Personal autonomy theorists seek to articulate the conditions under which one's life, decisions, actions, and values can meaningfully be said to be one's own. The “ownness” of one's life, globally speaking, and of one's decisions, actions, and values, however, is not indexed to the level of detachment one achieves from one's history, relationships, and sociopolitical context. Contemporary theorists recognize that we are thoroughly socialized and social beings who are in many ways dependent upon and depended upon by others. Contemporary theorists, that is, take seriously the intricate ways in which we are socially embedded.

Many theorists recognize not only that we are socially embedded, but also that the sociopolitical spheres in which we find ourselves are marked by subtle and not so subtle forms of inequality and oppression. The social realities of having a racialized identity in a racist context, and/or of being a woman in a sexist context, and/or of being queer in a heteronormative context, to give just a few examples, are rightly assumed relevant to considerations of personal autonomy. In what ways and to what extent being a member of a partially subordinated social identity is relevant to personal autonomy, however, is contentious.Footnote ¹

Attempts to account for the significance of oppression for personal autonomy can be divided, very broadly, into internalist and externalist frameworks. On internalist theories, the question of whether X's choices, decisions, motivations, and life are autonomous can be answered entirely on the basis of criteria internal to X.Footnote ² Externalist theories, while not discounting the importance of internal factors, hold that internal criteria are insufficient for determining autonomy; external factors are also relevant.

My interest here is in examining the senses in which membership in such partially subordinated social identities, identities that mark one for various forms of oppressive treatment, affect one's autonomy. I argue that internalist theories are insufficient for identifying the autonomy‐inhibiting aspects of being a member of a partially subordinated social identity and that at least one external criterion is required. Ultimately, I argue that one's global autonomy is narrowed when one is attributed a status incompatible with the assumption that one enjoys the freedom‐conditions necessary for autonomy. The externalist condition I propose is meant to articulate the sense in which aspects of some social identities constrain the autonomy of members of other social identities. I argue, more specifically, that the implicit social permissibility of subjecting members of socially subordinated identities to violent, or at least violating, forms of interference is incompatible with unconstrained global self‐determination.

I begin by articulating the general problem and the important motivating dilemma that internalist theorists are focused on addressing and suggest that despite significant differences among the various types of internalist theories, these theories share what I call “the damage model” of the relevance to autonomy of membership in partially subordinated social identities. On the damage model, sexist, racist, and heterosexist social environments affect one's autonomy by causing certain sorts of damage to members of partially subordinated social identities.

In the second section, I introduce a different sort of example for analysis, an example that generates a dilemma distinct from the one driving the debate among internalists. Briefly, recent literature on the theme of personal autonomy and oppression has been focused largely on resolving a particular dilemma about autonomy and women's agency: how are we to assess the autonomy of women who are motivated by values, have desires, and make decisions that coincide with particular features of sexist oppression? Although analyses directed at resolving this dilemma allow us to trace an important way in which oppressive social circumstances may affect autonomy, they do not exhaust the concerns for autonomy generated by subordinating social environments. I introduce examples that have to do not primarily with what an agent values or desires, but with problematic attributions of status that attach to one qua membership in certain social identities. In particular, I will focus on analyzing the significance to autonomy of living among others who are permitted, explicitly or implicitly, to be harassers, to engage in racial profiling, and to treat members of some oppressed social groups as expendable. I argue that an external criterion is needed if we want to identify the ways in which who some get to be constitute constraints on the autonomy of others.

In the third section, I articulate a positive argument for the claim that we should want to employ an externalist condition, and in the fourth section I draw on, but ultimately depart from, Marina Oshana's externalist account of autonomy in order to articulate a minimal externalist autonomy‐securing criterion.

Social Subordination and the Damage Model

Most discussions of the ways in which a person's autonomy is made vulnerable by membership in a partially subordinated social identity unfold from within an internalist perspective as part of an attempt to solve the socialization problem.Footnote ³ Contemporary autonomy theorists start from an acknowledgment that personal autonomy is not a matter of excavating or discovering a pre‐social self, untainted by socialization; we are all thoroughly socialized beings. Socialization per se, therefore, is not necessarily a threat to personal autonomy and in many ways enables one by providing the relationships, skills, and capacities necessary for autonomy. Socialization, however, even if understood as in some ways coercive, is not all the same. As Diana Tietjens Meyers explains:

Resolving the socialization problem, therefore, involves delineating the criteria by which we can distinguish benign or enabling socialization from harmful, autonomy‐inhibiting socialization.

The socialization problem has been especially vexing for feminists and has been complicated by the following dilemma. On the one hand, it has long been acknowledged that sexism works, in part, by conditioning women to desire and value the mechanisms of our own subordination; this acknowledgment raises suspicions about whether desires, preferences, and motivating values are autonomous when they coincide with or overlap with the set of desires, preferences, and motivating values that have typically been thought to perpetuate sexist inequality. On the other hand, feminists are also rightly wary of denying agency and a voice to women who claim that their desires and values are autonomous.

In response to this sort of dilemma, theorists attempt to articulate the conditions that some action, desire, motivating value, and so on must meet in order to count as autonomous. Internalist responses to this dilemma can be broadly divided into strongly substantivist,Footnote ⁴ weakly substantivist, and proceduralist positions. Internalist versions of strongly substantivist positions evaluate as necessarily nonautonomous certain values and desires because of their content, whereas proceduralist views are content‐neutral; any desire or value can be autonomous provided it passes an autonomy‐securing procedure. Weakly substantivist views, though not judging values and desires nonautonomous because of their content, hold that something more than a procedure is needed to decide the case, and this something more is something that must be true of the subject whose values and desires are being evaluated. My point here is not to enter into the debate among proponents of these different sorts of internalist views; instead I want to draw attention to a commonality that holds across the different views when articulating the ways that living in oppressive circumstances can undermine an agent's autonomy. In trying to understand how social subordination may diminish autonomy, they focus on the ways in which agents can be damaged by internalized oppression.

Natalie Stoljar and Susan Babbitt, for example, who represent strongly substantivist views, require agents to value certain things (autonomy itself, for example) and prohibit the valuation of other things (subordination, for example). Failures to value or to repudiate the right sorts of things are, for Stoljar and Babbitt, explicable in terms of the damage done to agents through oppressive socialization. Stoljar articulates the feminist intuition that “preferences influenced by oppressive norms of femininity cannot be autonomous” (Reference Stoljar, Mackenzie and StoljarStoljar 2000, 95). The reason they cannot be autonomous is because of their content (109). Subjects who have such preferences, Stoljar argues, are unable to see that they have internalized false norms. Their capacity to see these norms as false is what is lacking or damaged, on Stoljar's account (109). For Babbitt, autonomy requires that one satisfy one's objective interest in flourishing, but “[i]n situations involving ideological oppression, an individual may fail to possess preferences and desires that adequately reflect an interest in her own human flourishing because she has been beaten down by the circumstances of her situation” (Reference Babbitt, Alcoff and PotterBabbitt 1993, 246). Oppression, for Babbitt, can damage agents in ways that undermine their ability to value their own flourishing. “The effects of oppression” Babbitt explains, “may be such that people are psychologically damaged, possessing interests and desires that reflect their subservient social status” (246).

Most theorists, however, concede that desires, values, and actions, even when connected to gender stereotypes, expectations, and forms of oppression, have a more complex relation to autonomy than can be revealed simply by examining their content. These theories, however, share with strongly substantivist views a model in which oppressive circumstances are relevant to autonomy because of the damage they cause to agents. Specifically, agents can be damaged in the self‐reflexive attitudes required for autonomy, in the dispositions required for autonomy, and/or in the competencies required for autonomy.

Many theorists identify the possession of certain self‐referential attitudes as necessary for autonomy. Sonya Charles, for example, identifies damage to self‐worth, which makes a woman complicit in her own subordination, as the factor that is most important for considerations of whether some decision is autonomous or not (Reference CharlesCharles 2010). Catriona Mackenzie, as well as Joel Anderson and Axel Honneth, hold that oppressive socialization undermines one's autonomy competency by undermining the necessary self‐reflexive attitudes of self‐respect, self‐trust, and self‐esteem (Reference Anderson, Honneth, Christman and AndersonAnderson and Honneth 2005, 127–49; Reference Mackenzie, Mackenzie and StoljarMackenzie 2000, 124–50).Footnote ⁵

Paul Benson, in his weakly substantivist view, follows a similar trajectory; for Benson “autonomy's normative substance resides in agents’ attitudes toward their own authority to speak and answer for their decisions” (Reference Benson and Stacey TaylorBenson 2005a, 25). Agents’ attitudes toward this authority depend on having self‐referential attitudes of self‐worth and self‐confidence. Benson explains that internalized oppression may affect one's self‐worth and self‐confidence in a way that undermines an agent's ability to maintain the relevant sort of authority. He asks us, for example, to consider “someone who, on the basis of race, has systematically been treated as socially invisible, as lacking the dignity of a person and eligibility to participate in distinctly personal forms of relationship, such as friendship, or familial love” (Reference Benson, Christman and AndersonBenson 2005b, 111). On Benson's view, if agents in such a position internalize their social invisibility, this can “defeat agents’ capacities to take ownership of what they do” (112). Oppression, therefore, on Benson's view, impedes autonomy when it “defeats” the capacities agents need to take ownership of their actions.Footnote ⁶ Andrea Westlund holds a similar view; the difference between a woman who autonomously chooses a particular subordinate position and one who is nonautonomous in her choice depends, in part, on whether she has a disposition to hold herself answerable to others. Westlund explains that “autonomy in choice and action … relies (at least in part) on the disposition to hold oneself answerable to external critical perspectives on one's action‐guiding commitments” (Reference WestlundWestlund 2009, 28). Westlund does not specifically identify the ways in which living in oppressive circumstances may damage agents’ autonomy, but she does hold that “[r]elations of care are causal contributors to the developed capacity for autonomy” (36). It is plausible, therefore, to suppose that Westlund would accept that oppressive circumstances that deprive agents of sufficient care could impede the development of the disposition to hold themselves answerable in ways necessary for autonomous choice and action.

On Meyers's value‐utilizing, agential‐competence view, autonomy depends on the development and exercise of a set of autonomy competencies.Footnote ⁷ For Meyers, “autonomous people have well‐developed, well‐coordinated repertoires of agentic skills and can call on them routinely as they reflect on themselves and their lives and as they reach decisions about how best to go on” (Reference MeyersMeyers 2002, 19).Footnote ⁸ Oppressive circumstances can interfere with both the development and the exercise of these skills. As Meyers explains: “this view of autonomy acknowledges the institutionalization of male dominance and the gravity of internalized oppression, both of which impede women's ability to develop and exercise these skills” (21). Oppressive circumstances, therefore, are relevant to autonomy insofar as they damage agents by impeding the development and exercise of these skills.

Finally, strictly proceduralist views also adhere to a damage model. John Christman, for example, claims that the test for whether autonomy obtains with respect to a particular characteristic C depends on both competency requirements and authenticity requirements. With respect to the basic requirements of competence, Christman specifies the following two conditions:

As with the other internalist views, the autonomy of C could be affected by oppressive social environments if those environments undermine an agent's ability to meet either of these two conditions.

With respect to authenticity, Christman specifies a nonalienation condition. Specifically, C is autonomous if it meets the following conditions:

A further way, then, in which being a member of a partially subordinated social identity may affect autonomy on Christman's view concerns the concept of alienation. An agent may meet conditions 1, 2, 3, and 5, yet be alienated from the characteristic in question. She may, if she were to engage in the specified type of reflection, realize that she came to have this trait due to the internalization of oppressive norms, and this realization may (although it also may not) cause her to be alienated from the trait.

The predominant model, therefore, for dealing with the ways in which membership in partially subordinated social identities is relevant to autonomy on internalist theories is a damage model: a model that identifies the damage done to agents as a result of internalized oppression. Oppression may damage or impede the development of the self‐reflective attitudes, skills, capacities, and dispositions that are necessary for autonomy and thereby have a negative effect on an oppressed subject's autonomy.

Beyond the Damage Model

Because theorists are engaged largely with the task of distinguishing benign or enabling socialization from oppressive socialization, the examples employed in the literature tend to be those that have the following structure: a) person X thinks she is autonomous with respect to Y, but b) there are reasons to suspect that X is not autonomous with respect to Y, because c) Y is somehow problematically connected to the internalization of oppressive norms.Footnote ⁹ Or, to cast the point from the other side of the dilemma: a) some Z thinks that X is not autonomous with respect to Y, b) because Y is coincidental with some serious aspect of gender oppression, but c) Z is wrong in her assessment. Common examples of Y include decisions and life‐plans that reflect oppressive gendered norms: norms such as the expectation that women will be subservient or submissive, that women's worth or value is determined by physical appearance, or that women ought not to claim sexual agency. The animating worry in these examples concerns the coincidence of women's plans, decisions, and values, with gender norms that perpetuate women's subordination.

Internalist views deal in nuanced ways with the fact that living in oppressive circumstances can, but need not always, damage agents in ways relevant to their autonomy. The criteria offered on the various theories provide essential tools for articulating why one woman who values and is motivated by a certain subordinating value may be autonomous while another may not. The values of the former, for example, will pass the relevant procedural tests and the agent herself will have the relevant set of self‐referential attitudes and skills.

Although I think that internalists are right to look to internal factors when considering the primary motivating dilemma concerning choices that coincide with oppressive norms, identifying the ways that autonomy is affected by oppression nevertheless requires the articulation of an external criterion. To illustrate where the need falls, I want to consider an example that has a different structure from those that drive the dilemma that motivates the internalist debates.

In explicating her externalist position, Oshana offers an example that is fruitful to elaborate and engage further. Oshana argues for an externalist position, a position that takes the external requirements for autonomy to be the presence of appropriate social relations. The appropriate social relations that autonomous persons enjoy are those that “enable them to direct their lives with a minimum of interference” (Reference OshanaOshana 2006, 88). Oshana argues that “[i]nterferences threaten autonomy when they relegate persons to a position whereby, in order to live in a self‐managed, self‐directed fashion, persons must resist the interference, or at least resist the temptation to regard the interference as normal and legitimate, even as they adapt to its presence” (88).

To illustrate this problematic kind of interference, Oshana asks us to consider the adaptation and vigilance “found in response to the phenomenon of racial profiling known as ‘Driving While Black.’ Statistically, African‐American men (and darker complexioned men, more generally) are subject to arbitrary stop‐and‐searches by law enforcement officials to a disproportionate extent, frequently while driving to work” (88). Oshana explains: “the issue is that these persons are socially positioned such that they must adapt to being a person subject to racial profiling, by resistance, or by cunning, or by ingratiation” (90).

I will return in the final section of this paper to the significance of the claim that agents must adapt, but I want, for now, to consider how we should articulate to what one is adapting. Persons who are subject to racial profiling must adapt to the fact that they are socially positioned such that they have a certain status: the status of being preemptively assumed criminal. Such a person has this status precisely because other persons, in this case law‐enforcement officials, are permitted and/or encouraged to have the trait of being racial profilers.

Members of partially subordinated social identities often live in a context where they have to occupy a status as a result of who other people, in virtue of their social identities, are permitted to be. For example, black men and trans women of color live in a social context where they are socially positioned such that they have the status of being expendable. The unconscionable rates of both incarceration and murder that members of these social identities face reveal the presence of this status. Members of these social identities have this status precisely because other persons are permitted and/or encouraged to have traits that position these subjects as expendable. Law‐enforcement officials are permitted to engage in racial profiling, and cis‐gender men, for example, are permitted violently to enforce a naturalized gender binary. Women and queer subjects, to give another example, live in a social context where they are socially positioned such that they have the status of being harassable. Members of these social identities have this status precisely because other persons are permitted and/or encouraged to be harassers.

An externalist criterion is needed in order fully to articulate how being positioned as preemptively criminal, expendable, and harassable because of who members of other, super‐ordinate identities get to be constrain one's autonomy. Specifically, an externalist criterion is required in order to capture how this positioning in terms of status itself, and not just that one must react or how one reacts to this positioning, matters to autonomy. Internalist theories, though they can account for the ways in which such social positioning may damage agents in ways relevant to their autonomy, cannot say anything about agents who are undamaged yet nevertheless think their autonomy is constrained.

Consider what internalist theories can say about living in a social context where one is positioned as harassable, expendable, or criminal. One straightforward manner in which internalist theories could account for the way such attributions of status narrow one's autonomy would be to construe living in such a social context as conferring certain traits or characteristics, the traits of being harassable, expendable, and criminal, on the agent whose autonomy is being analyzed. If we cast the social situation in this way, then internalist theories could straightforwardly claim that such traits impinge upon autonomy to the extent that the agent who has those traits is alienated from them. Specifically, Christman suggests “that the proper test for the acceptability of the characteristic in question is one where the person does not feel deeply alienated from it upon critical reflection” (Reference ChristmanChristman 2009, 143). One could claim, therefore, that such designations of status, when construed as characteristics of an agent, are autonomy‐inhibiting whenever an agent is alienated from those characteristics.

I want to suggest, though, that there are good reasons to resist casting the point in this way. Rather than suggesting that in certain oppressive contexts the oppressed have the characteristics of being expendable, criminal, and harassable, we should, rather, cast the point in terms of characteristics and traits that other people have. Rather than claiming that oppressed agents are harassable, for example, we should instead claim that some members of society are permitted to be harassers. The trait or characteristic that needs analyzing is not in the oppressed subject but rather attaches to members of superordinate identities. This way of casting the point is important because it resists the problematic idea that there is something about oppressed subjects that explains their status. Being attributed the status of being harassable or expendable or criminal has nothing to do with the person to whom such a status is attributed and everything to do with the fact that members of other social identities are encouraged or permitted to have characteristics that enforce such status attributions.

If I'm right that we should resist suggesting that these status attributions constitute traits or characteristics of oppressed persons that can be analyzed in terms of alienation, then we should recast our question as: what can internalist theories say about the autonomy‐reducing features of living in a social context where other people are allowed or encouraged to be harassers, to engage in racial profiling, and to treat others as expendable? Since, on internalist views, we are confined to analyzing the internal states of agents, what we can analyze is how an individual responds to her social context: what she does, what desires she has, what her governing values are in light of these facts about her status.

Internalist theories enable us to claim that her autonomy is compromised when she is alienated from the traits she develops in response to her social position or when one or more of her self‐regarding attitudes or autonomy competencies are damaged by the internalization of oppressive norms. What internalist theories cannot capture, however, are the ways in which someone's autonomy is narrowed by the fact that others are allowed to have traits that position her as harassable, expendable, and criminal, in cases where she is neither alienated from her response to these facts nor damaged in ways that affect her self‐regarding attitudes or autonomy competencies.

For example, suppose a subject who is deeply embedded in caring and resistant communities such that she enjoys the kinds of support that will enable her to develop the necessary self‐reflexive attitudes, dispositions, skills, and competencies. This person is fortunate enough to enjoy the kinds of community that can stop one from internalizing messages about one's expendability, harassability, or presumed criminality such that her self‐regarding attitudes and autonomy competencies are undamaged by the oppressive context.

Moreover, suppose that a general governing value in this person's life is a commitment to being able to read her social situation and respond practically and with a range of affect to the reality of the various contexts in which she finds herself. In all spheres of her life, she is committed to taking appropriate action in response to what she understands of the world around her, and she is committed to maintaining a complex emotional life that does not ignore the circumstances in which she finds herself. Under the rubric of this governing value, she reacts to the fact that others are permitted and/or encouraged to be harassers by: i) engaging with others in social protests, education projects, and an assortment of resistant practices aimed at changing this feature of her social world, and by cultivating a healthy sense of outrage about this aspect of her social position, and ii) she employs a number of different strategies in her daily life depending on the context and what she's doing at the time. Sometimes she avoids situations where she suspects she will be harassed (she crosses the road rather than walk past a raucous and intoxicated group of men), sometimes she aggressively confronts those engaged in harassing behavior either of her or of other people, and sometimes, in the face of harassment, she tries as quickly as possible to forget it so she can return to daydreaming or planning her day. Suppose, also, that her governing value and the particular ways she instantiates this value in her daily life pass the relevant proceduralist test; she is satisfied with how she is governing her life.

On internalist theories, this person is autonomous. Since she has managed to escape the potentially damaging effects of having a social identity that positions her as harassable, and her governing values and the way she instantiates them pass the relevant procedures, there is nothing more to say about the ways her social context affects her autonomy.

There is, however, something more to say about cases like this. Specifically, what strictly internalist views miss about the ways oppression can narrow autonomy is the aspect of oppression that is constraining rather than damaging. In the next section, I support the claim that we should want to say that an agent's autonomy is narrowed even in cases like the above, but first let me briefly clarify what I'm not claiming. First, I am not claiming that internalists are wrong to claim that this agent's values, desires, and actions are autonomous. In order to articulate the sense in which her autonomy is constrained, we will need to look at her autonomy as a person rather than at the autonomy of particular desires, values, and actions. Second, I am not claiming, simply, that she is nonautonomous because her strategies of prudential rationality are adaptations to oppressive features of her social environment. My critique is not, that is, best cast in terms of the debate about whether adaptive preferences can be autonomous.Footnote ¹⁰ I would agree with those who judge this agent's adaptive preferences to be autonomous. What I am claiming is that even if she is autonomous with respect to her motivating values, her choices, and so on, there is still some significant sense in which she is not fully autonomous. There is a sense in which certain aspects of the way it is permissible to treat her in the world put a constraint on the extent to which we can say that her life is her own.

An Alternative Dilemma

Many internalists rightly worry about placing external constraints on autonomy because such constraints may obscure the agency of oppressed peoples.Footnote ¹¹ I, however, want to articulate a positive reason for introducing an external criterion. The dilemma internalist accounts of the role of oppression in limiting autonomy seek to resolve concerns the possibility of a coincidence between adopting oppressive and subordinating values and being autonomous. Taking care to articulate the conditions of this possibility is important so as not preemptively to silence what members of oppressed social identities have to say about their own lives. There is, however, another angle from which to consider the question of agency and autonomy. Just as we don't want to silence and disregard women who claim that they are autonomous despite making choices that coincide with subordinating values and desires, so too we don't want to silence those subjects who claim that their autonomy is constrained by who other people in their social context get to be.

Internalist theories can support a particular kind of story about such cases; specifically, they provide the tools needed for agents to examine their responses to who other people in their social context get to be. They can provide a language for articulating the problem in cases where who other people get to be damages agents’ autonomy competencies, dispositions, and self‐reflexive attitudes or causes agents to be alienated from some characteristic they have developed in response to their social position. What internalist theories cannot do, however, is provide conceptual space for articulating the conjunction of internally undamaged and not alienated from one's traits yet nevertheless lacking in full autonomy. We should take seriously those who would claim that the fact that other people are permitted and/or encouraged to be harassers, to engage in racial profiling, and to treat them as expendable narrows their autonomy without requiring that they articulate this narrowing in terms of internal damage or alienation from their own responses.

We should preserve conceptual space to accommodate claims of narrowed autonomy apart from damage and alienation mainly with a view to the kind of causal story we can tell about how oppression affects autonomy. Strictly internalist views of the autonomy‐impinging effects of the kinds of oppression I am considering necessarily implicate the agent in the causal story of her diminished autonomy. The causal story of the relevance of oppression for autonomy is one in which oppressive social circumstances cause damage to an agent by impeding the development or the exercise of the skills and capacities needed for autonomy, by damaging the self‐referential attitudes needed for autonomy, or by causing her to be alienated from some aspect of who she is. Although external oppressive social circumstances play a role in explaining such an agent's diminished autonomy, these social relations are not the immediate cause of her diminished autonomy. Instead X the agent is the immediate cause of her diminished autonomy; it is something about her, about her competencies or her self‐referential attitudes, for example, that explains why she is not autonomous. She is, in every instance, implicated in the causal story of her diminished autonomy. What is problematic about this kind of story is that there is no space to claim that it is only the agent's social context and not also the agent herself that causally contributes to her diminished autonomy. The causal story of someone's diminished autonomy will often involve an immediate cause in the agent, but we should preserve conceptual space for cases where the immediate and exclusive cause of an agent's narrowed autonomy is explicable in terms of her social context.

An Externalist Condition

In order to capture more fully the ways in which members of subordinated social identities find their autonomy threatened, it is necessary to consider how living in an oppressive social context can constitute agents as constrained rather than damaged. I suggest that, in addition to considerations internal to the agent, the articulation of at least one external condition is required in order to capture more fully the ways in which oppression is relevant to autonomy. I propose that members of social identity X find their global autonomy constrained when they are positioned in social systems of power such that some aspect of another social identity Y conflicts with the assumption that members of social identity X enjoy a practical right to self‐determination. In the specific cases I consider, cases where others are permitted to harass, engage in racial profiling, and treat as expendable, one's autonomy is practically constrained because one is positioned as someone with the status of being appropriately subjected to violent or violating interference; this status, I suggest, is in conflict with a freedom condition necessary for unconstrained global self‐determination.

Let me turn to Oshana's externalist criterion in order to clarify the external criterion I wish to articulate. Although Oshana does not undervalue or ignore internal conditions for autonomy, she claims that “[a]utonomous persons have de facto power and authority over the management of their lives within a framework of standards they set for themselves and their autonomy is partly constituted by exogenous phenomena—that is, by relations extrinsic to their psychology” (Reference OshanaOshana 2006, 87–88). These extrinsic features identify appropriate social relations where “appropriate” picks out social relations that “enable them to direct their lives with a minimum of interference” (88). Oshana argues that “[i]nterferences threaten autonomy when they relegate persons to a position whereby, in order to live in a self‐managed, self‐directed fashion, persons must resist the interference, or at least resist the temptation to regard the interference as normal and legitimate, even as they adapt to its presence” (88).

For Oshana, therefore, “[w]hat is at issue in the case of racial profiling and similar states of affairs is not simply the frequency or the predictability of interferences, nor that a person's self‐determination is abridged in what some regard as relatively minor ways, nor that only certain persons suffer such abridgements to a disproportionate extent. Rather, the issue is that these persons are socially positioned such that they must adapt to being a person subject to racial profiling, by resistance, or by cunning, or by ingratiation” (90). The problem, from Oshana's perspective, is that in order to live in a self‐directed way, a person subject to racial profiling must adapt to an inappropriate sort of interference. On Oshana's view, the issue is one of being problematically constrained with an imperative to do something.

I agree with Oshana's claim that being constrained by an imperative to do something because of one's social position narrows one's autonomy, but I want to suggest that the problem runs deeper than that. The problem is that who some people get to be because of their membership in particular superordinated identities puts a practical constraint on other people's autonomy. Insofar as, for example, men and cis‐gender straight people can be harassers of women and trans subjects, women and trans subjects have a social status that constrains their autonomy.

In order to explicate the nature of this constraint, I take up Mackenzie's suggestion that rather than viewing autonomy as a unitary concept, we should understand “that the concept of autonomy involves three distinct but causally interdependent dimensions or axes: self‐determination, self‐governance, and self‐authorization” (Reference Mackenzie, Veltman and PiperMackenzie 2014, 17). It is the distinction between self‐determination and self‐governance that is useful for articulating the scope and the content of my externalist condition. Mackenzie explains that self‐determination “involves having the freedom and opportunities to make and enact choices of practical import to one's life, that is, choices about what to value, who to be, and what to do” (17). Mackenzie explains that satisfying the freedom and opportunity conditions for self‐determination will involve identifying the “external, structural conditions for individual autonomy” (17). The second axis, self‐governance, “identifies internal conditions for autonomy, specifically competence and authenticity conditions” (18) and “involves having the skills and capacities necessary to make choices and enact decisions that express, or cohere with, one's reflectively constituted diachronic practical identity” (17). The advantage of distinguishing these two axes, as Mackenzie argues, is that one can identify external conditions that must be in place in order for one to be self‐determining, without thereby implying that agents who live in contexts where those conditions are not satisfied lack self‐governance (23–24). It leaves open, in other words, space to articulate certain kinds of threats to autonomy that come from living in oppressive circumstances without requiring that agents be internally damaged in the competencies, skills, and self‐reflective attitudes necessary for autonomy.

The content of the externalist condition I propose is best explicated as one of the freedom conditions required for global self‐determination. While particular instances of harassment, for example, undermine one's episodic self‐determination, living among those empowered to harass constrains one's global self‐determination. Specifically, I suggest that one of the freedom conditions necessary for global self‐determination is freedom from living among those whose identities include aspects that systematically and pervasively position one as someone for whom interference is appropriate/tolerated. If members of social identity A are implicitly or explicitly permitted to harass, subject to racial profiling, and/or treat as expendable members of social identity B, this means that members of social identity A are empowered to interfere with members of social identity B. If members of social identity A are empowered to interfere with members of social identity B, then members of social identity B are socially positioned as people for whom interference is appropriate/tolerated. And, being socially positioned as someone for whom interference is appropriate/tolerated is incompatible with enjoying the practical recognition of someone with the right to be self‐determining. Or, in other words, a real and practical right to self‐determination is conceptually incompatible with occupying a status that positions one as someone for whom interference is appropriate/tolerated.

Although internalist theories provide important tools for analyzing many aspects of the significance of oppressive social circumstances for autonomy, they do not adequately capture the problems that arise for autonomy from the traits and behaviors members of superordinated identities are permitted. Internalist theories are incapable of capturing this because no amount of work on yourself, no amount of making sure your desires and values are authentic, and so on, can change who others get to be, and sometimes who others get to be is incompatible with a freedom‐condition necessary for global self‐determination. My externalist criterion will, I hope, shift attention away from an exclusive focus on the inner lives of members of oppressed identities and onto the problematic traits that members of superordinated identities are permitted to have.