Relational and Distributive Discrimination

Abstract

Recent philosophical accounts of discrimination face challenges in accommodating robust intuitions about the particular way in which it is wrongful—most prominently, the intuition that discriminatory actions intrinsically violate equality irrespective of their contingent consequences. The paper suggests that we understand the normative structure of discrimination in a way that is different from the one implicitly assumed by these accounts. It argues that core discriminatory wrongs—such as segregation in Apartheid South Africa—divide into two types, corresponding to violations of relational and distributive equality; and that a pluralistic view of the internal structure of each type should be adopted. This normative structure serves to appropriately vindicate the mentioned intuitions; it also contributes to clarifying the normative underpinnings of legal concepts associated with discrimination (including the distinctions between direct and indirect, or ‘disparate treatment/impact’ discrimination, and intentional vs. unintentional discrimination), and developing a more productive public discourse around allegations of discrimination.

Introduction

Discrimination has long been a central topic of legal analysis and literature. By contrast, the purely philosophical reflection on the topic—a reflection that takes intuitions about the morality of discrimination, and not the content of anti-discrimination norms as its point of reference—is a fairly recent development. In fact, recent philosophical interest in discrimination has revealed that despite the apparent consensus about its wrongness, adequately accounting for moral intuitions associated with the phenomenon remains challenging.

More specifically, recent philosophical accounts fall short of accommodating some important intuitions regarding the particular way in which discrimination is objectionable. Common morality, as well as intuitions and assumptions that prevalently underlie the legal analysis of discrimination, maintain that actions representing the core wrong associated with the term—such as governmentally imposed policies of racial segregation—intrinsically violate equality. Moreover, such actions seem wrongful in a particularly discriminatory way, that is, in a way that is tied to their nature and structure. In other words, there are strong and pervasive intuitions pointing to the existence of an equality-related deontological constraint associated with discriminatory actions; specifically, one that isn't in some way derived from, or constituted by their contingent consequences.¹

The present paper argues that in order to plausibly and adequately account for these intuitions, the normative structure of core discriminatory wrongs should be understood in a way that is different from the one implicitly assumed by recent accounts. Namely, I argue that actions representing the core wrong of discrimination divide into two types—corresponding to violations of distributive equality on the one hand, that is, a value concerned with the egalitarian distribution of goods; and of relational equality on the other hand, that is, a value concerned with inegalitarian relations between people, such as those exhibited by illegitimate hierarchies or caste systems. The two types may accordingly be labeled 'relational discrimination' and 'distributive discrimination'. While they often co-instantiate, they are normatively distinct and independent of each other, and may instantiate separately in real-life situations; the paper thus suggests that the division is indispensable to understanding the wrong of discrimination. A case instantiating relational discrimination alone is an action of racial segregation in a public venue (such as a public pool), which has a strong symbolic meaning but no significant distributive effects; a case instantiating distributive discrimination alone is a healthcare policy which relies on incorrect information regarding the medical needs of a certain minority group, thus leading to unjust distributive outcomes without violating relational equality.

Congruent with a pluralistic view of both relational and distributive equality, I also argue that within each type a pluralistic view of the reasons that such actions are objectionable should be adopted; thus, each instantiation of each type may be objectionable due to a variety of reasons associated with each value. What unifies each type is its violation of either relational or distributive equality, and what unifies core discriminatory actions as a whole is their characteristic structure, along with their violation of equality more abstractly understood.

Viewing the normative structure of core discriminatory actions in that manner contributes to adequately and plausibly vindicating the mentioned robust intuitions, and, more generally, developing a clearer and more adequate understanding of the normative nature of discrimination, along with the content of allegations of discrimination. It also contributes to clarifying the normative underpinnings of concepts associated with the phenomenon that are commonly used in legal and public discourse—including the distinction between direct and indirect discrimination (or ‘disparate treatment’ vs. ‘disparate impact’ discrimination), and intentional vs. unintentional discrimination. Particularly, the paper suggests that some important moral intuitions underlying the legal direct-indirect discrimination distinction can be accounted for by the relational-distributive discrimination division suggested here; and that contrary to a common implicit assumption, both relational and distributive discrimination may be performed either intentionally or unintentionally.

The paper is structured as follows. Section 2 presents the conceptual assumptions on the nature of discrimination, and the theoretical assumptions on the nature of equality employed throughout the paper. Section 3 presents the main claims of the paper in detail and walks through central examples illustrating them. Section 4 discusses the relation between the structure of core discriminatory wrongs suggested here and the distinctions between direct and indirect discrimination, and intentional vs. unintentional discrimination. Section 5 concludes.

Conceptual Clarifications and Theoretical Background

Core cases of discrimination

The term "discrimination" refers in common discourse to a variety of objectionable social phenomena, whose normative underpinnings may diverge from one another. This ambiguity creates an ongoing challenge in addressing normative questions pertaining to discrimination.²With this methodological challenge in mind, the paper focuses on a particular subset of cases, with respect to which there is a (relatively) broad consensus that they represent the phenomenon of discrimination in its core sense; I label them 'core cases of discrimination' hereinafter.³Examples are the following:

Segregation: Governmentally-mandated racial segregation in public facilities (such as public pools) and important societal institutions (such as places of education) during the Jim Crow era in the United States;

Prison: In the prison system in Apartheid South Africa, short pants were given to Black inmates, while long pants were given to White inmates. Presumably, the goal behind the policy was to infantilize Black inmates; regardless, allowing inmates to wear shorts may have provided them with a material advantage;⁴

Employment or Educational Discrimination: Rejecting a potential employee or student (or a group of employees or students) based on their membership in a socially salient group—such as their race, gender, or ethnicity.

The conclusions of the discussion are applicable to core cases of discrimination that are not directly discussed here, but not to objectionable social phenomena associated with the term 'discrimination' in common discourse that do not exhibit the same normative traits. In order to simplify the discussion and facilitate the assessment of its scope of application, a definition of discrimination reflecting the nature of these cases will be useful. To this end, I use a (slightly revised) version of Lippert-Rasmussen's⁵ and Altman's⁶ conceptual analysis, which has closely and systematically examined the important features of these cases and the intuitions associated with them.

On their conceptualization, discrimination consists of disadvantageous differential treatment (of an individual or a group of people) which is based on, or related to the victim's membership in a socially salient group. This definition leaves open the possibility that such actions do not disadvantage the victim in any tangible way, that is, by harming them, depriving them of any important goods or opportunities, or subjecting them to an unfair distribution of resources; as will become clear, not all core discriminatory actions involve such tangible disadvantage.⁷

This conceptualization encompasses individual, identifiable actions—performed by identifiable individuals, collective entities or institutions—which directly wrong identifiable individuals or groups of people. Thus, I leave out of the discussion notions such as structural or institutional discrimination (to the extent that these conflict with the mentioned features), and expansive understandings of the term which include, for instance, social disparities that are not traceable to identifiable wrongful actions. It is also important to clarify that this basic conceptualization is non-moralized: it reflects the nature of actions that are potentially (or are suspected to be) wrongfully or objectionably discriminatory, but does not imply that every instance of disadvantageous, group-based differential treatment is wrongful. Rather, an additional feature is necessary for such actions to be an instance of wrongful discrimination; in line with the mentioned intuitions, I assume here that this additional feature can be generally characterized as their violation of equality (or people’s equality-related rights; this is further discussed below).

Three additional clarifications are needed. First, I use the term 'treatment' to refer to actions (including laws and policies) in a broad sense, to include, first, the underlying mental states or attitudes of the agent; and second, the action's immediate, non-contingent consequences, or those consequences that are naturally understood to be part of the action itself (in line with common terminology in discussions of discrimination, I refer to these as the action's 'effects', 'impact' or 'outcome').⁸ Second, the definition allows for a variety of ways in which the differential treatment may be based on or related to the victim's group membership. Particularly, it isn't limited to cases where the agent explicitly refers to the discriminatee's group membership in her reasoning (or where the rationales underlying a law or policy make such explicit reference); it extends to cases where, for instance, the action is causally influenced by a discriminatory mental state, or where there is correspondence between particular outcomes of a law or policy and the group identity of the people affected by it.

A final point concerns the group-based nature of discrimination. While the wrong of discrimination is commonly thought to relate to the victim's group membership, it remains controversial or unclear which groups are relevant to constituting wrongful discrimination. Here too, I adopt Lippert-Rasmussen's and Altman's view and assume that core cases of discrimination necessarily involve groups that are socially salient, that is, important across a wide range of social interactions. These may include ethnic, racial, gender-based, and other groups that are socially important in the relevant society; but not groups such as people who were born on Tuesday, or people with weird earlobes.⁹

There are open theoretical questions about which features of these groups make them socially salient, and what can ground or explain their importance in general, and in the context of discrimination (or particular discriminatory actions) in particular. One's positions on these larger questions may be tied to her judgments about whether particular (presumably, not clear-cut) instances of discrimination are wrongful; however, addressing them in full is beyond the scope of the present discussion. As with other components of the conceptualization of discrimination, then, I focus here on clear-cut cases of social saliency and rely mainly on the intuitive plausibility of the common, persistent thought (closely examined by Altman and Lippert-Rasmussen) that core discriminatory actions necessarily involve socially salient groups, without inquiring 'all the way down' into its justification. Thus, the discussion is open to a variety of views about which features of socially salient groups ground or explain their importance in the context of discrimination. These may include, for instance, members' possession of a distinct culture, set of attitudes, or meaningful commitments associated with their group membership; group membership playing an important part in determining its members' level of welfare; or everyday activities and projects being organized along group lines.

One important point of clarification is that past or present societal patterns of inter-group inequality or discrimination are not a necessary background feature in constituting a socially salient group (although the discussion is open to the possibility that this feature plays a part in constituting some socially salient groups). In line with that, I assume that discriminatory actions may be objectionable when they are directed at groups that are not at the lower end of some existing societal inegalitarian pattern or socioeconomic scale, and have not suffered from past discrimination or oppression. Discriminatory actions taking place between, say, members of different groups that are similarly positioned on such a scale, or directed at relatively well-off groups, may constitute objectionable discrimination as well (while discriminatory actions directed at, say, groups that have been subjected to persistent discrimination may be morally worse than these). In other words, while individual discriminatory actions are assumed here to involve a violation of an equality-related duty, this violation is independent of any pre-existing inegalitarian patterns in the society in which they take place (as will become clear, the moral objection to such actions is also independent of their contribution to larger inegalitarian patterns).¹⁰ This leads us to the next sub-section, which clarifies the paper's theoretical assumptions about equality.

Relational and distributive equality

The discussion here assumes a set of theoretical positions about the nature and normative significance of equality. These are meant, first and foremost, to accommodate the intuitions associated with core discriminatory actions; they thus may diverge from, or be in tension with some views defended in the broader egalitarian literature about the value of equality more generally. Here too, fully examining the theoretical basis (and larger implications) of these positions is beyond the scope of the present paper;¹¹I illustrate throughout, however, that they are plausible particularly in light of the robust and persistent intuitions associated with discrimination.

Extensive discussions have been devoted in the philosophical literature to the value of equality over the last several decades. In particular, egalitarian philosophers have focused on examining two different notions associated with it, or two different interpretations of equality more abstractly or generally understood. The first views equality as a distributive ideal: it maintains that equality is concerned with the distribution of certain goods, such as income, opportunities, resources, welfare, capabilities, etc. Violations of equality emerge, on this view, where the distribution of these goods does not conform to principles of just or egalitarian distribution; for instance, where the distribution of resources in a given society results from circumstances that are beyond people's control, or is not 'patterned' along features such as merit or responsibility. The second (and more recent) view maintains that equality is concerned not (or not primarily) with the distribution of goods, but rather with egalitarian societal relations; on this view, phenomena that are in tension with equality do not necessarily involve some objectionable distribution of goods, but rather things such as social exclusion, illegitimate hierarchies, oppression of certain groups or individuals, societal marginalization, and so forth.¹²

In the broader egalitarian literature these two views of equality are sometimes considered to be in tension with one another, or, at least, to generate conflicting judgments about some important contemporary societal or political questions, such that one should be preferred over the other.¹³As will become clear in Section 3, however, discriminatory actions do not typically give rise to such tension or competition between these two views of equality; on the contrary, intuitions associated with them give rise to the thought that both are independently needed to fully and adequately account for the moral objection associated with this particular phenomenon.

In line with these intuitions, then, I assume that these two notions are not competing interpretations of equality, but rather more particular views of it; both are associated with equality more abstractly or generally understood (whereas the precise relation between them is an open question). I assume further that relational and distributive equality are conceptually distinct and independently normatively important; that is, considering the nature of the cases examined below, I reject positions to the effect that objections to discrimination that are based on either of these values can be adequately restated using the other, or that when either of these values is fully laid out it emerges that it reduces to the other.¹⁴This is notwithstanding that there are often complex inter-relations between violations of these two notions of equality in real-life examples of discrimination; as illustrated in Section 3, these should not obscure the fact that the two are conceptually and normatively distinct.

Other theoretical assumptions are tied with the particular judgments associated with these two values in the appraisal of individual, identifiable discriminatory actions. As mentioned, persistent intuitions indicate that there is an equality-related deontological constraint associated with these actions, which is tied with the group-based differential treatment they involve.¹⁵By contrast, discussions in the broader egalitarian literature focus primarily on making evaluative claims about states-of-affairs, and not on deontological constraints associated with actions' nature. Thus, debates about distributive equality usually focus on the overall distribution of certain goods in a given society (or 'end-state', 'patterned' distributions), and discussions of relational equality focus on larger patterns of oppression or exclusion; to the extent that these discussions identify duties associated with particular actions, these are commonly thought to be derived from the action's influence on these larger inegalitarian states-of-affairs.¹⁶Contrary to this focus and in line with the intuitions associated with discrimination, then, I assume that there are deontological constraints generated by both values, making individual, identifiable discriminatory actions objectionable in a way tied to their particular nature and structure; that is, which are not derived from these actions’ contingent consequences, or those consequences that are not naturally understood to be part of the action itself.

It is particularly important to highlight, then, that discriminatory actions are intuitively objectionable regardless of whether they contribute to larger inegalitarian patterns in a given society (or another population). For instance, Prison seems wrongfully discriminatory regardless of whether it contributes to larger patterns of oppression or marginalization (while it may certainly be more severe if it also contributes to such patterns). Similarly, distributive discriminatory actions seem objectionable because of their immediate distributive consequences—e.g., the way an important societal institution distributes its budget or resources—and regardless of their possible contribution to larger patterns of distributive inequality. In fact, these immediate distributive consequences may be entirely isolated from larger distributive patterns, and discriminatory actions may be intuitively objectionable even in cases where their non-immediate, contingent consequences ultimately lead to more egalitarian larger distributive patterns.¹⁷

Consider, for instance, a selection process for an orchestra whose implementation leads to the unjustified rejection of qualified female musicians. Such a policy may have little to no influence on larger patterns of gender-based distributive inequality (and might, in some cases, lead to narrowing down gender income gaps in the long run, by incentivizing women to choose more lucrative professions). Nevertheless, it plainly wrongfully discriminates against women in virtue of its immediate distributive consequences—those that are naturally understood as part of the policy itself. Similar things can be said about employment or educational discrimination against well-off minorities.¹⁸

The adoption of these positions leads to some other assumptions that are important to highlight; note, however, that while the discussion favors the positions laid out below, it is compatible with a wide range of views about the matters discussed. First, it favors a particular view regarding the ‘currency’ of distributive justice (in the context of discrimination): individual, identifiable discriminatory actions typically involve the distribution of resources (that is, the action of distributing resources)—and not things such as welfare or capabilities, which have been offered as the appropriate 'currency' of distributive justice in discussions of larger egalitarian patterns. Employment or educational discrimination involves, for instance, distributing positions in the relevant institutions; other prominent contexts of discrimination involve the distribution of housing or medical resources, and, more generally, governmental budgets and benefits (from within a limited pool of the relevant resource).

Second, a pluralistic or complex view of both relational and distributive equality seems adequate in the moral appraisal of individual discriminatory actions. Thus, actions taking place in different contexts of distribution seem objectionable due to a violation of different distributive principles: while employment or educational discrimination may be objectionable because it violates principles of equality of opportunity or merit-based distribution, discrimination in housing or medical contexts may be objectionable because it doesn't conform to principles of need-based distribution. Similarly, relational discriminatory actions taking place in different contexts may be objectionable for a variety of reasons associated with this value. These particular objections are further discussed and illustrated below.

The Division between Relational and Distributive Discrimination and the Structure of Core Discriminatory Wrongs

The main claim

Recent philosophical accounts of discrimination face challenges in adequately vindicating the robust, persistent intuitions discussed above. Thus, accounts focusing on one particular equality-based explanation—such as the claim that discrimination is objectionable because it violates equality of opportunity, or because of its degrading social meaning¹⁹—match the intuition that discriminatory actions are intrinsically wrongful, but fail to accommodate central cases that do not satisfy these conditions.Other prominent accounts maintain that the wrongness of discrimination is derived from its objectionable contingent consequences, thus falling short of accounting for the intuition that individual, identifiable discriminatory actions are intrinsically wrongful (or that there is a deontological constraint associated with them) irrespective of such consequences.²⁰

To plausibly account for these intuitions, I suggest that we understand the normative structure of core discriminatory wrongs in a way different from the one implicitly assumed by these accounts. Namely, I argue that core cases of discrimination divide into two distinct and independent types, corresponding to violations of relational and distributive equality. Thus, any action exhibiting the core wrong of discrimination is objectionable either for reasons pertaining to relational equality, or distributive equality, or both at the same time; but contrary to what may be a common conceptual assumption, such actions do not have to violate both values—for instance, to involve the deprivation of a position in a place of employment, in a way related to a discriminatory mental state of the agent—for them to be objectionable in the manner characteristic of core cases of discrimination. Rather, the fact that the two types tend to co-instantiate together in one action might be due to historical or sociological contingencies that do not necessarily coincide with the nature of the wrong of discrimination; as illustrated below, in such cases the two types may be disentangled from one another. The suggested structure entails, then, that the division between relational and distributive discrimination is indispensable to understanding the nature of core discriminatory wrongs.

In line with the pluralistic view of both relational and distributive equality, on the suggested structure each instantiation of each type may be objectionable for a variety of reasons associated with each value; particularly, it is plausible that discriminatory actions performed in different contexts and by different types of agents are objectionable for different reasons associated with those values. These particular reasons are briefly discussed below and illustrated throughout the paper. Here too, fully examining the justification of each particular claim is beyond the scope of the paper; however, one may reject any of them (e.g., the validity of a particular distributive principle suggested, or the claim that mental states matter for the moral status of certain discriminatory actions) without rejecting the paper's main claim about the general structure of core discriminatory wrongs.

Similarly to distributive discrimination—where, as already mentioned, different distributive principles are applicable in different contexts—relational discriminatory actions may be objectionable for a variety of reasons. Those performed in the course of interpersonal interactions, for instance, may be objectionable due to agents' inegalitarian mental states, attitudes or motivations;²¹ these may include stereotypes,²² bias,²³group-based hatred, malicious intentions or animosity directed towards the group, etc.In other contexts, other relational equality-related objections seem appropriate. For instance, laws or policies promulgated by important political or societal institutions, which openly declare that a particular group will not be included in some important societal activity or venue, or has to wear humiliating clothing, seem wrongfully discriminatory because they have an objective-social meaning (or expressive traits) which degrades the group, or because they are disrespectful in a particularly discriminatory way (as claimed by Hellman and Eidelson respectively, regarding discrimination in general²⁴). Similarly, such laws or policies may have a degrading objective-social meaning, or be disrespectful, because they have the (immediate, non-contingent) effect of socially excluding a group from an important social venue or activity.

Hence, it is suggested that contrary to what seems to be implicitly assumed by recent philosophical accounts, we should not necessarily be looking for a single normative explanation that would unify all core cases of discrimination. Rather, what unifies each type is its violation of either relational or distributive equality; and what unifies core discriminatory actions as a whole is their characteristic structure (involving the differential treatment of people in a way that is based on, or related to their membership in a socially salient group), and their violation of equality in its more abstract or general sense.

The next sub-section illustrates the plausibility of these claims by closely examining a variety of central examples of core cases of discrimination. Sub-section B(i) focuses on showing that notwithstanding a common thought on which the two types suggested here must co-instantiate as a matter of conceptual necessity, they can and do instantiate separately—while still exhibiting the core wrong associated with discrimination. Sub-section B(ii) shows that in complex cases of co-instantiation, where both values are violated and there are interrelations between these violations, the two types may be disentangled from one another.

Illustration through cases

Cases instantiating only one type

Cases instantiating relational discrimination alone may be usefully divided into three categories. The first category includes cases where the action does not involve any significant distribution of goods; an example is not yielding to a Black pedestrian at a crosswalk (while others are yielded to).The second category includes actions that do involve the distribution of goods, but where the agent does not have a duty to abide by principles of just distribution, at least on some views of distributive equality. The most straight-forward examples involve individuals operating in their private capacity: a person who refuses to sell her bike to members of a certain group, or a small business owner who refuses to hire them.²⁵

The third category includes discriminatory actions that do involve the distribution of goods, where the entity distributing them does have a duty to conform to principles of just distribution—but where there is no violation of such principles, or where the unjust effects may be assumed away while the moral objection to the action is retained. Prominent examples include legally mandated racial segregation in public venues such as swimming pools or schools; and instances where a resource is being distributed, but the distribution is significant for symbolic reasons only—such as Prison. All of these actions seem unobjectionable from a distributive perspective, but still objectionable in a particularly discriminatory way—which is attributable to their violation of relational equality.

In the opposite direction, there are cases where an agent brings about an unjust distributive outcome without thereby engaging in relational discrimination. Prominent examples involve governmental agencies and other important political or societal institutions, which have a duty to abide by principles of distributive equality and do attempt to abide by them, but are mistaken about the distributive principles applying to the situation or the right balance between them. Consider, for example, policies such as the one discussed in the U.S. Supreme Court case of Personnel Administrator of Massachusetts v. Feeney.²⁶ There, a strong preference was given to military veterans in the selection for civil service positions, which resulted in the almost complete exclusion of women from these positions. Several valid (but conflicting) distributive principles or considerations seem to be at issue in that case: for instance, compensating veterans for their service on the one hand, and equality of opportunity for women on the other hand. Such a policy may be wrongfully discriminatory because it fails to appropriately balance these considerations, resulting in a distributive outcome that unfairly deprives women of job opportunities. Because of the challenging balance between different distributive considerations involved in the design of such policies, then, they may bring about unjust outcomes without being underlain by anti-relational equality attitudes or rationales, and without policymakers being motivated or influenced by such attitudes; in such cases, the policy would be objectionable only as an instance of distributive discrimination (while exhibiting the core wrong of discrimination).

Similarly to the case just described, the design and implementation of large-scale, highly complex and technical policies may result in outcomes that are distributively unjust even if they are not underlain by any anti-relational equality attitudes (including lack of due care for certain groups' interests). A plausible example involves governmental agencies acting on inaccurate or incomplete information simply because there is not enough reliable information or expert knowledge about issues relevant for determining the needs of a certain group in the context of a particular policy. Think, for instance, of a medical policy involving the distribution of certain resources (e.g., medical treatment or drugs), whose design takes place at a point of time where there is not enough research about the medical needs of a particular minority group (for example, about genetic diseases giving rise to needs diverging from those of the majority). Plausibly, the implementation of such a policy may result in an unfair disadvantage for group members without it being underlain by anti-relational equality attitudes. Due to the technical nature of such policies, the objection to them does not seem tied with any degrading meaning or disrespectful traits either; they are objectionable only for their violation of distributive equality. Another example of a policy instantiating distributive discrimination alone is discussed below.

Complex cases of co-instantiation from judicial decisions

Cases of discrimination that are prominent in legal discussions—normally these are laws or policies with law-like structure—often take a characteristic complex form. Namely, on the terms suggested here, they instantiate both relational and distributive discrimination, while each type is objectionable for a variety of reasons associated with each value, and where there are complex interrelations between the violations of the two values. Additionally, because such cases are typically discussed within the framework of anti-discrimination norms, their analysis in common discourse tends to conflate considerations relating to the normative nature of discrimination with considerations relating to its desirable regulation by legal norms, as well as the adjudication of claims of discrimination in legal proceedings (one of the latter type of considerations—relating to the evidentiary role of distributive discrimination—is discussed in Section 4a).

This characteristic complexity, along with the over-representation of such cases in discussions of discrimination, tends to skew the normative analysis of the phenomenon (in legal, philosophical, and public discourse) in several ways that are detrimental to illuminating the claims made here regarding the structure of core discriminatory wrongs. Thus, the present sub-section illustrates how these claims play out in such cases, while highlighting some possible sources of confusion. Particularly, I show that even in such complex cases the two wrongs may be disentangled from one another, that is, the former are objectionable for two distinct and independent groups of reasons associated with either relational or distributive equality. I do that by closely examining the landmark U.S. Supreme Court case of Griggs vs. Duke Power Co.²⁷

The policy in Griggs involved a general competency test used by a company to screen out job candidates; applying it resulted in screening out a disproportionate number²⁸of African-Americans. As background, the Court notes that African-Americans have suffered from prior discrimination resulting in inferior educational credentials, and that the policy under examination had been preceded by an overtly discriminatory policy, that is, one that openly declared that African-Americans will not be considered for the job.

As a case of relational discrimination, the policy may be objectionable due to the inegalitarian mental states or attitudes that the agents responsible for it hold, or are reflected in its underlying rationales. For instance, it may be underlain or motivated by a belief that African-Americans are, as a group, inferior to White-Americans, or by a desire not to associate with the group. Similarly, the policy may be objectionable because it reflects either consciously or unconsciously held objectionable stereotypes of African-Americans, or prejudice—feelings of hate and resentment—towards the group. Also within the group of relational equality-based objections, it may be argued that, regardless of the presence of such mental states, the policy has a degrading social meaning or is disrespectful towards African-Americans as a group—perhaps because, considering the cultural and historical context in which it takes place, it can reasonably be interpreted as reflecting anti-relational equality attitudes or as degrading, or because it results in the social exclusion of the group.

Considering the legal context in which the analysis of such cases usually takes place, it is important to note that notwithstanding what may be implied by a common formulation of legal anti-discrimination norms, the policy being 'facially neutral'—that is, not containing any explicit reference to the group identity of the alleged victims, or openly proclaiming the content of any anti-relational equality attitudes as its underlying reasoning or aim—does not necessarily mean that it is not objectionable as an instance of relational discrimination; nor do these facts have any necessary implications as to whether this would be an overt or covert form of relational discrimination, if it is indeed a case of relational discrimination. Rather, these features would vary depending on the nature of the relational equality-based objections that are applicable to the case. For instance, if in a certain context the underlying inegalitarian attitudes of agents responsible for a policy is plausibly what determines whether it is a case of relational discrimination, it would be an instance of relational discrimination depending on the existence of such attitudes, and an overt or covert form of relational discrimination depending on features such as whether these attitudes are openly stated or consciously employed—irrespective of the wording of the policy itself. As mentioned, in some contexts it may be that the policy’s degrading objective-social meaning is what plausibly determines whether it is an instance of relational discrimination; this feature too may be independent of the policy’s wording or openly proclaimed aims.

Independently and separately of these claims, the policy may be objectionable for a variety of reasons pertaining to distributive equality. For instance, if the test used by the company to screen out candidates does not adequately implement any reasonable requirement of competency (e.g., because the skills measured are not relevant for the job), the policy may be objectionable because it doesn't conform to principles of equality of opportunity, or of merit or desert-based distribution—in a way that disadvantages African-Americans in particular, by unjustly depriving them of positions in the company. Here too, such policy may lead to unfair outcomes even if the people responsible for putting it together do not hold any anti-relational equality attitudes. For instance, they might compose a competency test which is culturally biased, that is, inaccurately measures the skills of a particular cultural group, due to being unaware of certain educational practices prevalent among its members; the test may also fail to identify relevant, unique skills that are not common among the broader society. In such cases, distributive discrimination may instantiate independently of relational discrimination, despite the common relations between the two types in employment and educational contexts (e.g., where prejudice leads to rejecting group members from relevant institutions).

Another distributive equality-related reason to object to the policy may be that it is a special case of compounding injustice (that is, one that takes a particularly discriminatory form²⁹). As the Court notes, the lower scores of African-American candidates in the competency test may be attributable to a prior distributive injustice which resulted in inferior educational credentials; the outcomes of this prior injustice are then used to evaluate the candidate’s present competency—or as an evidentiary input in a present policy involving another, independent distribution of resources—which amounts to compounding the prior injustice. In such cases the policy may unfairly disadvantage candidates from a certain group even if the test accurately reflects their current competency; here too, then, it may lead to a distributively unjust outcome without violating relational equality at the same time.

A clarification relating to a common formulation of legal anti-discrimination norms is needed here as well. Norms pertaining to indirect discrimination (or ‘disparate impact’ in American jurisprudence) commonly maintain that a law, policy or practice are indirectly discriminatory if they (1) have the effect of disproportionately disadvantaging a socially salient group (protected by these norms); and (2) the effect is not reasonably related to a legitimate goal of the policy. The first condition is commonly understood to imply that the share of members of the protected group among those disadvantaged as a (direct) result of the policy is higher than their share among the natural 'recipients' of the goods distributed (for instance, among the pool of candidates to a place of employment). The satisfaction of the conjunction of these conditions, while not identical with the outcome being distributively unjust towards the group, is indicative of this feature in a range of cases that typically come before courts. For instance, in employment contexts it would indicate that the policy does not comply with principles of merit-based distribution, in a way that disadvantages members of a particular group. I return to this point in Section 4a.

Finally, note that the complex inter-relations that might exist between distributive and relational discrimination in such cases should not obscure the fact that the two wrongs are analytically distinct, and independent of each other. For instance, the fact that applying the competency test results in a distributive outcome that is unfair towards African-Americans might, at least in some circumstances, be indicative of the presence of anti-relational equality attitudes among those responsible for the policy, such as their desire to exclude the group, not associate with its members, or deprive them of resources.³⁰As noted above, however, such an outcome is not conclusive evidence of the existence of such attitudes, and, in any event, the former is objectionable independently of the latter. Similarly, cases where a group is excluded from participation in an important societal venue may be objectionable due to reasons associated with both values at the same time, as they may involve a deprivation of certain resources along with social exclusion; but the two objections are distinct and independent of each other. Lastly, while acting on certain negative stereotypes (e.g., associating certain groups with incompetence) might have distributively unjust effects (e.g., depriving group members of employment), such actions may be independently objectionable as an instance of relational discrimination.

Other Prominent Distinctions Pertaining to Discrimination

Direct and indirect discrimination

In many legal jurisdictions around the globe, anti-discrimination jurisprudence employs a distinction between direct and indirect discrimination norms (the latter are often traced to the U.S. Supreme Court ruling in Griggs vs. Duke Power just discussed). These norms vary in their precise formulation and the way they are implemented in the adjudication of particular cases. Generally, however, direct discrimination norms prohibit the explicit designation of socially salient groups (protected by anti-discrimination norms) for disadvantageous treatment. By contrast, indirect discrimination norms prohibit laws, policies, or practices which are 'facially neutral'—that is, that do not involve such explicit designation—but nevertheless: (1) bring about a disproportionate disadvantageous effect on group members; and where (2) the effect is not reasonably related to a legitimate goal of the policy.³¹In American anti-discrimination jurisprudence, the notion of ‘disparate impact’ largely corresponds to the formulation of indirect discrimination, when ‘disparate treatment’ corresponds to that of direct discrimination, while adding a condition of intentionality; I discuss this condition in the next sub-section.

Anti-discrimination norms are commonly associated with the value of equality, and their aim is generally understood to involve prohibiting actions that violate equality (or people's equality-related rights). As can be learned from the described formulations, however, these norms do not specify any more particular rationales, or a particular interpretation of equality they aim to protect. Instead, they put forward concrete rules whose implementation would, presumably, track cases of differential treatment that violate equality. Thus, direct and indirect discrimination norms are sometimes understood in legal analysis and literature as aimed at prohibiting (or tracking in their implementation) two distinct types of discrimination; however, the normative underpinnings of these two types have remained obscure. The argument developed here may serve to clarify these normative underpinnings: particularly, it seems that the usage of these two formulations in the adjudication of cases of discrimination would tend to map onto the normative distinction between relational and distributive discrimination.³²

Thus, applying the rule specified by indirect discrimination norms would tend to prohibit, or declare illegal or unconstitutional, cases of distributive discrimination: as discussed in the previous section, the satisfaction of the conditions specified by indirect discrimination norms is indicative of the presence of distributive discrimination in a variety of typical legal cases. Indeed, indirect discrimination is commonly—albeit often implicitly and vaguely—associated with violations of distributive principles in legal analysis and literature.³³By contrast, cases satisfying the condition specified by direct discrimination norms tend to be instances of relational discrimination: this condition is commonly interpreted as a requirement of explicit designation underlain by inegalitarian attitudes or rationales in particular³⁴ (while such cases would commonly also involve a distributively unjust outcome, or a tangible disadvantage).

Note, however, that the direct-indirect distinction is not always understood to be aimed at tracking or prohibiting two distinct types of discrimination; another prominent view is that indirect discrimination norms are aimed at uncovering covert instances of discrimination in legal proceedings.³⁵This thought is driven by the assumption that discriminators motivated by inegalitarian attitudes may hide these by adopting policies that lead to excluding the targeted group without openly stating any objectionable, anti-relational equality aims (e.g., by adopting general tests such as the one discussed in Griggs). Thus, as discussed in the previous section, distributive outcomes that are patently unjust, or involve the almost complete exclusion of a group from a certain venue, are sometimes (but not always) evidentiary of the fact that the agents responsible for the policy are motivated by anti-relational equality attitudes. This way of understanding the function of indirect discrimination norms seems to assume, then, that there is only one type of discrimination, which necessarily involves violations of both relational and distributive equality at the same time (on the terms suggested here); and then uses the latter as evidence of the former.³⁶

Intentional and unintentional discrimination

Apart from the distinction between direct and indirect discrimination, the literature and legal analysis contain a prominent distinction between intentional and unintentional discrimination. As mentioned, in (parts of) American anti-discrimination jurisprudence the explicit designation of a protected group has to be intentional—or, to use other common terms, constitute intentional discrimination, or be accompanied by discriminatory intent, aim, or motive—to qualify as an instance of disparate treatment (and come under the purview of disparate treatment norms). More generally, direct discrimination is often associated with intentionality, whereas indirect discrimination is associated with the feature of being unintentional.³⁷A related, less prominent distinction is between conscious and unconscious discrimination.³⁸ There is much vagueness as to the meanings of these terms; it isn't clear which actions would constitute, for instance, intentional discrimination, or unconscious discrimination.³⁹

Continuing the discussion from the previous sub-section, that is, assuming that some of the moral intuitions underpinning the direct-indirect discrimination distinction can be accounted for by the relational-distributive discrimination division suggested here, it should be clarified that there is no reason to view relational discrimination as more closely affiliated with intentionality, as compared to distributive discrimination.⁴⁰ Rather, both types may be performed either intentionally or unintentionally; the intentional-unintentional distinction cuts across the relational-distributive distinction, and similar things can be said about the conscious-unconscious distinction.

Important examples of unintentional relational discrimination (which may also be unconscious) involve agents who unintentionally act on unconscious bias consisting of mental states or attitudes with inegalitarian, degrading content. For instance, an agent's unconscious association between 'Black' and negatively-valenced concepts (e.g., 'Bad', 'Angry') may lead her to act in a less friendly manner around Black conversational counterparts relative to White ones, or not yield to a Black pedestrian at a crosswalk; as discussed above, such behaviors are degrading, but do not involve any significant distribution of resources. Common examples of intentional distributive discrimination, on the other hand, involve cases such as Feeney (discussed above), where the need to balance a variety of conflicting distributive considerations (and possibly, the interests of more than one socially salient group) may lead to unjust distributive outcomes. Assuming that the relevant entity intends to bring about the distributive outcomes of its policy and is mistaken about the right balance of considerations, the policy would constitute intentional distributive discrimination.⁴¹Note that this type of examples indicates that contrary to a common thought, the fact that a distributively unjust outcome is intentionally brought about does not necessarily indicate that the policy is motivated by (either conscious or unconscious) anti-relational equality attitudes or intentions (including attitudes such as indifference or lack of due care for different groups' interests); as discussed above, the entity responsible for the policy may be genuinely mistaken about the right balance of considerations (or act on inaccurate information).

By the same token, distributive discrimination may be performed unintentionally but with foresight, or completely without foresight, awareness or consciousness (but possibly in a way that involves negligence, or lack of due care for group members’ interests). For instance, again in a case such as Feeney, a governmental agency may bring about a distributive outcome that is unfair towards women due to its overlooking of women's interests or not paying due attention to possible disadvantageous consequences for women's employment. Alternatively, this might happen unintentionally but foreseeingly, for instance, if the agency predicts that the policy is going to unfairly disadvantage women, but decides to prefer conflicting considerations of compensating veterans (while, supposedly, intending the latter but not the former); or without awareness or consciousness, for instance, where the agency is unaware that the policy burdens a particular group until further information is gathered about its consequences.

Conclusion

This paper has illuminated an important normative distinction between two types of discrimination, corresponding to violations of relational and distributive equality. The distinction may have been overlooked in existing literature due to a variety of factors, including the tendency of the two types to co-instantiate in real-life situations, along with the dominance of the legal analysis of discrimination in the discourse about the phenomenon.

Beyond its contribution to developing better philosophical accounts of discrimination and clarifying the normative underpinnings of legal anti-discrimination norms, integrating the distinction into societal and political discourse may improve debates surrounding allegations of discrimination. Such debates often involve a dynamic where important institutions are found to have perpetrated distributive discrimination, and are then vaguely accused of discriminating against a certain group. Due to the tendency to conflate the two types, the accusation is often understood to refer particularly to relational discrimination; and because this type of discrimination is often implicitly viewed as more severe, or one for which the agent is more blameworthy,⁴² this sometimes leads to a backlash against the accusation, and to equally confusing counter-claims on which the institution did not engage in discrimination at all, or that the discrimination wasn't intentional (where the latter claim is presumably premised on the association of relational discrimination with intentionality). Such an unproductive dynamic may be mitigated by clearly specifying the type of discrimination that the accusation refers to in a way informed by the paper's arguments.