I. Disagreement and The Problem Of Autonomy Ascription

Autonomy ascription is about attributing autonomous status to agents, whatever that might entail. The attribution might involve calling someone autonomous, or acknowledging that someone is an autonomous person in routine interaction. The attribution of autonomy would involve a kind of respect for that person as an autonomous agent: as an individual with the authority to make their own choices and to speak on behalf of their own preferences and decisions. In this section, I will take for granted—for the most part—that ascribing autonomy to agents is a good thing. I then provide examples of potential disagreement over the ascription of autonomy and outline why this disagreement should be treated as a problem that warrants philosophical attention. I will then claim that dialogical exchange is a key process that should be instantiated to alleviate disagreement over autonomy ascription.

First, let me briefly outline why it is valuable to positively ascribe autonomy to agents. In my view, the process of autonomy ascription itself generates novel opportunities for agents’ autonomy to be enhanced or undermined. This holds independently of the theoretical conditions purported to describe autonomy. For instance, in medico-legal contexts, being ascribed with patient autonomy is not only a declaration of an agent's decision-making competence. Ascribing autonomy to the patient normatively grants that agent particular ethical treatments. For example, the status of patient autonomy legitimates the call to treat that agent as having the authority to make their own medical decisions without undue interference; autonomous status also calls for that agent's decisions to be respected and upheld regardless of how imprudent or irrational they may appear to outsiders. Ascribing autonomy to an agent thereby confers respect for that agent's decision-making authority. Assuming we have no countervailing reasons to consider otherwise, having autonomy ascribed in such fashion seems to have a positive ethical value. This itself appears quite intuitively to enhance that agent's autonomy. If we can at least accept that certain treatments follow or are congruent with being called autonomous, we should find the positive ethical value—and the high stakes—of autonomy ascription plausible.

Given that we are taking the ascription of autonomy to be ethically valuable, we are then naturally led to consider the question of when and how we might appropriately ascribe autonomy to agents. At first blush, we seem to already have a straightforward answer. We might point out that autonomy theories already do the work for us in terms of outlining who should be described as autonomous or nonautonomous. After all, is this not what the various alleged conditions of autonomy are meant to tell us? It seems almost redundant to talk about autonomy ascription provided that our autonomy conditions are correct—we would simply say that we should ascribe autonomous status to whoever fits the relevant description of autonomy, and refrain from autonomy ascription in cases where the agent does not fit the relevant description of autonomy.

Be that as it may, we should take seriously the potential disagreement that might arise between first-personal experiences of autonomy and third-party descriptions of autonomy, especially in the case that an agent wants to claim that they are autonomous against determinations to the contrary. This is the problem of autonomy ascription. Of course, people might contest all sorts of ascriptions placed on them; this would be nothing new.Footnote ² But autonomy is one of various ascriptive concepts that have particularly high stakes, given its multiple functional values (as a reason to find someone respect-worthy, as a right-claim, an identity, and so on) and its role as an emancipatory heuristic in our context of feminist philosophy. Being restrictive about autonomy ascriptions could in this sense be ethically devastating for the individual, and it seems to me sensible to set the burden of proof rather high if we wish to count someone as lacking in autonomy. Thus, the question of when it is proper to ascribe autonomy—even in view of plausible proposals about what autonomy is—remains open. Furthermore, how we should in practice ascribe autonomy in light of efforts to adjudicate the kind of disagreement mentioned earlier is uncertain—and this uncertainty is what I wish to redress for the remainder of my paper.

Recall Stoljar's normative competence view of autonomy from my earlier survey of autonomy theories. On Stoljar's view, agents must be able to critically exclude false and oppressive norms from their decision-making inputs. Now suppose that a woman, Jane, counts as nonautonomous on Stoljar's account of autonomy because she fails to exercise normative competence—she does not exercise critical exclusion of gendered oppressive norms in her choice-making. Jane appears, rather, to contentedly conform to stereotypes about feminine domesticity and to the desire to be a good “housewife.” She would, however, like to think herself perfectly autonomous. She claims something to the effect that her personal endorsement of domesticity, and her lack of interest in problematizing her traditionally feminized role as housewife, should be proof enough that her preferences are what she “really” wants. She would both detest and protest the idea that her preferences are reducible to some textbook instance of internalized oppression. She feels, as a matter of fact, alienated from variants of feminism that seem to her unreasonably intent on insisting that she is nonautonomous.

Although this example is fictional, it is a plausible enough, commonplace conflict that might arise for autonomy ascription in practice—especially in our feminist context of autonomy. We could frame such disagreement in several ways. One response we might have is to say that this is a trivial problem—a mere dispute over which accounts of autonomy individuals personally find plausible as representative of their circumstances. We could recognize that Jane really just expresses a proceduralist view about her autonomy by insisting that she freely identifies as a “housewife.” But we could still maintain that there is a fact about her autonomy—or lack thereof—independent of her first-personal view. We might say that Jane is simply mistaken to believe that she can truly identify as a housewife in a way that escapes her problematically gendered socialization. If we assume Stoljar's normative competence account is true, Jane would still count as nonautonomous. We can say of any account of autonomy, really, that it should not require a personal endorsement from the agent it describes for it to still hold true.

In my view, however, it would be a mistake to say that autonomous status is entirely independent of an agent's first-personal perspective, at least given my interest in autonomy ascription as an ethically valuable thing to confer on an individual and the high burden of proof it would take not to confer it. In this sense, my view is that it is appropriate to maintain an agnostic position on the question of whether somebody's autonomous status could be described entirely independently of the agent's first-personal input; prioritizing the normative value of soliciting the agent's first-personal viewpoint as a resource used to ascribe autonomy is more important. Just as we might sometimes set aside age-old questions about whether something like free will really exists—in favor of the normatively desirable practical supposition that it does—it seems to me that drawing on an agent's first-personal perspective is an important step in a fair adjudication between the first-personal perspective and third-party autonomy descriptions.

What I am getting at is this: the agent's first-personal perspective—no matter what it holds—should at the very least be elicited as part of any process of autonomy ascription. Ascriptions of autonomy should be reconsidered or revised in earnest, in light of points of departure in disagreements between the first-personal perspective and the third-personal autonomy ascription. Addressing this issue of disagreement requires a kind of dialogical bridge: we should ask and consult agents about what they find constitutes reasons, motivations, and justifications for their decision-making, as a way to shape our practice of autonomy ascription in a way that is attentive to agents’ own perspectives.

Take an ongoing tension over the issue of autonomy in feminist debates about sex work, for example. Much of the debate on sex work is couched in terms of a tension between oppression and sex workers’ lack of autonomy. Though liberal feminists might argue that workers have a right to offer sexual services in exchange for money or other goods, because of bodily autonomy, radical feminists would say that the sex trade is emblematic of violence against women (Lux Reference Lux2009). Against the more damning assumptions, sex workers may “[resent] the assumption that their work was necessarily demeaning and never freely chosen” (Overall Reference Overall1992, 705). Part of the grievance here, I believe, rests on issues that are exacerbated by a lack of dialogical exchange. Sex workers may feel that their insight is neglected, and their autonomous status reduced by others, without appropriate consultation. The language of autonomy, then, entrenches conflict between feminists who want to criticize sex trade by supposing the latter undermines autonomy wholesale, and sex workers who feel that their work is misjudged, simplified, and looked down on by feminists who denigrate their work. This conflict doesn't seem like it can be satisfactorily dealt with merely by sticking to the purportedly feminist motivation to determine sex workers as nonautonomous. It is also not resolved by treating sex workers as though they are autonomous for merely practical, political purposes, without really listening to the narratives and justificatory schemes these sex workers offer for their own decisions. A fair process of adjudication when it comes to autonomy ascription must involve not merely paying lip service to those whose autonomy is up for question, but earnest efforts to make sense of the disagreement by listening to their first-personal narratives.

In conflicts of the kind described above, we can pick out features we could take as disrespectful toward the agents whose autonomy is in question. Several paternalistic attitudes are implied in the insistence that someone is nonautonomous against first-personal protests to the contrary: it indicates that the agent's perspective can be trumped by another, that the agent lacks access to self-knowledge, that the agent's own perspective is irrelevant for their autonomous status, and so on. Perhaps there are cases where such paternalism is justified—it doesn't seem particularly controversial to call a currently intoxicated person nonautonomous, for example—but this is not so clear in cases like that of autonomy and sex work. In such cases, it seems that denying an agent their stake in the disagreement over their autonomous status would be problematic from an ethical perspective, and a serious blow to the emancipatory goals of any autonomy theory. Indeed, it seems to go against the feminist spirit of autonomy we surveyed earlier. If a concept of autonomy should recognize the potential for agents to be shaped by the social environment around them in both empowering and disempowering ways, a practice of autonomy ascription wherein an agent's authority to answer for themselves is diminished or silenced should not be seen as particularly consonant with the liberatory potential of autonomy.

Furthermore, it seems that some of this paternalism is potentially epistemically vicious by way of evoking intellectual vices. Linda Zagzebski points out that examples of intellectual vices include “intellectual pride, negligence . . . rigidity, prejudice . . . closed-mindedness, insensitivity to detail, obtuseness, and lack of thoroughness” (Zagzebski Reference Zagzebski1996, 152). The problem with these intellectually vicious traits, according to Quassim Cassam, is that they impede responsible inquiry (Cassam Reference Cassam2016, 160). We can see how such vice might be evoked in the context of autonomy ascription: in maintaining that someone is nonautonomous without consulting the agent whose autonomous status is in question (as happens to be the case in the sex work example), we may end up misrepresenting the psychology of the agent by making false assumptions about the content and causes of their desires and preferences. From an epistemic perspective this is potentially negligent, insensitive, and closed-minded toward relevant and valuable first-personal contributions an agent may be able to make in autonomy ascription.

And if our assumptions about the agents whose autonomy we determine are indeed mistaken, all the worse for our practice of autonomy ascription that theories of autonomy do not offer a corrective to the gaps in our knowledge of the individual. Not consulting and addressing agents’ perspectives may leave us with a mistaken base of presumptions on which evaluations of their autonomy are made. As I have mentioned already, the threshold for the burden of proof is high if we want to determine someone is lacking in autonomy. That is not to say agents themselves can't be mistaken about their own preferences or that they can't be self-deceived in some way. Perhaps this is also possible. But the point is that it is bad practice to inquire into an agent's autonomy, and to make judgments about their autonomy, on the presumption that acquiring knowledge about their circumstances without also consulting the agents in question is possible in the first place. A nondialogical practice of autonomy ascription would be overly optimistic, on a charitable reading, but more plausibly, it would be epistemically vicious. Although agents may be mistaken about their own autonomy, it surely does not compare to the worry that outside others are prone to make similar mistakes against the agents’ first-personal perspectives.

So issues of intertwined ethical and epistemic interest occur as part of the problem of autonomy ascription. Not only is it ethically dubious practice to exercise the kinds of paternalistic attitudes regarding autonomy mentioned herein, it also seems bad for epistemically responsible inquiry to do so. As the sex work example shows us, individual narratives should have a meaningful place in autonomy ascription. This means agents whose autonomy is called into question should be consulted directly for their views and given sufficient opportunities to answer for their decisions.

Enacting opportunities for dialogical exchange as part of the autonomy ascribing process, then, could resolve the question of what actions we ought to take for a fairer practice of autonomy ascription. As we saw in the previous section, descriptive but nondialogical theories of autonomy by themselves often did not sufficiently resolve the potentially paternalistic ways that ascriptions of autonomy might be imposed on persons. We can mitigate the issue by asking agents directly what they think, how they conceptualize their own autonomy, what their reasons are for holding their preferences, what issues they see with being denied autonomous status, and so on. We can outline substantive questions to elicit meaningful and relevant answers that might figure in our autonomy ascription of the agent. Just as we try to talk to our friends about their choices to understand them better, to empathize with them, to become accepting of their decisions, and to form a picture of the motivations that underlie their choices, it seems entirely appropriate to do the same with agents whose autonomous status we treat as subject to evaluation and judgment.

By giving agents the opportunity to answer for their own schemes of justification, we can also combat the worry that autonomy ascription involves epistemically vicious attitudes. In place of assumptive attributions of an agent's views and circumstances, which presume knowledge over the agent, we give agents proper epistemic authority by enabling them to answer for their own insights, reasons, justificatory schemes, and so on. Working with agents’ first-personal perspectives seems naturally to exhibit a kind of intellectual humility about the limits of what we should presume to know about other people. Overall, then, good practices of ascribing autonomy should invoke some process of dialogical exchange.

II. Dialogical Answerability

I've made the general point that exercising dialogical exchange in autonomy ascription would involve consulting agents whose autonomous status is in question for their first-personal perspectives. In this section, I will fill in the details of what such an exchange would involve and how it might transform our practice of autonomy ascription. I draw from and apply Westlund's dialogical account of autonomy to show how it might help us make progress on the problems of autonomy ascription.

Westlund takes autonomy to involve a kind of answerability. She says that the kind of self-governance relevant for autonomy requires a disposition to “hold oneself answerable to external, critical perspectives on the commitments or policies that guide one's practical reasoning” (Westlund Reference Westlund2011, 170). That is, we have a responsibility to respond to second-personal external queries about our practical reasoning and supply answers to others regarding our decision-making policies. She clarifies that this disposition for answerability is “a feature of the agent's psychology, and thus internal to the agent” (Westlund Reference Westlund and L2012, 66). Furthermore, this disposition is not merely necessary for an agent's autonomy; the very exercise of holding oneself answerable to external challenges just is what it takes to be autonomous. It is important to note here that unlike my claim about the role of dialogical exchange in autonomy ascription, Westlund's claims are much stronger; she takes her view to be a constitutively relational account of autonomy, by which she means that it “holds that some social or relational factor(s) play an ineliminable role in the definition of autonomy itself” (62). Her view is constitutively relational, she says, because the kind of reflectiveness the agent is required to exercise is “itself dialogical in form” (65). Finally, she points out that this type of relationality “carries with it no specific value commitments” (Westlund Reference Westlund2009, 28)—what matters for autonomy is not so much what the agent endorses or commits to, but how they discursively justify and take ownership over it.

Part of Westlund's motivation for advancing such a view is to offer a “principled” way to distinguish cases in which an agent is to count as self-governing, from cases where an agent is merely “in the grips of a practical perspective” (Westlund Reference Westlund and L2012, 65). When we say things like “That's the depression talking, not you” to others, or criticize others for self-abnegation or excessive deference, we are signaling a kind of failure of responsiveness to “interpersonal justificatory demands” (65). In the case of someone with depression, for instance, Westlund says, “her reasoning on these matters seems likely to be strongly psychologically insulated from confrontation with contrary considerations” (Westlund Reference Westlund2009, 33). So Westlund believes it is appropriate to settle the question of whether an agent has agential authority by looking at the attitudes the agent has about their own perspectives. But to figure out whether the agent's perspectives are really their own, and not merely something the agent is resigned to, follows uncritically, adheres to by force of habit, or is a result of conditions such as depression, we need to ask whether they are responsive to interlocutory challenges. Hence, the disposition to hold oneself answerable to external, critical challenges “marks the relevant distinction between being gripped by and governing the practical reasoning that guides one's actions” (Westlund Reference Westlund and L2012, 65)

I won't comment here on whether I find Westlund's constitutively relational view sufficient as a complete measure of someone's autonomy—I believe we can bracket this issue and focus on whether the application of Westlund's framework can help us model good practices of autonomy ascription.

I'll draw on work by Milton C. Regan Jr. as an example. I find the following example congenial as a challenging case for autonomy ascription that might be usefully transformed by Westlund's dialogical framework. Regan discusses complications we face in ascribing autonomy when we reconstruct different narratives regarding abortion: “Consider women who accept traditional gender roles because of their understanding of the demands of biology or religion. . . . Research indicates that many such women believe that biology represents ‘a natural order that should really be allowed to prevail’” (Regan Reference Regan1997, 454). Now we imagine that for these women, the “natural order” underlies the belief that men and women are inherently different, with specific roles to play, and that reproductive processes are not something over which we should gain complete control. Because these women identify as pro-life activists, hesitation to interfere with the “natural” biological order is reinforced by religious beliefs; they might see it as hubris that people want to intervene with pregnancy just because it happened when it was not convenient for them.

What do we make of this? Are such women autonomous? We'd get different answers depending on the theory of autonomy we adopt. A proceduralist view would look at whether the decision to endorse a pro-life stance was internally consistent and noncoerced. A more substantive internalist view might be skeptical about whether the beliefs underlying these women's choices could be adopted autonomously. Other nondialogical theories may yet make differing claims regarding this matter. But these accounts would be ethically and epistemically limitedFootnote ³ in their ability to tell a satisfying story to justify autonomy ascriptions regarding these women, precisely because they do not enlist these women's first-personal perspectives.

Faye Ginsburg suggests that pro-life women do not “passively acquiesce” to biological difference, but rather engage in an active affirmation of their biological sex, which they regard as the basis for their female identity. She says pro-life women: “often describe the trajectory of their lives as a process of overcoming initial ambivalence about or even resistance to pregnancy, which culminates in the willing embrace of one's female nature. . . . For some women, early identification with liberal feminism contributed to reluctance to accept their role. Coming to accept that role thus required them to take a critical perspective toward an initial commitment” (Regan Reference Regan1997, 454).

We can see here why the discursive contributions of these women would be invaluable in the process of evaluating and attributing autonomy to them. These pro-life women do not necessarily see themselves as perpetuating a system of male privilege. Rather, they positively seek to “preserve an intricate set of social relationships that valorize women by promoting feminine values of nurture and responsibility . . .” (455). Moreover, the thought is that linking sex to pregnancy and marriage limits the tendency to regard women as sex objects and, serves as “the linchpin for the material security of women with dependents” (455). On this narrative, legal access to abortion is seen as a threat to this security by taking pressure off men to step up to their financial and emotional responsibilities for the outcomes they've brought about by sex. On this story it does not seem entirely appropriate to not credit or attribute at least a degree of autonomy to such pro-life women and respect them as such: it seems that the justificatory schemes these women can articulate are reasonable enough, even if not agreeable to some.

Setting aside the issue of whether we morally or politically sympathize with these views, what a dialogical account of autonomy can accommodate are the fascinating differences that emerge between women of various leanings on the matter of abortion. On a dialogical view like Westlund's, which requires the agent to answer for their practical reasoning in decision-making, we can acquire a rich and nuanced picture of the first-personal narrative background that informs a woman's leaning toward a pro-life stance. This may help us exercise open-mindedness in autonomy ascription, and to resist the more naïve picture of the pro-life woman as one who has uncritically internalized sexist norms about the proper role and duties of a female body (and whom we may be tempted to call nonautonomous).

What Westlund's account offers, in the context of the practical purpose of autonomy ascription, is the fair opportunity—which would go some way toward alleviating the ethical and epistemic problems that might surface in nondialogical autonomy ascription—for agents to manifest a kind of answerability to claims that are given due consideration in autonomy ascription. The dialogical view naturally encourages the agent to have a central and active role in their own autonomy ascription. We can see why this would be in practice respectful to the agent: no epistemically vicious attitudes are directed at the agent, as we exhibit open-mindedness and curiosity about their respective positions. Paternalistic attitudes toward the agent, too, are for the most part avoided by ensuring that they are at least given adequate opportunity to take ownership over their practical reasoning process.

III. Limitations of the Dialogical View in Autonomy Ascription

The previous sections have shown how an emphasis on first-personal perspectives and individual answerability could help enhance the process of attributing autonomous status to agents, especially when there is disagreement between them. Asking for, and engaging with, an agent's first-personal perspective can be a way to ascribe autonomy responsibly. It may help us circumvent the charge of paternalism when it comes to the process of judging whether someone is autonomous, since the agents themselves are consulted and given a stake in the matter. Moreover, we enact open-mindedness when we treat the first-personal perspective as a relevant resource that can inform and enrich our attributions of autonomy. But though this dialogical focus may help us navigate and model better practices of autonomy ascription, I do not endorse the account wholesale, as there are limitations we should keep in mind.

In this section, I will survey some of these limitations. They include the idealization of articulacy and the masking of problematic oppressive influences. We should be careful not to lose sight of these worries when we utilize the dialogical framework as a way to improve our practices of autonomy ascription.

The first worry I'll mention is the idealization of justificatory dialogue. Expecting agents to be dialogically responsive in the first place appears to be a demand that would exclude agents who are not inclined to exercise this skill. Westlund addresses a case of this with an example from Ingra Schellenberg:

Westlund argues here that Betty is not self-governing with respect to her decision, but is rather in the grip of the value of independence and opposition against authority figures. Betty failed to be “moved by the feeling that she owes a response to the counter-considerations raised by others” (38). Yet this seems to suggest that an agent cannot disengage from those holding her answerable. And the idea that one must positively value dialogical responsiveness is a strong claim. As Schellenberg mentions, “the socially vulnerable may fall disproportionately into the group of those who feel threatened by justificatory dialogue” (38). We would not want our attempts to include agents’ first-personal perspectives in autonomy ascription to be threatening, alienating, or unduly burdensome to the agent who is called on to answer. As such, my view is that it would be acceptable not to insist on the value of dialogical exchange insofar as the conditions of Westlund's account may sometimes generate this problem.

Westlund tries to get around this by saying that responsibility for the self does not in fact “rely on valuing any specific justificatory practice” (38), like a willingness to cite reasons on demand. She says that answerability can take a more open form, and insists that agents who “feel threatened by a more articulate interlocutor” could reflect in a more private way, so they wouldn't be “disempowered by an imbalance in sheer argumentative skill” (39).

She also says that we would be right to feel that we do not owe anyone answers under “illegitimate” circumstances. Dialogical answerability is not something that we are required to exercise for every single challenge leveled at us, but rather something we owe to “legitimate” challenges. She says a legitimate challenge must be situated “in a way that makes relational sense of the intervention” (39). A sense-giving relationship might be one where one is positioned as a member of the moral community, someone's neighbor, and so on. What this means is that it must be clear “why it matters to my critic why I think and act the way I do, and it must matter to her in a way that she can reasonably expect to matter to me” (40). This is a condition Westlund calls relational situatedness. This condition clarifies the fact that part of the “burden” of dialogical exchange falls also on the interlocutors who pose such challenges.

Yet the demand for reasons and answers may still place too much of a burden on agents who aren't inclined to engage in argument with others, even where the challenge is “legitimate.” This worry, Westlund says, gives rise to the condition of context-sensitivity. On this condition, a legitimately placed challenge must be context-sensitive with respect to the kind of response it invites and tolerates. Furthermore, Westlund says, there are alternative ways of demonstrating that “one holds oneself to appropriately situated critical challenges” (40). She lists some ways one might do this: “outside the realm of the conversational, an agent may give explicit or implicit signals that she intends to reflect on what has been said . . . or that she is attempting to repair, restructure, or terminate a relationship or practice that has come into question” (40).

Her conditions of relational situatedness and context-sensitivity can be considered as ways to combat the idealization of dialogical responsiveness in the context of autonomy ascription. We can recognize that dialogical exchange is not necessarily a demand that must be met all of the time, because not all challenges to an individual's autonomy should warrant “answers” from the agent.

Furthermore, Westlund recognizes that it is possible to exhibit answerability (and thereby autonomy, on her account) by countering external critique. Let us consider, for instance, the way Westlund contrasts between the case of the Deferential Wife (DW) and the Anti-Feminist (AF). The DW defers to her husband's preferences. When the DW is asked by a friend about why she agrees to move for her husband's job when it would be disruptive to her life, the DW “casts about for things to say” (Westlund Reference Westlund2003, 487) and finally says that she just wants her husband to “have what he wants; that's all” (487). For Westlund, this is a classic case of deep deference: the wife's answers show us that her endorsements have no nondeferential basis. She abnegates herself as a “distinct and separate evaluator of practical reasons” (487). As such, the case of the DW is a case of a deficiency of answerability.

By contrast, Westlund imagines a case of deference that isn't a case of deep deference in the way shown in the DW case. She imagines an extreme Anti-Feminist (AF), who also thinks that women ought to demote their own interests and let their husbands make all the decisions. Unlike the DW, however, the AF gives the question of why she is deferential a proper uptake. Westlund says, “When questioned about her deference, AF has a justification at the ready. She may place herself within a religious community whose creed she shares and whose lifestyle she values. . . . Whatever we think of the arguments she offers, what's important is that AF is disposed to enter into justificatory dialogue about her deference . . .” (512).

Westlund's view is that we should consider the AF autonomous. She says that though we might think AF is living a life that is morally problematic, and disagree with the basis of her values, what matters is that we can continue a conversation with her, and even “take her on as a political or ethical adversary” (512). To this extent, we can ascribe autonomy to the AF in a way that we were unable to do with the DW. Dialogical exchange with DW did not yield much more useful information, but in the case of the AF we were directed to meaningful differences in the reasons for their deference that we could consider as legitimate first-personal challenges against any claims that they are deficient in autonomy.

But are these characters really so different? Westlund considers the possibility that we might end up in the exact same position with the AF if we took the AF's way of justifying herself as a more indirect way of arriving at deep deference. If the AF's reason for deferring to her husband is that the Bible tells her to defer to her husband, it doesn't really seem that she is any less deeply deferential than the DW, even if she is more articulate with respect to citing her reasons to do so. But for Westlund, a willingness to be “engaged in potentially open-ended self-evaluative dialogue” (514) would exhibit the appropriate responsibility for self despite the appearance of deference. In my view, however, we cannot divorce this answerability from the social background against which it is exercised: the first-personal component is one important part of a fairer autonomy ascription, and it may well be that we should ascribe autonomy to the AF in light of her rationales. Yet I would be careful not to commit to answerability itself as a constitutive criterion of autonomy. This is where my advocacy of the dialogical view would fall short of Westlund's metaphysical commitment to accept someone like the AF as descriptively autonomous.

This leads to my second worry regarding the role of dialogical exchange. The kind of “self-evaluation” Westlund is concerned with may itself be distorted by oppressive socialization. The dialogical account does not necessarily exclude the possibility that one's basis for decision-making rests on oppressive socialization. Thus, articulacy is a skill that may in such cases mask these problematic influences. In this case, turning toward the agent's first-personal perspectives for answers through dialogical exchange may not be particularly useful for their autonomy ascription. First-personal perspectives may sometimes not serve as reasons to attribute autonomy to agents; their justificatory schemes (or lack thereof) may rather reveal to us reasons we should hesitate to attribute autonomy to them (though of course we must have very good evidence to not attribute autonomy).

Be that as it may, I don't believe that the demand for articulacy generated by dialogical exchange, nor the potential that oppressive background influences might be obscured in answerability, outweigh the desirable features of utilizing dialogical exchange in autonomy ascription. Consulting agents’ first-personal perspectives would still be good practice, from an ethical point of view, and the fact that an agent may have limited insight into their own circumstances does not diminish the point that it would be epistemically valuable as outsiders making judgments about their autonomy to learn something about their perspective.

Overall, I have argued that ascribing autonomous status to agents is a good thing, which makes autonomy ascription a practice we should care about. I've suggested that we should specifically involve a dialogical account of autonomy—which involves consulting an agent's answerability toward questions directed at them—as part of the process of autonomy ascription. This should give us an ethical and epistemically desirable way to work out disagreements regarding autonomous status when it comes to autonomy ascription. Nonetheless, the focus on answerability may be undesirable or limiting in some ways: it potentially invokes a burdensome demand for articulacy, and articulacy can also obscure problematic features of an agent's decision-making bases such as oppressive socialization. Keeping these limitations in mind, however, dialogical exchange may serve as a tool that can calibrate our practices of autonomy ascription in promising ways.