A Journal of Philosophy, Applied to the Real World

Moral Education in the Liberal State

Department of Philosophy, Northwestern University


I argue that political liberals should not support the monopoly of a single educational approach in state sponsored schools. Instead, they should allow reasonable citizens latitude to choose the worldview in which their own children are educated. I begin by defending a particular conception of political liberalism, and its associated requirement of public reason, against the received interpretation. I argue that the values of respect and civic friendship that motivate the public reason requirement do not support the common demand that citizens “bracket” their comprehensive commitments in politics. Rather, citizens should seek to enact policies the justification of which is compatible with the truth of their fellow reasonable citizens’ worldviews. Next I argue that no single educational approach can meet this standard of justification. Many believe that state sponsored education in a pluralist, liberal society ought to present multiple worldviews in a neutral way. I argue that this aspiration is unrealizable, and no other educational model will plausibly meet the justificatory demand. Finally, I address two objections to my favored alternative: that it may allow for the inculcation of disrespect, and that it violates children’s autonomy. Against the first, I claim that political liberals have no grounds for thinking that reasonable citizens will seek to inculcate disrespect. Finally, I argue that there is no conception of autonomy that can sustain the second.


Educating a child is in part a matter of transmitting information, but also a matter of shaping her normative outlook.  Those who play a significant role in a child’s education will inevitably affect what she comes to value and to love, thereby influencing not only her intellectual but also her moral development.  The hallmark of the view called political liberalism is the affirmation that views about what is good—what is worth valuing and loving—are contested in ways that have normative implications for the use of state power.1  Most self-identified political liberals nevertheless champion common schools for all, and some even argue that such schooling should be required by law. Here I argue that the values that motivate political liberalism support a different conclusion. Its basic commitments are not best realized by sponsoring, much less prescribing, a single educational approach for all children. The liberal state does best by taking neutrality to a higher level, allowing parents considerable latitude in choosing the conception of the good in which their children will be educated.2

I begin in the first section by defending a particular conception of political liberalism for which I have argued at greater length elsewhere (Ebels-Duggan 2010b). In the second section I argue that no single educational approach can be justified to all reasonable citizens. I do not attempt a complete case for my favored alternative here, but in the last section I sketch responses to what I regard as the two most pressing objections to it.

What political liberalism is (and what it cannot be)

Respect is the foundational value of political liberalism. Endorsement of the significant value of political relationships that respect all reasonable citizens characterizes the view.3 But so far this is no more than a schema. Its practical import depends on what we mean by “respect” and whom we count as “reasonable.”

Political liberalism is a view about the respect due to others as citizens,4 and political liberals standardly understand it as requiring that exercises of state power over a person be justifiable to her. I adopt Rawls’ label, “civic friendship,” to name reciprocal relationships of this sort (Rawls 2001, p.579).5 To stand in relationships of civic friendship is, in an important sense, to use state power cooperatively together, rather than coercively against one another.6

Political liberals regard civic friendship as an ideal for political relationships, but only among “reasonable” citizens. That is, political liberals believe that we have reason to realize relationships of civic friendship with reasonable fellow citizens, using state power only in ways that can be justified to them. Clearly, much then turns on how we understand the class of the reasonable. If I were to regard others as reasonable just in case they share my own views, then the requirement to justify my political proposals to all reasonable citizens would look trivial to me. Any policy that I found justified by my own lights would thereby seem to me to be justifiable to all reasonable people, and purported reasons to justify proposals to others would do no normative work at all. Because political liberals claim to have a distinct view in which the ideal of mutual justification provides real guidance, they cannot understand the reasonable in this way.

For the purposes at hand “reasonable” is best understood as a term of moral rather than epistemic commendation.7 In particular, the reasonable person has a certain attitude about politics, which I think is best interpreted as comprised of the following two commitments: (1) Affirmation of the great value of civic friendship with others who seek this cooperative relationship. (2) Recognition that among these are some citizens whose convictions about what is valuable are different from and potentially in conflict with your own, and thus that the reasons to cooperate offer non-trivial normative guidance.8 A citizen who is reasonable in this sense will be willing, in principle, to forgo using political power in pursuit of some of the things that he values for the sake of realizing civic friendship. He can make sense of this restraint from his own point of view, that is without denying the values in question, because he also values civic friendship. A reasonable citizen thus recognizes that he cannot move directly from a value that he affirms to conclusions about the reasonableness of pursuing or promoting this value politically. He is committed to considering also what can be accomplished cooperatively.9

So far, then, we have the following characterization of political liberalism: relationships of mutual justification among reasonable citizens are valuable, giving rise to significant practical reasons. We should count among the reasonable all those who affirm the value of relationships of mutual respect. The upshot is that we have significant reason to limit the use of the power of the state to those policies that all reasonable people can accept from their own point of view.

To demonstrate that a policy meets this standard, political liberals must respond to objections to it by showing that either (1) the objections are unreasonable, that is no one who is reasonable in the sense just discussed could endorse them, so they need not be answered, or (2) the policy can be justified in a way that ought to move the dissenters by their own lights, so the objections can be answered.

This understanding of the argumentative burden requires a revision in the standard account of the content of public reason. Rawls’ public reason requirement demands that justifications for fundamental political policies appeal only to considerations that any reasonable citizen can recognize as reasons.10 Rawls presents this as requiring that citizens not appeal to the content of their worldviews in political reasoning.11 Most contemporary political liberals follow him in holding that, for the purposes of politics, such views must be “bracketed” or “set aside.” Call this the bracketing conception of public reason.12

Despite its popularity among self-identified political liberals, the bracketing conception does not provide the best way to realize the value that motivates the public reason requirement. What provides reason to respect the requirement, if anything does, is the value of relationships of civic friendship. In order to realize these relationships, justifications must be such that all relevant citizens can accept them even as they affirm their own worldview. Only so can each see himself as cooperating in, rather than coerced by, the policy in question. But this requires something different from bracketing. It requires that justification for political proposals be compatible with the truth of the comprehensive doctrines held by reasonable people in the society, such that reasonable citizens can accept these justifications without thereby denying their own convictions.13 Call this the shared reasons conception of the public reason requirement. While the bracketing conception emphasizes not affirming or relying on your own worldview in politics, the shared reasons conception emphasizes avoiding justifications that are incompatible with the worldviews of others.14

It should be clear that any violation of the shared reasons conception upsets relationships of civic friendship. No one can accept a justification that would hold only if her worldview were false. Any policy requiring this kind of justification cannot be enacted cooperatively with those whose view it denies, but only coercively over their objections. So adherence to the shared reasons conception is a necessary condition for realizing civic friendship. It also appears to be sufficient: if a policy can be justified in a way that is compatible with the truth of your worldview, then you can accept the justification from your own point of view.

One might worry that the resulting standard of justification is too strong. It is true that it will often be difficult to formulate political policies that appear justified from all of the relevant points of view. In fact, on the shared reasons conception it will sometimes—perhaps even often—be impossible to do so, and so impossible to realize fully relationships of civic friendship with all reasonable citizens. Political liberals standardly hold not only that we have significant reason to justify our policies to all reasonable citizens, but also that we have a general obligation to do so. In my view, there could not be such an obligation, because the availability of policies that all reasonable citizens can see as justified is a contingent matter.15 More generally though relationships of civic friendship are of great value, they are not the only thing of value, and the view so far does not determine a course of action when other values conflict.16

The idea that you must bracket your comprehensive doctrine in political discussion gains its plausibility from the fact that one way in which you might run afoul of the shared reasons conception is by appealing to a value that you affirm but others deny. But the bracketing conception is both too broad and too narrow. Appeals to values found in your comprehensive doctrine will not upset relationships of civic friendship if your fellow reasonable citizens also affirm these values. Moreover, you violate the requirements of civic friendship even without explicit appeal to your worldview if your justification would be undermined by the truth of a particular worldview held by some reasonable people in the society.17 Those who hold this worldview cannot then see the justification as sufficient grounds for the policy.

Meeting the standard of justification given by the shared reasons conception of public reason is a necessary and sufficient condition for realizing relationships of civic friendship, while adhering to the more widely affirmed bracketing conception is neither. The shared reasons conception is thus superior to the bracketing conception in that it better captures the underlying value that public reason is meant to serve. To realize this value, one needs policies that all can see as justified not when they set aside their comprehensive doctrines, but given their views, or in light of the fact that they take their own doctrines to be the true or best ones.18 Talk of bracketing is misleading because, while it is true that some appeals to our worldviews are inappropriate in politics, the content of these views is nevertheless of central importance in determining the limits of possible cooperation.

This leads to a final, underappreciated, point: In order to enact political liberalism so conceived, we need to attend to what our fellow citizens actually think, and what they can actually endorse from their own point of view. We need to do the hard work of conversation and co-deliberation. We cannot generate the content of public reason in our offices in the philosophy department. We can only generate it together.

In this section I have sketched the motivations for a political liberal view and briefly argued that interpreting political liberalism according to the shared reasons conception captures these motivations better than the alternative bracketing view. Political liberalism affirms an attractive political ideal: genuinely cooperative relationships among those who genuinely disagree. If you believe that significant reasonable disagreements are endemic to a free society then this seems to be the only alternative to voting each other down and implementing coercive policies that cannot be seen as justified by some reasonable people whom they govern. Even if the value of cooperative relations does not always override other values that we could possibly realize politically, it is a significant value that can and should do normative work in guiding our political decisions. It is also, in an important way, a foundational value. It is not just a value that we try to produce through politics, but one that governs the constitution of our political relationships.

Education Policy in the Liberal State

I turn now to the implications of this version of political liberalism for issues surrounding education in the pluralist, liberal state. Political liberals face the following question: How can state-sponsored and/or state-imposed schooling be designed so as to be justifiable to all reasonable citizens? If the forgoing is correct, this amounts to the question of how education policy can be justified to all those who value civic friendship without requiring them to deny their other values. In this section I argue that, due to the importance of schools in the development of children’s normative outlook, and the resulting inevitable entanglement of education with controversial worldviews, no single educational approach can be justified to all reasonable citizens.19

The central question for political liberals is how the state’s approach to education will relate to the various, conflicting worldviews reasonable citizens affirm. Citizens who are parents have a particularly deep interest in this question, and introduce unique issues about justification. Many, probably most, parents take the raising of their children to be as central, important, and defining a task in their lives as any other. More importantly for our purposes, most take themselves to have obligations to and with respect to their children. Parents standardly regard shaping their children’s character, helping them to value things worth valuing and to recognize appropriate considerations as reasons for acting, as central among these obligations.20 At the same time, in contemporary American practice most children spend a great deal of time during their most formative years in schools. On this widely accepted model schools participate significantly in the raising of children, shaping their normative outlooks in profound ways. So, without denying that educational policy should be justifiable to all reasonable citizens, we can acknowledge that the way in which schools relate to the worldviews of parents is of special interest. Consider then these four possibilities: A school might

    (1) Educate students exclusively in and about their parents’ worldview, while suppressing knowledge of other ways of life.
    (2) Teach children about many ways of life, in a context that treats the worldview that their parents embrace as true or best.
    (3) Present multiple worldviews, including parents’ own, in a neutral way.
    (4) Advocate a single, state-selected worldview.

The first of these two approaches share an important feature: both undertake to inculcate a particular worldview. But, despite this important similarity, the second model should not be confused with or assimilated to the first. The first approach uses suppression of information about the existence and content of competing worldviews as a tool. By contrast, the second model welcomes exposure to multiple worldviews. It occupies the wide space between refusing to expose children to different views and maintaining, or purporting to maintain, neutrality among them. I will return to the importance of this distinction shortly.

On the face of it, (3) looks to many like the most promising approach to education in a liberal, pluralist society.21 It is certainly the only plausible candidate for a single approach to education that could be justified to all.22 However, I will argue that (3) is not a genuine possibility.23 Any education will orient children towards some normative view, and for any such view there will be some reasonable parents who reject it. As these parents see it, their children are being educated to value the wrong things. Many will take cooperating in this to run counter to their obligations as parents, and thus will not be able to see it as justified. Since the other models do not have even prima facie plausibility as unique state-sponsored approaches to education that could be endorsed by all, I conclude that no single approach can meet that standard.

Consideration of the much-discussed case of Mozert v. Hawkins will help illuminate what is at issue. The case concerned an elementary school reading curriculum adopted by a school board in Tennessee in the early eighties. Though primarily designed to teach reading, the curriculum also explicitly aimed to expose children to various ways of life or conceptions of the good with the goal of developing virtues of citizenship. The school board adopted it in part to serve these purposes. A group of Christian parents claimed that allowing their children to be taught using this curriculum would violate their religious convictions, and so requested permission to opt out of it while still receiving the other benefits of public education. The case is helpful for our purposes in two ways. First, charitable interpretation of the parents’ complaint illuminates the distinction between the first and second models of education, a distinction that is both important and often overlooked. Second, the case raises concerns about the viability of model (3). The curriculum in question seems to have been designed with precisely this model in mind. Yet it incurred objections that, I will argue, are best understood as pointing to its lack of neutrality.

Whatever the merits of the substance of these objections, political liberals should recognize them as of the right form. The parents assert that participation in the curriculum would violate their obligations as parents. Their understanding of these obligations is an integral part of their overall normative outlook. They cannot then continue to sincerely affirm this outlook and take themselves to have sufficient reason to endorse or cooperate in the educational model in question. Their objection thus amounts to a claim that the curriculum, and so its monopoly on state support, cannot be justified within the terms of public reason on the shared reasons conception.24

If that is right, a political liberal response to the Mozert parents must accomplish one of the two argumentative tasks laid out above: either show that their objection reveals that they are not reasonable, or show that education on model (3) can be justified to them without requiring them to give up or violate their own commitments. Liberal theorists who discuss this case usually treat it as raising issues about whether and how to accommodate unreasonable views. Observe, for instance, Stephen Macedo’s construal of the central question Mozert raises: “Should public justification have a second stage?…a stage where, having constructed a reasonable public view, we consider pleas for accommodations and exceptions for marginal groups” (Macedo 2003, p.483). But in relegating justification to the Mozert parents to a potential second stage without first explaining why they are properly regarded as unreasonable, Macedo begs the very question political liberals need to ask: is the curriculum under consideration indeed part of a reasonable public view?

We may take Amy Gutmann to be pursuing the first of the two approaches when she claims, “The Mozert parents rejected the relevance of the distinction between exposure to knowledge and inculcation of belief. … They rejected the idea that schools may teach children to understand and thereby to evaluate different ways of life. They assimilated such teaching to indoctrination into false beliefs” (Gutmann 1995, p.571). On this reading the parents object that merely by exposing their children to normative views that conflict with their own, the state inculcates these views, thereby violating the parents’ obligations. Call this the exposure objection. To the extent that the parents raise the exposure objection, only the first model of education would satisfy them, and Gutmann claims that this is indeed what they seek. Elsewhere she argues at length that this sort of insularity is not compatible with a commitment to what I have called civic friendship.25 Communicating the value of civic friendship to children requires teaching them what Rawls calls the fact of reasonable pluralism, and this cannot be done without exposure to conflicting views.26 Since reasonable parents affirm the value of civic friendship, they are concerned to pass it, along with their other values, to their children, so they will not seek an education on model (1). I will treat this argument as decisive against the exposure objection and the first model of education. So, to the extent that Gutmann’s characterization of the parents’ objection is accurate, it is appropriate to treat the case as Macedo does, as one about whether and how a liberal society should accommodate certain unreasonable citizens.

But the strongest—and so most theoretically interesting—construal of the parents’ objection understands them differently in two ways. First, it does not take the exposure of their children to diverse views as the object of their concern.27 There is no doubt that the parents see themselves as obligated to pass on a particular normative outlook to their children, and reject model (3) because they believe that cooperating in this model is incompatible with fulfilling this obligation. But this commits them to the insularity of model (1) only if advocacy of a single view is incompatible with allowing exposure to others. So, while Gutmann accuses the parents of eliding the distinction between exposure to a view and the inculcation of that view, treating (1) as the only alternative to (3) apparently commits her to the closely related incompatibility thesis that one cannot present a single outlook as true or best, while also exposing students to alternatives. 28

In fact, both claims about the relationship between exposure and inculcation are false, as we can see by considering an example. Suppose we undertake to teach our children about racism, a view that we deplore. It seems that we should teach them about the history and content of racist views, and should not shield them from the fact that some people still hold such views. Thus we expose them to racist views. But we will also make clear our own anti-racist commitments, presenting these as correct, not merely as opinions or matters of taste or as one among other possible options. The case should leave no doubt that exposure to a view is distinct from inculcation of that view. It also presents a clear example of aiming to inculcate a particular view, while simultaneously exposing students to an incompatible outlook.

The lesson to draw from this case is that exposure is a red herring, a distraction from what is really at issue. It is neither the real focus of the Mozert parents’ objection, nor the correct characterization of what liberals like Gutmann and Macedo seek. In fact, the parents demonstrate that they recognize the distinction between exposure and inculcation when they say that they would endorse an education that exposed their children to alternative views, so long as it included teaching that the views in question were in error (Mozert, p.1064). The racism case is an example of just this kind. Here and elsewhere the parents demonstrate that they are not committed to model (1), but better interpreted as reaching for model (2). Their objection to (3) is then not due to the fact that it exposes children to alternative views. Rather, they assert that the curriculum in question goes beyond exposure to inculcate views incompatible with their own. Call this the non-neutrality objection.29

We cannot then dismiss the Mozert parents as unreasonable on the grounds rehearsed above. However, the racism example also makes clear that encouraging civic friendship requires more than mere exposure to alternative views. When we communicate the content of a racist view in order to repudiate it, we are not promoting civic friendship with those who hold the view. Since racism is a paradigmatically unreasonable view this isn’t a problem; we are not concerned to encourage civic friendship with racists. But in order to count as reasonable, the Mozert parents must acknowledge that not all cases of disagreement are like this. They must value civic friendship with some citizens with whom they disagree. Call the treatment of competing views needed to communicate the value of civic friendship with their adherents “respectful presentation.”

In the face of this, Gutmann could modify her case for regarding the parents as unreasonable. She might claim that they reject respectful presentation of alternative views, and in this way seek to educate their children in a way that betrays lack of commitment to civic friendship. She might then dismiss the parents as unreasonable even while acknowledging that they seek model (2) not model (1). This argument will go through if, but only if, model (2) is not compatible with presenting alternative views in a way that encourages civic friendship with those who hold them. I will set this question aside for the moment, returning to it in the next section. For now, notice that if (2) is compatible with valuing civic friendship, then we cannot dismiss the parents as unreasonable. If we cannot dismiss them as unreasonable, then political liberals must take seriously their claim that cooperation in the curriculum violates their parental obligations, and those who would defend the approach that the curriculum represents must try to answer it.

So far I’ve been arguing that the parents should be understood as advancing the non-neutrality objection rather than the exposure objection. In the remainder of this section, I explain the second modification of Gutmann’s interpretation of their position, and argue that the resulting claim provides reason to think that model (3) is unacheivable. In advancing the non-neutrality objection, the parents claim that the curriculum inculcates views incompatible with their own. Gutmann represents them as concerned that their children will embrace one of the views contained in the reader. They may have this worry, but the non-neutrality objection is better understood as raising a more sophisticated, second-order challenge with implications well beyond the case at hand: In teaching the material in these readers an instructor must either present the various views and lives represented there as on a evaluative par with one another or present some as superior to others. But neither option can be justified to all reasonable citizens.

Presenting some of the views as superior to others is obviously unjustifiable to citizens who affirm one of those in the latter category. But presenting all as on an evaluative par would also be problematic. Most everyone thinks that some matters are appropriately regarded as evaluatively discretionary while others are not. But, and this point is of central importance, which matters are evaluatively indifferent is itself an important, difficult and controversial normative question. This is just the sort of question over which reasonable citizens will persistently disagree.30 Many parents take themselves to be obligated to help their children get these matters right.

For example, most people in our society think that career choice should be largely determined by personal preferences. That is, almost everyone regards some wide range of career options as evaluatively indifferent. Most will thus regard presenting a wide range of options to children neutrally as an important aspect of education.31 But what exactly should be included is another matter, and can be an important one. How, for example, can one deal neutrally with military careers in a classroom that includes children from pacifist families? Endorsing the pacifist view that the military is an impermissible option is clearly inconsistent with neutrality. But treating choosing or rejecting a military career as an appropriate exercise of personal discretion would deny the truth of the pacifist convictions. Pacifists do not merely have a preference against their children choosing military careers, but view such choices as morally impermissible.32 Readers who are not pacifists themselves should substitute a career that they would regard as impermissible to see how this might look to such parents. Similar issues arise with respect to anything that incurs on matters that some citizens regard as determined by our obligations and others as rightly left to personal preference. To the extent that these questions are contested no presentation of the relevant material in formative moral education can be justified to all.

The Mozert parents take certain religious commitments and practices to be nondiscretionary, and understand their obligations as parents to include passing these convictions to their children. Presenting various religious commitments and practices to their children as appropriate matters for personal discretion, much as some careers are, conflicts with this obligation, and so cannot be justified them. As they recognize, the educational approach consistent with their understanding of their parental obligations with respect to such matters is not an appropriate candidate for unique state support. It would be unreasonable to seek to impose this education on all children. But an approach that treats a wide range of religious and non-religious views as equally correct or choiceworthy is no less contested. The more sophisticated reading of the non-neutrality objection in Mozert takes it to invoke this problem, and to object that the curriculum advances a second-order view that they reject.

Someone might object that this argument conflates two kinds of neutrality. On the one hand, there is neutral presentation of various worldviews or issues that incur on them. This is what schools would have to accomplish to stand a chance of satisfying the strictures of public reason with a single educational approach. On the other, there is the substantive normative position that the views in question are all equally correct or choiceworthy, and so evaluatively discretionary in the sense above. Call this substantive neutrality. Substantive neutrality is a position among contested views, and so not only different from neutral presentation but, as I have been arguing, in conflict with it. An advocate of model (3) might thus agree that treating contested matters like religious commitments as questions appropriate for personal discretion is not neutral among worldviews, but insist that the third model does not require, or even allow, this.

As a conceptual matter, these two sorts of neutrality are distinct. But, in practice disagreements concerning them will tend to arise together for good reason: in formative educational contexts, it is extremely difficult to present normative positions without explicitly or implicitly taking some position on their permissibility, value or truth.33 Presenting a view while contrasting it with another that one explicitly endorses would avoid communicating substantive neutrality, but model (3) does not allow this. The parents object that presenting a variety of normative outlooks in a way that aims to be neutral among them will tend to communicate substantive neutrality with respect to these views.34 This amounts to an education on the fourth model, one that inculcates belief in a single state-selected conception of the good.

Such concern might be warranted at all educational levels, but is most plausible with respect to early education. Avoiding the conflation of neutrality of presentation and substantive neutrality requires students to distinguish between (1) the teacher is talking as if all of these options are of equal merit, treating each even-handedly, and (2) the teacher believes, or advocates the position, that all of these options are of equal merit. This is a demanding cognitive task, and children who lack well-formed conceptions of the good cannot reasonably be expected to perform it.35

Interestingly, the importance of teaching the intellectual virtue of charity further complicates matters. Students must learn to put rival views in the best possible light, looking for what someone would find attractive in them, and to empathize with what it would be like to hold such a view. These skills are arguably pre-conditions for the ability to treat others respectfully in the political sphere, so all reasonable citizens should value them and endorse their inculcation in the next generation. Many theorists suppose that this shared value requires, or at least pulls in favor of, a single common curriculum. But there is good reason to think that it actually exacerbates the difficulty of designing such a thing by further blurring the distinction between the two sorts of neutrality. If merely presenting various views without advocating substantive neutrality is difficult, modeling these positive charitable attitudes without communicating substantive neutrality is more difficult still. In the next section, I will argue that the virtue of intellectual charity is easier, rather than harder, to model in an education that explicitly advocates a particular worldview.

It is then, at best, extremely difficult to avoid communicating substantive neutrality in the context of (3). Moreover, substantive neutrality is not the only controversial second order view about which such concerns arise. For instance, parents may have similar worries about what an education communicates about which normative questions should be regarded as open, and which we may, or even must, treat—like the racism question—as settled. The Mozert parents believe that certain religious questions are appropriately treated as settled, and may object to schools treating such questions as open. No doubt other parents would agree that these questions should be treated as settled, but disagree about how they have been settled. Some of these would object to traditional religious views being presented as serious options in a public educational setting. Still other parents will regard religious questions as exactly the sort about which we ought to lack confidence. Or, again, consider disagreements about which sorts of normative commitments are significant or of genuine moment.36 Here, too, questions about whether to take up a certain religious outlook provide a nice example. Some regard this as the most significant decision a person will make, while others think that it matters very little.

Reasonable citizens may have any of these convictions. And, while they are second-order, they are not necessarily esoteric or unimportant, but can figure importantly in a person’s overall normative outlook. Many parents will regard an educational approach that communicates what they take to be false views about important normative matters as in conflict with their parental obligations, and so as unjustified. But it is difficult to see how any attempt to implement (3) could negotiate all the relevant convictions simultaneously, avoiding taking or communicating a position on any of these issues. Any settlement amounts to siding with some of these parents and against others, and collapses again into model (4). Thus no single approach to education will be publicly justifiable.37 Political liberals, who aim for legislation that can be justified to all reasonable citizens, should thus seek an educational policy that provides parents with a wider range of options.

Toward Cooperation on Education

In the last section, I argued that any approach to education will effectively communicate some worldview. If this is right then someone must make the choice about what this worldview will be. There seem to be only two plausible contenders: the state and those with primary responsibility for raising the children, usually their parents.38 The former will certainly involve the state in action that not all reasonable citizens will take to be justifiable. This is not is decisive against it, since, as I acknowledged above, there are some questions that the state cannot resolve in any way that is justifiable to all. For all that I have said so far, how the state ought to involve itself in education might be such an issue. But accepting state interference with what many reasonable citizens take to be among their most important ends and obligations, in ways that cannot be justified to these citizens, should not be done lightly. We have weighty reason to seek a policy that all reasonable citizens could see as justified, and opt for such solution if we can find one.

I believe that we can. Within some limits, deference to parents’ choices about the worldview in which their children will be educated—a version of model (2) above—can be justified to all. I cannot give a complete defense of this claim here, but will respond to two important objections to it: First, that if parents are allowed to educate children into their own worldview they may inculcate intolerance and so undermine the development of essential virtues of citizenship on which the stability of a just political order depends. This amounts to a concern that educating children on model (2) cannot be justified to citizens generally. Second, that education on model (2) violates children’s autonomy. This amounts to a concern that such education cannot be justified to the children who receive it.

The basic response to the first worry is that political liberals have no grounds for supposing that reasonable parents who seek to pass on their worldviews to their children using model (2) will teach them not to respect others in politics. As reasonable, they are committed to the value of civic friendship.39 Given the opportunity to inculcate their own values these parents will, from their own point of view, have reason to pass this value, among others, to their children.

I think that this response is successful, but it will seem far too quick to some. Readers may worry that it begs the question set aside in the previous section: Is education on model (2) compatible with respectful presentation, or does seeking this sort of education itself reveal that a parent is not fully committed to the value of civic friendship? Recall that I granted that the latter is true of those who demand model (1): Insulating children from any exposure to alternative worldviews is plausibly incompatible with valuing civic friendship, since participating in cooperative political relationships requires some understanding of others’ views. Model (2) includes exposure to alternative views, but above I argued that exposure is not enough: Passing on the value of civic friendship requires, in addition, what I called respectful presentation. Opting for model (2) will thus render the parents unreasonable if—but only if—model (2) is not compatible with respectful presentation. By the same token, model (2) would fail to be justifiable to other reasonable citizens.

Why might one suppose that model (2) is not compatible with respectful presentation? The ground of this suspicion must be that, in self-consciously advocating a single worldview as correct or best, model (2) is committed to teaching that other views are wrong or sub-optimal where they conflict. Some may worry that teaching that a view is wrong is not compatible with presenting it respectfully. But political liberals should—by their own lights—reject this incompatibility claim. They should do so because it is a central commitment of political liberalism that the respect due to others in politics is compatible with disagreement over the truth or acceptability of their worldview. The whole point of political liberalism is to show that and how we can respect one another politically even in the face of deep and abiding disagreement over our worldviews. If this project has any hope of success then it must be the case that one can regard a view as wrong while yet respecting, in the relevant sense, those who affirm it. Rejecting this possibility amounts to abandoning political liberalism. But affirming it undermines the most plausible source of the doubt that (2) is compatible with full affirmation of civic friendship.40

In fact there is some reason, to which I alluded above, to think that the second model can communicate the value of civic friendship more effectively than the third, purportedly neutral, approach. There are certain intellectual and political virtues that a teacher can most effectively model for students only if he is not required to purport to neutrality among worldviews. On model (2), a teacher can explicitly affirm that a given outlook is correct, while also presenting others charitably, and explicitly encouraging political respect for those who hold them. The teacher can also directly affirm these virtues, and explain to students why someone who holds the worldview that he advocates should support them. By contrast, the teacher who must purport to neutrality cannot take any clear stand on matters of which view is correct, and so cannot model charity towards a view that he is known to reject, or political respect for people who hold this view.

Above I suggested that children who cannot distinguish between respect for those who hold a view and acceptance of that view may, as a result, come to embrace substantive neutrality among a variety of worldviews, as some parents fear. But another result is possible. Children who are not taught to distinguish between political and substantive neutrality may unnecessarily resist accepting the former on account of rejection of the latter. They may come to wrongly regard political neutrality as incompatible with their worldview, and so fail to develop virtues of empathy, charity and political respect. I take this to be the threat that political liberalism is designed to address, and a possibility that all reasonable people should seek to avoid. What I am claiming here is that one plausible strategy for doing so is to have teachers who make this distinction in their own outlooks teach and model it. If this is right, then it turns the first objection to model (2) on its head. Many political liberals appear convinced that model (3) is the best—or even the only—way to teach students how to be reasonable citizens. But, if valuing political respect is compatible with these various worldviews, it isn’t at all clear why model (2) shouldn’t serve this purpose at least as well, if not better.

Turn now to the second objection. There are at least three constituencies to whom educational policy would need to be justifiable if it is to be a cooperative endeavor among all citizens: parents, citizens generally, and children. The second objection claims that education on model (2) cannot be justified to the last of these, the children who receive it.41 In particular, it asserts that such an education violates these children’s autonomy, and that they cannot reasonably accept this.

The very idea of justifying policy to children raises difficult questions. Political liberals understand respect for an adult citizen as requiring that we enforce against her only those policies that she can see as justified given her own worldview. But it seems wrong to understand children as having any particular worldview. Thus it is inherently difficult for political liberals to determine what should count as respecting a child, or as justifying a policy to her.

Nevertheless, any plausible view must recognize that a child is a distinct individual with her own claims on the state. To point out the difficulty above is not to deny this, but only to recognize a complication about how to understand the content of children’s legitimate claims. Since standard notions of justification do not apply easily to children, we might instead try to articulate their claims in terms of their interests, treating a view as justifiable to children just in case it gives their interests proper consideration. But this appealing suggestion brings complications of its own, since the reasonable disagreement that is the starting point of political liberalism entails reasonable disagreement about how to conceive these interests. Understanding children’s claims in terms of their interests thus immediately raises the question of whose conception of children’s interests we should use.42

Neither the range nor the significance of these disagreements should be overstated. It is widely agreed that children need basic material provision and nutrition, safety and the sense of safety, and stable and loving relationships especially with the adults who are their primary caretakers. And it is extremely plausible to hold that any residual disagreement on these fronts should be overridden, even if this requires coercing some who seek cooperative political relationships. Thus we can justify many important child protection and welfare laws by appeal to children’s interests.43

The objection at hand might be understood as holding that children’s interest in autonomy also belongs on the list of those so fundamental that we should be willing to override the decisions of even reasonable parents who would oppose it. So we need to consider whether children do indeed have such a significant interest in autonomy, and if so whether model (2) would be distinctively prone to violate it. We cannot settle either question definitively without a complete conception of autonomy, something that I can’t possibly offer here. But, even in advance of trying to fill out such a conception, the argument of the last section gives us reason to be skeptical about the claim that (2) is distinctively problematic. Someone might think that (2) threatens autonomy in virtue of the fact that it directs children to a particular conception of the good. But if the argument above is right then any education will have this effect.44

Many readers will nevertheless find tempting the view that an education that at least aims at model (3) is more conducive to autonomy, so it is worth considering the objection in greater detail. The best response would be an argument that no plausible conception of autonomy can support the objection. That is, there is no sense of autonomy for which it is plausible both that children have an enforceable claim to its development and that it is in principle in tension with an education on model (2). Here I can do no more than sketch the case for this claim.45

A leading theme in discussions about autonomy and education is that of thinking for one’s self or, what I take to the be same, self-direction according to one’s own reason. Eamonn Callan’s view is representative here. He characterizes autonomy this way: “Autonomy in the sense that counts here is about the capability and inclination to reason for one’s self, and to shape one’s life on the basis of the deliverances of reason.”46 This elusive idea is subject to both a strong and a weak reading. On the strong reading one directs one’s self according to the deliverances of reason only if one restricts one’s value commitments to those one can vindicate by reason alone, without appeal to any non-rational factor. But the claim that this could be done for any, much less all, normative commitments is an extremely controversial position in moral philosophy, associated with one of the most ambitious programs in contemporary ethics.47 It seems to imply, for instance, that we should be able to provide a moral skeptic with a complete refutation, one that he should, by his own lights, accept. Careful political theorists rightly ward off the idea that their view depends on the prospects for success of this sort of grand project.48

Steering clear of this unattractively demanding interpretation leaves us with a more plausible, but weaker—and vaguer—standard. Note Callan’s looser talk here: “…whatever reflection autonomy requires surely does not demand that we detach ourselves from all our ends. The requirement is only that we be capable of asking about the value of any particular end with which we currently identify and be able to give a thoughtful answer to what we ask” (Callan 1997, p.54), emphasis mine.49 Elsewhere he tells us “I am autonomous to the degree that I have developed powers of practical reason, a disposition to value these powers and use them in giving shape and direction to my own life, and a corresponding resistance to impulses or social pressures that might subvert wise self-rule” (Callan 1997, p.148). In these characterizations the autonomous person is just the thoughtful person, and “thinking for one’s self” seems to be nothing other than “thinking.” If this is what autonomy means, then any approach to education should indeed seek to develop it. No one should deny that the well-educated person will be able to think well. But we’ve seen no reason to think that this version of autonomy is in any tension with model (2). It would simply beg the question to suppose without further argument that the second model of education cannot produce thoughtful people.

Many readers will think that this is surely too quick. Some will object that I have not considered the idea that distinctively autonomous thought requires the availability of alternatives to one’s current normative outlook.50 Now there are two clear senses in which students educated on model (2) do have access to normative alternatives. First, since this approach does not seek to insulate them from the existence and content of disagreement, they are aware of multiple points of view in the culture around them. Second, they understand that citizens with a wide variety of views have equal political status, so they both have—and know that they have—options politically. They know that no coercive sanctions will be brought to bear on them if they alter their worldviews.

But some will think that this is not enough, that there is a further sense in which these alternatives need to be live options, genuine possibilities for normative outlooks that they might adopt as their own, in order to enable a student’s autonomy. Callan makes this requirement explicit:

The essential demand is that schooling properly involves at some stage sympathetic and critical engagement with beliefs and ways of life at odds with the culture of the family or religious or ethnic group into which the child is born. Moreover the relevant engagement must be such that the beliefs and values by which others live are entertained not merely as sources of meaning in their lives; they are instead addressed as potential elements within the conceptions of the good and the right one will create for oneself as an adult. (Callan 1997, p.133).

For alternative views to appear as live options in this sense, one must be able to gain a certain distance on one’s own normative commitments, reflect critically on them, bring them up for review, or view them as contingent and revisable.51 This sense of autonomy strikes many as in fundamental tension with an education that seeks to nurture allegiance to a particular conception of the good.

But it is more difficult to specify this idea in a compelling way than one might suppose. Like the first aspect of autonomy, this one is subject to a strong reading that most autonomy advocates repudiate. On the strong view, autonomy requires alienating ourselves from all of our commitments and occupying a neutral standpoint from which we can reconstitute them. This seems to require an act of “criterionless” or existential choice. But it is widely recognized that any choice must be made on the basis of some normative commitments: views about what is valuable or worthwhile, and the considerations that count as a reasons. Such convictions are aspects of a conception of the good, not independent of them. This sort of argument convinces many that the idea of a neutral standpoint from which we could choose among worldviews is conceptually confused.52

Moreover, supposing we could specify the relevant standpoint of neutrality, it seems unlikely that anyone has an interest in occupying it, much less an interest so important that we should override the value of civic friendship to protect it. At the very least the standpoint in question would need to be carefully distinguished from normative uncertainty, skepticism or confusion, since it is not plausible to suppose that children have an enforceable right to any of these. The strong reading under consideration makes it unclear how to draw these distinctions. None of these points are novel, and they explain why friends of autonomy tend to distance themselves from identifying it with criterionless choice.

Many do so by suggesting that, though all of our commitments should be viewed as up for revision, they should not all be thrown into question at once. MacMullen tells us that, “The key to autonomous reflection is that one can always temporarily treat as fixed one’s allegiance to a particular belief or value while questioning others” (MacMullen 2007, pp.82-83). But if it were really meant to apply to all of our commitments this would still be too strong. Consider again the egalitarian rejection of racism. An adult whose education has rendered him incapable of regarding revision of this commitment as a real possibility is, to my mind, to that extent a well-educated adult. I see no appeal in an education that would encourage children to achieve a degree of critical distance from, seriously question, or entertain as realistic the possibility of abandoning this particular commitment.

Such wholehearted normative commitment is compatible with autonomy as thoughtfulness. It is compatible with encouraging students to consider such questions as What does equality really mean? What does honoring it demand of us? What might be our reasons for affirming it? Asking these questions does not require any sort of doubt or ambivalence about the authority of the commitment in question; it is perfectly compatible with finding abandoning that commitment unthinkable. Notably then, model (2) can welcome and encourage such questions.

So far I have argued that if autonomy means merely being thoughtful about one’s value commitments, and informed about the presence and political status of normative disagreement, then it is not in tension with model (2). Moreover, it is hard to articulate a further value the term could name in this context. Even advocates deny that autonomy entails the absence of any non-rational determinates of one’s commitments and actions, or the occupation of a neutral standpoint from which one could choose all of one’s values. And while it might seem initially plausible that it requires something like the possibility of revising one’s commitments, on reflection it seems too strong to claim that this is generally of value. It is normal, and commendable, to hold that there are at least some value commitments that we do not want to treat, or encourage others to treat, this way. So we have not found a conception of autonomy that is both in tension with model (2) and plausibly among the most important interests of children.

But one might still ask whether there is some range of particular commitments the content of which it is inappropriate to instill—or aim to instill—in a way that makes them unrevisable. The answer is that of course there is some such range, but reasonable people disagree over its boundaries. We might put this point in the following, familiar sounding, way: which normative matters should be unrevisable is itself an important and difficult normative question, any answer to which will be controversial among reasonable citizens.53

Someone might then still hold that the state should enforce a particular view on which commitments children should be educated to regard as revisable. But, if the argument here is correct, this cannot be defended by appeal to an interest in autonomy independent of further controversial normative commitments. It will need to depend instead on appeal to children’s interests in being taught the normative truth, including the truth about which normative matters to treat as revisable, and a substantive position about what that truth is. That is, it will depend on precisely the kind of normative claims that political liberals hope to avoid in politics. Even supposing that excellent normative direction is indeed an essential interest of children, one ought to worry about assigning this task to the state. To do so is, at any rate, to abandon the political liberal hope for cooperation among reasonable citizens.


I’ve argued that no single educational approach can be justified to all, and that there is good reason to think that granting parents greater discretion could be so justified. But I’ve not made any particular policy recommendations for how to apply this insight in practice. Ad hoc arrangements, like the one that the Mozert parents propose, state sponsored and regulated home schooling, vouchers or a voucher like system, increased charter schools with particular stated worldviews, and some combination of these are all among the possibilities. There are simply too many empirical and logistical issues surrounding each of these for me to make any credible assessment. I claim only that any consideration of such policies must take the case made here in to account.

It is impossible to raise a child without taking substantive stands on controversial matters of value. Someone must make these choices, and political liberals have important reasons not to assign them to the state. Parents have a very deep interest making these choices, and I have argued that we have no compelling reasons not to trust fellow reasonable citizens to do so. Because they endorse the value of civic friendship such citizens will aim to pass this shared value on to their children, in the context of a particular wider and richer ethical view about which there is disagreement. In so doing citizens both enact and aim to reproduce the overlapping consensus of which Rawls speaks.

Acknowledgements: My thanks to Mark Alznauer, Stacy Clifford, Ryan Davis, Candice Delmas, R.J. Leland, Ian MacMullen, Stephen Macedo, Joseph Mazor, Bryan McGraw, John Oberdiek, Alan Patten, and Laura Valentini for comments on earlier drafts. Thanks to audiences at the Northwestern University Practical Workshop, the Vanderbilt Political Theory workshop, the Society for Applied Philosophy conference at St. Anne’s College in Oxford in June of 2012, and the Virtue of Justice Conference at Hendrix College in March of 2012. I am also grateful to the Princeton University Center for Human Values and the Spencer Foundation for providing financial support during my work on this article.


Ackerman, B. A. 1980. Social Justice in the Liberal State. New Haven, CT: Yale University Press.

Audi, R. and N. Wolterstorff. 1997. Religion in the Public Square: The Place of Religious Convictions in Political Debate. Lanham, MD: Rowman & Littlefield.

Brighouse, H. 2000. School Choice and Social Justice. Oxford/New York: Oxford University Press.

——— 2006. On Education. New York, NY: Routledge.

Burtt, S. 1996. In Defense of Yoder: Parental Authority and the Public Schools. In: I. Shapiro and R. Hardin (eds.) Political Order. New York, NY: New York University Press.

——— 2003. The Proper Scope of Parental Authority: Why We Don’t Owe Children an “Open Future.” In: S. Macedo and I.M. Young (eds.) Child, Family and State. New York, NY: New York University Press.

Callan, E. 1997. Creating Citizens: Political Education and Liberal Democracy. Oxford/New York: Clarendon Press.

Clayton, M. 2006. Justice and Legitimacy in Upbringing. Oxford/New York: Oxford University Press.

Cohen, J. 2009. Philosophy, Politics, Democracy: Selected Essays. Cambridge, MA: Harvard University Press.

Dombrowski, D. A. 2001. Rawls and Religion: The Case for Political Liberalism. Albany, NY: State University of New York Press.

Ebels-Duggan, K. 2010a. Awarding Custody: Children’s Interests and the Fathers’ Rights Movement. Public Affairs Quarterly, 24, 257-78.

——— 2010b. The Beginning of Community: Politics in the Face of Disagreement. The Philosophical Quarterly, 60, 50-71.

——— forthcoming-a. Autonomy as Intellectual Virtue. In: H. Brighouse and I.M. McPherson, (eds.) Justice and Aims in Higer Education.

——— forthcoming-b. 2014. Educating for Autonomy: An Old-fashioned View. Social Philosophy and Policy, 31:1.

Freeman, S. 2000. Deliberative Democracy: A Sympathetic Comment. Philosophy and Public Affairs, 29, 371-418.

——— 2006. Justice and the Social Contract. New York, NY: Oxford University Press.

Galston, W. 2011. Parents, Government, and Children: Authority over Education in a Pluralist Liberal Democracy. Law & Ethics of Human Rights, 5 (2), 285-305.

Gutmann, A. 1995. Civic Education and Social Diversity. Ethics, 105, 557-579.

——— 1999. Democratic Education. Princeton, NJ: Princeton University Press.

Korsgaard, C. M. 1996. The Sources of Normativity. New York, NY: Cambridge University Press.

——— 2009. Self-constitution: Agency, Identity, and Integrity. New York, NY: Oxford University Press.

Kymlicka, W. 1989. Liberal Individualism and Liberal Neutrality. Ethics, 99, 883-905.

Larmore, C. 2003. Public Reason. In: S. Freeman (ed.) The Cambridge Companion to Rawls. New York, NY: Cambridge.

Macedo, S. 1995. Liberal Civic Education and Religious Fundamentalism: The Case of God v. John Rawls? Ethics, 105, 468-496.

——— 2003. Diversity and Distrust: Civic Education in a Multicultural Democracy. Cambridge, MA: Harvard University Press.

MacMullen, I. 2007. Faith in Schools?: Autonomy, Citizenship, and Religious Education in the Liberal State. Princeton, NJ: Princeton University Press.

Mozert v. Hawkins. (1987) Court of Appeals, 6th Circuit.

Nagel, T. 1987. Moral Conflict and Political Legitimacy. Philosophy & Public Affairs, 16, 215-240.

Neufeld, B. and G. Davis. 2007. Political Liberalism, Civic Education, and Educational Choice. Social Theory and Practice, 33, 47-74.

——— 2010. Civic Respect, Civic Education, and the Family. Educational Policy and Theory, 42, 94-111.

Nussbaum, M. 1997. Culitvating Humanity. Cambridge, MA: Harvard University Press.

Patten, A. 2012. Liberal Neutrality: A Reinterpretation and Defense. Journal of Political Philosophy, 20, 249-272.

Rawls, J. 1993. Political Liberalism. New York, NY: Columbia University Press.

——— 2001. Collected Papers. Cambridge, MA: Harvard University Press.

Stolzenberg, N. M. 1993. He Drew a Circle that Shut Me Out: Assimilation, Indoctrination, and the Paradox of a Liberal Education. Harvard Law Review, 106, 581-667.

Wolterstorff, N. 2002. Religion and the Schools. In: G. G. Stronks and C. W. Joldersma, (eds.) Educating for Life. Grand Rapids, MI: Baker Academic.

1. Compare Eamonn Callan’s statement of the problem: “If the role of the state in education is to keep faith with its constitutive morality, a path must be found between the horns of a dilemma. The need to perpetuate fidelity to liberal democratic institutions and values from one generation to another suggests that there are some inescapably shared educational aims, even if the pursuit of these conflicts with the convictions of some citizens. Yet if repression is to be avoided, the state must give parents substantial latitude to instill in their children whatever religious faith or conception of the good they espouse” (Callan 1997, p.9).

2. No doubt some will regard a successful argument for this conclusion as reason to abandon political liberalism in favor of a more comprehensive liberal approach. Nothing in the structure of my argument blocks this response. I do not aim to defend political liberalism as such here, but address my case to those who find the view antecedently attractive.

3. Cf. Stephen Macedo: “The basic motive behind political liberalism, it should be emphasized, is not fear of conflict or a desire to exclude religious speech from the public realm but the desire to respect reasonable people” (Macedo 1995, p.474).

4. On the need for a distinctively political liberal conception of respect, and the implications of this for arguments about education, see (Neufeld and Davis 2010) and (Neufeld and Davis 2007).

5. Cf. (Rawls 1993, p.li) Cf. Cohen on political community (Cohen 2009, pp.163-164).

6. I do not mean to suggest that this is the only thing that we might mean by cooperation, or that when the ideal of civic friendship is realized there is no sense in which exercise of state power might remain coercive.

7. Here I take myself to be following Rawls (Rawls 1993, pp.48-54). Some of the arguments in the literature on liberalism and education depend on a more epistemic interpretation of reasonableness, one that treats Rawls’ comments about the burdens of judgment as making significant epistemic commitments. See, e.g. (MacMullen 2007, p.55) and (Callan 1997, p.40). I defend the superiority of the moral interpretation in (Ebels-Duggan 2010b).

8. These roughly map onto Rawls’ two aspects of the reasonable person, as he states them in (Rawls 1993, p.54): “The first basic aspect of the reasonable, then, is the willingness to propose fair terms of cooperation and to abide by them provided others do. The second basic aspect…is the willingness to recognize the burdens of judgment and to accept their consequences for the use of public reason in directing the legitimate use of political power in a constitutional regime.” In (Ebels-Duggan 2010b, p.61) I put the two requirements this way: reasonable people “acknowledge the value of justifying policies on grounds common among all reasonable views, and they recognize the existence of reasonable views in society besides their own.” I don’t mean the first requirement here to be any different from my earlier statement, though I prefer the present formula since it removes an apparent circularity by eliminating the term “reasonable.” This revision of the first requirement may make the second redundant. I previously believed that it had to be included to guard against an overly narrow interpretation of the reference to the reasonable in the first requirement.

Compare Cohen: “Let us say that people are reasonable, politically speaking, only if they are concerned to live with others on terms that those others, as free and equal, also find acceptable.” (Cohen 2009, p.226). In an earlier paper Cohen has a different, more epistemic, view of the reasonable as those who have a stable disposition to affirm their views about value in the face of new information and critical reflection, but he seems to have subsequently abandoned this view. See “Deliberation and Democratic Legitimacy” in (Cohen 2009, p.52).

Contrast Freeman: “To be reasonable ultimately is to have a settled disposition to reason and act from the requirements of a liberal political conception of justice” (Freeman 2000, p.400). This understanding seems to subject the view to circularity that renders it unhelpful, though to be fair Freeman is clear that he does not regard this statement as a definition of the reasonable. He believes that no such definition is needed. Cf. (Freeman 2006, pp.227, 238).

9. I believe that this is the best way to understand Rawls’ principle that we should not try to “enact the whole truth in politics.” Cf. (Rawls 2001, p.574). Those who embrace civic friendship as a genuine and significant value do seek to enact the whole truth, as they see it, in the sense that they think that the value of civic friendship has an important place among the values to which political life appropriately responds. They may forgo the policy that they would enact if they presumed that everyone shared their worldview. But they do this on the basis of another value that they accept. So they seek to enact the policy that they think is best, all things (including the value of civic friendship) considered.

10. See, e.g., (Rawls 1993, p.226), (Rawls 2001, p.579).

11. See, e.g., (Rawls 1993, pp.141, 214, 236), (Rawls 2001, p.435).

12. The bracketing conception also follows from the combination of the limitation of public reason to a political conception of justice, e.g., (Rawls 2001, pp.581-584), and the requirement that the political conception must be presented as freestanding (Rawls 1993, pp.12-14, 144-145). For my worries about the “freestandingness” requirement see (Ebels-Duggan 2010b, p.64).

Samuel Freeman seems to embrace the bracketing conception in a thoroughgoing way in “Public Reason and Political Justification,” in (Freeman 2006). Cf. also (Freeman 2000). For other versions of the bracketing conception see (Larmore 2003) and Robert Audi’s contribution to (Audi and Wolterstorff 1997).

Joshua Cohen’s view is more ambiguous. In one place he speaks of “restricting ourselves to common ground in the face of diversity” (Cohen 2009, p.55), and surrounding context suggests that he has the bracketing conception in mind. However, he elsewhere recognizes the need for public justifications to take account of the force of each reasonable citizen’s comprehensive view from her own point of view. This different conception of the role of comprehensive doctrines in public reasoning is on display when Cohen defends a right to religious liberty (Cohen 2009, pp.164-166). In making this argument he relies on the idea that appropriate public justifications take seriously the fact that some convictions result in felt obligations and not merely preferences. So, he writes “…reasonable adherents cannot accept, as sufficient reasons in support of a law or a system of policy, considerations that would preclude their compliances with fundamental religious demands or require that they treat these demands as a matter of choice” (Cohen 2009, p.244). And later, “…we need to let political ideas of burdensomeness track the weight of reasons within the reasonable views of those we are regulating” (Cohen 2009, p.313). These statements differentiate his view from one in which we determine in a vacuum, or by exclusive appeal to the political culture, which considerations count as “public” and only later apply these to policy questions. I have argued elsewhere (Ebels-Duggan 2010b) that the latter approach leads to mistaken reasoning precisely on issues in which some citizens see themselves as having obligations. But these statements of Cohen’s seem to me to be in considerable tension with the following (lifted from one of his discussions of abortion): “because of the pluralism of philosophies of life among politically reasonable citizens, some bases for regulating conduct are politically weightless. To take the clearest case, people hold some commitments on faith, and take those commitments to impose overriding obligations…because they are expressly held as truths known through faith, they are matters on which reasonable people disagree, and adherents cannot reasonably expect others to accept those considerations as having any weight and, therefore, cannot use them in justifying regulations” (Cohen 2009, pp.309-310).

The bracketing conception figures prominently in liberal discussions of education. E.g., “These ultimate beliefs need not be bracketed in the sense of being denied, but they should be bracketed in the sense of not being invoked or relied upon as the public ground of decision for matters of basic justice” (Macedo 1995, n.41).

13. This is very much in the spirit of Rawls’ formulation: “We try, so far as we can, not to assert nor to deny any particular comprehensive religious, philosophical, or moral view, or its associated theory of truth and the status of values” (Rawls 1993, pp.55-56). Compare (Brighouse 2000, p.7) on the value that motivates liberal neutrality: “This is the aspiration that when the state resorts to coercion as it often must, it should do so without expressing disrespect for the persons coerced, and that part of what it would be to be disrespectful is to presume false the deepest commitments of the individuals against whom coercion is used.”

14. Sometimes political liberals talk as if they are most interested in avoiding the assertion of a particular comprehensive doctrine. Macedo claims that “While political liberalism cannot avoid ruling out some accounts of what has ultimate value, it does not rest on a particular comprehensive account of the truth or the good as a whole” (Macedo 1995, p.492). And cf. (Gutmann 1995, p.559): “Political liberalism[’s]…neutrality is limited to its refusal to invoke any particular conception of the (nonpolitical) good life, including individuality, autonomy, and devotion to the divine, as a grounds for justifying …liberal principles…”

But a policy that does not rely on the truth of any particular doctrine for justification may still rely on an implicit denial of some, as Macedo explicitly acknowledges. This strikes me as a problem, given the values that political liberalism purports to serve. Consider a policy that could be justified only by appeal to a general theism. The justification would not depend on any particular theological doctrine—Lutheran Christianity, Sunni Islam, or the like. But clearly it would still fail to be a justification that all reasonable people could accept. This, rather than lack of dependence on any particular doctrine, seems to me to be the most relevant feature of the policy for our purposes.

My way of interpreting the public reason requirement strikes me as in agreement with a different formulations of Macedo’s: “…the critics of political liberalism seem consistently to miss the essential point: when determining the basic shape of the awful coercive powers of the modern state, should we not try and offer our fellow citizens reasons that they ought to be able to accept without making the absurdly unreasonable demand that they first accept our convictions about the ultimate ends of human life?” (Macedo 1995, p.495). To this I would add “…or reject their own views about this?”

15. If one wants to reserve the term “respect” to designate an obligatory way of regarding and/or treating others, then one can formulate this view as holding that respect for our fellow citizens does not generally require realizing civic friendships with them. It requires only appropriately valuing this relationship, and so taking seriously the reasons that it provides.

16. In (Ebels-Duggan 2010b) I argue that civic friendship, while important, is not important enough to justify cooperation in significant injustices.

17. Many political liberals’ arguments about abortion rely on the bracketing conception and exemplify this problem. This is already present in Rawls’ own brief comments. See (Rawls 1993, p.243n) and (Rawls 2001, p.606). See also, e.g., (Freeman 2006, pp.244-250) and (Dombrowski 2001), chapter 9.

18. The shared reasons conception can be read as an interpretation of the ideal of liberal neutrality. I take it to be a version of neutrality of justification, though standard interpretations treat neutrality of justification as roughly equivalent to the bracketing conception. See, for example (Kymlicka 1989) and (Patten 2012).

19. I am interested not only in curriculum, narrowly construed, but—among other things—in the institutional organization of the school, and the norms—formal and informal—that govern classroom and other interactions. This includes what Callan describes as a school’s “hidden curriculum.” See (Callan 1997, p.168).

20. Thus when parents aim to communicate their values to their children they do not usually understand themselves as indulging a preference their own, or engaging in an act of self-expression. I think that this important point goes missing in many discussions of the possible grounds of parental rights. See, e.g., (Callan 1997, pp.132-145), (Brighouse 2000, pp.13-18) and (MacMullen 2007, pp.113-124). For a much better framing of conflicts between state and parents over educational issues see (Burtt 1996).

21. See, among others, (Macedo 1995), (Gutmann 1995), (Gutmann 1999), (Callan 1997), and (Ackerman 1980).

22. It might be thought that, in a limited way, prominent political liberals endorse a version of (4). Many think that the state need not remain neutral about everything, but can and should promote moral and civic values of tolerance and respect. (Gutmann 1999, pp.53-64) endorses a version of this. Cf. (Macedo 2003, p.237). I agree that the state has a legitimate interest in teaching these values. But, as I argue in the final section, reasonable parents will endorse these values by definition, and so seek to hand them on to their children. Any instantiation of (2) that my proposal endorses thus promotes civic respect, and in this way my proposal captures the limited appeal that (4) may have.

23. At least some parents take their obligations to include seeing to it that their children’s formal education is positively embedded in a particular normative outlook, not merely that it does not conflict with this outlook. Such parents present the clearest cases in which the state will have difficulty justifying a single approach to education to all. See (Wolterstorff 2002).

24. Some readers may be tempted by the view that it is not of the right form, because it makes reference to outcomes, here the outcome of education. This may suggest to some conceptions of neutrality widely rejected by liberal thinkers as too demanding. Among these are the idea that government action should not increase or decrease the number of adherents to a view (See (Patten 2012) on neutrality of effects), that all conceptions of the good should be burdened or benefited equally by government policy (See (Brighouse 2000) on neutrality of effects), or that all should be made equally easy to pursue (See (Brighouse 2000) on neutrality of outcomes). But the parents’ complaint does not depend on a demand for neutrality in any of these senses. Rather their central objection is that participation in this policy would violate their own parental obligations. A citizen cannot both think that supporting a policy violates an obligation that he has, and see the policy as justified. So the parents claim that the policy violates the shared reasons conception of public reason.

25. See, e.g., (Gutmann 1995), especially p. 571.

26. Cf. (Macedo 1995, p.485).

27. In an under-discussed concurring opinion in Mozert, Justice Boggs draws the distinction that I am pressing here. He chastises the majority for construing the parents’ objection as concerning mere exposure to particular beliefs that conflict with their religious commitments, and is impressed by the fact that they object to the Holt curriculum “overall.” He interprets them as holding that, taken as a whole, it tends to denigrate or oppose their religion (Mozert, p.1074), and to inculcate commitments opposed to it. He argues that we must accept their assessments that they are forbidden to submit their children to this educational program, and so regard the curriculum as requiring them to engage in conduct forbidden by their religion as a condition of receiving the benefit of public schooling.

Boggs sees the case, thus construed, as demanding an answer to the question: does the majority have a right to dictate the content of a single state-sponsored curriculum not limited by parameters set by the requirements of minority religious convictions? As he puts it, “[t]he school board recognizes no limitation on its power to require any curriculum no matter how offensive or one-sided, and to expel those who will not study it, so long as it does not violate the Establishment Clause.” (Mozert, p.1073) He concludes that there are no constitutional protections against the majority imposing its views in this significant way. But he draws this conclusion only “reluctantly,” and strongly suggests that, were he asked to declare on the question of what justice, rather than positive law, requires he would draw a different conclusion.

Late in his opinion Boggs provides an vivid illustration what he takes to be the profound implications of the court’s decision: “Contrary to the position of amicus American Jewish Committee, Jewish students may not assert a burden on their religion if their reading materials overwhelmingly provide a negative view of Jews or factual or historical issues important to Jews, so long as such materials do not assert any propositions as religious truth, or do not otherwise violate the Establishment Clause” (Mozert, p.1080). Unless we think that such a curriculum could be justified to a reasonable Jewish citizen, political liberals must share his regret that the law allows such an imposition on those who dissent from the majority viewpoint.

For another discussion of Mozert that takes Boggs opinion seriously, see (Stolzenberg 1993).

28. Like most citizens, the Mozert plaintiffs were not trained political theorists, so unsurprisingly their complaint amalgamates several different strands. One strand undeniably does seek something like the first model of education for their children. At some points in the proceedings, parents assert that simply by exposing their children to the content of religious and non-religious worldviews that conflicted with their own the school violated their religious convictions. One parent clearly asserted in courtroom testimony that there were at least some ideas in the curriculum that she regarded in this way (Mozert, p.1064).

29. They also make a third, distinct, complaint: lack of balance. This complaint seems sustained by the facts as I understand them. In a concurring opinion, one judge cites the plaintiffs’ claim that “of 47 stories referring to, or growing out of religions…only 3 were Christian, and none Protestant” (Mozert, p.1081). The judge who wrote the main opinion rejects this objection on the grounds that “Balance in the treatment of religion lies in the eye of the beholder” (Mozert, p.1065). But this might not be a reason to throw out complaints based on balance, but rather be a reason to acknowledge reasonable disagreement about what counts as balance and accommodate more pluralism in state-sponsored education. Callan believes that the complaint of lack of balance is the most interesting part of the parents’ case (Callan 1997, p.160). I will not address the balance complaint independently of the non-neutrality complaint since I think that the theoretical interest of the former collapses into that of the latter.

30. (Nagel 1987) comes closer than anyone else I am aware of to explicitly addressing problems arising from this fact. He attempts a principled account of which controversial views are to be excluded from the realm of public reason, and which controversies are rightly addressed there. His proposed solution deserves careful consideration, but—as he acknowledges—remains subject to vagueness at important points.

31. This example suggests a worry about girls raised in conservative communities where parents affirm narrow limits on the career choices that should be presented to their daughters as open possibilities. I suspect that this problem has no solution defensible in public reason. There is no education that both the parents in question and citizens concerned for the autonomy interests of the children can regard as justified. My view acknowledges the possibility of such cases, and it is beyond the scope of the paper to say how they should be addressed. Here my purpose is only to explore the limits of reasonable agreement.

32. Cohen’s contrast between preferences and obligations is again relevant here. See (Cohen 2009, pp.244, 313).

33. This is part of what motivates resistance to the exposure/inculcation distinction. Cf. Callan on the “hidden curriculum:” “The hidden curriculum comprises the attitudes, beliefs, or the like that the very fact of participating in the institution will tend to instill, and which are yet, at least typically, not consciously acknowledged by those who participate” (Callan 1997, p.168). Callan recognizes that “common schools” will inevitably have such a hidden curriculum and that it will be in tension with the commitments of some citizens in a pluralist democracy.

34. Cf. (Burtt 1996).

35 Cf. (MacMullen 2007).

36. Cf. (Wolterstorff 2002) on how an educational approach may communicate through silence that an issue is unimportant.

37. Cf. (Callan 1997, pp.169-171). Cf. (Wolterstorff 2002, pp. 194-195): “It must be noted, in the first place, that the public school in our present-day American society has no choice but to infringe to a significant degree on the religious freedom of some members of society. It cannot possibly be fully impartial and nondiscriminatory among all religions and irreligions in our contemporary society, for in our society we find a large group of people who believe, as a matter of conscience, that the education their children receive should not be set in the context of a religion or irreligion and should not incorporate religious practices. But we also find in our contemporary society large groups of people who believe, as a matter of conscience, that the education their children receive should be set in the context of a specific religion and should include the religious practices appropriate to that religion. The public school must discriminate, coercively, against the members of one or the other of these groups.”

38. It might look as if there is a third option: the classroom teacher. Some people find attractive a model on which the state enacts its neutrality towards the good by employing a diverse teaching staff who advocate various and conflicting worldviews. But the problems with model (3) surveyed above apply to this proposal as well. The state is still trying to present a variety of worldviews neutrally, though now diachronically, over the course of a student’s education, rather than synchronically in a single year. Some parents may welcome this, but some reasonable citizens will find that an education consisting in this kind of cacophony of views does not fulfill their parental obligations as they understand them. There’s no way that they can see this kind of education as justified from their own point of view. I thank Stacy Clifford for articulating this option for me.

39. They can thus answer the worry of (Gutmann 1995, p.560). There she argues that mutual toleration is not sufficient for virtuous liberal citizenship. We need, beyond this, the virtue of respect. Cf. (Neufeld and Davis 2010) and (Neufeld and Davis 2007).

40. MacMullen argues that political liberals have, in any case, no grounds for giving development of civic virtue absolute priority over other values. He claims that arguments for such priority tend to overlook the fact that the polity can remain robust even if not all of its citizens are virtuous. See (MacMullen 2007), Chapter 2. (Brighouse 2000, p.68) is also unsympathetic with according this sort of authority to a collective interest in social reproduction.

41. Indeed, on some views, this means that I have not so much as begun to justify education policy. Harry Brighouse argues that education policy decisions must always be justified primarily in terms of the interests of the children whom we seek to educate. See (Brighouse 2000), especially pp 13-18 and Chapter 3. Others think that the justifications in question should appeal to some of the interests of children, parents, and citizens. For the latter view see, e.g., (Galston 2011).

42. Cf. (Ebels-Duggan 2010a). And cf. (Burtt 1996).

43. For a similar view see (Burtt 2003).

44. Some writers suppose that (2) is especially problematic on account of the unity between the child’s home and school experiences that results. On this view, normative dissonance is good for autonomy. See (Gutmann 1999, p.69), and see Chapter 2 more generally. And cf. (Macedo 2003, pp.226-238). Also cf. (Brighouse 2006, p.22). My own view is that whether such dissonance is salutary or not depends entirely on the content of the views that are communicated to the child by the home and school respectively. Moreover, we cannot expect agreement on what this content should be in the midst of a pluralistic society.

45. For more on how to understand autonomy in educational contexts see (Ebels-Duggan forthcoming-a) and (Ebels-Duggan forthcoming-b).

46. (Callan 1997, p.58). Cf. (Brighouse 2000, p.67) and (MacMullen 2007, pp.69-73).

47. Here I have in mind something like Kantian constructivism, which attempts to derive substantive normative standards from the mere form of practical reasoning or the logic of agency. Perhaps the most ambitious attempt to do this is found in (Korsgaard 1996) and (Korsgaard 2009).

48. See, e.g., (Callan 1997, p.54). Others are clearer. MacMullen holds that it is inevitable, and not cause for regret, that non-rational factors will play a role in determining which normative outlooks we will find compelling. He sees no reason why attachments to our parents and culture of origin formed early in our development are not suited to play this determining role. (MacMullen 2007, pp.77-80). It is interesting to contrast Clayton, who agrees that reason cannot settle normative issues, but believes that the remaining indeterminacy ought to be by individual desires and dispositions. See (Clayton 2006), Chapter 3. Brighouse also denies that reason is sufficient to determine a particular way to live. He accepts the possibility of a determinative role for the parental culture but worries that one’s individual dispositions may make this culture a bad fit, so one needs to be given other possibilities. See (Brighouse 2000), Chapter 4 and cf. (Brighouse 2006), Chapter 3.

49. Cf. (MacMullen 2007, p.23) and (Clayton 2006, pp.11-13),

50. Here I mean to suggest this as a constitutive, conceptual claim about autonomy: to be autonomous is (perhaps among other things) to have alternatives. I take (Clayton 2006) to be endorsing this view when he speaks of the “environmental” conditions of autonomy, pp 11-13 and cf. p.89.

51. This also seems to be what Callan has in mind when he characterizes the autonomous person this way: “…[He] learns…seriously to examine the basic commitments of his life. These are no longer taken for granted as the virtual fate of someone born and reared in particular circumstances; he now sees them as revisable elements within one way of life among others, to be embraced or set aside according to the verdict of his own reason” (Callan 1997, p.52). Cf. (MacMullen 2007, pp.81-86), (Clayton 2006, pp.11-13), (Brighouse 2006, p.14), (Macedo 2003, p.237), and (Nussbaum 1997), chapters 1-2.

52. E.g., (MacMullen 2007, p.76).

53. It is tempting to think that political liberals can circumvent this disagreement by appeal to the idea of the reasonable. As we’ve seen political liberalism rests on a distinction between reasonable and unreasonable disagreements, and the assertion that the norms for dealing with these differ. Some theorists try to employ these same categories to arrive at a determinate view about the commitments on which education should encourage critical distance. They hold that children are entitled to an education that allows them to gain distance on just those commitments over which there is reasonable disagreement. This apparently allows one to agree that aiming to inculcate an effectively unrevisable commitment to racial egalitarianism does not violate a child’s autonomy, because disagreement about this value is unreasonable. But, on this view, aiming to nurture the same sort of commitment to, say, a certain religious outlook does violate autonomy, because religious disagreement falls within the scope of the reasonable.

This argument relies on an important confusion. The distinction between the reasonable and the unreasonable that figures in political liberalism is tailored to define the line between those views that the state has reason to treat as one among many equally good ways to live and those that it does not. It is a very different matter to suggest that we should encourage individuals, including our children, to regard these same ways of life as live options, real possibilities for themselves. To think this is not to transcend the reasonable disagreement about which normative commitments should be unrevisable, but to take a substantive and controversial stand on this disagreement. Moreover, it is not a substantive stand that anyone should find attractive. If Rawls’ grass counter seeks cooperation in political relationships, we should treat him as reasonable for political purposes. But this should not commit us to encouraging anyone to regard emulating him as a real possibility for their own lives. The interpretation of political liberalism that I defend in the first section makes the mistake especially easy to see. On my reading, the view requires respect for, in the first instance, people, not their views. It is our fellow citizens whom we identify as reasonable, or not, depending on their willingness to seek cooperative relationships with us. The respect due to particular normative outlooks is strictly derivative from the fact that respecting our fellow citizens is a matter of treating them in ways that they can understand as justified, and this in turn requires not enacting policies that could be justified only if their value commitments were false or bad. But this yields only a view about what to treat as reasonable in politics, not further views about which value commitments we can sensibly encourage people to treat as up for review.