Gender Unrealism

1. Introduction

One of the most striking ideas in Catherine MacKinnon’s pivotal body of work from the late 1980s is that claims involving the concept woman express a masculine perspective because the concept was formed under patriarchy.¹ Given that claims involving the concept woman are parochial in this way and that objective reality is not, such claims do not describe objective reality according to MacKinnon. Rather, what is presented as objective reality is male reality instead—women are not objectively real, but rather real only relative to the male perspective.

MacKinnon’s ideas are influential in law and popular culture.² However, their influence in analytic philosophy, even in analytic feminism, is relatively muted. This contrast is partly owed to a prominent rejoinder from the influential analytic feminist, Sally Haslanger. On Haslanger’s (1995) reading, MacKinnon’s idea that claims about womanhood are real only relative to the male perspective—or, as Haslanger puts it, “unreal”—collapses “the epistemology-ontology distinction” (118). Rather, claims Haslanger, “what we believe to be real may be deeply conditioned by our point of view; but what is real is another matter” (114). Haslanger continues, stating the founding assumption of a now-dominant current in analytic feminism:

We must distance ourselves from the objectivist tendencies to limit our vision of what’s real, but we must be careful at the same time not simply to accept perspectivist limitations in their place. I would propose that the task before us is to construct alternative, modestly realist, ontologies that enable us to come to more adequate and just visions of what is, what might be, and what should be. (119)

In this concluding passage of her response, Haslanger articulates a vision of gender according to which its reality is not diminished by its emergence from oppressive social practices. Accordingly, feminist emancipation does not require us to resist the reality of gender, as MacKinnon seems to urge. Rather, it requires us to create novel social realities—and, in particular, gender realities—that permit us to discard oppressive ones and “ameliorate” gender-based social inequities. Haslanger’s positive vision of feminist theorizing is theoretically rich and politically significant.³ But, as the quotation above expresses, her ameliorative project is underpinned by a contestable assumption: that gender depends on, but is importantly distinct from, how we think and talk about it and that we can change the former by changing the latter.

We understand MacKinnon’s project differently than Haslanger. As we understand her, MacKinnon denies that there is anything more at all to gender than the way that we think and talk about it. On our interpretation, MacKinnon advances a limited form of anti-realism about gender, according to which thought and talk about being a woman, man or the like carries no ontological commitment about the nature of women, etc. This paper’s main aim is to develop MacKinnon’s anti-realist insight that someone has a gender just when and because they are correctly thought and talked about as such. Although this aim is strictly constructive, we also discuss some (perhaps fatal, perhaps superable) problems with Haslangerian, Haslanger-inspired and Haslanger-adjacent views to motivate our Mackinnonite alternative. Polemically, we use Haslanger’s own words to name this MacKinnon-inspired alternative Gender Unrealism.

We begin by highlighting deficiencies with well-known self-identification and social position views of gender (§§2–3). Reflecting on these deficiencies illuminates a particular interpretation of the insightful norm-relevance account of gender developed in a pair of recent papers (2016; 2018) from Katherine Jenkins (§4). Drawing on lessons that emerge from our discussion of the account, we advance the related but novel view that someone has a gender just when and because they are correctly thought and talked about as having it, and that such thought and talk is correct, very roughly, to the degree that its subject non-instrumentally desires to be thought of and talked about in that way (§5). In the final stretch of the paper, we highlight various ways in which the account can be extended while also drawing out some potential meta-metaphysical implications (§6). In particular, we observe that our view of the correctness conditions for gender talk and thought generalizes in a highly plausible way when combined with Kate Manne’s (2016; 2017) “Democratic Humean” treatment of desire-based reasons. Finally, we combine these ideas with an expressivist treatment of gender talk and thought into a comprehensive metaphysical, semantic, and ethical framework for thinking about gender, which reframes some of the recent history of analytic feminism and gender theory.

2. Gender as Self-ID

Since you’re reading this paper, it’s likely that you identify as a philosopher. It’s also likely that you identify as a member of the academy. Being an academic philosopher is part of who you are; it’s part of your identity. Moreover, your students, colleagues, family, friends, and so on see you as you identify, suggesting that your identity is not merely private or personal, but also to some extent public or social. Likewise, there’s a good chance that you see yourself as a woman, genderqueer, man, or the like, and that you may be similarly seen by those around you. So, gender also appears to be part of your social identity. These banalities suggest an interesting view of what it takes to have a gender, such as exhibiting the property of being a woman, genderqueer, man, or the like.⁴ The basic idea is that there’s nothing more to exemplifying womanhood than self-identifying as a woman. As a result, understanding when and why someone is a woman requires understanding how to self-identify as a woman.⁵ Call the view that explains having a gender with facts about how individuals self-identify,

Gender as Self-ID: Necessarily, someone is a gender iff and because they self-identify as that gender.⁶

To properly assess Gender as Self-ID, it must be distinguished from nearby views.⁷ For example, Gender as Self-ID is distinct from, though related closely to, the following:

Epistemic: Necessarily, you are in a position to know that someone has a gender if you know that they self-identify as that gender.⁸

We find this principle appealing, but we will not digress in defending it.

Likewise, just as facts about self-identification can have epistemic import, they can also bear on how we should treat each other as captured by,

Deontic: Necessarily, you should treat someone as a gender iff you know that they self-identify as that gender.⁹

We are also sympathetic to something like this deontic principle, proposing it without further support.

However, Gender as Self-ID is more metaphysically ambitious than Deontic or Epistemic. It aims to explain facts about gender using facts about the attitude of self-identification. Analyses like this, which rely on facts about attitudes in metaphysical analysis, invite familiar metaphysical concerns inspired by Plato’s Euthyphro. Famously, Euthyphro imagines that what’s pious and what’s loved by the gods are the same—that what’s pious is loved by the gods and what’s loved by the gods is pious. But, how, if at all, does one property relate to the other? Euthyphro eventually concedes to Socrates that the attitude of divine love does not make something pious or explain why it is pious. Rather, some other feature(s) of that thing does. Euthyphro highlights how some properties that are closely related to attitudes are nevertheless ill-suited to analysis by facts about those attitudes.

To be sure, not all properties resemble piety in this regard. For example, being popular is just a matter of being widely liked, so it’s well suited to analysis in terms of attitudes, unlike piety. The following diagrams illustrate this structural difference between these properties:

If Euthyphro’s argument is sound, whether something is pious depends ultimately on facts about what merits divine love, or the normative reasons (we’ll drop ‘normative’ from here on) for the gods to love it, not on whether they in fact love it (Figure 2). Conversely, whether something is popular depends ultimately only on the distribution of the appropriate attitudes (Figure 1). With this distinction in mind, reconsider Gender as Self-ID as represented below:

Figure 1

Figure 2

Figure 3

Because Gender as Self-ID fundamentally explains the instantiation of a property (gender) with the presence of an attitude (self-identification), it implies that gender is more like popularity than piety. This gives us pause. Consider a case where someone identifies as a gender merely instrumentally—a case in which such self-identification, as the prominent trans philosopher of gender Talia Mae Bettcher (2009: 110) puts it, “reflects political choices made for tactical reasons.”¹⁰ Gender as Self-ID appears to imply that identifying as (say) a woman merely instrumentally makes one a woman. Like Bettcher, we find this implication implausible. As a result, we deny that the bare attitude of self-identifying as a woman suffices for being a woman. As Bettcher points out, it seems that one’s self-identification must also be sincere to count. Crucially, the difference between sincere and insincere self-identification depends on the reasons for which one self-identifies. So, the ultimate analysis of someone’s gender, like of something’s piety, rests in the reasons behind an attitude and not in the attitude itself (Figure 4).¹¹

Figure 4

Thus, at best, Gender as Self-ID offers an intermediate explanation of why an individual has their gender. The ultimate explanation rests in the reasons that make their self-identification sincere.¹²

We anticipate several ways of resisting our Euthyphronic objection to Gender as Self-ID. Addressing resistance from two directions in particular helps to clarify our concerns about the view. First, one could accuse the objection of mischaracterizing self-identification. That objection presupposes that self-identification is a genus of which sincere and insincere self-identification are species. However, self-identification can be understood differently. A distinct but closely related ‘disjunctivist’ understanding distinguishes between genuine self-identification and ersatz self-identification, rather than distinguishing between sincere and insincere self-identification. On this approach, roughly, the condition that underlies being a woman is not a conjunction involving self-identification and sincerity; it is self-identification simpliciter. Those who claim a gender for tactical reasons do not challenge the sufficiency of self-identification for being a gender, on this approach. For just as faux fur is not fur, insincere self-identification is not self-identification. It is ersatz, defective, or failed self-identification, much as some characterize hallucination as ersatz—not genuine—perception. On this disjunctive interpretation, self-identifying as a gender just is what it is to have a gender. Hence, it’s both necessary and sufficient for having a gender.

We concede that this distinction exempts theorists from explaining what makes some instances of self-identification sincere. However, it does not relieve us of accounting for the new distinction between genuine and ersatz self-identification. It’s plain that this distinction also depends on a distinction in the reasons behind each act or attitude, inviting the Euthyphronic worries stated above. Consequently, disjunctivism about self-identification offers no real advantage.

A second way of rebutting our objection to Gender as Self-ID begins with an idea from Bettcher (2009)—that any full characterization of gender must account for the “existential” nature of gendered self-identification, or what we might describe more broadly as the sense in which identifying as a gender reflects our agency and autonomy.¹³ According to this rebuttal, grounding one’s gender in one’s reasons for self-identification rather than in one’s attitude of self-identification incorrectly diminishes one’s authority over one’s gender. After all, our reasons are not “up to us” in the way that some acts and attitudes are. Since our gender is “up to us” and our reasons are not, gender is not grounded in reasons.

We have several responses. First, this view’s appeal is best understood in terms of one’s reasons for self-identification. Imagine a case in which Dwayne “The Rock” Johnson self-identifies differently from how he actually does (i.e., as a woman, rather than a very manly man). Because The Rock’s gender is up to him, one might accept the following counterfactual:

If The Rock were to identify differently, holding all else equal, The Rock’s gender would correspond to his counterfactual, not actual, identification.

The counterfactual seems true. But it doesn’t mention The Rock’s counterfactual reasons for identifying differently. So one might conclude that all there is to the nature of someone’s gender is self-identification. Reasons don’t matter.

But we think that the preceding counterfactual is easily confused with a similar one:

If The Rock were to identify differently, The Rock’s gender would correspond to his counterfactual, not actual, identification.

The difference between the two counterfactuals turns on the implications of holding all else equal. The second counterfactual implies a claim we endorse: in the scenario most like our own where The Rock identifies as a woman, The Rock is a woman. We think this is true. After all, were The Rock to self-identify as a woman, he’d have good reason for doing so. But The Rock’s self-identifying as a woman and The Rock’s having reason to self-identify as a woman are logically distinct; the latter does not entail the former.

Second, because we embrace Deontic and Epistemic, we hold that self-identification could play an important role in governing what people do and believe, which ratifies much of which this objection seeks to preserve. We agree with Bettcher that when you know how someone self-identifies, you know their gender and you should respond appropriately, e.g., with the relevant pronouns, etc. We deny only that self-identification plays a foundational metaphysical role.

Third, and more importantly, the possibility of insincere self-identification demonstrates that having a gender is not entirely up to us. Only certain exercises of self-identification affect one’s gender, so one’s gender cannot be entirely grounded in an existential choice. In sum, we simply don’t have groundless and unlimited authority or agency over having our own gender.

Finally, consider the possibility of groundless or reasonless self-identification—that is, of self-identifying for no reason at all. It’s hard to imagine the possibility initially, but one way of doing so is to think of self-identification as spontaneous, like a reflex, or sub-personal, like sleepwalking. Neither reflexes nor sleepwalking are expressions of a person’s agency. But self-identification, insofar as we understand the notion (more on this in §5), is deeply embedded in agency. So, we suggest, reasonless self-identification fails to secure a role for agency in gender. Indeed, it seems to fare much worse than grounded or reason-based self-identification, at least on the highly natural view that our agency is reflected in the reasons for which we act, including when we self-identify.¹⁴ While not exhaustive, these arguments push us to look beyond self-identification for an analysis.

3. Gender, Sex, Social Position, and Beyond

The previous subsection’s Euthyphronic objection shows that the bare attitude of self-identification cannot fully explain gender. For example, whether Dolly Parton’s self-identification explains why Dolly is a woman depends on Dolly’s reasons for identifying as a woman. This section asks what those reasons are and how they figure in self-identification.

Some doubt that reasons for self-identification explain gender because they doubt the connection between self-identification and gender. Instead, taking inspiration from mainstream linguistic practice with words like ‘woman,’ ‘man,’ etc., some suggest that it’s facts about human sexual reproductive anatomy that explain gender.¹⁵ Even Bettcher (2009: 105) claims, in commenting on what it is to be a woman, “On the face of it, the definition ‘female, adult, human being’ really does seem right. Indeed, it seems as perfect a definition [of ‘woman’] as one might have ever wanted.”¹⁶ Correspondingly, it might be said that what it is to be a man is to be an adult male human. Call the view that one is a gender just when and because one has a certain configuration of human sexual reproductive features, Gender as Sex.¹⁷

Gender as Sex is a natural view. After all, sex seems to at least partly constitute the very idea of gender—it’s what distinguishes the subject matter, at least partially.¹⁸ But it’s also a view that is vulnerable to a wide range of counterexamples.¹⁹ Moreover, we think the view fails to treat various forms of evidence sufficiently seriously. Consider the following testimony:

I loathe being talked about as though I’m a woman. But I don’t really think I’m a man. I’m a person. For fifty years I’ve gazed into the void of me and tried to work out my gender, and nothing looks back. I still can’t find it… And that’s how I decided I was agender. I went looking for my gender, and I kept not finding anything. Surprise surprise, there’s actually a word for that, and I am very relieved to be able to finally stop, and say “Stuff it, I’m not a man or a woman, I’m just me.” Other people can do gender from now on. I don’t get it at all. (emphasis added)²⁰

Taking Marnie’s testimony seriously implies that there are genders that are neither the gender woman nor the gender man—for example, genders like agender or non-binary. It seems that the only way to make sense of non-binary genders on Gender as Sex is to posit more sexes beyond the binary of adult female human or adult male human. But while the existence of the sex intersex leads us to agree that there are more sexes than the sex male and the sex female, testimony from non-intersex and non-binary gender folk like Marnie suggests that proliferating sexes does not account for all genders. More simply, it seems to us that since there are actually more genders than sexes, genders are not sexes.

Like our objections to Gender as Self-ID, these concerns about Gender as Sex are not decisive. We view them instead as licensing a search for an alternative view that doesn’t face these prima facie problems. So, in light of Gender as Sex’s apparent shortcomings, we are tempted to explain gender in a different way—one influenced by many contemporary feminist theorists. These theorists deny that gender depends on reproductive anatomy. Rather, genders are the product of the relationship between presumed human reproductive anatomy and societal organization. Very roughly, genders result from how individuals with various sexual characteristics are conventionally treated in society. Different proponents of these Gender as Social Position views express their views differently, depending on how they theorize the relationship between sex and social treatment.²¹ For example, Haslanger (2000) influentially argues that the relationship between sex and social treatment is one of social hierarchy—what it is to be a woman is, approximately, to be subordinated in society on the basis of one’s presumed female sexual features and what it is to be a man is to be superordinated in society on the basis of one’s presumed male sexual features.

Gender as Social Position is appealing for several reasons. First, while it falls short of identifying sex with gender, Gender as Social Position nevertheless apprehends gender through the lens of sex. As a result, it captures much of Gender as Sex’s superficial appeal. Second, Gender as Social Position is consistent with the testimony of people like Marnie, which reveals that, in addition to the genders man and woman, there are non-binary genders like agender. As we saw above, views in the tradition of Gender as Sex struggle to accommodate non-binary gender, seemingly (and falsely) implying that all agender people are intersex. In contrast, theorists who accept Gender as Social Position can simply claim that to be agender just is to be neither subordinated nor privileged on the basis of one’s presumed sexual features.²² While much more needs to be said, it’s at least clear that, unlike Gender as Sex, Gender as Social Position has the resources to begin theorizing about genders beyond those conventionally associated with sex.

Moreover, the analysis stresses the resemblance between social kinds like class, race, and gender in an effort to illuminate new modes of resistance to oppression. Basing our understanding of gender on such resemblances leads to the idea that just as money or the bourgeoisie came into being when cowrie shells were treated as tokens of fixed value or when the economy created capital-owners, respectively, women, men, etc. came into being when it became conventional to treat individuals grouped on the basis of perceived or presumed sexual characteristics in certain ways. So the analysis can satisfyingly explain the contingency of gender.

However, while we are tempted by Gender as Social Position, we think there’s good reason to resist the temptation. Many have observed, as Mari Mikkola (2011: 71) does, that women like the Queen of England aren’t subordinated but are no less women for it. Mikkola’s observation may not be fatal to Gender as Social Position. Theorists can respond that gender corresponds to particular axes of subordination or superordination, which exclude the Queen’s case. The challenge for developing this response is identifying the axes that matter without appeal to claims about gender, which would circularly undermine the analysis. Or theorists can simply bite the bullet, as Haslanger does.²³

But the point is more general. This is shown by a similar argument that there are trans women who are not subordinated in the relevant way but who are also no less women for it.²⁴ More broadly, the problem with social position views generally seems to be that they fail to capture the psychological or more broadly “internal” element that seems so closely interwoven with belonging to a gender and that motivates Gender as Self-ID.²⁵

Fortunately, there is another, independently motivated way of thinking about the relationship between Gender as Self-ID and Gender as Social Position. Consider dancers, for example. What is it to be a dancer? It is to be someone who dances. We can also ask: How does someone become a dancer? Likewise, it seems, by being someone who dances. A single consideration answers both questions about dancers.

However, answers to these two kinds of questions come apart for other social categories. Imagine a freemason-like secret-society called The Stonecutters. The Stonecutters are known in their community for some prominent volunteer work, for wearing a peculiar uniform when they publicly appear as Stonecutters, and for secretive meetings in their centrally-located hall. If a prospective member were to ask, “What is it to be a Stonecutter?”, both members of the group and of the broader community would respond by listing these qualities.

However, these qualities do not answer a different question: How does someone become a Stonecutter? Like other secret societies, it’s up to the Stonecutters to decide what it takes to join their ranks. Suppose that they’ve settled on a secret handshake: anyone who knows the secret handshake is, ipso facto, a Stonecutter. Of course, the group teaches candidates the handshake only after a gauntlet of grueling tasks; but if a candidate is wily enough to learn the handshake without completing the tasks, the Stonecutters reward the candidate’s shrewdness with membership. Plainly, then, facts about learning the secret handshake answer questions about the membership conditions for being a stonecutter.

These answers are logically independent. It is neither sufficient nor necessary for being a Stonecutter that one wears a certain uniform while volunteering and that one attends the meetings. After all, you can steal the uniform and sneak into the meetings. Nor is performing volunteer work necessary for being a Stonecutter. The Stonecutters do not expel those too ill to volunteer.

So it seems to us that these two questions carve at subtly distinct metaphysical joints.²⁶ One asks about the nature of a certain category and the other asks about membership in that category. In particular, we distinguish:

Fact about Gender Category Membership Facts about what it takes to be a given gender, e.g., the features in virtue of which someone is a woman rather than a man or some other gender,

From,

Facts about Gender Categories Facts about the nature of a given gender, e.g., about how womanhood differs from other gender categories like agenderhood and manhood.²⁷

Because this distinction is easily elided—questions about gender are often ambiguous between these two questions—it should not surprise us that gender has proved resistant to analysis. Indeed, it’s natural to think that these two questions are so intermingled that little of philosophical value comes from distinguishing them. And this is true for many kinds, such as dancer—we gain nothing by distinguishing the two questions with respect to dancer.

But we think that it’s relatively straightforward that other social kinds can be properly understood only by distinguishing these questions. The question of how one becomes a Stonecutter differs from what it is to be a Stonecutter. This difference tracks the difference between facts about who has learned the handshake and the causal conditions/functional role/network analysis/etc. that circumscribes the nature of Stonecutterhood.

Moreover, conflating these two questions leads to absurdities. For example, from the fact that all and only Stonecutters know the secret handshake, we might be wrongly tempted to infer that knowing the secret handshake is somehow part of the nature of Stonecutterhood. But it’s generally bad to infer that some property is essential to a kind from the fact that all and only members of a kind share that property. For example, all and only Stonecutters—let’s suppose—are prone to a particular kind of hand cramp that comes from overusing the secret handshake. Being prone to that cramp is not part of the nature of Stonecutterhood. And since the secret handshake is—well—secret, were the handshake part of the nature of Stonecutterhood, then that nature would seem to be secret too. Non-members would have only a poor grasp of the concept Stonecutter. But that’s implausible.

Furthermore, if we conflate the two questions and include facts about the secret handshake in the nature of Stonecutterhood, then the social kind becomes implausibly fragile, too easily destroyed by small changes to the kind’s membership conditions that are irrelevant to its nature. For example, suppose that students at the University of East Dakota (UED) are known colloquially as ‘Easties.’ Part of the nature of being an Eastie is attending UED and anyone who’s graduated from UED since its inception in the 19th century is, at least intuitively, an Eastie. But UED has changed its admissions procedures over the years. Alongside Harvard, UED began using the SAT for admissions decisions in 1934 under the belief that it tracked academic merit. But UED recently discontinued use of the SAT after learning that the SAT unfairly advantaged students who could afford specialized SAT tutors. So UED’s admission procedures have changed over time; that is, the membership conditions of being an Eastie have changed over time. But, we submit, the nature of being an Eastie has remained unchanged. After all, to be an Eastie is to attend UED.

Conflating our two questions—conflating what it is to be an Eastie with what it is to become an Eastie—implies that there’s one set of graduates (either the pre-SAT cohorts or the post-SAT cohorts) that are Easties and that the other cohort is something else.²⁸ We find this implausible. So we conclude that these questions must be distinguished.

Other social ontologists, most prominently Brian Epstein in his excellent The Ant Trap (2015), also examine social phenomena through a two-dimensional lens, vaguely similar to our own. For example, Epstein distinguishes the question of What are the conditions for a person to be in the kind Stonecutter? from the question of What makes those the conditions for a person to be in the kind Stonecutter?²⁹ In his terminology, the first is a question about grounding; the second is a question about anchoring.

Epstein’s first question is very much like our first question. However, Epstein approaches this question in starkly realist terms, such as grounding. Grounding is a relation that, at least as it is standardly understood, obtains between facts. Facts are paradigmatic instances of the kind of metaphysical bloat that anti-realists typically decry. So Epstein’s metaphysical project stands opposed to the “gender unrealism” that interests us, since, as will become clear, we deny that understanding gender requires (irreducible use of) realist metaphysical tools such as grounding. We concede that the gender unrealist’s paths are largely untrodden (or at least overgrown), so her tools are a bit rougher.³⁰ But they can still mount a workable account of gender.

Epstein’s second question, concerning anchoring, is orthogonal to our first and second questions. Claims about anchoring purport to explain social kinds by explaining how, very roughly, the conditions underlying a particular social kind are themselves explained by social practice. For example, the fact that porchetta is pork grounds the fact that it’s not Kosher, a fact which is itself anchored in the social and religious practices of the Jewish people. Consequently, claims about anchoring might explain why certain facts about the Stonecutters’ distinctive characteristics (volunteering, peculiar uniforms, secretive meetings) answer our first question. They might also explain why, for example, the Stonecutters settled on the particular secret handshake that they did. Consequently, despite our common interest in two-dimensional approaches to social phenomena, Epstein is ultimately interested in different questions than we are; and his realist assumptions circumscribe his approach in ways that we hope to avoid.

Distinguishing these two questions nuances our understanding of gender. Because what it takes to become a member of a kind can differ from the qualities that characterize that kind, an individual can belong to a kind while lacking its characteristic qualities. In particular, in the sense at issue in the first question, a woman is, inter alia, someone addressed by female personal pronouns. That’s undeniably part of what distinguishes womanhood, as a kind or property, from manhood or other gender categories. But being someone addressed by female pronouns is no part of what makes someone a woman because there are women who are not addressed by female personal pronouns—for example, trans women in oppressive societies. While association with a particular set of pronouns helps distinguish the kind woman from the kind man, it does not help us distinguish whether someone is a woman or not.

Distinguishing these questions suggests that views neighboring Gender as Social Position answer questions about the nature of particular genders, but they don’t answer questions about how one comes to have a gender. Likewise, the distinction places us in a position to appreciate that Gender as Self-ID offers purchase on the question of how one comes to have a gender but not on the nature of particular genders. Transposing these views onto the two questions yields the following account:

A gender’s nature supervenes on facts about how society is arranged but membership in those categories does not; it depends on the reasons for which individuals self-identify.³¹ Framed in this manner, Gender as Social Position and Gender as Self-ID offer complementary insights concerning the nature of gender. This is worth stressing in part because many recent discussions of gender appear to assume that the views are in tension.³² The next section explores a recent view that appears to be an exception.

Figure 5

4. Gender and Norm-Relevance

Previously, we distinguished two questions that a theory of gender ought to answer—(1) What’s the nature of man, woman, etc., and the gender categories themselves?—and (2) What are the conditions under which one becomes a member of these categories? We argued that several natural views of gender are best thought of as answering one but not both questions. Consequently, they are best understood as offering partial accounts of gender. In this section, we examine an approach based on Jenkins’s (2016; 2018) recent and influential discussion of gender, which looks like it can answer our concerns with other views.

The Norm Relevance Account draws on the plausible thought that, for example, part of being a woman is feeling bound by certain norms, perhaps in a manner that recalls Gender as Self-ID, and that those norms emerge from precisely the kind of social arrangements that figure in Gender as Social Position. The Norm Relevance Account introduces the metaphor of a map to express the web of norms experienced by members of a given gender.³³ According to the view,

S has a gender identity of X iff S’s internal ‘map’ is formed to guide someone classed as a member of X gender through the social or material realities that are, in that context, characteristic of Xs as a class. (Jenkins 2016: 410)

Consider the following illustration. A gender identity is like a standard-issue map in the sense that “entering a space that is designated on your ‘map’ as a ‘no-go’ area for people like you might result in feelings of bodily awkwardness, tension, and even the physiological responses associated with fear” (2018: 729). These responses constitute the experience of “a norm being relevant to oneself” (730). Given that certain norms are gender norms, the relevant collection of these experiences will constitute the relevant gender map.

Norm Relevance Account’s gender maps appear to mediate the “internalist” insights, so to speak, of Gender as Self-ID and the “externalist” insights of Gender as Class Membership. Representing one’s space as governed by certain gender norms rather than others through the experiences described above appears to suffice for being a gender—that’s the internalist-y Self-ID bit. But exactly which norms count as gender norms is strictly a matter of how society is arranged—that’s the externalist-y Gender as Class Membership bit. Furthermore, gender maps provide the glue linking the two questions that we distinguish above, explaining the connection between what’s distinctive of womanhood and what it takes to be a woman. Social norms constitute the gender kind, woman. Possessing a map that navigates these norms makes one a member of the gender kind, woman. There’s a lot to like about the Norm Relevance Account.³⁴

However, the Norm Relevance Account is contested, despite its considerable promise. For example, Barnes (2019) worries that gender’s underlying norms are complex. Being guided by a gender map might then be equally complex, thereby requiring a high level of cognitive sophistication. As a result, Jenkins’s account implies a demanding cognitive threshold for gender membership. But, observes Barnes, having a gender doesn’t obviously require cognitive sophistication.³⁵ The account thereby appears to exclude the cognitively disabled.

Figure 6

A natural way for advocates to reply is to (re)emphasize that a gender map guides an individual in response to social norms when they “experience” such norms as “relevant” to them.³⁶ We might expand on these remarks by claiming that what it is to possess a map that guides an individual’s responses to social norms is for them to be disposed to feel, in a straightforward phenomenological sense, bound by those norms. Because feeling bound plausibly doesn’t require much in the way of cognitive sophistication, advocates might be able to rebut Barnes’ objection.

This response, however, foregrounds a second objection advanced separately by Bettcher (2017) and Andler (2017: 891–892). Imagine that Sam is a trans individual in the process of transitioning. If Sam has not yet come out as a woman (or, indeed, even after having come out), she may still feel bound by the expectations of friends and colleagues to act as a man. Worse still, consider societies that are indifferent or hostile to the idea of transitioning. Many trans individuals in repressive and regressive societies may feel bound more by the norms associated with their perceived gender rather than by those associated with their actual gender. Jenkins’ view suggests that such individuals possess multiple maps and hence have more genders than they plausibly do.

In response to Bettcher and Andler, Jenkins (2018: 733) emphasizes that the Norm Relevance Account lacks the epistemic implication that everyone knows their own gender, but has the ethical (c.f., Deontic, above) implication that it is bad not to respect an individual’s gender avowals. But neither claim responds to the metaphysical concern implicit in Bettcher’s and Andler’s objection: on a highly natural interpretation, the Norm Relevance Account implies that trans individuals have more genders than they in fact do.

Whether the account is doomed by these problems turns on its explicit details. But we haven’t presented the account in explicit detail, relying instead, as do its foremost advocates, on the evocative metaphor of maps together with talk of “experiencing norms as relevant.” This is not a flaw. Metaphors often convey ideas more economically and accessibly than drab prose; they allow us, in good Wittgensteinian fashion, to show what we cannot say. But challenges like those described above from Barnes, Bettcher, and Andler require us to clarify the metaphor and to articulate its details explicitly.

We think it’s unpromising to explain gender using the metaphor of a map, regardless of how we unpack that metaphor, because it gets the direction of explanation wrong. We’ve been appealing to the idea of a direction of explanation all along. The Euthyphro objection is supposed to show that facts about what’s loved by the gods cannot (ultimately) explain what’s pious. Although it may be true that something is pious iff it’s loved by the gods, it is false that something is pious because it is loved by the gods—at least if the Euthyphro objection is sound. Rather, it’s far more plausible that something is loved by the gods because it is pious. So analyzing piety with divine love seems to get the “direction of explanation” wrong. We think that the Norm Relevance Account makes a similar mistake.

Let’s understand ‘map’ perhaps more literally than its advocates intend to highlight the trouble. Suppose that you’re vacationing to a series of wonderfully tranquil but formidable desert landscapes. You have Quine Travel’s famed collection of desert landscape maps in your pocket to avoid getting lost, which includes, e.g., a Sahara map, a Gobi map, etc. Suppose further that you’re in the Sahara. As a result, the Sahara map rather than the Gobi map is the right map for you; that’s the map to guide you around your material reality. It would be quite odd to hold the inverse view, viz., you’re at such-and-such location in the Sahara because the Sahara map is the right one for you. That gets the direction of explanation wrong. The opposite view is clearly true: the fact that you trekked for three days from the city of Marrakech to the desert explains why one map rather than another should guide you. More generally, your map is right because you’re at such-and-such position; it’s not true that you’re at such-and-such position because your map is right.

Now, recall the more formal Norm Relevance analysis provided by Jenkins:

S has a gender identity of X iff S’s internal ‘map’ is formed to guide someone classed as a member of X gender through the social or material realities that are, in that context, characteristic of Xs as a class. (2016: 410)

Because this is advanced as an analysis,³⁷ it can be made fully explicit:

S has a gender identity of X iff [and because] S’s internal ‘map’ is formed to guide some-one classed as a member of X gender through the social or material realities that are, in that context, characteristic of Xs as a class.

But just as it’s quite odd to say that you’re in the Sahara because the Sahara map is the right map for you, it’s also odd to say that you are such-and-such gender because such-and-such map is the right one for you. It’s the other way around; the fact that someone has their particular gender explains why their map is the right one to guide them around the norms constituting that gender.

To be clear, this objection to the Norm Relevance Account is quite broad. It is not directed at the particulars of any philosophers’ account. It may be possible to develop the Norm Relevance Account that soften or evade the objection. So we do not mean to suggest that Jenkins, or anyone else in particular, must respond to our concerns about the Norm Relevance Account’s direction of explanation.

Nevertheless, for views like the Norm Relevance Account of gender to work—views which, we’ve been suggesting, are best positioned to address the challenges that we’ve been developing—we need an explanation of the gender-endowing link between individuals and the norms constitutive of various genders like those that inscribe gender maps. In the metaphor of the Norm Relevance Account, we need an account of how one comes to “possess” a gender map, of what makes a map the right one for an individual. That’s the task of the next section.

5. Gender Membership in the Key of Desire

In the previous section, we observed that the Norm Relevance Account has the right structure to answer our two questions, one about the nature of gender categories and another about the membership conditions for belonging to them. While we’re confident that the view can answer questions about the nature of gender because of its social position analysis of gender categories, we argued that it doesn’t properly answer the question of gender membership. While gender maps or norms may constitute a gender kind, gender-conferring power—the conditions of gender membership—lie not in norms, not in maps, but in us. The Norm Relevance does not properly account for this distinction for it does not explain why one gender map rather than another counts as the right map to ground one’s gender. Further, the idea of having a quality, such as being at a location or being a certain gender, because one’s map is the right one gets the direction of explanation between the map and the quality wrong. As a result, gender maps aren’t the right sort of fundamental “internal” analysans for illuminating gender category membership.

In this section, we defend an account of gender that builds on the Norm Relevance Account. Like that account, it also incorporates a social position analysis of the nature of gender categories. However, this view departs from the Norm Relevance Account by advancing an alternative account of gender category membership and by jettisoning the metaphor of gender maps.

Gender classification—gendering behaviour of which pronoun use is paradigmatic—is informed by knowledge of the norms characteristic of various genders. We’ll represent each set of norms as a set of imperatives, e.g., {“respond to these pronouns!”, “use bathrooms marked this way!”, “walk this way!”, “talk this way!”}, etc.³⁸ This way of understanding gender norms foregrounds a way that agents relate to commands, namely, by accepting them. Importantly, accepting a command requires something different from other forms of acceptance, such as accepting an assertion. If you assert that the window is open, accepting that assertion requires believing it is open. But accepting your command to close the window requires something different; it requires intending to close the window, or planning to close it, or, in some very broad sense, desiring to close it. It’s possible to believe that you are instructed to close the window but not see any reason to close it or feel any impulse to close it. In that circumstance, you do not accept the instruction. Consequently, accepting a command or instruction differs from believing that you are so commanded or instructed. Thus, accepting a set of gender norms, construed as a set of imperatives, is different from representing space in a certain way with (say) a map.

On our stipulative usage, accepting those norms requires, broadly speaking, intrinsically or non-instrumentally desiring to conform to them.³⁹ This account of gender norm acceptance leads us to favour a view according to which, roughly, someone is correctly classed as being a gender just when and because they intrinsically desire to conform to the norms that constitute the corresponding gender category. More formally (but still roughly),

A Desire-based Account of Gender Category Membership: Necessarily, S has gender G at world w to the degree that and because S intrinsically desires to conform to the set of norms of gender classification N that constitute the gender kind G at w.

This account easily explains complex cases like Sam’s. We saw that recently-transitioned Sam can represent her space’s norms using the gender maps associated with both men and women—she experiences both sets of norms as relevant. However, while she understands the norms represented by the male map, she does not intrinsically desire to conform to them. So, she is not a man. Indeed, though Sam may live as a man, conforming to those norms if the social costs of living as a woman are sufficiently high, she does not intrinsically desire to live as a man. She only instrumentally desires to live as a man, given the high costs of living otherwise. So, she is not a man, though she may live as one. Likewise, Bettcher’s individuals who self-identify insincerely, for “tactical” reasons, do not accept the norms associated with their insincerely professed gender. Their desires to satisfy those norms are only instrumental because they are grounded in tactical reasons and they are, therefore, irrelevant to the individual’s gender.

Basing gender on intrinsic desires is also appealing because it allows for variation along two axes: individuals can differ not only in the proportion of a gender’s norms that they accept but also in the strength or intensity of the intrinsic desires with which they accept them. This is attractive because it accommodates the thought that gender comes in degrees or consists of a spectrum without entailing it, for the account allows for sharp boundaries or cut-offs along each axis of variation.⁴⁰ Second, the account’s two axes of variation explain various attitudes one can bear towards one’s own gender identity. For example, some are ambivalent about their gender, accepting norms from different genders at different times. Others are indifferent to their gender, accepting their gender’s norms only weakly. These simple explanations of complex lived relationships with one’s gender follow straightforwardly from the account. That is evidence for its truth.

Like Jenkins (2018: 740), we demur on specifying exactly which norms belong to any particular gender’s set. This is at least partly an empirical question perhaps better answered by social scientists. Further, we also demur on the question of why a particular set of norms, rather than some other set, constitutes a particular gender kind, say, the kind woman. However, we have positive philosophical reasons for doing so in this latter case. On a natural elaboration of the norm-relevance account, a gender kind asymmetrically metaphysically depends on its norms by being constituted by them, in the way that being a triangle is constituted by having a certain number of sides.⁴¹ From this point of view, asking why triangularity is constituted by that number of sides, rather than some other number, is confused. All there is to triangularity, we might answer, is being three-sided—there is no further answer. Likewise, on our understanding, the question of why, for example, the category woman is constituted by that set of norms, rather than some other set, is confused.⁴² All there is to the gender category woman, we answer, is to be the gender category of which those norms are characteristic, indeed, constitutive.⁴³

Our desire-based answer to the gender category membership question resembles Gender as Self-ID.⁴⁴ Figure 7 makes this clear. In particular, both our account and Gender as Self-ID explain one’s gender through one’s attitudes. The view of gender category membership that we propose above analyzes membership in terms of intrinsically desiring to conform to the norms that constitute the corresponding gender kind. But we argue that grounding gender in attitudes like self-identification is vulnerable to the Euthyphro objection. So, it’s natural to wonder whether intrinsic desires provide the most fundamental account of gender category membership, or whether such desires are grounded in something more fundamental.

Figure 7

There are several reasons to think that our view is less susceptible to Euthyphronic difficulties. First, we distinguish two distinct questions arising from distinct readings of ‘what is it to be a woman?’. One question concerns the nature of being a woman or of womanhood. Our answer to this question does not depend on claims about attitudes. As a result, Euthyphronic worries do not apply to a substantial portion of the analysis.

In contrast, Gender as Self-ID theorists typically do not distinguish these two questions. As a result, they are vulnerable to Euthyphro-style worries because their view suggests that the nature of a gender category is given by self-identification with that gender category and that self-identification suffices for membership in a gender category.⁴⁵ The circularity arises because Self-ID theorists must explain what all instances of self-identifying as a woman have in common without appealing to the concept woman. This forces Self-ID theorists to look elsewhere for the commonality. Reasons for self-identifying as a woman are a natural place to look. In contrast, because we offer an account of the nature of womanhood based in norms, we can rely on that concept to explain what all instances of desiring to follow a certain set of social norms have in common. As a result, our account does not face the same pressure to appeal to reasons.

Second, philosophers, most notably British sentimentalists and their followers, have found that desires are particularly suited to grounding normative categories. Doing so is appealing partly because there is already a clear sense in which some desires are more fundamental than others. Since one has instrumental desires for the sake of one’s non-instrumental or intrinsic desires, appealing to intrinsic desires starts theorizing about individuals’ motives at a fundamental—arguably the most fundamental—level.

In contrast, there is relatively little discussion about the metaphysical nature of the attitude of self-identification. If it is something belief-like, a natural elaboration of Gender as Self-ID holds that a person is a member of a gender category just when and because they believe that the set of norms constitutive of that gender category apply to them. It is, in other words, to have a kind of cognitive or representational attitude with a so-called “mind-to-world” direction-of-fit. But now imagine a trans woman, Bo, who was born with a penis, XY chromosomes, and other sexual features that are typical of male humans. For as long as Bo can remember, she has wanted to conform to the norms constitutive of the gender kind woman, though she would not put things this way—she simply claims to feel like a woman. However, Bo belongs to an extremely conservative household. Consequently, Bo has internalized the belief that only the various norms constitutive of the gender kind man apply to her. Indeed, the conflict between her desires and her beliefs about these norms is a considerable source of misery for Bo and for many trans individuals like her. Holding that Self-ID is something belief-like seems to imply that Bo is a man, even though she’s always wanted to conform to the gender kind woman. That strikes us as an objectionable prediction that generalizes across analyses of ‘self-identification’ in terms of cognitive attitudes.

The first point and the second point combine to form a third. The first point is that because our account of the nature of womanhood is distinct from our account of membership in the gender category woman, we can appeal to the nature of womanhood to explain what unites all members of the gender category woman, namely, sharing certain intrinsic desires. The second point is that the significance of one’s beliefs to one’s social identity is sometimes a matter of how they are acquired. In contrast, the provenance of one’s intrinsic desires matters less to issues concerning one’s social identity. It doesn’t matter which song made you like the Beatles, if you like The Beatles because of that song, you’re a fan of The Beatles. Likewise, it doesn’t matter why one intrinsically desires to conform to the norms that constitute a certain gender category, all the matters for one’s membership in that category is that one possesses those desires. Consequently—and this is the third point—it is possible that people desire to conform to those norms for varied and non-overlapping reasons, corresponding to people’s varied and non-overlapping personal histories. There is no guarantee that those reasons will resemble each other in theoretically useful ways. Rather, those reasons are significant only because they give rise to intrinsic desires. These desires form the most basic common unit of analysis for membership in a gender category. That is why we have chosen to base our analysis of gender membership in intrinsic desires.

To illustrate, it doesn’t matter if Dwayne “The Rock” Johnson suddenly intrinsically desires to conform to woman-category norms because (say) he was hit by gamma rays. We embrace the implication that those gamma rays have made him a member of the category woman by making him intrinsically desire to conform to the norms that constitute the category. By contrast, if gamma rays cause Dwight to believe that he is a woman or to believe that woman-category norms apply to him, he is not a woman; he has simply been confused by gamma rays.⁴⁶

As a result, we explicitly embrace the idea that analysis of an individual’s membership in a gender category begins in their intrinsic desires, not in the scattered reasons for which particular individuals have those desires. This exposes our account to a second challenge. While we explicitly decline to speculate about which norms constitute a given gender category, Haslanger’s (2000) discussion of gender-based oppression deserves special focus—even if we’ve already signaled our skepticism about it. According to her, as discussed above, womanhood is closely associated with norms of subjugation on the basis of perceived sex (39). If these norms constitute the category woman, does our account, to put it bluntly, necessitate that women desire their own subjugation?

If the desires that underlie one’s gender membership are understood as taking de re scope, as we suggest in footnote 39, then even if some norms subjugate people like you, desiring to conform to those norms doesn’t imply desiring to be subjugated. After all, if all wage labor is exploitation under capitalism, then you are subjugated by the norms that characterize your employment. But it seems false to say that by desiring to do your job, you desire capitalist subjugation.⁴⁷

Moreover, you can want to do something without wanting to want to do it. So one can desire in conformity with a subjugating norm while rejecting that desire from a higher-order desire to want to resist one’s subjugation. As a result, our account is consistent with women resisting their oppression.

Feminists have long noted that women’s socialization often involves inculcating self-abnegating desires or, as Khader (2011) puts it, women participating in their own deprivation. If so, then it is a feature, not a bug, of our account that it implies that women possess subjugating desires since it is wholly unsurprising (a) that it benefits the patriarchy for women to become their own jailors, and (b) that shaping young women’s desires through socialization is the most direct means to that aim. Theorizing about gender through desire thus explains how socialization imparts gender partly by incentivizing the adoption of certain gendered preferences. Our account thus promises to illuminate not just what it is to become a woman, but how we become women.

6. Higher-Order Reflections on Gender

Let’s take stock. In the first half of this paper, we observed that questions about the metaphysics of gender are subtly ambiguous, sometimes expressing questions about what makes someone a member of a gender category and at other times expressing questions about the nature of the categories themselves. Observing this ambiguity recasts several prominent views about gender. We argued that views like Gender as Social Position offer promising answers to questions about the nature of gender categories but lousy answers to questions about gender category membership, and vice versa for views like Gender as Self-ID. From this perspective, we argued that the best approach to the metaphysics of gender reflects the ambiguities that we identify. On views such as the Norm Relevance Account, gender categories receive a kind of Gender as Social Position analysis, while gender category membership receives a style of analysis broadly akin to Gender as Self-ID. Being a member of a gender, according to an influential formulation of the Norm Relevance Account, involves possessing a kind of ‘map’ that navigates one around the norms that constitute the corresponding gender category or categories. We also argued, however, that the Norm Relevance Account wrongly inverts the order of explanation of two facts. A map of the Sahara is a guide to one’s material reality because one is in the Sahara; one isn’t in the Sahara because one’s map is correct. However, the Norm Relevance Account suggests that one occupies a particular location in the space of genders because one’s gender map is a good guide to one’s social and material reality. But this left untouched the question of why that map, rather than another, is the right one. This led us to embrace an analysis of gender category membership in terms of intrinsic desires on which being a member of a gender category involves intrinsically desiring to conform to the norms that constitute the corresponding gender kind.

However, any analysis based in paradigmatically non-cognitive mental states like desire invites familiar “meta” questions about its metaphysical commitments. In particular, it’s natural to wonder whether desire-based analyses are genuinely committed to the reality of gender. Gender’s reality seems indisputable. After all, gender helps us to explain otherwise obscure social relations, why attributions of gender and talk of gender itself appear truth-apt, why gender categories appear to have causal powers, and so on.⁴⁸

However, the ‘modest’ realism about gender implied by these observations is less obvious outside analytic philosophy. We trace that difference partly to the influence of Haslanger within analytic philosophy and to MacKinnon’s influence outside of it. As we show below, explaining gender with desire lets some of MacKinnon’s light back into analytic discussions of gender. In this final stretch of this paper, we draw on longstanding discussions in moral philosophy to develop one way of being “anti-realist” about gender and so articulate the final theses of the package of theses we dub Gender Unrealism.

Realism and anti-realism diverge in several ways. In particular, realists and anti-realists about a phenomenon disagree about what’s fundamental to explaining it.⁴⁹ Realists’ explanations ultimately involve (non-deflationary) claims about truth, properties, and other entities thought to obtain independently from the mind.⁵⁰ By contrast, anti-realists’ explanations ‘bottom out’ in claims about mental states, such as intentions, desires, and so on. For the anti-realist, “the extra ingredients the realist adds … are pulling no explanatory weight: they just sit on top of the story that tells how our sentiments relate to natural features of things.”⁵¹ Of course, this disagreement about what’s fundamental to a phenomenon does not excuse realists from explaining certain claims about mental states. Nor does it excuse anti-realists from explaining certain claims about truth, properties, and worldly states. That’s because the difference between the two isn’t a difference in what each account explains. It’s a difference in where those chains of explanations end.

Philosophers defend a wide range of anti-realisms (e.g., mathematical, scientific, etc.). We’ll use moral anti-realism to illustrate our approach for exegetical reasons, but also for substantive reasons that allow us to develop the view offered above. According to moral anti-realism, once we’ve explained how the natural, non-moral features of things relate to sentiments like blame and praise, and once we’ve used those sentiments to explain thought and talk that only apparently refers to moral properties, we’ve explained everything about morality that’s worth explaining. Oversimplifying considerably, anti-realists about morality argue that no questions about morality have answers that require looking beyond the non-moral world and how we react to it.

One historically central challenge to anti-realism about a phenomenon is interpreting discourse about that phenomenon in a manner that does not imply realism. For example, well-known arguments involving the T-schema applied to gendered language seem to entail the existence of gender properties, undermining gender anti-realism. To illustrate, if “S is a woman” is truth-evaluable, then the T-schema implies that “S is a woman” is true just when S is a woman. The right-hand side of the biconditional commits us to the property of being a woman. Commitment to properties such as these is a hallmark of realism.

As a result, arguments in this vein push many moral anti-realists to non-standard accounts of moral language, according to which moral talk does not represent acts as right or wrong. If moral talk does not purport to represent acts, it’s not obvious how it could be true or false. As a result, anti-realists can deny that the T-schema applies to moral language and thereby avoid commitment to moral properties of rightness and wrongness. Analogously, gender anti-realists can avoid similar difficulties by embracing a semantics for gendered language that does not represent individuals as women, men, genderqueer, etc.

Many possible semantics for gender terms fit the gender anti-realist’s needs. However, given the analogy we’ve drawn between moral and gender anti-realism, an especially natural choice is moral expressivism, in the tradition of A.J. Ayer, C.L. Stevenson, R.M. Hare, Simon Blackburn, and Allan Gibbard. Expressivists explain an utterance’s meaning through the mental state it expresses, not its truth conditions. We will not focus on the subtleties that distinguish various expressivist proposals compatible with gender anti-realism, but we’ll sketch one possibility for concreteness.⁵² On this interpretation, predicates such as ‘is a woman,’ ‘is a man,’ ‘is genderqueer,’ etc., express, on their gendered uses, plans to classify an individual according to one of the sets of norms thought by some to constitute each gender kind. For example, uttering “S is a woman,” according to this proposal, expresses a plan to use ‘she’ rather than other pronouns when referring to S, and similarly for the other behaviours involved in classifying S as a woman, whichever those are. A contrast helps make expressivism about gender terms vivid:

Gender Realist Semantics: Necessarily, “S is a woman” is true at world w just when S is disposed to accept (in quantity q to degree d) the set of norms of gender classification N, which constitute the gender kind woman at w.

Gender Anti-Realist Semantics: Necessarily, “S is a woman” expresses a plan to classify S according to the set of norms of gender classification N, which are constitutive of classifying someone as a woman.⁵³

Likewise, on one way of thinking about gendered pronouns, use of ‘she’ presupposes that its referent is a woman.⁵⁴ However, on an expressivist account, use of ‘she’ presupposes a plan to classify its referent as a woman. And so on for other bits of gendered language.

Assertions can be correct or incorrect. Realists typically assess an assertion’s correctness by its truth. However, anti-realists cannot easily do so. They often instead appeal to blame and praise, maintaining that, for example, the utterance “killing is wrong” is correct relative to a judge just when the judge approves of the utterance or praises it, in some sense. Transposing this approach to gender anti-realism yields the view that “S is a woman” is correct, relative to a judge just when the judge approves of it.

The standard moral anti-realist approach applied to the case of gender therefore offers a judge-relative account of the correctness of a gender attribution. While this view is coherent, we think that the analogy between gender and moral anti-realism breaks down here. We deny that the correctness of an attribution of gender can differ between judges. But, at the same time, we do not think that there are fundamental gender properties that fix the judge-independence of such attributions. Rather, we think that, very roughly, something like morality fixes a gender attribution’s correctness and we assume that only one moral point of view exists.⁵⁵ Hence, insofar as judges disagree about which gender attributions merit praise or blame, only the judge who occupies morality’s point of view is correct; that is, whether a gender attribution merits praise or blame, in some broad sense, is judge-independent. Nevertheless, positing a single standard of correctness for gender attributions does not contradict or undermine gender anti-realism’s anti-realism. Gender attributions are not correct (or incorrect) by being accurate or true. Rather, there is simply a fact of the matter about which token gender attributions merit approval or disapproval in a circumstance—that is, are correct or incorrect.

This anti-realist account of correctness for gender attribution requires further explanation. Happily, basing gender category membership in desire, as we did in the previous section, pays explanatory dividends here. It is natural to connect the normative to our desires if for no other reason that we should desire certain things, like justice, fairness, the well-being of our family and friends, etc. But a central strain of metaethical theorizing goes further, explaining what we should do, broadly understood, with claims about what others want. Perhaps this is so because welfarist utilitarianism is true and desire-satisfaction is an important component of welfare. Perhaps desires underlie the hypothetical agreement that explains our moral duties. Perhaps desire’s reason-giving authority lies in a deeper practice of interpersonal reason-giving that Darwall (2006) calls the “second-person standpoint.” There are many other such possible, variously controversial connections between desires and obligations. However, regardless of which is correct, we find it plausible that others’ desires constrain how we should act to at least some degree.

Although each of these accounts may suffice for our view, we’ll assume one to simplify our presentation, just as we did with expressivism. We’ll assume a limited form of Democratic Humeanism, defended recently by Kate Manne (2016; 2017). According to it, others’ desires give me reasons to act; or, slightly more precisely, my desire for an end gives you a reason to act in some way when doing so promotes that end. We need not assume Democratic Humeanism, much less the full Humean programme, as the unique explanation of why certain reasons for action exist. Nevertheless, it seems highly plausible that others’ preferences give us reasons to act. If I know that you prefer chocolate ice cream, yet I choose to give you vanilla when I could have just as easily given you chocolate, it seems highly plausible that I’ve ignored an important reason in favour of giving you chocolate, which is explained by your desire for chocolate. In the previous section, we defended the idea that thinking of yourself as, for example, a woman is accepting certain norms of gender classification, where one accepts a norm only if one intrinsically desires to satisfy it. For example, thinking of yourself as a woman involves an intrinsic desire to be addressed with ‘she,’ ‘her,’ and ‘hers,’ rather than other gender pronouns.

Pairing our account of what it is to think of oneself as a woman with a limited Democratic Humeanism illuminates an anti-realist account of the correctness conditions for gender attributions. According to our account, thinking of oneself as a woman involves, inter alia, intrinsically desiring to be addressed by (and this is just an example) female pronouns. According to Democratic Humeanism, my desiring that end gives you a reason to do what promotes it. If I intrinsically desire to conform to the norms associated with being female, then e.g. you have a reason to address me with female pronouns. If you have a reason to do something, then other things equal, you ought to do it. And if you do as you ought, your acts merit approval—that is, they are correct. Consequently, other things equal, uttering “S is a woman” is correct when S intrinsically desires to conform to the norms associated with being a woman.

7. Conclusion: Introducing Gender Unrealism, or, Haslanger and MacKinnon, Redux

In the previous section, we explained three elements of gender: what it is to think of someone as a member of a gender, the meaning of gender terms, and when use of those terms, and the application of related norms, is correct. This explanation is anti-realist, foregrounding our “sentiment” and backgrounding entities like truth and properties insofar as these are understood to be mind-independent.⁵⁶ In the sections preceding the previous one, our arguments led us to embrace a limited social position view of gender according to which social norms constitute gender categories, and we argued that intrinsically desiring to conform to these social norms is necessary and sufficient for being a member of those categories. Putting it all together yields the following package of theses:

Gender Unrealism

The Nature of Gender Categories: Gender categories such as man, woman, genderqueer, etc., are constituted by sex-involving social norms.

A Desire-based Account of Gender Category Membership: Necessarily, S has gender G at world w to the degree that and because S intrinsically desires to conform to the set of norms of gender classification N that constitute the gender kind G at w.

Gender Anti-Realist Semantics: Necessarily, “S is a woman” expresses a plan to classify S according to the set of norms of gender classification N, which are constitutive of classifying someone as a woman.

These theses constitute Gender Unrealism. It captures part of what’s compelling about MacKinnon’s approach to gender. Her influential idea that claims about women express a masculine perspective need not, as Haslanger suggests, be understood as collapsing “the epistemology-ontology distinction.” Rather, claims about women express plans about classifying individuals according to various social norms, at least some of which at least some of the time, on MacKinnon’s feminist analysis, involve patriarchal social norms.

Gender Unrealism echoes two of MacKinnon’s core ideas, dressing them in analytic garb. First, the category woman depends entirely on contingent social practices, some of which are oppressive. Second, concepts like truth, reference, property and real don’t help us understand what it is for an individual to belong to a gender. Because Gender Unrealism ratifies these ideas, we take it to offer a MacKinnon-inspired alternative to the founding assumption of a now-dominant current in analytic feminism that we identified in the introduction.

However, MacKinnon’s legacy surpasses mere metaphysical theory, offering concrete prescriptions for improving the plight of women and, by extension, other oppressed groups. In contrast, we’ve focused on theory so far, rather than praxis. So, we cannot conclude without briefly mentioning some of Gender Unrealism’s practical implications, even if these fall considerably short of MacKinnon’s political legacy. We focus on two.

First, Gender Unrealism is particularly well suited to explaining how morality and gender affect each other. We went through a lot of high-powered metaphysical thinking to get to this point, but it boils down to a very simple lesson. To put it bluntly, we think, absent countervailing concerns, that we should treat others the way they want to be treated since someone’s desires—at least the desires to be treated as a member of a gender—give us normative reasons to treat them that way. Misgendering is not mere misdescription. It can be contested on moral and political grounds.⁵⁷ An explanation of this fact follows straightforwardly from the account. Predications of ‘is a woman,’ ‘is a man,’ ‘is genderqueer’ are not descriptions, according to our view. Rather, they express plans to treat people in gendered ways. Misgendering someone, then, expresses a plan to treat them in ways that they do not want to be treated. Absent countervailing considerations, treating someone in a way that they do not want to be treated is mistreatment or a form of disrespect.⁵⁸ Consequently, misgendering someone expresses a plan to mistreat them, in ordinary circumstances. That is clearly objectionable.

The second concerns the differing aims apparently prescribed by Gender Unrealism and by more orthodox realist analyses. The latter hold that, “the task before us is to construct alternative, modestly realist, ontologies that enable us to come to more adequate and just visions of what is, what might be, and what should be.” But we think that the realist framework distracts us from what matters. For the gender realist, gender justice requires us to construct more just gender realities. This reflects the realist’s distinctive methodology of treating concepts like reality, truth, etc. as fundamental to theory. In contrast, for the gender unrealist, gender justice requires nothing more than treating people well, according to some of their most basic desires. We think realism’s focus on gender reality is distracting. We ought not to treat each other well simply to bring more just gender realities into being, as the realist approach’s foundational commitment to reality seems to suggest. Rather, I ought to treat you well for the sake of treating you well. Gender justice is its own end, not a mere means to reshaping reality.

Acknowledgements

This paper was first conceptualized in 2018 and drafted for presentation to the LOGOS group at the University of Barcelona in July 2019. Thanks to Esa Diaz-León, Dan López de Sa, Adam Sennet, and others in attendance for their feedback. Special thanks to Laura and Francois Schroeter and Mark Schroeder for their feedback and encouragement. Apologies to anyone we missed. It’s been a long journey.