Without Exemptions: Reconciling Equality with the Accommodation of Diversity

When generally applicable rules clash with one’s cultural, religious or moral commitments, should exemptions be granted? The debate on exemptions raises the question both of what it means to treat people equally and of what it means to protect diversity adequately. The objective of this paper is to defend the no-exemption argument and to make it a more attractive position for liberals. I first argue that exemptions violate the principle of equal treatment because they rely on distinctions that cannot be neutrally justified. I then argue that diversity can be adequately accommodated if we use a more demanding interpretation of the justification of generally applicable rules. There should be no exemption to fully justified rules, but rules are only fully justified when the particular demands that they impose on individuals are justified, i.e. when they are either neutral or pass both a Necessity Test and a Sufficiency Test. When rules are not fully justified, they should be repealed or modified. Based on this focus on the demands of rules, equality can be reconciled with an adequate accommodation of diversity.


Introduction
Rules are there to get people to do things that they might not have done without them. There is sometimes a significant conflict between what a rule requires and what (some) individuals want to do. The legal system usually gives priority to the former over the latter. Is this unfair? Are there cases in which what individuals desire should have a priority over what rules require that they do? When a generally applicable rule clashes with religious or cultural commitments, should an exemption 1 3 be granted? Sikhs, for instance, have asked to be exempted from the obligation to wear a helmet when riding a motorcycle, to allow them to wear a turban. On the one hand, those who defend exemptions argue that they are necessary for an adequate accommodation of diversity. Without an exemption, Sikhs would have to choose between wearing a turban and riding a motorcycle. On the other hand, those who oppose exemptions argue that exemptions entail a violation of equal treatment. The states that grant an exemption to Sikhs have not granted an exemption to bikers who would prefer to not wear a helmet and who might therefore feel that they are not treated equally. At the heart of the debate, then, are the questions of both what it means to treat people equally and what it means to protect diversity adequately.
My objective in this paper is to defend the no-exemption argument and to make it a more attractive position for liberals. I first argue that the pro-exemption position should be rejected because it violates equal treatment (1). I then argue that if the no-exemption position is based on a more demanding interpretation of the justification of rules, then it can adequately protect cultural and religious diversity without exemptions (2). This requires that we examine the particular demands that generally applicable rules impose on individuals. Next, I show that this allows us to accommodate diversity without exemptions and therefore without violating equal treatment (3). Finally, I respond to two possible objections that could be raised against my argument (4).

Exemptions Versus Equality
Those who defend exemptions do not deny the importance of generally applicable rules, but they believe that in specific cases, the reasons to have the same rule for all are outweighed by others. We may have good reasons, for instance, to give priority to what people want to do, because it is crucial to their moral integrity (Bou-Habib 2006;Laborde 2017;Seglow 2017). Consequently, they defend what Barry has called the rule-and-exemption approach (Barry 2001, p. 33): the rule itself should be upheld, but it should simply not apply to some.
Although exemptions are usually justified as a way to re-establish equality, they can also be opposed on the grounds that they introduce unequal treatment: granting an exemption means treating people differently based on the kind of cultural, religious or moral commitments that they have.
The problem is not exemptions in themselves. All kinds of exemptions exist in our legal systems and play a necessary role to make laws sensitive to important differences. For instance, numerous exemptions are made for medical reasons, and rightly so. The problem is with what Simon May has called 'volitional exemptions', i.e. exemptions that are 'granted on the basis of an individual's objection to complying with the law' (May 2018, p. 59 Of course, nobody has ever argued that objecting to a rule is enough to be exempted from that rule. This would defeat the purpose of rules entirely. The kind of objections that are considered as potentially justifying an exemption are those that come from 'core or meaning-giving beliefs and commitments' (Maclure and Taylor 2011, pp. 12-13), or 'integrity-protecting commitments' (Laborde 2017, pp. 203-204). What makes these commitments potential candidates for objections is not simply that people object to a rule, but that they object for specific reasons: obeying the rule would mean renouncing living their life as they believe, in their heart of hearts, that it should be lived. Let us call this the integrity argument. As Laborde explains, 'these commitments cannot be sacrificed without feelings of remorse, shame, or guilt, by contrast to preferences, which can ' (2017, p. 204). Cultural, religious and moral commitments are considered as paradigmatic examples of such integrity-protecting commitments. What is protected through exemptions, then, is this interest that people have in being able to 'act in accordance with [their] perceived duties' (Bou-Habib 2006, p. 117).
Equal treatment means that those who are in situations that are similar in the relevant way should receive the same rights and benefits. When different things are treated differently, then, there is no violation of equal treatment. Those who are immunocompromised have a right to be exempted from mandatory vaccination, but those who are healthy do not. Organizations pursuing charitable or public aims do not have to pay taxes, for-profit organizations do. Those who defend the rule-andexemption approach argue that those who are granted an exemption are in a different situation than the others because there is a relevant distinction between deep commitments and superficial preferences.
The problem, I argue, is that the distinction unfairly privileges deep commitments which are not always more meaningful to people than so-called superficial preferences. Those who want to engage in a particular conduct based on hedonist grounds, for instance, would be denied an exemption that might be granted for others to engage in the same conduct based on their conscientious commitments (Fornaroli 2019, p. 665). Some people have cultural or religious commitments that are not crucial to their moral integrity. Others have preferences that play an essential role in their understanding of who they are, even though they could 'be sacrificed without feelings of remorse, shame, or guilt'. Not everyone understands their conception of the good life in terms of moral integrity. But just because something is not perceived as a duty does not mean that it can be easily renounced. The opposite is also true: just because something is perceived as a duty does not mean that not doing it would be a great sacrifice. We sometimes fail to do what we believe that we have a duty to do, because we are too lazy or because there is something else that we just enjoy doing more. The priority given to integrity can only be justified if integrity matters more than other considerations for the well-being of individuals. This is, however, a controversial claim that the liberal neutral state cannot rely on to distinguish between people, and therefore differential treatment based on such considerations would be a violation of equal treatment.
The unequal treatment entailed by the distinction between core commitments and superficial preferences can be illustrated by the 1994 case of Nof v. The State of Israel: During the first Gulf War (1990), the state of Israel made gas masks available to its citizens. In this context, a bearded man called Akiva Nof petitioned the Israeli Supreme Court. He argued that the state's denial of his request to receive a free gas mask specially adjusted for bearded men was a violation of his rights. Such masks are more than double the price of regular masks, and they were given free of charge to Ultra-Orthodox Jews, who grow beards for religious reasons. Nof argued that once one financial exemption had been granted, a refusal to grant it to him as well violated his right to equal treatment. (Perez 2009, p. 201) Israel's distinction between religious and non-religious commitments would be considered as unfair by those defending the integrity argument. If anyone believed that they had a moral obligation, based on secular commitments, to wear a beard, then they should also be exempted. This is, however, not the case for Akiva Nof. Nof's beard is not based on religious reasons, but it is not based on secular deep meaning-giving commitments either. At no point does he claim that his integrity is at stake or that he has a moral duty to wear a beard. The integrity argument, then, could not justify an exemption to Nof.
The court's decision in favor of Nof is in line with my argument that a distinction between integrity and non-integrity is just as unfair as the distinction between religion and non-religion. Instead of the integrity argument, Justice Mazza, writing the majority opinion, relies on what can be called the importance argument: In my opinion, a person's right to grow a beard is a part of his human dignity, irrespective of his beliefs or religious convictions. (…) a beard often forms part of one's self-image and very identity, especially to one who has worn it for years.
The relevant and fair distinction is between things that really matter to the individual and things that do not. Because what is relevant here is what is considered as important by the individual, i.e. based on the individual's own preferences, the importance argument does not face the same objection as the integrity argument: it does not single out any specific value or project as particularly valuable. The importance argument is therefore compatible with the idea that certain things can be extremely important to individuals even though they are not central to their integrity and even though they are not based on deep or meaning-giving commitments. Based on the importance argument, then, those core commitments that are not really important to people would not be favored over preferences that are. Mazza argues that there is no relevant distinction between those who grow a beard for religious reasons (or, one could add, for similar secular reasons) and those who grow a beard as part of their 'lifestyle'.
The importance argument sounds compelling in theory, but it raises considerable problems in practice. The fairness of granting an exemption to some but not to others relies not only on the assessment of the importance of preferences for different individuals, but on the comparison between them. This comparison, I argue, cannot be made reliably and fairly.
The issue of interpersonal comparisons is a notoriously difficult one (Elster and Roemer 1991). The satisfaction of preferences, which considers the importance of preferences to the individual, is often used to measure levels of well-being but economists and philosophers disagree on whether such comparisons make sense. 2 To compare two cases, we need some kind of metric. Let us say, for instance, that the level of importance of a preference can be graded from 1 to 10. The individual grades could then be compared: if the preference for wearing a beard is a 10 for Nof as well as for Ultra-Orthodox Jews, then we know that they should be treated equally. But who can say which level the preference has? There are only two options: the grade is given subjectively by the individuals themselves, or it is given objectively by an independent observer. The results of the subjective strategy would be unreliable: individuals might have irrational preferences, preferences might change over time and the grading of preferences might say more about how uncompromising or flexible people are than about the actual importance of their preferences. In fact, even those who believe that interpersonal comparisons are possible argue that it should not be based on subjective evaluation (Elster & Roemer 1991, pp. 6-7). 3 The results of the objective strategy, on the other hand, would be unfair: like the integrity argument, they would prioritize certain types of preferences over others. Objective measures that make interpersonal comparisons possible are necessarily based on value judgments (Scanlon 1991), and such value judgments should be avoided in a neutral liberal state.
Mazza suggests such a problematic objective measure when he argues that the state could distinguish 'between bearded men whose beard forms part of their identity and way of life (…) and other bearded men, for whom the growing of a beard was a recent initiative, taken for reasons of aesthetics or of convenience for a limited period of time'. How long someone has been doing something, he suggests, tells us how important this is to them. Someone who would have decided recently to grow a beard would therefore not be exempted. But although it matters if a practice is longstanding, it cannot be the decisive feature to draw the distinction. There might be men who have been wearing beards for decades who do not believe that the beard is important to their self-image, and there might be men who recently decided to grow a beard who do.
The importance argument might correctly identify the relevant distinction between things that matter and things that do not, but it would be too difficult to operationalize. Not only are the results of the different measuring strategies either unreliable or unfair, but it is also unlikely that such strategies could be applied in large-scale societies. Even if we could agree that Nof and Ultra-Orthodox Jews have a similarly strong preference for wearing a beard, the state does not have the time or resources to compare the preferences of all beard-wearing persons to decide who else deserves an exemption. Exemptions are only compatible with equal treatment if we can distinguish between those with an important preference and those with a 1 3 superficial preference. Since it seems unlikely that such a distinction can be made accurately in practice, then any exemption is likely to treat at least some people unequally.
One last strategy to avoid this issue would be to reject the idea that the focus should be on the individual preferences and instead argue that exemptions should be understood as group-differentiated rights. The identity of groups can be associated with specific commitments. This is particularly true for religious groups but could also apply to certain cultural or political groups (as opposed to, say, geographical groups or age groups). The kind of intercultural dialogue defended by Parekh (2006, pp. 268-273) could then provide a way to identify such commitments and to assess their weight for different groups. The advantage of such a strategy is clear: it is easier to compare the preferences of a very small number of groups than those of a very large number of individuals. Exemptions would then be granted based on group membership rather than on individual preferences. The potential unfairness of this strategy is also clear: it would grant an exemption to certain group-members who might not actually need it and it would not grant an exemption to those who might need it but do not belong to the right groups. There is no group of bearded men that Nof could join and that could then be included in intercultural dialogue. Such strategy can only work if the commitments associated with certain religious and cultural groups are considered as more protection-worthy than similarly important commitments that are not-and this would be an unfair distinction, which would then lead to unequal treatment.
By comparison to the pro-exemption position, the no-exemption position maintains a non-negotiable commitment to the principle of equality. The no-exemption position is best expressed in Brian Barry's words: 'either the case for the law (or some version of it) is strong enough to rule out exemptions, or the case that can be made for exemptions is strong enough to suggest that there should be no law anyway' (Barry 2001, p. 39). There is no risk of misidentifying where the line between legitimate and illegitimate cases of exemptions should be, because no exemptions should be granted as long as the generally applicable rule is justified. The only issue for no-exemption theorists, then, is to identify the conditions under which a generally applicable rule can be considered as justified. The approach adopted by these theorists is apply-or-repeal: either the rule, or a modified version of the rule, should apply to everyone, or it should apply to no one.
Refusing the rule-and-exemption approach does not mean that the ways in which a rule might clash with the projects or commitments of individuals cannot be taken into consideration. What it does mean is that these are taken into consideration ex ante, when evaluating the justification of the rule, rather than ex post, when evaluating the justification of an exemption. This is a fundamental principle of liberalism: restrictions on individual liberty are possible, but only when they can be justified. The more a rule infringes upon an individual's freedom, the stronger the justification must be. For instance, making military service mandatory for all individuals of a certain age significantly limits their freedom. Whether they have a moral objection or not, many of them prefer to spend their time doing something else. Mandatory military service clashes with many important interests and commitments that individuals can reasonably have. The reasons justifying mandatory military service must be strong enough to hold in the face of such interests and commitments. Prohibiting littering, on the other hand, is a fairly trivial limitation of individual freedom. For this reason, the threshold for justification of the rule is much lower than in the previous case. No-exemption theorists do not deny the importance of religious, cultural or moral commitments. They deny, however, that the mere fact that a generally applicable rule clashes with one's religious, cultural or moral commitments means that there is a prima facie case for an exemption. Burdens must be justified, and the required strength of the justification depends partly on the heaviness of the burden. However, if the burdens are appropriately justified, then no exemption should be granted to anyone.
There is something very appealing about the no-exemption position. First, it invites individuals to bear the costs for their own beliefs (Jones 1994). If the beliefs people hold happen to conflict with a generally applicable rule (e.g. the religious obligation to wear a turban and the legal obligation to wear a helmet when riding a motorcycle), then they have three options: they can decide to not obey the rule and take the risk of being sanctioned for it (e.g. get a fine because they rode a motorcycle without a helmet); they can put themselves in a position where the rule does not apply to them (e.g. not ride a motorcycle); or they can amend their beliefs so that they are compatible with the rule (e.g. accept that religious duties should only apply when they are not illegal or potentially harmful for oneself or for others). The noexemption position encourages people to reconcile their beliefs and practices with generally applicable rules, whereas the pro-exemption position encourages people to overstate the incompatibility of their beliefs and practices with these rules.
Second, and more importantly, the no-exemption position fits better with the egalitarian and democratic principle that the law should be the same for all and should not unfairly privilege certain projects or commitments over others. No distinction is made between individuals based on the religious, cultural or moral commitments that they have. The no-exemption position does take into account the importance of commitments or interests in the evaluation of possible liberty-limiting rules, but it faces neither the difficult question of identifying what should count as a religious commitment (Sullivan 2005) and of what should be done with parodic religions (Martin 2020), nor the problem of justifying the distinction between moral and nonmoral projects (May 2017). Whether the rule is repealed, applied or modified, it is the same for all. The no-exemption position, then, is more egalitarian than the proexemption position.
Pro-exemption theorists reject the idea that exemptions are a violation of equality, and actually defend the use of exemptions precisely as a form of equal treatment (Parekh 1998). This argument, however, is based on the assumption that not granting an exemption would in itself entail unequal opportunities for some individuals due to their cultural or religious commitments. In other words, critics of the no-exemption position have objected that it does not go far enough to accommodate cultural and religious diversity and in particular that it does not protect religious freedom adequately: the no-exemption position is 'too stingy in its approach to accommodating religion' (Patten 2017, p. 205). This is the traditional multiculturalist objection: difference-blind policies are problematic precisely because they ignore important differences between groups (Taylor 1994). The inability of the difference-blind approach to take cultural particularities into consideration leads to unfair treatment as there are, critics have argued, many neutrally justified rules that burden some religious believers more than other people. One of the cases that critics often use is dress code. This is for instance what Cécile Laborde writes, as an objection against Dworkin's no-exemption argument: What if an organization-say, the police force-has a policy that requires all members to wear a suitable uniform, and be bare-headed and clean-shaven?
The policy clearly has a neutral, non-discriminatory justification. (Laborde 2017, p. 48) If there are neutral, non-discriminatory justifications for requiring motorcycle drivers to wear a helmet, for requiring police officers to wear a uniform or for prohibiting cruel slaughter practices, then it seems that the no-exemption position must accept that some justified rules, despite their non-discriminatory justification, can have a discriminatory effect, and that this does not amount to unfair treatment. Does this mean that the no-exemption position fails to adequately protect religious freedom?
Pro-exemption theorists argue that the kind of equality before the law defended by no-exemption theorists fails to deliver an adequate accommodation of cultural and religious diversity. This is the objection that I want to challenge in this paper: I believe that the commitment to equal and neutral laws and the absence of exemptions is compatible with a fair treatment of cultural and religious minorities. The noexemption position, I contend, can be made much more accommodating of cultural and religious diversity than its critics assume by using a more demanding interpretation of what it means for a rule to be justified.

Focusing on the Demands of Generally Applicable Rules
When considering whether a particular rule is justified, the focus tends to be on the reasons that support this rule: does it aim at a generally shared objective? Is it based on a public reason? What counts as an appropriate, public reason remains highly contested, but the idea that reasons matter is itself compelling.
Reasons do matter, and I believe the principle of public justification is a sound one (Bardon 2018). But an exclusive focus on reasons is misguided. For a rule to be fully justified, it is also crucial to examine the particular demands that it imposes on individuals. These demands should be either neutral (1) or both necessary and proportional (2).
(1) Demands are neutral when they do not make it easier or harder for some people, compared to others, to engage in whatever activity they want to engage in based on their conception of the good. 4 Certain activities, such as wearing a turban or not eating pork, are required by people's conceptions of the good. Any policy that prohibits, regulates or otherwise directly or indirectly affects the possibility to engage in such an activity, then, imposes non-neutral demands. Some activities are not closely related to any controversial conception of the good, and therefore if they are prohibited or imposed, the particular demands of these policies would be neutral. Not being allowed to violate the basic rights of others, or having to drive on the right side of the road are neutral in that sense, as the demands themselves are not associated with any (permissible) controversial conception of the good. 5 Most policies, however, impose non-neutral demands. For instance, taxation is a justified policy but its demands are far from neutral, since it affects those whose conception of the good life depends on the accumulation of individual wealth more than others.
It is important to note that demands cannot be considered as neutral once and for all. In some cases, the demands of a rule can be neutral in a certain situation, and later become non-neutral (Balint 2015, p. 502). This is particularly likely to happen when new individuals join a society, bringing with them cultural or religious practices that belong to their conceptions of the good and that did not exist in that society previously. The demand for policemen to be clean-shaven is an example of this: until Sikhs and Muslims joined, the demands of the uniform policy were neutral. 6 (2) Demands that are themselves not neutral, i.e. that make it easier or harder for some people, compared to others, to engage in whatever activity they want to engage in based on their conception of the good, can still be justified if two conditions are met: (i) the demands are necessary for the achievement of the objective of the rule, and (ii) the objective of the rule is important enough to justify a restriction of individual liberty. The two conditions must be met in order for non-neutral demands to be considered as justified, and therefore for the rule itself to be justified. Both conditions are assessed by looking at the relation between the demands and the objective of the rule.

(i) The Necessity Test
The particular demands of a rule are necessary if it would not be possible to achieve the same objective with different, but maybe equally weighty, demands. This expresses the basic liberal idea that the least restrictive way to achieve any objective should always be favored. Compulsory education is justified in that sense. Although it does privilege some controversial conceptions of the good over others, it is necessary in order to guarantee to all children the right to an open future and to provide them with the basic skills necessary to develop their own conceptions of the good life. This objective being intrinsically related to education, it could not possibly be achieved through any other means. In this case, the first condition for the justification of demands is met.
(ii) The Sufficiency Test Objectives can be more or less important, restrictions of individual liberty can be more or less severe, and it is necessary to assess the importance of the rule in light of the burdens that it imposes on individuals. The particular demands of a rule are justified if the importance of the rule outweighs the severity of the burden. 7 Compulsory education will be seen as a significant restriction of the liberty of parents to decide how they want to raise their children. The objective is, however, sufficient to justify this heavy burden, and therefore the second condition for the justification of demands is also met in this case.
Evaluating the burdens placed on individuals requires assessing the importance of preferences. When a preference is very important to an individual, not being able to satisfy it creates a heavy burden. When the preference is very superficial, not being able to satisfy it creates a minor inconvenience. It is crucial that we take into consideration the experiences and testimonies of individuals to assess their preferences, but there are also objective components to this evaluation. For instance, social mobilization that aims at explaining how much some practice matters to people or the readiness of any individual to face extra burdens in the hope of satisfying a preference should be considered. In this context, Parekh's intercultural dialogue could provide a useful opportunity for minorities to explain their preferences. These assessments might be difficult and controversial but, crucially, they do not require the use of any precise metric since no comparison is necessary. There is no need to know how much a preference for wearing a beard matters for every single individual in society. To evaluate the burdens of any law that would prohibit beards, it is sufficient that we know that it is something that strongly matters for at least some individuals.
The Sufficiency Test always implies a balancing between the demands and the objective, which is why the same objective can be sufficient in one case but not in another. Consider the example of physical activity, which is known to be beneficial for individuals' health. In a liberal state, public health is a public interest, which can justify some policies. It is, for instance, sufficient to justify state-supported campaigns praising the advantages of physical activity, as well as mandatory physical activity for all children in school, but it is not sufficient to justify the obligation for all individuals to practice one hour of physical activity daily, in which case the objective would be outweighed by the heaviness of the burden.
The focus on the relation between the demands and the objective of the rule has been largely ignored in the literature on public justification. Yet this is something that legislatures and courts have paid serious attention to in dealing with cases of restrictions of human rights, in particular religious freedom cases. In the US, both the Necessity Test and the Sufficiency Test are identified in the 1993 Religious Freedom Restoration Act, which states that the restriction of religious freedom can only be legitimate provided it '(1) furthers a compelling governmental interest; and (2) is the least restrictive means of furthering that compelling governmental interest' (H.R. 1308). 8 Based on the European Convention of Human Rights, only restrictions of rights that are 'necessary in a democratic society' (ECHR,Art. 8,9,10,11) for the protection of public interests can be legitimate. In order to assess this relation between the demands of the policy and the importance of the public interests that it aims to protect, the European Court of Human Rights uses a proportionality test that has recently relied more and more on the idea of the least restrictive means (Brems and Lavrysen 2015;Gerards 2013).
objective. For instance, although the police uniform cannot be designed in a fully justifiable way, it is essential for the police to have a uniform. In this case, the uniform should include a choice between wearing a turban, a hijab, a kippa or some kind of headgear that is not associated with any specific conception of the good, such as a cap. Any member of the police force should then be allowed to choose any of these options. Likewise, vegetarian meals or kosher meals in prisons, schools or hospitals should be available to everyone.
Modifications of the demands have implications in terms of costs, as it is more costly to provide several options than it is to provide a single one: several uniforms for the police need to be designed and made, several menus should be prepared, etc. Limiting the non-neutral dimensions of policies and adequately protecting diversity and individual liberty has a price. This should be taken into consideration when deciding how particular demands should be changed, and how the costs will be supported and distributed. In any case, the egalitarian principle should also apply here: the costs should be borne by the individuals themselves or by the state, but the costs should be not borne by the state for members of certain groups only.
What I propose, then, is accommodation of diversity without exemptions. Accommodation can be defined as a response to a situation in which a certain religious or cultural group is 'disadvantaged compared with others' (Jones 2015, p. 544) by a specific rule. This can be done by lifting generally applicable rules for certain individuals, in the case of an exemption, or by modifying these rules so that the conflict with religious or cultural commitments is diminished or eliminated.
The distinctive feature of the position defended in this paper is that accommodation should be done without differential treatment. Religious and cultural accommodation is usually thought of as implying group-differentiated rights, in which case there is differential treatment. For instance, only those who have a religious objection to eating meat are offered a vegetarian alternative menu. Whenever accommodation is done through differential treatment, it violates equal treatment just like exemptions. 9 But this does not have to be the case: generally applicable rules can be modified in response to concerns about the religious freedom of specific individuals or groups, and the new modified rule can still apply to everyone in the exact same way. It is equal treatment since it applies to all, irrespective of the specific cultural or religious commitments that people have, but it can still be described as accommodation since it remains 'difference-sensitive rather than difference-blind' (Jones 2015, p. 545). Whatever modifications of the rule are required, then, should be valid for all.
This argument overlaps strongly with the claims recently defended by Balint (2015Balint ( , 2017. I share his conclusions that 'neutrality can be made sufficiently and fairly accommodating ' (2017, 60) and that exemptions are therefore not necessary. There are, however, two main differences between Balint's arguments and mine. First, Balint focuses on existing conceptions of neutrality-of justification, intent and outcome (2015, pp. 498-499)-which I believe are insufficient. Instead, I have argued that we should look at the specific demands that generally applicable rules are imposing on individuals. The demands-centered conception of neutrality might be compatible with Balint's position, but it does suggest a different way to explain what is non-neutral about certain rules. Second, Balint wants to 'be neutral between mere preferences and deeply held beliefs ' (2015, p. 506). Although I agree that differences between preferences are difficult to assess and should therefore not be used as a basis for differential treatment, I believe that such assessments are overall justified and necessary in the making of generally applicable rules. Balint claims that 'the question is not whose beliefs are more serious, but whether the mainstream practice that is being challenged can be (…) justified neutrally' (2017, p. 8): on my account, knowing how serious the individual preferences are is crucial, and it is part and parcel of how we can know if the rule is neutrally justified since we need to be able to evaluate the burdens that the rule imposes on individuals to determine whether its demands are neutrally justified.
One important upshot of my argument is that demands can only be justified by reference to a particular objective in the context of a particular rule. This means that the same required behavior can be justified for one rule but not for another. Consider the requirement to be clean-shaven for specific jobs. This requirement has unequal effects: it burdens those who want to wear a beard more than those who do not. But can the requirement be fully justified? Consider the cases of John and Jim, who both want to wear a beard for religious reasons. John is a policeman, and as part of the police uniform he has to be clean-shaven. Jim works at a nuclear power plant, he does not have to wear a uniform but he has to be clean-shaven for safety reasons: in case of emergency, Jim will have to wear a respiratory mask, and those masks can only work properly on people who are clean-shaven. In both cases, there is a shared public objective and the regulations are supported by neutral reasons. In both cases, the way in which John and Jim are treated is the same: the requirement to be cleanshaven accommodates some conceptions of the good (those that do not require that men wear a beard) more than others (those that do), and is therefore likely to make the former more successful than the latter. These two cases are, however, significantly different because even though the demands are the same, they are justified in one case but not in the other.
In the case of John, the requirement to be clean-shaven does not pass the necessity test: whatever is achieved by the existence of the uniform, could be equally achieved if the uniform said nothing about facial hair, or if in fact it required all men to wear a beard. In the case of Jim, on the other hand, the removal of facial hair is necessary to guarantee that respiratory masks will work: the safety of employees of nuclear plants could not be achieved by different means. Provided the safety of employees is considered as important enough to justify the requirement to be clean-shaven, which sounds reasonable, this means that the requirement to be clean-shaven should be modified or repealed in the case of John, but that it should be applied in the case of Jim. Applied to well-known examples in the literature on religious freedom, this means that policemen, Sikhs or not, should be allowed to wear a turban, but that nobody, Sikh or not, should be allowed to ride a motorcycle without a helmet. Only in this latter case are the demands of the rule, i.e. the obligation to wear a helmet, necessary for the achievement of the objective. 10

Two Objections Against Accommodating Diversity Without Exemptions
Two objections can be raised against my argument. First, it might be argued that the focus on demands is still too stingy. After all, some rules are fully justified (they are supported by good, public reasons, and the demands pass the necessity and the sufficiency tests) yet they still make it impossible for certain cultural or religious groups to engage in practices that are key to their identity. The 1990 US Supreme Court case Employment Division v. Smith illustrates this point: the Oregon prohibition on the use of drugs was held to be a 'neutral law of general applicability' by the Supreme Court. This decision has often been criticized as an excessive restriction of the rights of members of the Native American Church to ingest peyote, which they consider as a religious practice (Eisgruber and Sager 2007, pp. 78-79, 92-93). Two different responses can be offered. The first is to bite the bullet: if the demands are considered as both necessary and sufficient, then the application of the rule should be general, and the rights of Native Americans are limited, but in a justifiable way. There are many instances in which religious practice is restricted because of justified general rules, and this might simply be one of them. The second possible response is to challenge the claim, made by the State of Oregon as well as by Scalia in the majority opinion, that the rule is fully justified, either because drugs should not be restricted in general, or because the criminal prohibition on the use of drugs should not include peyote, which is a hallucinogen, with effects that are fairly limited in time as well as in seriousness. The two responses are compatible with my position: either the Oregon rule is fully justified and no exemption should be granted, or it is not and the rule should be repealed.
Second, it might be objected that, in cases in which the demands are necessarily arbitrary, such as dress codes, my argument leads to 'anything goes' and ends up accommodating too much, making it impossible to distinguish religious headgear from clown hats (Laborde 2017, pp. 199-200). How can I justify that the list of possible headgears is limited to a few options, and does not extend to whatever people choose to put on their heads? Because my argument only justifies the accommodation of practices that deeply matter to individuals (such as religious headgear), not the accommodation of more superficial preferences (such as clown hats). The Sufficiency Test entails a balancing approach that considers the importance of particular practices in order to evaluate the severity of the burdens. Not all preferences are given the same weight and, it follows, not all preferences should be accommodated.

3
Without Exemptions: Reconciling Equality with the… To have one's preferred headgear included in the list, one must make the case that the preference is a serious one, that it matters greatly for one's understanding of the good life, for instance by showing that it is central to one's integrity. Those who wear a hijab or turban for religious reasons have a weightier preference than those who wear a clown hat for fun or those who wear a fedora for aesthetic reasons, and that explains why hijab and turban, but not clown hat and fedora, should be added to the list of possible headgears. 11 This response to the second objection, however, might make it look like my argument is collapsing into the pro-exemption position. This is, however, not the case because, unlike the rule-and-exemption approach, I also argue that the reasons that justify the existence of a right can be different from the reasons that explain the uses of that right. One needs to show the importance of wearing a hijab or a turban to justify that these are included in the uniform; but once this has been acknowledged, one does not need to demonstrate the importance of the headgear to be allowed to choose to wear a hijab or a turban as part of the uniform. The difference that it makes in the case of dress regulations seems pointless since, admittedly, only Muslims and Sikhs will want to have these options. But the difference matters in other cases, such as the provision of food in places such as schools, prisons or hospitals: once a non-pork option exists, justified by religious dietary requirements, that option should be offered to all, whatever their reasons might be to prefer it over the pork option, including if they simply happen to not like pork.

Conclusion
Do liberal philosophers have to choose between the principle of equal treatment and an appropriate accommodation of diversity? The way the debate on exemptions is framed often suggests that they do. In this paper, I have argued that this need not be the case. The pro-exemption position necessarily violates the principle of equal treatment, as it entails that rules that should be generally applicable in fact do not apply to some individuals, although the differential treatment cannot be reasonably justified. It is, however, possible to combine the no-exemption position, committed to the egalitarian principle of having the same rules apply to everyone, with an adequate protection of individual freedom and an adequate accommodation of cultural and religious diversity. What this requires is an understanding of the full justification of rules that takes into account the demands that rules impose on individuals, and how these demands relate to the objectives at which the rules are aiming.
With this focus on demands, rules that pursue a shared, public objective and that are based on good, public reasons can still fail to be fully justified because the demands that they impose on individuals are themselves not justified. Rules that are not fully justified should be either modified or repealed.
The evaluation of the particular demands of rules is a complex and necessarily contextual task. It involves balancing the importance of the objective of the rule with the severity of the burdens. It involves making judgments regarding how valuable it might be for people to be free to engage in the particular practice in question. It involves examining the potential costs of changing the demands, and making decisions regarding whether these costs are justified and how they should be distributed. All of these assessments will be based on controversial claims. But this is how politics works and how rules are written. The point of this paper is simply that, whatever decisions we make regarding the evaluation of these demands, either the demands are considered as fully justified, in which case the rule applies to all; or the demands are not fully justified, in which case the rule should be rewritten for all, or should be simply dropped. This difference-sensitive conception of the justification of rules reconciles the principle of equality before the law with an adequate accommodation of cultural and religious diversity.