near to us, and negative partiality which means that we are permitted to harm ourselves but not other persons In other words, we have three possibilities and not only two, the first involving self-other symmetry and the two others involving self-other asymmetry.
1. Impartiality I should have the same benevolent attitude toward everyone, myself included.
2. Positive partiality I am permitted to treat myself, my friends, my relatives, my fellow countrymen better than other people, systemically or in certain specific circumstances.
3. Negative partiality I am permitted to harm intentionally myself but not other persons.
If we dig deeper, we find other possibilities: 4. misanthropy: it is a kind of impartiality which consists in having the same malevolent attitude toward everyone, myself included. The problem is of course that such form of impartiality could hardly be called "moral". 5. self-hatred: it is a kind of negative partiality which consists in treating myself, my friends, my relatives, my fellow countrymen worse than other people. Blacks, Jews, and other representatives of minorities are sometimes claimed to exhibit this attitude. The question is again if such a form of negative partiality could be called "moral".
6. self-sacrifice: here we have a "supermoral" or "supererogatory" negative partiality which consists in a complete ban on selfbenefits .
I will insist on negative partiality, especially on this specific kind which permits harm to self but not to others.

MORAL ASYMMETRY IN COMMON-SENSE MORAL THINKING
Actually, I have borrowed the words "self-other symmetry" and "selfother asymmetry" from Michael Slote 9 . Sometimes, he uses the words "moral symmetry" instead of "self-other symmetry" and "moral asym- metry" instead "self-other asymmetry". For him, these expressions have the same meaning. I will follow him on this account as well. What he calls either "self-other symmetry" or "moral symmetry" is a system that brings together care to others and something like selfcare or reasonable selfishness. What he calls "common-sense moral asymmetry" has two aspects. It consists in self-abnegation or self-sacrifice, that is, in avoiding some personal benefits where we would not prevent another person from receiving the same benefit. But it consists also in permitting to harm ourselves where we would not be permitted to harm another person in a similar way 10 . When he comes to examine moral theories according to these standards he insists on the first aspect of common-sense moral asymmetry, that is, on self-abnegation or selfsacrifice. He thinks that common-sense and Kantian morality value only self-abnegation or self-sacrifice while utilitarianism and virtue ethics value equally care to others and self-care or reasonable selfishness. He claims that "the common avoidance of self-other asymmetry allows utilitarianism and virtue ethics to go forward securely in a manner that is not possible for either Kantianism or commonsense morality" 11 .
I have difficulties with both the classification and the conclusions.
First, it seems to me that through the second main formulation of the categorical imperative, Kantian moral system is so deeply committed to self-other moral symmetry that it is not very fruitful to insist on some elements of self-other moral asymmetry we can find in it 12 .
Second, although Mill's is utilitarian, this doesn't prevent him from proposing a Harm principle, which is clearly asymmetric,. And third, I think that Michael Slote should have treated in a completely different manner the two aspects of what he calls "commonsense moral asymmetry".
"Common-sense moral asymmetry" consists in avoiding some personal benefits where we would not prevent another person from receiving the same benefit, as well as in permitting to harm ourselves where we would not be permitted to harm another person in a similar way.
One can say that there can be something "moral" in avoiding personal benefits where we would not prevent another person from receiving the same benefit. But can one say, in a similar manner, that there can be something moral in permitting to harm ourselves where we would not be permitted to harm another person in a similar way? I don't think so. This second kind of asymmetry tells us something interesting about the limits we set to what can be called "moral" or "immoral". It seems to fall outside the scope of morality altogether.
Let me present some examples. Suppose that instead of cutting his own ear, Van Gogh had jumped on some innocent passer-by to cut the ear of this unfortunate fellow. It seems natural to me to say that common-sense moral thinking would treat these two cases differently. The harm done by Van Gogh to himself could only be called crazy, irrational but not immoral, while the harm done to the innocent passer-by could be called immoral as well. Now, let's compare the following statements.
1. You should read or practise some sport instead of spending all your days on the couch, watching stupid programs on television and stuffing yourself with chocolate cookies. I don't force you, I don't threaten you. I just tell you that it would be better for you.
2. You should read or practise some sport instead of spending all your days on the couch, watching stupid programs on television and stuffing yourself with chocolate cookies. I don't force you, I don't threaten you. I just tell you that so staying in front of the television would be immoral.
3. According to many witnesses, you have spent more than 30 days on a couch, watching stupid programs on television and stuffing yourself with chocolate cookies. By so acting, you have violated the law criminalizing unhealthy behaviour. You are fined 100 000 dollars and from now on, you will be under medical supervision at your own expenses.
Let's concentrate on the second statement. It seems to me that there is something queer, if I may say, in asserting "It is immoral to spend all your days on the couch, watching stupid programs on television and stuffing yourself with chocolate cookies" because it consists in blaming a person for harming himself or herself, and harming oneself is morally indifferent.
In order to prove that what's wrong with this judgement is that it concerns only harm to self, it could be sufficient to show that if it were formulated a little differently, with a hint to the fact that such behaviour would also harm, even indirectly, some other person, it would not seem queer at all. For example, the judgement "It is immoral to stuff yourself with chocolate cookies when you could save so many lives by sharing them with hungry little children", wouldn't seem queer, I believe, but it is, of course, because it directly implies other persons. So much for the idea that the self-other asymmetry is plausibly grounded in our common-sense moral thinking, as a sort of intuition which does not require justification.
Some philosophers have tried to ground self-other asymmetry in something "deeper", more "philosophical". They have tried to show that self-other asymmetry could be derivative from the moral importance of consent. Could it be the case?

GROUNDING COMMON-SENSE MORAL ASYMMETRY
In a footnote attached to his "Some Advantages of Virtue Ethics", Michael Slote writes that it has been suggested to him that the reason why we are allowed to harm ourselves where we would not be permitted to harm others in a similar way lies in the consent implicit in whatever we do to ourselves. If I harm myself I presumably do it willingly, whereas the agent I am harming does not consent to the harm 13 .
According to this explanation, the self-other asymmetry is not a substantial feature of common-sense morality but derivative from the moral importance of consent. Shelly Kagan has also observed that a defence of moral asymmetry could be grounded in the moral importance of consent. He writes that if what is in question is "only my treatment of myself, it is obvious that I will always be acting with the consent of the person I am affecting" 14 . But in what sense does consent affect my action? Well in case of harm, it does in a very special way. According to Kagan, normally, it is not permissible to perform an act if it would involve harming someone. There is a constraint against acting so. But when I deliberately harm myself, I have my own permission to do so. And for that reason, I am removed from the scope of the constraint 15 . The argument would be clearer if it were put in legal terms. We should make a difference between causing physical or psychological pain to someone as it may happen in medical procedures or in violent sports like ice hockey or boxing, that is with some sort of consent of the victim to the risks of being injured, and causing to someone some physical or psychological pain to which he has never consented in any sense. In both cases some pain has been caused but it is only in the second case, that is, where there was no consent, that one can speak of rights violations, torts, or harm in the moral sense. The common law doctrine "Volenti non fit injuria" means "one cannot be wronged by that to which one consents". It could be interpreted in the following way, I think. When there is tacit or explicit consent, there might be pain but no harm, no violation of rights, no wrong. The same could be said for harm to self. You can intentionally cause pain to yourself but you can't harm yourself intentionally, because where there is consent, there is no harm, in the sense of there being no wrong. Shelly Kagan thinks that the argument can obviously be resisted by those who deny the moral relevance of consent, and also, but less obviously, by those who do accept the relevance of consent but deny that in cases like those supposed to illustrate harm to self, true consent obtains.
When someone is about to commit suicide or self-mutilation can we say that he is really acting willingly? Can we say that he is really consenting? I think that we don't have to enter into these psychological and conceptual complexities because there is a better argument against grounding self-other asymmetry in consent. It has been proposed by Michael Slote, actually. According to him, common-sense morality makes a difference between negligently causing harm to another person and negligently causing harm to oneself. We keep making a difference between harm to self and harm to others even when the harm was not done intentionally or when there was no consent to the harm done. And this should bring us to think, in the most simple way, that after all, self-other asymmetry is not grounded in intention or consent.
I think that Michael Slote is right and that self-other asymmetry is not derivative from the moral importance of consent, nor reducible to it 16 . It should be taken as a basic moral intuition which can be used against traditional impartialist moral theories.
Until now I have presented some arguments to the effect that self-other asymmetry is a reasonable feature of common-sense moral thinking and need not be grounded in the moral importance of consent. Now, I will try to show how wrong one can go when one departs from this common-sense view about self-other asymmetry. I will present to this effect Kant's views on "duties to the self " or "duties to oneself " as they are more often called.

MORAL SYMMETRY
Kant "duties to self " are divided into perfect duties which concern us as natural beings and perfect duties which concern us as moral beings. Examples of perfect duties which concern us as natural beings are prohibitions on suicide, on masturbation and on excessive use of food and drink. Examples of perfect duties which concern us as moral beings are prohibitions on lying, avarice and servility. We have also imperfect duties not to leave our natural talents or capacities idle and, as it were, rusting away. All these duties to ourselves are presented as duties not to downgrade ourselves by denying our own humanity. They are duties not to harm ourselves. This is why I am especially interested in them. It has been recently noticed that, despite Kant's claim that duties to oneself are of primary importance in his system, his idea has not been widely discussed 17 . Actually, after Marcus Singer published a short and brilliant paper on duties to oneself in 1959, there has been an intense exchange of arguments on the subject 18 . But it didn't last, and I don't know why. The only thing I can say is that it is not due to the fact that Singer's replies were so good that nothing could be added to them. Nothing of that sort has ever happened in philosophy, especially in moral philosophy, and it would be almost supernatural that it had happened then! I am personally very interested in this issue because I am trying to build a sort of "minimalist ethics" in which there are no duties to self and no self-regarding virtues. I will return to this very shortly in my conclusion.
But before that, I will present what I take to be the best argument against duties to the self as Kant conceived of them. It aims to show that the notion of duty to the self is contradictory. Other arguments could have being raised as well. For example: 1. Duties to the self are not really duties toward the self as a particular person but duties toward abstract entities like nature or the human species. When, for example, Kant argues against masturbation, it is partly because according to him human species would disappear if it were a unique and universal sexual practice. So Kant's ban on masturbation is less a duty to oneself than a duty to the human species.
2. As moral duties, duties to the self are only derivative from duties toward other people. I may have a moral duty to keep fit, if for example it may help me care about other people: my lover, family, children, fellow citizens, etc. Another example: I may have a moral duty to stay sober if I am the pilot of the airplane, but it is a moral duty to the passengers more than to myself. 3. Duties to the self are not real duties but natural preferences. There can be no moral prohibition to ruin your own talents because no normal human being wants to ruin his own talents.
I insist on the formal argument because if duties to the self are inconceivable, we will naturally turn to these other arguments which are meant to show that talk about duties to the self is always talk about something else, something more conceivable, if not more acceptable.
The fact is that it has often been noticed that the notion of duties to oneself is conceptually problematic in a way that the notion of duties to others is not.
There are duties which can clearly be regarded as relative to other persons: duties arising from contracts, agreements, debts or promises for example, but duties to oneself do not seem to be that clearly conceivable. Why?
Against Kant who claimed that our duties to ourselves are of primary importance and should have pride of place and who added that the prior condition of our duties to others is our duties to ourselves 19 , Marcus Singer has argued, for example, that it is "actually impossible for there to be any duties to oneself in any literal sense because, if taken literally the idea involves a contradiction" 20 .
His argument goes like this. If in general "A has a duty to B, then B has a right against or with respect to A. But it follows from this that to have a duty to oneself would be to have a right against oneself and this is surely nonsense. What could it mean to have a right or a claim against oneself? (Could one sue oneself in a court of law for return of the money one owes to oneself)?" 21 Singer is right to stress the absurdity of these implications of the idea that one can have a duty to himself. But here, there is no obvious contradiction, just some kind of category mistake. The contradictory character of duties to self is more obvious when we consider the structure of promises or debts.
As a duty to another person a promise has the following features. Suppose you have promised to lend me your vacuum cleaner. You can, of course, break this promise, by hiding yourself, destroying the vacuum cleaner, or even by telling me that you have simply changed your mind: you realised that you don't like me after all and that you don't want to lend me anything anymore. But none of this will release you from your specific promise or cancel it. It will just be a promise which you have not kept. Actually you can't be released from your promise by yourself. However, you can be released from it by me, that is, by the one to whom you made it. I can cancel your promise because, for example, I realised that I am too lazy to vacuum my place, or because I realised that I don't like you after all and I don't want to feel I owe something or for whatever other stupid reason.
In short, the person who promised something cannot release himself from the promise, but the person to whom the promise has been made can release from his promise the person who promised. Now suppose that it is the same person who made the promise and to whom the promise has been made.
He is not bound by his promise and he is bound by it. He can release his promise and he can't. Isn't it contradictory? When we move from promises to debts, the contradiction is as obvious when it is the same person who is creditor and debtor. Suppose you lent me 20 000 dollars. I am under the obligation to return them to you. Of course you are free to tell me in a fit of generosity or vanity: "I decided to cancel your debt. You don't owe me 20 000 dollars anymore". But I am not free to answer. "Well, it doesn't matter. I didn't plan to return them to you. In the meanwhile, I had cancelled my debt." So the creditor is free to release the debtor from his obligation, but the debtor isn't free to release himself from the obligation. Now, suppose that it is the same person who is creditor and debtor. He is free to release himself from the debt ant not free to release himself from the debt. Isn't it contradictory?
One could deny that the model of the contract or the promise is relevant when we deal with duties to oneself. One may think that duties to the self belong to the category of duties you can't be released from, like duties not to torture or duties not to enslave 22 . But could all duties to the self be seen that way? I doubt it. An argument is obviously missing here.
In any case, although Kant claimed that "our duties to ourselves are of primary importance and should have pride of place" and added that "the prior condition of our duties to others is our duties to ourselves", he was quite aware that duties to oneself involve, at first sight at least, a contradiction.
In the Metaphysics of Morals, Kant insists on the foundatio-nal role of duties to oneself. He presents them as precondition of all obligation to others as well 23 . But at the beginning of the section where he develops this point, he mentions a puzzle about the possibility of duties to the self 24 : does the self not have to be active and passive at the same time, which is impossible? If the "one imposing obligation and the one put under obligation were identical, could not the former always release the latter -we ourselves-from an obligation?". Wouldn't it involve a contradiction? For Kant the contradiction is only apparent, because actually it is not the ordinary or phenomenal self, that is, the one governed by the laws of nature, who is placing himself under the obligation, but what he calls the noumenal self, the one that is endowed with inner freedom. It is only qua noumenal self that one can freely place oneself under an obligation. But as fantastic as the powers of the noumenal self might be, it is hard to see how they may allow him to have a duty to himself, if it is a contradictory notion. It might be that it is in order to block this objection that Kant appeals to a totally different argument. "For suppose there were no duties to oneself; then there would be no duties whatsoever, not even external duties" 25 .
At first sight, Kant tries here to make use of his notion of autonomy, that is the idea that we can freely give a law to ourselves, in order to show that duties to the self are conceivable 26 . But, as it has been noticed, "it would be surprising if the notion of autonomy and the notion of duty to oneself turned out to be identical" 27 .
In Kant's eyes, the idea that we can freely give a law to ourselves may ground moral duties to others as well as duties to the self. But from this it does not seem to follow that, for him, there is no difference between duties to others and duties to the self, or that the idea of autonomous lawgiving is identical with the idea of duty to the self. In any case, it has been suggested that we should take Kant's reasoning here as a reductio ad absurdum and read it: "If they were no duties to the self, there would not be duties at all. But there are duties, therefore there are duties to the self " 28 .
Marcus Singer calls this argument a blatant "non sequitur". I agree. It seems to me that we would still have external duties, that is, duties to others, even if we had no duties to ourselves. It might be that if we had no duties to ourselves, we could not fulfil our duties to others, but this is a totally different story. And even this story doesn't seem promising. After all, what we need in order to fulfil our duties to others is not to be able to respect our duties to ourselves but to have the motivation or the will to fulfil our duties to others. A number of other arguments have been raised against the idea of duty to the self. Bernard Williams, for example, claims that there is nothing more in the notion of duty to the self than the idea of selfinterest. For him, using the word "duty" is just a fraudulent and pernicious way of speaking of self-interest 29 . It could be a perfect example of the fallacies of modern moral philosophy. But of course, by mocking duties to the self, Williams is not at all pleading for some kind of self-other asymmetry. As a supporter of Ancient ethics, he will not deny that self-care has an ethical importance. I myself take the argument against duties to the self as a first step in a defence of common sense moral asymmetry, as it is expressed by the harm principle. The second step could be a criticism or self-regarding moral virtues like temperance or endurance. But I leave it for another occasion.

CONCLUSION
Instead of presenting as a conclusion what I have already said, I will try to push my questions a little further. Could we conceive of an impartialist theory which would make room for self-other asymmetry? 30 Actually, impartialist moral outlooks could make room for selfother asymmetry in a very simple way. The impartialist may say that if we reason from the impartial point of view, we will be brought to the conclusion that harm to self is morally indifferent. But why would we be brought to this conclusion? The claim would need further justification and it is not sure that it could be given. Another way to reconcile impartiality and self-other asymmetry seems to be more promising.
It consists in specifying first what can be called the "scope of morality". In this area, I would separate "maximalists" and "minimalists" 31 . I call "maximalists" those who give the largest scope to morality. They include in morality what we owe to ourselves, what we owe to others and what we owe to abstract entities like the needs of society or symbolic entities like the flag of the nation. By contrast, for "minimalists", the scope of morality is very narrow. It covers only what we owe to each other as individuals.
Once the scope of morality is fixed, our moral concepts apply in this limited area only.
For a maximalist, moral impartiality will apply to everyone, myself included and to many other things as well. A maximalist won't be able to be committed to common-sense self-other asymmetry and an impartialist at the same time. If he is an impartialist, he will have to deny common-sense self-other asymmetry.
For a minimalist, moral impartiality will apply only to our relations to other people. A minimalist will be able to be, at the same time, an impartialist in his relations to others and to be committed to common-sense self-other asymmetry.
I think this is a good point for moral minimalism.