(Case 1) On the evening of June 17, 2015, Dylann Roof burst into the Emanuel African Methodist Episcopal Church and killed nine people, all African Americans. Roof’s despicable action was not the result of a sudden and uncontrollable outburst that randomly targeted innocent churchgoers. Roof was, by his own admission, a white supremacist: on his website one could find a manifesto containing his (negative) opinions concerning African Americans and other racial groups as well as pictures of himself wearing a jacket with two white supremacist emblems: the flags of former Rhodesia and that of Apartheid-era South Africa. There was also a photo of Roof with a handgun and the Confederate flag. His actions on the day of the shooting, were planned and his goal was to ignite a civil war (Mindock 2015; Costa et al. 2015).
This is a clear, undisputable case of racism. Agreements on claims concerning racism are not always so easy to obtain. Take the following cases:
Getting clear on what racism is, is thus an important issue. But we should not be overly optimistic about the prospect of attaining agreement of the definition of racism. One reason for that is that the meaning of racism has evolved over time and needs to cover phenomena that it was not intended to cover initially. Some might not be aware of these new changes of meaning or might resist them for theoretical reasons (one such reason to resist change is the perceived danger of conceptual inflation, that is, the possibility that the concept will encompass so many different things which are different from each other that it ends up meaning “nothing”). Another reason is that people working on racism and providing definitions of the phenomena might have different research agendas that make it impossible to agree on such a definition. For instance, historians and sociologists might hold that society-specific cultural factors contribute to producing specific forms of racisms. If such is the case, racism is likely to take different forms, not necessarily being unified by much. Others, for instance moral philosophers, might be more interested by what is wrong with racism and will be quite happy to work with a generic version of racism that encompasses historic and sociologic variations.
In the following, I will present the different forms of racism that have been identified in the literature. I’ll use this as a background for the discussion of the different philosophical accounts of racism. Philosophical accounts have generally a double objective: to capture the common (or core) use of the concept and to correct incorrect uses. As I will show, no account can pretend to capture all phenomena described as “racism” in the literature. I’ll consider solutions to this problem.
If we are interested by racism, it is because of its negative effects on some individuals and on some groups. If racism never had led to any actions, had not created and perpetuated inequality between groups, it is uncertain we would care about it as much. It is not that there is a necessary link between racism and actions (as we will see later), but rather that because of its effects that racism has become important to track and to understand.
To say that someone or something is racist raises a flag or attracts attention to a kind of moral ill related to race. Claims of racism usually serve to deplore, denounce, or condemn a situation, an idea, an individual, and so forth. They are also ways to raise consciousness about the effects of some practices or behaviors on others. Often times, they are also a way to ask for a change (and/or a reparation from) either in the person (or group) who is accused of racism or in the situation that is qualified as racist. So racism is not a mere descriptive concept, it is a normative one. For instance, to say that an idea or a joke is racist is not just to say that it belongs to a class of ideas or a class of jokes (like one would say that this is an existentialist’s idea or that this is a blonde joke). To say that an idea or a joke is racist is also to negatively evaluate that idea or that joke. As Arthur (2007) puts it, the concept of racism is a bit like the concept of “terrorist” or “genocidal”: “the evaluative conclusion … is embedded in the concept itself… . Once we define a person as a terrorist, we have committed ourselves normatively” (12).
I have said that the concept of racism is a normative concept used to refer to a kind of moral ill, but we might want to be more specific. Racism is a moral ill that happens to an individual by virtue of the fact that he is perceived as belonging to a racial group (hereafter, a racialized individual), or to a group that it is perceived as a racial group (hereafter, a racialized group). So if a comet were to strike the earth and in the process eliminated a racialized group, this would not count as racism because it is not an ill that is caused to the group by virtue of being racialized; the latter is just bad luck happening to a racialized group. For something to count as racism, the fact that the individual or the group is racialized has to be a causal factor in the production of the event. But we might want to leave open until later the question of knowing if all moral ills that happen to individuals or groups by virtue of the fact that they are racialized should be labelled racism (Blum 2002a, 2002b, 2004; Garcia 2010).
Contemporary social sciences and philosophy have identified many different forms of racism. Though there are sometimes disagreements concerning the existence or the relative importance of these different forms of racism, I think it’s worth mentioning them because they are the background against which the philosophical discussion takes place. I’ll follow Zack’s (2003; see also Atkin 2012) classification of the different forms of racism in what follows.
The first form of racism is classic or mens rea racism (sometimes called “overt racism”). Case 1 is an illustration of that kind of racism. It is grounded in a negative attitude (of hatred or contempt) or a combination of such attitudes (for instance, a mix of hatred and contempt) towards another racialized group that results in intentional actions/or inactions whose goal is to harm in one way or another (for instance, through violence, insults, or discrimination) racialized individuals or the racialized groups as whole.
In many countries, norms now favor the open expression of more egalitarian ideas and chastise the open expression of prejudice. While the extent to which this is true varies, old-fashioned, overt racism is nowadays frowned upon in many societies, enough that many people would not dare to openly express such attitudes. This has led racism to go underground, that is, to operating more covertly. Philosophers as well as researchers in psychology and social sciences have proposed different concepts seeking to capture the new forms racism is taking; I will mention a few in what follows.
Ikuenobe (2010) proposed two variations of classical racism: closet and tolerant racism. A closet racist is someone that “has the relevant racist beliefs and attitudes but does everything possible consciously to conceal them, but those racist attitudes and beliefs may be expressed occasionally, albeit inadvertently and unconsciously” (172). A closet racist has what psychologists call an “external motivation” not to discriminate or to respond without prejudice against other’s racialized individuals (Amodio 2008). Someone is “externally motivated” if their motivation is to avoid negative reactions from others who may disapprove of prejudice rather than being motivated by internal standards (like the desire to be fair or equitable). A tolerant racist, for his part,
has the relevant racist beliefs or attitudes, expresses them, but is willing to coexist with Blacks [or some other racialized group]. He will try not to openly oppress or discriminate against Blacks [or another racialized group]. But if he can get away with discriminating against them, he will do so.
has the relevant racist beliefs or attitudes, expresses them, but is willing to coexist with Blacks [or some other racialized group]. He will try not to openly oppress or discriminate against Blacks [or another racialized group]. But if he can get away with discriminating against them, he will do so.(Ikuenobe, 2010, 172)
Finally, there are many people who think that racism has mutated in contemporary society. Think of Case 3: is the case of the assaulted Muslim woman a case of xenophobia, or should we talk here of racism? (note the two are not incompatible; see Kim and Sundstrom 2014). Some have seen in Islamophobia a new form of racism that might be called cultural racism (Rattansi 2007). This racism is different because it does not use old racial categories, does not presume intrinsic biological differences, does not explicitly make reference to hierarchical positions of different groups. As Rattansi puts it, “the emphasis is on cultural difference and the genuine fears of ordinary citizens that their national character and, by implication, way of life may be in danger of being overwhelmed and marginalized [by different ethnic or religious groups]” (2007: 98). It might be argued that this is not enough to be considered racism. Indeed, some would claim that cultural racism is racism only when the group that is the object of negative feelings and/or discriminatory behaviors is “psychologically” treated as a “race” (one might say, is psychologically racialized), that is, that membership in the group is considered a basis for rich inductive generalizations, is inherited from parents, is rather immutable and so forth (Machery and Faucher, 2017; Mallon 2010; Moya and Boyd 2015).
Previously stated forms of racism involve individuals who are conscious of their racial beliefs and attitudes. The next forms of racism included involve unacknowledged forms of racism (Zack, 2003: 255; Atkin talks of “avert racism” to contrast it with “overt racism”; 2012: 118). There are two grand families of unacknowledged racism: less-than-conscious racism and unconscious racism. Let’s start with the first family, which comprises color-blind racism (Bonilla-Silva 2006; Bonilla-Silva and Dietrich 2011), laissez-faire racism (Bobo and Smith 1995) and symbolic racism (Sears and Henry 2003). These theories are usually grounded in two observations: (1) overt racism has been on the decline in the American society in general since at least the second half of the twentieth century (this is true despite sporadic surges of overt expression of racism by some groups in the American society); (2) the disparities between racialized groups that were explained by overt racism are not disappearing. To explain this apparent paradox, sociologists and psychologists postulated that racism is taking new forms.
The new forms of racism share two features. They postulate that (1) because of the advancement of civil rights and the adoption of anti-discrimination laws, whites now believe they live in a “post-racial” or “color-blind” society where racial minorities are not suffering from discrimination or prejudice anymore (so any complaints about racial discrimination or any demands of positive discrimination to rectify racial injustice are perceived as unfounded and thought of as a way to bring back race in the fore, therefore can be seen as a form of racism); and (2) whites use socially available ideological “scripts,” “frames,” or sets of beliefs to explain and maintain disparities. These scripts, frames, or sets of beliefs are either exculpating whites of their responsibility for racial disparities and/or faulting cultural features of racial minorities for the existence of these disparities (for instance, blaming disparities on a lack of effort from their members, or loose family organization).
Individuals exhibiting these new forms of racism might not recognize their own racism, they might have what is sometimes called a “false-consciousness,” rationalizing their negative affects toward other races by blaming members of these groups for their own condition. Moreover, their negative affects are not necessarily grounded in hate or hostility, rather, as Sears puts it, “it may be experienced subjectively as fear, avoidance, and a desire of distance, anger, distaste, disgust, contempt, apprehension, unease or simple dislike” (Sears 1988: 70; cited by Sears and Henry 2003: 260). Not being motivated by racial animus and resorting to abstract ideological concepts (like free market or laissez-faire), they might not recognize that their own actions perpetuate a system of racial domination. For this reason, it is legitimate to talk about “unrecognized” or less-than-conscious racism.
Aversive racism is different from the previous forms of racism because it is espoused by individuals who are often well-intentioned and ostensibly unprejudiced (for a recent review, see Pearson et al. 2009). Indeed, contrary to closet racists, aversive racists explicitly reject racists attitudes and beliefs and try to be fair and equitable in their actions. As Pearson and colleagues put it: “Aversive racists … sympathize with victims of past injustice, support principles of racial equality, and genuinely regard themselves as non-prejudiced, but at the same time possess conflicting, often non-conscious, negative feelings and beliefs about Blacks [or other races] that are rooted in basic psychological processes that promote racial bias” (2009: 3; my emphasis). Some explain cases like our Case 4 as the product of aversive racism. Police officers, white and black alike, have stereotypes about young black males, about the danger they represent, the likelihood that they carry guns, and so forth. Some think that it is the activation of such stereotypes that explain (at least partially) the asymmetry of shooting rates of black versus young white males (this is called the “weapon bias” or “shooter bias”; Payne 2006). So, to create a (fictional) example of how aversive racism would work, in Case 4, it would be possible to posit that perhaps Officer Freeman, who shot David Joseph, was not an overt racist nor a closet or tolerant racist, maybe that he was even working to advance the cause of Blacks in the United States. However, he might have harbored, unbeknownst to him, stereotypes about black males, and these stereotypes might have led him to overestimate the danger that Joseph posed, which led the officer to shoot him. These unconscious (or not perfectly conscious) stereotypes or attitudes are called “implicit biases” in social psychology, and they are thought to partly explain disparities in health, employment, housing, and education. While the size of these effects has been the object of a recent debate (see for instance, Greenwald et al. 2009; Greenwald et al. 2015; Oswald et al. 2013), the fact remains that despite the decline of overt racism, substantial discrimination and racial disparities persist.
The final form of racism that I will consider in this section is institutional racism or structural racism, a term introduced by Carmichael and Hamilton (1967) to refer to social, economic, or political inequalities disproportionally affecting a racialized group. This form of racism is called “institutional” or “structural” because it is thought that these inequalities are produced by social structures or cultural practices or political institutions (such as schools, health systems, voting laws, judicial nominations procedures, etc.) that create, perpetuate, or accentuate the unfair advantages of a dominant racialized group on a dominated one. Theorists of institutional or structural racism recognize the fact that mens rea or unacknowledged racism is often a factor in the production of the inequalities that they denounce, but these same theorists also state that these factors alone are not sufficient or necessary to produce racial inequalities. These theorists emphasize the fact that some inequalities depend, in part or in total, on social or political structures or institutions or practices, which, because of some of their features, give an unfair advantage to one racialized group over another. For instance, in Case 6, inequalities in educational attainment are thought to be caused by many factors: (a) attitudes of the parents toward education as well as the time they spend reading to their kids, (b) attitudes of the teachers (the fact that they have lower expectations of kids of certain racial groups than others), and (c) quality of the school attended by the children. Among those factors only (c) can be thought as being directly the result of institutional racism (some [Haslanger 2004] also argue that (b) might also be thought as being a form of institutional racism, since despite the good intentions of the teachers, their expectations lead to the perpetuations of an oppressive situation for some racial groups). Indeed, it has been shown that African Americans have access to lower quality schools because of housing discrimination. For a number of reasons (having to do with earning and assets, loan regulations, etc.), African Americans are often relegated to poorer neighbourhoods with less established, less well-equipped and staffed schools than those frequented by whites. Kids studying in those schools have less chance of getting the kind of support and environment they need to achieve good results in school. This is thought to be an important factor in educational disparities that will affect future income prospects that will prolong racial disparity.
As I said earlier, there are debates concerning the relative importance or even the existence of some of the forms of racism that I have mentioned in this section. For instance, some doubt that it makes sense to talk of aversive or unconscious racism (Arkes and Tetlock 2004; Garcia 2010) because phenomena thus qualified might be the result of shared stereotypes rather than racial animus. If the disparities are produced unintentionally by mechanisms that produce unfair advantages for a (usually dominant) racialized group, then it cannot be racism because it is not the result of racial animus (Garcia 2004). It seems therefore that what is at the root of some of disagreements concerning racism is the different conceptions people have of what racism is.
In the previous section, I mentioned a number of phenomena that are called “racism” in the literature. I also pointed out that there are disagreements as to what really “counts” as racism. Disagreements of the latter sort are only possible if one holds that there is a concept of racism (one and only one) that is the right one. People who believe in the above, may hold one of two positions: either they assume that the above concept is the one that we folks share, or that the concept is significantly different from our folk concept (in that sense, this concept of racism would be “revisionist,” in that it would be significantly different from our folk concept, and not necessarily capture the same phenomena; see Haslanger  for an example of such a concept). Philosophers have until recently believed the former and they have tried to identify the core component(s) (sometimes called the “constitutive elements” or the “essence” or “nature”) of our folk concept of racism through conceptual analysis. Conceptual analysis is a philosopher’s tool and by using it, philosophers seek to provide the set of necessary and sufficient conditions for the application of a concept. A philosopher’s proposed analysis is usually tested against what he thinks are common intuitions. Indeed, generally, a philosopher would propose a fictional or a real case (or many of them) where one condition deemed to be necessary for the application of a concept would be missing, or where all conditions deemed necessary would be present, and see if the concept is judged to apply in those circumstances, or if it generates counterintuitive consequences. If the latter, the proposed analysis would have to be rejected (or modified substantially).
Before presenting the different philosophical accounts of racism, it has to be noted that trying to identify the core components of racism is something different from studying the forms that racism takes in different places or at different times (the latter being what historians and sociologists set out to do). Philosophers using conceptual analysis often explore the “limits” of our concepts by using imaginary cases to test our intuition; while historians, psychologists, and sociologists are interested in “real” cases, and are happy with operational definitions that capture the phenomena they are trying to explain (without being exhaustive or trying to capture the “essence” or “nature” of the concept). So, it is possible that philosophers will find the later definitions to be lacking. I think it would sometimes be unfair to accuse historians, psychologists or sociologists of failing to spell out the necessary and sufficient conditions for something to be racist: they are, so to speak, playing a different game than philosophers. At other times, philosophers might be disturbed by an extension of the use of the concept to cases that they think are not racism; it is there that one encounters real and substantial points of disagreement. One question that such disagreements raise is “By virtue of what do philosophers think that they can impose a particular definition of racism?” I won’t try to answer that question here, but that is something one might want to hold in mind while considering various arguments for or against certain definitions of racism.
There are two grand families of philosophical accounts of racism: one locates the core elements of racism in individuals (we can call these accounts “agent-based” or “input accounts”), while another locates them in the effects of certain actions or processes (for this reason we can call these “output accounts”). Until recently, philosophers have mainly been concerned with the first family of accounts of racism.
There are three types of agent-based accounts: the doxastic, the behavioral, and the affective/volitional account. I will present them shortly in what follows; then I’ll explain what their limitations are (I am expanding here on work I have previously done with Edouard Machery; Faucher and Machery 2009).
According to the proponents of a pure version of the doxastic account such as, for example, Zack (1999) or Shelby (2002), a necessary and sufficient condition for racism is a set of beliefs (that includes minimally the belief that there are races). For instance, according to Zack, racial thinking is fairly recent, dating from the eighteenth century. Race is at the center of a cluster of ideas (a kind of theory) proposed by philosophers like Hume or Kant to explain human difference, ideas which have since been used to rationalize European domination of other groups of humans. Zack also thinks that these ideas are articulated in the form of Kuhnian paradigms, ensconced in sets of beliefs (some concerning the existence of race, some others about the specific groups called “races”), as well as ways of seeing and categorizing, and so forth. As she puts it:
The ingredients of a racial paradigm at any given time would include a taxonomy of race, the criteria for membership in different races and their application to individuals, social customs and laws that pertain to race, moral beliefs about different race relations, expectations for change in social areas pertaining to race, ideologies of race, and beliefs about the connections between physical race and human physical attributes.
The ingredients of a racial paradigm at any given time would include a taxonomy of race, the criteria for membership in different races and their application to individuals, social customs and laws that pertain to race, moral beliefs about different race relations, expectations for change in social areas pertaining to race, ideologies of race, and beliefs about the connections between physical race and human physical attributes.(Zack 2002: 110)
Shelby for his part thinks of racism as a type of ideology. As he puts it, “ideologies are widely accepted illusory systems of beliefs that function to establish or reinforce structures of social oppressions” (2002: 415; my emphasis). What seems to distinguish ideologies from other sets of beliefs is their function, which is to rationalize oppression. The precise content of the ideology underlying racism is rather dynamic in that “with the possible exception of the belief in the reality of races, no one belief is essential to the legitimating function of the belief system” (2002: 417).
Shelby’s theory can be conceived as a pure form of the doxastic approach. While according to other “impure” forms, including Zack’s previously cited and Appiah’s (1990), racism essentially involves believing in a particular set of propositions. They also involve having specific dispositions with respect to these beliefs (see also Lengbeyer 2004). To be racist, people must believe that races exist (a belief sometimes called “racialism”) and that they are morally significant (either because races are correlated with morally significant properties, or because races are intrinsically morally significant). As Appiah noted (1990: 8; our emphasis), the dispositions themselves consist of
assent[ing] to false propositions, both moral and theoretical, about races—propositions that support policies or beliefs that are to the disadvantage of some race (or races) as opposed to others … and to do so even in face of evidence and argument that should appropriately lead to giving those propositions up.
assent[ing] to false propositions, both moral and theoretical, about races—propositions that support policies or beliefs that are to the disadvantage of some race (or races) as opposed to others … and to do so even in face of evidence and argument that should appropriately lead to giving those propositions up.
The behavioral account insists that racism essentially involves behaving or being disposed to behave in a way that is harmful to the members of a racial group. Michael Philips has proposed a version of this model in which the expression “racist” is used in “a logically primary sense” to qualify actions (what he calls “Basic Racist Acts”). Thus, he wrote:
P performs a Basic Racist Act by doing A when: (a) P does A in order to harm Q because Q is a member of a certain ethnic [or racial] group; or (b) (regardless of P’s intentions or purposes) P’s doing A can reasonably be expected to mistreat Q as a consequence of Q’s being a member of a certain ethnic [or racial] group. Note that, on this account, P’s motives, beliefs, feeling, or intentions need not be taken into account in determining that P performed a racist act.
P performs a Basic Racist Act by doing A when: (a) P does A in order to harm Q because Q is a member of a certain ethnic [or racial] group; or (b) (regardless of P’s intentions or purposes) P’s doing A can reasonably be expected to mistreat Q as a consequence of Q’s being a member of a certain ethnic [or racial] group.(1984: 77)
Note that, on this account, P’s motives, beliefs, feeling, or intentions need not be taken into account in determining that P performed a racist act.(1984: 77)
In his book Racism: A Short History, Fredrickson has defended another version of the behavioral model:
Racism exists only when one ethnic group or historical collectivity dominates, excludes or seeks to eliminate another on the basis of differences that it [the first group] believes are hereditary and unalterable. Racialists do not become racists until they make such convictions the basis for claiming special privileges for members of what they consider to be their own race, and for disparaging and doing harm to those deemed racially Other.
Racism exists only when one ethnic group or historical collectivity dominates, excludes or seeks to eliminate another on the basis of differences that it [the first group] believes are hereditary and unalterable.(2002: 170)
Racialists do not become racists until they make such convictions the basis for claiming special privileges for members of what they consider to be their own race, and for disparaging and doing harm to those deemed racially Other.(2002: 154)
This version of the affective/volitional account aims to correct this lacuna by positing different forms of racism tied with different contexts.
accounts of racism tend to have a generality that obscures important particularities of group-specific types of racism: e.g., the genocide-based racism against Native Americans differs notably from the slavery-based racism against African Americans, and the racist anti-Semitism directed at Jews is distinct in expression and historical effect.(2014: 21)
The previous form of the affective/volitional account is at present not the dominant one. The most influential version of it is found in a series of articles written by Garcia (1996/2003, 1997, 1999, 2004, 2010, ms) in which he defends a version of (2) above. He has proposed that a behavior (an institution, a thought, etc.) is racist if and only if it results from (or is maintained by) a vicious attitude, namely hate or malevolence (but also of lack of benevolence, disdain, or callous indifference) toward the members of a particular race, and that a person is racist if and only if she feeds on this hate and malevolence. As he has put it, “in its central and most vicious form, it [racism] is a hatred, ill-will, directed against a person or persons on account of their assigned race” (1996/2003: 259). As Garcia puts it, his account “concentrates on the disposition of certain volitions, along with desires, preferences, choices, and affective states” (2010: 253). For him, what underlies racial judgments and actions deemed to be racist is malevolence, a “deformation of character” (2004: 41) that is opposed to benevolence and justice (as he sometimes puts it, opposed to “benevolent love”) and which leads to disregard of the welfare of racial others. Importantly for later discussion, Garcia posits that one does not need to have beliefs about the existence of races to have a vicious attitude toward members of a particular race (2001: 134).
Each of the previous accounts can be criticized on the grounds that it does not capture each and every aspect of racism. First, contrary to what advocates of the doxastic account assert, it does not appear to be necessary to believe in the existence of races to be racist. Someone can sincerely profess that races do not exist, and yet still be racist if they have a malevolent attitude toward a particular racial group. Indeed, this is the reason why doxastic accounts do not fare very well with the new forms of racism I mentioned in the previous section (like color-blind racism or symbolic racism). These new forms of racism are predicated on the idea that nowadays agents believe that biological races do not exist, but still entertain negative attitudes towards racialized groups.
Second, and to counter Fredrickson’s behavioral account, it seems quite conceivable that we would think of someone as a racist even if they were powerless and could not harm the members of another racialized group, provided that they demonstrated hate or lack concern for the other racialized group. For instance, one could imagine someone living in an isolated area, without any contact with members of a specific racialized group, yet we would consider that person to be racist if they had a vicious attitude toward them (Garcia 1996/2003: 268). We also might want to say that the tolerant racist we mentioned in the previous section is “racist,” even if they structure their environment and control their actions in such a way that their racism would not have any effect on anyone.
More generally, behavioral accounts suffer from two types of problems. First, in many cases, it is not exactly clear what can reasonably be expected of someone (recall that the second part of Philips’s definition implies that someone is racist if it could be expected that their actions would have a negative impact on members of a racialized group), and who is entitled to settle the question (Garcia 2001). For instance, would well-intentioned teachers who, to counter the malnutrition of their mainly black students, proceed to distribute milk (not knowing that it is likely that many of their students are lactose intolerant), reasonably be expected to know about their students’ lactose intolerance—therefore should these teachers be considered racists? Second, behavioral accounts do not fare well with reverse discrimination (that is, discrimination against members of a dominant group in favor of members of a discriminated group; see Gracia 2010 on these questions).
Third, affective/volitional accounts face a problem with “benevolent” forms of racism (sometimes also called “paternalistic racism”; see Dixon et al., 2012). A benevolent racist is someone who thinks that members of a racialized group R are inferior (either morally or intellectually) and need to be taken care of. Though it is obvious that benevolent racism can be harmful, it is easy to imagine—despite Garcia’s assessment that benevolent racism is necessarily malevolent (Garcia 1997, 2001)—cases of benevolent racists who have a really pure heart (for instance, someone brought up to believe that members of another racialized group were unable to govern themselves and were better of governed by members of “a more able race”). Accounts similar to Garcia’s also face another problem: it is not clear if one can have ill-will against members of a racial group without having any beliefs (Shelby 2002; Mills 2003). At the minimum, one must believe that these individuals belong to a racial group. Accounts like Faucher and Machery are better equipped to face that kind of problem since they are built around the idea that racism is shaped by what individuals believe about the specific dangers posed by racialized groups. Unfortunately, they face other problems: for instance, one can imagine the case of someone who has no negative feelings toward members of race R, but who has been raised to believe that the other racialized group is inferior in some way (for example, they think that members of race R are lazy). This belief will produce a host of negative effects, like depriving members of this race R of work, or bonuses related to work performance, and so forth, that one might want to consider as a form of racism.
Because of these limitations, it is hard to figure out where racism is really located (in the mind, in the heart, or in the actions of agents); Glasgow calls this the “location problem” (2009: 69). If this problem is not solved by the accounts mentioned above, it is also not solved by hybrid accounts in which racism is seen as being a combination of the elements of the individual accounts. For instance, Corlett (1993, 2003) and Dummett (2004) have proposed understanding racism as a combination of prejudice (i.e., a negative attitude) and behavior (i.e., discrimination). Though these accounts might better capture the kind of phenomena that got people interested in “racism” in the first place, or the phenomena that are most socially relevant (if racism had been confined to powerless individuals living on desert islands, racism would not be a pressing issue), they end up excluding more cases than non-hybrid accounts of the kind we mentioned in the previous paragraph (the more elements one adds to our definition, the more restrictive it is). Another possible solution to the problem, which avoids the previous objection to hybrid accounts, is a disjointed analysis: for instance, an analysis of racism as the product of either inferiorization or antipathy (Blum 2002a; 2002b). Glasgow faults this kind of solution for not being able to accommodate all cases of what we consider racist: for instance, stereotypical remarks concerning a racial group coming from someone with a pure heart and who harbors no beliefs in racial inferiority of that racial group. In such a case, Glasgow argues, the remark would be racist, even if the person uttering it is not. It is not clear that this could not be patched by adding more elements in the disjunction. But then, one would run the risk to trivialize the concept of racism as it would lose its precise meaning.
A more general problem for agent-based accounts comes from cases of institutional racism. It is generally admitted that some cases of institutional racism are forms of racism by virtue of the intentions of those who built the institutions in the first place, or those who help maintain them now. Yet some cases of institutional racism cannot be so explained. For instance, imagine university admission policies that would privilege students from specific high schools (because students from these high schools are more likely to have success at the university level) and that, as a side effect that was not foreseen by those who adopted the policies, it results in harming students from a particular racial minority that has an historical background of racist mistreatment (Atkin  calls these forms of racism “indirect”). According to agent-based accounts, these policies would not be racist if they were adopted (or maintained) via a process involving no racist beliefs, malevolent motives, or individual behaviors that could be labeled racist. However, some might want to insist on calling the policies “racist” (Glasgow 2009: 76).
At this point, it seems that different accounts can accommodate different intuitions about what is racist and what is not, but that no accounts can accommodate all of our intuitions. The solution to this situation might be either to abandon the idea of producing a unifying account that would encompass all our intuitions or to produce a different one. I’ll come back on the first solution at the end of this section; let’s look at the second solution for now.
Some philosophers have proposed a type of account that does not try to locate racism inside the agent, but rather in the effects something or someone has on people of a particular race(s). For instance, Glasgow has proposed that something or someone is racist if and only if that thing or person is disrespectful towards members of a racialized group qua members of that group (we can call his approach “the racial disrespect” view; for a similar account, but in terms of disregard, see Taylor 2004) According to Glasgow, this kind of account can accommodate all of the various agent-based cases, plus cases not covered by them (like the peculiar form of institutional racism I mentioned earlier or the behaviors caused by unconscious biases). As he puts it: “One of the reasons that disrespect is a useful analysans for racism is that disrespect is abstract enough to take as many forms as racism takes” (Glasgow 2009: 92). One problem with this analysis is that it might be over-inclusive as there are many disrespectful acts that we might not want to call racist, for instance, the fact of feeling uneasy in the presence of someone perceived as belonging to another race might be interpreted as disrespect. Another problem, noted by Garcia (ms), is that according to some analyses, disrespectful acts depend on the imputation of mental states of an agent, more specifically, they seem to flow from an attitude of disrespect. If so, the racial disrespect account is not really different from the input-based account.
Given that all the accounts we presented in this section suffer from some defects, and that none seem to capture all our intuitions, one might be tempted to give up the idea that “racism” refers to a set phenomenon that could be captured by one and only one analysis. This is a strategy adopted by Haslanger (2004). She observed that:
Persistent institutional injustice is a major source of harm to people of color. Of course, moral vice—bigotry and the like—is also a problem. But if we want the term “racism” to capture all the barriers to racial justice, I submit that it is reasonable to count as “racist” not only the attitudes and actions of individuals but also the full range of practices, institutions, policies, and such like that, I’ve argued, count as racially oppressive.
Persistent institutional injustice is a major source of harm to people of color. Of course, moral vice—bigotry and the like—is also a problem. But if we want the term “racism” to capture all the barriers to racial justice, I submit that it is reasonable to count as “racist” not only the attitudes and actions of individuals but also the full range of practices, institutions, policies, and such like that, I’ve argued, count as racially oppressive.(2004: 122)
There are reasons to think that we will never agree on what racism is, and I think it would be sobering if we could agree on that. One reason why I think it would be important to recognize this (other than the ones mentioned in the previous sections), is the question of trying to establish who is right and who is wrong about the meaning of the term would be foregrounded by the question of identifying the different kinds of moral ills that are involved in the different phenomena people call racism. One advantage of adopting such a policy, as Lawrence Blum once observed (2004: 77), would be to attract attention to the larger class of “moral ills” which have been neglected by philosophers.