2. Protecting Fundamental Choices

It is often assumed that discrimination law exists to protect only those traits or characteristics which one has no choice about. With increasing societal awareness of race- and sex-based inequality, and sensitivity to the impact of discrimination in sustaining that inequality, the salience of these characteristics gives natural support for this assumption. Plainly enough, in including race and sex, discrimination law protects some characteristics which one has no choice about, and, further, were the assumption true, political belief would thereby be excluded. But it is nonetheless false, for discrimination law also protects some characteristics which are chosen. As a result, a disjunctive ground for legal recognition has been recognised, so that a characteristic may be a matter either of ‘immutable status’, as race and sex are standardly understood to be, or one of ‘fundamental choice’ (Wintemute Reference Wintemute1995; Khaitan Reference Khaitan2015, 50).

The legal recognition of fundamental choice is particularly important for the characteristics of religion and sexual orientation. For some, these traits remain fixed throughout their lives, while for others, they are mutable. Individuals may change their religious practices and affiliations or sexual behaviours, often reflecting shifts in underlying attitudes and dispositions. Regardless of whether changes to these underlying attitudes and dispositions are voluntary or involuntary – and they may be either – their expression through behaviour and practice is typically a matter of choice, and including these traits in discrimination law protects individuals’ ability to act in accordance with them. Religion and sexual orientation are not the only characteristics protected under the fundamental choice test rather than that of immutability. Tarun Khaitan cites approvingly legal cases that extend protection to pregnancy, marital status, and citizenship (2015, 60). Considering the UK’s Equality Act 2010, its primary source of anti-discrimination law, a majority of protected characteristics (five of nine) are better classified as matters of fundamental choice than as immutable.Footnote ² Recognizing a characteristic as a matter of fundamental choice does not deny that, for many, the way that they instantiate the ‘universal order ground’, for example, of religion or sexual orientation, in its ‘particular order ground’, in terms of a specific religious commitment or orientation, is involuntary (adopting Khaitan’s useful terminology; 2015, 31). Rather, it affirms that individuals should have the right to make choices in these domains without being subject to disadvantage imposed by others.

Including a characteristic in the protectorate is a legal response to the claim that, as Moreau describes it, that characteristic should be ‘normatively extraneous’: that is, people ‘should not have to factor into their deliberations … the broader social costs of having that trait’ (2010, 149-150, also 2020, 84ff.; cf Khaitan Reference Khaitan2015, 56 on ‘normative irrelevance’, so that ‘our possession of these grounds should not affect how successful our lives are’). For Moreau, the point of discrimination law is thus to secure deliberative freedom, so that people can make decisions without having to take account of their normatively extraneous features, such as skin colour or sex; we may say that one’s decision making should be ‘unburdened’ by having such traits. And the proper effect of discrimination law is, indeed, to render specified characteristics extraneous in this way. It achieves this by the combination of the various reasons for action that it creates: some comply with discrimination law because it specifies how one should comply with an acknowledged but imperfect moral duty not to discriminate; some comply with it because they recognise a general duty to comply with the law as such; others do so out of prudence, fearing legal or social sanction; and those who are the victims of others’ unlawful activity have, through the courts and regulatory bodies, means of redress and compensation. Collectively, these factors reduce the severity of harm due to discrimination and its incidence. As Khaitan observes, the requirement that characteristics must be either immutable or a matter of fundamental choice is puzzling if viewed from a choice-based perspective, for the law simultaneously protects characteristics over which one has no choice and those where choice is crucial, which ‘appears contradictory’ (2015, 60). The apparent contradiction is dissolved, however, by recognising that the disjunctive requirement represents a disjunctive justification. Immutable characteristics are normatively extraneous because individuals should not be disadvantaged for traits they cannot change and which are morally irrelevant. But what justifies treating some domains of fundamental choice as similarly extraneous, to the extent that they warrant protection under discrimination law?

Khaitan’s answer to this question is brief. ‘[T]he choice in question is important because it is fundamental to the person whose choice it is’ making its exercise ‘positively valuable, rather than merely not-immoral’ (2015, 60; italics original). This is dubious, however. Even as a necessary condition – where a choice must be fundamental to the individual to warrant protection under discrimination law – the claim is implausible. Would it be a valid legal defence that it was not particularly important to the complainant that she was religious or gay, despite all parties agreeing that she was treated disadvantageously on either of those grounds? Surely not. Not only is the fact that an individual perceives a choice to be fundamental not necessary for the relevant domain of choice to be properly included in the protectorate, it is also not sufficient. Many choices are deeply significant to a few but are too idiosyncratic to warrant inclusion, whether it be preferences for international travel, tattooing, Dungeons and Dragons role-playing, and so on. If fundamentality alone justified protection, no characteristic would be excluded from the protectorate.

While Khaitan here draws on John Gardner’s work, Gardner’s own position does not have such expansive implications. Explicitly inspired by Joseph Raz’s perfectionist political morality, Gardner avers that the basis of the fundamental choice test is ‘the familiar liberal ideal of an autonomous life. This is the ideal of a life substantially lived through the successive valuable choices of the person who lives it, where valuable choices are choices from among an adequate range of valuable options’ (1998, 170; see his fn 8 on Raz). The Razian nature of Gardner’s argument is further reinforced by his brief discussion of pregnancy. ‘It is because pregnancy is so worthwhile that choosing it should be a possibility for all who can, in principle, choose it. … [So] it should not be effectively ruled out as an option by prohibitive costs like the destruction of other aspects of women’s lives’. This argument ‘depends on the value of choosing as well as the value of pregnancy’ (1998, 171, fn 9; italics original). It is not just the value of worthwhile options which gives reason to ensure that they are available for people to choose from, but that the act of choosing from worthwhile options is also, itself, valuable. Since the task of the state is to promote valuable lives, it must also protect individuals from discrimination that would penalise the exercise of autonomy. This Razian justification of the fundamental choice test provides a principled basis for determining which characteristics should be protected as matters of fundamental choice – namely, when it is valuable that individuals should be able to choose, of a universal order, what their particular order ground should be.

The Razian justification is unsatisfactory. It still fails to solve the problem of over-inclusion; while it restricts some putative grounds from the protectorate because they are subjectively important to a small number of individuals but not objectively significant, the range of valuable choices available to an autonomous individual is vast. So, it still permits a considerable expansion of the protectorate. A more significant qualification, however, is that it is needlessly committed to a controversial political morality. Take, again, religious belief and affiliation. The Razian liberal may affirm that a variety of religious traditions all present valuable options and that the exercise of choice between these options is itself valuable. But religious doctrines may deny exactly this: it is not difficult to imagine some which are properly part of a liberal society but view rival religious traditions as disvaluable, affirming instead that (by their lights) those others are heretical, idolatrous, or similar, and regret that there is so wide a choice between different ways of being in error. Yet, adherents of such religions may nonetheless support a principle of religious non-discrimination, perhaps principally on grounds of religious freedom. It is true that, in prohibiting disadvantageous treatment because of a specified characteristic, discrimination law requires that people should be treated equally, regardless of differences across that trait. In this sense, it may be said to assert, expressively, that those who possess a given particular order ground are the equals of those who possess a different particular ground (Anderson and Pildes Reference Anderson and Pildes2000, 1533–45). But the kind of equality it asserts is civic equality only – equality in the public sphere, with that sphere specified as per the social contexts subject to that body of law – and not all-things-considered equality.

Discrimination law need not imply a commitment as to the objective value of a given particular order ground, nor to that of exercising one’s choice in a domain specified by a universal order ground, as a means of promoting autonomy. That there is reason to protect individuals’ ability to choose, all things considered and in a relevant domain, may be justified on a quite different basis, and there is no reason to assume that a single criterion uniquely determines the boundaries of the protectorate. As Moreau remarks, of the search for a single criterion which would determine whether a characteristic should be included, ‘All that we can say is that each ground reflects a judgment that people have a right to make decisions in these social contexts in a manner that is insulated from the burdens imposed by these traits, or by other people’s assumptions about them. The normative facts about these traits that form the basis for such a judgment will be diverse’ (2010, 157). Moreau’s ‘deliberative freedoms’ account, which I endorse, is thus thin. It correctly identifies that anti-discrimination law protects individuals’ freedom to deliberate without being burdened by either their possession of a given characteristic or a fundamental choice they have made. And it asserts that, if the violation of a right to make unburdened decisions or the risk of its being violated constitutes social ills of sufficient severity, that justifies including a characteristic in the protectorate. Even if no such moral or natural right to unburdened decision making exists, the existence of a significant social harm arising from such burdened decision making may justify establishing such a right in law. But it is thin because, for any properly protected characteristic, it does not say why that deliberative freedom matters.Footnote ³ I now turn to discharge the resulting argumentative burden, showing why this deliberative freedom matters for political affiliation and belief – that is, why a right of unburdened political deliberation and decision making justifies including political belief in the protectorate of anti-discrimination law.

3. Unburdened Political Agency as a Democratic Right

By disadvantaging individuals based on their political affiliations or beliefs, political discrimination imposes costs on political activity and restricts (negative) freedom. How might this happen? Imagine a society in which two parties, the Freedom Party and the Equality Party, compete for power. But the competition has turned nasty: the historic ideological foundations of each party have faded, and partisan affiliation and identification are now more often driven by hostility toward the opposing leadership and increasingly its supporters. Passionately committed to distributive equality, Simona knows that to advocate publicly for higher marginal tax rates will enduringly associate her with the Equality Party. With the party’s leader having once been a communist whose populist rhetoric has led to a spate of ‘eat the rich’ graffiti, negative stereotypes of its members now abound. Opposing partisans perceive it as the party for those motivated by envy and spite who, consumed by self-loathing and irrational guilt, hate the society which has created the conditions for their own material prosperity, which they intend to destroy. Simona works in finance, where most of her colleagues do not just disagree with her preferred policies, but also subscribe, to varying degrees, to these negative stereotypes. Those are reinforced by the news sources standard in the sector, and a workplace culture shaped by off-hand remarks at the water cooler and in the margins of meetings.

In this scenario, should Simona join the Equality Party, advocate online for higher marginal tax rates, or go on protest marches for the same? It depends on her career ambitions. If success in finance is her priority, it would be prudent for her not to join that party, nor do anything that might lead to her being identified as supportive of its aims. Suppose that the sector is highly competitive, with tens and sometimes hundreds of nearly equally qualified people applying for jobs and positions; all it takes is for one member of a committee to express reservations about a candidate for that person to be unlikely to be appointed or promoted. It may be further supposed that success within the sector often comes from ad hoc invitations to join teams that are working on cutting-edge opportunities; belonging to informal, trusted networks is essential to receive and credibly make these invitations. Simona risks all this by being public about her political beliefs and being active in their pursuit. While most of her colleagues might be people of good faith, able to separate political disagreement from professional cooperation and workplace friendships, some are still affected by unconscious bias primed by negative stereotypes. Making her political beliefs public would, in others’ minds, prime in them an association for her with some severely negative stereotypes. And, some of her colleagues are not fair-minded and may actively seek to penalise her for her political beliefs. While it is difficult to know when and what costs might be imposed on her, that some costs will be is likely. She is vulnerable, in expectation, to political discrimination.

Simona’s freedom is restricted because she lacks what Matt Kramer calls ‘conjunctively exercisable opportunities’ (2003, 38–40; also, Carter Reference Carter1999, 180–82). Freedom can be denied outright if an option is physically blocked; one is not free to travel along a road if a concrete barrier has been placed across it. But freedom can also be diminished without physical obstruction. There may be no direct obstacle to Simona travelling along the road, but her freedom is much reduced if getting caught will result in a year in prison. Severe threats are not the only way that freedom may be compromised, however; some difficulty in exercising an option, which requires resources to overcome, also limits it. Introducing a toll barrier on the road means that someone with only £10 must choose between travel and a fish and chip supper. One’s freedom is diminished when travelling by that road, and having a fish and chip supper is no longer conjunctively exercisable. On Kramer’s useful elucidation of ‘pure negative’ liberty, freedom thus consists in the sets of conjunctively exercisable options. And it can be diminished without being eliminated when the number of those sets reduces, but is not eliminated. The more valuable the set of conjunctively exercisable options, the more valuable one’s freedom. In Simona’s case, the risk of political discrimination diminishes her freedom because she cannot simultaneously engage as a citizen in political activity while also pursuing her career in finance unhindered. She can choose one, but not both.

Political discrimination thus poses a threat, and an obvious one, to some central ‘operating principles’ for democracy. At the heart of normative democratic theory is the claim that each citizen has an equal right to participate in the governance of their polity. In a society of equals, political power must be distributed equally (Kolodny Reference Kolodny2014a, Reference Kolodny2014b). While free and fair elections are the most evident institutional expression of this claim, the essential additional freedoms are those of freedom of association and of belief and expression. (These imply the corollary freedoms of assembly and of the press.) Political agency requires more than just the right to vote. Citizens must also be free to organise, advocate and persuade others in their pursuit of political objectives; only in a context of such political activity are elections meaningful events. The freedoms of association and belief thus recur in the various constitutional or otherwise foundational legal documents that establish democratic polities because only with such freedoms are all guaranteed that they can exercise, as of right, the political agency that belongs to the citizens of a democracy. Deliberative theories of democracy give these freedoms particular importance. While elections aggregate the views of the public and appoint its representatives, it is only with the deliberative processes that precede and succeed elections that democratic rule is exercised legitimately (Habermas Reference Habermas1996; Cohen Reference Cohen, Hamlin and Pettit1989), with this deliberation essential for ensuring that it is the people that govern, not merely elites (Landemore Reference Landemore2020). Thus, deliberative democracy is viable only if the freedoms of association and belief are real and protected. More generally, these freedoms are core elements of what one may call the ‘operating principles’ of democracy, and they are so central that it is a test for any proposed normative justification of democratic governance that it should explain their significance.

In the example, Simona’s right to free political association and expression is compromised. Although she is legally free to associate politically with the Equality Party and advocate for its policies, she cannot do so without cost; her political decision making is burdened. While Simona’s freedom as a democratic citizen is not denied, it is restricted, and she has an interest, as a democratic citizen, in being able conjunctively to pursue her career unhindered and to engage in the wider process of political deliberation and contestation core to democratic politics. What is true for Simona is true generally. For democratic citizens to enjoy their right to affiliate and express their political views in an unburdened way, costs must not be imposed on them by others in public and quasi-public interactions because of their political activity. This right is violated by political discrimination. When members are at risk of being disadvantaged because of their political affiliation and beliefs, so, in accordance with the costs in expectation of such discrimination, their decision making is burdened. The risk of such discrimination is itself a restriction on their freedom. Yet the free exercise of such political agency, without such costs, is essential to any polity which affirms the right of every member to equal participation in governance – in short, to democracy itself.

To see how essential it is that citizens should be able to engage in political activity without fear of penalties, consider the consequences if such penalties, or the fear of them, were widespread. Other things equal, citizens would decline to join parties, campaign in elections, attend meetings, or advocate publicly for policies or reforms according to the risk that they are exposed to in their personal affairs for so doing. Society would likely divide into a smaller group of members who are ‘out’ politically, perhaps because their jobs are explicitly political, they have nothing to lose, or there is ideological homogeneity in their social and professional networks, and a larger group who keep their views private to avoid professional risks. Such a polity could still hold elections, and political debate might appear vibrant. But politics would have become a spectator sport, something from which much of the citizenry is kept from participating in, and not by any restriction imposed by the government but by citizens’ mutual hostility towards others’ political activism. Alternatively, such a society might ‘balkanise’ into mutually antagonistic, politically defined spheres, where there is solidarity within but suspicion and hostility across them. To prevent such outcomes, citizens must be assured that they will not be penalised for political participation. While free elections require secret ballots to prevent retaliation for voting choices, deliberative democracy demands forms of political participation that cannot be concealed, being essentially public. So, other legal remedies are required to prevent others from imposing costs on someone for the exercise of their political agency. Discrimination law serves precisely this function: it prohibits adverse treatment on matters of fundamental choice, and provides remedies for those treated. Just as it protects religious affiliation and belief, so it is apt for protecting political affiliation and belief.

That political non-discrimination is required because only so do citizens enjoy the freedom of association and belief required for full political participation is a democratic argument for including political belief in the protectorate, and not a liberal one. It may be thought that a liberal argument for the same conclusion is available, with that argument proceeding as follows. If the liberal ideal familiar to Gardner is the strong, perfectionist one of the autonomous individual making a succession of valuable choices, there is an alternative liberal ideal, no less familiar and intentionally weaker, which is the attempt for those with contradictory and competing conceptions of the good to find fair terms of cooperation. This latter, anti-perfectionist liberalism provides an alternative basis for including some matters of fundamental choice within the protectorate, most notably religion. On this view, religion is included in the protectorate of discrimination law because it violates fair terms of cooperation for others to impose costs on a religious adherent, disadvantaging her in matters of employment, marketplace transactions and so on because of her religious commitments, so that religious non-discrimination contributes to religious freedom.Footnote ⁴ But freedom of religion is really a shorthand for the right to live and act in accordance with one’s comprehensive conception of the good, and while religious commitments have been the most salient form of expression of this historically, there are principled reasons to include political commitments too in these.

Most obviously, many religious traditions are directive not just for their adherents’ lives in terms of ideals of character, conduct, and relationship, but also for their political engagement. That is, they have a political theology, which is viewed as enjoying the same justificatory support as the ethics which that tradition promotes, being based on a set of metaphysical and axiological doctrines, whether those are revealed or otherwise identified. With political affiliation and belief thus often part of someone’s comprehensive conception of the good, as significant doctrinally as the metaphysics of religion, so the same reasons that justify including religious belief in the protectorate also justify including political belief. As further support for this conclusion, it is plausible that, for many people, political affiliation has become as important a basis of their self-conception as religion (Hoffman Reference Hoffman2011, 1536). All this would provide an alternative, liberal argument for including political affiliation and belief in the protectorate.

While this liberal argument is available in principle, it offers a much narrower scope of protection for political affiliation and belief than that derived from the democratic argument. The liberal argument justifies protection only for political beliefs that function like religious beliefs in shaping an individual’s self-conception. However, there is much political activity and engagement that is not foundational to someone’s self-conception or their conception of the good in the way that, by contrast, religion is, so not qualifying for protection on this basis, and yet is essential to the normal operating of a democracy. Rawls himself marks this difference by distinguishing between two moral powers – the capacity for a sense of justice, and that for a conception of the good – with the principles that should govern political life derived by the exercise of the former only (2001, 18–19). An extensionally similar distinction is marked by UK discrimination law, in which the relevant protected characteristic of ‘religion and belief’ explicitly allows that some non-religious and otherwise philosophical beliefs may enjoy the same protection as religious beliefs (Equality Act 2010, sec. 10). The characteristic is interpreted by five criteria that enable courts to determine whether a non-religious belief qualifies as a ‘Protected Philosophical Belief’, and which have the effect of protecting non-religious beliefs that are functionally equivalent in someone’s life to religious belief. Known collectively as the ‘Grainger test’, the ruling which gave rise to the criteria allowed that some political philosophies might qualify, but, nonetheless, specifically excluding ‘support of a political party’ from protection.Footnote ⁵ The liberal argument provides for patchy protection only for political affiliation and belief, while the democratic one justifies protection for political affiliation and belief as such.

4. Political Polarisation

If protection from political discrimination is a prerequisite for a well-functioning democratic public sphere, one may ask why there has been no significant public or academic discussion of including political belief in the protectorate. The primary reason, I suggest, is that political priorities have been elsewhere; in the domestic jurisdictions in which discrimination law has been most developed since the 1960s, combating the social ills of racism and sexism has been more pressing. Although the protectorate has expanded as awareness of unequal treatment of disparate groups in society has grown, and the utility of discrimination law has become evident, political discrimination has generally not been perceived as a significant problem, nor has protection from it seemed necessary.Footnote ⁶

This political assessment may be buttressed by a principle of legal parsimony. Given the social costs of law’s operation – including public and private resources spent on litigation, compliance, and associated bureaucracy – new laws should be introduced only when there is some widespread public harm, and when the costs of inaction exceed those of intervention. Larry Alexander applies this principle to discrimination law. ‘Discriminatory preferences are extrinsically morally wrong if their social costs are large relative to the costs of eliminating or frustrating them. And if a discriminatory preference is morally wrong – and if there is no moral right that protects its exercise – then there is a case for legally prohibiting its exercise if the costs of legal prohibition and enforcement are low relative to the social gains to be achieved.’ (1992, 219; also, Epstein Reference Epstein1992, Reference Epstein1995). For the sake of argument, suppose this principle of legal parsimony is correct. This imposes an empirical condition on the normative claim defended by this paper. Granted that people should not be treated disadvantageously because of their political beliefs, in the public and quasi-public contexts subject to discrimination law, is this nonetheless happening? What concrete social harm would be addressed by making political discrimination unlawful?

While the incidence of political discrimination may not have met this threshold historically, the current landscape of political polarisation alters the assessment. Political polarisation fosters the social conditions in which political discrimination may become widespread, and with polarisation intensifying in many countries, this empirical condition is increasingly met. The harms arising from such polarisation are both individual and societal. The individual harm that would be directly addressed by the proposal is the constrained political agency widely experienced under conditions of political polarisation. And, depending on relevant causal relations, including political belief in the protectorate, may also help to ameliorate its societal harms indirectly.

The largest body of evidence of political polarisation comes from the USA where, alongside the policy disagreement and ideological competition which is a sign of a healthy democracy, political scientists have in more recent years focused on some more concerning phenomena, referred to variously as ‘affective polarisation’ (Iyengar et al. Reference Iyengar, Lelkes, Levendusky, Malhotra and Westwood2019), ‘social polarisation’ (Mason Reference Mason2018), or ‘political sectarianism’ (Finkel et al. Reference Finkel, Bail, Cikara, Ditto, Iyengar, Klar, Mason, McGrath, Nyhan, Rand, Skitka, Tucker, Van Bavel, Wang and Druckman2020). Evidence indicates that Americans increasingly dislike ‘or even loathe’ both the party they oppose and those partisans who identify with that opposed party (Iyengar et al. Reference Iyengar, Sood and Lelkes2012, 405), so that by 2016, negative sentiments towards the opposing party had replaced positive sentiments towards one’s own party as the main driver of political participation (Iyengar and Krupenkin Reference Iyengar and Krupenkin2018, 213).

The immediate effect of political polarisation is to increase the incidence of political discrimination. The causal link is clear: political polarisation consists in a prevalence of negative attitudes towards opposing partisans, and political discrimination occurs when people act on such negative attitudes. A plethora of recent work documents this effect. Partisans may favour co-partisans, or penalise opposing partisans, in settings as diverse as academia (reviews of papers and grant applications, and hiring, Inbar and Lammers Reference Inbar and Lammers2012, Honeycutt and Freberg Reference Honeycutt and Freberg2017, Peters et al. Reference Peters, Honeycutt, De Block and Jussim2020; college admissions, Munro et al. Reference Munro, Lasane and Leary2010; scholarship allocations, Iyengar and Westwood Reference Iyengar and Westwood2015; grading, Rom and Musgrave Reference Rom and Musgrave2014), criminal prosecutions (Gordon Reference Gordon2009), the marketplace (hiring, Gift and Gift Reference Gift and Gift2015, Roth et al. Reference Roth, Thatcher, Bobko, Matthews, Ellingson and Goldberg2020; one-off contracts, McConnell et al. Reference McConnell, Margolit, Malhotra and Levendusky2018), employee performance evaluations (Patel Reference Patel2023), behaviour in economics games in the laboratory (Iyengar and Westwood Reference Iyengar and Westwood2015, Lelkes and Westwood Reference Lelkes and Westwood2017), residential decisions (Carlson and Gimpel Reference Carlson and Gimpel2019), charitable giving (Kazemian Reference Kazemian2023), dating choices (Huber and Malhotra Reference Huber and Malhotra2017, Nicholson et al. Reference Nicolson, Coe, Emory and Song2016), and friendship preferences (Iyengar et al. Reference Iyengar, Konitzer and Tedin2018). Partisan prejudice may affect decisions about life and death, with people willing to withhold COVID-19 vaccines from opposing partisans (Stoetzer et al. Reference Stoetzer, Munzert, Lowe, Çali, Gohdes, Helbling, Maxwell and Traunmüller2023). For present purposes, there is no meaningful difference between favouring a co-partisan versus penalising an opposing partisan. By conferring an advantage on co-partisans, a disadvantage is thereby conferred on all others; it is a descriptive question whether one sprinter has a 5-second head-start, or everyone else has a 5-second penalty.

While it is difficult to estimate the prevalence and impact of political discrimination, one test is to compare the willingness to engage in it with the willingness to engage in other forms of discrimination for which there is a well-recognised sense of its significance. Shanto Iyengar and Sean Westwood collect implicit, explicit, and behavioural measures of partisan affect and compare them with affect based on racial identity, ‘the most salient social divide in American society’. Their striking finding, from data gathered in 2012-13, is that ‘implicit affect and behavioural discrimination based on partisanship are just as significant as affect and discrimination based on race’ (2015, 703, 690). Westwood et al. extended the result: across not only the USA, but also the UK, Belgium, and Spain, ‘partisans discriminate against their opponents to a degree that exceeds discrimination against members of religious, linguistic, ethnic or regional out-groups’ (2018, 333). By 2018, in the USA, ‘Out-group denigration based on race, religion, or gender does not match the level of animus directed at opposing partisans’ (West and Iyengar Reference West and Iyengar2022, 810).

All this evidence suggests that the case of Simona given earlier, while hypothetical in its details, accurately reflects broader patterns. In the USA – the country for which there is the most robust data – although there is variation in the degree to which this is true, it is prudent for many individuals to avoid publicly identifying their political views or partisan sympathies. Their professional lives will go better if they avoid political engagement. To state the obvious, this is not a sign of a healthy civic sphere. And, there is a principled basis for presuming political discrimination to be prevalent according to the degree of political polarisation in a given country. As well as those countries identified earlier in which political discrimination exceeds religious, linguistic, ethnic, or regional discrimination, there is evidence of polarisation having increased across other OECD countries including Switzerland, France, Denmark, Canada, and New Zealand, while data indicates that affective polarisation is higher in Greece, Portugal, and Spain than in the USA. Non-OECD countries are less well represented in the research, although there is some work on Latin American countries, for instance (respectively, Boxell et al. Reference Boxell, Gentzkow and Shapiro2022; Gidron et al. Reference Gidron, Adams and Horne2020, 7; and Bergman and Fernández Reference Bergman and Fernández2025). In all those countries which exhibit signs of significant political polarisation, making political discrimination unlawful would protect individuals from the personal harm of being disadvantaged due to their political affiliation and belief. And, crucially, it would restore to them the freedoms of association and belief, so that they may exercise their democratic rights of political agency unhindered.

Further investigation is needed to determine whether making political discrimination unlawful would help to combat the societal harms caused by political polarisation, by reducing polarisation itself. Recent evidence suggests that polarisation has deeply destabilising societal and political effects. Polarised voting publics have attitudes supportive of democratic backsliding: ‘The other side cheats, so our side would be foolish to adhere to long-standing democratic norms’ (Finkel et al. Reference Finkel, Bail, Cikara, Ditto, Iyengar, Klar, Mason, McGrath, Nyhan, Rand, Skitka, Tucker, Van Bavel, Wang and Druckman2020, 535). Braley et al. identify this as giving rise to a ‘subversion dilemma’. If voters are led to believe that they are competing with an opposition who will renege on democratic procedures, even those who value democracy may feel rationally compelled to ‘save democracy’, to support leaders who advocate non-democratic means (2023). Alongside democratic backsliding is a potential link between political polarisation and political violence (Kalmoe and Mason Reference Kalmoe and Mason2022; Kleinfeld Reference Kleinfeld2023). Could prohibiting political discrimination help to reduce polarisation or ameliorate its harmful effects? This is an empirical question, but there are plausible causal pathways worth investigating. If political discrimination is not only a symptom of polarisation, but also a cause, then other things equal, reducing its incidence would ameliorate the latter’s severity. The causal relationship is intuitive – those who are the victims of political discrimination, or suspect they are, likely feel anger and resentment at their disadvantageous treatment, leading to increased animus towards opposing partisans. Another causal pathway might be through the impact of discrimination law on social norms. It has been suggested that political discrimination is prevalent in part because there are no social norms which constrain animus towards the partisan out-group (Iyengar and Westwood Reference Iyengar and Westwood2015, 690; Bougher Reference Bougher2017, 732; McConnell et al. Reference McConnell, Margolit, Malhotra and Levendusky2018, 7; but see Shafranek Reference Shafranek2020, 45). Introducing political belief into the protectorate could be the crucial intervention that would catalyse the emergence of such a norm.

If a causal link were to exist between including political belief in the protectorate and reducing political polarisation, this would provide the crucial premise for a separate normative argument in favour of the reform. In endorsing the principle of legal parsimony, Alexander advances the conjunctive claim that there is a case for legally prohibiting discrimination if discriminatory preferences have large social costs and they are morally wrong. This is certainly plausible. But it is compossible with the weaker claim, which is independently grounded, that there is a case for legally prohibiting those discriminatory preferences which have large social costs – so long as there is no moral right that protects their exercise and that the costs of legal prohibition and enforcement are low relative to the social gains, as before – even if there is no moral wrong in the preference as such. Given this weaker claim, and assuming that prohibiting political discrimination would ameliorate the societal harms of polarisation, there would then be a supplementary reason to expand the protectorate to include political belief. Notably, such an argument would be independent of the democratic one I have defended, on which the constraining effects on individuals’ political agency of political discrimination justify prohibiting it. That political discrimination should be unlawful is plausibly, argumentatively over-determined.

The principle of legal parsimony was granted earlier for the sake of argument. But is it sound? As a general constraint on legal innovation, I am sympathetic to the idea, but in this case, there are principled reasons to reject it. Some freedoms are so fundamental that individuals have a claim to legal protection of the relevant rights, regardless of how often that right is violated by others. Suppose the incidence of murder fell to such low rates that each violation of someone’s right to life was an astonishing surprise. Should laws concerning murder be abolished because of the social costs of their enforcement? Surely not. But this example is not yet decisive. The freedoms of association and belief, which I described as core ‘operating principles’ of democracy, are unlike the right to life, in that many people are politically disengaged and still live flourishing lives. That these freedoms should be legally protected regardless of how often they are violated does not, therefore, turn on their (absolute) value to the individual; rather, it turns on the value to the polity of their being securely enjoyed by each person. Each violation of these rights does not merely reduce the freedom of the individual involved; everyone’s freedom is impaired, because they do not securely enjoy the relevant rights with the ‘backstop’ provided by law. And a democratic polity has a core interest in ensuring that these rights are enjoyed securely. As indicative support, one way to construe the purpose of constitutional and otherwise foundational legal documents is as articulating those rights whose enjoyment must be secure in this way; and, as noted, the freedoms of association and belief recur widely in these. (The same point applies to religious freedom.Footnote ⁷ This also provides an answer to those who might be concerned that expanding discrimination law to include political affiliation and belief would result in the ‘proliferation of the protectorate’, in a phrase from Owen Fiss (Reference Fiss1974, 751), with no basis for not also including, for instance, people who have red hair. Those documents reflect a considered and widely shared judgment that these freedoms are appropriately core.) That such rights may be violated infrequently is no argument for removing the protections which ensure that they are enjoyed securely.

5. Objections

I conclude by replying to some of the most important objections.