I. A NEW JUSTIFICATION FOR EXEMPTING ADULTS FROM PATERNALISM

Appealing to a person's interests, and to the special knowledge that she has of them, is not a new way of justifying the impermissibility of paternalism. For instance, John Stuart Mill and Joel Feinberg both argue that a person's well-being is served by having her interests promoted, and that individuals are the best judge of their own interests. Mill rejects paternalistic intervention into the lives of adults on the grounds that, ‘with respect to his own feelings and circumstances, the most ordinary man or woman has means of knowledge immeasurably surpassing those that can be possessed by anyone else’,Footnote ⁴ while Feinberg writes that ‘a given normal adult is much more likely to know his own interests, talents, and natural dispositions . . . than is any other party, and much more capable therefore of directing his affairs to the end of his own good.’Footnote ⁵

Both Mill and Feinberg are right to note that individuals have privileged epistemic access vis-à-vis external observers to many internal facts about their own lives, and this should indeed go some way towards protecting persons from paternalism. After all, reasonable people disagree widely on the question of what has value, and what one person takes to be of the deepest value may be taken by another to be downright harmful. If we reject the more unsophisticated forms of objective list theories of well-being that would allow us easily to identify which party in the conflict is wrong, then epistemic humility should require us to recognize that we can rarely be certain that another person has misjudged her own interests. A person's privileged epistemic access to her own life should accordingly support a presumption in favour of treating each person as an authority on her own interests – and, accordingly, against paternalism.

The great majority of accounts of well-being, however, allow that we can be wrong about what our interests are. Privileged epistemic access to one's own interests lends support to the presumption against paternalism because persons obviously have an interest in not being forced to violate their commitments or other interests, and third parties who lack epistemic access to what those commitments and interests are may inadvertently encourage this kind of violation. But if persons can be wrong about their own interests, then they may themselves make choices that inadvertently violate those interests – and they surely have an interest against this kind of violation as well. And when the objects of commitments and other interests that a person chooses seem to most third parties to have obvious disvalue, there is less prima facie reason to think that paternalism would result in a violation of interests, rather than a protection of interests. Think here of the person committed to martyring herself for an unworthy cause, or the person committed to gender roles that deny her important freedoms. While Mill and Feinberg may be right that individuals generally have privileged epistemic access to their own good, and so should not have their lives paternalistically micromanaged, as long as we allow that persons can sometimes be wrong about their interests, there will seem to be cases in which enough third-party evidence compounds to create an isolated presumption in favour of paternalism.

To strengthen the presumption against paternalism in these more contentious cases, we will need to explain why even broad agreement by third parties is likely to provide poor evidence about the well-being of the person in question. I argue that we can provide such an explanation if we look closely at how and why commitments benefit their possessors. I follow Cheshire Calhoun's definition of a commitment as ‘both an intention to engage with something . . . and a preparedness to see to it that that intention to engage persists’, at least across a certain range of cases.Footnote ⁶ Commitments can benefit their possessors in two ways. First, one can have an interest in having some commitments rather than no commitments. Many take it to be uncontroversial that at least the great majority of adults have an interest in having some substantive commitments, since a life in which one cared deeply about nothing, or flitted constantly from goal to goal, may seem impoverished. Calhoun's article aims to show that this will not be the case in every single circumstance, but she does not deny that a great majority of persons rightly take commitments to play an indispensable role in their lives.

Second, and relatedly, one can have an interest in the objects of particular commitments in a way that goes beyond the interest that one would have in that object if one were not committed to it. While we can quibble about whether, say, great works of art or animal rights have more intrinsic value, it seems uncontroversial to say that a person with a commitment to animal rights will be more deeply invested in ending animal testing than in securing a rare sculpture for a local gallery, and that the opposite will hold for the person committed to the arts. And if we adopt any kind of informed desire approach to well-being,Footnote ⁷ having their respective commitments will give each person a stronger interest in their respective objects than the other has.

But to understand fully how commitments can benefit their possessors, we need an understanding of the way in which commitments are developed. At some point in life, each of us must recognize that we cannot have it all. We are creatures with finite lives and energies, and these facts prevent us from pursuing all of the options available to us. Instead, we unavoidably ‘settle’ for some subset of them. But while this may at first seem unfortunate, it should not: Robert Goodin, in praise of settling, writes that ‘fixed points in our lives are what provide the structure that we need to plan other aspects of our lives. Settling some things – treating them as fixed, at least for a time – is for that reason important for human agents to be able to form and pursue plans for their lives.’Footnote ⁸

Three related elements of Goodin's treatment of settling are particularly important for our purposes. First, settling allows us to develop the commitments that constitute our identities and allow us to have self-respect.Footnote ⁹ It is the way in which commitments contribute to identity and self-respect that explains why we have an interest in having commitments simpliciter. From among the vast array of experiences available to me, I have the opportunity to choose a set that will become distinctively my own, and that will allow me to make sense of and esteem myself as a unique individual with a coherent identity. In order to respect myself, I must have a sense of self that sets me apart from others: settling on commitments is important for self-respect because, as Goodin notes, ‘having fairly firmly fixed commitments to some principles, values, and ongoing projects is bound up with what most people ordinarily think of their “self” as involving.’Footnote ¹⁰ And the fixity of such identity-conferring commitments is important: if we did not settle on commitments – that is, if we were just as happy for our engagement with them to end at any moment – then they would neither be commitments at all on Calhoun's definition, nor contribute to a sense of self that endured in any way.Footnote ¹¹ I call this type of settling ‘principled settling’.

Second, and relatedly, principled settling also gives us something for which to strive.Footnote ¹² While having a sense of self requires knowing who I am now, our identities are not static. Actually respecting ourselves requires taking charge of ourselves, achieving things, and growing as persons, and the commitments on which we settle also help to direct this growth and attempted achievement. The fact that achievement and growth take time explains as well why we must settle on our commitments in order for them to play this role: commitments inform our goals, and a goal that constantly changes is no goal at all. This need for striving explains why the objects of our commitments contribute more to our well-being than many other objects: the objects of our commitments provide goals for the projects that allow us to take ownership of our coherent identities, and when we advance those objects, we either reaffirm or redesign those identities.

Third, and once again relatedly, settling in more mundane ways is also necessary for striving. While we need goals for which to strive, we also need the resources to do so, and this will often require settling in the rest of our lives.Footnote ¹³ I call these instances of ‘pragmatic settling’. Food, shelter and other basic necessities are preconditions for meeting virtually any goal, and many people accordingly settle on jobs that they do not love or live in affordable locations that may not excite them in order to have the resources to focus on their other goals. And since time and attention are necessary for achieving any goal, persons must always triage their commitments and preferences, putting some aside, and taking others as given, in order to focus on those they choose to pursue. Indeed, pragmatic settling may even become an end in itself, since we are all likely to absorb the standards of our communities, and many communities place a high premium on supporting oneself and avoiding placing burdens on others.

We have established, then, why it is important for persons to have commitments. We have not yet, however, established which commitments it is in a person's interest to have. As we saw above, some commitments seem clearly harmful. I will now argue that given our interests in settling, even seemingly harmful commitments can come to be in a person's interests.

Consider the analogous case of friendship. It is uncontroversial both that we all have an interest in having friends, and that what makes a friend a good friend for me depends not only on the features of a friendship that generally count as good-making (loyalty, honesty, generosity, etc.), but also on features particular to my own life (a person interested in poetry may count as a good friend for me but not for you, if I, unlike you, am interested in poetry). The more interesting point, however, relates to features that are generally understood to be bad-making (for instance, intolerance, impatience, rudeness). If I am choosing a friend from an array of strangers, these sorts of bad-making features are likely to rightly count against a person – but having already befriended some person who has those features, they may actually come to contribute to the friendship's being good for me. I can come to love my friend sincerely for features that I previously saw as shortcomings, and I could genuinely regret it if such a character trait were lost at some later time. After all, this character trait is just part of who my friend is.

Interests in commitment are the same. If I begin by looking at a field of commitment-objects, there will be many that I will prefer because of good-making features that many others would agree upon, and others that I will prefer for more idiosyncratic, but still readily understandable, reasons. But if I turn to the commitments that I already have, then I might see as good-making some features that I would not have seen as good-making if I were choosing them from a set to which I had no previous attachment. Indeed, I might see as good-making features that I previously saw as down right bad-making, and that others continue to see in this way. Once again, these commitments are now just a part of who I am, and it is often perfectly reasonable not to regret who one is – even in cases where one could, counterfactually, have been better.

Though Goodin does not address this point, we can nevertheless justify my somewhat counter-intuitive claim by appealing to the strong interest that adults have in settling on commitments. When Goodin defends the concept of settling, his aim is to show that settling is not a regrettable necessity, but is rather a central part of what allows all of our lives to be good. Accordingly, he is swift to show that both compromise (which involves permanently forsaking some of one's goals in order to achieve others)Footnote ¹⁴ and resignation (which involves permanently forsaking some of one's goals because they are ultimately unachievable)Footnote ¹⁵ are unfortunate subsets, rather than the whole, of settling. Most settling, unlike compromise and resignation, involves treating some open options as if they were fixed for some limited period of time, and leaves open the possibility of returning to and reconsidering those options in the future. Accordingly, for Goodin, settling should not generally require settling for what is bad.

Yet, while Goodin is right to note that instances of resignation and compromise are often undesirable, we must also note that they may nevertheless sometimes be necessary – sometimes the only way of achieving one of one's dearest goals really is (or really is for the moment, or rightly seems to be, from the available evidence) to permanently forsake another one, and sometimes one must indeed become resigned to unfortunate and even unjust situations that one genuinely cannot change (or cannot reasonably foresee changing, for a long time or ever, even with significant work). But those who find themselves in such unfortunate positions have no less need than others of the benefits that settling provides: like others, they need to adopt a set of commitments that provide them with a coherent identity; they also need goals for which to strive; and, like others, they also need to free up the resources with which to do so.

Settling under these circumstances is clearly not the ideal way of developing commitments. Indeed, one might think that compromising or becoming resigned is not genuinely adopting a commitment in Calhoun's sense at all, since neither seems to involve a real intention to engage with an object – much less a preparedness to see that the former intention persists. Perhaps this is strictly correct. Perhaps persons forced to (or who simply take themselves to be forced to) compromise or become resigned only ambivalently adopt the objects of commitments in a way that might better be described as simply living in accordance with a commitment out of necessity. But since persons in such situations still need a sense of self and goals for which to strive, they might well find, when later and unexpectedly given the opportunity to change their minds, that they have lived in accordance with even a reluctantly adopted commitment object for long enough that it has become a part of who they are – that what once sprang from an unwelcome compromise now plays an important role in shaping the coherent identity of the person they have become. Indeed, we often find that settling happened behind our backs while our attention was elsewhere, and that we are now persons we did not realize that we had become.

All of this affects our interests. In a discussion of how to make prospective judgements about what one ought to do, Elizabeth Harman cautions us to remember that ‘as identity changes, what it is permissible to desire changes, too.’Footnote ¹⁶ Harman makes this point in order to argue that, prospectively, we have reason to focus on agent-neutral values as we decide who to become in the future – but the converse of this point is that we can never make any decisions without simultaneously considering, retrospectively, who we already are. And what is best for me as I actually am may not be what is agent-neutrally best for persons. But once again, this should not be regretted: even if what is best for me is not agent-neutrally best, and even if what is now in fact best for me would not be counterfactually best, the great interest that adults have in having settled identities and enacting plans in the world should prevent us from seeing our actual adult interests as lamentable.

How does this affect the permissibility of paternalism? So far I have argued that persons have special epistemic access to their own interests, and that in adults, there is legitimacy to those interests that they have developed from settling upon a commitment – or at least living in accordance with it for a long time – even when the commitment-object in question would have been bad for them both counterfactually and before they developed that commitment. When we add the additional widely accepted premise that there are stringent prohibitions against actively doing harm to persons, we have a strong case against paternalistic intervention into the lives of adults: intervening in a person's life in a way that directly violates her interests harms her, and third parties necessarily lack the epistemic access that would allow them to determine whether a person's unlikely or unseemly preferences have reached the level of commitment so important to her interests.Footnote ¹⁷

The case against paternalism becomes even stronger when we turn again to the temporal dimension of settling. I have also argued that given our nature and the world in which we live, adults – including those forced to compromise or become resigned – must almost always engage in both principled and pragmatic settling in order to achieve the kind of coherent identity that allows them to make sense of themselves and secure a sense of self-respect across the course of their lives. Accordingly, the danger of paternalistic intervention is not simply that a third party may misjudge a person's interests in isolated instances and so prevent her from furthering them in the moment. Rather, paternalistic intervention runs the risk of disrupting the whole course of her life, and of robbing her of both the sense and the tools of agency that leave her able to esteem and make sense of herself. While third parties may be well placed to say which commitments most persons would counterfactually be better off having, they cannot say which of a particular person's commitments are most central to the identity that she in fact has, and which necessary compromises she relies upon to pursue those commitments. Since external actors necessarily lack this internal epistemic access, and since the interest involved is so important, the stringent prohibition on harming makes paternalistic action even more clearly impermissible.

Note, however, that this justification does not render it always impermissible to intervene in an adult's life for her own good.Footnote ¹⁸ These considerations merely place limits on the methods of intervention that are permissible. Since adults may be wrong about their interests, they may have commitments which, upon reflection, they would rather attempt to revise. Accordingly, it is permissible to appeal to their sense of reason, or to make information about alternatives available to them. And if they decide to attempt to change their commitments, it is certainly permissible to provide them with resources or to create the structures within which they can most easily do so. The justification, however, limits the permissibility of manipulating them or coercively encouraging them to adopt new behaviours, since doing so risks harming them in full light of the evidence offered by their own privileged testimony that their interests lie elsewhere.

One final note: although I take no stand on whether all types of paternalism are impermissible, my justification may rule out more cases than it at first appears to. In particular, while it clearly rules out many of the more value-laden interventions that a perfectionist state might undertake, it might not obviously rule out the interventions more frequently associated with paternalism, which are aimed at trivial behaviours like smoking and drinking soda. In response, recall first that persons must also engage in pragmatic settling. Even if smoking is not itself related to a commitment, it may well be a necessary means of stress-relief which a person has pragmatically settled upon in order to allow her to reach her other goals. Second, consider that behaviours may be either directly or indirectly related to commitments: a person's soda-drinking may be commitment-relevant not because she is committed to soda itself, but because she has libertarian political commitments that cause her to reject all government intervention into her life. In both cases, more common paternalistic interventions violate her commitments, and the justification therefore counts significantly against paternalistic intervention.

That said, the justification also suggests that some kinds of paternalistic intervention might be more permissible than others. For instance, close friends with intimate knowledge of one another's commitments might be more justified in intervening than the state. This would require more argument, but I do not take it to be obviously implausible, and so do not count it as a weakness of my account that it might allow this.

I turn now to three additional benefits of my account.

II. BENEFIT ONE: PRESERVES THE DISTINCTION BETWEEN ADULTS AND CHILDREN

As I noted in the introduction, the standard justification for exempting persons from paternalistic treatment appeals to the connection between the capacity of autonomy and autonomy as a right. As I also noted there, whether the descriptive claim holds will depend entirely upon the features of the account of the capacity of autonomy that one uses, and different accounts of the capacity of autonomy will accordingly take very different sets of persons to be owed autonomy as a right. This becomes especially problematic if one wants to justify the standard position that adults are owed autonomy as a right while children are not.Footnote ¹⁹ The problem is this: if one uses a very demanding account of autonomy,Footnote ²⁰ then one succeeds in ensuring that most children cannot meet the bar for the capacity of autonomy that would ensure them autonomy as a right – but in doing so, one also sets the bar too high for many adults to clear it; if one adopts a weaker account of autonomy, then one includes most so-called ‘normal’ adults, but does so at the expense of also including many (especially older) children. In either case, appealing to the capacity of autonomy fails to draw a clear descriptive line between children and adults that makes children liable to paternalism and adults exempt from it.

My well-being based account does a much better job of drawing this descriptive line between children and adults. Recall that, on my account, adults are exempt from paternalism on the grounds that settling on commitments makes some surprising objects of commitments contribute to their well-being, and that adults have a strong interest in settling on commitments. These reasons, I argue, do not apply to children.

We can begin with the descriptive claim that children are less likely than adults to have settled on commitments. While children, like adults, may have unique epistemic access to the things that they care about, unlike adults, many children cannot meaningfully be said to have commitments. Below a certain age, they may have likes and dislikes, but they cannot be said to have a genuine intention to engage with objects, let alone an intention for that engagement to persist. And even older children's genuine intentions, once they can be meaningfully called as such, change frequently. Given these facts, in conjunction with the short duration of their lives, children are less likely than adults to live with commitments for long enough that they shape the course of the children's lives and accordingly give rise to strong interests in their objects. When this is descriptively the case, there is significantly less reason to privilege children's epistemic access to their own interests over the access of third parties than there is in the case of adults. Since they have not had a chance to incorporate into their identities the kinds of commitments that most third parties would take to be harmful, those third parties will be able to say with much more accuracy when a child is likely to be wrong about her own best interests. And if children's interests are far more transparent than adults’, requiring them to engage in behaviours and functionings that seem generally to be valuable runs far less risk of harming them than it does in the case of adults whose interests are altered by long histories with their commitments.

But of course, this descriptive claim might not always hold. Older children, especially, may settle on commitments, and even children too young meaningfully to form their own commitments may live in accordance with the commitments of their parents and community members for long enough that they become incorporated into their identities. Accordingly, I turn to offering normative reasons for preventing children from settling on commitments during their childhood. In this section, I’ll offer a negative reason; in the next I’ll offer a positive one.

In the previous section, I showed that adults were harmed by being prevented from settling on commitments. In order to show that it is permissible to use paternalistic action to prevent children from settling on (at least some) commitments, I must first show that doing so will not similarly harm them. Happily, the same features that explained why adults were harmed by being prevented from settling explain why children are not. I take each in turn.

First, unlike adults, children do not generally require a set of commitments in order to have an identity that secures for them stability and self-respect. We need only think back to our own childhoods to be reminded that the goals and commitments of children change frequently and radically. Despite this, however, most children retain many fixed points around which to marshal their identities: home towns, parents and other family members, schools, and beloved pets all play major roles in the lives of children, which provide them with continuity and sense of self even in light of constantly changing commitments and interests. Of course, children may feel deeply about many issues in ways that may sometimes persist for long periods of time, and such feelings are likely to increase as children get older. But even in older children, such fixed commitments are not necessary for self-respect in the way that they are for adults. While older children may experience distress about not knowing what they want to be when they grow up, adults rightly reassure them that this is normal and even positive, in a way that they would not with other adults. Conversely, even those adults who can be happy without knowing what they want to do with their lives, or with flitting endlessly from project to project, will be treated with some distrust or pity by many of their fellow adults. And while the respect of those around us may not always be sufficient for self-respect, many of us experience it as necessary.

In part, this first dissimilarity can be explained by the second and third dissimilarities between adults and children. The second difference relates to the necessity of settling on goals for which to strive. No one can fully achieve a goal until they have settled on it, but for children, many available means of striving-in-aid-of-achieving are ‘multi-purpose’: finishing school, becoming more thoughtful and better-read people, and further developing a wide range of social and practical skills count as the first steps at striving towards an overwhelming number of possible goals. As time passes, this ceases to be the case: as one narrows down their set of goals, they must adopt increasingly finely tuned methods of striving for them, and adults who fail to do this fail to engage in striving. Children, who can engage in multi-purpose striving, need not similarly narrow their methods of doing so.

Finally, children also do not need to settle in order to ‘free up’ resources in the same way that adults do. Adults must generally provide the necessities of life for themselves, and this generally makes it necessary for a given adult to spend much of her time in consistent paid work, whether she wants to or not. If she does not, then she not only lacks ambition: she is also a drain on others. And taking on consistent paid work, of course, requires closing off many other options. Children generally do not have this responsibility, since their parents standardly provide for them. And while they do need to attend school (whether they like it or not), they need not close off their options in the same way: they can explore more – and more varied – courses of action and possible commitments over the course of a school day than the average adult can over the course of a work day.

Children, then, are not harmed if they are prevented from settling on commitments while they are young enough to have financial support, to be engaged in basic forms of multi-purpose striving, and to retain a sense of self without a sustained engagement in projects. If they do not settle on commitments, then their interests remain relatively transparent. And if their interests are transparent, then requiring them to engage in or refrain from certain behaviours runs far less risk of harming them than it does in the case of adults whose interests are altered by long histories with their commitments. When it comes to the permissibility of paternalism, then, we have what we previously lacked: a justification for drawing a bright line between children and adults.

I should note, however, that the justification for the bright line holds most clearly in wealthy, Western-style democracies. In particular, in many parts of the world (including in some sub-sections of wealthy, Western-style democracies), children may not be able to rely on their parents to support them, and may instead need to engage in pragmatic settling earlier in order to support themselves and their siblings. Where children are forced to settle earlier for this reason, my proposed justification may not apply. However, where those resources can be provided, I assert without argument that they ought to be – and in so far as they are, the justification will apply. Since my account is based in well-being, and since features such as these can so significantly affect one's well-being, I do not take it as a limit of my account that its implications will change when empirical well-being-relevant facts change systematically. Further, the point up to which multi-purpose forms of striving are available and effective will vary according to institutional arrangement, and this will affect the appropriate age at which to stop considering individuals as children. I do not take there to be anything special about the age of 18, and think that childhood could be defensibly understood to end at a variety of different ages under different institutional arrangements. At what age the bright line between children and adults should be drawn must depend on this fact, and may vary from society to society.Footnote ²¹

III. BENEFIT TWO: EXPLAINS THE VALUE OF THE CAPACITY, AND CLARIFIES THE RELATIONSHIP BETWEEN CAPACITY AND RIGHT

But showing that paternalism is permissible in the case of children does not yet show that we ought to use it. To show this, I need to reintroduce the capacity for autonomy. And in doing so, we find the second benefit of my account: it explains the close association between the capacity of autonomy and autonomy as a right, and it provides a resuscitated justification for seeing the capacity of autonomy as valuable. Given the problems discussed in the introduction with using the capacity directly to justify the right, I will argue only for an indirect relationship between the two: rather than claiming that possession of the capacity is necessary for the right, I will argue that developing a strong version of the capacity at the appropriate time renders one better able to take advantage of the full benefits that can come from exercise of the right. The value of the capacity, in turn, provides a positive reason for using some forms of paternalism on children: namely, those that help them to develop the capacity.

To see how this is so, we should make explicit the difference between autonomy and settling. While we can negatively appraise either a person's capacity of autonomy or her settledness, we mean very different things by these appraisals. When we say that someone's capacity of autonomy is poor, we mean that she lacks that capacity – that is, she is not really autonomous at all. When we say that she has settled poorly, however, we are not denying that she is in fact settled – rather, we are denying that the things upon which she has settled are good for her. As persons’ opportunities for multi-purpose striving shrink, they cannot really help but settle into some course of action and sense of character. And while, as I argued above, adults have an interest in settling whether or not they do it well, they clearly have a stronger interest in initially settling on objects that are better for them than on those that are worse.

This difference explains the connection between autonomy and settling: the capacity of autonomy is valuable because it plays a unique role in ensuring that the interests that we take ourselves subjectively to have (and so move towards settling on) actually track our objective interests.Footnote ²² There are of course many competing accounts of which interests should count as objective,Footnote ²³ but I take to be least controversial those informed desire accounts according to which a person's objective interests are the ones that she would take herself to have if she possessed all relevant knowledge and reasoned perfectly. Her subjective interests, on the other hand, are the interests that a person takes herself to have in her actual life where knowledge is never full and reasoning is never perfect.Footnote ²⁴ Clearly, subjective and objective interests can come apart. Indeed, this is what we meant above when we said that a person could be ‘wrong’ about her interests: the subjective interests that she takes herself to have fail to track her objective interests. Since non-autonomous preferences are, as we saw in the introduction, foisted on a person by her environment, they are unlikely to track closely her objective interests. If the environment is bad, her preferences are likely to be bad as well, and so to differ significantly from her objective interests; even if it is good, the preferences it imposes are likely to be generically good, and will be insufficiently specific to track any particular person's objective interests. To see how autonomy can help subjective interests to track objective interests, let's consider a particular, relatively demanding account of autonomy.Footnote ²⁵

Theorists concerned with children's freedom and autonomy turn frequently to the 1972 U.S. Supreme Court case Wisconsin v. Yoder.Footnote ²⁶ In this case, the Court decided in favour of a group of Old Order Amish who challenged a Wisconsin state law that made education to age 16 compulsory, claiming that it violated their right to religious freedom. Most theorists writing on children hold that this case was decided wrongly, often on the grounds that it violated the children's right of exit from their community. Across the literature, two mechanisms are consistently identified as necessary for the child's meaningful right of exit,Footnote ²⁷ and I take these mechanisms to be jointly sufficient for a strong account of the capacity of autonomy. They capture two of the requirements that comprise other standard accounts of autonomy,Footnote ²⁸ and they help to illustrate the importance of a strong account of autonomy in the real world where people's choices can be so easily and pervasively influenced by their social circumstances.

The first mechanism required for autonomy is access to a set of alternative options. If children are unaware of other lifestyles and systems of values, it will be very difficult for them to consider these alternatives as possibilities that they might want to pursue. As Rob Reich notes, when a person has been ‘rigorously secluded and shielded from value diversity, his or her horizon of choice is severely limited . . . it is impossible to pursue an end or endorse a value that one does not know exists.’Footnote ²⁹ If children do not know what the options in the world are, then they will be in no position to choose for themselves which options are most valuable. And if children are not even aware of the options that correspond to their objective interests, then the subjective interests that they take themselves to have cannot possibly track them, even by accident.

But how many – and indeed, which – options does autonomy require? After all, a child's option set cannot be fully closed in even the most extreme circumstances. Even secluded Amish communities allow their youths to explore the outside world in a ritual called ‘rumspringa’ before they choose to become adult members of that community.Footnote ³⁰ But a child's option set cannot be fully open, either. Every child is born into a social circumstance, and every concrete circumstance will make some options more immediately available than others. In Claudia Mills's words, it is simply impossible to raise children ‘without steering them, however imperceptibly, toward one option rather than another’.Footnote ³¹ Given that many of a child's early activities will be chosen by his or her parents, the path dependence of most lives means that activities chosen by someone else in childhood are likely to be the ones that persons continue to engage in and identify with.

If a child's set of options can be neither fully closed nor fully open, then choosing autonomously will require possessing the tools to decide between the unavoidably limited set of options that she has, and to recognize the possibility that she might need to seek out others. Without these tools a child's chance to direct her own life would be more formal than substantive: if I am given the formal right to leave the Amish community in which I have been raised, I will only have any real chance of taking advantage of this opportunity if I am able to weigh the value of staying against the value of other alternatives with which I have little familiarity. When so many factors weigh in favour of staying in an insular community, the genuine possibility of leaving requires significant independence of mind.Footnote ³² This, then, is the second important mechanism of autonomy: the possession of the independence of mind that allows a person to interrogate critically the options that are actually before her and to seek out the alternatives less obviously available.Footnote ³³

To show that independence of mind and a variety of options can help subjective interests to track objective interests, we need only one further, and uncontroversial, premise: that every individual will have a different set of objective interests. While we may all share many broad interests, such as life and health, each of us will also have many more specific interests dictated by (among other things) our unique abilities, circumstances and values. An independent mind and a broad variety of options, in addition to epistemic access to the people who we already are, will put us in a position that better approximates the situation of full knowledge and perfect reasoning that allows us to identify our unique set of objective interests. That is, armed with these tools, the interests that we take ourselves to have will better approximate the ones that we actually have.

The ability of the capacity of autonomy to align our subjective and objective interests also provides us with a positive reason for treating children in at least some paternalistic ways. Although children develop autonomy at different points (and some may never develop it), there is a certain point before which all children necessarily lack autonomy. Before a certain age, they can simply have no independence of mind, and cannot meaningfully grasp the options open to them – let alone seek out others. At such a young age, however, they can nevertheless live in accordance with the commitments of others around them for long enough that they start to become settled in, and so develop interests in those commitments – and if the commitments that they live in accordance with are bad ones, then they would counterfactually be better off not to have settled on them. For adults, this was not to be lamented, since they were harmed by being prevented from settling; but since children are not similarly harmed by such prevention, it makes sense to encourage them to avoid forming commitments until they have had the chance to develop the autonomy that will allow them to commit themselves to objects that track the best set of objective interests that they might have. While they will almost surely eventually settle either way, possession of the capacity of autonomy will allow them to settle well.

Several methods of paternalistic action will facilitate development of these sorts of commitments. First, we can prevent them from actions that seem clearly to make many likely valuable future courses of action inaccessible. Such actions may include prohibitions on using cigarettes, alcohol or other drugs that will affect their future health or the ways in which their young bodies develop, as well as prohibitions on sexual relationships with adults that are likely to lead to lasting emotional distress. Second, we can require children to engage in activities that will help to strengthen their independence of mind and keep many future courses of action open to them. Such actions may include requiring them to take a broad set of subjects in school, so that they learn to look at problems in different ways, and may choose to study any one of them further. Third, we can expose children to a wide variety of values, ideas and practices from a young age, so that they still retain a broad set of live options from which to choose when and if they develop the autonomy that will allow them to choose well between those options. Such actions might include taking them to live in or visit different parts of the world, bringing them to different kinds of religious services, or enrolling them in extracurricular activities unlike the ones they currently pursue.

Note, however, that none of this is to say that we should discourage children from developing strong interests. It is merely to say that we should actively encourage them to see their options as open, and to develop a wide enough set of interests that they become well-rounded people exposed to many different ideas. This proposal will certainly count against, for instance, encouraging a child to devote herself exclusively to study of the violin – but I take this to be an acceptable price to pay for giving that child the best chance she can have to direct the course of her own life.

Of course, some children will develop enough autonomy to choose for themselves well before they reach the age of 18. While it might seem unfair to continue to subject them to paternalistic intervention at this point, such subjection nevertheless will not harm them, for the reasons discussed in the last section. Conversely, we must also recognize that, while the point of this proposal is to keep children's options open until they can develop the autonomy to choose for themselves, some children will simply never develop this strong type of autonomy. This is perhaps unfortunate, but children who do not will at least have had the best opportunity that they could to do so, and will have been provided with a set of practical skills that will open many doors for them – as well as having been protected from harmful courses of action that might have closed many doors.

One might object that, if children receive such benefits from developing the capacity of autonomy, then adults who have failed to develop the capacity will also benefit from being pushed to do so, and that this will justify some paternalism towards them. I find it plausible that it is permissible to use the rational means discussed in section I to suggest to adults that they develop greater independence of mind, or consider a wider variety of possibilities. I doubt, however, that it will be permissible to use any coercive or manipulative means to do so. Some of the commitments which adults might adopt are commitments which specifically prohibit serious consideration of alternatives, as in the case of a commitment to a marriage partner, or perhaps, to a certain religion. And even if these commitments are ones which persons might, counterfactually, be better off not having, they are also the sorts of commitments that get at the heart of who many of us are as individuals.Footnote ³⁴ So while adults ought to be free to decide for themselves to leave their partners or reconsider their faith or countenance other radically new life options, they must not be pushed into doing so from the outside. The strong prohibition against harming makes it impermissible to take such a reckless chance on someone's interests for her own good.

Possession of the capacity of autonomy, then, cannot do the work of drawing a bright line between children and adults when it comes to assigning one group, but not the other, autonomy as a right. Those theorists who draw a close connection between the capacity and the right, however, are correct to do so in at least one sense: the capacity makes the right much more valuable, and helping those who are not yet owed the right to develop the capacity will provide extra reason for granting them the right when they reach the point at which they have a strong interest in settling on commitments. Note, however, that I am only claiming that the capacity is instrumentally valuable. This is of course compatible with holding that the capacity is also intrinsically valuable, but I remain uncommitted on this point. Doing so allows my proposal to be attractive both to perfectionist liberals who take autonomy to be intrinsically valuable, and to political liberals who deny this claim for political purposes (although they leave the question open for each individual).Footnote ³⁵ Since I take ecumenicism to be a desirable trait of any justification of paternalism, I take my ability to remain uncommitted on the question of the intrinsic value of autonomy to be a strength of my own account.

IV. BENEFIT THREE: GUARDS AGAINST PATERNALISM EVEN IF THE CAPACITY OF AUTONOMY IS IMPLAUSIBLE

While political liberals treat the value of the capacity of autonomy as an open question that ought to be answered by individuals rather than states, other critics more stringently deny both the plausibility and desirability of the capacity. I will only briefly introduce these criticisms here, and will not aim to either defend or evaluate them. If they succeed, then they raise further problems for the ‘standard’ link between the capacity and right of autonomy, and my account has an additional benefit; if they fail, the benefits previously discussed remain.

First, one might be sceptical about the plausibility of the capacity of autonomy. Some literatures in psychology provide reasons to think that decisions and judgements about values are generally not the result of a reflective and deliberative self, but are instead primarily driven by intuitions which are heavily socially and culturally influenced.Footnote ³⁶ This literature suggests that rational deliberation and justification, of the kind often associated with the capacity of autonomy, tends to be post hoc and to aim at defending those socially influenced intuitions rather than genuinely critically evaluating them. If this is so, then at least one standard understanding of the capacity of autonomy seems descriptively implausible.

Second, one might be sceptical about the desirability of the capacity of autonomy, at least as it is traditionally construed. Feminists have often criticized the view of the self that holds that ‘logically, if not empirically, human individuals could exist outside a social context.’Footnote ³⁷ Holding that ‘[persons’] essential characteristics, their needs and interests, their capacities and desires, are given independently of their social context and are not created or even fundamentally altered by that social context’Footnote ³⁸ denies both that elements of our identities are and should be socially constituted and that vulnerabilities and personal relationships do and should place obligations and limits upon us. Even if the capacity could be plausibly exercised, such failures might cast doubt on the desirability of developing such a capacity.Footnote ³⁹

If the right of autonomy is only appropriately conferred when the capacity for autonomy is present, and these theorists are right about the plausibility and/or desirability of the capacity of autonomy, then political philosophy will find itself in a dire predicament: there would no longer be a reason for either the state or individuals to refrain from paternalistic action towards citizens. But happily, if these theorists are correct, section I of my article still provides a reason to refrain from paternalism towards adult citizens. (Section II would also make paternalism towards children permissible, although given that section III would fail, the desirability of paternalism towards children might come into question.)

I do not claim that my proposal is unique in providing this benefit. One could, for instance, argue against paternalism on the grounds of equal respect for citizens.Footnote ⁴⁰ But an equal respect justification only claims that citizens have an interest in being treated as if they were equally well equipped to direct the course of their own lives. It does not provide any reason to think that this is in fact an accurate statement to make about persons, and since it does not, their interest in being treated equally may well come into conflict with their interest in living a life that is otherwise good for them. My account removes this tension, by explaining how even those who seem to third parties to be living lives that are bad for them are nevertheless the best judges of how their lives should go – and ought, accordingly, be allowed to act on their own judgements. My justification for exemption from paternalism, then, works in tandem with equal respect for persons, rather than counting against it.