Seeing Race Like a State: Higher Education Affirmative Action Verification Commissions in Brazil

ABSTRACT A growing body of literature has focused on how different states continuously “make race” by legitimizing certain racial categories while invisibilizing others. Much less has been written on the actual processes of transforming race into a bureaucratic category when implementing antiracist public policies. This article focuses on the recent use of verification commissions to validate the racial self-identification of potential beneficiaries of racial quotas at federal higher education institutions in Brazil. We argue that through their choices, particularly through their definition of what race is, of who can see race, and of how to see race, these commissions are transforming not only understandings about affirmative action’s aims but also understandings of race. The study focuses on three potential consequences of commission practices for Brazilian racial boundaries: the disciplining of racial identifications, the decontextualization of race, and the individualization of racial injustice.

literature has focused on how different states continuously "make race" or "transform race" by validating or invisibilizing certain racial categories (e.g., Bailey et al. 2008;Loveman 2014;Roth 2017). 1  These studies have shown the key role of state institutions in "seeing" (or not seeing) racial inequalities (Bailey et al. 2013) and promoting different types of public policies (Skrentny 2015).Much less has been written on the actual bureaucratic processes of delimiting the boundaries of racial categories for antiracist goals.Such processes are particularly important in the case of Latin America, where narratives of mestizaje and "blurred boundaries" have historically encouraged people to self-identify as mestizos, or, in the case of Brazil, as morenos and pardos.
This article investigates how race becomes a bureaucratic category of policy practice when racial boundaries are not clearly marked, and the consequences of this process.Empirically, it focuses on the use of verification commissions to validate the racial self-identification of beneficiaries of racial quotas at federal higher education institutions in Brazil (instituições federais de ensino superior, hereafter IFES).These commissions have been introduced in response to the demands of black social movements and student collectives for mechanisms to control fraud in affirmative action and have been legalized through state regulations since 2016. 2 Building on James C. Scott's idea of state legibility (2020), we ask, How do these state-mandated commissions make decisions on how to standardize contested racial categories into clear-cut bureaucratic categories?And how do particular choices of legibility, by definition simplifying, disciplining, and partial, remake the categories they aim to "see," in this case transforming racial boundaries, identifications, and meanings?
Given the growing presence of racial categories in Latin American censuses (Loveman 2014), as well as the expansion of different types of affirmative action in the region (Paschel 2016), the predicaments Brazil faces may shed light on the dilemmas other Latin American countries face when implementing antiracist policies.As one of the first countries to implement large-scale affirmative action policies in Latin America, as well as one where the narrative of mestizaje has played a prominent role in nation building, Brazil is an ideal case for analyzing the dilemmas of "seeing" race in contexts of contested racial boundaries.As such, the country's IFES verification commissions can be analyzed as sites where race is "seen" by the state through bureaucratic procedures.We argue that through their choices on how to "see" race, these commissions are redefining racial identification in the context of racial quotas and, at the same time, the meanings of affirmative action policies originally envisioned by black movements themselves (Feres Júnior and Campos 2016;Alberti and Pereira 2006).
Starting with an overview of the literature on antiracism policies, in particular affirmative action policies, this article then examines the Brazilian case and maps the discourse that led to the creation of racial verification commissions.After a discussion of our data and methods, we take a closer look at how these commissions define race as a bureaucratic category (what is race?), how they select their members (who can see race?), and which procedures they use to operationalize it (how to see race?).We then turn to what we see as the three main consequences of the commissions' procedural changes: the disciplining of racial boundaries through external verification, the decontextualization of race by reducing it to phenotype, and the individualization of racial injustice by defining it as personal experiences of stigmatization.
It is important to note that by focusing on these consequences, we are not arguing against the legitimacy of commissions, a demand of black social movements that was authorized by a 2012 Supreme Court ruling.Our interest instead is in understanding how state initiatives to reduce categorical inequalities can also create new dilemmas for antiracism policies in Brazil and in other Latin American countries where official racial categorization was absent for most of the twentieth century and most people identify as mixed.

HOW STATES MAKE AND REMAKE RACE
Much has been written on how formal and informal state practices have contributed to the creation and reproduction of the racial inequalities we see around the world.From de jure segregation and discrimination practices of the United States and South Africa throughout most of the twentieth century (Marx 1999;Massey and Denton 1993) to de facto discriminatory policies of European and Latin American nations (Lavanchy 2013; Telles and PERLA 2014), researchers have shown how state practices are nearly always racialized.Much has also been written about the practices of state ethnoracial categorization, which serve to reveal certain inequalities but also hide others (Nobles 2000;Loveman 2014).As states increasingly acknowledge the need to implement antiracism policies, attention to the racialized consequences of color-blind and colorconscious policies has been growing (Skrentny 2015;Paschel 2016).Much less attention has been devoted to understanding how states actually "see" race in ways that go beyond self-identification, particularly in contexts in which they seek to redress racial inequalities.
Affirmative action is probably one of the most widespread antiracist policies across the world (Warikoo and Allen 2020).In the case of the United States, by far the most studied in the affirmative action literature, lively debates have occurred over the legality of the use of race-based affirmative action and its specific implementation procedures (e.g., quotas, points, or indirect measures) (Orfield and Miller 1998;Sabbagh 2011).Nevertheless, North American debates have shown considerable consensus regarding which groups are discriminated against and are thus deserving of affirmative action (e.g., African Americans, Latinos) and which individuals belong in these groups. 3 Ironically, despite its centrality in the affirmative action literature, the United States may be the exception rather than the rule.In most other contexts, there is much less consensus both about which groups are deserving of reparation, as in the case of the Scheduled Castes in India (de Zwart 2000;Jenkins 2003), as well as about which individuals belong in each of the groups, as reflected in the changing ethnoracial demographics in Latin America (Loveman 2014).
Indeed, Latin America is often presented as the region where boundaries between ethnoracial groups are the most difficult to draw, partly because state institutions abandoned racial categorization after independence, due to the failure of immigration whitening policies (Graham et al. 1990).Having adopted the language of mestizaje and mixture for most of the twentieth century, many Latin American countries took a "multicultural turn" at the beginning of the twenty-first century and are increasingly implementing affirmative action policies for indigenous and Afro-descendant populations (Hale 2002;Paschel 2016). 4Still, the majority of the population in the region identifies as mestizo or mixed, categories with unclear and moving boundaries (Telles and PERLA 2014).

FROM SELF-IDENTIFICATION TO VERIFICATION COMMISSIONS: REDEFINING THE BOUNDARIES OF BLACKNESS IN BRAZIL
Since the beginning of the twenty-first century, Brazil has implemented affirmative action policies in the form of quotas for low-income and black students at public universities, the most selective higher education institutions in the country. 5 Affirmative action policies are commonly described as a major turn in Brazilian state practices related to race (Htun 2004;Lehmann 2018).Although the origins of Brazilian affirmative action can be traced back to the late 1990s, public universities incrementally adopted these policies in the aftermath of the 2001 United Nations World Conference Against Racism (Daflon et al. 2013).
Brazilian quotas for public higher education have a social component (for alumni of public high schools and for those of low income) and a racial component (for black and brown students).But since the beginning, the most controversial dimension has been the racial one (Fry et al. 2007).Nearly 20 years after their initial implementation, racial quotas have significantly increased the numbers of black students at the IFES (Senkevics and Mello 2022).In addition, few significant racial performance gaps have been identified, and then only in close correlation with socioeconomic factors (Pelegrini et al. 2022;dos Santos et al. 2022).In fact, university racial quotas have recently been expanded to graduate programs (Venturini 2017), made mandatory in the selection of civil servants when three or more positions are available (Oliveira and Targino 2017;de Faria 2016), and, in 2020, were likewise made mandatory for political party electoral lists (GEMAA 2020).In 2022, a national survey showed that half of the Brazilian population approved of racial quotas at public universities (Saldaña 2022).
Until 2016, Brazilian racial quotas relied mostly on self-identification to define their beneficiaries-those who identified as black (preto) or brown (pardo) according to Brazilian census categories were entitled to racial quotas. 6The use of selfidentification as the basis for operationalizing race in affirmative action was the result of long political negotiations between social movements, international actors, academics, and state bureaucrats.Following ILO C169-the Indigenous and Tribal Peoples Convention (ratified by Brazil in 2002)-the right to self-identification was the milestone that made possible the recognition of the land rights of quilombolas and indigenous groups in Brazil, as well as in other Latin American countries (French 2009).The right to self-identification was also used by Brazilian supporters of affirmative action as a strategy to counter criticism from the policy's detractors, who framed racial quotas as a threat to people's identities and privacy and to the fundamental idea of race as a social construct (Fry et al. 2007).By defining the beneficiaries of affirmative action on the basis of self-identification, policymakers avoided such criticism, at the same time advancing the case for affirmative action in Brazil (Guimarães 2017).
By the early 2010s, however, black student collectives (Guimarães et al. 2020) and sectors of the black movement, in dialogue with the federal public prosecutor's office (CNMP 2015), had started to question whether self-identifying as preto, and in particular pardo, was sufficient to be considered a legitimate beneficiary of racial quotas, and to demand verification commissions.Broadly speaking, there were three dimensions to their criticism.First and most salient in the debate was the suspicion that people socially regarded as "white" were cheating the system by strategically mobilizing the idea that in Brazil racial boundaries are flexible (Leme and Caetano  2016; Neves 2022). 7 Second, many critics claimed that racial quotas were benefiting people who were not aware of the history of the Brazilian racial struggle.Several activists expressed frustration over the perception that most beneficiaries do not demonstrate awareness of racial discrimination in Brazilian society and that the goal of racial quotas to promote black consciousness and demolish epistemological racism was being lost (Bernardino-Costa and Blackman 2017; dos Santos 2023).
Third, many activists claimed that differences between dark-skinned and lightskinned pardos had to be considered, as darker skin color is increasingly perceived as a determining factor of racial discrimination, independent of racial identification (Telles 2014;Monk 2016).This argument echoes the epistemological criticism of taking state-defined categories as the best indicators of inequalities.It was a call to take race cues seriously in better understanding and redressing discrimination (Monk 2022).
From a legal perspective, the use of commissions to validate racial selfidentification had already been legitimized by a decision of the Brazilian Supreme Court (STF 2012).But it was only in 2018 that verification commissions became mandatory for civil servant recruitment.Two official decrees govern the implementation of verification commissions in Brazil: a 2016 directive (DOU 2016) and a 2018 normative ruling (DOU 2018).The 2016 directive made commissions mandatory to validate the racial self-identification of candidates for federal civil servant jobs. 8The 2018 normative ruling followed, declaring that only a candidate's phenotype should be considered when verifying their self-identification.It also listed the procedures that should be used to assess this phenotype; these procedures are mandatory for civil servant selection and highly recommended for student selection processes.Such procedural changes implied a major shift in how state institutions saw race in Brazil.Even if self-identification as preto or pardo is still a necessary condition to benefit from quotas, it is no longer a sufficient one.
Although a number of case studies of these commissions have been published, providing insight for this article, we have been unable to find any extensive empirical review of their implementation and procedures, or of their potential consequences. 9Therefore this study focuses on the procedures followed by these commissions and the dilemmas they have created for affirmative action as an antiracism policy and discusses the consequences of verification commissions for affirmative action policies in Brazil.

DATA AND METHODS
Between April 2018 and February 2019, we searched all IFES websites, reviewing documents related to affirmative action, quotas, and racial diversity, and the results of keyword searches, including comissões de verificação, comissão de identificação racial, identificação de fraude, and identificação falsa. 10As other researchers have acknowledged, these documents are not always visible or easily accessible, making this mapping a laborious manual task (Silva et al. 2020).
Taking the verification commissions as our starting point, we consulted a myriad of official documents (portarias, resoluções, editais), reports, news clippings, and statements from various administrations and student collectives, as well as individual faculty, students, and staff (in writing as well as in YouTube videos), which gave us more information about their practices to operationalize race or phenotype.In particular, this included the composition of the commissions, their procedures, and the criteria they used.In total we collected more than 700 sources and, for this article, analyzed 278 that referred specifically to the undergraduate selection verification commissions.Among these 278 sources, 180 were official documents and 98 were unoficial documents.The latter were journal articles and news clippings and statements from individual faculty and collectives that provided information on the commission in a specific university.
We found 35 IFES (33 percent) with permanent validation commissions for black and brown undergraduate candidates. 11This represents remarkable growth compared to 2012, when Daflon et al. (2013) found that only 6 universities with racial quotas had race verification commissions.In 2016, after the 2012 federal law but before the 2016 directive, this number was even lower: only 3 IFES had verification commissions.Importantly, although verification commissions for undergraduate selection are not mandatory, debates about their need started at the undergraduate level.This is also the level that receives the largest number of candidates through quotas-hence the greater need for standardization of procedures. 12 It is important to stress that the implementation of these commissions is recent and ongoing; as such, we followed the developments of the commissions in these 35 universities through September 2020 and found that 20 of them had changed their practices, largely to adapt to the 2018 ruling. 13In presenting our data, we capture the most current procedures but also mention some of the changes we believe are indicative of future trends for the commissions' practices.
To complement the findings from the documentary analysis, we triangulated the information from the official documents and internet searches with information from other sources.We conducted ten interviews with key sources in the federal administration and with members of the commissions and analyzed recent academic publications based on narratives by commission members or case studies of particular universities, usually published only in Portuguese. 14

WHAT IS RACE? DEFINING BLACKNESS AS PHENOTYPE
At a 2015 public hearing of the National Council of the Public Prosecutor, Frei David dos Santos, the founder of the activist organization Educafro and one of the main advocates for racial quotas in Brazil, stated that the reliance on self-identification as the single criterion for identifying quota beneficiaries had created the possibility of fraud, in particular because candidates misunderstood their racial self-identification and "do not consider that you need to have black phenotypes, and not just black genotypes" to be a legitimate beneficiary of the policy (CNMP 2015).Echoing the centrality of black phenotype, the 2016 directive instituting the verification commissions stated that "criteria to evaluate the truthfulness of the [racial] self-identification should consider, only, the phenotype aspects of the candidate, which should be verified only in the presence of the candidate." 15 Similarly, the 2018 normative ruling stated that "the verification commissions will use exclusively the phenotypic criterion to assess the condition declared by the candidate in the public contest."We found that, reflecting this principle, in 2020, 34 out of 35 of the IFES analyzed declared using phenotype as a criterion and only 11 declared using other complementary criteria, such as official documents or written statements (down from 16 in 2018).These numbers suggest a convergence in the commissions' definition of race as phenotype.
The idea that phenotype is an obvious proxy for racial classification in Brazil is not new.In the 1950s, Oracy Nogueira (1955) claimed that while in the United States blackness was defined by origin, in Brazil it was defined by appearance (marca).The understanding of race as skin color is also echoed in the ethnoracial question in the Brazilian census.Before 1991, the ethnoracial question in the census was "What is your color?" and the response options were color categories (branca, preta, parda, and amarela).It was only in 1991, with the inclusion of the term indígena, that the question was changed to "What is your color or race?" More recently, scholars have argued that skin color is a better measure to capture the discrimination that affirmative actions are supposed to redress.Relying on a color palette, the Project on Ethnicity and Race in Latin America (PERLA) also found that skin color was more closely associated with education, occupation, and income indicators than self-identification according to racial categories (Telles and PERLA 2014;Monk 2016).The centrality of phenotype in public understandings about racial identification in Brazil is also confirmed by a 2008 survey conducted in five large metropolitan areas in Brazil on perceptions about racial identification.The results showed that when asked to justify their racial self-identification, Brazilians most frequently pointed to phenotype as their first justification (70.4 percent), in particular skin color (55.5 percent) (IBGE 2008).
Despite the long history of race as phenotype, as well as its widespread public acceptance, the reliance on phenotype as the only legitimate criterion to see race in Brazil, along with the demand to more clearly police the boundaries between preto and pardo, is a remarkable development.In fact, during the second half of the twentieth century, black movements urged people "not to let their color pass as white" and celebrated the rising number of people who identified as pretos and pardos as evidence of the strengthening of the racial consciousness of black Brazilians (Powell and Moraes Silva 2018).The understanding of race as phenotype, in contrast, restricts blackness to dark-skinned individuals and underplays, if not completely rejects, the political definition of black identity as a refusal of whitening and an affirmation of black ancestry.
The rise and fall of the term afrodescendente illustrates this change.Until 2010, it was common to refer to the beneficiaries of affirmative action in Brazil as afrodescendentes-a term that evokes black ancestry and kinship (Feres Júnior et al. 2018).Today the term is rarely used in Brazil, and a number of campaigns from black collectives have denounced "claiming black ancestry to benefit from racial quotas" as a form of fraud.For example, in a widely circulated campaign, black and brown students held signs that said "my grandmother is white, therefore I am white?" (Coletivo Negrada 2016).The assumption was that if their dark skin did not allow them to claim whiteness, light skin should prohibit those students from claiming blackness in the context of affirmative action.Echoing the statement from Frei David dos Santos about the difference between genotype and phenotype, these collectives rejected ancestry as a sufficient basis for black identification, claiming that in the same way that those who had a white grandmother but who looked black would suffer discrimination, those who had a black grandparent but looked white did not experience it.This contrast is clearly captured in the term afro-conveniência, used in campaigns by black student collectives to refer to students who have ancestors who were black but who are too light-skinned to be perceived as black.In this view, they are only strategically identifying as black and should not be entitled to quotas. 16 Our results show that universities rely on traditional census color categories to define legitimate beneficiaries of affirmative action.Among the 35 commissions analyzed, nearly all referred to preto (33) and pardo (33) and most referred to indígena (28).The more political term negro was mentioned by only half the universities (18).Nevertheless, assuming that these census color categories reflect a clear-cut phenotype plays down the dynamic nature of racial identification in Brazil, particularly among those who identify as pardo.

WHO CAN SEE RACE? SOCIETAL, SKILLED, AND MILITANT GAZES
In her study of verification commissions for municipal jobs in a town in the northeastern state of Bahia, Sarah Lempp (2019, 9) found that commission members may expect candidates to identify not only as pardo but also as negro in order to qualify as affirmative action beneficiaries.She argues that for commission members, race is sometimes defined as something that every Brazilian can see, what she calls a "societal gaze," and sometimes as something that only specialists can see, what she terms a "skilled gaze."Building on Lempp's findings, we argue that by defining who is a desirable member of these commissions, or who can see race, the commissions are also defining how race should be seen by the state.
According to the 2016 directive, the selection of verification commission members should consider diversity of gender, color, and preferably, place of birth.The 2018 ruling, which should have provided more details on the composition of the commissions, is actually rather vague on this point.It simply states that members should have a flawless reputation (reputação ilibada), reside in Brazil, participate in workshops on racial equality, and preferably, have experience in the promotion of racial equality and antiracism.Not all commission members have to be civil servants, but these commissions are mandated and regulated by the state.In fact, a few commission members, such as the one cited in the next section, explicitly mentioned that on the commissions they are acting as "state agents." 17At the same time, like most street-level bureaucrats, commission members have much discretionary power on how to "see race," and understanding who they are is key to analyzing how the state will see (or not) potential racial quota beneficiaries as deserving or not.
Although the 2018 ruling specifies that members' CVs should be publicly available, we were rarely able to find these CVs.In fact, in five cases, we were unable to find any information at all about the composition of the commissions.For the 30 IFES for which we had information, we found that most (28) relied on their employees-professors and administrative staff-to serve on commissions.Student members were also mentioned, although less frequently.Probably echoing the 2016 directive, 24 IFES mentioned the need to have a diverse commission.For example, the Federal Institute of Goiás (IFG 2018) mentions that commissions should be heterogeneous; should observe criteria of gender, color, and preferably, state origin; and should consist of at least two-thirds pretos/as, pardos/as, or indígenas.
The reference to diversity acknowledges that people may see racial boundaries differently, but also suggests that these commissions should be able to perform Lempp's "societal gaze" andwhat Roth (2018, 1094) calls "collective racial appraisals" or "societal-level norms of contemporary racial classification or assessments of the societal racial order."One of the public administrators we interviewed, for example, claimed that if the idea was to capture how society sees people, the commission ought to be composed of ordinary people, and potentially even people who discriminate on racial grounds in their everyday lives, offering the example of jury duty in the United States as an ideal template.
At the same time, the documents we reviewed also commonly referred to social movements and scholars.Scholars who were part of research groups called Núcleo de Estudos Afro-Brasileiros (NEAB, Afro-Brazilian study centers) were often mentioned (19 cases); 7 universities added a reference to NEABs between 2018 and 2020.NEAB research centers have been encouraged by Federal Laws 10639/03 and 11645/08 as "a strategy to strengthen educational processes oriented toward Afro-Brazilian values, worldviews and knowledge" in IFES (MEC 2009, 45).
NEAB membership is a proxy for being a "race scholar," suggesting that race is something academic experts are more skilled to identify correctly, using methods that are more likely to be reliable and objective.At 22 of the IFES, commission members were also invited to take courses on race and ethnic relations (which usually meant reading and discussing social science authors); this suggests that "seeing race" is something people should be trained for, akin to Lempp's "skilled gaze."By virtue of this skilled gaze, expert members are perceived to be in a better position to understand who should benefit from quotas, especially when candidates' racial identification is called into question.
In addition, 15 IFES mentioned the presence of civil society and social movements on the commissions.For example, IFPR (2019) states that commissions should be composed of "member(s) of research groups on racial issues of IFPR, representative(s) of racial diversity, representative(s) of study groups and/or research groups and other representatives of organized civil society."The presence of social movements on commissions was a demand of university student collectives, and, as celebrated on Facebook by black collectives from UniRio and UFRGS, was achieved through intense negotiations with the university administration (Coletivo Balanta 2018; DCE Unirio 2018).Their presence is legitimized by their historical participation in the fight for affirmative action policies, and more recently, for the verification commissions, adding to the commission what we call, complementing Lempp's 2019 typology, a "militant gaze." Because most commissions we analyzed included members with at least two of the three types of gaze (societal, skilled, and militant), it is fair to assume that different types of gaze were present within them.This does not mean that commissions are necessarily conflictual: although some of our interviewees told us about strong disagreements among committee members, others said there was more consensus than disagreement in their categorizations-a perspective validated by most published narratives of the commission members.On the one hand, this might reflect a broader pattern found in studies comparing interviewee racial self-identification and interviewer racial categorization: although there is some disagreement about who is pardo and who is preto, there is strong consensus about who is not black (i.e., who is white) (Telles 2004).On the other hand, disagreements may be less common when group boundaries between commission members are not as clear-cut as the documents and Lempp's typology suggest.
Published testimonies of commission participants show that faculty who participate on the commissions are often scholars of race and ethnicity, and staff more frequently active in black social movements.In addition, the boundaries between activism and academic knowledge on issues of race are particularly blurred.For example, the Federal University of Maranhão (UFMA 2019) states that commission members "should have links with research groups, study groups or organized social movements related to the ethnoracial issue."This blurs the boundary between the skilled and militant gazes.Similarly, one interviewee told us that the courses that provided a "skilled gaze" were often taught by black activists.Furthermore, emerging consensus may also be a result of the 2018 ruling, which operationalizes race as phenotype.But even if commisson members may often agree on who are the deserving beneficiaries, light-skinned pardo students not considered black enough by commissions often disagree with that assessment and feel unfairly treated (Neves 2022).The different ways that procedures need to be redesigned and justified suggest that operationalizing race as phenotype may not be as straightforward as surveys relying on the color palette may suggest.

HOW TO MEASURE RACE? SEEING PHENOTYPE AND PERFORMING BLACKNESS
When setting up their first verification commission for civil service positions in 2016, the Instituto Federal do Pará (IFP) provided evaluators with a form to be completed in case there was doubt about a candidate's racial self-identification.The form brought additional criteria, such as the shape of the nose, lips, jaw, cranium, and zygomatic arches, as well as hair type, facial hair, and the color of teeth and gums (IFP 2016).The form became public, and the institute was accused of racism for mobilizing categories from nineteenth-century eugenics.Representatives of the institute argued that they were just trying to find a way to operationalize race as phenotype, as recommended by the 2016 directive.
It was partly as a consequence of this case that the 2018 normative ruling was issued.The broader goal of the document was "to guarantee human dignity in the procedure, the right to self-defense, standardization and equal treatment, transparency and public access, legal norms and the effectiveness of affirmative action in guaranteeing access to black candidates" (DOU 2018). 18In order to do so, the ruling also defined a number of procedures to be respected by these commissions, such as face-to-face interviews and recording of interviews.
We were able to find information about the procedures of the verification commissions at 33 of the 35 IFES we analyzed.Although the 2018 ruling stressed the need to take account only of candidates' phenotypes, we found that at least 11 IFES also referred to "official documents" when evaluating eligibility.We also found contradictory statements about the role of official documents in IFES' official rulings and on their websites.For example, at the Federal Fluminense University (UFF), official documents of various kinds, including written statements, were accepted and examined by the commission, provided that they included the person's color, as do birth certificates and military documents from several northeastern states (UFF 2019). 19Others, such as the Federal University of Espírito Santo (UFES 2018), stated explicitly that "previous records or documents eventually presented, including images and certificates related to confirmation in heteroidentification procedures carried out in federal, state, district, and municipal public competitions should not be considered."In addition to official documents, we found a few references to pictures (six cases) and written documents (seven cases) as documents that could be used to verify candidates' racial identification.
Interviews with key sources also point to ambiguity on the part of commission members toward documents that refer to census categories, along with the difficulty in denying their legitimacy, given that this racial classification was conducted by a local state agency.As one commission member stated, In some places in the Northeast, the skin color of the person is written on the [birth] document.And then, if preto or pardo is written on the document, there's no way to deny [admission], you're already in.Because, as we [Federal University] are part of the state, we cannot contradict the state.
Nevertheless, to a very significant degree, and even before the 2018 ruling, most commissions based their decisions on interviews with candidates.By 2020, 31 IFES were relying on interviews to verify candidates' racial identification.One visible change that happened during our two years of data collection was the increasing use of filmed interviews (adopted by eight IFES between 2018 and 2020) that could be used in case of an appeal against the commission's decision.By 2020, 24 IFE commissions were using filmed interviews.
Interviews with candidates allow commission members to see the candidates (avoiding possible photo manipulation), but they also give an opportunity for candidates to perform and potentially justify their racial identification. 20Lightbrown-skinned candidates who can show that not being perceived as white has had negative consequences on their lives may be seen as more deserving of racial quotas, even if such a definition might be challenged on purely phenotypical grounds.For example, in an early experience with a verification commission, a blonde woman with straightened hair and light skin was validated as black because she affirmed that she had had to straighten her hair to find a sales job (Silveira 2015, 5).In contrast, and more controversially, "wrong" narratives about race can also exclude candidates who might potentially have been included (i.e., who could be considered nonwhite).
In short, the interview performance may exclude candidates who are not familiar with the "right" repertoires of race and racism while benefiting those candidates who have had access to those repertoires-even if they might share similar phenotypes.This is actually one of the reasons why many commission members and black collectives support the emphasis on phenotype, even if this emphasis also has unintended consequences of other kinds.

SEEING AND (RE)MAKING RACE THROUGH VERIFICATION COMMISSIONS
We have seen how verification commissions have challenged racial self-identification as the sole criterion for defining quota beneficiaries, have adopted phenotype as the way to see race, and through societal, skilled, and militant gazes have identified deserving beneficiaries of racial quotas as those who would be targets of racial discrimination.These policy implementation changes brought about by the commissions have generated notable consequences.

Seeing Race Like a State: Verifying Blackness and Disciplining Racial Identity into a Policy Category
Verification commissions were created to validate self-identification and avoid fraud in affirmative action policy.Such a move created tensions and brought about the need to define what is race, who can see it, and how "seeing" can be done.More broadly, this move was also a U-turn from the historical perspective of the Brazilian black movements that all people of African descent should embrace their blackness.Such a turn, we argue, is related to a move away from understanding racial identification as a contesting and contested political category, aimed at increasing the constituency and claim rights, to an understanding of racial identification as a policy category that (re) allocates scarce resources.
In arguing for the need for commissions, supporters pointed to the urgency of excluding "white fraudsters," who were benefiting illegitimately from the racial quotas meant for blacks-officially pretos and pardos, according to the census categories.However, the focus on fraud as a judicial category underplays the historical ambiguity of blackness as a social category in Brazil.In disciplining the black identity, commissions face the hard task of defining who can claim black (or nonwhite) identity to benefit from quotas in a country that has historically encouraged people to define themselves as mixed or, according to the official census category, as pardo.
Between 1940 and 2010 (before racial quotas were widespread), the percentage of Brazilians who identified as pardos in the census doubled (from 21.32 percent to 43.3 percent), and today, according to the 2019 official household survey (PNAD, Pesquisa Nacional de Amostra Domiciliar), 47 percent of Brazilians identify as pardos and would, in theory, be entitled to racial quotas.In fact, the percentage of people who identify as pretos and pardos defines the percentage of racial quotas in each Brazilian federal unit or state.We also know from previous studies that not all pardos see themselves as negros (Moraes Silva and Leão 2012;Daflon 2017).However, as argued by Bacelar da Silva (2022), Lehmann (2018), andLempp (2019), among others, with racial quotas people are, for the first time, gaining benefits when choosing not to whiten.Relying on survey experiments, Bailey (2008) has shown that if only half of those who identified as pardos chose to identify as negro when facing a dichotomous option of white versus negro, this percentage significantly increased when racial quotas were explicitly mentioned at the beginning of the questionnaire.
But to self-identify as pardo is different from "proving" to be pardo or negro.Because the category pardo is largely ambiguous and unstable, people may feel particularly uncertain about its meaning and boundaries.In their article on the writing of UFF candidates, Miranda et al. (2020) found that pardo candidates consistently searched for objective criteria to convince commission members of their racial identification.This uncertainty is also evidenced by figure 1.Using data from Google Trends, we found that the number of web searches since 2004 for the term pardo has grown steadily, especially close to 2016, when commissions became mandatory.The search terms with the greatest increase were "brown skin" (pele parda), "brown person" (pessoa parda), "brown color" (pardo cor), "brown race" (raça parda), "what is brown skin" (o que é cor parda), "brown skin color" (cor de pele parda), and so on.
Importantly, pardos not only have higher chances of having their selfidentification denied by commissions, but they also might not see racial quotas as meant for them.Asked to be certain about their identities, or phenotypes, and knowing that they will have to pass in front of a commission may lead those who are less familiar with repertoires of blackness (and more familiar with traditional repertoires of racial democracy) to self-exclude.Although there are no conclusive results about that idea, Batista and Figueiredo (2020) found that the number of applicants who self-identify as preto and pardo has declined since commissions were implemented at a federal university in the South.
In addition, because access to repertoires about black identification and the inclusion of pardos in the negro category are also shaped by education and socioeconomic status, chances are that it is low-income pardo students, not uppermiddle-class white or pardo students, who might not be attending verification commission interviews.Even if many of these students can still benefit from the quotas aimed at those candidates coming from public high schools, they are denied the recognition of a positive racial identification, ironically one of the initial goals of black movement activists who asked for racial quotas.
If racial quotas and explicitly mentioning race seem to encourage people to embrace their blackness, commissions restrict who can claim such an identification, at least in the context of affirmative action policies.As put by Neves (2022), if the white fraudsters are the villains of commissions, the light-skinned pardos are the victims-or collateral damage-of this disciplining of racial boundaries.

Seeing Race as Phenotype and Decontextualizing Race
We have seen that verification commissions have generally followed the federal 2016 directive and 2018 regulation and increasingly used phenotype as the sole criterion to define who should benefit from racial quotas.Phenotype, and its consequential Figure 1.Internet searches for the term "pardo" between January 2004 and January 2018 emphasis on skin color, allows an emphasis on colorism as a form of discrimination, along the lines of what Monk (2022) has termed a shift from nominal categories to the cues of categories.Nevertheless, privileging phenotype means discarding other dimensions of race, such as ancestry, territory, and family (Roth 2016) and changes the very understanding of what "race" means.This is what we mean by decontextualizing race: an understanding of race as a category that can be seen outside of social interactions and relations and assessed at the individual level.By seeing race as phenotype, the commissions also validate an understanding of race that potential candidates see as more legitimate or objective.In other words, when "seeing" race as phenotype, commissions also are legitimizing particular meanings of race and changing the way beneficiaries see race and public policies (de Souza Leão 2022).An anecdotal example of how this change can happen emerges from the case of a student who was initially rejected for racial quotas at the Brazilian School of Diplomacy and decided to appeal the decision.When the case became public, he presented himself as "a very typical Brazilian, and I've always been very proud of it.In my dad's family, my grandfather is black, my grandmother has Indian and white roots.And on my mother's side, they are mostly white, mostly Portuguese."It is clear that he justified his identification based on ancestry, arguing that his father's black family would entitle him to identify as black and thus meet quotas.
This can be seen as a typical case of afro-conveniência, but one could also argue that if the student's black grandfather had been discriminated against, it might have had consequences in his father's life and therefore in his own.But our goal here is not to evaluate the merits of his claim but instead to show that because phenotype is the only accepted dimension for justifying claims, in his appeal the student changed his narrative about his racial identity, focusing on skin color and phenotype as allegedly objective measures.In his words, "Apparently on my face I'm a type 4 [Fitzpatrick].Which would be like Jennifer Lopez or Dev Patel, Frida Pinto or John Stamos.On my limbs I would be type 5, which is Halle Berry, Will Smith, Beyoncé, and Tiger Woods."According to him, if those people who had the same skin color as his were considered black, why shouldn't he be?
Even if such a medical understanding of phenotype has been largely rejected by commissions (AC Silva et al. 2022), moving from a subjective (a typical, i.e., mixed Brazilian) and relational definition (based on ancestry) to one that is purely phenotypical still allows for a reinterpretation of race as an objective measure, rather than a social construction.In other words, limiting the definition of race to phenotype creates the illusion of a reliable and absolute measure of race.However, phenotype markers vary between countries (e.g., between the United States and Brazil) and within countries (e.g. between the Northeast and Southeast of Brazil) as well as across different socioeconomic strata.In short, seeing race as phenotype may overlook the fact that in societies constructed through a narrative of racial mixture, phenotype is also partly socially constructed, not least because it depends on how it is perceived by society and experienced by individuals.

The Individualization of Racial Injustice and the Changing Meaning of Affirmative Action
By relying on how others see race or phenotype, verification commissions hope to capture what Roth (2016Roth ( , 1325) ) calls the "observed race, [which] more closely mirrors experiences of discrimination."A definition along these lines echoes an ironic Brazilian saying, "if you want to know who is black in Brazil, ask a policeman"-since the police consistently and violently discriminate against the black population (see Miranda et al. 2020, who claim that this was an important argument on the UFF commission).
In the case of racial quotas, verification commissions are expected to make similar assessments and define who should benefit from affirmative action.We found a consistent emphasis in the documents reviewed and on the commissions that commission members should be diverse, so as to be able to perform what Lempp (2019) calls a societal gaze.But commissions are also composed largely of university professors and "race experts," underlining the centrality of the skilled gaze.As evidenced by in-depth interviews and published case studies written by commission members, most commissions acknowledge the multidimensionality and complexity of race.The fact that "workshops on racial equality" are also required further suggests that expert knowledge is needed to be able to see race.Furthermore, the inclusion of a militant gaze seeks to guarantee that affirmative action will remain an antiracist policy.
Yet although both the skilled and militant gazes have historically defined racism as structural, the goal of the commissions is to find which individuals deserve racial quotas and which do not.The idea of a structural, relational, or institutional racism is thus narrowed down to individual and everyday experiences of racism and discrimination.In our research, we found that there is room for nuance and disagreement among commission members regarding how narrow the definition of racism should be and, consequently, the role of affirmative action (Silveira 2015).But regardless of their gaze, what matters for these commission members is whether a given individual is a potential victim of discrimination in the labor market and at school institutions; that is, whether they could lose opportunities for individual social mobility on the basis of their appearance.This view recurs in several reports from commission members in Brazil (see note 9) and from some sectors of the black movement (Dias and Tavares Júnior 2018).
The idea that affirmative action should target individuals who have suffered (or will suffer) racial discrimination and racial prejudice and use their phenotype or individual experiences as a proxy for this potential discrimination is new.It is an important change in the way affirmative action policies have been historically understood in Brazil.In fact, the inclusion of pretos and pardos in racial quotas was based on the idea that discrimination in Brazil works along white versus nonwhite lines and that structural racism in the Brazilian society prevents all nonwhites from ascending socially (Hasenbalg 1979).Thus, this new understanding is transforming the original goals of the policy, which were to rediscuss racism itself, reduce inequalities, and promote racial justice and equal opportunity as a societal project (Feres Júnior and Campos 2016).As French argues (2022), it was the idea of collective rights that shaped the revolution of Brazilian affirmative action in the first place.By narrowing the goal to individual discrimination, commissions reframe racial injustice as prejudice and racism as attitudes, moving away from a broader understanding that would account for forms of intergenerational transmission of advantages and disadvantages, accumulated assets, distribution of social and symbolic capital, and networks of social relationships, among other factors.

CONCLUDING REMARKS
States continuously make race and (re)create racial boundaries as they simultaneously employ practices of racial exclusion, such as police violence against blacks, and policies of racial inclusion, such as affirmative action.In its nearly two decades of existence, affirmative action has democratized Brazilian IFES by allowing more black students to enter highly selective spaces.Racial quotas have transformed Brazilian racial boundaries by making it possible for black Brazilians to benefit from their racial categorization as preto or pardo and potentially encouraging people to do so (Bailey et al. 2008;Francis-Tan and Tannuri-Pianto 2015;Senkevics 2022).
The mandatory nature of racial quotas at prestigious universities and the expansion of racial quotas to other realms have come with a growing number of accusations of fraud.Black movements and collectives have argued that white people -or those socially perceived as white-were claiming black and brown identities and benefiting from affirmative actions not meant for them.Lehmann (2018, 196), however, proposes a different way to see this issue.
The very ambiguity about who qualifies as a black person and the generous official quota criteria of pretos, pardos, and indígenas broaden the constituency of the negro movement still further, and to this may be added people who have never thought of themselves as preto or pardo but come to do so not for opportunistic reasons but because they genuinely begin to interpret their disadvantage in racial terms.
Following Lehmann, if we take seriously the fact that awareness of racial discrimination is one of the goals of multicultural policies in Latin America, we also should take seriously the fact that affirmative action may create a new framework for people to see how racism has shaped Brazilian society and thus, individual trajectories (Giraut 2023).The role of verification commissions is therefore to separate those who identify as negro out of opportunism from those who have become conscious of their racial disadvantage.
Through our description of the emergence and practices of these commissions, we have focused on three issues: how commissions define race as phenotype; who, at the intersection of the societal, skilled, and militant gazes, can see race; and how race can be seen, in particular through a reliance on phenotype.We have argued that these policy implementation choices bring with them the risk of disciplining racial boundaries, decontextualizing race, and individualizing racial injustice.
To be sure, in calling attention to these largely unintended consequences, we are not arguing against the need for affirmative action, racial quotas, or verification commissions.As one of our interviewees put it, if there is evidence that the policy is benefiting people whom it was not intended to benefit, it is the state's obligation to verify racial identification, as it does school credentials, income, age, and so on.Nevertheless, when the state "sees" race across certain dimensions, it loses sight of others.This is the trade-off of every public policy and, as researchers committed to social inclusion in Brazil and beyond, our role is to make these unintended consequences visible and question the assumptions that might lead to them.
As argued by Miranda et al. (2020), maybe the problem lies with the terms of the Quotas Law itself, which does not clarify the boundaries of what racial justice is, and instead forces candidates to define themselves if they are to be considered deserving of racial redress.Above all, the controversies around verification commissions do not in any manner call into question the importance and necessity of racial quotas for democratizing higher education in Brazil.Referring once again to Scott (2020, 4), "[state legibility practices] are as vital to the maintenance of the welfare state as they are to the designs of a would-be despot."It is the strength of democracy and the potential of civil society and public debates to constantly put these state categorization practices under scrutiny that make the difference.

NOTES
We are grateful to our interviewees, who cannot be named but were key to our understanding of the dilemmas of the verification commissions.We would also like to thank the organizers and participants of the Global Race Project (INED/SciencesPo), the Núcleo Interdisciplinar de Estudos sobre a Desigualdade (NIED/UFRJ) seminar series and the Geneva Latinolab (UNIGE) seminar, as well as Chana Teeger, Luciana de Souza Leão, and the anonymous reviewers for their comments and suggestions on earlier drafts of this article.The Swiss Society of Sociology and Bruno Milanez and Beth Pasin organized wonderful writing retreats during 2022 that allowed us to finalize this article.We also acknowledge the financial support provided by the seed money from the Geneva Graduate Institute during the initial phases of this project, as well as FAPERJ, which supports Veronica Toste Daflon through a JCNE grant, and SNSF, which supports Camille Giraut through a Doc.CH scholarship.
1.The term race is always a contested category.Following a social constructivist approach, we understand race as a social construction.By focusing on how public policies remake race, our idea is exactly to emphasize the changing meanings, dynamics, and consequences of racial categorizing.
2. Following Paschel (2016), this article understands "Brazilian black movements" as a plurality of organizations and collectives unified by an imagined community that is multinational and ideologically and culturally diverse.
3. The growth of the Latino population has brought some challenges to these racial boundaries, as discussed by Mora (2014).Similarly, the recent scandals of Rachel Dolezal and Jessica Krug have raised new questions about the growing mismatch between self-identification and external racial appraisals, as discussed by Roth (2018).But these remain marginal cases, and crossing racial boundaries remains largely contested in the United States, as discussed by Brubaker (2016).
4. Due to space limitations, the present article focuses on the commissions' procedures to verify the identification of those who claim black (preta or parda) identification.Indigenous identification also goes through verification, but very often through a different procedure, more attached to official documentation, such as the RANI (Registro Administrativo de Nascimento de Indígena) document or a statement issued by the indigenous group to which the candidate belongs.See Silva et al. (2020, 336) for details.
5. Because these policies started in a decentralized fashion, different public universitiesstate and federal-had different types of quotas.While some relied only on census racial identification (i.e., identifying as preto or pardo), others relied on a combination of socioeconomic indicators (e.g., enrollment at a public school) and ethnoracial ones.However, since the beginning, the most controversial element has been ethnoracial, not socioeconomic (Daflon et al. 2013).In 2010, census respondents were asked, "What is your color or race?" and given five options to respond: branca (white), parda (brown), preta (black), amarela (yellow), or indígena (indigenous); 36.6 percent of the population identified their color or race as parda, and 14.1 percent as preta.
6. Between 2003 and 2010, the implementation of affirmative action policies was decentralized, and a few federal universities, like the University of Brasília, also relied on pictures to select racial quota beneficiaries.The 2012 federal law (Lei 12.711/2012) made racial quotas mandatory and standardized their implementation for undergraduate selection at all IFES.In the 2012 federal law, ethnoracial self-identification was combined with other socioeconomic markers, such as household income per capita and high school enrollment at a state school (most middle-class students in Brazil go to private schools).Racial quotas were established only for state school students, and the percentages varied by state, according to the racial demographics.
7. In January 2018, the newspaper Estado de São Paulo contacted 63 federal universities to ask about fraud allegations, of which 53 responded.Of those 53, 21 affirmed that they had received allegations of fraud and that investigations were currently being conducted.In total, 595 students were under investigation, and the majority had had their enrollment canceled, even if many were later able to return to the universities on the basis of legal appeals.See Toledo 2018.8.This applies only to selections that offer three or more positions.See DOU 2016.9.For the case studies, see Silveira 2015;Santos et al. 2019;Elísio et al. 2019;Marques et al. 2019;Maciel et al. 2019;Fonseca and Costa 2019;Dias and Tavares Júnior 2018;Passos 2019;Nunes 2019;Miranda et al. 2020;Silva et al. 2020;Silva and Custódio 2021;Silva and dos Santos 2021;Silva et al. 2022;Rodrigues 2022;. [Batista and Figueiredo 2020?]We also found two other papers published in Portuguese that analyze specific dimensions of the verification committees.Rosa et al. (2021) analyzes how committees at 17 universities dealt with COVID restrictions, Neves (2022) compares the debates about commissions in 3 universities, and A. Silva et al. (2020Silva et al. ( , 2022) ) provide a very general overview of the practices of these commissions across 20 universities but largely focus on one case and do not account for the 2018 ruling.
10.As of 2018, the website of the Brazilian Ministry of Education listed 109 IFES.Although all IFES have been legally obliged to adopt racial quotas since 2012, verification commissions were not mandatory for undergraduate selection processes in 2018.
11.The complete list of the IFES analyzed can be provided by the authors upon request.Due to space limitations, we have excluded universities that had only temporary commissions to analyze fraud allegations (comissões de sindicância).
12. For example, when the Federal University of Rio de Janeiro, one of the largest in the country, appointed a commission for the first time in 2020, it had to review 1,549 applications of students accepted through racial quotas; 1,270 attended their interviews over seven days, and