II. BACKGROUND

Let me begin with some background information. Nearly everyone believes that showing respect for persons is a fundamental moral duty, but theorists differ on how to interpret it. Utilitarians, for example, argue that the egalitarian aspect of their theory – the idea that each is to count for one and nobody for more than one – suffices for showing respect for persons. Some Liberals argue that they respect persons by protecting individual rights and by insisting on state neutrality between competing conceptions of the good life.Footnote ¹ Naturally, there are many disagreements over who is actually complying with this duty. For instance, deontologists often criticize utilitarians for not taking the separateness of persons seriously, and thus failing to show persons the proper respect.Footnote ² These disagreements, however, are over the principle's interpretation, and not over its appropriateness as a foundational principle. Most believe that any political system that, on a fundamental level, does not treat persons with respect is deeply flawed.

Many also believe that the choices people make for themselves, and the preferences on which these are based, should also be respected. Indeed, the contractarian view of political obligation would make little sense without this idea, for unless choice is worthy of respect, the fact that people would voluntarily choose to abandon the state of nature and live in civil society would merely be a psychological fact without the desired normative implications.Footnote ³ Many utilitarians also believe that individual preferences should be respected. Some think this because they believe that preference satisfaction is intrinsically good, whereas others tend to think that persons are happiest when allowed to choose their own ends.

It is also natural to suppose that many who believe that preferences ought to be respected believe this because of their commitment to respecting persons. Historically, Kant argued that treating persons, or humanity, as an end in itself requires that we treat the ends of others as our own.Footnote ⁴ Thomas Hill Jr., for instance, interprets Kant to mean that ‘in respecting the dignity of humanity in a person, one is to value another's achievement of a (morally permissible) end because it is an end he adopted rather than because one expects it will bring him pleasure or something regarded as intrinsically valuable apart from his choice’.Footnote ⁵ Other contemporary scholars also endorse the connection between respecting persons and respecting their preferences. For example, Robert Goodin argues that ‘the primary reason we talk in terms of respecting people's choices is to remind ourselves of the reason for respecting them. The reason is not that they are choices but rather that they are choices of people.’Footnote ⁶ Of course, there are important views, such as mental state utilitarianism, that deny this connection, but it is nevertheless natural to suppose that there is a link between respecting persons and respecting their preferences, and that the latter derives its value and importance from the former.

Yet even if this connection is granted, its nature remains obscure and contentious. In order to show a person the respect she is due, which of her preferences must the state respect?Footnote ⁷ I will argue that a preference's autonomy determines whether a duty to respect persons entails a further duty to respect it. I begin by rejecting the view that a duty to respect persons entails that preferences qualify for respect in virtue of being constitutive of our identity. This naturally leads into the positive arguments for my view. I then argue against the view that the state has only negative preference-based duties and Ronald Dworkin's view that showing respect for persons obligates the state to consider only personal and not external preferences. I conclude by considering an objection to my view which concerns whether immoral autonomous preferences are worthy of respect.

III. SHOULD CONSTITUTIVE PREFERENCES QUALIFY FOR RESPECT?

Some might argue that treating persons with respect requires that their ‘parts’ be respected, especially if these are constitutive of their personal identity. Indeed, it seems true by definition that in order to respect a person, we should respect his essential qualities, or that which makes him the person that he is. Therefore, if some preferences can be constitutive in this sense, then, on this view, a duty to respect persons would generate a further duty to respect these preferences.Footnote ⁸ Though this view is initially plausible, I will argue that it fails because it overlooks the counterfactual nature of the duty to respect persons.

Showing respect for you as a person does not solely entail that we respect that which makes you a person, or even that which makes you the particular person that you are, but also the person you would have been had you not been exposed to certain distorting influences. Intuitively, preferences that are constitutive of your personal identity but patently non-autonomous do not appear to qualify for respect. For example, you might have an unshakable preference for a certain drug that you desperately wish to be rid of, or at least not be motivated by. Even if this preference is constitutive of your identity, it seems absurd that a duty to respect you as a person would generate a reason to help you satisfy it, or that not helping you satisfy it would be showing disrespect for you as a person. That is, not helping you satisfy this preference does not seem to show you any disrespect at all. Indeed, the more plausible view is that showing respect for you as a person requires helping you overcome this addictive preference rather than helping you satisfy it. Or suppose that you emerge from a compulsory ‘re-education’ with radically different preferences and goals. Indeed, to many, you appear to be an entirely different ‘person’ altogether. If you emerge from your confinement with, say, vastly different political preferences, surely the extent to which these are the product of coercion and manipulation should enter into the state's deliberations when determining their proper political role even if they are now constitutive of your personal identity. If, as a direct result of your confinement, you have a new constitutive preference that a particular politician be elected, a duty to respect persons does not seem to obligate the state to promote its satisfaction, nor is it obvious that facts about this preference itself, apart from the effects of its satisfaction, provide reasons why it should be satisfied. There may be a reason to satisfy your preference if, say, the candidate you endorse is superior to all the others, but this reason does not stem from your preference's intrinsic features. Finally, suppose that a powerful wizard can directly implant preferences into people and also manipulate their preference profiles to render these preferences constitutive of their personal identity. While the state may have welfare-based reasons to take these preferences seriously, there seems to be no reason to press for their satisfaction if the consequences of doing so are neutral or negative with respect to welfare. If this wizard implants in me a constitutive preference, say, to fly kites, but the satisfaction of this preference would have no consequences for anyone's welfare including my own, I maintain that there is no reason to press for its satisfaction.

In short, a preference can be both constitutive of your identity but have a dubious history or current status that makes it unworthy of respect. Accordingly, showing respect for persons compels the state to consider other facts about a person's preferences aside from whether they are constitutive of his personal identity. While there may be many reasons for taking these preferences seriously, merely being constitutive of your personal identity does not make them worthy of respect. Facts about these preferences themselves, apart from the effects of their satisfaction, do not provide reasons why they should be satisfied. Nor is being constitutive of one's personal identity a necessary condition of a preference being worthy of respect, as I show in the next section.

IV. AUTONOMOUS PREFERENCES

In this section, I will argue that a preference's autonomy makes it worthy of respect, and not other facts about it such as whether it is constitutive of one's identity. I offer two arguments for this claim. The first is a direct appeal to intuitions. The second makes note of an overlooked feature of autonomous preferences – namely the unique way that they are connected to their bearers. Before proceeding, I should note that I do not provide a comprehensive account of what makes preferences autonomous. Instead, I focus on paradigm cases of non-autonomous preferences to make my case. Since, however, even the notion of a paradigmatically non-autonomous preference may be somewhat obscure, a brief look at the current state of research on autonomy may be helpful. For present purposes, an informal discussion highlighting the main features of the dominant views should suffice.

Modern conceptions of personal autonomy come in two varieties.Footnote ⁹ First there is the capacity view, according to which being autonomous consists in nothing more than having certain capacities such as an ability to formulate a life plan or an ability to identify with your first-order desires.Footnote ¹⁰ The relevant capacities can be described as latent psychological abilities possessed by most humans.Footnote ¹¹ Typically, this view is accepted by those who want autonomy to play a crucial role in moral argument, such as in defining the nature of personhood or grounding universal human rights. The concept of autonomy, however, must also play an important role in political argument, and here the capacity conception falters. A full defense of this claim would take us too far away from this article's main topic, but briefly, note that it seems possible to undermine someone's autonomy without affecting any of his aforementioned capacities. For instance, on the capacity view it appears that someone involuntarily chained to a dungeon wall is not even likely to be non-autonomous because his latent psychological abilities, say, to identify with his desires or to formulate a life plan, can be fully intact. All else being equal, a person who is free to live according to his conception of a worthwhile life appears considerably more autonomous than the aforementioned prisoner, but this clear intuition cannot be explained with the capacity view.

An alternative to the capacity view is the achievement conception. On this view, merely having certain latent psychological abilities does not suffice for autonomy. Rather, autonomy requires that one actually be self-determining, self-directing, or the author of one's life.Footnote ¹² On this view, autonomy requires actual self-government, and not merely its capacity.

There are at least three kinds of achievement conceptions: historical, structural, and reason-oriented. According to the historicists, a person is autonomous if his preferences or motivations originated or developed in the right way. One widely shared view of a ‘wrong’ way for a preference to come about is under conditions of extreme manipulation or social pressure.Footnote ¹³ According to the structuralists, a person is autonomous if his preferences or motivations presently cohere in the right way with other elements of his character and values. Many structuralists argue that the right coherence relation is that of identification, where, roughly, one is said to identify with one's preferences if one's second-order preferences are consistent with one's first-order preferences.Footnote ¹⁴ A third school maintains that there is a tight connection between autonomy and practical rationality.Footnote ¹⁵ On this view, persons are autonomous if, in some sense, they are responsive to reasons.

While these views are typically presented as accounts of autonomous persons, there is room within all three for the notion of an autonomous preference.Footnote ¹⁶ On the historical view, we can say that a preference is autonomous if it came about or was developed in the right way. On the structural view, a preference is autonomous if it coheres in the right way with other elements of one's character and values. On the reason-oriented view, we can loosely say that a preference is autonomous if its bearer has good reasons for having it, for endorsing it, or for making it effective in his actions.

Call a preference paradigmatically non-autonomous if it is clearly non-autonomous, or nearly so, on all three accounts. Call a preference paradigmatically autonomous if it is clearly autonomous, or nearly so, on all three accounts. I will make my case using these kinds of preferences as examples. To be sure, some of my examples feature preferences that clearly violate only two out of these three accounts, but when all is said and done all three views will be adequately represented.

V. THE CASE FOR RESPECTING AUTONOMOUS PREFERENCES

Two paradigmatic instances of non-autonomous preferences are those that are secretly implanted into the agent without the usual rational screening process and those that are unwillingly compulsive or uncontrollable. If we consider these, it should be clear that a duty to respect persons does not entail that either the state or anyone else must respect them. Consider first a case of the former. Suppose that Jim, who is an avid reader of biographies, is surreptitiously hypnotized to prefer only detective fiction by an unscrupulous publisher.Footnote ¹⁷ Being a voracious reader, he proceeds to plow through this genre's many offerings. If the state is considering a new tax that will make this preference much harder to satisfy, should Jim's newly acquired preference be taken into consideration? All things considered, it might be better if Jim's preference is satisfied for this might make both him and the publisher happier. But surely the reasons in favor of its satisfaction will not stem from facts about the preference itself. Apart from the expected effects of its satisfaction, there are no facts about this preference that make it the case that it ought to be satisfied, or that the state should reconsider the proposed tax. Insofar as the state has an obligation to promote the general welfare it may reconsider its tax because Jim may suffer if it is enacted. But if no favorable or unfavorable consequences would follow from its satisfaction, I maintain that there is no reason why it ought to be satisfied.

Indeed, if we consider a more sinister version of this case, a duty to respect persons might entail that we help Jim overcome this preference rather than help him satisfy it. Suppose that this preference is detrimental in the long run because Jim now spends all of his disposable income on detective novels. Suppose that the state is considering enrolling Jim in a publicly funded program to overcome the effects of hypnosis. Should the fact that this conflicts with his current preference give the state a reason to reconsider? Again, perhaps it does if Jim would be greatly distressed by the ordeal. But, putting aside these worries, the mere conflict with his non-autonomous preference should give the state no pause whatsoever.

Our intuitions pull us in the other direction when the crucial preference is autonomous. Suppose that Mary has recently acquired a new preference to paint. Suppose she took several art courses in college, became interested in painting, and began exploring the city's galleries. Recognizing both a talent and a love for it, and noting the happiness and tranquility it brought her, she decided to pursue it professionally.Footnote ¹⁸ Surely here we are far more sympathetic to the claim that a duty to respect persons gives the state a reason to take this preference into consideration, apart from the effects of its satisfaction, when weighing policies that may affect it.Footnote ¹⁹ If the state is considering raising taxes on paintbrushes, facts intrinsic to Mary's painting preference give the state a reason to reconsider. One might argue that this is because Mary has made a large investment in painting and would be harmed by the new tax. But although a new tax policy's effect on personal welfare is relevant to whether it deserves state sanction, this is not the issue here. Presumably, one can have one's preferences thwarted and yet not suffer in other respects or have them satisfied but not have more positive experiences as a result. My claim is that even if the state's proposals would have no effect on Mary's well-being aside from the satisfaction of her preference, or even a negative effect on her well-being, there is still a reason to give weight to her preference, and the same does not hold when the crucial preference is non-autonomous.Footnote ²⁰ At the very least, there would be less reason to give her preference weight if, all else equal, it resulted from hypnosis, if it didn't cohere with other elements of her character, and if there was no reason to have, to endorse, or to act on it.

Consider now a case of a compulsive preference. Suppose that I unwittingly develop a nearly irresistible desire for a certain drug. In a sober moment, I beg you to keep me away no matter how forcefully I demand it. If I then come to you and insist that you give me access to it, should you refuse my plea? Surely we can imagine many reasons for or against doing so. You might grant me access despite your promise because you judge that, in my current state, my life is not worth living. Or, you might refuse my request because of the binding force of your promise. While what you should do may be unclear, it is clear that, independent of the expected effects of this preference's satisfaction, the fact that I now prefer this drug does not give you any reason at all to satisfy my current preference or to believe that it ought to be satisfied. While there may be reasons to take my preference seriously, there are no reasons to respect it.

So far, I have been treating respect for persons as a deontological notion. But this principle plays a central role in teleological theories as well. One could reach a similar conclusion by considering it as an element of the good. On this view, we could say that if a preference is worthy of respect then there are facts about this preference itself, apart from the effects of its satisfaction, that make its satisfaction a good thing. Intuitively, the mere satisfaction of certain kinds of preferences does appear to be good even if it is not good for anybody. Consider again Jim's preference for detective fiction. If we imagine two worlds that differ only with respect to this preference's satisfaction, then, intuitively, I claim that neither of these worlds is any better than the other. To be sure, we must imagine that his preference's satisfaction has no other effects such as changes in Jim's happiness, pleasure, or well-being. Intuitively, the matter is different for Mary's preference. If one world differs from another only with respect to the satisfaction of her preference for painting, I contend that this world is also slightly better, all things considered.Footnote ²¹ Or suppose that, in the relevant respects, world A is full of people like Mary while world B is full of people like Jim. I contend that, in at least one respect, world A is better than world B.

To reinforce this point, consider some more typical cases where an autonomous preference's satisfaction is expected to make its bearer worse off. Some people have powerful desires to participate in dangerous activities or experiment with dangerous substances. Suppose that you control the supply of some drug, and two people approach you and request equal portions of it. From a moral point of view, it seems that you should consider several factors in deciding whether to accommodate them, such as the benefits they would receive from the drug, the possibility of them harming themselves, and the possibility of them harming others. But surely the nature of their desire for the drug is also relevant. To see this, suppose that these people are equal in every respect but one – customer X is an unwilling addict whereas customer Y is not. In some sense, X is barely able to control his cravings while Y can, even though their cravings are equally intense. Surely in this case there is more reason to grant Y's request than X's. Indeed, even if you know that Y's drug use is causing him greater harm than X's, there still appears to be more reason to grant Y's request. As I will argue in the next section, this reason is intimately linked with the notion of showing respect for persons.

The matter is similar for people who participate in dangerous sports. The autonomy of their desire is a crucial factor in determining whether to facilitate their activity. A world of autonomous extreme mountain climbers is better than a world of non-autonomous ones, all else being equal. Indeed, the former seems better than the latter even if all else is not equal. Even if there is more pain and suffering in the former world owing to climbing-related injuries, there is still something to be said in favor of that world. If my desire to climb Mount Everest is just an inexplicable compulsion or is secretly implanted in me by a powerful wizard, surely you have less reason to encourage me – perhaps no reason at all – than if my desire is paradigmatically autonomous.

Accordingly, non-autonomous preferences do not qualify for respect. Facts about these preferences themselves, apart from the effects of their satisfaction, do not provide reasons why they should be satisfied or why it would be good if they were satisfied. Intuitively, the same does not hold for autonomous preferences. Thus far, however, my case rests entirely on intuitions that some might not share. Thus, in the next section, I attempt to provide a more principled justification for my claims.

VI. WHAT MAKES AUTONOMOUS PREFERENCES RESPECT-WORTHY

Autonomous preferences qualify for respect in virtue of the nature of their attachment or connection to their bearers. I will argue that there is a unique way in which autonomous preferences are connected to their bearers that both distinguishes them from their non-autonomous counterparts and also makes them worthy of respect. This view starts with what may be called the deep insight of the view that constitutive preferences qualify for respect. That view began with the premiss that a duty to respect persons entails a further duty to respect that which is connected or attached to persons in a certain way. My view accepts this premiss but not the further claim that the respect-generating relationship between a person and an attachment is one of constitution.

Begin with the observation that, in a sense, our preferences are attached to us, but not all of our preferences are necessarily attached to us in the same way. Some settle into us without our recognition, such as preferences for certain foods and climates. Typically, these preferences are, as it were, non-consciously absorbed from our environment rather than actively cultivated or endorsed, though this is not always the case. Other preferences come attached to us from birth. Many believe that some of our preferences are innate, such as those for certain body types, certain arrangements of light and shadow, and perhaps even for landscape design.Footnote ²² Some preferences, such as those that involve addictions, remain attached to us despite our best efforts at dislodging them. When our preferences fit these descriptions, we can say that we are merely their bearers. They are ours, but only in a superficial sense. However, when we take an active role in cultivating them during their development or we endorse or identify with them, we become more than just their bearers. In a sense, we become their owners.Footnote ²³ By cultivating them during their development, presently endorsing or identifying with them, and maintaining them for good reasons, we change the nature of their attachment to us. By cultivating a preference, we are, in a sense, attaching it to ourselves. By endorsing or identifying with it, we are approving of its attachment to us. By actively engaging with a preference in this manner, we are making it our own, and hence our connection to it is significantly different from our connection to those preferences that we merely bear.

Now consider the metaphors typically used to describe autonomy in the achievement sense: self-government, self-determination, self-direction, or being the author of one's life. These metaphors, which play a central role in our understanding of autonomy, share an important common feature. Each ascribes autonomy only to those persons whose ‘selves’ exhibit some control over their motivations, actions, and lives. Since, presumably, you are not in control of your addictive preferences, innate preferences, or those that non-consciously settle into you without your recognition, these are not good candidates for autonomy.Footnote ²⁴ At the least, being in control requires an active evaluation of your motivations. An autonomous life is opposed to a thoroughly passive existence characterized by a lack of self-reflection or the lack of critical scrutiny of your preferences and goals.Footnote ²⁵ the structuralist view, for instance, it is not enough that you merely have coherent first- and second-order desires. Rather, you must identify with your first-order desires, which requires an active evaluation of them. On the reason-oriented view, it is not enough that there are reasons in some abstract sense that support your actions. Rather, you must recognize these reasons and make them effective in your actions. A more plausible historical view would maintain that it is not enough for your preferences to have originated in the absence of certain distorting influences. After all, this could be entirely a matter of luck. Rather, this view should maintain that you must cultivate your preferences during their development in order for them to be autonomous. Thus, while all three views disagree about how you make a preference your own, they all agree that you must make a preference your own for it to be autonomous. That is, they all agree that autonomy requires an active engagement with your preferences. Hence, an autonomous preference is one that the agent's ‘self’ has actively engaged with in a manner that makes it her own. Whether she does this by recognizing reasons for having it, by cultivating it during its development, or by identifying with it, is not at issue here. The crucial point is that, in some way, an agent must make a preference her own in order for it to be autonomous.

If we consider the aforementioned cases of Mary and Jim in this light, crucial differences emerge in the nature of their attachment to their preferences. In some sense, Mary chose and cultivated her painting preference, whereas Jim had his preference for detective fiction thrust upon him. Mary, but not Jim, attached her preference to herself, or at least endorsed its attachment to her. She is the owner of her painting preference, while Jim is merely the bearer of his preference for detective fiction. In the addiction case, I was overcome by the preference's force, and, in a sense, it had control over me rather than the other way around. Since it remained attached to me despite my best efforts at dislodging it, I am more aptly described as its bearer rather than its owner.

These differences in the nature of an autonomous preference's attachment explain why it is worthy of respect. To see this, consider the following case. Imagine that, while out for a walk, a leaf happens to settle on top of Bob's head without his recognition.Footnote ²⁶ If you were to come over and flick it off, it does not seem that you would be showing disrespect for Bob as a person. After all, his attachment to this leaf is purely accidental and not deliberate. But suppose that Pierre, an avant-garde artist, spends hours in front of a mirror arranging a leaf on his head so that it is in just the right position to convey his aesthetic sensibilities, and, knowing this, you flick it off his head anyway. This would be an act of disrespect, and it seems that what distinguishes these cases is the nature of the attachment between the person and the leaf. Unlike Bob, Pierre ‘actively engaged’ with the leaf. He deliberately placed it on his head and endorsed its presence, whereas the leaf just happened to settle on Bob's head without his recognition. This shows that the nature of the connection between a person and his attachments is crucial for determining our other-directed obligations. We can conclude that if we want to show a person the respect she is due, we must show respect for some of her attachments.

But which attachments? If they are thrust upon her, merely settle into her without our recognition, or persist despite her best efforts at dislodging them, then, as in the case of Bob and the leaf, showing respect for persons does not obligate us to respect these attachments. But if she cultivated them, presently endorses them, and acts on them for good reasons, then respecting her as a person, and as the person that she is, entails respecting these attachments as well. Notice that, in the relevant respects, Mary's attachment to her painting preference is very much like Pierre's attachment to the leaf whereas Jim's attachment to his detective fiction preference is very much like Bob's attachment to the leaf. To respect Pierre as a person, some attention should be paid to the attachments he has formed, and not to the attachments he merely happens to have. Similarly, to respect Mary as a person, and as the person that she is, some attention should be paid to the attachments she has formed, and not to the attachments she merely happens to have. This is simply what's involved in treating Mary as a free and rational being capable of controlling her attachments. Mary's autonomous preferences have a certain connection to her self that her non-autonomous preferences do not. If Mary is deserving of respect because she is free, rational, and responsible, then the attachments she freely chooses and is responsible for maintaining must also qualify for respect. The person-based obligations we have toward Mary transfer to her autonomous preferences.

On my view, then, autonomous preferences qualify for respect in virtue of the unique way they are connected to their bearers. In the next section, I argue that the state's duties toward these preferences are not entirely negative. I then consider and reject Ronald Dworkin's view that respect for persons obligates the state to consider only personal and not external preferences, and I conclude by weighing the state's obligations toward immoral autonomous preferences.

VII. THE STATE'S POSITIVE PREFERENCE-BASED DUTIES

Some might argue that the respect due to autonomous preferences should be cashed out wholly in terms of negative duties. Indeed, the case of Pierre supports this conclusion since, in that case, our duty was not to interfere with Pierre's leaf, which seems like an instance of a negative duty. I, however, maintain that the state also has positive duties towards autonomous preferences and that these duties are grounded in the state's obligation to respect its citizens as persons.Footnote ²⁷ What justifies this conclusion?

First, notice that the case of Pierre can be modified so as to involve a positive duty. Suppose that I am aware that Pierre is searching for a certain rare leaf to express his aesthetic sensibilities. If I should come across this leaf while strolling through the park, I take it that I would have a reason to pick it up and give it to him irrespective of welfare considerations. That is, the mere fact that his desire for the leaf is autonomous provides me with a reason to retrieve it for him. No such reason exists if his desire is, say, just an inexplicable compulsion or was implanted in him by a powerful wizard.

Of course, one could reject this intuition and argue that the state should show respect for persons by remaining neutral in all matters of preference. Indeed, several theorists have suggested that the hallmark of the liberal state is its commitment to neutrality between competing conceptions of the good life, and that this can be justified by appealing to the value of autonomy.Footnote ²⁸ But, applied to preferences, this view is implausible. To see this, first note that the idea of neutrality can be interpreted in many ways, such as that the state should refrain from influencing preferences during their formation or that it shouldn't alter people's preferences. But any view that literally requires the state to avoid influencing preferences is implausible because, in light of the state's size and scope, it can't help but affect individual preferences. The state is simply too large a player in our lives to remain neutral in this sense. Of course, we might interpret this negative duty of neutrality to mean that the state should not justify its policies on the basis of preferences. But while this might be a reasonable view of the state's role vis-à-vis conceptions of the good life, it is deeply implausible as an account of the state's preference-based duties. This is because we rightly believe that, typically, popular support provides a compelling reason for state action, and that, in sufficient quantity, preferences can create a mandate for state policy. If a large majority prefers, say, a single-payer health care system, then that is a strong, though perhaps not conclusive, reason to implement one. The opposing view, however, leads to the opposite conclusion – that overwhelming popular support provides a compelling reason for the state to withdraw its support and attention. On this view, the fact that a large majority prefers a single-payer health care system provides no reason at all for the state to implement one because otherwise the state would be justifying its policies on the basis of preferences. That is, if the state cannot justify its policies on the basis of preferences, then an issue's popular support provides no reason at all for the state to advocate it. This can't be right. While mere popular support may not provide a conclusive reason for state sanction, surely it provides at least some reason for it.

Since the view that the state has only negative preference-based duties is implausible, we are left with the question of which preferences the state has positive duties towards, and not whether the state has any such positive duties. For all of the aforementioned reasons, I suggest that autonomous preferences are extremely plausible candidates for the objects of these positive duties. First, as I argued in section V, the fact that a preference is autonomous provides a reason why it should be satisfied and why it would be good if it were satisfied. Second, as I argued in section VI, the state already has positive duties toward these preferences in virtue of the nature of their attachment to persons. The nature of their attachment to persons explains why autonomous preferences should be singled out as objects of the state's positive duties.

I now consider and reject Ronald Dworkin's proposal regarding which preferences the state has positive duties toward.

VIII. RONALD DWORKIN ON RESPECTING PREFERENCES

Dworkin's argument that the state should consider only its citizens’ personal and not their external preferences is perhaps the most notable attempt to specify which preferences should qualify for political respect. His arguments are especially relevant here because they are based on the premiss that the state must treat its citizens with equal concern and respect, which is his interpretation of the duty to respect persons. Admittedly, his arguments are somewhat dated and it is not clear whether he still endorses them. Still, they are the most thorough examination of this issue, so they warrant careful attention.

Dworkin provides a lengthy discussion of the personal/external distinction in the context of affirmative action.Footnote ²⁹ Generally, he holds that when rights are not at stake, public policy should be determined by appeal to the common interest, which, on his view, is determined by the policy that ‘would satisfy on balance more preferences, taking into account their intensity, than alternative policies’.Footnote ³⁰ On his view, nobody has a right to be admitted to a professional school, so the legitimacy of affirmative action programs must depend on the utility calculus. But then it seems that racist admissions policies may be justified if enough people prefer them with sufficient intensity. For instance, if a white majority strongly prefers that blacks be denied admission to law school, then it seems that, on his view, this would be justified. Yet he claims that these kinds of preferences can be justifiably excluded from the calculus because they are external, and only personal preferences need to be counted when crafting policy. Personal preferences are preferences for ‘one's own enjoyment of some goods or opportunities’, while external preferences are preferences for ‘the assignment of goods and opportunities to others’.Footnote ³¹

Assuming that this distinction can be clearly drawn, why exclude external preferences from the utility calculus? Dworkin's answer is that otherwise utilitarianism's egalitarian component – the idea that each is to count for one and not for more than one – would be corrupted. He explains that the problem with counting external preferences is that some will be treated as ‘more than one’. If others outvote me by combining their personal and external preferences, Dworkin argues that I am no longer being treated as an equal, and thus no longer being treated with the respect I am due as a person. His argument for this appears in the following passage about a community decision over whether to invest in a pool or a theater:

This passage suggests that Dworkin rejects counting external preferences because this will result in double-counting. Those with mutually reinforcing personal and external preferences will get two votes, whereas everyone else will get only one, and thus utilitarianism's egalitarian component will be undermined.Footnote ³³ To put it another way, Dworkin is suggesting that this form of double-counting is analogous to some literally getting more votes than others by being permitted to vote twice. Since this is an obvious infringement of our right to be treated as equals, then, if these cases are indeed analogous, counting both personal and external preferences also violates this right.

But are they analogous? Consider a different analogy. Suppose that we can distinguish short-term from long-term preferences, and it turns out that some have only short-term preferences regarding the pool and the theater. Just as those with mutually reinforcing personal and external preferences will be double counted, it appears that those with reinforcing short and long-term preferences will also be double counted. After all, they will, as it were, insert two tickets into the hopper – one for their short-term and one for their long-term preferences. If they vote sincerely, those with only short-term or only long-term preferences, but not both, will insert a single ticket into the ballot box. Or suppose we distinguish national from local political preferences. Again, it appears that those with reinforcing national and local political preferences will be ‘double counted’ in exactly the same sense that those with personal and external preferences are double counted in Dworkin's example.Footnote ³⁴ Yet clearly this is not objectionable, and it can't be argued that this constitutes a violation of one's right to be treated as an equal. In Dworkin's system, the fact that some cast more votes than others does not by itself constitute a rights violation, so long as everyone's vote is counted and nobody is denied an opportunity to vote.

Indeed, why can't those who are suffering because of others’ external preferences also cast two votes – one for their personal and another for their external preferences? There is no clear answer in Dworkin's essay. Time and again he discusses how racist attitudes on the part of some cancel out the personal preferences of others if they are counted, but there is no explanation for why the latter group cannot offset this with their own external preferences. One might argue that the minority might not have the requisite external preferences, but while this may be unfortunate for them, it is no more unfortunate than someone who is outvoted because his preferences lack intensity, because he's in the minority, because he lacks the necessary preferences, or because his short-term and long-term preferences are not mutually reinforcing. If people were allowed to cast two votes for the president, most would vote twice for the same candidate. Some, however, may vote for each of the two candidates, or just cast a single ballot, thereby diminishing their vote's influence. But this is not the same as their vote not counting at all. If they are barred from the voting booth, then their right to be treated as an equal has been violated, but if they choose to split their vote, or not cast it, then this right is not infringed.

Indeed, though framed in the language of double-counting and mutually reinforcing preferences, one gets the impression that Dworkin's actual concern lies elsewhere. The problem is not that some will have more influence than others on the final outcome, but that some can secure just as much welfare for themselves by thwarting the preferences of others as by having their personal preferences satisfied. Suppose that the welfare I would receive from attaining an important office is identical to the amount I would receive from thwarting my enemy's similar ambition. In terms of overall welfare, suppose that the best outcome is that we both attain this office, second best that I attain it and he does not, and third that neither of us attains it.Footnote ³⁵ On a strict preference view, however, the second outcome may turn out to be the best, the third the second best, and the first the worst. Indeed, by registering my external preference that my enemy be rejected, I may be able to secure as much for myself as I would by registering my personal preference that I get the job. This outcome is clearly inferior for the welfare utilitarian but not necessarily for the preference utilitarian. Perhaps, then, Dworkin is worried that by counting external preferences we are more likely to create a situation that is perfectly acceptable on preference utilitarian grounds but markedly inferior with respect to overall welfare.

Though I find it hard to believe that this is his actual worry, some of his comments suggest this interpretation. For instance, when defending himself against Herbert Hart's objections in a later essay, he asks us to consider the case of Sarah, who is so beloved by her countrymen that many prefer that her preferences be counted twice.Footnote ³⁶ Naturally, she reaps the benefits of being so adored, and her admirers are greatly distressed when their wishes are thwarted. As before, Dworkin claims that this violates utilitarianism's egalitarian component.

He has two arguments for this claim, only one of which is relevant here.Footnote ³⁷ Hart objects that Sarah's preferences would not be double-counted, but that if the Sarah-loving preferences of others were discarded this would undercount their preferences. Dworkin, however, argues that this is mistaken because preferences are not like votes in that they are exhausted after being registered. Rather, one can always add external preferences on top of personal ones, and thereby increase the influence of one's preferences. As Dworkin explains,

Interpreted as a case of multiple counting, my previous objections still apply. On another reading, however, Dworkin's claim that those who give their votes to Sarah are securing more for her without themselves sacrificing anything may reveal his actual concern. By registering their external preferences, the Sarah-lovers may be creating an inferior situation with respect to overall welfare since, had they exercised their personal preferences, they presumably would have attained a commensurate degree of welfare for themselves, less welfare for Sarah, but much more for those whose preferences have been thwarted by the weight of Sarah's amplified personal preferences. To clarify, in Dworkin's system, if everyone prefers a larger slice of pie to a smaller one, but a large number of people also prefer that Sarah's preferences are satisfied, Sarah might end up with most of the pie while a sizeable part of the population receives virtually nothing at all. This might be acceptable on preference utilitarian grounds, but it is unlikely to maximize overall welfare.

This is clearly a legitimate worry for anyone espousing a preference utilitarian framework, but it is doubtful that it can be avoided by appealing to utilitarianism's egalitarian component. If the problem with preference utilitarianism is that it can result in non-optimal welfare distributions, the solution cannot lie in arguing that its victims are being denied their right to be treated as equals. The problem, rather, is that on this interpretation, being treated with equal concern and respect creates sub-optimal welfare distributions. This cannot be resolved with the personal/external distinction.

IX. AN OBJECTION

An initially compelling objection to my account concerns the state's obligations toward the autonomous but immoral preferences of its citizens. If the state has an obligation to help persons satisfy their autonomous preferences, and if autonomous preferences can be immoral, then, on my account, it appears that the state has an obligation to help murderers murder and thieves rob if their desires to do so are autonomous. This seems deeply counterintuitive.

First, let me note that many modern accounts of autonomy do not preclude the possibility that an immoral preference can be autonomous. Presumably, such a preference can have the right history to satisfy the historicists, it can cohere with other elements of one's character to satisfy the structuralists, and there can be good reasons for acting on, having, or endorsing such a preference to satisfy those who connect autonomy to practical rationality. So, on my account, it appears that the state has an obligation to help persons satisfy immoral preferences if they are autonomous. To address the obvious worry this raises, I must first remind the reader that I am only defending a prima facie obligation. While there may be some reason for the state to help autonomous murderers to murder, presumably this is outweighed by the reasons to prevent them from doing so. Still, my critic may not be placated, for perhaps it is implausible to maintain that there is any reason at all to help the autonomous thief or murderer, even if that reason pales in strength against the reasons to prevent him from robbing or killing. On the basis of intuitions alone, however, it is difficult to distinguish cases where there is no reason to do X from cases where there is some reason to do X, but this reason is swamped by opposing reasons. How should we proceed?

Here it is useful to recall previously discussed cases, such as the case of Mary, where the satisfaction of an autonomous preference had no effect on anyone's well-being. In those cases, I argued that a preference's autonomy provides a reason to press for its satisfaction despite the fact that nobody's well-being is affected. I then argued that even when the satisfaction of an autonomous preference will make its bearer worse off all things considered, such as in the case of someone engaged in a dangerous sport, there is still an autonomy-based reason to press for its satisfaction. I believe the same applies when the satisfaction of an autonomous preference leaves others slightly worse so long as the relevant action is not wrongful. Presumably, if the autonomy-based reason survives when the consequences of the desire's satisfaction are slightly negative for others, then it also survives when others are made significantly worse off. It is just that the more worse off others are made, the less overall pull the autonomy-based reason in favor of the preference's satisfaction has. One might extend this reasoning to suggest that even when the preference in question is immoral, the autonomy-based reason in its favor still survives simply because it would be odd for this reason to disappear entirely. The more natural view is that this reason survives but is dwarfed by opposing reasons in the final tally.

Still, upon further reflection, this objection raises problems. To add a perplexing wrinkle to the above analysis, consider the fact that, all other things equal, autonomous criminals seem worse than non-autonomous ones.Footnote ³⁹ By autonomously choosing to commit an immoral act, the perpetrator appears more worthy of condemnation than if he committed the same act non-autonomously. Thus it might be held that the murderer's autonomy actually makes him more contemptible, and this provides a further reason to thwart him rather than a reason to respect his preference. On this analysis, a preference's autonomy counts against its satisfaction if the preference is immoral.

Indeed, the matter is still more complicated since it is possible for an immoral preference's autonomy to provide both some reason to press for its satisfaction in light of its autonomy and another reason to thwart its satisfaction in light of the fact that autonomous wrongdoers are worse than non-autonomous wrongdoers, all else being equal. That is, an immoral preference's autonomy may provide two reasons – one to press for the preference's satisfaction and another to thwart it – perhaps with the latter usually outweighing the former.

We now have three views under consideration, none of which are obviously false. These are that an immoral preference's autonomy:

1. (1) provides some reason to help its bearer satisfy it, but this reason is typically outweighed by other considerations;

2. (2) provides an additional reason to thwart its satisfaction because autonomous wrongdoers are worse than non-autonomous wrongdoers;

3. (3) provides both a reason to help its bearer satisfy it and a reason to thwart its satisfaction, with the latter reason typically outweighing the former.

Here it is important to notice that the proponents of these views will agree far more often than they will disagree. On all three views, it follows that, for the most part, the state should not help autonomous murderers kill people. Whether there is some reason to do so, no reason at all, additional reason not to do so, or both a reason to help and an additional reason not to, the end result is more or less the same as far as the state's duties are concerned. Of course, there will be cases where the right analysis makes a difference, but rather than worry about these exceptional cases I suggest we turn our attention to a deeper and more worrying issue raised by these views. This concerns the strength of the autonomy-based reasons to press for a preference's satisfaction. On views (1) and (3), these autonomy-based reasons seem especially weak. On view (2), these reasons disappear entirely. One might suspect that if these reasons are typically outweighed or completely obliterated by competing considerations such as a preference's immorality, they do not warrant serious attention. I will argue that this impression is mistaken.

If view (2) is correct, then it follows that the autonomy-based reasons I am defending vanish in certain contexts. But this is not that embarrassing for my view as long as these reasons have weight in other contexts. Here it might help to recall the arguments in section V which show that these reasons have considerable weight when the welfare of others is not at stake, such as in cases where someone participates in a dangerous sport. Moreover, even if these reasons disappear in certain contexts, this would not show that these reasons are altogether unimportant. For consider. Someone who defends equality as a criterion of a just distribution of wealth might concede that the demands of equality have little or no weight in situations where this could only be achieved by bringing the more advantaged to the level of those whose lives are barely worth living,Footnote ⁴⁰ or in possible worlds inhabited only by the rich and the super rich. Similarly, someone who defends the difference principle might concede that it has little or no weight in situations of extreme scarcity or extreme plenty, but nevertheless maintain his commitment to it in other contexts.Footnote ⁴¹ Thus, even if the autonomy-bases reasons I am defending have little or no weight in certain contexts, this is not deeply problematic as long as they have weight in other contexts, as the arguments in section V show.

On views (1) and (3), the autonomy-based reasons to press for a preference's satisfaction appear to have little weight when the preference in question is immoral. Or so it seems if we focus on examples of the most grievous moral wrongs such as murder. But if we focus on ‘lesser’ kinds of immorality, the autonomy-based reasons do not appear so weak. To see this, consider again Jim's non-autonomous preference for reading detective fiction. Suppose I have an autonomous desire to steal Jim's detective novels. Also suppose that doing so would cause Jim no distress and that this act will have no other untoward consequences such as encouraging others to do the same. In short, I have an autonomous desire to commit a harmless wrong. In this case, my intuition is that the theft, despite its wrongfulness, may be permissible. I believe the same can be said for other ‘smaller’ wrongs such as lying, particularly when no harm is done to the victim. For instance, should my friend, who witnessed my theft, inform Jim or help me establish a false alibi? Surely the fact that my desire to steal is autonomous, and that Jim's preference for detective fiction is not, is relevant here. To see this, compare this theft with stealing Mary's paintbrushes. All else equal, I take it that stealing from Mary is considerably worse than stealing from Jim, and this is explained by the autonomy of Mary's painting preferences. Lying to Mary in order to protect a friend's false alibi also seems far worse than lying to Jim on the same topic.

If this is correct, then the autonomy-based reasons to press for a preference's satisfaction have weight even when the crucial preference is immoral. It might have little weight in the context of a grievous wrong like murder, but it has considerable weight in cases of ‘lesser’ wrongs – especially when these involve harmless acts. In principle, one could argue that these reasons have weight in the context of lesser wrongs like petty theft but not in the context for grievous wrongs like murder, but this strikes me as implausible. The more natural view is that if these reasons have weight in the context of lesser wrongs, they also have weight in the context of more serious wrongs – it is just that, in the latter case, these reasons pale in comparison to the opposing reasons.

Finally, let me add that because of the connection between promoting autonomy and respecting persons, it is natural to suppose that the reasons to press for an autonomous preference's satisfaction are quite strong and do not always lose out to other considerations. Indeed, since showing respect for a person's autonomous preferences is a crucial aspect of showing respect for him as a person, any state that fails to make special provisions for the autonomous preferences of its citizens is failing in a crucial aspect of its duty to respect its citizens as persons. This is no trifling matter.

X. SUMMARY AND CONCLUSION

I have argued that part of what's involved in the state treating its citizens with respect is helping them satisfy their autonomous preferences. I have argued that the state has positive prima facie duties toward the autonomous preferences of its citizens, and it has these duties in virtue of the autonomy of these preferences. To be sure, I have addressed neither the precise extent of the state's duties to promote the satisfaction of autonomous preferences nor the means the state should employ in discharging them. Nevertheless, I think that merely establishing their existence constitutes significant progress. After all, some have argued that the state's autonomy-based duties culminate in a neutrality constraint between competing conceptions of the good life. Others have argued that they extend to providing the conditions in which autonomy can flourish.Footnote ⁴² Notwithstanding the merits of these views, I have argued that these duties are considerably more extensive than anyone has recognized. They extend all the way to helping persons satisfy their autonomous preferences.Footnote ⁴³