Intersectionality as a Regulative Ideal

2.1. Intersectionality as a Piece of Social Metaphysics or Social Science

Intersectionality is frequently glossed as a metaphysical claim about the mutual construction of certain social phenomena. We’ll call this the metaphysical co-constitution thesis. Precisely which entities are said to be co-constituting varies. The co-constitution thesis is often applied to social identities or categories (e.g., being a woman and being black), and to social structures and systems (the positioning and treatment of individuals and groups in a broader network of social relations). For example, Stephanie Shields defines intersectionality as “the mutually constitutive relations among social identities” (2008: 301), and Jennifer Nash asks if it provides “a general theory of identity” (2008: 10).[10] As Elizabeth Spelman puts it, we must reject the notion that:

Spelman’s point is that it is a mistake to think that a person may tease apart the different strands that constitute the complex yarn of her self-conception, as if, now, she identifies as a woman, later as black, and later still as gay. It is not just that a person may be all of these things together, but also that, in attempting to tease them apart, we obscure the ways they mutually constitute one another.[11]

Such claims about intersectional co-constitution are frequently made at the level of social metaphysics, that is, about the nature of identities, categories, or structures. This construal of intersectionality is also adopted by social scientists (in sociology, political science, psychology, economics, etc.).[12] Given that these categories are deemed to be co-constituting, intersectional social scientists further conclude that the categories cannot be studied, theorized, or generalized-about independently, as when Greenman and Xie state that there is no “pure” race or gender effect on earnings.

Claims about co-constitution reappear in other leading glosses of intersectionality as well—as when intersectionality is defined as part of a political theory of oppression, that is, as a normative concept for capturing certain forms of injustice. In this context, intersectionality is seen as more about the natures of racism and sexism than about race and gender.

2.2. Normative Co-Constitution and Structural Oppression

Many accounts of intersectionality treat it as a reminder to conceptualize sexism, racism, and other forms of oppression in terms of unjust social structures. In Crenshaw’s example (§1), the structural forces operating on undocumented, non-native-English-speaking women suffering from domestic violence are different in kind from those faced by documented women and American citizens. The Combahee River Collective coined the phrase “interlocking systems of oppression” to characterize such phenomena, while Collins (1990) refers to the “matrix of domination,” and Crenshaw (1991) to “structural intersectionality.” On these interpretations, intersectionality is a way of identifying the multiply disadvantaging dimensions of laws, institutions, and norms.[13] In this vein, Crenshaw (1989) focused on the inadequacies of existing law and judicial precedent for “the conceptualization, identification and remediation of race and sex discrimination” (1989: 140).

The normative-structural approach to intersectionality has also been used to criticize the marginalizing tendencies embedded within the leadership structures and priority-setting mechanisms of traditional social-justice movements, political discourses, and identity politics. Termed “political intersectionality” by Crenshaw, the claim here is that the political interests of women of color “tend not to be represented within the discourses of either feminism or antiracism” (Crenshaw 1991: 1243–1244; see also 2008; and King 1988), which have historically focused instead on advancing the interests of white women and black men, respectively. Political intersectionality points to a failure on the part of those attempting to change social reality to acknowledge and prioritize the interests of those who live on the margins.

2.3. Epistemology and Experience

Claims about social identity and structural oppression are also connected to epistemic, phenomenological, and psychological interpretations of intersectionality, which refer to the experience and knowledge of members of multiple disadvantaged groups. For example, Collins refers to “race, class, gender, sexuality, and national belonging as mutually constructing categories of experience” (2003: 209), and Crenshaw refers to “intersectional experiences” (1991: 1244). Here the idea is that what it’s like to be, for example, a woman and black cannot be neatly articulated in terms of this or that single-factor social grouping (Barvosa 2008). Such experiential claims are, in turn, related to epistemic interpretations of intersectionality, which enrich and complicate feminist epistemological traditions such as standpoint theory (Collins 1990; Grasswick 2018). The insight here is that the distinctive experiences of individuals who occupy multiply disadvantaged social positions give them privileged epistemic access to certain regions of social reality.

There are various ways of drawing the different interpretations of intersectionality discussed in §2.1 to §2.3 together. For example, Crenshaw argues that individuals’ intersectional experiences reflect their corresponding normative-structural intersectional oppression. She writes that “the experiences of women of color are frequently the product of intersecting patterns of racism and sexism” (Crenshaw 1991: 1243; see also Else-Quest & Hyde 2016: 319). And Collins (2003: 207) writes that, “Race, class, gender, and other markers of power intersect to produce social institutions that, in turn, construct groups that become defined by these characteristics.” Intuitively, our personalized social identities, group affiliations, and experiences are shaped (at least in part) by broader social structures. If the systems of oppression are fundamentally intertwined, then, one might think, so too will be people’s experiences of discrimination—and their understandings of what it means to inhabit various social identities. The intersection of structural inequalities will shape the psychological identities and experiences of individuals who occupy those intersections (and individuals who occupy different intersections). This then in turn explains why multiply marginalized individuals may have distinctive epistemic windows onto oppression.

2.4. Buzzword?

Yet given the diverse array of interpretations of intersectionality, some have grown skeptical of defining it at all. Davis (2008) argues that intersectionality is a “buzzword,” akin to a viral meme glommed onto by social-justice activists, and largely devoid of substantive theoretical content (cf. Hancock 2016: Chapter 1; Knapp 2005: 254–255). Building on a footnote from Crenshaw (1991: Note 9), Carastathis (2014a: 60) suggests that intersectionality is an irredeemably “provisional” or “transitional” concept, on the way toward a richer theory of social identity, structure, and justice. Similarly, Phoenix and Pattynama (2006: 187) claim that intersectionality is “useful as a handy catchall phrase that aims to make visible the multiple positioning that constitutes everyday life and the power relations that are central to it.”

2.5. A Commitment to Focusing on Women of Color?

Others object to the “traveling” of intersectionality beyond its roots in black feminism. In response to the viral-buzzword status of intersectionality and its application across a widening range of (sometimes depoliticized, mainstreamed, and institutionalized) contexts, some argue that intersectionality is, or should be, specifically about women of color. Thus Nikol Alexander-Floyd defines intersectionality as,

We’d like to echo Alexander-Floyd’s commitment to grounding intersectionality in the experiences of women of color, which is to say, in its ongoing living history. Any intersectional inquiry that is oblivious to its history and original aims risks the very erasure intersectionality means to resist. And we must, as Alexander-Floyd suggests, center our research and activist efforts on demarginalizing women of color. Yet we believe that intersectionality has much to teach and that its full epistemological import extends to a plurality of marginalizations (cf. Cho 2013; Táíwò 2018). The lessons to be learned from intersectionality must be historically grounded in the experiences of black women, but they are instructive for theorists interested in social reality and power in all of its detail. For example, we find evidence for ways of approaching social reality intersectionally that do not speak directly about women of color, such as Pedulla’s (2014) study of the intersection of race and sexuality in discrimination against different groups of men.

Intersectional thought reveals that we are all members of multiple social categories and that everyone exists within a network of relations of privileges and oppressions. For this reason, we believe that intersectionality is essential for inquiring into any dimension of social reality, including the positioning and experiences of members of multiply advantaged groups. White privilege, for example, is different for white men and white women, for rich whites and poor whites, and so on. Similarly, male privilege is different for cis men and trans men, for straight men and gay men, for white men and black men. It is intersectionality that teaches us these lessons. With these many lessons in mind, our model aims to respect intersectionality’s breadth as well.[15]

We are also moved by Jennifer Nash’s caution against treating intersectionality as coextensive with all research by or about black women:

2.6. Methodological Tool

There is, finally, an alternative set of approaches to intersectionality, which define it in methodological terms, as a claim about how to do social science and activism (Crenshaw 2011; May 2015: 19), rather than as a theory or hypothesis about how social reality is or ought to be. In this context, intersectionality is often referred to as a “heuristic device.” It is elsewhere described as a “critique of deeply entrenched cognitive habits” (Carastathis 2014b: 305), a “knowledge project” (Collins 2015: 3), and even as a Kuhnian “research paradigm” (Hancock 2007a). Similarly, Crenshaw (2015) has recently described intersectionality as “an analytic sensibility, a way of thinking about identity and its relationship to power.” Among existing interpretations, we are most attracted to (and influenced by) these, and we will say more about them in §4. First, however, we show how the various interpretations highlighted above threaten to lead to a vicious regress.

3.1. Regress Problems for Normative Intersectionality: Law and Policy

Perhaps the most notorious allusion to the regress problem appears in the ruling in DeGraffenreid v. General Motors Assembly Div., Etc. (1976), discussed by Crenshaw (1989: 141–3). In this case, a group of black women alleged that GM discriminated against them by failing to hire any black women before 1964 and by firing all the black women hired after 1970, when GM downsized. Noting that GM employed women and blacks (white women and black men!), the district court issued a summary judgment that black women are not “a special class to be protected from discrimination” (Degraffenreid v. General Motors Assembly Div., Etc. 1976, cited in Crenshaw 1989: 141). The plaintiffs had to prove “race discrimination, sex discrimination, or alternatively either, but not a combination of both” (Degraffenreid v. General Motors Assembly Div., Etc. 1976). Among their reasons for this decision, “The prospect of the creation of new classes of protected minorities, governed only by the mathematical principles of permutation and combination, clearly raises the prospect of opening the hackneyed Pandora’s box.”

The court’s concern seems to be that permitting claims of combined discrimination will allow all manner of gerrymandered collections of individuals to claim group-based mistreatment. While it strikes us as obvious that black women can suffer wrongful discrimination specifically as black women, other combinatory classes may not be equally obvious candidates for discrimination. The following categories are currently protected against discrimination by U.S. federal law: age (being 40 or over), disability, genetic information, national origin, pregnancy, race/color, religion, and sex.[16] Are all combinations of these categories potential targets of discrimination? For example, suppose instead that GM had hired people over 40 and Muslims, but no Muslims over 40. Should federal law protect individuals not just from ageist and religious discrimination, but also from combined, age-religion discrimination? Can a group be discriminated against specifically as Muslims over 40? From the armchair (or the bench), initial intuitions about this example may be less clear cut, but it is not at all difficult to imagine a situation in which such a case could be made. (Imagine, say, that an ignorant and xenophobic CEO had sent a memo to HR asserting that older Muslims, unlike younger Muslims and unlike older Christians, were less likely to assimilate and “fit” into their workplace culture, and therefore should not be hired.)

But now suppose that GM had hired black women, people over 40, Muslims, pregnant people, and American citizens originally from Canada—as well as black Muslim women over 40, and Canadian-born pregnant women, etc.—but no black Muslim pregnant women over 40 from Canada. Can a group be discriminated against specifically as Canadian-born, black, Muslim, pregnant women over 40? It’s certainly logically possible. However, if intersectionality literally means that all the various forms of group-based discrimination, in principle, cannot be understood in isolation, and must instead be understood as co-constituting, then at first glance it seems the answer must be not “possibly yes” but “necessarily yes.” The answer will be yes for every possible permutation of every intersection of every social category.

The problems here are, in fact, more acute and important than they may initially appear. It is not just that the courts must allow any potential combinatory classes to pursue discrimination claims in court. Discrimination is subtle and insidious, and we may welcome general legal norms that recognize this. Yet it is also that, even if we limit the axes to the limited number recognized by federal law, class-based discrimination claims become nearly impossible to bring. Worse, if we open the door to further axes beyond those recognized by federal law, allowing indefinitely many co-constituting factors to make a significant difference to the nature of discrimination, class-based claims verge on incoherence. To see the problem, suppose a group of twelve Asian-American women allege race-based sexual harassment. If a co-constitutional intersectional model of discrimination is adopted into the legal system, it is then open to the defendants to argue, precisely because there are indefinitely many other factors, that discrimination of this kind is in-principle impossible, given the many real differences among the plaintiffs. Some, they might say, are of Vietnamese heritage, some are of Chinese heritage; some are from middle-class backgrounds, others are from working-class backgrounds; some are queer, some are straight; some speak several languages at home, some speak only English. The worry is that these twelve women cannot represent a protected class, because the discrimination suffered by the middle-class Vietnamese woman cannot in principle be the same as the discrimination facing the working-class Chinese woman. They could not claim discrimination qua Asian-American women, because each of their Asian-American, gendered experiences would be distinct. They could not purport to represent all Asian-American women, let alone all women of color, all women, or all people of color (cf. Crenshaw 1989: 143–149).[17]

So, while the initial worry might be that intersectionality entails that any intersection, however gerrymandered it may seem, may constitute a protected group—and thereby strain plausibility—the second and more troubling worry is that intersectionality, taken as a general principle, entails that almost no one suffers from discrimination like anyone else. The worry is that intersectionality, taken to its “logical conclusion,” entails the dissolution of social categories and therefore group-based discrimination claims altogether.

3.2. Regress Problems in Social Science: Down the Rabbit Hole of Difference

Analogous worries arise if we read intersectionality as a general social-scientific hypothesis, theory, or law. Recall Greenman and Xie’s (2008) general claim that there is no such thing as a “pure” effect of race or gender when it comes to earnings, and their corollary insistence that the two social categories must be considered simultaneously. Why think that the important intersections stop with race and gender? Why, for the purposes of social-scientific generalization, should considering the interactions of race and gender be sufficient? Surely attention to class, religion, geography, education status, ability, citizenship/immigration status, and age, just to name a few, will reveal that there is no “pure race-and-gender” effect either.

Take sexuality, for example. Recall Pedulla’s (2014) experimental finding that straight black men and gay white men suffered a wage penalty relative to straight white men, but that gay black men suffered no such penalty. Outside the lab, meta-analyses show that gay men are paid significantly less on average than straight men, whereas lesbian women are paid more than straight women (Bagri 2017; Klawitter 2015).[18] These studies suggest that there is no “pure” effect of sexuality on earnings independently from race and gender. Given these intersectional complexities, what justification can social scientists give for considering only race and gender? Worse, if we take intersectionality seriously as a universal law in social science, then it seems like true generalization just as such is suspect. If intersectionality entails that social categories are indefinitely mutually co-constituting, then social reality is, at bottom, highly particularistic, which calls into question the possibility of social knowledge itself.[19]

One might think that the regress problem for social knowledge is a merely “academic” concern, with limited implications for research. Many social scientists, especially in informal conversation, are ready to admit that they are in the business of making knowingly “rough” generalizations, whether about single-factor social groups (e.g., all black people) or double-factor social groups (e.g., all black women). The regress worries, they might say, can be deflected simply by including what might be called an “intersectional ceteris-paribus clause” that their generalizations will break down when more specific social groupings are examined. However, this response to the regress problem may have unacceptable political (and epistemic) effects. If intersectionality is a mere asterisk to the rough generalizations of traditional social science, then the door is once again reopened to further disappearing those who live on the margins. One of the primary impetuses for the intersectional turn is that too often abstract generalizations about the “whole group” reflect the experiences and positioning of otherwise-privileged group members, and thus render others all the more invisible, especially the smaller numerical minorities within these groups, such as LGBTQAI+ or disabled individuals.

3.3. Metaphysical, Normative, and Epistemic Regress Problems

With the examples from §3.1–2 in view, the general difficulties become apparent for the various interpretations of intersectionality and the various guises of the co-constitution thesis summarized in §2. Taken, for example, as a claim about the nature of oppression, intersectionality entails that phenomena such as “black oppression” and “white privilege” are not genuine kinds, because gender and race are “mutually constitutive” (Shields 2008: 301). According to this thesis, that is, claims about the nature of black oppression or white privilege are bound to be false or profoundly misleading unless they are articulated in relation to gender, because what it means to be oppressed in virtue of blackness differs for black men and black women. By the same token, however, “black women’s oppression” isn’t a genuine kind either, because gender, race, and class intersect: what it means to be oppressed in virtue of black-womanhood differs for rich and poor black women. The same goes for sexuality, ability, religion, and a host of other significant social categories, potentially ad infinitum.

The problems facing epistemic and experiential interpretations of intersectionality are analogous. We would be forced to draw the untenable and absurd conclusion that, for example, black people don’t know what it’s like to be black, that Latina women don’t know what it’s like to be Latina women, and so on. They can only know what it’s like to be, say, cis queer able-bodied . . . etc., members of these groups.

In fact, the range of social categories and phenomena invoked in relation to intersectionality has been remarkably extensive (and far exceeds the EEOC’s grievously brief list of protected classes), to include race, ethnicity (which is, in turn, a complex construct involving numerous intersecting subcomponents, e.g., related to cultural and ancestral heritage as well as appearance, including skin color, eye color, hair, eye and nose shape, etc.), gender (which of course is not a binary phenomenon, and must be further partitioned to reflect differences related to being, e.g., cis, non-binary, trans—e.g., Bettcher 2017; Howansky, Wilton, Young, Abrams, & Clapham 2019—or intersex—e.g., Carpenter 2014), sexuality (which is itself a complex, multidimensional spectrum), religion, geography, neighborhood, nationality, language, borders (e.g., Anzaldúa, Cantú, & Hurtado 2012; Moraga & Anzaldúa 2015), capitalism, colonialism (e.g., Lugones 2007; 2010), employment, education, age, generation, ability (e.g., Frederick & Shifrer 2018), class (e.g., Brannon, Higginbotham, & Henderson 2017), culture, height, weight (e.g., van Amsterdam 2013), relationship status, personality, facial symmetry, citizenship, migrant status, carceral status, and so on.

Surveying all these categories, one might be tempted to conclude that the co-constitution thesis ultimately entails the dissolution of groups and a return to individualism. This is the threat that Iris Marion Young (1994: 718–721) and Nancy Ehrenreich take the regress problem to pose. Ehrenreich’s focus is particularly on the potential obstacles intersectionality creates for political activism and coalition building, pointing to “the tendency of all identity groups to split into ever-smaller subgroups, until there seems to be no hope of any coherent category other than the individual” (2002: 268). It would seem that intersectionality, which begins as an insistence on the importance of group-based categories for understanding social identity, knowledge, and justice, would devolve into another form of individualism, such that the individual becomes “the only unit of analysis, making group-based critiques of power hierarchies impossible” (2002: 271).[20]

But the regress does not stop at the individual. Each individual “intersects” with these categories in different ways across time and space. One cannot, for example, generalize about the political interests, privileges, or experiences of a given individual (call her Reema) without considering Reema when she is employed versus unemployed, young versus old (and her correlative changes in ability), single versus in a committed relationship, and all the changes she undergoes in neighborhood, education, health, pregnancy, parenthood, class, nationality, religion, personality, mood, and contextual specificities (e.g., what it’s like for Reema when she is the only member of a marginalized group in a given context—“solo status”—versus when she is part of a “critical mass,” or a slim majority, or one among many in an overwhelmingly homogeneous social setting, or when she “passes” as a member of a different group, etc.).[21] Does the intersectional effort to advance our understanding of social kinds and injustices instead devolve into a social-theoretic commitment to nothing but isolated, atomic space-time points?

We take the foregoing considerations to suggest that intersectionality should not be understood as a general theoretical claim: neither an in-principle metaphysical thesis about social kinds, nor a universally applicable normative thesis about oppression, nor a general law in social science, etc. We turn next to a more promising set of interpretations.

5.1. Regulative Ideas

It is internal to the nature of reason, according to Kant, that, as inquirers into nature, we aim at an ideal of systematicity (Kant 1999: A645/B673).[24] In the advancement of knowledge, we should have an eye toward whether our newly formed concepts and beliefs systematically cohere with our prior concepts and beliefs. Operating with the aim of systematicity directs us how and where to look for new information. But this ideal, for Kant, is merely regulative. That is, it is an aim that we must assume to advance inquiry, but does not thereby also theoretically posit anything about what nature is actually like, or about what we, necessarily, will be able to achieve through inquiry.[25] In fact, Kant denies that we are or ever will be in a position to assert that nature, as we know it empirically, is in fact globally systematic. That is, he denies that we can know whether the elements of nature actually cohere as a unified system describable by a unified overarching empirical law. The ideal of systematicity instead has a purely regulative use.

To elaborate the content of the ideal of systematicity, Kant introduces three criteria, which he refers to as the principles of homogeneity, specification, and affinity. An empirical inquiry aims at systematicity only if it:

1. homogeneity: aims to unify diverse concepts or findings under higher, or more general, concepts or laws
2. specification: aims to specify difference or deviation within existing empirical categories or classifications
3. affinity: pursues a continuity of forms, according to which:
   1. for any two given distinguished species, or groupings, a and b there is some higher genus, or grouping, g (if we ascend high enough) that reflects some property p shared by a and b, and
   2. for any given species or grouping s there are distinct subspecies x and y that have different properties c and d while sharing the property e essential for membership in s. (See Guyer 1990: 24)

Let’s consider some (simplistic) examples. We pursue the principle of homogeneity when, for example, we seek higher genera given a certain diversity of species. Suppose we’ve distinguished two groupings of animals, feline and canine. Homogeneity as a principle of reason tells us, in seeking new knowledge, to search for some way of unifying the two. So, we investigate the properties they share and ultimately posit the higher genus, mammal. Similarly, suppose I am committed to the existence of two given fundamental forces, such as electricity and magnetism. The aim for rational systematicity, via the principle of homogeneity, directs me to seek a higher more general fundamental force (in this case, electromagnetism) that explains and unifies the lower forces.

Now consider the principle of specification. Suppose I’ve posited the animal grouping, canine. Specification compels me to seek important differences within this grouping such that I discover a diversity of subspecies, for example, domestic dogs, coyotes, and wolves. Similarly, due to the way specification guides inquiry, we might seek out variety in the ways a given general empirical principle or law is instantiated in different contexts. I may be, for example, compelled to inquire into whether there are important differences in the way, say, gravitational forces apply to large celestial bodies as opposed to teeny tiny bodies.

Notice that in these examples, the regulative principles operate to guide inquiry in light of some antecedent concepts, classifications, or principles, but with the aim of uncovering new higher general concepts or principles or new lower further-specified categories. They are “maxims” for gathering new knowledge in light of existing classifications and so are, at least in part, of a higher-order nature. In neither case is it posited, from the outset, that we necessarily will discover this or that unity or diversity in any given case. Rather, while we bring our assumptions and principles of reason to nature, what we empirically find may or may not itself be systematic.

Kant’s discussion of the regulative principles focuses on natural science, but we can see how the principles of unity and specification apply in more obviously social domains. We find the principle of homogeneity at work in, say, inquiring into the similarities in the ways that both pregnant and disabled people are excluded from the workplace via architectural design and social stigma. We find the principle of specification at work when we inquire into the disparate impact of facially race-neutral voting requirements on members of different racial groups. As guiding principles of reason, we should expect to find instances of them in all areas of inquiry.

5.2. Intersectionality as a Regulative Ideal

At this point, readers may have already noticed similarities between intersectionality’s imperative to “ask the other question,” and Kant’s principle of specification. One aspect of the intersectional critique is that feminists, race theorists, and social scientists historically pursued homogeneity (i.e., unification and simplification) too single-mindedly, to the exclusion of specification. Generalizations about the nature of these social categories, and how they relate to power and inequality, tend to ignore the specific experiences and forms of discrimination faced by members of multiple disadvantaged groups. There is some precedent for understanding intersectionality in this way. For example, Warner, Settles, and Shields (2016: 173) write that:

Intersectional approaches to social reality and oppression very often reveal significant diversity where there had been an assumed homogeneity. In many cases the assumed homogeneity is the result of a problematic generalization, for example, when white middle-class women treat their experiences and interests as representative of all women. This problematic universalizing can occur along many axes of privilege, including whiteness, maleness, able-bodiedness, cisness, and wealth. Intersectionality directs us to challenge this (false or misleading) homogeneity; similarly, according to Kant’s regulative principles, coming to a more systematic view of things requires that we must always seek diversity within apparently homogeneous categories.

But the principle of specification is not equivalent to intersectionality. For Kant, systematicity is the ideal or aim of knowledge as such, understood as a neutral, unbiased, and largely individual enterprise. The histories of critical feminist philosophy and sociology of science have taught us that science, in fact, does not operate in this ideal way (Crasnow, Wylie, Bauchspies, & Potter 2018). Kant simply did not imagine the ways that political and social-cognitive factors influence which unities and specificities actually get pursued. Intersectionality, by contrast, emerges from and is situated within traditions of social thought that emphasize that the generalizations we make, and who gets to make them, are inevitably political. Kant’s regulative principles do not account for this.

Given the realities of bias and privilege, an inquirer into social reality might sincerely and wholeheartedly pursue Kant’s regulative principles of unity and specification without adopting a genuinely intersectional methodology, and perhaps without ever arriving at an intersectional question or claim. Consider the study of sexuality. Suppose a naïve theorist originally believed different-sex attraction was the only kind of sexuality, but eventually (following the principle of specification) hypothesized that there were two types: different-sex and same-sex attraction. Enter the Kinsey scale of sexuality, which challenges this traditional sexuality binary. On the original 7-point Kinsey scale from 1948, all males are said to fall within two poles of sexuality, “exclusively heterosexual” and “exclusively homosexual,” with relatively few males falling at the poles themselves (Kinsey 1998). Arguably the Kinsey scale follows the principle of specification, revealing new diversity within an existing simplistic binary classification scheme. Yet the Kinsey scale itself fails not only with respect to gender diversity (initially focusing only on cis men and the traditional gender binary),[26] it also fails to investigate the interaction of other classifications with its central categories of sexuality. So, although it specifies, it is clearly not intersectional.

Similarly, we can imagine a researcher who devotes their career to studying race and along the way discovers, as a result of following the principle of specification, that whiteness has been constructed, interpreted, and demarcated in different ways across time and space. Suppose the researcher comes to see that the present-day experiences and privileges of “American whiteness” differ in important respects from those of, for example, “Australian whiteness” and “European whiteness.” Then the researcher might realize that European whiteness itself must be further specified into Scandinavian, German, Eastern European “whitenesses,” and so on. Conversely, the researcher might come to identify previously unappreciated similarities between different manifestations of whiteness, for example, exploring the shared legacy of segregation on the present-day experiences and privileges of South-African and Southeastern-American whiteness, and so on. The researcher could indefinitely seek out such similarities and differences among the multifarious constructions of whiteness, and thereby advance our understanding of whiteness in numerous respects, without ever investigating the intersection of whiteness with gender, sexuality, class, or ability. They might, despite conscientiously following the principles of specification and homogeneity with the best of epistemic intentions and to the best of their investigative abilities, never stop to consider whether, say, whiteness confers different privileges on men versus women, on the wealthy versus the working-class, and so on. In this example, the sincere social inquirer searches for unity and diversity within a single classification scheme (namely, a geographically-distributed ethnic-racial classification scheme) without considering forms of unity and diversity that might emerge across separate classification schemes (i.e., what emerges when racial and gender classification schemes are considered in tandem).

Pursuing intersectionality is thus not reducible to pursuing any of Kant’s principles, or their combination. This should not be surprising. It is possible to earnestly pursue specification and homogeneity (i.e., diversity and unity) without adopting an intersectional approach or raising an intersectional question. Seeking out specification and homogeneity is, we believe necessary, but certainly not sufficient, for an intersectional approach.

We believe that intersectionality fills some of the gaps in and corrects some of the mistakes of traditional social inquiry because it is premised on the lived realities of injustice. It is a concept first born out of the experiences of practical, epistemic, and hermeneutical injustice faced by women of color at the hands of white feminisms and traditional race theories. In seeking unity and diversity, intersectionality looks for the ways existing concepts or categories interact, but it looks with a particular purpose, namely that of correcting injustice (e.g., erasure, marginalization) within those existing categories. Intersectionality therefore should be considered an independent regulative ideal on all forms of human and social inquiry. It is a regulative ideal for non-ideal scientific and activist communities in a non-ideal world.

5.3. The Content of the Ideal

As we understand it, the intersectional regulative ideal has two components or aspects. One component of the ideal calls for treating existing classification schemes as if they are indefinitely mutually informing. The directive is to seek out new or underappreciated ways in which different, ostensibly comprehensive but single-factor schemes for classifying individuals (by race, gender, class, etc.) can be analyzed in tandem. Following this directive will reveal new or underappreciated unity and diversity within and across existing theoretical and practical schemas. Like the Kantian regulative principles, the ideal does not imply a priori that these distinctive schemes are necessarily or will inevitably turn out to be mutually informing, but rather that it is an imperative that we must seek out the ways they might be, in gathering new knowledge about human beings and social reality. Whether an inquiry reveals an actual interaction in a given case, and when an intersectional analysis will be fruitful, are, on this model, both empirical questions. It is possible that we may not find the “intersectional result” that any two or more schemes are in fact mutually informing.

However, to interpret the intersectional ideal as nothing but the mandate to treat classification schemes as if they are mutually informing would be to strip it of its intrinsic critical meaning. Intersectionality emerged from critical discourses the central aims of which were to understand and resist injustices, including marginalizing tendencies in social theory and practice (Bilge 2013; Dhamoon 2011). Thus a second component of the intersectional ideal is explicitly and essentially political. In understanding human beings and their positions in the world, intersectionality aims to train our theoretical and practical gazes on the ways that injustices influence the production of knowledge and meanings, the formation of identities and experiences, and the constructions of social institutions, practices, and policies. The ideal must make essential reference to questions related to power, privilege, inequality, and injustice. In respecting both of these aspects of intersectionality, we suggest the following formulation of the ideal:

The content of the ideal thus involves two essential aspects, the first (a), much like the Kantian regulative principles, tells which general kinds of things to look for (interactions between existing classifications or groups), while the second (b), very much unlike the Kantian principles, tells how and why to look for them (with the aim of resisting injustice and inequality, which may be involved in the very construction of those classification schemes or categories).

This general maxim of intersectionality can then be applied to particular domains to generate more specific heuristics. For example, applying the ideal to the epistemological domain, we might generate the following more specific maxim:

Similarly, applying the ideal to the production of meanings and interpretations, we can generate the following heuristic:

Likewise, we can apply it to the formation of political coalitions:

We can generate similar intersectional heuristics for theorizing about the creation of identities or subjectivities; for the construction of social institutions and policies; and for the analysis of psychological categories and kinds (including desires, stereotypes, and biases). For this reason, modeling intersectionality as a regulative ideal represents one way to bring some unity to the various and multifaceted interpretations and applications of intersectionality. We can see how pursuing the ideal (i.e., following these maxims) can help us to uncover the ways certain social categories or identities can be mutually informing (e.g., blackness, womanhood); how certain forms of oppression interact (e.g., homophobia, sexism); how certain experiences are constructed (e.g., the painful “in-betweenness” of the experiences of the historically colonized; Anzaldúa et al. 2012); and how certain identities end up being marginalized in political discourse. It also explains how intersectionality is as much about uncovering intergroup similarities and shared struggles as it is about revealing differences (§5.6; see also Collins 2003: 207; Cole & Luna 2010; Doetsch-Kidder 2013; Moraga 2015: 29). It guides us to search for common forms of oppression for members of groups that are disadvantaged along different axes, for example, sexual and racial minorities (Cohen 1997; Cortland et al. 2017). By applying intersectionality as a general regulative ideal to particular domains we also arrive at some unity with respect to the different heuristics currently associated with intersectionality, including Matsuda’s (1991) normative heuristic, “Is there x-ism in this y-ism?”[28]

Interpreting intersectionality as a regulative ideal models it as a methodological principle, but diverges from some existing methodological approaches in significant ways. For example, Ange-Marie Hancock (2007a; 2007b) argues that intersectionality is a Kuhnian research paradigm. There is much to recommend Hancock’s methodology-centered analysis, which is one of those most closely aligned with, and influential upon, ours. (Consider, e.g., Hancock’s, 2016: 16, recent description of intersectionality as “a pathbreaking analytical framework for understanding questions of inequality and injustice.”) However, Hancock has elsewhere identified some limitations with interpreting intersectionality as a research paradigm (Hancock 2013), and we suspect there are others.[29]

Specifically, classifying intersectionality this way risks being both too narrow (by not covering enough cases) and too broad (by incorporating commitments that, we think, intersectional theorists should resist). It may be too narrow insofar as scientific paradigms are usually disciplinary and domain-specific, and inherently committed to a narrow set of methods, whereas intersectionality is a methodological imperative reaching across existing research paradigms, disciplinary boundaries, and methods of knowledge production and activism. Intersectionality is as much an ideal for studying unjust social institutions and the nature of oppression as it is for research in psychology or medicine. It is a guiding aim of both quantitative and qualitative research, as well as various hermeneutical and artistic projects, including memoir, poetry, and countless others. Conversely, modeling intersectionality as a Kuhnian paradigm may also be too broad and theoretically committal, insofar as scientific paradigms fundamentally involve commitments to general theories, laws, and principles, some of which are taken to directly describe reality in general ways. Operating within a research paradigm means engaging in what Kuhn calls “normal science,” which depends on a consensus about the basic laws governing the objects of study, and precisely how best to study them (taken-for-granted “rules that tell the practitioner . . . what both the world and his science are like”) (Kuhn 1996: 42).[30] In effect, we worry that treating intersectionality as a paradigm threatens to smuggle in the co-constitution thesis as a general law internal to the paradigm, which would lead to the regress problem (§3, §5.7).

5.4. Why Unify?

Our interpretation represents one way among many to see how different invocations of intersectionality hang together (see, e.g., §2.3 for others). However, it is reasonable to step back and ask why a unified interpretation is desirable in the first place.[31] After all, as we mentioned in §5.2, some argue that one of intersectionality’s principal teachings is to resist efforts to articulate overly parsimonious empirical theories or to form artificially unified coalitions, insofar as these efforts threaten to obscure intragroup differences (Warner, Settles, & Shields 2016: 173). Intersectional scholars have rightly highlighted the perils of certain projects of unification, or “universalizing tendencies” (Alexander-Floyd 2012). Crenshaw (1991: 1244), for example, stresses that she is not offering “some new, totalizing theory of identity.” Perhaps pursuing a unified account of intersectionality, just as such, faces similar pitfalls.

While we agree that the perils of unchecked unificationism are real (indeed, these pitfalls are central to our analysis), there are perils of unchecked anti-unificationism as well, as existing scholarship by women of color teaches. Consider Uma Narayan’s (1997: Chapter 3) analysis of the racist tendency in the United States to interpret dowry murder as an “exotic” cultural practice endemic to Hinduism or South Asia, when it is in reality highly similar and continuous in profound ways with patterns of domestic violence in America and elsewhere. This “othering” of “Hindu culture” is itself oppressive and leads to false exoticizing ideas.

Similarly, there are risks in not pursuing any unity in the way we think about intersectionality, especially given existing power relations in the construction of knowledge. For one, the term may become so variously used that it is, or is viewed by some as, “thinned out,” and thus consequently abused, dismissed, or relegated to vacuous “buzzword” status (§2.4-5). For another, there are risks of epistemic loss and missed opportunities for coalitions when we fail to see an intersectional phenomenon as intersectional. Collins (2015: 15) writes about the puzzled reactions she received after giving a talk on intersectionality at the Afro-Latin and Afro-Caribbean Women’s Festival in Brazil, when she was approached by a group of Afro-Brazilian women scholar-activists who thought that intersectionality “had nothing to do with” them. According to Collins, they were engaged in an intersectional social justice project without knowing it and under the impression that intersectionality was just for white feminists. At a minimum, cases like this illustrate the live possibility of talking past each other in scholarly and activist intersectional endeavors (a practical cost) and of losing knowledge (a theoretical cost). The implications of such misunderstandings should not be underestimated. Not being able to see one’s movement or project as intersectional could mean missing out on the existing wealth of intersectional resources and history to draw on. Recognizing what’s shared across diverse intersectional endeavors is thus also a matter of hermeneutical justice, of having conceptual resources adequate to the tasks of understanding and articulating social experiences of oppression. Such misunderstandings can, moreover, clearly put up obstacles to coalition-building. Two grass-roots intersectional movements are presumably less likely to seek opportunities to work together to the extent that one proudly adopts the label “intersectional” while the other interprets that label as meaning “for white feminists only.”

We therefore conclude that unified interpretations of intersectionality (especially those that remain cognizant of the perils of unificationism) can offer both theoretical and practical value. Specifically, interpreting intersectionality as a regulative ideal makes it clear that intersectionality is not a label or badge that a movement or lab can earn once and wear forever; it is an ever-present imperative, guiding research and activism over time. For example, it is not enough to invoke intersectionality in the initial articulation of a movement platform or social theory; the regulative ideal applies continuously and critically as movements and research programs develop. It directs all practitioners to stay vigilant about the risks of further marginalizing less visible members of groups, as well as to stay on the lookout for intergroup similarities and shared aims. In the final three sections, we illustrate how the intersectional regulative ideal applies in social science (§5.5) and activism (§5.6), and makes generative, forward-looking contributions to inquiry and activism (§5.7).

5.5. Applying the Ideal in Social-Scientific Practice

For a better sense of what intersectional methodology entails, it may help to see how the regulative ideal applies in practice. Recall Pedulla’s (2014) study on salary discrimination. For the purposes of illustration, we might imagine that the research proceeded along the following lines. Pedulla could have started with the observation that race affects income. This is apparent in the overall average data. The intersectional ideal then directs him to treat social classification schemes as if they are mutually informing, in a way that reveals and resists the effects of inequality and injustice. The ideal thus instructs him to consider the way that current understandings of inequality might interact. In this case, Pedulla considers whether race and sexuality are mutually informing when it comes to income, and looks for unjustly marginalized group members whose experiences and positioning are wrongfully obscured by the overall average. So he asks, “Is there heterosexism in this racism?” At this stage he is pursuing diversity (specification), hunting out subtypes and exceptions to existing generalizations. He might have hypothesized an additive disadvantage here (sometimes referred to as “double jeopardy”), such that gay black men suffer an increased penalty relative to straight black men. However, he finds the surprising result that gay black men do not (in this specific context) suffer an additive disadvantage and, unlike straight black men and gay white men, are not offered a lower salary than straight white men (see also Wilson et al. 2017).

Now that he has uncovered this diversity, he is led to ask why. He transitions to a new stage of inquiry (unity, or homogeneity), seeking out more general social hypotheses that might explain the deviation from the more simplistic generalization. In this case, Pedulla had (in fact, not just in our fictionalized illustration) hypothesized that psychological stereotypes intersect. He measured participants’ perceptions of how threatening and feminine the job applicants were. Black men are stereotyped as threatening and hyper-masculine (e.g., Johnson, Freeman, & Pauker 2012), whereas gay men are stereotyped as feminine (e.g., Petsko & Bodenhausen 2019). In the case of the gay black male applicant, these two stereotypes “canceled out”: he was perceived as less threatening because of his perceived femininity, and these perceptions in turn predicted his salary. In this case, a broader, unifying explanation refers to the cognitive operations of participants’ stereotypes. In this way Pedulla’s exploration of stereotypes about femininity injects a further intersectional dimension into the inquiry, by unearthing sexism within existing racist homophobia. The three classification schemes of racism, sexism, and homophobia are mutually informing in the psychological context of stereotype activation and application.

To reiterate, we do not mean to suggest that intersectionality is only theoretically at home in quantitative social-scientific investigations like Pedulla’s. The ideal applies across a broad range of epistemic projects. Thus we part ways with Nash who regrets that we have “not yet developed a rigorous method of examining multiple subject positions” (2008: 4–5). On our view, this is a feature, not a bug. Social scientists, metaphysicians, and other theorists can hunt for mutually informing categorization schemes, in-group differences, and between-group similarities in myriad ways. There is no need to “prioritize” multifactor quantitative analysis over, say, qualitative interviews and case studies of specific groups. Modeling intersectionality as a regulative ideal is thus consistent with Bright, Malinsky, and Thompson’s (2016: 62) call for a pluralistic understanding of intersectional methodologies.

5.6. Applying the Intersectional Ideal in Activism

It is also important to see how intersectionality as a regulative ideal is a maxim for activism and coalition-building. Consider Veronica Terriquez’s (2015) rich analysis of the complex intersections between immigrant and LGBTQ activism. In the early 2010s, undocumented immigrant youth (DREAMers) sought to both destigmatize their status and catalyze the movement toward immigration justice. For the purposes of illustration, we might imagine that the intersectional ideal directed them to ask how different identity-based movements might mutually inform their respective struggles to resist injustice. In this case, the activists effectively asked, “how can queer activism inform the immigration rights movement?” They began by seeking common ground across social difference (unity, homogeneity), finding inspiration from the queer “coming out of the closet” narrative. DREAMer activists then organized a “Coming Out of the Shadows” initiative, whereby they “publicly declared their undocumented status in order to combat the stigma associated with their precarious legal situation and humanize their experiences in the eyes of broader audiences” (Terriquez 2015: 344).

Concurrently, the DREAMer movement also turned inward (specification), toward the multiply marginalized individuals within their community, asking (in effect), “Is there heterosexism within immigration activism? Is our movement too focused on undocumented youth who are ‘otherwise privileged’ in virtue of being straight and cis?” In fact, queer undocumented youth face distinctive forms of heterosexism. As Terriquez (2015: 352) and one of her interviewees explain:

To both draw attention to and empower undocumented queer youth, immigration groups invested in several strategies. They created the Queer Undocumented Immigrant Project, which aimed to promote LGBTQ leadership in immigration groups, and Queer Dream Summer, which set up internships for undocumented queer youth in LGBTQ organizations (Terriquez 2015: 355). Immigration rights groups also organized workshops, on campuses and in wider communities, to introduce “straight members to the experiences of the movement’s queer-identified members,” including “testimonials from queer members about their LGBTQ coming out experiences and the impact of homophobia on their lives” (2015: 355). Notably, this attention to intragroup difference (specification) did not function primarily to subdivide immigration groups (much less did it form an endless regress toward individualism), but had just the opposite (unifying) effect, drawing “connections between living in the shadows and living in the closet” (2015: 355). Lifting up the voices of multiply marginalized group members did not tear the movement apart. It ultimately highlighted points of common humanity and struggle across crosscutting dimensions of social difference.

These strategies paid numerous dividends. Adopting the coming-out narrative not only inspired undocumented youth to leave “the shadows,” but also inspired queer undocumented youth to come out as queer. As the coming-out narrative “traveled” out of its original LGBTQ context, it also traveled “back around” and reciprocally reinforced queer activism. Terriquez refers to this as a “boomerang effect,” whereby “strategies originated by movement A are borrowed by movement B in pursuit of separate goals; yet in employing that strategy, movement B ends up furthering the aims of movement A” (2015: 347).

The courage and strength that undocumented queer youth found in coming out not once but twice helped these doubly-marginalized individuals become doubly empowered, or, as Terriquez says, “intersectionally mobilized.” Specifically, she offers tentative grounds to think that undocumented queer youth are even “more civically engaged than their straight-identified peers” (2015: 344), and especially well-represented in leadership positions in immigrant rights groups. “LGBTQ individuals, who first become empowered around their sexual orientation as a result of their ties to the LGBTQ movement, were primed to come out with respect to their other identities and play a leadership role in another movement” (2015: 347). Thus, recognizing multiply disadvantaged subgroups need not lead to internal division within a movement, but rather may very well increase activism within that subgroup, the members of which may then take on leadership roles in the organization (see also King 1988: 54). Terriquez concludes “that attention to the interests, needs, and unique experiences of and by groups who experience multiple identity-based hardships . . . can not only assist these groups in overcoming barriers to political activism, but also inspire high levels of commitment and activism” (2015: 358).

This insight is firmly in keeping with the black-feminist tradition, and resonant with, for example, Audre Lorde’s celebrated argument for the productive rather than destructive power of difference:

She called instead for “learning how to take our differences and make them strengths.”

Of course, Terriquez also emphasizes that reciprocally reinforcing boomerang effects are by no means guaranteed to occur when one movement “borrows” another’s strategies. First, such strategies risk appropriation as much as collaboration (see also §2.5). Second, and relatedly, boomerang effects don’t just happen; they depend on cooperative agency, and likely “require intentional efforts by activists to bridge two movements” (2015: 348). But lastly, the “intersectional result” in this case, of generating mutually reinforcing activist momentum across distinctive movements, may simply never materialize. Again, qua regulative ideal, the intersectional activist ideal directs us to look for such reciprocal movement-enriching opportunities. The benefits of boomerang effects are obvious—making intersectional social movements stronger than the sum of their parts—but there is no guarantee that they will appear.

5.7. Escape the Regress Problem?

We are now in a position to see how modeling intersectionality as a regulative ideal avoids the regress problem. A vicious regress arises only if we commit to a version of the co-constitution thesis according to which some aspect of social reality (or experience, identity, oppression, etc.) is in all cases and indefinitely intersectional. The vicious form of the regress arises only if we adopt the co-constitution thesis as a general law—that is, as universally true of a given domain. Treating the co-constitution thesis this way calls into question the validity of social categories, kinds, or generalizations as such. If all social categories are mutually co-constituting and there are indefinitely many social categories, then the regress takes off a priori, and thereby demands that we find a stopping point, or solution, for the regress from the armchair. Worse, insofar as the co-constitution thesis is precisely intended to be a thesis about the nature of social groups (kinds or generalizations), and because the regress entails that there are no such groups, the co-constitution thesis is unavoidably self-undermining.

Yet if intersectionality is a regulative ideal, then we do not commit to the co-constitution thesis, and there is no vicious regress. Pursuing the ideal will sometimes be fruitful and sometimes not, and, in any given case, we may not arrive at “intersectional results.” It’s always an empirical question (not one that follows from a general law a priori) where intersectional inquiry (on a given topic) will stop (see also Garry 2011; Hancock 2007a; 2007b; Haslanger 2014; May 2015; McCall 2005).

In another sense, however, the methodological view does entail a regress—but it is productive rather than vicious. In pursuing the intersectional ideal, we will continue to seek out further discoveries, to continually ask new “other questions.” But this is a good thing. We want our scientific methodologies as well as our modes of activist engagement to be generative, such that each answer generates a new set of questions.

Intersectionality compels us to seek out further practical and theoretical discoveries: unity across groups (intergroup similarities) as well as variety internal to groups (intragroup differences), to find out whose experiences are obscured by prevailing theoretical and activist schemes of representation, to seek out new grounds for solidarity and boomerang effects across crosscutting dimension of identity, and so on. But it does not imply that every particular category will be indefinitely diverse, or that, in literally every case, no meaningful and true generalizations can be made about a racial group, or a gendered racial group, without cross-referencing the claim with other classification schemes. Investigations into further intersections may not be useful or fruitful, and so may not change the results obtained from higher-level generalizations. In these cases, the social world may stop the regress on its own (see also Yuval-Davis 2006: 202–203).

To the extent that there is a methodological regress, then, we should embrace it. Just as Kant wrote that “cognition requires an advance to the always still remaining differences” (A656/B684), so, too, have intersectional scholars argued that the intersectional regress is productive. Kathy Davis (2008: 77) writes, “The infinite regress built into the concept [of intersectionality] – which categories to use and when to stop – makes it vague, yet also allows endless constellations of intersecting lines of difference to be explored.” Modeling it as a regulative ideal shows, we hope, that intersectionality is not so vague after all.