1 Introduction

According to the relational egalitarian theory of justice, justice requires that relevantly situated agents relate as equals.Footnote 1 Recent work on the topic suggests that this ideal does not only require that people treat each other as equals, but also that people form attitudes about others in ways that are constrained by a concern for relational equality (e.g., Hojlund 2022; Lippert-Rasmussen 2018a; Ross 2022). On some ways of interpreting this idea, it is not that hard to see that attitudes could come about in objectionably, because inegalitarian, ways. For instance, if Marika deems Maxi of lower moral worth because of Maxi’s gender, this certainly feels like a type of attitude that is non-egalitarian in spirit and objectionable for that reason. But this case narrowly concerns beliefs about moral status. Some relational egalitarians go further, suggesting that the ways in which we esteem or appraise one another are subject to egalitarian constraints as well (Ross 2022).

In this paper, we argue that background injustice gives rise to distinctive relational egalitarian requirements of affirmative action in terms of how we confer esteem and disesteem. Specifically, when faced with evidence suggesting that those belonging to disadvantaged groups have traits worthy of disesteem, we should display further resistance to believe and hold out for more evidence. Similarly, when faced with evidence suggesting that those belonging to advantaged groups have traits worthy of esteem, we should be displaying further resistance to believe this and require more evidence. However, a similar resistance is uncalled for when evidence suggests that people from a disadvantaged background have traits worthy of esteem and people from advantaged groups have traits worthy of disesteem. This dynamic can be justified by appealing to the ideal of equality of opportunity, something that is seen by many relational egalitarians as partly constitutive of their favored ideal of justice as relating as equals. Just as affirmative action can be seen as offsetting the effects of background injustice and thereby promoting equality of opportunity in acquiring valuable goods and positions, affirmative action in terms of how we apportion (dis)esteem helps offset the effects of background injustice on people’s ability to acquire traits worthy of esteem and disesteem. Accordingly, we call this view doxastic affirmative action.

The plan is as follows. In the next section, we introduce affirmative action and relational egalitarianism. In relation to the latter, we explain why esteem is of importance to how we relate to each other. In Section III, we argue that relational egalitarians have good reason to pursue doxastic affirmative action when it comes to apportioning (dis)esteem. Before concluding (Section V), we respond to the objection that our argument is not really relational egalitarian since relational egalitarians are collectivists, and not individualists, when it comes to (dis)esteem practices (Section IV).

2 Relational Egalitarianism, Affirmative Action, and Esteem

We want to argue that relational egalitarians sometimes have good reason to pursue doxastic affirmative action. Before explaining why, we should start by clarifying what relational egalitarianism and affirmative action are and saying something about the nature of esteem and its significance for relational egalitarianism.

Starting with the former, relational egalitarianism is a theory of justice according to which justice ultimately requires that people relate as equals, or at least that they do not relate as inferiors and superiors (e.g., Anderson 1999; Lippert-Rasmussen 2018a).Footnote 2 Paradigmatic instances of relational inequality include discrimination, domination, exploitation and racism. Given this focus, relational egalitarians argue, they are able to capture the concerns of real-life egalitarians (Anderson 1999; Schemmel 2021).Footnote 3 Relational egalitarians typically argue that for two people to relate as equals, they must regard each other as equals and treat each other as equals (Anderson 1999; Fourie 2012; Lippert-Rasmussen 2018a; Miller 1998; Voigt 2018). Whereas the first is an attitudinal requirement, the second is a behavioral requirement. These may come apart. An opportunistic racist may treat someone whom they regard as inferior as their equal to avoid criticism from others. Conversely, an egalitarian may treat someone whom they regard as their equal as their unequal to satisfy their racist boss (cp. Lippert-Rasmussen 2018a: 72). In this paper, we will be concerned with the attitudinal component of relational egalitarianism. We will say more about this component as we move along, including why, according to relational egalitarians, attitudes are constitutive of how people relate to each other. But before we do so, we would like to say a bit about affirmative action.

As it is usually understood, affirmative action is a non-ideal phenomenon in the sense that it is used when and because there is background injustice.Footnote 4 Such background injustice may be a result of historical wrongs, such as slavery, but also continued injustices, such as racism and sexism. Affirmative action is used to benefit those who are disadvantaged due to such injustices. For example, it is used in university admissions to benefit those who have had worse opportunities to earn the qualifications necessary for being admitted. And in a similar way when it comes to hiring for prestigious jobs. But there are many more ways in which affirmative action can be pursued (as we will see later).

As we are not particularly interested in definitional matters, given our purposes in this paper, we will simply understand affirmative action in the following way:

Affirmative Action

“A policy, an act, etc. amounts to affirmative action if, and only if, in a particular site of justice (i) the agent of the policy, etc. ultimately aims at reasonably increasing the representation of minorities in the relevant area or aims at reasonably addressing the disadvantages they suffer in the relevant area in at least some, but presumably not all, ways other than by boosting their representation, or (ii) the relevant policy, etc. will in fact, or is believed to, address a disadvantage of a certain minority group in the relevant area using certain means, e.g., quotas, that go beyond eliminating direct discrimination against the group but not beyond eliminating the relevant disadvantages.”Footnote 5

This suffices for introducing affirmative action (we will say more as we move along). Now, we will argue that relational egalitarians, under conditions of background injustice, have good reason to pursue affirmative action in the attitudinal domain.Footnote 6 We saw above that relational egalitarians argue that how we regard each other is important for how we relate to each other. How people appraise or esteem each other is an attitudinal matter. As such, how people appraise or esteem each other matters for relational egalitarian justice. Here, we shall use these terms—appraise and esteem—interchangeably and take them to refer to a certain kind of emotion directed towards an object (Zagzebski 2017).Footnote 7 In the case of esteem (or positive appraisals), we can identify this with the emotion of admiration which again can be thought of as a specific kind of pro-attitude towards the object under consideration. We will follow Zagzebski (2017: 31) and use contempt to refer to the corresponding con-emotion involved when an attitude of disesteem (or negative appraisal) is taken towards the object under consideration. Of course, it is an open question how to best describe the family of emotions here, but for our purposes, we can stick with this simple picture.

Notice that one can also use ‘appraisal’ and ‘esteem’ to refer to practices of esteeming or appraising people via actions that conventionally signal such attitudes (e.g., building a statue of someone will often conventionally signal positive appraisal of the person depicted). And plausibly, relational egalitarians should also have something to say about the structure of such practices (see, e.g., McTernan 2016). But here we will confine ourselves to the question of fitting attitudes as it seems more foundational and therefore central to inform the more practical, derivative question.

As mentioned, esteem can be (and typically is) directed towards different kinds of objects. The attitude can be directed towards an agent (e.g., Gandhi or the French Army), or one or several intrinsic traits of the agent, or simply their achievements (e.g. Gandhi’s admirable traits or the things he did). It is also important to stress that these attitudes can differ in scope. Admiring Heidegger for several traits he had or what he achieved need not imply a global evaluation of Heidegger as admirable in a way that goes beyond the specific things that made him a fitting object of admiration. And correspondingly, denigrating or disesteeming a criminal person for their criminal activities need not imply a global negative evaluation of the person. But as Fruh (2023) points out, more restricted evaluations of traits or achievements can often ‘spill over’ or ‘leak’ into more global evaluations of entire persons. This is true as a matter of psychological propensity, but also seems plausible when reflecting upon our (dis)esteem-conferring social practices. Building a statue of Einstein to signal admiration for his scientific achievements will often end up signaling a more global admiration of Einstein. In so far as we think that such leaks are problematic, there may be a prima facie case for preferring social practices that only signal narrow characterological admiration or even only admiration of achievements (Fruh 2023; see also Kolodny 2014; McTernan 2018).

Thus construed, why should relational egalitarians be concerned with how people esteem each other as a part of an ideal of justice as relating as equals? We can think of two compelling reasons why (dis)esteem is something to care about in the first place. In laying out these reasons, it is important to stress that we are not taking a substantive view on what kinds of traits and acts relational egalitarians should deem worthy of esteem and disesteem. Instead, our argument should be read as conditional upon there being traits that relational egalitarians should have independent reason to want people to base their esteem and disesteem upon. Alternatively, our argument could be read as conditional upon the claim that esteem-conferring practices are likely an inescapable part of social life. Given that we likely cannot do away with such practices, we should prefer that they are tempered by a concern for relational equality.

Now, the arguments. The first proceeds from the observation that when we (dis)esteem people, we are in fact evaluating them along some dimension. And a core commitment of relational egalitarianism seems to be that evaluations of people should be constrained by equality. To motivate this point, we can draw on a parallel to how relational egalitarians object to hierarchies of social status based on race, caste, or class. Such hierarchies, notice, are also in part constituted by evaluations of people, namely ascriptions of social status (van Wietmarschen 2022). Relational egalitarians disagree over why such hierarchies are objectionable, but a widespread commitment seems to be that they are objectionable because of their inegalitarian nature. We can then make progress by pointing out that allocations of esteem can be objectionable in much the same way, even though esteem typically only involves evaluating aspects of a person as opposed to their (typically broader in scope) social status. To see this, imagine a person who easily jumps to conclusions about people partly based on their demographic group membership, e.g., judging that a person is dangerous (and correspondingly judging them worthy of disesteem) only because they belong to a certain demographic group that is overrepresented in crime statistics.Footnote 8 Or suppose that a societal majority believes that women, on average, are worse drivers than men, and for that reason interprets even the faintest signals of bad driving (e.g., slightly wobbly driving) as decisive evidence of a specific woman being a bad driver.Footnote 9

By contrast, this person would be much more cautious when forming opinions about people belonging to other demographic groups, e.g., by looking for further signs of bad driving skills before forming a judgment of a man. From a principled perspective, it is hard to see what the moral difference is between according to someone a low social status because of their membership of a certain caste and being much more inclined to judge people worthy of disesteem because of their demographic features. If there is a moral difference here, it seems to be one of degree rather than kind as both judgments seem to be in tension with a commitment to equality in that different standards—seemingly absent good reasons—are applied to different (groups of) people.Footnote 10 This suggests that relational egalitarians should object to at least some ways of apportioning esteem and disesteem (namely those ways that are inegalitarian).

Here is the second reason why relational egalitarians should care about how we apportion (dis)esteem. In making the above argument, we have not committed ourselves to anything substantive about the value or disvalue of being (dis)esteemed by others. But one could easily argue that it is non-instrumentally good (for people) that our evaluations of them are appropriately based on features of their person. Moreover, in many societies, including one such as ours, there is a significant correlation between how we (dis)esteem people and how they are otherwise benefitted or deprived. This establishes the instrumental significance of esteem. For instance, we award prizes to those we deem most worthy of our esteem within a given domain (e.g., grants based on scientific excellence), and within other domains, we impose costs on those we deem worthy of certain forms of disesteem (e.g., the criminal system). One reason for relational egalitarians to object to social structures based on caste, race, and class is that they distribute benefits and burdens in non-egalitarian ways (see, e.g., Kolodny 2014). And insofar as our ways of esteeming one another confer benefits and burdens, either instrumentally or non-instrumentally (or both), they should be concerned with how (dis)esteem is conferred as one instance of this broader commitment.

Thus, the first reason points to why relational egalitarians have reason to care that evaluations of people are constrained by equality. The second reason points to why relational egalitarians have reason to care about (dis)esteem; because esteeming, and disesteeming, confers burdens and benefits. Taken together, these reasons explain why (dis)esteem should be an object of concern for relational egalitarians. But these reasons do not tell us precisely what an ideal allocation of esteem would look like from a relational egalitarian point of view.Footnote 11 What might that ideal look like? A plausible suggestion is based on the ideal of equality of opportunity.Footnote 12 Indeed, what goes wrong in the cases presented above seems to us aptly described as failures of affording people an equal opportunity for earning esteem and avoiding disesteem. When different standards for conferring esteem are applied for different demographic groups, this brings these different groups of people on an unequal footing in their ability to earn our (dis)esteem. In the case of the female driver presented above, a slight misstep will mean that she is disesteemed, whereas men in comparison might be given a much more charitable interpretation of their driving. Of course, inequality may obtain for other reasons as well, but different standards make up one important source of inequality of opportunity.

Focusing on (denying) equality of opportunity as the diagnosis makes sense within a relational egalitarian perspective, as many relational egalitarians see a tight connection between this ideal and the idea of relating as equals. To give a few examples, Kolodny (2014: 291) argues that, to secure equal relations, an equality constraint must be satisfied in the sense that “if a procedure gives anyone a say, it should give everyone an equal say.” This is why democracy—which grants everyone an equal opportunity for influence—is a particularly important constituent part of relational equality according to Kolodny (see also Peña-Rangel 2022: 23). Similarly, Lippert-Rasmussen (2018a: 211) says that “we find both in Anderson and Scheffler [two of the most prominent relational egalitarians] suggestions to the effect that equality of opportunity is not just desirable as an instrument for promoting suitably egalitarian relations, but that it is either independently desirable or desirable as a constitutive part of the relational ideal … this implies that any critique of equality of opportunity is a critique of a constitutive element of relational equality.” Finally, and of particular relevance to our purposes, Ross (2022: 815) says, “What relational equality does require is a level playing field with respect to earning the esteem of your fellow citizens.”Footnote 13

These arguments express the importance of equality of opportunity for (some) relational egalitarians. With this in hand, it is easy to see that an inequality of opportunity when it comes to earning esteem, especially when esteem evaluations are also evaluations of persons, might be constitutive of relational inequality. And this is one reason why relational egalitarians should care about how we esteem people.Footnote 14

We will assume that these relational egalitarians are right that equality of opportunity is an important part of (their conception of) relational egalitarianism: that a level playing field is necessary for people to relate as equals. Other relational egalitarians may be skeptical that equality of opportunity is required for relational equality.Footnote 15 Although we will proceed with basing our argument on equality of opportunity (given its importance for the relational egalitarians mentioned above), it is important to stress that our argument might not even require that equality of opportunity in esteem evaluations is necessary for relational equality. The reason is that even if equality of opportunity in esteem evaluations is not required, it might still be the case that, for people to relate as equals, at least people’s race, gender, and similar traits should not make a difference to how we esteem them (cp. Lippert-Rasmussen 2022). The latter requirement is weaker than the former: there might still be inequality of opportunity when people’s race, gender, etc., do not make a difference to how they are esteemed. So, the relational egalitarians who may be skeptical that equality of opportunity is required may at least agree that traits like race and gender should not make a difference to how we esteem people. Therefore, they may still agree with us that relational equality requires doxastic affirmative action under conditions of background injustice since, if we do not pursue doxastic affirmative action, race and gender will affect how we esteem people since the background injustice is structured around such traits. This shows that our argument (suitably modified) should be of interest to relational egalitarians of different stripes.Footnote 16

Our interest is whether there ever is a case for engaging in affirmative action when forming attitudes of (dis)esteem about others.Footnote 17 Before we propose a way in which this might be done, it will prove helpful to indicate another possible way that is unattractive and which also points to an asymmetry between affirmative action as standardly practiced (e.g., in filling job positions) and affirmative action in the attitudinal case. This asymmetry arises because attitudes of (dis)esteem—as many other attitudes such as beliefs, desires and emotions in general—are subject to requirements of fit. That is, there is an internal or constitutive standard that must be met for an attitude of, say, esteem to be fitting.Footnote 18 For instance, you cannot (appropriately at least) admire a person if you do not believe this person has the traits that merit admiration, just as there is something deficient about fearing what is not in fact fearsome (e.g., fearing a toaster). More generally, this feature of esteem attitudes seems to block the possibility (or desirability, or both) of engaging in affirmative action in ways that fail to respect the fittingness requirements. By contrast, it is not obvious that we are violating a constitutive standard of fit if we prefer to employ a person from a historically disadvantaged group over a more skilled, non-disadvantaged person.Footnote 19 This raises the intriguing question of whether there is a way of practicing affirmative action in our (dis)esteem practices that do not violate constitutive standards regulating the fittingness of such attitudes. We will turn to this question now.

3 Doxastic Affirmative Action

Here is an observation that will structure most of this section. In an unjust world, people have had very different opportunities for developing traits and acting in ways that command our fitting esteem. Correspondingly, in an unjust world, the opportunities for developing traits worthy of our disesteem and acting in ways that are worthy of our disesteem are unequally distributed. As an example of the former, it often takes time, practice, and resources to develop traits and make achievements worthy of our esteem. As an example of the latter, lack of adequate opportunities may easily bring about the cultivation of traits and deeds worthy of disesteem. Some may engage in criminal activity out of bare necessity, and the lack of adequate access to moral education and guidance may in and of itself bring one onto a path towards disesteem (Tadros 2020). There is both an absolute and a comparative point here. It might be especially regrettable that we disesteem people that have not had adequate opportunities to follow a more esteem-worthy path. But insofar as esteem-judgments are evaluations of people, it seems just as regrettable that we esteem some (and disesteem others) when opportunities for earning this have not been equally distributed. One response to this could be to reform our practices of allocating esteem entirely and purge them of any connection to character or persons. We think this would be both unfeasible and too radical, since characterological admiration and contempt are clearly sometimes appropriate, even obligatory, emotions to entertain.Footnote 20,Footnote 21

Another response that we will develop here is the following. When forming judgments about whether people have a relevant kind of (dis)esteem base—the traits that warrant responses of esteem and disesteem—we should be attuned to the risk that this base may be a product of background injustice, rather than the kind of agential facts that makes it appropriate to evaluate persons based on this trait. To motivate the idea, compare two individuals who have each committed a crime, assuming that this is the kind of thing that would normally warrant disesteem. However, upon closer inspection, we learn that one of the persons committed the crime only because of suffering unjustly from poverty. It seems appropriate, we suggest, to evaluate these two persons rather differently despite the outcomes under consideration being identical. While we might deem the action worthy of contempt in both cases, we only think it is appropriate to take the action as a basis for forming an attitude of contempt targeting persons in the case where background injustice did not influence the outcome.

One may agree with this but retort that this observation is of little practical relevance since we, most of the time, lack the evidence needed to discriminate between outcomes that make persons worthy of genuine (dis)esteem (because their agency is appropriately connected to an outcome worthy of (dis)esteem) and cases where people are not worthy of genuine (dis)esteem because of background injustices. Instead, one might infer that this should push us towards skepticism about the practice of conferring esteem or disesteem upon people under conditions shaped by background injustice. And yet, we think there is a more nuanced option (one where we do not have to give up making (dis)esteem judgments). Specifically, we think we should pursue doxastic affirmative action. We should be more cautious when forming negative evaluations of people from groups that are subject to background injustice; and, on the other hand, be more cautious when forming positive evaluations of people from groups that are subject to background privilege.

To describe how this could work in practice, we need some terminology (some of it reflected in the table below). First, we can distinguish between positive and negative traits by which we mean traits worthy of esteem and disesteem, respectively. Second, we can distinguish some doxastic attitudes: Belief (and disbelief) and suspending judgment. As Ross (2022; see also McGrath 2021) points out, we should not think of judgment suspension as being agnostic, but rather as the attitude of deciding not to make up one’s mind—a form of commitment to neutrality on the answer to some question. Third, we need a distinction between a minority group (disadvantaged due to background injustice) and a majority group (advantaged due to background injustice). This third point mirrors how people tend to discuss affirmative action as giving preference to members of a disadvantaged (typically) minority group over members of an advantaged (typically) majority group. Finally, we need the idea that we could have more (and less) demanding standards to the body of evidence we must have available to us to be warranted in forming a specific doxastic attitude. There are many ways of cashing out this idea. But one way of doing so is that when the stakes that flow from being mistaken increase, we should require more and better evidence to be warranted in forming a belief in the proposition under consideration. This idea is familiar from the literature on pragmatic encroachment, but is not wedded to this view (Bolinger 2020; Lackey 2021; Ross 2023).

Let us consider some cases to flesh out this idea. Consider first the cases of believing that a member of a disadvantaged group has a trait worthy of esteem and that of believing that a member of a disadvantaged group has a trait worthy of disesteem. We propose that evidence of background injustice in the ability to earn esteem at the group level calls for doxastic affirmative action in the form of letting oneself be constrained by a more demanding evidential standard for believing that a minority person is worthy of disesteem. By contrast, believing that such people are worthy of esteem for having a positive trait is comparatively less demanding. To illustrate this, consider:

Testimony, positive variant. Sheela hears from a friend at work that a colleague—a person belonging to a disadvantaged minority—is “self-made.”

Testimony, negative variant. Sheela hears from a friend at work that a colleague—a person belonging to a disadvantaged minority—has committed a crime.

When considering these cases, we need to keep some things fixed. First, we must take for granted that were things in fact as Sheela depicts them to be, then the appropriate response would be to deem her worthy of esteem in the first case and worthy of disesteem in the second case. That is to say, agents responsible for being self-made—that is, having become successful or wealthy by their own efforts—are worthy of esteem and people responsible for committing crimes are worthy of disesteem. Second, we are assuming that testimony from others is under normal conditions (that is, absent evidence of unreliability) sufficient to justify rational belief in a proposition but might under certain conditions be insufficient when the stakes are high (e.g., Lackey 2021). Finally, we are assuming that Sheela is evaluating the colleague “from a distance” in the sense that she has no other evidence available and that she is well aware that the colleague belongs to a minority group subject to background injustice affecting their opportunities for developing traits worthy of esteem. On this background, we suggest that Sheela should believe that the minority person has the positive trait based on the available evidence, but she should not believe that the minority person has the negative trait in the latter case, even though the evidence is of a similar type and strength. Instead, she should inquire further, and, absent further evidence, suspend judgment on the esteem-worthiness of the colleague. By contrast, we suggest things should be different if we know that the person belongs to a group that is comparatively advantaged due to background injustice:

Testimony, positive variant. Sheela hears from a friend at work that a colleague—a person belonging to an advantaged majority group—is “self-made.”

Testimony, negative variant. Sheela hears from a friend at work that a colleague—a person belonging to an advantaged majority group—has committed a crime.

In this case, we propose, it should be harder for Sheela to believe that the majority person has the positive trait and easier to believe that they have the negative trait because she knows the person belongs to an advantaged group (we explain why below). The pattern of verdicts, and how they are sensitive to facts about background injustice at the group level, can be schematized as follows (where ‘easier’ and ‘harder’ refers to the demandingness of the evidential standard that must be met for justifying the requisite attitude):

Person from disadvantaged group


Person from advantaged group



Withholding judgment



Withholding judgment

Positive trait




Positive trait



Negative trait




Negative trait



How can we explain and justify this pattern of verdicts? One way is by focusing on the evidential value of being aware of background injustice structured along the lines of group membership. Such evidence should make us more attuned to the fact that what we might initially be inclined to think of as traits worthy of esteem might in fact be an inappropriate basis for such evaluations because the trait does not appropriately reflect agential contributions. More specifically, the ordinarily accepted amount of evidence required for judging people worthy of esteem for having certain traits is insufficient to discriminate between (i) the scenario in which they have the trait but are not worthy of esteem, and (ii) the scenario in which they have the trait and are worthy of esteem. In other words, background injustice should attune us to the fact that the kind of evidence we ordinarily take as a basis for esteem judgments cannot rule out salient possibilities of error.

These errors, of course, would result in us ascribing both esteem and disesteem based on people having traits that for the most part are the result of background injustices—i.e., inequality of opportunity—rather than agential contributions. Relational egalitarians, given that equality of opportunity is constitutive of relational equality, should prefer that we actively aim for avoiding making such errors. To appreciate the relationship between equality of opportunity and errors better, notice that were a person to develop a trait worthy of disesteem due to having comparably bad opportunities, this would in itself be regrettable from the point of view of the ideal of equality of opportunity. But were we then to go on and further disesteem the person because of having this trait—which would be erroneous because we would be ignoring that the origin of this trait makes it unfit to esteem the person on this basis—we would effectively let the underlying injustice not only affect the allocation of opportunities, but also the allocation of esteem. To avoid making such errors, we must be cautious in the way we have suggested above.Footnote 22

Furthermore, not all types of errors need be concerning from the perspective of equality of opportunity. If the possibilities of error are distributed randomly or evenly across the population, there may still be equality of opportunity. But when injustice has affected opportunities for earning (dis)esteem, the possibilities of error will not be randomly distributed across the population. They will be distributed to the disadvantage of those who have been subject to injustice, that is, we will be more prone to disesteem them when we should not, and less prone to esteem them when we should (compared to when we (dis)esteem those who have not suffered from injustice).Footnote 23

Notice that evidence of background injustice at the group level affects the evidential standards in a way that is conditional upon the valence of the target proposition. How may this be explained? We can see this by focusing on the relationship between background injustice and the propensity for acquiring traits worthy of esteem and disesteem. In the beginning of this section, we assumed that background injustice makes members of some groups have a less than equal opportunity for acquiring esteem-worthy traits and an increased propensity for developing traits worthy of disesteem. For disadvantaged groups, then, the salient risk is that we fail to eliminate the effect of unjust social structures in how we disesteem people. But there is no comparable risk when focusing on positive valence propositions for minority people. On the other hand, the bias might be said to work in the other direction when focusing on people from privileged groups. Their esteem-worthy traits are at a heightened risk of being the product of unjust social structures rather than agential contributions. This explains why we should be more cautious when forming beliefs that they have traits worthy of our esteem.

Before ending this section, let us make three further remarks. First, as should hopefully be clear, our proposal amounts to a form of doxastic affirmative action in the sense that different evidential standards are taken to apply depending on the group membership of the person under consideration (and conditional upon the valence of the proposition under consideration). A way to see this is by noting that endorsing such different standards would clearly be an objectionable form of differential treatment on a background where there was no background injustice affecting the opportunity for earning traits worthy of esteem. Indeed, this was the result we indicated in the previous section. Group-membership should only affect evidential standards where this is a sufficiently reliable proxy for inequality in opportunities.Footnote 24

Second, notice that our proposal—by operating at the level of evidential standards—avoids the worry that affirmative action in esteem judgments could come apart from the esteem base that makes such judgments fitting in the first place. Hence, our proposal does not engender the potentially worrisome implication that we ought to (dis)esteem people for things that are in fact not true of them.

Third, our results offer an important nuance to a recent exploration of how relational egalitarians should think about esteem-judgments. In a recent paper, Ross (2022) takes up the question of what is wrong with forming beliefs about people that are predominantly based on statistical regularities. We are not interested in Ross’ specific explanatory ambitions here, but in his conclusion that there is something distinctively relational egalitarian to what he calls evidential parity:

The beliefs that result from demographic profiling undermine the attitudinal requirements of social equality. By harbouring antecedent beliefs about the esteem-relevant characteristics of our fellows before an individual has had the opportunity to personally distinguish themselves in one way or the other, we are not providing those whom we encounter with a level playing field. Not only are such attitudes at odds with social equality in themselves, but they also have deleterious downstream cognitive effects. Attitudes of social equality are important because they facilitate what we can term evidential parity. Evidential parity has two parts—one concerning positive assessment, and the other concerning negative assessment. Social equality demands that we ought not (i) to give members of certain groups a heightened benefit of the doubt, or (ii) to tend to interpret evidence as confirming the worst about certain groups over others (Ross 2022: 816).

While we agree with Ross’ verdicts on the significance of what he calls ‘personalized evidence,’ and that some beliefs resulting from demographic profiling will undermine the requirement of equality, we find ourselves in disagreement with the idea that relational equality could never call for giving some groups a heightened (and, in comparison, giving other groups a lowered) benefit of doubt. Ross might be right about our esteem practices under conditions of background justice, but his proposal overlooks that background injustice makes us prone to make certain errors in our esteem judgments depending on the distribution of disadvantage at the group level. In such cases, we have suggested, it may be better, from a relational egalitarian point of view, to steer away from strict neutrality in our evidential standards. Moreover, our argument also reveals that there might be a role for “profiling” evidence for egalitarian minded people. Statistical evidence about how the distribution of opportunity for earning esteem correlate with group membership may help us tell when it is appropriate to raise or lower our evidential standards.

4 Is this Really Relational Egalitarianism?

At this point, some might wonder whether what we have put forward is really a relational egalitarian argument for doxastic affirmative action. One might reason as follows. You, the authors, are in fact addressing a distributive question, namely the question of whether two individuals have had equal opportunities for earning esteem (and, if not, what this means for how we should apportion (dis)esteem). But this is to take an individualized and compensatory view of apportioning (dis)esteem on which (dis)esteem is something to be equalized. A relational egalitarian, on the other hand, would take a collective and expressive view of apportioning (dis)esteem. It is a matter of asking, “which practices do we, as a community, value and disvalue?” We should not esteem or disesteem people because they deserve it, but because it upholds our relational values. This is not, as it is on the individualized view, a matter of fairness. It is about esteeming the right kinds of egalitarian actions and disesteeming the wrong ones. Esteem and disesteem are not something to be distributed equally on the relational egalitarian view, but instead are essential tools for regulating behavior in a relational egalitarian society. In this way, your account of doxastic affirmative action is not really relational egalitarian.Footnote 25

We agree with the objection in the sense that one relational egalitarian view of apportioning (dis)esteem takes the collectivist, expressive form suggested in the objection. We take it that this is the view defended by McTernan (2016; see also 2014).Footnote 26 As she says, “When considering whether and how to incorporate considerations of responsibility into an egalitarian theory, I propose that in each instance one should ask whether this particular responsibility practice is one that egalitarians should find valuable … the responsibility-sensitive egalitarianism I propose is best characterised as fundamentally ‘social’: determining who is responsible and for what requires that we address which forms of responsibility practice have value and, further, one of the central ways in which a responsibility practice can have value is in shaping our social relations” (McTernan 2016: 750–752).

We agree that this is a relational egalitarian view. But we think it is wrong to assume, as the objection does, that there is only one relational egalitarian view of apportioning (dis)esteem. Indeed, relational egalitarianism is not merely one view. There are different relational egalitarianisms. To give an example, Bengtson and Lippert-Rasmussen (2023: 6) distinguish the following two views:

Telic Relational Egalitarianism. It is, in itself, good (bad) if egalitarian (inegalitarian) relationships between people exist.

Deontic Relational Egalitarianism. It is morally required that people relate as equals, not unequals.

These two views of relational egalitarianism are different.Footnote 27 Whereas the former is an axiological view—saying why it is (dis)valuable that relations are (in)egalitarian—the latter is a deontic view which does not say anything about whether (un)equal relations are (dis)valuable. Instead, it says that it is a moral requirement that people relate as equals, and not unequals. As this shows, relational egalitarianism is not merely one thing. And, thus, the objection wrongfully assumes that only the collectivist, expressive view is a truly relational egalitarian view.Footnote 28 But as we have seen, many relational egalitarians—including Kolodny and Ross—care about (in)equality of opportunity. Indeed, they believe that equality of opportunity is constitutive of relational equality. In caring about individuals enjoying equality of opportunity, their view is individualized in the way described in the objection. But this does not mean that their view is not a relational egalitarian view. It just means that their relational egalitarian view is different from the collectivist, expressive view put forward by McTernan.Footnote 29

Moreover, the objection says that a relational egalitarian view is not about fairness. If this is meant to describe the relational egalitarian views “out there,” this is not true. Lippert-Rasmussen (2018a): ch. 7) grounds his relational egalitarian view in fairness. As he understands fairness, “it is unfair if people are differently situated if the fact that they are differently situated does not reflect their differential exercise of responsibility” (Lippert-Rasmussen 2018a: 207). Applied to apportioning (dis)esteem, one can say that it is unfair if people do not enjoy equal opportunity to earn esteem and avoid disesteem because when that is the case, how they are situated in relation to each other will not reflect their differential exercise of responsibility. In this way, one could ground the equality of opportunity argument for doxastic affirmative action in fairness. It is enough for our argument that equality of opportunity is constitutive of relational equality. But it is compatible with adding this fairness component. Thus, insofar as the objection assumes that a relational egalitarian view cannot be grounded in fairness, at least partly because fairness is an individualized notion, it is wrong, as we have just seen. And insofar as it assumes that no actual relational egalitarian view is grounded in fairness, it is wrong, as Lippert-Rasmussen’s view illustrates.

Finally, the objection assumes that a distributive view is individualist whereas a relational view is collectivist. And while we agree that a distributive view can be individualist, and that a relational view can be collectivist, this is not a matter of necessity. The individualist/collectivist distinction cuts across the distinction between distributive and relational views. Indeed, we could imagine a distributive collectivist view which distributed responsibility in line with what would be valuable for the collective. Nothing in principle hinders such a view. And, as we have just seen, there are also individualist relational views. Thus, while we agree with the objection that a relational egalitarian view can take a collectivist form, this is not a matter of necessity. We have defended doxastic affirmative action on an individualist view of relational egalitarianism.

We would like to end on a more ecumenical note. As we explained earlier, our argument, suitably modified, could appeal to relational egalitarians who do not believe that equality of opportunity is constitutive of relational equality. It could do so in the sense that such relational egalitarians might agree that even if equality of opportunity is not necessary, at least race and gender (and similar traits) should not make a difference to how we apportion (dis)esteem. And this is actually all we need to get the argument going. Other relational egalitarians might believe that equality of opportunity is important, but for a different reason than the one we pointed to. They may argue that equality of opportunity is important because of what inequality of opportunity expresses. Inequality of opportunity, they may suggest, expresses that those with worse opportunities are less important, morally speaking, than those with better opportunities. The state thereby sends the wrongful message that those with worse opportunities are less worthy of respect than those with better opportunities. Thus, whereas the expressivist view says that equality of opportunity is needed to express that everyone is equally worthy of respect, the non-expressivist view (the view we have defended) says that equality of opportunity is needed because it is a constitutive part of relating as equals. But both relational egalitarian views care about equality of opportunity, so both views, at least suitably modified, should agree that doxastic affirmative action may be needed, under conditions of background injustice, to secure relational equality. We suspect that this distinction between expressivist and non-expressivist views might be what was actually driving the objection with which we started this section (i.e., that the important difference between McTernan’s view and our view may be that the former is expressivist, whereas ours is not). In any case, the preceding remarks show that our argument—that relational egalitarianism under conditions of background injustice may require doxastic affirmative action—should be of interest to relational egalitarians of different stripes.

5 Conclusion

In this paper, we have suggested that relational egalitarians ought to engage in affirmative action in the allocation of (dis)esteem. We have proposed a plausible model for how this should happen and why it would be desirable. This contribution fleshes out relational egalitarianism as a theory suggesting that equal relations also require that we regard one another in certain ways. Moreover, and surprisingly, we show that affirmative action for relational egalitarians is not only a matter of action, but also a matter of thought.Footnote 30