Paternalistic Affirmative Action

Typical examples of affirmative action involve introducing gender quotas in boards of directors, reserving spots for qualified minority students at universities, or giving preference to an applicant belonging to a designated social group when two or more applicants have equal qualifications for employment or admission to education. More generally, Anderson defines affirmative action as ‘any policy that aims to increase the participation of a disadvantaged social group in mainstream institutions, either through “outreach” (targeting the group for publicity and invitations to participate) or “preference” (using group membership as criteria for selecting participants)’ (Anderson Reference Anderson2010, 135).Footnote ^2, Footnote ³ Interestingly, and seemingly unnoticed in the literature on affirmative action, such initiatives will often be paternalistic towards those beneficiaries who prefer to be without the relevant initiatives. Before we show this, we will define paternalism and elaborate on the elements of the definition.

Standard definitions of paternalism include three elements: interference, lack of consent, and a benevolent motive. An example of a so-called standard definition is the following: A acts paternalistically towards B if and only if:

1. (I) A interferes with B

2. (II) A circumvents or overrides B's agency or judgement, and

3. (III) A is motivated by promoting B's interests, good, or well-being.

In the literature on paternalism, many adopt a broad understanding of what counts as ‘interference’ (cf. condition I). According to Scoccia, interference involves any ‘means other than rational persuasion’ (see also Hausman and Welch Reference Hausman and Welch2010, 128–9; Midtgaard Reference Midtgaard2016; Scoccia Reference Scoccia2008, 352).Footnote ⁴ Hanna focuses on intervention and describes it as ‘any way in which one might involve oneself in another person's decisions, behaviour, or sphere of agency’ (Hanna Reference Hanna2018, 4). Shiffrin argues that A may act paternalistically even when A aims to avoid an effect on B or B's sphere of legitimate agency (Shiffrin Reference Shiffrin2000, 218).

We adopt a broad understanding of interference, leaving room for a wide range of intuitively paternalistic actions that would not be considered paternalistic on a narrower understanding (see also Hanna Reference Hanna2018, 4, n6). Consider Shiffrin's example where A does not help B build a set of shelves because A thinks it would be better for B to build them themself. However, even such omissions seem paternalistic (Shiffrin Reference Shiffrin2000, 213, 218). Dworkin provides another example of ‘a wife who hides her sleeping pills so that her potentially suicidal husband cannot use them’ (Dworkin Reference Dworkin2017). According to Dworkin, the wife seems to act paternalistically, but he grants that this may not be captured by his narrower understanding of interference (ibidem). Other examples of broadly interfering acts and policies include earmarked subsidies (for example, food stamps) or nudges using ‘flaws in human decision-making to get individuals to choose one alternative rather than another’ (Hausman and Welch Reference Hausman and Welch2010, 128).

A second feature commonly considered necessary for an action to be paternalistic is that the paternalistic agent somehow circumvents or overrides the agency or judgement of the person interfered with (cf. condition II). This is the case if A interferes with B against B's will. However, as Groll (Reference Groll2012, 707) has argued, even if A acts in accordance with B's will, A may still treat B paternalistically if A does not treat B's will as authoritative in matters concerning B's interests or well-being. For example, if A acts on the basis of balancing B's will and considerations for B's own good, A does not treat B's will as authoritative (as a trump (cf. Groll Reference Groll2012, 701) in the relevant matter. In other words, there is a relevant sense in which A bypasses B's agency or judgement.Footnote ⁵

Finally, for A's act to be paternalistic, it must be the case that A is motivated by promoting B's interests, good, or well-being. Often, A will be motivated by different reasons, where the paternalistic reason appealing to B's interests, good, or well-being figures among other reasons, such as appealing to financial considerations. Even in such ‘mixed cases’, A acts paternalistically if the paternalistic reason ‘plays the role as a reason’ in favour of the interfering act (see Grill Reference Grill and Schramme2015).Footnote ⁶ For an action to be non-paternalistic, the paternalistic reason should be excluded from the motivation behind the interfering act (see, for example, Enoch Reference Enoch2016, 44–5; Groll Reference Groll2012).

We will argue that affirmative action policies ‘using group membership as criteria for selecting participants’ (Anderson Reference Anderson2010, 135) will often satisfy the conditions of the above definition. First, affirmative action involves ‘means other than rational persuasion’, and these means are ‘aimed to … have an effect on’ the intended beneficiaries. Moreover, the ‘means’ deprives its targets of the opportunity to apply for a seat, position, education, etc. without the advantage associated with the selection criteria. In this way, affirmative action removes an option that the intended beneficiaries would have had without the relevant policy. This applies even if some potential applicants could never get the seat, position, etc. without affirmative action. For these reasons, affirmative action involves measures of a kind that are potentially paternalistic (depending on whether conditions (II) and (III) are satisfied).

It might be argued that if most disadvantaged group members welcome affirmative action, such policies would not be paternalistic. Grill (Reference Grill, Hanna and Grill2018, 55) describes this as ‘the willing majority view’. However, as Grill puts it, if the conditions of paternalism are satisfied for A but not for B and C, ‘it is not clear what additional information is conveyed by saying that the action itself is or is not paternalistic’. It is, after all, still an unwelcome benevolent interference with A. Similarly, even if only a minority of black adults oppose a certain affirmative action policy, the policy seems paternalistic towards this minority.

However, the larger the proportion of people who support the policy, the more it seems likely that the motive behind the policy is to cater to the interests of the willing majority (and not the interests of the unwilling minority).Footnote ⁷ In this way, the proportion of the intended beneficiaries opposing the policy seems relevant to whether the policy likely satisfies the third condition of the definition of paternalism. On the other hand, as Grill (Reference Grill, Hanna and Grill2018, 55) argues, another problem with the willing majority view is that this view ‘disregards the reasons for why people consent to interference with a group to which they belong … Proponents tend to assume that the consenters consent out of self-interest, but they may instead be altruistically motivated, consenting for the sake of the non-consenters, whom they see as failing to act in their own best interests.’ A policy where majority consent is given on this basis does not avoid charges of paternalism. In the context of affirmative action, some of the intended beneficiaries may support the measure because they want to benefit other group members (including those who do not welcome the measure).

Whether affirmative action's intended beneficiaries oppose the policies is an empirical issue. According to Swain (Reference Swain, Mitchell, Wilson and Smelser2001, 335), going over a range of surveys shows that ‘[b]lacks are by no means enthusiastic supporters of racial preferences, and in recent years have given affirmative action a less than ringing endorsement’. Recent public opinion surveys provide similar results. For example, a 2022 Washington Post-Schar School poll of US adults asked: ‘Would you support or oppose the Supreme Court banning colleges and universities from considering a student's race and ethnicity when making decisions about student admissions?’: 47 per cent of Black, 60 per cent of Hispanic, 65 per cent of Asian/Pacific Islanders, and 66 per cent of White adults indicated support (Anderson et al. Reference Anderson2022). Moreover, a 2022 Pew Research Center Poll shows that 59 per cent of Black, 68 per cent of Hispanic, 63 per cent of Asian, and 79 per cent of White adults respond that race or ethnicity should not be a factor when colleges and universities make decisions about student admissions (Pew Research 2022). Since people's attitudes are highly sensitive to question framing, context, and policy measures, there is reason to be cautious when interpreting such findings (Swain Reference Swain, Mitchell, Wilson and Smelser2001, 335). Still, the above studies support the claim that many potential beneficiaries oppose specific affirmative action policies.

Affirmative action, as explained, ‘aims to increase the participation of a disadvantaged social group in mainstream institutions’ (Anderson Reference Anderson2010, 135). There are good reasons to believe that this aim is often based on an underlying aim of promoting the interests of the group members (tellingly, often described as the ‘beneficiaries’). Interestingly, in his examination of discrimination law, Khaitan (Reference Khaitan2015, 222) writes: ‘The raison d’être of affirmative action measures is to benefit members of protected groups. If a measure is likely to result in net harm to these groups rather than a net gain, it will have no leg to stand on.’ In our argument, however, we will grant that it is possible to aim at increasing the participation of disadvantaged groups without aiming to benefit the group members. Some possible motives for increasing the participation of disadvantaged groups are:

1. (1) Benevolence: concerns for promoting the disadvantaged group members' interests, good, or well-being.

2. (2) Equity: the view that we are all equally entitled to participating in mainstream public institutions regardless of the socially salient group we belong to.

3. (3) Representation as a norm: the view that it is valuable if mainstream institutions represent the surrounding society.

4. (4) Productivity: the view that diversity positively affects productivity.

5. (5) Reputation: considerations by mainstream institutions to avoid sanctions or other adverse reactions from the surrounding society (for example, from clients, consumers or voters) due to lack of diversity.Footnote ⁸

Affirmative action motivated by benevolence seems paternalistic towards those group members who do not welcome the initiative. Indeed, it is introduced precisely because the policy-maker believes it benefits them. But what to say about a situation where, for example, a prestigious university recognizes that many of their applicants oppose affirmative action but uses it anyway because the relevant policy-makers believe that increasing the representation of members of the disadvantaged group at the university will have trickle-down effects and thus benefit the rest of the group? In this case, the intended beneficiaries are not those admitted to the university but the remaining part of the disadvantaged group. Is an affirmative action policy motivated in this way paternalistic?Footnote ⁹

One potential reason for thinking that the policy is not paternalistic is that there seems to be a mismatch between the group members interfered with and the intended beneficiaries. However, on closer inspection, it is not clear that such a mismatch exists. It is still the case that the policy deprives everyone in the relevant group of the opportunity to apply to the university without having their group membership used as a selection criterion. In this way, it interferes with all potential applicants from the group, irrespective of whether they will use the opportunity to apply. If A removes an option from B's choice set, A seems to interfere with B regardless of whether B would have used the option without A's removal.Footnote ¹⁰ The upshot is that an affirmative action policy motivated by promoting the interests of those group members who are only indirectly benefitted by the selection criteria can still be considered paternalistic towards these group members.

Let us turn to the second type of motivation. Is it paternalistic if motivated by an equity concern; that is, that all are equally entitled to participate in mainstream public institutions? We must first ask why people are entitled to participate in mainstream public institutions. A stone, for instance, is not entitled to participate. So, their entitlement must have to do with their moral status. And if they are equally entitled, that must be because they are moral equals. One might say, then, that when there is inequality of opportunity, this means that, all things being equal, the interests of moral equals are not equally protected. If one then pursues affirmative action to ensure better protection of the interests of those with worse opportunities than others, the affirmative action policy would be paternalistic. We believe that this is often what goes on in the equity case.

A potential worry with this argument is that one may think that the aim of promoting justice is different from the aim of promoting people's interests or well-being, even if promoting justice (say, equality of opportunity) involves promoting the interests of disadvantaged groups.Footnote ¹¹ However, in our view, nothing precludes that policies introduced to promote justice can be paternalistic. For example, imagine a state restricting access to cigarettes because smoking has disadvantageous effects among disadvantaged groups. In their view, preventing smoking would promote justice (cf. Voigt Reference Voigt2010). In this case, it seems that members of the disadvantaged groups who do not welcome such restrictions may still, with good reason, complain that they have been treated paternalistically. In other words, paternalism can be seen as a potential means to promote justice (see, for example, Arneson Reference Arneson2005, 275).

Still, an equity rationale does not have to be paternalistic. For example, if one simply believed that equality of opportunity was valuable in itself, and one pursued affirmative action to secure equality of opportunity, the affirmative action policy would not be paternalistic.Footnote ¹² It would not be undertaken with the motivation to promote the targets' interests, good, or well-being. So, when it comes to equity-based reasons for affirmative action, we will have to look closer at motivation to determine whether the affirmative action policy is paternalistic.

A similar analysis can be provided regarding representation: it is valuable that mainstream institutions represent the surrounding society. If black people comprise 20 per cent of the population they should also be present in mainstream institutions. Why is that valuable? One might say that it is valuable for group members as a whole that they are present in mainstream institutions, even if some individual members of the group might not benefit; for example, a person being stigmatized for being the beneficiary of affirmative action. In this case, affirmative action benefits more individuals from the excluded group than not pursuing affirmative action would. If this is why one believes affirmative action is valuable when it comes to representation, the affirmative action policy would be paternalistic (cf. our discussion above). But a representation-based argument does not have to be paternalistic. Some, both in academic and legal contexts, have provided a diversity-based argument for affirmative action (see, for example, Appiah Reference Appiah2011, 278; Bowen and Bok Reference Bowen, Bok and Cahn2002, 179; Grutter v. Bolinger 2003; Lippert-Rasmussen Reference Lippert-Rasmussen2020, 6; Regents of the University of California v. Bakke 1978). If one were to argue that diversity is good and that we should pursue affirmative action because it secures more diversity, which is good in itself, this would be a non-paternalistic representation-based argument in favour of affirmative action.Footnote ¹³

It is often argued that diversity has positive effects on productivity. For instance, Appiah (Reference Appiah2011, 278) says, ‘why is diversity a good? Because diversity of social identities makes education and research better.’ This is what we would classify as a productivity argument for affirmative action. If one were to justify affirmative action by appealing to this reason, then the benefit condition in the definition of paternalism would not be met, and the affirmative action policy would be non-paternalistic. But suppose that one were to argue that diversity is good for productivity and that one could use these extra resources, resulting from increased productivity, to benefit the worst off. Then, if one were to motivate affirmative action by saying that it would lead to more diversity, which would lead to more productivity, which would lead to better protection of the interests of worst-off individuals, the affirmative action policy would be paternalistic if the worst-off opposed the policy.

Reputation is the most clearly non-paternalistic motive for pursuing affirmative action. It is hard to see how a reputation-based affirmative action policy could be said to be pursued to benefit the targets. Indeed, in such cases, affirmative action is pursued precisely to avoid adverse reactions from the surrounding society and not to help those who will be affirmative action targets. The beneficiaries' interests do not even play a role in the deliberation.

We have now seen five different motives for increasing the participation of disadvantaged groups in institutions through affirmative action. Based on the analysis, we suspect that most actual cases of affirmative action are justified by appeal to how the interests of the beneficiaries are promoted; that is, most cases of affirmative action are paternalistic towards those from the targeted groups who do not welcome the policies. In what follows, we will be interested in cases of paternalistic affirmative action.

The Disrespect Objection to Paternalism

An often-raised objection to paternalism is that the agent acting paternalistically insults or shows disrespect for the person interfered with. Paternalistic acts and policies involve the paternalist subordinating the autonomy of another person to the person's interests, good, or well-being. In doing so, the paternalist fails to respect other people's status as the authors of their own lives. As Feinberg puts it, ‘[t]he life that a person threatens by his own rashness is after all his life; it belongs to him and to no one else’ (Feinberg Reference Feinberg1986, 59). This status as the ‘authors’, ‘owners’, or ‘controlling force’ of our lives is not respected when others interfere with us to our benefit. For example, as Eidelson (Reference Eidelson2015, 142) explains with reference to the Jehovah's Witness who does not want a blood transfusion even if this could prevent an accident from having a tragic outcome:

Suppose we introduce an affirmative action initiative, say gender quotas, which is motivated by promoting women's interests in equal opportunity. Now imagine a woman applying for a managerial position. She wants the job only if she would have been given the position without the presence of affirmative action. If she could not get the job ‘on her own’, she would rather not have it at all. She prefers her situation without the advantage associated with the selection criteria. The woman's project in this situation differs from what motivates the selection criteria. Through the initiative, we seem to substitute the woman's view on what would best promote her interests (non-affirmative action) with our idea of what would be best for her (affirmative action). In other words, we fail to recognize that her project differs from ours. That is, we call her status as ‘the controlling force in her own life’ into question.

One might question whether it is reasonable to describe paternalistically-motivated affirmative action as disrespecting a woman's autonomy. For example, reserving seats for women may not be seen as interfering with any of the woman's autonomy rights.Footnote ¹⁵ Whether or not an institution introduces gender quotas does not seem to be within the woman's discretion or legitimate decision-making domain. It might plausibly be the case that the woman does not have a claim to the opportunity of applying for the position without the advantage associated with the selection criteria.

However, the disrespect objection applies even if gender quotas do not violate any such right. Consider, for example, the parallel case where I refuse to lend you £50 because I think you will spend the money buying heroin. According to Quong (Reference Quong2010, 79–80):

Since it is usually up to me whether I want to lend others money, one may ask why this is a case of disrespectful paternalism. The answer is to be found in the specific reasoning on which I act. As Enoch (Reference Enoch2016, 46) argues:

If, instead, I refuse to give the money because of my interest in not contributing to my friend's heroin habit (cf. Shiffrin Reference Shiffrin2000) or because I want to spend the money myself, then my action is not a case of disrespectful paternalism. So, reserving a seat for the woman who prefers her situation without preferential treatment because this would be good for her is to act for a reason that fails to respect her as the author or controlling force in her own life (see also Shiffrin Reference Shiffrin2000, 214). But this, of course, leaves many other reasons for action open. Thus, according to this objection, paternalistic affirmative action is disrespectful towards the intended beneficiaries even if it is not generally within the applicant's discretion which criteria should be used by assessment committees.

Respectful Affirmative Action

While paternalistic affirmative action seems disrespectful towards its intended beneficiaries for the above reason, we will argue that it is not necessarily disrespectful all things considered. In our view, there is a relevant sense in which (even paternalistic) affirmative action is a means of respecting the moral worth of individuals. Specifically, in the literature on paternalism, authors have argued that if A could easily prevent B from harming their vital interests, but fails to do so, then there is a relevant way in which A, because of A's inaction, can be said to disregard B's moral worth (see also Hojlund Reference Hojlund2021; Pedersen Reference Pedersen2021, 430). The underlying thought here is that to respect a person's moral worth, one must show appropriate concern for them. In some instances, when people act imprudently, not interfering with them will be tantamount to not giving due weight to their vital interests.

Vital or moral interests, as we may call them, are different from the person's preferences. They are interests that attach to the person's moral values and are thus objective in a way that preferences are not (cf. Velleman Reference Velleman1997). To illustrate, suppose a person is about to walk onto a dilapidated bridge. Even if the person acts voluntarily – if, that is, the person has a preference for crossing the bridge – it seems that there is a relevant sense in which we do not give due weight to the person's vital interests in adequate physical and mental capacities when we allow the person to continue but could easily have prevented them from doing so. Failing to show appropriate concern for a person's vital interests when one could easily have done so is to disregard that person's moral worth (Pedersen Reference Pedersen2021).Footnote ¹⁷

In our view, a similar argument can be made in the case of affirmative action, which aims to increase the participation of disadvantaged groups in mainstream institutions to promote equality of opportunity. Since individuals are moral equals irrespective of their group membership, failing to mitigate the inequality of opportunity that exists between different groups, if one is in a position to do so, is to disregard the vital interests of people from the disadvantaged groups compared to the vital interests of people from the advantaged groups. In this way, it disrespects the moral worth of people from disadvantaged groups.Footnote ¹⁸ Importantly, this is also the case for group members from disadvantaged groups who prefer their situation without affirmative action since they disregard their vital interests, not their preferences. Even those who do not prefer affirmative action still have vital interests in equality of opportunity.

Of course, it is relevant to the respect-based assessment of affirmative action that some disadvantaged group members do not welcome the policy since we, by pursuing affirmative action, thereby fail to respect their status as the authors of their own lives (cf. the arguments in the previous section). But this does not change the fact that failing to mitigate inequality of opportunity when one is in a position to do so does not show proper concern for their vital interests (compared to the interests of others). Therefore, it disrespects them qua disregarding their moral worth. In this way, it can be disrespectful in one sense not to pursue affirmative action. Therefore, this consideration should also be included in the overall assessment of the policy's (dis)respectfulness.

The above argument assumes that affirmative action is needed to mitigate the inequality of opportunity between different social groups. Indeed, some studies find that affirmative action is particularly useful in combating inequality (Adams Reference Adams2021; Gulzar, Haas, and Pasquale Reference Gulzar, Haas and Pasquale2020; Khaitan Reference Khaitan2015, 8). However, if this can be achieved as effectively in other (perhaps less interfering) ways, we are entirely open to other possibilities.Footnote ¹⁹ Yet, we find it challenging to provide good examples of such alternative, effective measures (see Adams (Reference Adams2021) for why affirmative action, at least in the US context, is (at least part of) what is needed to mitigate inequality). Because the playing field is not equal for members of advantaged and disadvantaged groups, we need a policy like affirmative action – which changes the playing field to provide better playing conditions for disadvantaged people – to get closer to equality of opportunity. But basically, our argument concerns the disrespect associated with standing idly by when one could relatively easily have done more. In some cases, affirmative action is the most relevant example of how one could have done more.

At this point, one might point out that we appeal to both respect and equality of opportunity but ask whether respect is doing any explanatory or justificatory work beyond that done by equality of opportunity. Why appeal to respect?Footnote ²⁰ In our view, the notion of recognition respect helps explain why we should secure equality of opportunity. Specifically, to respect people's moral worth as persons is, in Darwall's terms, to show them recognition respect. To show recognition respect for persons, one must ‘take seriously and weigh appropriately the fact that they are persons in deliberating about what to do’ (Darwall Reference Darwall1977). Since people's vital interests attach to their moral worth as persons (Eidelson Reference Eidelson2015, 97; Hojlund Reference Hojlund2021, 523), we disrespect the disadvantaged persons if we ignore their vital interests in equality of opportunity – or fail to take these interests as a reason for action. In this way, respect ultimately explains why we must promote equality of opportunity.Footnote ²¹

It is also relevant to consider whether the disrespect objection to paternalism can necessarily be avoided if affirmative action measures are not introduced. Having focused on those members of the disadvantaged groups who oppose affirmative action, we now shift our focus to those members of the disadvantaged groups who welcome affirmative action (the members towards whom affirmative action is not paternalistic). The disrespect objection to paternalism presented in the section ‘The disrespect objection to paternalism’ does not apply here – at least not if affirmative action is introduced. However, as shown below, the objection may arise if affirmative action is not introduced.

As described in the introduction, one common objection to affirmative action is that it stigmatizes its intended beneficiaries. As Cohen dramatically puts it:

This is known as the stigma objection to affirmative action. That beneficiaries of affirmative action are stigmatized finds support in some experimental studies. For example, in two studies conducted by Heilman, Block, and Stathatos (Reference Heilman, Block and Stathatos1997, 603):

It would be unfortunate if the introduction of affirmative action had such unintended stigmatizing effects. However, if policymakers decide not to introduce affirmative action for the reason that such initiatives would lead to stigmatization of the beneficiaries of affirmative action, and they believe it would be better for these people not to be subject to such stigmatization, then this is paternalistic towards those potential beneficiaries of affirmative action who prefer their situation with such initiatives in place (regardless of the stigma that this would potentially entail). As described above, even unwelcome benevolent omissions are paternalistic. Accordingly, avoiding affirmative action out of concern for the interests, good, or wellbeing of the potential beneficiaries does not avoid the disrespect objection presented in the section, ‘The disrespect objection to paternalism’. Interestingly, this shows that although it can be paternalistic to implement affirmative action, it can also be paternalistic not to implement affirmative action.

Nevertheless, we have just argued that it can be disrespectful to stand idly by when people are making decisions that jeopardize their vital interests. Is introducing affirmative action, knowing that this will likely lead to stigmatization of the intended beneficiaries, not disrespectful because it disregards the beneficiaries' strong interest in avoiding such stigmatization? In our view, it would be if nothing is done to counteract or mitigate these unintended effects. Here, however, the disrespect will not lie in introducing affirmative action as such but in the absence of actions that serve their interests in avoiding stigmatization. Indeed, as Lippert-Rasmussen (Reference Lippert-Rasmussen2020, 184–7) argues, the stigma resulting from affirmative action will often result from epistemic injustices committed against members of the targeted groups. If so, the stigma is simply an unjust response to affirmative action being implemented to combat injustice rather than something inherent in affirmative action itself. Thus, it is not affirmative action that is disrespectful (but people's responses).Footnote ²⁴ Therefore, this objection does not threaten our main claim in this section, namely that although paternalistic affirmative action can be disrespectful qua paternalistic, it can also be disrespectful not to pursue affirmative action.

Adaptive Affirmative Action Preferences

In the preceding section, we argued that it does not follow from the fact that affirmative action is paternalistic that it is, for that reason, disrespectful all things considered. Indeed, it can be disrespectful not to pursue paternalistic affirmative action. In that section, we took the preferences of those who oppose and favour affirmative action as given, at face value. We did not consider how these preferences were generated in the first place. We want to do so in this section. This is because affirmative action responds to injustice.Footnote ²⁵ And then potential beneficiaries' preferences about affirmative action may similarly be shaped by injustice. If they are, this makes a difference to the (dis)respectfulness of (not) pursuing affirmative action. At least, we argue so below.

As we said about the disrespect objection to paternalism, paternalism is considered disrespectful because it violates the recipient's autonomy. Enoch recently argued that there are two distinct autonomy considerations: autonomy as non-alienation and autonomy as sovereignty.Footnote ²⁶ Autonomy as non-alienation is to be the author of one's life. As he says, ‘you're autonomous in the sense of non-alienation vis-á-vis an action or a decision that concerns you to the extent that the relevant matter is determined by your values, or your deep commitments’ (Enoch Reference Enoch2022, 144). Autonomy as sovereignty is to have the last say on the matter. To illustrate the difference, Enoch (Reference Enoch2017, 31–2) imagines that his daughter sometimes refuses to pass him the salt at the dinner table when he asks for it. Sometimes, he insists and explains to her that he is aware of the health risks, but it is his body, his life, and that he wants the salt. If she continues to refuse, she is offending his autonomy as sovereignty but not his autonomy as non-alienation.Footnote ²⁷ His deep commitments or his ability to write his life story in accordance with them are not threatened by his daughter refusing to pass him the salt – his deep commitment is, after all, to be there for his children in the future – but he is not given the last say in this case.

Both autonomy concerns may be disrespected in cases of paternalism. In so far as X paternalizes Y because X believes it would be better for Y to live by X's conception of the good and not Y's values and deep commitments, the paternalistic act would disrespect Y's autonomy as nonalienation. In being paternalized in this way, Y is not given the final say on the matter, so the paternalistic act would also disrespect Y's autonomy as sovereignty.

Let us now return to potential beneficiaries who oppose affirmative action. We argued that pursuing affirmative action would be paternalistic towards these people as it would circumvent their agency or judgement (in addition to interfering with them to promote their interests, good, or well-being). Indeed, that they preferred their situation without affirmative action was a necessary condition for it to be paternalistic to pursue affirmative action towards them. So, their preferences are important in this context. But preferences can be generated in different ways.

In an illuminating discussion, Enoch (Reference Enoch2020) has recently shed new light on the phenomenon of adaptive preferences. He presents the following case:

Is the starving woman's preference autonomous? Not according to Enoch. He presents a moral, historical account of when preferences are non-autonomous. A preference is non-autonomous when it is ‘shaped (in the appropriate way) under the causal influence of unjust conditions, conditions that violate the rights or entitlements of the relevant agent’ (185).Footnote ²⁸ The starving woman's preferences ‘were shaped under the causal influence of a social order that routinely violates her rights, that wrongs her … The causal role played by injustice here explains why it is that these preferences (or consent or choice based on them) do not manifest the value of autonomy (as nonalienation), for it shows an important sense in which these preferences are not the agent's (185)’. Her life is being shaped by her oppressors instead of by herself (188).

Let us now suppose, as was the case with the starving woman, that some potential beneficiaries who oppose affirmative action hold this preference due to the causal influence of unjust conditions. Although it is ultimately an empirical question to what extent this is the case,Footnote ²⁹ there is at least some reason to believe that adaptive preference opposition to affirmative action is not a marginal phenomenon. It does not seem unlikely that some members who live in societies with racist, sexist, and other prejudiced norms internalize these norms.Footnote ³⁰ As Scanlon (Reference Scanlon2018, 64) says, ‘one objection (not the only one) to racist and sexist attitudes in a society is that they undermine equality of opportunity by discouraging members of these groups from thinking of various worthwhile careers as appropriate for them’. Similarly, they may internalize that they are worth less and that their vital interests are less important because of their race or sex. For example, as Lippert-Rasmussen (Reference Lippert-Rasmussen2006, 173) puts it, ‘some of the most damaging forms of discrimination are those in which discrimination is, so to speak, internalized by the victims of discrimination and involve some sort of abhorrence of what one is.’ Imagine, in this context, that the woman who prefers her situation without gender quotas grew up in a patriarchal society with men rather than women being leaders and, therefore, does not believe that she, as a woman, deserves the position unless she can get it without affirmative action. This would be an example of the cases we have in mind in this section. What would such influence of adaptive preferences change, if anything, about affirmative action being disrespectful qua paternalistic?

First, it would mean that the affirmative action policy would not be disrespectful in terms of violating the beneficiaries' autonomy as non-alienation. After all, the preference would not in the relevant sense be theirs, but their oppressors' (those upholding the unjust circumstances). In expressing or acting on this preference, they would be ‘more like a pawn in someone else's game’ (Enoch Reference Enoch2020, 188) than the author of their life. So, in paternalizing them through affirmative action, one would not limit the extent to which their lives would be determined by their values and deep commitments. Instead, one would limit the extent to which the person's life would be determined by the oppressors' values and deep commitments (or the preferences conducive to the oppressive system). This is significant. It means that if the disrespect concerning paternalism has to do with disrespecting autonomy as non-alienation, affirmative action would not be disrespectful qua paternalistic in such cases. As Enoch puts it, ‘you don't respect them by accepting their choices when those are motivated by preferences they only have because causally implemented in them by their oppressors. Perhaps you respect them precisely by refusing to play along’ (201). And we take it that a focus on autonomy as non-alienation is, at least sometimes, what is at stake when some people charge paternalism with being disrespectful.

But even if so, it is still the case that in pursuing paternalistic affirmative action, one is offending against the autonomy and sovereignty of those who oppose affirmative action. The beneficiaries are not given the final say on whether there should be affirmative action for their sake. So, to the extent that the disrespect concern, when it comes to paternalism, has to do with disrespecting autonomy as sovereignty, how the preferences came about is irrelevant. As this helpfully illustrates, we must be clear on why paternalism is disrespectful autonomy-wise. Paternalism is not necessarily disrespectful of autonomy as non-alienation, but it is always disrespectful of autonomy as sovereignty.