/ 'Yep, I'm Gay': Understanding Agential Identity


What’s important about ‘coming out’? Why do we wear business suits or Star Trek pins? Part of the answer, we think, has to do with what we call agential identity. Social metaphysics has given us tools for understanding what it is to be socially positioned as a member of a particular group and what it means to self-identify with a group. But there is little exploration of the general relationship between self-identity and social position. We take up this exploration, developing an account of agential identity—the self-identities we make available to others. Agential identities are the bridge between what we take ourselves to be and what others take us to be. Understanding agential identity not only fills an important gap in the literature, but also helps us explain politically important phenomena concerning discrimination, malicious identities, passing, and code-switching. These phenomena, we argue, cannot be understood solely in terms of self-identity or social position.

In April 1997, Ellen DeGeneres officially ‘came out’ in Time magazine, the cover boldly stating, with characteristic nonchalance, “Yep, I’m Gay”. In one sense, DeGeneres’s coming out was not informative: it was widely known that DeGeneres was gay. But in another sense, it was hugely informative. It communicated to the world that DeGeneres wanted this knowledge to move outside people’s heads and beyond closed doors, and to impact how she was treated in her public life.

DeGeneres’s coming out is but one instance of someone attempting to revise their social position, or how they are seen and treated by others, to match their self-identity, or who they take themselves to be. How should we understand this phenomenon? Social metaphysics has given us many tools for understanding what it is to be socially positioned as a member of a particular group, such as women or people with disabilities.[1] And a growing literature asks what it means to self-identify as belonging to such a group.[2] But there is little exploration of the general relationship between these two questions.

We take up this exploration, which leads us to develop a model of agential identity. On our model, agential identities are the self-identities we make available to others—they bridge what we take ourselves to be with what others take us to be.

Agential identities are unexplored in social ontology, particularly in comparison with social position and self-identity. We think a concept of agential identity is central to understanding how individuals navigate social landscapes. As it stands, this literature has a conceptual lacuna concerning how individuals purposefully shape the way they are perceived. It has neglected individuals’ ability to place themselves into particular social positions by choice, rather than simply be placed into them.

Agential identity helps us explain certain politically important phenomena that cannot be understood solely in terms of self-identity or social position. We discuss four such phenomena in detail in Section 2. The first concerns an easily misconstrued (or simply missed) form of discrimination: discrimination targeting neither an individual’s self-identity nor their social position, but rather their ability to revise their social position on the basis of their self-identity. As a case in point, we focus on regulatory policy preventing trans persons from achieving social transition, which brings their public perception in line with their gender. Such discrimination, we argue, is best understood as targeting people on the basis of their agential identity.

Is it always wrongful to place barriers in front of someone’s expression of identity? We think not. The second issue we address concerns our obligations toward others’ identities, particularly when those identities are malicious. We agree with a general sentiment that we have a duty to recognize and respect others’ identities. But how far does this duty extend? For example, if a friend begins to claim a white nationalist identity, does this identity deserve respectful, socially cooperative uptake? It seems not, but on what basis may we refuse social uptake of this (or any other) identity claim? Here, we suggest that the difference between self- and agential identities is crucial. While it may be that someone’s self-identity is non-voluntary, manifesting this self-identity as an agential identity often is cultivated or chosen. When an agential identity causes unjustified harms, this defeats any obligation to respect that identity. In short, by distinguishing between self- and agential identity, we leave room for circumstances in which it is appropriate to refuse social uptake of an identity claim.

The third puzzle concerns a phenomenon known as ‘passing’. Standardly, passing is defined as occurring when a person achieves the social position of a social group membership without in fact belonging to that group—that is, when they are falsely taken to be a member of a social group. A paradigm case of passing, in this sense, was when John Howard Griffin successfully travelled through the American South in the guise of a black man while writing Black Like Me. To understand passing only in this sense, though, ignores a nearby phenomenon: one where someone’s social group membership is perceived accurately, but they nevertheless wish to revise, eliminate, or otherwise change their social position. There is an important sense in which such an individual is passing, but one that cannot be accounted for only in terms of group membership or self-identity. Here too, we argue, we need a concept of agential identity.

Finally, we address the issue of ’code-switching’. Code-switching is a ubiquitous phenomenon. When we switch what language we speak as we cross borders, or what clothes we wear between work and the pub, we are code-switching. Code-switching often is glossed as the ability to behaviorally adapt to various social contexts. We think glosses like this do not get to the heart of code-switching, and that our account of agential identity allows for a sharper, clearer analysis. Here, we argue that code-switching can be helpfully understood as the intentional or automatic revision to real or perceived agential identities as one moves between contexts. That is, in other words, code-switching occurs when (intentionally or not) we revise the way that we perform agential identities, whether or not those identities are genuine.

1. The Dimensions of Agential Identity

What is an agential identity? How does agential identity relate to social groups? What is distinctly agential about them? Answering these questions is our present goal, and one that is essential for what we take to be the heart of our project: the normative upshots discussed in Section 2.

Agential identities are, on our picture, a relation between individuals, their self-identities, and social positions. They act as a bridge between one’s internal identification and one’s preferred public perception.

1.1. Foundations of Agential Identity: Social Positions

In order to understand this picture of agential identities, we need to understand social positions—roughly, positions into which individuals are sorted when they are perceived as belonging to a certain social group.

Here, we follow Sally Haslanger’s proposal, on which social positions are occupied by persons on the basis of their engagement in social practices, or coordinated responses to other persons, ideas, and the material world.[3]

These coordinated responses arise from our shared cognitive attitudes. For example, consider the Eruca sativa plant. One of the social practices associated with Eruca sativa is referring to it by a common name. In the United Kingdom that name is ‘rocket’. Americans, however, have a different practice, instead referring to Eruca sativa as ‘arugula’. Even though these two contexts share the view that Eruca sativa is edible, the difference in their referring practices means that they do not have a coordinated response to that resource, at least when it comes to naming.

We’ll call the clusters of beliefs, concepts, attitudes, and so forth that give rise to such concrete practices a ‘social blueprint’.

Blueprints facilitate social interaction by allowing us to “interpret and organize information and coordinate action, thought, and affect, ” (Haslanger 2016: 126). In this way, they serve as the basis for many behavioral and emotional dispositions. Moreover, blueprints are public and, generally speaking, resistant to change—you can’t get Americans to start calling Eruca sativa ‘rocket’ by simply insisting that they’re wrong.

The ability to coordinate with others requires awareness of and responsiveness to a vast network of interrelated social blueprints. We can think of these networks as the social contexts we occupy. They build coordinated behavior, thought, and even affect in ways that generate social practices. When we (willingly or unwillingly) participate in these social practices, we gain social properties, or properties that are significant within our shared blueprints. For our purposes, social properties are important because of how they relate to social positions. We are assigned a social position when we are regularly perceived as possessing (enough of) the properties that are, within shared blueprints, associated with a group corresponding to that position. For example, it is plausible that a number of properties are associated with persons who are women, such as particular kinds of dress, speech, occupation, and presenting as having a female-sexed body. It is not necessary that one is perceived as possessing all of these properties to be assigned a woman-coded social position—it is sufficient to be perceived as possessing enough of them with sufficient regularity.[4]

The social positions we are assigned further affect which social properties we have. Consider the relationship between a professional football player and a professional football team. By signing a contract with a professional football team, an individual (say, Dan Marino) immediately comes to stand in a new social position—that of a professional football player. By occupying this position, Marino in turn gains new, additional social properties, such as being an employee and having fans. Similar things can be said of the properties parent of, convicted of, or spouse of —if perceived as possessing these properties, they typically result in immediately being assigned a corresponding social position, along with the additional social properties, expectations, or rights associated with that position.[5]

Social positions, in short, arise from being perceived as having certain social properties—namely, relations to persons, ideas, and the material world that are, according to our shared blueprints, associated with a social group. And they may, in turn, give rise to additional social properties.[6] Because these properties exert profound influence on our lives—constraining our relationships with others, the environment, institutions, and ourselves—they are extremely important.

Before using this framework to construct a concept of agential identity, it is important to flag that, for simplicity’s sake, we have been speaking as though a unique set of social properties—and also a unique social position—is associated with each group. But, of course, this is not always, or even usually, the case. Many groups (perhaps most) have distinct (though overlapping) sets of social properties associated with them. So, too, multiple social positions often are associated with a single social group. And while we understand that this sketch of social positions leave much of this complexity under-explored, we leave a richer examination to future work.

1.2. Agential Identity: Externalizing Self-Identity

Social positions are determined by how we are perceived and treated by those around us. But, this does not mean that we are without agency in the matter. Consider, for example, a young man deciding whether to come out to his parents as gay. For all they know, he is straight. So, he currently occupies the social position of being straight, at least at home.[7] But, having realized that he is gay, he comes to self-identify as gay.[8] This internal identification is crucial to agential identity, but it is the choice to come out, the choice to make that private identity public and allow others to perceive him as gay, that would create an agential identity. This relationship between self-identification and preferred public perception is at the core of agential identity.

Of course, agential identities may not have uptake. If the young man’s parents ignore him or tell him that he’s confused, and then continue to ask after his girlfriend, they in effect refuse to perceive him as gay. This, in turn, denies him access (in that context) to the social position of a gay man. In such a case, he has the agential identity of being gay in virtue of his intentions and actions, but lacks the social position. Similarly, after being outed as having white parents, Rachel Dolezal was no longer taken as being black, despite asking to be. She ceased to occupy the social position despite still self-identifying as black and still making that self-identification public—despite still having the agential identity.[9]

From these cases, we can see that having a particular social position is neither necessary nor sufficient for having the associated agential identity. Rather, the following conditions give rise to agential identities:


  • In order to have an agential identity, one must self-identify as a member of the group in question.

Position-directed Externality

In order to have an agential identity,

  1. (externality) one must, consciously or unconsciously, make their self-identification externally available, whether by engaging in certain behaviors or displaying certain readily perceptible features, where those behaviors or features outwardly manifest (or, in certain cases, are merely intended to manifest) social properties associated with a particular group,[10] and
  2. (position-directedness) one must allow, or at least accept, that others take them as occupying a group associated social position as a result of fulfilling (a).

The self-identification condition states that, for example, one cannot have the agential identity of being gay without actually identifying as gay. The position-directed externality condition adds to this that one cannot have the agential identity of, say, being a woman without outwardly manifesting (or intending to manifest) social properties associated with women (according to one’s blueprint for this role). ‘Behaviors’ here includes not only actions, but also expressed attitudes, preferences, physical appearance, etc.[11] These require a good deal of unpacking, of course. We say more about each of these in turn.


Bill Clinton self-identifies as a Democrat. Ellen DeGeneres self-identifies as a lesbian. George R. R. Martin self-identifies as a fantasy author (and probably a nerd). Each of these identities shapes the attitudes, beliefs, and preferences of those who possess them. Moreover, each of these individuals expresses a sort of affirmation of these identities and kinship with the associated group. By contrast, while all three of them would identify themselves as over 5’5”, it’s highly doubtful that any of them self-identify with that trait in any meaningful sense.

While a thorough exploration of self-identification is beyond the scope of this paper, we should say at least a few words about what it is to self-identify with a social group. Common-sense notions of self-identification abound, but perhaps the most common is the idea of the ‘true self’, which is taken to be the instinctual source of creativity and authenticity.[12] We take this view to oversimplify our interaction with the social practices we encounter. It suggests a process of unconscious development or discovery, as if the self is a fully-formed artifact merely awaiting archaeological introspection. On this view, the obfuscated true self is not subject to development through our choices (social or otherwise). But this is clearly at odds with much of our experience.

Self-identifying with a social group, especially doing so strongly, changes the individual we take ourselves to be. To see this, consider the difference that self-identification can make in the case of ethnic identity. Suppose a pair of twins has some Native American heritage. The first, call her Anya, strongly self-identifies as Native American. As a result, she wishes to learn more about Native history, attends events aimed at Native people, and participates in Native cultural activities. The second twin would identify herself as Native American if asked, and perhaps weakly self-identifies as Native American, but does none of Anya’s cultural study. These two women have very different self-identities.

This is difficult to explain on the “true self” view, however. To begin, it is unclear how something like the “true self” could govern traits as nuanced and culturally specific as ethnic identity. Even granting that, it is difficult to see how a true self view would explain the very different self-identities of two people with such similar backgrounds without resorting to the suggestion that one (or both) of them has gone wrong somewhere along the line. On such an account, one of these women might lack authenticity, hiding her true self. But this seems both unlikely and unjustly chastising.

If not the ‘true self’, however, then what?

We cannot fully address this question here. However, the sense of self-identity we deploy here is one distinguished by a sense of kinship and norm-relevance. Individuals who self-identify with a group see themselves fitting it, as the sort of person who ought to be in it, and they actively see themselves as a part of it (or as a would-be part of it, for cases in which public knowledge is necessary for membership and those in which the group is not yet broadly recognized).

Richard Jenkins defines self-identification as “the systematic establishment and signification, between individuals, between collectives, and between individuals and collectives, of relationships of similarity and difference” (Jenkins 2004: 19). Our notion of felt ‘kinship’ picks out a similar notion, shifting focus to internally-directed perceptions of similarity. To feel kinship with a social group is to see oneself as relevantly similar to other members of the group, where one takes this similarity to mean that one does (or should) have properties that are sufficient for group membership. In the cases we’re concerned with, this feeling of kinship goes hand-in-hand with norm-relevance.

Norm-relevancy builds from this sense of group membership. Borrowing from Katharine Jenkins’s view of gender self-identity, we can understand norm-relevancy as experiencing a substantial subset of the norms associated with a particular group (in one’s social context) as applicable to oneself (Jenkins 2018). In keeping with the groundwork laid in Section 1.1, one’s sense of these norms comes from the shared blueprint for that group’s social position(s). Importantly, experiencing a set of norms as relevant to oneself does not entail complying with those norms. Someone who self-identifies as a woman, for example, might take a norm about leg-shaving as relevant to her, and thereby experience herself as violating or rejecting a norm by refusing to shave her legs. While she need not see herself as following the prescriptive norms of her group, her attitudes and preferences are nonetheless shaped by them in this way.

Neither kinship nor norm-relevance mean that self-identity is fully explained by internal cognition. The ability to experience kinship with a particular group is inextricably bound to that group’s shared blueprint for responding to the world. In Section 1.1, we described these blueprints as guides to social meaning, which consist of habits, concepts, and beliefs. Self-identifying with a group entails using these blueprints to read oneself as the sort of person who does (or should) have the social properties necessary to be part of that group.

Several features of this view are worth highlighting. First, as we just saw, group membership is not a necessary condition on self-identity. If African ancestry is necessary for membership in the social group of blacks, then Rachel Dolezal is not a member of that group. Nevertheless, this does not prevent her from self-identifying as black, however erroneous that may be. Questions of understanding and meeting the actual conditions for group membership are distinct from the question of what it takes to have the corresponding self-identity.

Second, identifying-oneself-as is distinct from self-identification. We identify ourselves as many things—a certain height and weight, a particular gender, a participant in certain athletic activities, a True Fan of certain television shows, a lover of various foods, etc.—but very few of these properties are stalwart, persistent, or central enough to become part of our self-identity. Most are merely properties we know that we have.

Third, self-identification varies in strength. Both of the central features we’ve discussed, norm-relevancy and kinship, can be felt to varying degrees. While someone may self-identify as both a Democrat and a Trekkie, the degree to which they experience norm-relevance and kinship pertaining to Democrats may be outweighed by those pertaining to Trekkies, or vice versa.

Finally, while feeling norm-relevancy and solidarity requires having concepts of the relevant group, it does not require having a label for that group. That is, all one needs is an ability to recognize the group and to recognize one’s own relevant similarity with it.[13]

Position-Directed Externality

This is the feature that makes agential identity agential. There are two aspects of role-directed externality: one’s perceptible features and one’s attitude toward those features. We begin with externality.

The externality condition states that one must manifest or intend to manifest social properties associated with a given group. This amounts to outwardly conforming or intending to conform to shared blueprints for the target social group. Of course, many social groups have multiple positions associated with them and, therefore, multiple (possibly quite disparate) or complex shared blueprints. Role-directed externality need not target more than one of those positions and need not occur under the guise of conforming. We often unconsciously behave in ways that conform to our agential identities without any intention to do so.

For example, someone with the agential identity of a Trekkie must, in some way, conform or intend to conform to the particular blueprint they’ve internalized for Trekkies. That might mean that they attach a Star Trek logo to their keychain, attend a fan club meeting, or simply talk with others about the series. None of this behavior needs to be undertaken by the agent under the guise of conforming to this blueprint. More often than not, these behaviors are merely byproducts of self-identifying with some group. They are artifacts of the fact that self-identities influence our preferences, attitudes, and behaviors. This is clearly the case, for example, for most women’s blueprint-conforming behavior—seldom is wearing a skirt explicitly meant to conform to norms for ‘woman’.[14]

Sometimes, however, intentions do come into play. There are two ways this might happen. First, someone might attempt to outwardly conform with the relevant blueprint, but have yet to figure out exactly how to do so.[15] Take, for example, a budding Trekkie who, though they know what they want to do—be fluent in Klingon and other Trekkie lingo—hasn’t yet mastered these things. In such cases, an attempt to conform to certain blueprints may fall short of in fact conforming. Such instances still satisfy the externality condition.

Second, intentions to conform may be forward-looking, involving partial planning for the future (Setiya 2015). This sort of forward-looking intention is sufficient for agential identity so long as the position-directedness condition is met. This condition specifies an attitude one must bear toward their action, namely that they allow and accept others taking them as occupying a certain social position as a result of that intended behavior. So, in the case just mentioned, the budding Trekkie in question has an agential identity as a Trekkie only if, having expressed their intention, they allow others to treat them as a Trekkie on the basis of those intentions. Without this, they may still self-identify as a Trekkie, but they lack the agential identity. They are in the Trekkie closet.

It is also important to note that simply allowing others to take them as a Trekkie is not sufficient for meeting this condition—they must allow others to take them as a Trekkie as a result of their externally manifested properties. The connection between these properties and attitude is critical to the agential nature of agential identity, and marks another substantial distinction between agential identity and social position. While social positions can be assigned to us without our consent, agential identities cannot. Thus, regardless of how often people misgender a feminine-presenting genderqueer person (e.g., Ruby Rose), and regardless of whether they do so on the basis of behavior that does outwardly conform to a blueprint associated with those who have a ’woman’s social position’, Rose’s refusal to accept that position means that position-directedness is not met.

Finally, note that this condition does not require uptake. It may be that no one in fact recognizes or accepts the individual’s agential identity, either because they do not take them to be a member of the relevant social group, or because they assign to them a social position that, though associated with that group, is not one to which the individual allows or accepts assignment. In the case of the budding Trekkie who is trying but failing to actually conform, for example, she satisfies the position-directedness condition, regardless of anyone else taking her to be a member of the social group ‘Trekkie’. Similarly, early feminists such as Susan B. Anthony, who sought a different position for women than the one assigned by the pervasive cultural blueprint, still satisfied the position-directedness condition, even if they were unsuccessful in achieving their aims.

1.3. Nuances

Together, self-identification and role-directed externality give rise to agential identities. This relatively simple picture might belie the complexity of agential identity, however. Social phenomena, we reiterate, are messy. Intention, allowing, and acceptance are attitudes that admit of borderline cases. Someone who is just coming out as trans, for example, might well be uncomfortable with people taking her as a woman—she might be a borderline case of accepting and allowing. In this case, her agential identity with respect to gender will be vague (though her self-identity is not). Alternatively, she might be comfortable with strongly externalizing her self-identity in some contexts, but not at all in others. In this case, she may be disposed toward ‘code-switching’ with respect to gender. These complexities, we take it, are to be expected—insofar as social phenomena are messy, that mess must find some place in any adequate theory of agential identity. We return to the case of code-switching in Section 2.4. We think one important upshot of our theory of agential identity is that it helps us make sense of what code-switching is and how code-switching helps us negotiating between self-identities and social contexts. In this section, however, we focus on certain nuances of agential identity per se. In particular, we are interested in ways that agential identities are more complex than we’ve suggested so far, and vary along several dimensions. While the following sketches are meant to provide a sense of these dimensions, we take it that there is much more to be said in each case, which we leave to future work.


For one, agential identity, social position, and group membership are independent to an extent. This is because agential identity depends neither on how others perceive you, as social position does, nor on social group membership. We can be wrong about whether our agential identities align with our social group memberships. For example, in Section 2.3 we discuss the case of Sean Cole, an American teenager who sincerely, but falsely, insisted that he was British. Similarly, we can have agential identities that clash with our social positions: this occurs whenever an agential identity does not receive adequate social uptake. Cole’s parents refusing to treat him as British, and instead continuing to treat him as American, is but one example of social position and agential identity coming apart.

This is not to suggest that social position and group membership do not influence agential identity, or vice versa. It is also not to suggest that our agential identities are not, in most cases, a reliable guide to our group memberships. We typically have privileged (though not infallible) access to facts about ourselves. Insofar as we are epistemically privileged with respect to identifying features of ourselves that are relevant to certain group memberships, such as the nature of our sexual attractions, our racial ancestry, or our religious beliefs, those facts are likely to influence our agential identities. Moreover, how others perceive us can affect how we perceive ourselves. This is perhaps most obvious during stages of childhood development in which we learn our group memberships from those around us, and come to self- and agentially-identify according to those memberships.


In addition to independence, agential identities are often ‘sticky’, where stickiness concerns the ease of entering and exiting the agential identity. Extremely sticky agential identities are difficult to enter or exit.

Stickiness may obtain for myriad reasons. Moreover, it can occur either at the level of entering or exiting a particular self-identity, or at the level of entering or exiting the externalization of that self-identity.

Gender provides a clear example of agential identities that typically are difficult to enter or exit in both senses. Though not without exception, gender self-identities usually are stable across time. Research on trans youth suggests that this is the case even in the face of immense social pressure and contradiction. For most people, gender self-identities are not quickly or easily changed, much less changed voluntarily.

In addition, gender externalization also is difficult to revise. In dominant contexts, the pervasive shared concepts and norms for classifying someone’s gender rely on features, such as developmental conditions and secondary sex characteristics, that are hard to change. Many people struggle to even comprehend changing these features (even if one can minimize them), and—as we discuss in Section 2.1—social practices surrounding such features put up additional barriers to changing them.

But many agential identities are not so sticky. Take, for example, the agential identity of being a triathlete. For many people, their self-identification as a triathlete is not extremely difficult, much less impossible, to revise. If they do not have this self-identity, they have the capacity to commit themselves to triathlon training and competition in order to develop this identity. If they do have it, they can turn their attention to alternative sports or hobbies in order to minimize this identity. Moreover, whether someone who self-identifies as a triathlete externalizes this self-identity is, despite appearances, voluntary. Even if triathletes struggle to keep this self-identity to themselves, they could do so.

A final point to note is that, while we are speaking generally, the stickiness of an agential identity usually is not uniform across individuals. Some people are gender fluid, and their gender self-identity changes frequently. For a transradial amputee, gaining the agential identity of a triathlete likely requires a complex process to obtain a prosthesis capable of supporting triathlon training. We expect to find similar exceptions with respect to most agential identities.


Finally, agential identities also exhibit varying degrees of salience. A strongly salient agential identity is one that is easily perceivable to others when expressed. For a given agential identity, there may be contexts in which this identity is not at all salient, as well as contexts in which it is highly salient. Gender and race are cases of agential identities that are often strongly salient across most or all contexts. But other identities, such as professional identities, are more likely to vary in salience across contexts.

In some cases, an agential identity’s salience is up to the individual, who can control this salience in various ways. For example, consider a gay teenager who comes out to his parents, but otherwise acts conservatively at home in order to minimize the salience of his gay identity. This same teenager might, in other contexts, deliberately talk and behave in ways that persistently signal and emphasize his gay identity.

In other cases, individuals have little to no control over their agential identity’s salience. Consider someone who externalizes their self-identity of being a Trekkie, but unfortunately is surrounded by people who do not know anything about Star Trek or Trekkies. In this case, the person does have the agential identity of a Trekkie, but this identity is invisible to those around them—that is, it is not at all salient in their immediate context.

2. Work for Agential Identity

We have argued that agential identities should be understood as having two components: self-identification and position-directed externality. We emphasized that agential identities reveal agents’ chosen orientation toward particular social groups on the basis of self-identification. The model, then, describes agential identities as pertaining to psychological features and individual agency, but within constraints determined by the available (or potentially available) social positions. On one hand, this is to make the fairly mundane point, emphasized by many philosophers, that choice is exercised within constrained options. But, by providing a link between agents’ social structures and their preferences, it also allows us to provide structural explanations for unique forms of discrimination, for our moral and political obligations concerning agential identity, for puzzles surrounding ‘passing’ and ‘traveling’, and for the phenomenon of ‘code-switching’.

Before exploring these upshots, we should first introduce our understanding of a structural explanation. Here, we again follow Haslanger (2016: 2), who takes a structural explanation to be one in which, rather than limiting focus to the individual whose behavior we want to explain, we look to the system in which the individual is embedded for clues about how their behaviors were constrained. So construed, structural explanations can be fitting for something so minute as why you extend a hand when first meeting someone. A useful explanation of this action will not simply focus on your arm movements and the neurological processes. Rather, it will explain the social meaning of this action (or lack thereof)—that is, it is considered polite to shake someone’s hand when first meeting, and impolite not to do so.

We think that a similar style of explanation is most useful when we want to explain individuals’ social behavior more generally. Recall the earlier discussion of shared blueprints as collections of shared habits, concepts, beliefs, and the like that facilitate interpretation of social meaning. Among the many important consequences of shared social meaning are what we have been calling ‘social positions’. And, with this in mind, we can begin to see why a structural explanation of individual social behavior is available. Social interactions are at least in part driven by persons’ agential identities, because agential identities are the externalization of self-identification, aimed toward a social position. And so, if we want to predict or explain someone’s behavior, we suspect that it may not be illuminating to merely point to their mental states in isolation—that is, to their self-identification. We will achieve a richer explanation if we look instead to the agent’s agential identities and how their behaviors relate to the social positions toward which those identites aim.

2.1. Agential Identity Discrimination

Many feminist writers have pointed out that important forms of discrimination occur at the structural, and not only individual, level.[16] The most familiar kinds of structural discrimination, such as hiring practices that are discriminatory against persons perceived to be women, or educational practices that discriminate against students perceived as black, affect persons who are perceived as members of marginalized social groups. We think an additional form of structural discrimination has gone largely unnoticed. This discrimination primarily targets persons with certain agential identities rather than persons with certain group memberships. When this kind discrimination occurs, it manifests as an effort to frustrate an individual’s success in externalizing their self-identity, making it more difficult for them to come to occupy the social position associated with that identity.

We can first approach this phenomenon through illustration. One glaring example occurs within the current American healthcare system, particularly with respect to trans persons’ access to medical care necessary for what is sometimes called ‘social transitioning’, or revising what gender one is perceived by others to have. For better or worse, this change in social perception is, in most contexts, required in order for trans men and women to successfully begin to occupy social positions associated with the group ‘men’ or the group ‘women’. And yet, many trans persons are unable to access this medical care. Consider the following excerpt from a newspaper article about a trans woman named Alena:

To be able to [be taken to be a woman], she requires a steady, affordable supply of prescription hormones—something impossible to come by in Albany. Alena finds herself in a double-bind that makes securing affordable prescription hormones for transition virtually impossible. She is one of 600,000 Georgians who have been left uninsured for healthcare as a result of the Republican state’s refusal to expand Medicaid under Barack Obama’s Affordable Care Act . . . Even if she were entitled to Medicaid, Georgia—along with 15 other states—specifically excludes health coverage for transition-related treatment, so she wouldn’t get anything anyway. (Pilkington 2015)

The first thing to notice about this case is that whatever explanation lies behind the discriminatory treatment toward Alena, it must be structural in nature. This is because no individual attitudes are needed in order for the discrimination to persist: even if everyone surrounding Alena was without explicit or implicit bias against trans women, these policies discriminate against her.

Second, the medical barriers affecting Alena cannot be explained as discriminatory on the basis of either her perceived group membership or her self-identity. That is, these structures are not discriminatory against her for being assigned male at birth, for self-identifying as a woman, or for her perceived social group membership (which, against her wishes, is that of a man). Even holding these things fixed, were Alena to suffer from prostate cancer she would most likely be able to receive estrogen-based therapy as a treatment for that cancer (American Cancer Society 2018).[17] The difference comes down to intention: given that she was assigned male at birth, identifies as a woman, and wishes to revise her perceived group membership, she is denied access to hormone therapy. In other words, it is her agential identity—the combination of her identity as a woman (or as a trans woman) and the effort to make that identity public—that triggers these discriminatory policies. Such policies systematically target persons who were assigned male at birth, but who want to transition to occupy the social role corresponding to the group women. Similar policies target persons who were assigned female at birth, but wish to occupy the social role corresponding to the group men. As Dean Spade points out, “Medicaid provides all of the gender-confirming procedures and medications that trans people request to nontrans people and only denies them to those seeking them based on a transgender diagnostic profile” (Spade 2015: 148). For example, in many American states, Medicaid policies provide hormone therapy, chest surgery, breast, testicle, and penis reconstruction, electrolysis and other procedures for nontrans, but not trans persons (Spade 2015: 148–149). We think that these policies, which prevent trans persons from accessing affordable healthcare that would facilitate social transition, are unjust.

This level of analysis is valuable because it sidesteps ontological issues concerning what it means to be a woman. Public debate over this ontological question often impedes progress in rectifying cases like Alena’s. Rather than discussing whether someone ought to have a (prima facie) right to revise their perceived gender social position, public disputes over trans medical care are often bogged down in vicious debates over whether trans women ‘really are’ women. Given the analysis of Alena’s situation we have provided, we think this debate is a red herring. Agential identity does not depend on social group membership: rather, agential identity concerns one’s preferred perceived social group membership, regardless of whether or not one in fact had that membership prior to the perception. The ontological question of Alena’s ‘true’ gender, then, is irrelevant: what matters is that Alena ought to have access to a perceived group membership—and, thereby, a social position—that facilitates her mental and physical wellbeing.[18] The discrimination is wrong because it unnecessarily and detrimentally frustrates Alena’s (and other trans persons’) autonomy over their social position.

Discrimination against trans women is not the only form of agential identity discrimination. Similar discriminations can be seen in identification requirements that prevent persons from wearing certain religious clothing such as burqas, or laws that prevent queer persons from adopting foster children. Historical examples abound as well. Consider, for example, past suppression of languages other than English under British domain. English education policies in India, Scotland, and other locations have, at various times, forbidden the use and teaching of the native language. During the early 18th century, the Society for Propagating Christian Knowledge, an Anglican missionary organization, set up a series of “Charity Schools” directed toward a “reformation of manners” for the Scottish clans (Mason 1954: 3). Schoolmasters were instructed to punish students caught speaking Gaelic, even while students were playing in the schoolyard (Mason 1954: 8). Such cases of linguistic imperialism, in which a dominant culture’s language is transferred to a subordinate group, are (among other things) forms of agential identity discrimination (Phillipson 1992). By eradicating native languages, they strip away an important means by which people externalize ethnic self-identities.

All of these practices systematically target persons with self-identities who are attempting to be seen by others in accordance with those identities. Arising out of historical prejudice or ignorance, these practices consistently either require persons to ‘stay in line’—that is, not revise their assigned social position—or else ‘stay in the closet’—that is, not allow their self-identification to have public uptake.

While we suspect others will share the intuition that discriminatory practices like these are wrong, it is worth examining in more detail why this is the case. We think this is because they violate a prima facie obligation to allow and respect an individual’s determination of the social positions they occupy. In bridging the gap between self-identification and perceived social group membership, agential identities play a critical role in self-determination. For this reason, we take it that the exercise of agency that is central to agential identities is morally significant in its own right and, unless defeated by other considerations, ought not to be prevented. Such potential defeaters arise from the many roles—legal, epistemic, practical, and so forth—that social positions play in society. Because potential defeaters are unique to each social position, however, we cannot give a general answer to the question of whether the prima facie obligation we’ve identified outweighs them—nor do we suspect such an answer will be forthcoming. This question is further complicated by the fact that different moral views may accord more or less weight to these potential defeaters. It is not our intent to stake out a position on such debates. Rather, we mean to point out the structure of agential identity discrimination and to suggest that the cases we’ve identified are, indeed, wrongful.

Recognizing and accommodating agential identities can have significant implications for the social positions, social categories, and societies that underlie them. In some cases, agential identity discrimination may perpetuate harmful social positions that would be bettered by allowing the position to be determined by persons who socially identify with the relevant group. This latter thought is captured by Wesley Morris in his New York Times article, “The Year We Obsessed over Identity” (2015):

[I] wonder if being black in America is the one identity that won’t ever mutate. I’m someone who believes himself to have complete individual autonomy, someone who feels free. But I also know some of that autonomy is limited, illusory, conditional. I live knowing that whatever my blackness means to me can be at odds with what it means to certain white observers, at any moment. So I live with two identities: mine and others’ perceptions of it.

Here, Morris describes a situation in which his agential identity aims toward a social position associated with blackness that differs from the position ‘white observers’ associate with blackness. We would be better off if we allowed the social position of black persons to be determined by persons who identify as black, rather than by ‘white observers’. That is, we think that agential identities of black persons deserve uptake, and that this would improve the social positions available to them.

Similarly, it is reasonable to think that we would do well to (continue to) expand the social positions for ‘women’ to accommodate those who agentially identify as women. As we have seen over the last five to ten years in America, changing the entry conditions for a social position can in turn affect the everyday concept of the associated group that fits that role. That is, by changing who will be seen as a woman, we affect our everyday concept of what a woman is, shifting it away from a biological concept toward an identification concept.[19] Our concepts and their associated social positions seem to be, in this respect, situated in a feedback loop, with changes to one affecting the other. And so, we would like to suggest, it may be that accommodating agents’ agential identities can have positive effects far beyond the individual, and may in some cases further the cause of social justice by making our social positions more inclusive, and determined to a greater extent by persons occupying them.

This is not to suggest that all such accommodations are positive, however. For example, the high-profile cases of Rachel Dolezal and of Emile Ratelband, the 69-year-old Dutch man who requested that his legal age be changed to 49 (Domonoske 2018), are less clear. In Dolezal’s case, social acceptance of her agential identity as a black woman would require revision of racial categories. Such revision would involve shifting away from current understandings on which race is at least partially a matter of ancestry toward a version on which race is a matter of desire and personal sentiment. In Ratelband’s case, this kind of accommodation would require a rethinking of many of the legal and practical rights and obligations that track records of age, including age limits surrounding drinking, marriage, voting, retirement, and so on. While these two cases, and many like them, are structurally analogous to those we’ve considered, the broader ramifications of the changes they involve are quite different. Because of this, they require focused analyses of their own, which are beyond the scope of this paper. More generally, as these cases demonstrate, the questions of when and why an agential identity ought to be accommodated are complicated. And, as we discuss in the next section, some instances of accommodation may even be harmful.

2.2. Malicious Identities

The previous discussion motivated the notion of agential identity discrimination, and in particular the idea that it is prima facie wrong for social structures, norms, or practices to frustrate a persons attempts to have their perceived group membership (or social position) align with their self-identity.

With that said, however, one might worry that these barriers are not always wrong. What about self-identifications with malicious groups? Suppose, for example, that you are family of Peter Tefft, a young man from North Dakota who proudly identifies as a white nationalist, and even goes so far as to claim that “Nazi is a racial slur against whites ” (White 2017). Perhaps Peter would claim that he has a right to have his identity respected.[20] After all, it is not uncommon to hear tenets along the lines of ‘Respect everyone’s identity’ or ’Don’t question other people’s identities’. At the same time, it is hardly controversial (we hope) that white nationalism is a malicious ideology. Given this, what obligations—if any—does Tefft’s family have toward his identity? What are our obligations toward malicious identities more broadly? How do we reconcile the intuition that there is something wrong about his case with these respectful truisms about identity?

We think this question is difficult to answer only using the conceptual tools of social position and self-identification. The “noble” social position that white nationalists envision for themselves is one we hope goes unfilled: it would be best if there were no practices particularly supportive of this ideology, no individuals whose actions fit with it, and no corresponding social groups. And yet, there are such social positions. There are such practices. And there are such groups. So, it is unsurprising that some people come to self-identify with them. Moreover, it might be the case—at least with respect to some self-identifications—that agents do not have voluntary control over those self-identifications. If this is the case, then it is at least unclear what the basis for judgment against individual agents’ self-identities might be—even if the basis for judgment against the racist social norms, practices, and structures that give rise to such identities is quite clear.

And yet we think there is a clear sense in which someone like Tefft is blame-worthy for his identity, even if the self-identity is non-voluntary. That is, we think that we do not owe Tefft respect or acknowledgement of his identity.[21] We can even denounce this identity. But what is the basis for this response? Where, in other words, does an individual’s responsibility for their identity lie, especially if the self-identity is non-voluntary? We propose that this puzzle is resolved by looking to the moral significance of persons’ agential identities.

Two claims strike us as wrongheaded ideas about agential identities: First, that they are not morally evaluable (or are always good) just because they are (among other things) a chosen expression of self-identity. Second, that they are not morally evaluable because we have no control over these identities. The reason we dismiss the first suggestion is, in a way, simple: agential identities are (in part) ways of being in the world. And we think that there are better and worse ways of being in the world. Consider this example from Anthony Appiah:

[Suppose] I adopt a life as a solitary traveler around the world, [and] my parents tell me that I am wasting my life . . . You don’t have to be a communitarian to wonder whether it is a satisfactory response to say only that I have considered the options and this is the way I have chosen . . . It is one thing to say that [others] ought not to stop you from wasting your life if you choose to; but it is another to say that wasting your life in your own way is good just because it is your way. (2005: 14)

Appiah’s tale illustrates that, assuming we have some choice in the way that we live, it is not a carte blanche justification to say ‘I live this way because it’s how I see myself as meant to live’. Our actions are not performed in a bubble—someone who wastes their education and talent harms themselves and others, even if indirectly.[22] And, to return to the original example, no matter how much someone self-identifies as a white nationalist, it would be better for them to not express this identity with their words and actions. More generally, the externalization involved in having an agential identity that is associated with a malicious group is harmful and, insofar as that harm is under our control, it is therefore subject to moral evaluation.

Tefft’s case also illustrates why we reject the second thought—that we are not responsible for our agential identities because we have no say in what those identities are. We admit that (at least in many cases) persons might have no control over their self-identifications. But, because we hold to the idea that persons exercise agency over what they do and say, as well as the social positions they aim to occupy, we are committed to the idea that we exercise agency, even if limited, in the development of agential identities. And this is just because, on our model, agential identities are formed when an agent externalizes self-identification.[23] Tefft is responsible for his white nationalist agential identity. His demand for a respected corresponding social position—instead of one that acknowledges the dangerous, racist nature of his group’s ideology—should not receive social uptake.[24] So, while the moral evaluation of malicious self-identities is a difficult and distinct question, we believe that having a malicious agential identity—an agential identity associated with a malicious group, the externalization of which causes undue harm—is subject to negative moral evaluation.

As in the case of agential identity discrimination, however, to deny uptake is to cause harm to the individual denied. But, not all cases of causing harm are wrongful. Causing harm may be morally neutral, as in the case of a lightning that damages one’s home, or even good, as in the case of giving someone a needed vaccination. So, the question becomes: is the harm of denying a malicious agential identity—the harm caused, in this case, by refusing to allow Tefft to occupy the social position of noble white nationalist—wrongful?[25] A full discussion this question is beyond the scope of this paper, and the details will depend on one’s normative ethical commitments. However, we take it that the central point is that any harm done to Tefft is vastly outweighed by the good of preventing uptake of Tefft’s agential identity as a noble white nationalist. Individuals exercise choice when it comes to adopting agential identities. Where such identities are malicious, causing undue harm in their externalization, we are under no obligation to facilitate that externalization through uptake of those identities.

2.3. Passing and the Ambiguity of ‘Is’

In addition to the structural explanations discussed above, our model of agential identity has the virtue of being able to contribute to an ongoing discussion of ‘passing’ in social ontology. Passing is standardly described as occurring when a person is taken—even consistently taken—to be part of a social group to which they do not belong.[26] Here are two high-profile examples:

  • Rachel Dolezal living as a black woman for years, despite having no black ancestry.
  • Caitlyn Jenner (prior to ‘coming out’) living as man, despite not being a man.

These particular examples are controversial, as is the possibility (and frequency) of passing. One version of this controversy centers on how to make sense of passing alongside various social constructionist views of social groups. If, for example, one holds that someone’s race is a matter of how they are socially positioned, then it seems that racial passing is a conceptual impossibility: to regularly ‘pass’ as, say, black, just is to be socially positioned as black. On a ‘social position’ view of blackness, then, to pass as black is to be black. Contrast this with a common view of gender on which one’s gender is a matter of self-identification. On this view, someone who was assigned female at birth, self-identifies as a man, but makes no attempt to externalize this self-identity is perpetually in a state of passing.

We think that both views capture something true and important about passing, but nevertheless give deficient pictures of passing. In particular, we think this is because both rely on the standard description of passing (seen above), which describes passing as if it were single phenomenon. In fact, though, we think this description is ambiguous, and that removing this ambiguity reveals multiple forms of passing. In particular, the description does not distinguish between three importantly different (though potentially overlapping) kinds of passing.

When we say that someone is a man, ‘is’ might refer to one of several kinds: group membership, social position, self-identity, or agential identity. By disambiguating these possibilities, we discover that passing occurs in three forms:

Group Passing: An individual group passes when they occupy a social position associated with a social group to which they do not belong.

Self-Identity Passing: An individual self-identity passes when they occupy a social position associated with a self-identity that they do not have.

Agential Identity Passing: An individual agential identity passes when they occupy a social position associated with an agential identity that they do not have.

To see how these relate, consider the case of NPR’s Nancy Updike (Glass & Updike 2004). For two years, Updike sincerely adopted the agential identity of being a lesbian despite, in fact, not being sexually attracted to women. While she identified as a lesbian, she was group passing as a lesbian to those who knew and accepted that identity because, as she explained, she was wrong about belonging to that social group. But, she was not identity passing as a lesbian at that time, in either the self- or agential sense, because her agential identity really was that of a lesbian.

While she does not describe such a situation, Updike might also have group passed and agential identity passed while not self-identity passing. This could have occurred while she self-identified as a lesbian but had not yet come out, so that she lacked the agential identity of being a lesbian. Were her friends or colleagues to nevertheless identify her as a lesbian, she might thereby come to occupy a social position associated with being a lesbian despite lacking that agential identity. In this case, then, she would group and agential-identity pass as a lesbian, because she would occupy a social position associated with a group to which she did not belong or agentially identify. However, she would not thereby self-identity pass, because she did self-identify as a member of this group.

When Updike came to realize that she was not gay, the situation changed. In contexts in which she still occupied the social position of being a lesbian, she passed as a lesbian in all of the ways we’ve identified. First, she group passed in a context like this for the same reason she did in the other cases—she was not, in fact, a lesbian. Second, she self-identity passed because she no longer self-identified as a lesbian. Finally, she agential identity passed in this context because, once she self-identified as straight, she could not have the agential identity of being a lesbian. Without that agential identity, occupying the associated social position—continuing to be taken as a lesbian—meant that she agential identity passed as a lesbian.

In each case, passing is a matter of mismatch between the social position (of the kind described in §1.1) one occupies and some other aspect of one’s identity. What determines the kind of passing that occurs is whether that mismatch relates to the individual’s social group, their self-identity, or their agential identity. That said, group passing will not be possible with respect to cases in which occupying a social position is sufficient for social group membership (e.g., celebrity). But if social position comes apart from group membership conditions, all three forms of passing apply. Moreover, in many cases, these forms of passing occur together. In the previously mentioned case of John Howard Griffin, for example, it is plausible that Griffin ‘passed’ as black in all three senses.

There are two important observations to be made about the relationships between these forms of passing. First, we saw in the discussion of Updike that agential identity passing does not presuppose self-identity passing. But the converse is not true: self-identity passing does presuppose agential identity passing. If one occupies a social position associated with a self-identity that they do not have, then—because agential identity requires self-identity—they also occupy a position associated with an agential identity that they do not have. A closeted gay man, for example, may make successful efforts to pass as straight, but he does not have an agential identity as straight because he does not self-identify as straight. This passing, then is both a matter of self-identity passing and agential identity passing (as well as group passing). In this case, he convincingly fakes the agential identity of being straight, but does not have this agential identity because he lacks the self-identity. As a result, he both agential and self-identity passes. We can compare this with a case where a non-closeted gay man—that is, one who agentially identifies as gay—is nevertheless treated as straight. Here too, all three types of passing occur, but in a way that explicitly denies him agency over his social perception and position.

Second, group passing is independent of either form of identity passing. It is important that a social ontology be able to describe each phenomenon. Consider, for example, Sean Cole, an American teenager who spoke with a British accent from the ages of fourteen to sixteen, and agentially identified as British despite having no British ancestry and having never been to Britain (Glass 2000). Cole was not a member of the group of British people, and did not occupy a corresponding social position: his family and friends knew that he was not British. Despite knowing facts about his ancestry, Cole thought “there has to be some way that [his being British] could be true.” For years, he insisted that he was British. It seems, then, that while Cole did not group pass as American (rather than British) because he really did belong to the social group of Americans, he did agential and self-identity pass as American because his agential and self-identity were British.

2.4. Code-Switching

We’ve seen that a theory of agential identity helps us understand varieties of passing, as well as important normative issues surrounding agential identities. In addition, we think this theory can help us make sense of a phenomenon commonly known as ‘code-switching’.

Code-switching is well recognized within linguistics, psychology, sociology, and other social sciences. Various glosses on how to best understand code-switching abound. Most often, it is described as something like “the ability to adapt one’s behavior as a response to a change in social context”, and do so in a way that allows the code-switcher to navigate multiple communities (Morton 2014: 259).

These definitions of code-switching are useful but do not, we think, get at the heart of the phenomenon. To see why, consider H. Samy Alim’s personal example of code-switching while abroad in Germany:

I land in Berlin . . . [and] once I’d found an Italian place, I walked up to the counter and was greeted in German by the cook, whom I assumed was Italian . . . I responded in English, ‘Are you still open?’ In response to my English, he looked at me and said in Italian-accented English, ‘Ah, American Latino. Español.’ To which I said, in a sigh of relief, ‘Si, si, español es major porque no hablo alemán yo.’ (2016: 40)

In this story, Alim describes the experience of code-switching using particular language, behavior, and even silence in order to be read by the cook as ‘American Latino’. He goes on to describe distinct modes of code-switching that enabled him to be interpreted as “‘Indian’, ‘Algerian’, ‘Mexican’, ‘Turkish’, ‘American Latino’, ‘Columbian’, ‘Arab’, and ‘Black’” within five days across three countries.

Alim’s case is but one example of how an individual, intentionally or automatically, can use linguistic, aesthetic, or behavioral signals in distinct contexts in order to communicate, emphasize, or downplay particular social identities. And it is this toggling between or within social identities that is, we think, at the heart of code-switching. Additional familiar examples of code-switching include wearing different clothing when teaching than when at a pub with friends, unthinkingly shifting into one’s childhood accent when home for the holidays, or intentionally avoiding profanity or cultural slang in professional contexts. Code-switching is, to put it lightly, an extremely pervasive phenomenon, and one that is especially salient to persons from disadvantaged backgrounds, multilingual persons, and persons who regularly move between stratified social groups.

We think that a theory of agential identity is useful in developing a general account of code-switching. We often choose carefully where to have particular agential identities (or the appearance of a particular agential identity). We also often manifest distinct agential identities (or their appearance) in select contexts without realizing it. This, we think, gets at the core phenomenon that is code-switching:

Code-Switching: The intentional or automatic revision to real or perceived agential identities as one moves between contexts.

This account explicitly encodes two important features of code-switching. First, a code-switcher may revise their agential identities intentionally or automatically. For example, a trans woman who is deliberately closeted at work, but not with her close friends, is someone who intentionally code-switches with respect to her gender agential identity. In contrast, someone who unknowingly picks up on the regional accent of their immediate geographic area is someone who automatically code-switches with respect to regional agential identity.

Second, depending on their self-identities, a code-switcher may switch between multiple genuine agential identities, between both genuine and apparent agential identities, or between multiple apparent agential identities. For example, someone who identifies as biracial may code-switch as they move between predominantly white and predominantly black communities, but manifest genuine agential identities in each context. This is because, while they select the contexts in which to externalizes their white and black self-identities, they do in fact have both self-identities.

In contrast, someone who self-identifies as Mexican, but who adjusts their clothing, language, and behavior in order to be read as Iranian in one context and as Columbian in another, moves between multiple apparent agential identities. Because they do not self-identify as either Iranian or Columbian, this code-switching manifests the appearance of, but not actual, Iranian and Columbian agential identities.

Finally, a trans woman who hides her trans identity at work, but not with friends, code-switches between genuine and merely apparent agential identities. While she does self-identify as a woman, she does not self-identify as a man. At work, then, where she is careful to be perceived as a man, she manifests the mere appearance of a male agential identity. When with friends, however, and externalizing her self-identity as a woman, she code-switches in order to manifest her genuine female agential identity.

The context-sensitivity of many agential identities is an important feature to add to the nuances of agential identity discussed in Section 1.3. Often, our agential identities vary not only in their independence, stickiness, and salience, but also in their pervasiveness. We frequently downplay or hide our agential identities in certain contexts, and even perform agential identities that we do not in fact have in order to cope with various social demands, norms, and expectations. For this reason, code-switching is particularly familiar to persons who belong to disadvantaged or otherwise marginalized social groups.[27]

This analysis of code-switching is one that we think is rich for further philosophical analysis. It is not only metaphysically interesting, but also incredibly important for understanding normatively loaded phenomenon such as cultural imperialism and hermeneutical injustice, as well as the social practices that reinforce systems of race-, gender-, and class-based oppression.[28]

3. Conclusion

In the above, we present a theory of agential identities as position-directed externalities of one’s self-identification with members of a social group. We argue this theory does important explanatory work with respect to discrimination against and moral attitudes toward persons’ identities, as well as the phenomenon of ‘passing’.


We would like to thank Elizabeth Anderson, Elizabeth Barnes, Liam Kofi Bright, Esa Diaz-Leon, Peter Finnochiaro, Zoe Johnson King, Ishani Maitra, Elizabeth Victor and many others for comments and conversation during the development of this paper.



1. See, e.g., Haslanger (2000), Barnes (2016).return to text

2. See, e.g., Jenkins (2018).return to text

3. Haslanger (2016: §6). While we have some disagreements with details of Haslanger’s account, we are in overall agreement that social positions are importantly connected to social practices: the conditions for occupying a social position and these positions’ features will depend on practices. It is also worth noting that we diverge from Haslanger in terminology, simplifying as much as possible for our purposes.return to text

4. We draw an analogy here with the notion of ‘genericity’ common in linguistics. Generics are fault-tolerant constructions like “birds fly”, for which the existence of individuals that do not meet the description does not falsify the description. We take it that many of the membership criteria for social groups are best understood as being fault tolerant in a similar way.return to text

5. On this point, we disagree with Haslanger, who requires regular participation in a social practice in order to have the corresponding social position, and who does not allow that one can occupy a social position simply by virtue of being perceived to participate in a social practice.return to text

6. It is worth noting that being assigned a social position does not require that the individual members of a social group possess sharply defined concepts or linguistic labels of that role. See, e.g., Haslanger (2012a: 312–314).return to text

7. We take it that social positions are context-sensitive: just because the young man is not recognized as gay at home does not mean he could not have the social position of being gay other social contexts. For more on the complexities and nuances of agential identities, see §1.3 and §2.4. A further complexity to this picture, noted by an anonymous reviewer, is that there is no single answer to how much uptake is needed for an agential identity to revise an individual’s social position. The efficacy of an agential identity will depend, we think, on the context, what identity is at issue, how the agential identity manifests, and the degree and amount of uptake. For example, in some cases, only select individuals may have the authority to determine someone’s identity-related social position. In other cases, gaining a new social position may require that the majority of persons in one’s context affirm the agential identity at issue.return to text

8. This should not be taken to suggest that one must be correct about one’s social group membership in order to self-identify as a member of the group. One can, for example, self-identify as Scottish owing to misinformation about their place of birth.return to text

9. As we’ll see below, whether she is black is a different question entirely.return to text

10. This condition does not require that manifesting social properties be intentional. One’s skin color, for example, is not manifested intentionally. Moreover, we take it that many manifested behaviors will be intentional, but not intentionally directed at a particular social position. Rather, they will be the result of internalized identities. In other words, (a) merely requires that such behaviors are non-accidentally related to a social group.return to text

11. These conditions are ‘cognitivist’ only to the extent that, in order to have a particular agential identity, someone must possess a blueprint for the social position that their externalized identity aims toward.return to text

12. Such views originate with Winnicott (1965).return to text

13. In this, we depart from Appiah (2005), who requires that social identity have a basis in socially salient and labelled “kinds of persons”.return to text

14. Note also that ‘woman’ is a fairly clear case in which the blueprint one has may come apart from the blueprint guiding the position they are assigned, though they are associated with the same group. In this case, the blueprint to which a particular woman is conforming is unlikely to be associated with the same blueprint that the editor for, say, Hustler Magazine associates with women.return to text

15. Without getting mired in debates surrounding knowledge how, it suffices to say that, insofar as knowledge-how and knowledge-that are independent, we should expect cases like this to be fairly common. (See Fantl 2014 for further discussion.)return to text

16. E.g., Iris Young, Ann Cudd, bell hooks, Sally Haslangerreturn to text

17. This, despite the fact that such therapies have significant long-term physiological effects (Basaria et al. 2002).return to text

18. Focusing on cases like Alena’s, the National Center for trans Equality and the National Gay and Lesbian Taskforce report that after receiving gender-confirming treatment, “78% of trans people had improved psychological functioning” and suicide rates for trans individuals drop from a range of 29% to 19% to 6% to 0.8% (GLAAD 2018).return to text

19. Depending on one’s view about the relationship between concepts and social kinds, we expect there will be varying views about how these shifts to our everyday concepts may or may not additionally change what the social kind in fact is.return to text

20. This example is drawn from an actual interview with white nationalist Richard Spencer.return to text

21. Respecting an identity, in this context, should not be understood as involving any endorsement of that identity. Rather, respecting an identity involves acknowledging the individual’s identification and contributing to the social uptake of that identity. Failing to respect an identity might involve simply refusing uptake of an identity or, in more explicit cases, assigning the individual a different identity in its place. In this case, that identity might be racist Nazi rather than noble white nationalist. We thank an anonymous referee for drawing this clarification to our attention.return to text

22. Nagel (1979: 5) picks out a similar point, observing that someone who “wastes his life in the cheerful pursuit of a method of communicating with asparagus” is not living well, despite being happy.return to text

23. This is not to suggest that the social practices surrounding certain roles do not exert substantial pressure on individuals to adopt agential identities corresponding to those roles.return to text

24. Tefft’s case shows that the veridicality of an agential identity is not sufficient for justifying social uptake. While we will not argue for this claim here, we tentatively also hold that veridicality also is not necessary for justifying social uptake. Social groups’ membership conditions may unjustly exclude certain persons, and in ways that could be ameliorated by giving uptake to (strictly speaking) non-veridical agential identities that push the boundaries of those groups. For further discussion of such cases, see discussion of ontological oppression in Dembroff (in press).return to text

25. It is important to distinguish this question, which is about responding to an agential identity, from the nearby question of allowing someone to have a malicious agential identity in the first place. We do not address the latter in this paper.return to text

26. Though see Silvermint (2018) for an interesting account of passing as “a permissible form of self-regarding complicity”.return to text

27. Thank you to an anonymous reviewer for bringing the fluidity of agential identity to our attention.return to text

28. We think there are also important relationships to explore between code-switching and passing (see §2.3), though considerations of space require that we leave this discussion to future work.return to text