SAILING ALONE: TEENAGE AUTONOMY AND REGIMES OF CHILDHOOD

. Should society intervene to prevent the risky behavior of precocious teenagers even if it would be impermissible to intervene with adults who engage in the same risky behavior? The problem is well illustrated by the legal case of the 13-year-old Dutch girl Laura Dekker, who set out in 2009 to become the youngest person ever to sail around the world alone, succeeding in January 2012. In this paper we use her case as a point of entry for discussing the fundamental question of how to demarcate childhood from adulthood. After summarizing the case, we identify a ‘demarcation dilemma’ that frames much of the public and expert debate. On the one hand, it seems morally imperative ‘to treat like alike’, which means that both children and adults should be allowed to undertake all actions for which they have the relevant competences. On the other hand, requiring proportional treatment of children and adults seems to neglect the special nature of childhood as a distinct stage in life that ends at a speciﬁc age. We introduce the notion of a ‘regime of childhood’ to deal with this problem. This regime includes several dimensions, including the limited liability for children, the supervisory responsibilities of parents, the role of age-based thresholds, and the overarching purpose of childhood as a context for developing autonomy. We argue that, all things considered, there are good reasons not to shift to a regime that offers individual children the option of qualifying for adulthood on the basis of age-neutral criteria.

fundamental questions about the nature of autonomy and the grounds for treating children differently from adults. In particular, should society restrict skilled and experienced teenagers' risky behavior when it could not permissibly restrict the same behavior by less competent adults?
As our point of entry for discussing these problems, we bring in a case that arguably falls into this category. In 2009, Laura Dekker (age thirteen) announced -with her father's support -her plans to sail around the world alone. The journey would take approximately 2 years. If successful, she would become the youngest person ever to sail around the world. She completed her quest on January 21, 2012. 1 Her journey was preceded, however, by a lengthy legal battle that began when the Dutch Council for Child Protection asked the courts to take custody of Dekker away from her father (with whom she was living) so as to prevent her from departing. The legal proceedings were widely reported, and the case generated heated debates in the Dutch media. Dekker's plans brought to the surface -and indeed came to symbolize -widespread disagreements about the appropriate norms for the upbringing of children. Those arguing in favor of Dekker's freedom to do something adventurous and inspiring were staunchly opposed by those arguing that this was an irresponsibly risky undertaking that any sensible parent (and in the last instance, the state) should prohibit. Intuitions about Dekker's case diverged widely, among experts as well as the wider public.
Our aim in this paper is not to adjudicate whether Dekker should have been allowed to sail or not. Rather, we will use the case as a stepping stone to assess the different normative strategies that could be used to get a grip on practical questions like these and to show how these strategies reveal different ways of conceiving childhood. First we introduce the case of Dekker in some detail, presenting the main features of the courts' treatment of her case (Sect. II). We then formulate the central moral dilemma of paternalistic interference in children's lives. On the one hand, it seems to be morally imperative 'to treat like alike', which means that both children and adults should be allowed to undertake all actions for which they have the relevant competences (this is the 'competence-based position'). On the other hand, such an imperative of proportional treatment of children and adults seems to neglect the special nature of childhood as a distinct stage in life that ends at a specific age (this is the 'age-based position'). Our main research question in this paper is how to escape this 'demarcation dilemma': how should decision-making authority over children's actions be allocated (Sect. III)?
In the remainder of the paper we argue that both horns of this dilemma -competence-based and age-based -should be avoided, or rather integrated on a higher level, in the context of justifying a specific 'regime of childhood'. We begin by discussing Tamar Schapiro's Kantian argument about why childhood should be seen as a separate stage in a child's development, a stage that requires the room for experimentation that can be provided only by not holding children fully responsible for their actions. While Schapiro's argument is important, it still is vulnerable to (a reformulated version of) the demarcation dilemma (Sect. IV). Therefore we propose a broader framework within which to understand demarcation disputes. We suggest possible ways to arrange and institutionalize what we call a 'regime of childhood'. The elaboration and justification of a regime involves demonstrating the relevance and interconnection of several aspects, especially the purpose of children's autonomy development, the limited liability for children, parental supervisory responsibility, and age-based demarcation (Sect. V). We argue that, all things considered, there are good reasons not to shift to a regime with an age-neutral basis for lifting parental responsibility and child tutelage (Sect. VI). Finally, we will comment on Dekker's case in the light of this broader framework (Sect. VII).

II. THE CONTROVERSY OVER DEKKER'S JOURNEY
In all of the court proceedings, Dekker and her father were opposed by the Council for Child Protection of the Netherlands. Child neglect cases often involve courts having to decide whether to support the point of view of the parent or the child. Here however both are SAILING ALONE: TEENAGE AUTONOMY AND REGIMES OF CHILDHOOD united in a struggle against the state's agencies. 2 The relevant legal norm is that the parent(s) should not show 'grave neglect' in their care of their children. The main question for the courts was whether Dekker's father had shown such neglect by not prohibiting her from carrying out her plans to sail around the world. In ruling on the father's duties, however, the court had to consider whether the plan itself posed such a threat to Dekker's well-being that paternalist intervention was necessary. Between August 2009 and July 2010 three different courts gave six verdicts on Dekker's case. 3 The first five placed and kept Dekker under temporary surveillance, so that she was not allowed to start her journey. The last verdict, in June 2010, lifted these restrictions, so that she was allowed to go. She began her solo journey from Gibraltar on 21 August 2010, sailing first to the Dutch Antilles, to which she returned, after 27,000 nautical miles, at 16 1/3 years of age.
The case started with the Council for Child Protection demanding permanent surveillance, both because it believed 13-year-old children in general do not have the so-called 'coping capacities' (dealing with heavy weather, loneliness, lack of sleep) required for such a journey and because they have a series of 'developmental tasks' which would come under serious threat. The court, however, was dissatisfied with a general description of the coping capacities and developmental tasks and abilities of 13-year-olds. It granted temporary surveillance, so that experts could investigate to what extent Dekker herself had these capacities and had satisfactorily prepared for this journey. An investigation of Dekker's specific abilities would determine her right to sail. The courts split these issues into two sets of requirements.
On the one hand, the courts were worried about Dekker's cognitive development, her safety, and her coping capacities. The courts regarded appropriate arrangements for her cognitive development (schooling), adequate safety measures and training of her coping capacities as sine qua nons for permitting her to sail. Initially the court saw serious problems with respect to each of these requirements.
The court-ordered surveillance was intended to give Dekker the opportunity to make better arrangements regarding distance-learning, safety precautions, 'sleep management' and so on. 4 After almost a year of preparations, the last court judged that these arrangements had been made adequately. Dekker had prepared to follow the World Schooling program, written a sailing plan which detailed the various stages of the journey, and taken other necessary precautions. 5 On the other hand, the courts looked at Dekker's social, emotional and identity development. Here the courts took a less straightforward approach. The psychologist who was asked for an assessment concluded the following. With respect to her social development Dekker 'is inclined to operate independently (…), so that friendships take on a somewhat functional character and do not seem to rest upon reciprocity (on the basis of reciprocal interests)'. 6 Dekker's emotional development was characterized by 'onesided[ness]', 'flatness' and 'little emotional sensitivity'. She was judged to 'wave aside emotions like fear and sadness'. Moreover, 'the emotional side is poorly integrated in her personality development as a whole'. With respect to her identity development Dekker shows 'remarkably little interest in acceptance by her peer group, is self-centered and very self-satisfied'.
The Council for Child Protection took all of these psychological judgments to indicate that the solo journey would threaten her further development: she would be able to engage in 'superficial and functional contacts only', would not have occasions to work on her emotional development, and would lack the contact with people of her own age needed to develop her identity. By contrast, the court found no 'serious developmental threat'. The court also stated that her emotional development is not as one-sided as the Council believed, and questioned whether this alleged one-sidedness would be reduced if she were not allowed to undertake her solo journey. The court further judged that Dekker's identity development might be abnormal compared to that of her peers, but that her development did not pose a threat to her. Moreover, 'she will have fewer problems from not being able to mirror herself to her teens than would be the case with an average child'.
All in all, the court in Utrecht did not see an impediment to the journey because of Dekker's current development. As long as the journey did not aggravate her supposed developmental issues, it was not a reason to keep her home. However, this conclusion was not followed straightforwardly by the other courts. The Court of Appeal was much more critical of the threat to Dekker's development. 7 The Court of Middelburg had the final word. It concluded that insufficient information prevented it from making a good judgment of the matter. Meanwhile, the relation between the governmental agencies and Dekker and her family had seriously deteriorated to the point where Dekker and her family refused any further cooperation. On the basis of these circumstances, the court concluded that further prolonging the surveillance arrangement would likely be more of a threat to her social, emotional and identity development than lifting it. As a consequence, it allowed her to go. 8 In conclusion, the court's verdict to let Dekker undertake her journey is based on handling the two sets of requirements differently. The court was strict on what we will refer to throughout the paper as a child's 'technical competences'. These are always related to a specific activity that a child wants permission to undertake (here: sailing). The court had considerably more difficulty in determining the relevance of the second set of issues, the threats to Dekker's social, emotional and identity development (later we will take these together in our notion of 'autonomy development'); in the end it even denied that a positive evaluation of these was a necessary condition for allowing her to sail.

III. THE DEMARCATION DILEMMA
Central to our analysis of the Dekker case is what we will refer to as the 'demarcation dilemma'. This dilemma turns on whether to use age or competence as the primary threshold criterion for decision-making authority. What we will call 'age-based' theories treat the person's age itself as decisive (be it 12 or 16 or 21), such that all children of a certain age will be subjected to the same rule. This provides a practicable and clear criterion but also seems to do injustice to competent or precocious minors by denying them authoritative decision-making status on the basis of morally irrelevant considerations. By contrast, 'competencebased theories' treat each individual's actual decision-making competence or proficiency as determinative. The courts in the Dekker case did not decide the matter by a fixed age limit (as most countries do in determining a well-specified set of often-recurring risks like drinking and driving), nor did it tie Dekker's fate to the average competencies of people of her own age. Rather, Dekker's actual technical competencies were regarded as decisive. In taking this line, the courts adopted a competence-based view, while many opponents in public debate took an age-based view. In this section we hope to show that both age-based and competence-based theories are morally problematic.
The problem with age-based theories is that they seem to rely on ageist discrimination against children. Our current legal systems regularly rely on systematic restrictions on the individual liberty and political representation of a large segment of the population, simply on the basis of age. This fact clearly requires a convincing argument for why such matters ought to be decided on the basis of how many years one has been alive. It's not, of course, that age per se is morally relevant. Age-based theories are characterized by their use of age as a proxy for morally relevant factors that often coincide with youth, such as inexperience, impulsivity, or ignorance. 9 This linkage is what critics of age-based criteria for decisionmaking authority do not accept. Empirically, cases of precocious children demonstrate how unreliable age can be as a proxy. And as a matter of principle, respect for the equal dignity of persons as individuals would seem to require that they not be lumped together on the basis of a morally irrelevant criterion. Instead, these critics insist that treatment be tied directly to the individual's possession of the requisite proficiency or competence in the domain in question. 10 This seems to suggest that competence-based views are superior. But advocates of this position encounter a powerful set of objections regarding what it would actually mean in (legal) practice. As Feinberg puts the point, 'the law cannot do without rigid lines dividing ''standard persons'', who because of their age are presumed to have sufficient capacity to play some given legal role, from those below that age who are not. That is because direct tests of capacity in particular cases without recourse to such rules would be cumbersome to administer, or unreliable, or both' (Feinberg 1986, p. 326). 11 There is a troubling potential for bias in competenceassessments. Determining someone's age is ordinarily a value-free and reliable affair. Processes of determining whether someone is 'sufficiently competent', by contrast, are ordinarily value-laden, highly contested, and riddled with problems of reliability -all of which open the door to systematic biases.
In some cases, of course, procedures of competence-assessment are rather well established, such as getting a driver's license. But matters quickly become much more controversial once we have to assess issues such as Dekker's psychological preparedness to make her solo voyage. These determinations create the potential for bias and abuse, with vague criteria being interpreted in ways that go against the interests of the agents themselves. In addition, competence-criteria turn out not to be neutral between different ways of leading a life, where each way has its own distinct prerequisites. For example, does Dekker's cool affect and emotional distance from peers mean that she is at psychological risk or does it rather mean that she is especially well suited for a solo journey around the world? Trying to answer that question quickly draws one into substantive issues of how she can best lead her life, issues that governments and courts are rightly keen to avoid.
Up to this point there seems to be an irresolvable dilemma between ageist discrimination against children, on the one hand, and the risks of unworkable and controversial competence tests, on the other (Schrag 1977, p. 336). Even so, defenders of age-neutral, competence-based approaches might still claim to have the moral high ground. After all, they will say, the difficulties and even political risks involved in administering tests for psychological maturity or decision-making competence pale in comparison to the moral violation involved in systematic discrimination on the basis of the morally irrelevant criterion of age. If the choice is between pragmatism and morality, the advocates of pragmatism seem to be on shaky grounds, especially when they are adults defending an automatic privilege.
This conclusion is too quick, however, for it overlooks the fact that there is a more principled objection to a competence-based approach. As a principle, age neutrality cuts two ways. Competence is competence, and so it would not only be those under the age of 18 who would have to demonstrate they make the grade for decisionmaking authority. On this approach, 40-year-olds who fail to meet the standards would fall into the same category with incompetent 12-year-olds. But is it really a good idea to start assessing the emotional maturity and relationship-skills of adults in determining whether they should be allowed to sail solo around the world? An age-neutral approach has the unpalatable result of opening the door to rampant paternalism towards adults.
Defenders of age-neutrality could, of course, bite the bullet here, candidly welcoming an expanded scope for intervening in the lives of immature and incompetent individuals above the age of 18. At this point, however, it becomes clear that age-neutral, competence-based criteria for adulthood sit very uneasily with the fundamental liberal principle that adult persons have a special authority over their own self-regarding choices, including what others deem 'bad' decisions. If the liberal distinction between harm to others (a basis for interference) and harm to self (no basis for interference) is to have any bite, at least some class of persons needs to have the possibility of actually harming themselves. The paradoxical feature of this class of persons, normally called 'adults', is that they all have a 'right of autonomy' to an equal degree -even though they may differ markedly in the extent to which they have developed the skills, character, and capacities typical of fully autonomous persons. In other words, it is a SAILING ALONE: TEENAGE AUTONOMY AND REGIMES OF CHILDHOOD core commitment of liberalism that, although autonomy, in one sense, develops over time and admits of gradations, in another sense it marks a threshold. Everybody above that threshold has decisionmaking authority and 'rights of autonomy', such that further differences in their capacities are of no moral significance (Feinberg 1986). In this second sense, 'having autonomy' is a categorical distinction in status: one either is or isn't an autonomous agent. By ruling out any role for age thresholds, competence-based approaches have trouble avoiding a problematically tight linkage between degrees of decision-making authority and degrees of competence. 12 In light of the foregoing discussion, it becomes clear that resolving the demarcation dilemma ultimately requires finding a way to preserve a liberal policy towards adults that doesn't involve arbitrary discrimination against children. For this end, however, we need a principled way to explain why treating childhood as a separate stage in life is not merely determined by pragmatic considerations.

IV. CHILDHOOD AS A SEPARATE STAGE IN LIFE
Up to this point, we have presupposed the demarcation dilemma is about whether age or competence should mark the cut-off point between childhood tutelage and adult decision-making authority. What this focus on a specific criterion misses, however, is the larger context within which the child-adult contrast is supposed to make sense. What we propose is to reverse the order of explanation, justifying an approach to demarcation in light of the meaning and distinctive purpose of childhood. What first needs elaboration, then, is an overall account of what 'being a child' is about, including the rights, duties, powers, immunities, and entitlements which are jointly constitutive of what it is to be a child.
Particularly useful in this context are two articles by Tamar Schapiro: 'What is a Child? ' (1999) and'Childhood andPersonhood' (2003). Schapiro begins with the claim that childhood is not a natural category, the way the ability to walk or sail is. Developing such 'proficiencies' is not sufficient for being an adult. Rather, adulthood (like childhood) is an attributed, normative status that exists only in virtue of being ascribed and earned. Just as physically possessing an object is insufficient for being able to claim it as one's property, developing certain abilities (or reaching the age of 18) is not by itself what makes one an adult. In the absence of a wider set of practices and institutions, neither property nor adulthood exists (Schapiro 2003, p. 585). Thus Schapiro rejects the proficiency argument (which is a species of what we have called 'competence-based views'), but not in favor of an age-based view. Rather, she presents what seems to be a third alternative that has the potential to transcend the previous two views. In this alternative, the distinction between children and adults is one of status. But what does this status consist in?
In explaining her position, Schapiro draws on Kant's understanding of the transition from the state of nature to civil society. Civil society is not simply a condition in which certain features present in the state of nature become more pronounced. Rather, the emergence of civil society involves a group of individuals taking one another to be members of civil society, as constitutive of a new way of relating to one another. It is a fundamental shift in kind, not just in degree. For our purposes, then, the key intuition is this: just as civil society isn't merely an incremental expansion of the political features present in the state of nature, adulthood isn't merely an incremental expansion of a child's abilities. Both involve a seismic shift whereby a fundamental and distinctively normative transformation is brought about -even if the properties and characteristics of the composite parts change only gradually.
Schapiro describes the parallel shift from childhood to adulthood as the emergence of a kind of 'constitution', a state of being organized as a self-guiding agent around a core set of commitments, principles, and perspectives. What children lack is their own perspective on their choices, a constitution on the basis of which to adjudicate between conflicting inclinations or impulses: 'Thus the condition of childhood is one in which the agent is not yet in a position to speak in her own voice because there is no voice which SAILING ALONE: TEENAGE AUTONOMY AND REGIMES OF CHILDHOOD counts as hers' (Schapiro 1999, p. 729;2003, pp. 587-589). 13 Crucial to Schapiro's account is that, as with the transition from the state of nature to civil society, the emergence of a perspective or voice is not merely a matter of having a little more voice or perspective, since a point of view is something one has only in virtue of one's ways of approaching and responding to the situations being integrated. 14 This position makes it necessary for Schapiro to explain what children are actually doing when they act. Contrary to Kant's position that children are characterized by mere 'animality' and cannot actually be said to engage in action proper, Schapiro characterizes what children are doing as a kind of pretending or 'play'. 'By engaging in play, children more or less deliberately 'try on' selves to be and worlds to be in' (Schapiro 1999, p. 732). Just as, in pretending, one can make the same movements that one would make 'for real'think of a bride and groom exchanging vows during a wedding rehearsal -what happens when children express a point of view or plan a course of action is fundamentally different from what happens when the expressing and planning is done by adults. The difference is normative: children's doings and sayings do not have the same status or meaning. Many children (especially teens) would understandably chafe at the idea that what they are doing is pretend-play, but the suspension of full responsibility is part of the social reality within which children act. In this sense, even a high level of competence in practical reason, self-understanding, and critical reflection is not sufficient for being an adult. Only self-constitution qualifies.
The other part of Schapiro's account is that, as with other forms of rehearsal and pretend-play, children learn a great deal from the increasingly serious rehearsal of activities such as expressing a perspective or deliberating about significant choices. Playing is a form of 13 Schapiro's general line of argument is not completely new. Earlier Geoffrey Scarre argued that paternalism towards children is legitimate since they lack 'the ability to plan systematic policies of action' (Scarre 1980, p. 123). Also more recently Robert Noggle suggested that while children sometimes have the capacity for 'simple agency', they lack 'fully stable moral selves' (Noggle 2002, pp. 100-101;also Noggle and Brennan 1997). 14 We here follow her 2003 article. In the 1999 article, Shapiro gave a different account. There she argued that children gradually acquire authority over different 'domains of discretion'. In some areas, we allow them to act according to their own wishes while in other areas we are more reluctant to grant them this kind of authority. As children grow older, the number of domains in which they achieve discretion grows until finally all domains are under their control (Schapiro 1999, pp. 733-734). We agree with Schapiro that the new, 2003 model fits better with her overall theory. The 1999 position is mistaken in suggesting that decisions to grant discretion in particular domains relied on the basis of 'local attributability arguments'. The competent child should not be seen as acting as 'herself' in any domain, however well she acts. experimentation which allows the child to adopt different roles and positions, so as to build up experiences about the world and one's place in it. According to Schapiro, a kind of bootstrapping takes place, by which children prepare themselves for adulthood by rehearsing for the time at which they step onto a different stage and play the role of adults. Her discussion suggests that what is distinctive of childhood is that the social world within which children play and rehearse is oriented toward that role -and that means that there is room for experimentation and a gradual expansion of challenges. But children are not held responsible to the same extent as adults, and above all they require supervision. Without these elements in place, the behavior of children no longer is that of play and rehearsal, but rather hit-or-miss attempts at doing what adults door mere attempts to survive. Indeed, in such cases we tend to speak of individuals being 'denied a childhood'.
Where does all of this leave us with respect to the demarcation dilemma -our central issue in this paper? On the one hand, we think Schapiro's account is very helpful in making the shift from technical competencies (proficiency) to the global status of childhood. On the other hand, however, she hasn't yet developed the arguments to ward off a threatening 'second demarcation dilemma'. In this second dilemma, a pragmatic case for age limits yet again stands opposed to a moral case for competence tests -only this time the object of the dilemma is autonomy development itself, not any set of technical competences. Schapiro herself is surprisingly silent on this issue. Selfconstitution marks the transition, and requires that self-constitution is attributed to someone. But it remains unclear whether or not this status is attributed on the basis of age. As long as this issue remains unresolved, Schapiro's account cannot be applied to a case like Laura Dekker's.
The claim that self-constitution is what is constitutive of adulthood might suggest that Schapiro is actually defending a competence-based criterion for demarcating childhood from adulthood. Indeed, some of her formulations suggest that she believes selfconstitution is sufficient for adulthood. But if we view self-constitution -understood as comprising the ability to assess matters critically from an integrated point of view -as the sufficient criterion for adulthood at any age, we are back to the foregoing difficulties faced SAILING ALONE: TEENAGE AUTONOMY AND REGIMES OF CHILDHOOD by competence-based views. Nine-year-olds who precociously develop mature decision-making skills and a distinctive point of view would no longer be children -which would put Schapiro squarely in the camp of the child-liberationists. This implication not only runs against her stated aim of refuting child-liberationism but is also morally unattractive. On the other hand, if we stipulate a specific age as the point at which self-constitution is attributed to individuals, we seem to be licensing treatment of individuals on the basis of a morally irrelevant category. Doing so would ignore the fact that selfconstitution involves capacities that some will not yet have developed and others will have already long had. Laura Dekker, for example, is not claiming merely that she is a very good sailor but that this trip matters deeply to her and that she is competent to make that choice.
How ought one to deal with this second demarcation dilemma, where self-constitution (or, as we will henceforth say, 'autonomydevelopment') has taken the place of technical competencies as the competence criterion? To answer this question, in the next two sections we will argue that Schapiro's account of the distinction between adulthood and childhood and the importance of childhood tutelage needs to be integrated in a wider context, what we call a 'regime of childhood'. V. REGIMES OF CHILDHOOD As we are using the phrase 'regime of childhood', a regime comprises the set of norms, practices, institutional arrangements, guiding ideals, criteria regarding thresholds, etc. on the basis of which a particular status is ascribed to individuals -in our case, the status of 'being a child'. A regime is constituted by institutionally and culturally backed understandings of what this status licenses bearers of the status to do, what others are obligated or forbidden to do, and so on. 15 The regime of childhood centrally includes tutelage and limited decision-making authority. To speak of a regime is to speak of the entire network of normative interrelations. As in the case of other normative statuses (such as citizenship or mental competence), a regime of childhood is identified and justified as a package.
As an illustration, take the mundane 'regime' governing selling one's car. The most relevant change in 'normative status' resulting from the sale is that the seller no longer has any claims to the use of the car and that the car is now the personal property of the buyer. In many countries, the sale is embedded in a further set of relations and institutions. The sale thus also transfers obligations to pay vehicular taxes, insure the car, and get regular safety or emissions inspections. At the same time, the seller retains responsibility for paying parking fines incurred before the sale. And so on. These facts are rarely specified explicitly in the terms of sale but rather are part of the relevant 'regime of vehicular property' comprising the laws, regulations, and customs that set the terms of what is entailed by a car sale. This regime describes what it means to become the owner of a car, and neither buyers nor sellers may alter this regime at will. Different jurisdictions may decide to arrange matters differently, but the justification for any particular regime depends on the regime as a whole.
A regime of childhood comprises an even more complex network of relations, governing the status change from childhood to adulthood. We propose to focus on four distinctive, interlocking aspects of the specifically 'modern regime of childhood': an orientation towards autonomy development, limited liability for children, parental supervisory responsibilities, and age-based demarcation. 16 What is ultimately at issue in the Dekker case (and others like it) is whether this regime of childhood is preferable to an alternative in which adulthood is demarcated less on the basis of age and more on the basis of competence. Our position is that the modern regime is less vulnerable to competence-based challenges than defenders of Dekker's claim have suggested. In elaborating this position, we describe how the first three aspects can be seen to fit together. In the next section we turn to the fourth aspect, the specific issue of how to handle the diverse ways in which the precocious development of competence might be viewed as grounds for exemption from agebased limits on decision-making authority.
First, the guiding purpose of the modern regime of childhood is a commitment to autonomy, not as the criterion for a normative status but as a set of capacities that are important for adult life and that require a period of development. The current regime of childhood is thus by its nature geared toward promoting and making possible a developmental process. This fact gives it the paternalistic and 'perfectionist' character that so irritates many liberals (Scarre 1980, p. 117). But surely a regime of childhood should enable a process of maturation. We take it that Schapiro's articles have sufficiently established this point. We will take the paramount importance of autonomy as given, and not defend this guiding purpose against other (say, pre-modern) regimes of childhood in which autonomy does not play such a fundamental role (e.g. in which adulthood is connected primarily to the maturation of other capacities). A full defense of the modern regime would of course require such a comparative investigation of the value of autonomy. We also omit a discussion of which skills precisely are necessary for being an autonomous adult in a given society, and what level of development is necessary for decision-making authority and full responsibility. For example, as societies become more complex and require more complex autonomy-capacities, this may very well require a longer trajectory of development, and therefore a higher age of majority, at least relative to less complex societies. 17 Second, the current regime circumscribes the responsibilities of children. Minors are not fully accountable for their actions, given that they are not fully their own. Although this accountability increases with age, minors never have ultimate responsibility. As Schapiro's discussion of 'play' suggests, limited liability creates a protected space for the duration of childhood in which there is room to experiment, explore, and practice with a safety net in place. The guiding intuition here is that autonomy development is facilitated by such a circumscribed period of experimentation. Although this intuition would ultimately have to be supported by empirical research in developmental psychology, support can be found in the familiar idea that there are costs to moving too quickly to a more difficult and demanding level of activity. In the case of musical skills or sports skills, coaches and teachers are regularly cautioning against children doing exercises or activities for which they are not yet prepared. Aside from possibilities of injury, the concern here is that once an athlete or a dancer or pianist has started to put all the different component parts of the performance together the individual moves to a different perspective on the performance, and any attempt to look at the component parts again will always be from that perspective. There is in this sense, then, no going back. The 'opportunity costs of premature advancement' involve a lost opportunity for a kind of development that occurs best (or perhaps exclusively) within the framework provided by a particular level of development. This is just one example of the general point that having a period of being a novice can be functional for the development of certain abilities.
Third, the current regime of childhood can only facilitate modes of exploration by arranging for a period of tutelage, since the skills and dispositions required for autonomous adulthood are best acquired as a novice and with a safety net in place. The justification of parental authority and responsibility fits in here, in that parental fiduciary duties are not merely arbitrary impositions of power, but partly constitutive of the possibility of exploration, rehearsal, and play as such, since this involves supervision being in place so that the full responsibility regarding things going wrong does not have to lie with the child. Parents are responsible not only for giving a child the possibility to act upon her own decisions as she gradually becomes capable. They also are responsible for protecting the interests of the adult the child is to become. Harry Brighouse has usefully drawn a fourfold distinction between immediate and future welfare interests and immediate and future agency interests (Brighouse 2003, p. 701). 18 Parents have to take all of these interests into account.
With these three elements of the modern regime of childhood in the picture, we now can approach the fourth aspect: the demarcation of childhood on the basis of age. Within the context of a regime of childhood, the question about demarcation now becomes: What scheme of supervisory responsibilities on the part of parents and what duration of the period of childhood tutelage best serves to realize the fundamental interest in autonomy development? No longer are we talking about a dilemma of demarcation. Rather, we are talking about selecting the package of age-based thresholds for adulthood that best promotes the autonomy interests of individuals. In the next section we take a closer look at this fourth element of the regime of childhood.

VI. EMANCIPATION
Any regime of childhood should be explicit about how it deals with children's gradual acquisition of competences, both more technical and autonomy-related competences. To what extent should children be allowed to act upon their growing competencies and thus be emancipated from their childhood status? In this section we propose a way of answering this question by invoking a distinction between local emancipation and global emancipation. 'Local emancipation' will refer to specific exemptions from age-based status-ascriptions while still remaining under tutelage more generally. 'Global emancipation' will refer to exemptions whereby those below the age of majority are no longer under tutelage at all (although they may not automatically acquire all of the rights typically accorded to adults, such as the right to purchase alcohol).
Local emancipation serves to accommodate precocious children who have reached certain technical competencies in specific domains. Thus, for example, a regime of childhood might allow for numerous procedures by which minors can apply for exemptions from default age thresholds. Differing degrees of responsibility and decision-making authority might be accorded differently across different domains of activity. Commercial and occupational choices, for example, might be treated differently from choices in medical contexts, and experimentation with hobbies and sports might be treated differently from experimentation with drugs or alcohol. In some cases, for example, liability or responsibility for foolish or harmful actions might be shared between children and guardians. Local emancipation allows for these competence-based exceptions. Making arrangements in which children gradually take on responsibility and 'suffer the consequences' of poor choices are necessary for autonomy development to be effective. But, as we saw, the very idea of childhood as a separate status presupposes that parents retain final JOEL ANDERSON AND RUTGER CLAASSEN responsibility for the overall process of their children's development. The buck stops with them. 19 Global emancipation is a different matter, because decision-making authority is no longer limited to specific contexts. Here, a child is granted the status of an adult before reaching the legally prescribed age of majority, and the supervisory role of parents or guardians is terminated. A given regime of childhood might implement global emancipation in two different ways. The first is to restrict global emancipation to exceptional circumstances, such as an insurmountable conflict between parents and children in the absence of suitable guardianship arrangements, or (more controversially) when a child marries. This is how the law in most countries handles global emancipation. 20 In such cases, global emancipation is not viewed as valuable per se, but as part of the best solution to a difficult situation (call this 'exceptional global emancipation'). The second approach is to make global emancipation into a straightforward, elective option for qualified children. A regime that takes this path effectively abolishes the age of majority and endorses an age-neutral, competence-based approach to demarcation. Most likely this would involve retaining a default age of majority but permitting children to apply for an assessment of their level of maturity and autonomy. In a regime of childhood that permits this 'elective global emancipation', it would not be necessary to provide evidence of exceptional circumstances such as familial hardship. Allowing the option of suing for adulthood would not abolish the status-distinction between children and adults altogether. Everyone still first goes through a phase of childhood before reaching adulthood. But the moment of transition would be radically transformed along the lines of a competence-based, age-neutral approach. 21 Our position is that elective global emancipation is problematic, and that both local emancipation and exceptional global emancipation are more defensible. In supporting this position, we must explain why we oppose giving children who meet the minimal requirements for autonomy the corresponding decision-making authority. If we are willing to endorse exemptions from age restrictions in specific domains on grounds of technical competence, why not allow exemptions for the restriction that ultimately matters most, childhood tutelage itself? Of course, the kinds of pragmatic considerations mentioned in Sect. III -administrative costs and dangers of biases in testing -play a role here. Elective global emancipation would be an addition to the current legal system and raise administrative costs. The biases when judging autonomy development are likely to be much more severe than when testing competencies for relatively circumscribed sets of activities (like driving). As we mentioned in Sect. III, the courts' treatment of Dekker's developmental opportunities was much more contentious than its assessment of her sailing competencies. However, these pragmatic considerations alone do not provide a definitive objection to elective global emancipation and therefore still keep us trapped in the second (higher-level) demarcation dilemma.
Within the context of a regime of childhood, we are now in a position to supplement these pragmatic considerations with inherently relational considerations. We have to decide whether allowing some children to leave the childhood phase earlier than others undermines the beneficial effects of the regime on all children. We think there are three regime-specific reasons to be critical of elective emancipation.
A first reason is the need for stable expectations with respect to the parent-child relationship. A regime needs to give parents feasible structures to perform the responsibilities attributed to them as part of the regime. It is plausible that a strongly competence-based approach to demarcating childhood and adulthood would undermine the ability of parents to carry out their responsibilities effectively (thereby jeopardizing the long-term autonomy interests of the children involved). Allowing the child's status to be a constant focus of negotiation and renegotiation creates uncertainties that strain parent-child relationships. If so, one of the advantages of age-based criteria is clear. Publicly recognized rites of passage at particular ages introduce much-needed stabilizing points of reference into the complex process by which parents navigate with their teenage children the transition to adulthood, as in, 'When you're 18, you can decide that for yourself, but not yet'.
A second reason for not allowing global emancipation is that it would tend to undermine healthy in-group relations between children. Compare this phenomenon to the controversy about mandatory health insurance. In many countries, the state requires non-indigent individuals to purchase health insurance. This requirement enables the state to pool many different risks, so that health coverage can be extended to all at a moderate price. It may be advantageous for 'good risks' (those with good health expectations) to be allowed to exit the collective pool and insure themselves. The 'bad risks' that remain behind, however, will pay a much higher price. 22 A similar argument can be made against the elective emancipatory option. If most children turn out not to be interested in early emancipation, there would be no negative consequences for the regime. 23 But if many would try to exit, a schism might be created among children, who are often very anxious about belonging within their peer group. Some would feel inferior because they haven't qualified for the adulttest (or, even worse, have failed it). Others would be stressed about trying to meet the test. These results could severely strain relations among children.
A third reason relates to the potential for inequality between children. In an elective emancipatory regime, children from stable and fortunate family backgrounds are the most likely to succeed in opting out. Existing inequalities of power, wealth, and social status can be expected to worsen. The practical consequences of a generalized practice of differentiation in seeking adulthood status are difficult to foresee. Still, it would be surprising if the sphere-transgressing influence of money would not come to play the role it has in so many other aspects of life (Walzer 1983).
Before continuing, we would like to address two potential objections. The first is that we are exaggerating the problems with permitting elective global emancipation because the number of cases will be vanishingly small. But are there reasons to believe it will be rare?
Two possibilities are worth discussing. The level of global emancipation might be low because few children request it or because most requests for emancipation would be denied on the merits. Take the first possibility. If the criteria are set at a relatively relaxed level, such that 10% of 12-year-old children qualify, it might still be thought that few would actually take advantage of the exemption before the age of 16 or 18. But it is very risky policymaking to permit an option that would be problematic if pursued widely on the basis of an assumption about the low numbers of people who would be inclined to take advantage of it. Trends can change, and trying to roll back these permissions would be politically difficult. Alternatively, it might be thought that few children would qualify even if they wanted to do so. Indeed, a regime of childhood could ensure this result by setting the bar very high. But suppose it then turned out, as we think likely, that the only way to keep the number of qualifying 15-or 16-year-olds to low levels would be to require aspirant adults to demonstrate exceptionally well-developed skills in self-discipline, rational planning, emotional maturity, and so on. If so, we are back to the objections raised earlier by critics of competence-based approaches to demarcating childhood. Not only are young persons required to meet a higher standard for full-fledged adulthood, but defining full-fledged adulthood in such demanding terms would put pressure on the touchstone liberal assumption that individuals above the age of 18 (or 21) ought to count, by default, as genuine adults.
Furthermore, we do not accept the assumption that elective global emancipation is problematic only if large numbers of minors take advantage of the opportunity. We have compelling reasons to take global emancipation off the table as an elective option (and to permit it only when doing so is required to avoid a significant harm). The foremost such reason is grounded in the expressive value of a competence-neutral boundary between childhood and adulthood. As several legal philosophers have argued, part of the purpose of law consists in what it expresses about the significance of certain actions, institutions, and practices (Anderson and Pildes 2000;Sunstein 1995). A policy of elective global emancipation would express a view of childhood as defined by thresholds of competence. It thus equates childhood with a condition of inability that children can leave as soon as they have acquired the necessary skills. It becomes a golden cage that children escape by convincing their parole board they are ready for life on the outside. Rejecting elective global emancipation in favor of the approach we have proposed replaces this message with an emphasis on a period of time in which the development of individual autonomy is central as a societal value and goal. It is a phase of life devoted to the structured and supervised acquisition of autonomy.
It might be objected, second, that our approach adds nothing significant to an approach based exclusively on the practical advantages of a clear line of demarcation. As we acknowledged in Sect. III, it is often appropriate to rely on a notion of the 'standard person' (Feinberg 1986, p. 326). This reliance minimizes administrative and juridical responsibilities that drain public coffers, overburden the legitimacy of the courts, or expose individuals to the risk of biased treatment. Thus the courts and legislatures could just say that the 'standard 15-year-old' is ill-prepared for global emancipation and leave it at that. We believe that such an exclusive focus on administrative reasons is inadequate, for it fails to take seriously both the strongly held objections of teens such as Dekker and the real importance of childhood as a context for autonomy development.
Having now formulated a principled set of objections to elective global emancipation on the basis of a reconceptualization of the demarcation dilemma in terms of regimes of childhood, we finally return to the Dekker case.

VII. SAILING ALONE?
Although we do not aim to resolve the question of whether Laura Dekker ought to have been allowed to sail, in this section we will point to relevant features of the case that would need to be decided in order to answer that question. What is intriguing about Dekker's case is that part of what has been in dispute is whether allowing her to sail alone constitutes what we have called local or global emancipation. Our position is that it would be less problematic if she (and her father) were suing for merely local emancipation. In that case, SAILING ALONE: TEENAGE AUTONOMY AND REGIMES OF CHILDHOOD provisions would have to have been made to enable the father to carry out his supervisory duties. Otherwise, we have a situation of de facto global emancipation.
Above, we argued that an essential component of any regime of childhood is the suspension of full responsibility. One way to bring this out is by imagining the public response to news that Dekker had drowned as compared with the news that someone twice her age, undertaking the same feat, had drowned. In the case of adults, we would view it as a tragic loss, but perhaps also partly the result of foolishness by the adult sailor. In the case of a young teen, we suspect that any doubts about the judgments involved would be directed not primarily at the child but at her parents. For the reasons discussed earlier, this shift is entirely appropriate. Children take risks, but the results of this risk-taking are at least partly the responsibility of parents. And it is, we have suggested, central to the currently accepted regime of childhood that children be protected from the full brunt of responsibility for their exploratory and risk-taking activities in order to develop into autonomous persons.
In this regard, it is interesting to consider the role of Laura's father, Dick Dekker, who supported his daughter throughout the proceedings. 24 Dick Dekker's willingness to let Laura undertake the solo trip could be interpreted as entailing a willingness to accord her the status of an adult. On this view, he could be seen as supporting his daughter's plan by effectively ceding his guardianship. And, indeed, given a regime of childhood in which children's immunity from full responsibility is tied to the parents bearing that responsibility, Dick Dekker's relinquishing his responsibilities amounts to an attempt to effect his daughter's global emancipation. Importantly, however, a parent cannot decide on his or her own to relinquish these responsibilities, even if the child wants him to do so. It is essential to the regime of childhood as we know it that no one can be removed from the status of being a child by parental fiat. Dekker's father can decide to relinquish guardianship, but the state auto- 24 We see no reason to doubt that the impetus and driving force behind the plan to sail around the world lay with Laura Dekker and not her father, although this may well play a decisive role in analyzing other cases, such as those of child performers. In Dekker's case, it is the child who is seeking (some form of) emancipation. Note that a potentially complicating factor that we bracket is the opposition of Laura's mother, from whom she was estranged. Of course, since we do not have full access to the court documents, we cannot here make any adjudication of what Mr. Dekker's actual motives, intentions, or judgments are in the case. We have focused instead on the implications of possible scenarios. matically steps into the resulting vacuum. Were Mr. Dekker's aim to end his daughter's childhood, the courts would be correct to deny his suit.
Alternatively, Mr. Dekker might actually understand himself to be still exercising his parental responsibilities by adequately and appropriately supervising his daughter, albeit at a distance. In our opinion, this route is more fruitful. Dekker's case has to be interpreted as one of local emancipation. What does this mean for judging the permissibility of her journey?
First, this means courts cannot avoid saying something about the scope of the responsibility that parents must exercise. Using the fourfold distinction introduced by Brighouse (see Sect. V above), they should take into account the probability that Dekker will normally enjoy her trip (that she has an immediate welfare interest in it). Furthermore, assuming all goes well, the trip could be expected to train several parts of her agency (representing an immediate agency interest). However, what courts still must squarely face is the question of whether the trip could also be expected to advance the interests of the person she is later going to be, both in terms of agency and welfare. It might just as well turn out to thwart her future agency (e.g. if her personality development has been stunted) or her future welfare (e.g. if she later regrets not having spent her teenage years among peers). The court treated these kinds of future interests in terms of Dekker's social, emotional and identity development. As mentioned in Sect. II, the court could not properly assess these interests because Dekker and her family did not fully cooperate. On our account, however, this assessment is imperative and the fact that the court waived it can be criticised as too permissive.
Second, the development of children's autonomy requires parental supervision. Parents need to be able to intervene if they judge the child to be insufficiently competent to make decisions or carry out specific plans. Whether this supervision requires parents to conduct more or less continuous monitoring of children depends on several circumstances: the precise details of the planned journey Dekker is undertaking, the technological resources that are available for monitoring at a distance, and so on. We are not in a position to judge whether the relevant criteria are met in this specific case; we simply argue that these considerations should have been central to the SAILING ALONE: TEENAGE AUTONOMY AND REGIMES OF CHILDHOOD court's deliberations. Note that this issue is different from Dekker's technical competencies -the issue on which the court focused. The point about supervision is not merely about safety. Rather, it is an issue of whether or not the supervisory responsibilities constitutive of parenthood can be exercised under the circumstances of a solo circumnavigation. Childhood requires guardians, whose role is oftenparticularly in adolescence -that of a safety-net, background monitor, and 'sleeper' supervisor, who swings into action when needed. To fulfill this role, there needs to be not only enough communication, but also an ongoing relationship that provides a supportive context for constructive intervention.
Of course, the supervisory responsibilities of parents change as the child approaches the age of majority. Hence parents must judge the extent to which their children are able to handle new responsibilities, even if they lack ultimate responsibility. So it might not be problematic that Dekker has a lot of decision-making responsibility on her journey if she turns out to have developed her personality to such an extent that she can independently handle a great many unexpected situations. One might still worry, however, that the waiver from actual supervision that Dekker gets from her father is so generic and across-the-board that she seems to gain something like decisionmaking authority on her trip -just the thing that distinguishes global emancipation from local emancipation. Whether this is the case is something the courts must decide. If Dekker's sailing plan excluded global emancipation in words only, but in effect would be a case of global emancipation under the guise of local emancipation, the courts would have had reason to prohibit her plans.

VIII. CONCLUSION
In this paper we have taken the Dekker case as a point of departure for examining the legal, moral and political status of childhood. We have shown how the debate about the Dekker case deadlocked by being framed in terms of the dilemma 'age versus competence'. We have argued that we should move beyond this dilemma by reconceptualizing the distinction between childhood and adulthood as one of status. This opens up a more fruitful -though decidedly more complicated -debate over various ways of arranging and institu-tionalizing a 'regime of childhood'. The issue of emancipation is one important part of the specific regime of childhood in modern Western societies, which strongly emphasizes the idea of maturing into an autonomous person. It is our hope that the usefulness of thinking of childhood as a normative status accorded within a regime extends beyond the issue of emancipation. Other controversial issues and court cases surrounding childhood can also be analyzed in these terms.

OPEN ACCESS
This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited. ACKNOWLEDGEMENT For feedback on previous versions of this essay, we wish to thank audiences at the annual meeting of the Dutch-Flemish Political Science Association, as well as at the University of Amsterdam's Colloquium in Philosophy and Public Affairs, where we benefitted especially from Peter Rijpkema's prepared remarks. Joel Anderson also gratefully acknowledges the Netherlands Institute of Advanced Study for the fellowship support during much of the writing of this paper. We also wish to thank two anonymous reviewers for pressing us to clarify several key points.