Toggling: How Shifts between Deontology and Virtue Ethics Undermine Public Moral Discourse about Gender and Race

Abstract

This article argues that shifts between deontological and virtue-ethical moral reasoning hamper public moral conversations about misogyny and racism. Using examples ranging from sexual assault to police violence, I show that when someone is accused of committing a moral wrong, they often respond by shifting from deontological to virtue-ethical modes of analysis. This kind of “toggling” between the two enables a person to claim that they are “a good person” even when they have violated important moral rules. In this way, toggling shuts down or makes incoherent our efforts to call one another to moral account, especially (but by no means exclusively) when we are attempting to address issues of systemic oppression and discrimination, including misogyny and racism.

Deontology and virtue ethics are often presented as alternative, or even mutually exclusive, approaches to ethical theory, but moral actors often move between them when they analyze their own or others’ moral conduct. This article analyzes how shifts between deontological and virtue-ethical analyses of moral situations hamper public moral discourse, especially discourse about race and gender. Here, I argue that these shifts, or “toggles,” trade on an artificial separation of the analysis of individual moral actions from analysis of a moral actor's character, which in turn limits the possibility of future moral change. I propose that toggling has the following structure: When people are accused of violating a moral rule, they often respond by using a contrasting moral reason, thereby toggling from one mode of moral reasoning to another. In the most common form of toggling, those violating a deontic norm respond by claiming that they are virtuous. Toggling can also occur in the opposite direction: those accused of having some vicious trait can attempt to shift our moral attention to a specific good action that they have performed, or to argue that they have not violated any explicit moral or legal rules.¹

Toggling can also change the way that we relate to the victims of moral wrongs, with observers attempting to defuse a deontological moral claim by arguing that the victim is of poor character. In this way, toggling between deontological and virtue-ethical modes of moral reasoning shuts down or makes incoherent our efforts to call one another to moral account, especially (but by no means exclusively) when we are attempting to address issues of systemic oppression and discrimination, including misogyny and racism.

In order to clarify the phenomenon's structure, I analyze several cases of toggling taken from recent political events. Here, I focus on describing toggles that shift from deontological to virtue-ethical modes of reasoning, in order to analyze how they shape public discourse about gender, race, and their intersection. I show that some forms of toggling create a significant “aretaic deficit” between different moral actors by linking some aretaic assessments with other, unrelated traits; like the “credibility deficits” that hamper the epistemic credibility of some knowers, especially women and people of color, “aretaic deficits” place women and people of color at significant disadvantage in some moral analyses. After outlining toggling's structure and effects, I offer a more integrated account of the relationship between deontological and virtue-ethical moral reasoning. I argue that this integrated account can be used to combat the underlying moral psychology that makes toggling between deontological and virtue-ethical modes of reasoning seem plausible.

The Anatomy of Toggling

Several moral theorists have recognized the potential for something like the toggling I am describing: In his essay, “Two Faces of Responsibility,” Gary Watson distinguishes between two kinds of moral evaluation, one that analyzes the action in terms of a person's accountability for it and one that treats it as evidence of a person's character; Watson calls the latter “aretaic appraisal” (Watson 1996, 232). Building on, but also critiquing, this distinction, Stephen Darwall argues that although Watson and others treat aretaic evaluations as distinct from claims of moral responsibility, there are really two kinds of responsibility: “responsibility as attributability” (or aretaic appraisal) and “responsibility as accountability” (Darwall 2016). Darwall notes that a failure to distinguish between these two kinds of responsibility can be morally problematic when actors “avoid taking responsibility for their culpable actions by saying to themselves that their record of moral actions is sufficiently strong otherwise that the effect of the action in question on their overall goodness is negligible” (Darwall 2016, 19). Here, Darwall captures part of the broader phenomenon that I wish to describe. Below, I show that those responding to moral claims often switch both from deontic to aretaic modes of analysis and vice versa when analyzing both the perpetrators of moral harms and the person(s) who suffer(s) those harms.² Just as someone might self-servingly claim that they are not responsible for a given “culpable action” because of their own moral track record, they might also claim that they are not responsible because of some feature of the harmed party's moral track record.

Before proceeding, though, it is worth noting that my underlying moral psychological concern differs slightly from Darwall's. Darwall's primary concern seems to be that skepticism about whether a given action is really “attributable” to a person (that is, whether it indicates that a person possesses a given character trait) might undermine the coherence of asking questions about “responsibility as accountability” at all. In other words, both Darwall and Watson (and P. F. Strawson, on whose work they both draw) are replying to a philosophically driven skepticism about the concept of moral responsibility, or even of the possibility of free human action more generally.³ This is certainly a prominent philosophical worry, but here I am more interested in describing a broader nonphilosophical intuition that there is something exculpatory about appealing to a virtue-ethical moral reason in the face of a deontic moral claim. In what follows, I analyze the different ways that this kind of “toggle” is performed, in order to then offer a clearer account of the relationship between deontic and virtue-ethical moral thinking.

Toggling by Appeal to the Actor's Character

In Down Girl: The Logic of Misogyny, Kate Manne analyzes the moral discourse surrounding misogyny and sexual violence. Manne provides an instructive account of the kinds of moral moves that are often used in the aftermath of sexual assaults (Manne 2017). Though Manne does not characterize them this way, these moves can be understood as a shift from deontic to virtue-ethical modes of moral reasoning.

Manne uses the case of Brock Turner, who was accused of attempting to rape Chanel Miller behind a dumpster outside a campus party, as the paradigmatic example of what she calls the “honorable Brutus problem,” which uses an ill-formed syllogism to make irrelevant evidence seem exculpatory.⁴ This problem, Manne argues, plagues many efforts to hold the perpetrators of sexual violence to account. This is precisely what occurred in the Turner case, when Turner was sentenced to only six months in a county jail. He was later released after serving only half that time (Levin 2016). During the case's sentencing phase, Turner's father testified that the conviction, and the possibility of prison time, posed a threat to the life “that [Turner] dreamed about and worked so hard to achieve,” making it “a steep price to pay for 20 minutes of action out of his 20 plus years of life” (Xu 2016). Manne argues that the father's testimony, along with similar testimony from one of Turner's friends, were examples of the “honorable Brutus problem”:

Both Brock's friend and the judge in the case seemed to be reasoning in accordance with the following inference pattern: a golden boy is not a rapist. So-and-so is a golden boy. Therefore, so-and-so is not a rapist. This the honorable Brutus problem. It is high time to give up the myth of the golden boy, to reject the major premise, and learn to tollens the ponens when appropriate. Turner was found actively violating his victim, who was unconscious and intoxicated, in an alley behind that dumpster. That is rape. Somebody who rapes is a rapist. So Turner is a rapist—as well as a golden boy. Therefore. . . . (Manne 2017, 198)⁵

For Manne, the key problem with the logic deployed by Turner's father and his friend is that, rather than assuming that because Turner raped someone he was no longer a “golden boy,” they assumed that, because Turner was a “golden boy,” he could not have raped someone. Manne argues “The implicit modus ponens here is too seldom tollensed” because the term rapist is often defined differently from the way Manne defines it herself in the passage quoted above: rather than simply being “someone who rapes,” a rapist is often assumed to be a sociopathic personality (180). Manne writes,

One reason for this denialism is a mistaken idea about what rapists must be like: creepy, uncanny, and wearing their lack of humanity on their sleeve. “Brock Turner is not a monster,” wrote one of his female friends, in a letter blaming his condition on political correctness. He was the victim of a “camp-like university environment,” in which things “get out of hand” due to alcohol and “clouded judgment.” Turner's crime was “completely different from a woman getting kidnapped and raped as she is walking to her car in a parking lot.” “That is a rapist” she wrote. “I know for a fact that Brock is not one of these people.” (198)

Elsewhere in her statement, the same friend also asked rhetorically, “Where do we draw the line and stop worrying about being politically correct every second of the day and see that rape on campuses isn't always because people are rapists” (Levin and Walters 2016). Manne is correct to suggest that there are two disparate definitions of the term rapist at work here, but this semantic dispute masks a deeper disagreement about the kind of moral analysis that ought to be performed in both the legal case and the public discourse surrounding it. Moving from one definition to the other marks a shift between deontological and virtue-ethical modes of moral reasoning. Manne supplies her own definition, “Someone who rapes is a rapist,” whereas Turner's friend suggests that a rapist is a kind of stereotyped, crazed sociopath who kidnaps and rapes women at random in a public place.⁶ Manne uses rapist as a deontic term, whereas the friend uses it as a virtue concept, albeit a negative one.

This shift between using rapist as deontic concept as opposted to an aretaic assessment explains why these two definitions interact differently with the term golden boy. Someone labeled a golden boy cannot be a rapist in the sense of crazed sociopath, since having the character of a crazed sociopath is incompatible with having the character of a golden boy. Because these are both virtue concepts, they are in competition with each other; one cannot be both a crazed sociopath and a golden boy. So if someone is a golden boy, then they are not a crazed sociopath. On the other hand, though, it is possible for a golden boy to be someone who has raped someone, since this one action is claimed to be of minimal significance against the backdrop of the broader character of a golden boy. In fact, given the constellation of traits Manne associates with being a golden boy (white, nondisabled, often athletic, and academically high-achieving), and the high rates of sexual assault on college campuses, it seems likely that, reasonably often, people who are golden boys are also rapists in the deontic sense.⁷

In their defense of Turner, his father, his friend, and the judge deployed two different sets of virtue concepts, neither of which was relevant to the original deontic claim that attempted to hold Turner responsible for attempting to rape Miller. First, as Manne notes, Turner's father and friend sought to cement his status as a golden boy, citing his “easygoing personality” and his “gentle and quiet nature,” as well as his academic and athletic achievement (Xu 2016). These traits are meant to help us understand “what kind of person” Turner is in some general sense. In this way, whether or not we grant its validity, golden boy is implicitly deployed here as a virtue concept, and subsidiary virtue concepts like easygoing are meant to point toward a broader wholistic account of Turner's character. When Turner's friend argues that the “evidence of his character up until the night of this incident . . . has been positive,” she helps him be seen as a golden boy rather than a rapist (Manne 2017, 198).

Second, what starts out as a deontic analysis designed to hold Turner responsible in the sense of being “accountable” for rape becomes a character-based analysis of the term rapist. A rapist, then, is not “someone who has committed a rape” but a monster or sociopath; deployed this way, rapist is also a negative virtue concept. The question has been turned from one of “accountability” for having performed a morally proscribed action of rape to whether someone is a rapist. Once the term rapist has been transformed into a virtue concept, it becomes possible to say that “a golden boy cannot be a rapist.” However, this inference goes through only once the work of moving into a virtue-ethical register has been done in preparation. In order for the propositions to be in place for the modus ponens that Manne criticizes to exculpate Turner, we first have to move into the realm of virtue-ethical analysis; we have to move from asking about “responsibility as accountability” to “responsibility as attributability.” Having framed the question of Turner's responsibility in terms of virtue rather than in terms of the violation of a deontic norm, his defenders are able to claim not that Turner did not attempt to rape Miller, but rather that his actions were not sufficiently “attributable” to him to warrant saddling him with the lifelong consequences of being publicly recognized as “someone who rapes.” In fact, they challenged this attributability by arguing that Turner's actions were better explained by a “promiscuous” and alcohol-fueled campus culture than they were to the moral agent named Brock Turner (Stack 2016). By toggling from a deontic to a virtue-ethical mode of analysis, Turner's father and friend attempt to turn the judge's attention away from the act of rape that Turner performed. Once his long track record as a golden boy had been established, even a weighty moral wrong begins to seem morally insignificant on a virtue analysis.⁸

Our analysis of toggling from the Turner case can also be applied to responses to claims of racism; identifying the structure of toggling can help us see that racist, like rapist, is sometimes used deontically and sometimes used as a virtue concept. Because the term racist relates to a wider range of actions than rapist, we might mistake racist as a virtue concept that refers only to a repeated, psychologically deeply rooted pattern of behavior. But, following Manne's effort to “ameliorate” the definition of misogyny to refer to a set of social relations rather than to a set of attitudes held by a moral actor, we can see that the term racist also can be used deontically, to refer to an action that enforces or exploits systemic racial prejudices and hierarchies.⁹ In many race-related toggles, the term racist is deployed as a virtue concept that describes a thoroughgoing psychological state, rather than as an adjective that describes an action's role in a system of racial prejudice and oppression.

In a striking example of this from May 2020, a birdwatcher named Christian Cooper asked Amy Cooper (no relation) to comply with a policy requiring that dogs be kept on a leash in that section of Central Park. Amy Cooper responded by saying “I'm calling the cops . . . I'm gonna tell them there's an African American man threatening my life” (Vera and Ly 2020). Amy Cooper made good on her threat, which Christian Cooper captured on video. When she called 911, she told the dispatcher, “There is an African American man—I am in Central Park— he is recording me and threatening myself and my dog. Please send the cops immediately!” (Vera and Ly 2020). Many saw the incident, which occurred on the same day as a Minneapolis police officer murdered George Floyd, as an example of the ways that Black men are often assumed to be dangerous. These accounts emphasized that an interaction with the police could have ended violently, or even fatally, for Christian Cooper. The next day, Amy Cooper issued a short apology:

I am well aware of the pain that misassumptions and insensitive statements about race cause and would never have imagined that I would be involved in the type of incident that occurred with Chris. I hope that a few mortifying seconds in a lifetime of forty years will not define me in his eyes and that he will accept my sincere apology. (Cooper 2020)

Later, she told a CNN reporter: “I'm not a racist. I did not mean to harm that man in any way,” adding that she also did not intend to cause harm to the Black community as a whole (Vera and Ly 2020).

Amy Cooper's response uses a very similar logic to Turner's—she accepts the basic fact pattern of the incident, but makes two counterclaims, both in a virtue-ethical key. First, she claims that these “few mortifying seconds” ought not be considered as morally determinative against a much longer record of (presumably good) behavior. In addition to offering a positive assessment of her own character, she also treats the term racist as a virtue concept, which she takes to name a longstanding trait marked by an intense, internal hatred, which Cooper claims not to have felt. Whether or not the incident itself made use of, or was a product of, racist social structures is, in Cooper's analysis, not especially relevant here.¹⁰ In order to set up the modus ponens that would prove that Cooper was not a racist, one first has to move the discussion away from the act-centered question of whether her behavior constituted a racialized threat, but rather, whether Cooper is a racist moral agent.

Some toggles make these moral psychological assumptions more explicit—rather than simply citing instances of previous virtuous behavior or generalized statements about “disposition,” some toggles make a direct appeal to a “real” or “genuine” self that (supposedly) could not have performed the relevant actions. We can see this clearly in another example of toggling in response to accusations of racist behavior. Two years before the incident in Central Park, Aaron Schlossberg, a New York lawyer who was videotaped threatening to call ICE (Immigration and Customs Enforcement) on workers who were speaking Spanish in a New York restaurant, also used a similar toggle, arguing that “What the video did not convey is the real me. I am not racist” (Robbins and Salam 2018). Instead, Schlossberg claimed that the “real him” was best understood by the fact that “one of the reasons he moved to New York is because of immigrants and the diversity of cultures immigrants bring to this country” (Robbins and Salam 2018). Later, several other videos surfaced of similar behavior, prompting state representative Adriano Espaillat and then-Bronx borough president Ruben Diaz Jr. to file a formal complaint with the New York State court system (Robbins and Salam 2018). I will argue below that claims of the form “That is not me” are only properly made aspirationally rather than descriptively. It is obvious that Schlossberg is indeed the person who said and did what was captured on video, but it might nonetheless be the case that Schlossberg does not wish to be the kind of person who does this. Recognizing the gap between these two versions of oneself then opens up the possibility that one can, perhaps in dialogue with the harmed party, do the relevant psychological work in order to become the kind of person who would not have performed the morally proscribed action. Responding with a descriptive claim that “That is not me” actually makes this work impossible, because it refuses to recognize that any character change is necessary in the first place; it simply assumes that I am already the kind of person who would never do this, even if all evidence points to the contrary.

Toggling in Order to Characterize the Victim of a Moral Wrong

In the examples considered thus far, I have shown how toggling can be used to try to exculpate someone accused of a moral wrong. However, “the honorable Brutus problem” has an equally insidious analogue that analyzes the victim's character. Those responding to a deontic moral charge can also shift to a virtue-ethical analysis of the victim. Manne notes that in sexual assault cases, “small violations may be blown out of all proportion, and taken to indicate something damning about a woman's character. She may be represented as breaking promises, telling lies, or reneging on her side of the bargain—and hence as deeply untrustworthy, duplicitous, irresponsible, and so on” (Manne 2017, 54). In these cases, toggling treats the victim as “promiscuous” or “asking for it”—once the question has been shifted to a virtue-ethical one, then one appealing way to answer it is by comparing the character of the victim and the accused—so Turner is a “golden boy,” who happens to be caught up in a “promiscuous” campus “culture” that is not representative of his character, while Chanel Miller is, herself, “promiscuous.”

Manne comes close to recognizing this shift to a virtue-ethical analysis of the victim's character when she observes that “we tend to think of victims as innocent, blameless, and (worse) as needing to be” (122). “Often,” Manne continues, “there is a subsequent reluctance or failure to recognize someone as a victim when they are, or are even suspected of being, guilty of some minor perfidy. By the same token, often we are inclined to deny the minor perfidies of someone who [sic] we take to be a genuine victim of some major offense” (225). However, we can now see that this move is made possible by toggling from a deontological analysis to a virtue-ethical one. We are much more likely to need the victim to be “blameless” if what we are really doing is comparing the relative virtuousness of the two individual actors. Some philosophers have argued that women and people of color suffer from a “credibility deficit” that puts them at an epistemic disadvantage.¹¹ My analysis of toggling shows that once a toggle has been performed, these women (and others alleging harm) also suffer from a crippling “aretaic deficit”; not only are their statements given less epistemic weight, their character tends to be judged more harshly. Whereas we are asked to assume that the accused's misdeeds were not representative of their character, any potential missteps on the part of the victim are assumed to indicate poor overall character, thereby purportedly undermining the moral claim against the person who harmed them.¹²

A similar pattern often occurs during the response to charges of racism. Manne notes that, in the aftermath of the killing of Michael Brown by Ferguson police in August of 2014, police released video footage in which Brown appears to steal a pack of cigarettes from a convenience store just a few minutes before he was shot. Manne argues that this footage, seemingly “irrelevant to the question of whether Brown was the victim of an egregious civil rights violation, in the form of no less than a homicide, at the hands of a state actor” was released to discredit Brown's character, to make it seem like he could not fit into the role of a necessarily “blameless” “victim” (Manne 2017, 226–27). Manne writes,

Once Brown was depicted as any kind of criminal or aggressor (in however trivial a way) via the footage of him in the convenience store, many white people couldn't or wouldn't see him as the victim of police brutality or misconduct. The two narratives—Brown as having committed a minor wrong, and Brown as victim—seemed to compete with each other, even though these two possibilities are of course compatible. (227)

There is more than a conflicting set of “narratives” here, however; focusing on the “character evidence” produced by the video made it so that Brown would never be deemed the “more virtuous” agent, no matter who held the gun and whose body bore the bullets. This made it impossible for Brown to be seen as the victim.

Disturbingly, this trend has continued in the response to more recent police killings. An autopsy report showed that George Floyd had fentanyl in his body at the time of death, and that there was evidence of previous methamphetamine use.¹³ This led some, including a then-sitting member of Congress, to argue that Georgy Floyd “did not deserve protests,” because he was a “druggie,” “thug,” and “armed robber” (Brooks 2020).¹⁴

Some police departments have even institutionalized the practice of gathering negative character evidence on the victims of police shootings. In 2015, The New Yorker reported that the Albuquerque police department would develop a “red file” in response to police shootings; a former Albuquerque police sergeant told The New Yorker that after a police shooting, “The special-investigations division did a complete background [check] on the person and came up with any intelligence to identify that, you know, twenty years ago, maybe, the person got tagged for shoplifting. . . . Then they gave the red file to the chief” (Aviv 2015). The approach worked. Renetta Torres, whose son was later killed by Albuquerque police officers, told reporters that when she saw local news reports of police shootings, she “just assumed that these men must have done something to merit being killed. On the news, they relayed these really sinister stories about the men, and they'd flash these horrible pictures. They looked frightening” (Aviv 2015).¹⁵ The narratives and images Torres saw on local news assessed the victims’ character, implicitly leading viewers to suspend the generally applicable deontic norm that prohibits police officers from killing unarmed citizens.

The fact that a person needs to be a “psychopath” or a “monster” to be a rapist can be used to exculpate “golden boys,” but a similar kind of aretaic assessment is used to call those murdered by the Albuquerque police “monstrous” or “psychopathic,” or to refer to George Floyd as a “druggie.” Falling into one of these kinds of character assessments has a strong deontic effect—if a victim is a “psychopath” or a “monster,” then the deontic norms that might otherwise protect them can be suspended; at the same time, though, being “not a monster” can keep someone from being held accountable for attempted rape. So, even as togglers rely on a supposedly clear distinction between virtue-ethical and deontic moral assessments (otherwise a virtue-ethical moral assessment would not be exculpatory in cases where everyone agrees on the basic fact pattern), they also quietly make strong connections between some aretaic assessments and the applicability of certain deontic norms.

A Note about Gender as an Aretaic Concept

We can now see that one way that toggling works is by grouping certain sets of aretaic assessments.

There is a long-standing philosophical debate about whether the virtues are all conceptually connected (Vlastos 1972; Wolf 2007; Toner 2014), but it seems that many moral actors (and analyzers of moral action) do make strong associations between certain virtues. This means that, if someone can claim that they have a given virtue, they can also implicitly lay claim to another set of virtues, even without producing evidence from their own conduct that relates directly to those virtues. So, rather than claiming that “I am a person with the virtue x, because I am a person who x's,”¹⁶ an actor can implicitly lay claim to the virtue x by claiming that they have some other virtue y, which tends to be linked to x. These linkages are socially produced through the narratives we tell about virtuous (and vicious) people, both fictional and nonfictional, and through the “lists” of virtues that are sometimes collected in both philosophical and nonphilosophical contexts.¹⁷

The linkages can, of course, be gendered. Moral subjects might be quicker to link virtues that are often coded as feminine with other “feminine” virtues than they are with those that are usually coded as masculine. So “caring” might tend to go more with “polite” than with “courageous.” In addition, some traits might be linked to different constellations of traits when they are taken to fit into certain gender categories; “strong” might imply different sorts of things about a woman than about a man. In turn, these linkages help produce the “aretaic deficits” alluded to earlier by making toggles more accessible, or more likely to succeed in exonerating some actors and rending others culpable. Thus, a woman is less likely to be “exonerated” by an appeal to her athletic or even academic prospects than a man would, because possessing the traits “strong,” “powerful,” and “intelligent” are not as strongly linked to other positive aretaic assessments for women as they are for men.

It is even possible for gender concepts themselves to be deployed as aretaic assessments that also reinforce these deficits. In their work on “real men” Robin Dembroff argues that patriarchy rewards “real men” while punishing both those who are not men and men who fall short of the expectations of masculinity (Dembroff 2023). This enables Dembroff to show how patriarchy harms some men as well as nonmen. In this way, Dembroff argues, “[M]anhood is something to achieve, prove, or pursue” (Dembroff 2023). Patriarchy polices both men and nonmen by deploying men as an aretaic concept; being a man is something one can be good or bad at.

When deployed aretaically, gender concepts can be used to facilitate toggles. To be called a real man either explicitly or implicitly (or, to be called a related label like family man), is actually a way to lay claim to a certain set of virtues, even without necessarily demonstrating all of the relevant traits. Turner's athletic prowess, or even his disregard for women's bodily autonomy, might implicitly render him a real man, thereby allowing him to claim unrelated (or even contradictory) aretaic descriptions like “personable,” “bright,” or “promising.” Similarly, a person might claim to be a “family man” by referencing a positive relationship with his wife or daughter, even as he also harasses women in the workplace.¹⁸

The links between some aretaic assessments can be so strong that people who disrupt them can be seen as inherently threatening or disturbing. In an interview over a year after the incident in Central Park, Amy Cooper deployed a version of these “grouped” aretaic assessments, and then, in almost the same breath, expressed dismay when these groupings were disrupted. Amy Cooper initially described Christian Cooper as having a “very physical posture” and having “been very dominant towards me” (Weiss 2021). Although each of these assessments might seem to code Christian Cooper as a “real man,” they also need to be interpreted in terms of the way racial categories intersect with grouped aretaic assessments. A black man who is described as being “dominant” might be more likely to be seen as a criminal than as a real man. Either way, though, Amy Cooper's initial assessment of Christian Cooper matches the way that she expects aretaic assessments to be linked together. But then, Amy Cooper says, something changed: “suddenly out of him comes this voice from [a] man who's been very dominant towards me. Suddenly, you know, almost this victimized voicing . . . like, almost like he's terrified of me. . . . To me that's even more terrifying now” (Weiss 2021). Amy Cooper's deepest dismay seems to come not from the fact that Christian Cooper presents as “dominant” or “physical,” but rather than he eventually shows his fear. This kind of behavior does not fit with Amy Cooper's assessment of him as a real man, or, for that matter as an “African American male” (the term Amy Cooper used in her call to the police).

Understanding that gender concepts can function aretaically, and recognizing the emotional response that can be trigged when groups of aretaic assessments are disturbed, also helps us understand what is driving the allegiance to what Dembroff calls the “Real Gender assumption,” the view that “Gender classifications should track the operative gender kind membership facts” (Dembroff 2018, 29). As I have just argued, gender categories can facilitate moral inferences by linking certain aretaic assessments to one another. But if these categories are not grounded in something “real” that can be picked out in the world, then it is harder to trust inferences based on gendered links between virtues, because there is no fact about the world “out there” that is grounding these associations. Conversely, an attack on the “Real Gender assumption” is also an attack on a set of inferences about grouped virtues. Losing the stability of these linkages makes moral analyses seem much more unpredictable—if we deny the “Real Gender assumption,” and the linkages between virtues that it provides, then toggles will be less likely to exculpate actors who have aretaic advantages, and less likely to condemn actors who suffer from aretaic deficits.¹⁹ A fear of this instability may help explain some of the backlash we see in response to efforts to undermine the supposed empirical basis for the “Real Gender assumption.” We may also be able to see similar sorts of reactions to efforts to deconstruct racialized categories like “criminal” as not grounded in some concrete trait “out there” in the world that some actors have but other actors lack.

Manne's Response: An “Ameliorative” Definition of Misogyny

In an effort to respond to the “honorable Brutus problem” and other logical moves that bolster misogyny, Manne offers an “ameliorative” conception of the term. Manne begins by noting that on the “naïve conception,” “misogyny is primarily a property of individual misogynists who are prone to hate women qua women. . . . Misogyny is as misogynists are, then. And misogynists are agents who fit a certain psychological profile” (Manne 2017, 18). Though Manne does not explicitly note it, on the “naïve conception,” misogyny is treated a virtue concept; to be a misogynist to be a certain kind of person, and “misogyny is as misogynists are.”

Manne then goes on to propose a new definition, which will help identify misogyny's role even in cases whether there is no “deep” or “psychological” account of what is going on in an individual person's head. Manne suggests that we adopt the following “ameliorative conception” of misogyny: “we should instead understand misogyny as primarily a property of social environments in which women are liable to encounter hostility due to the enforcement and policing of patriarchal norms and expectations” (19). Again, though she does not note it here, the “ameliorative conception” of misogyny is an act-centered conception, used for deontological analysis rather than agent-centered conception used for virtue-ethical analysis.

Our analysis of the cases considered above enables us to recharacterize Manne's ameliorative definition as an effort to shift our analysis of misogyny from a virtue-oriented one to a deontic one. Faced with people who toggle from deontological moral claims to virtue-ethical ones, Manne suggests that we stand our ground and reassert that this was never about virtuous agents, it was always about actions: for the victims, the relevant data is the set of actions that were carried out against them, rather than any other morally blameworthy actions they may or may not have done in other situations; and similarly, for the perpetrators, the relevant question is whether they performed a given series of actions in violation of some recognized deontic norm, rather than whether we can make a positive aretaic assessment of their character overall.

This approach may work in some cases, but it limits the range of moral claims that are available to those who want to fight misogynistic behavior. Not all moral claims—including those made in the name of fighting misogyny and racism—are made in an act-centered register, and not all claims of moral responsibility are claims of “responsibility as accountability,” in Darwall's sense. In fact, when an accusation of “racism” or “misogyny” is made, it can be made in either a virtue-ethical or deontic key. Sometimes when we say that someone behaves poorly, this claim is really meant as a moral assessment of the agent, and not just the action; sometimes, to use Darwall's terminology, we make moral claims in an effort to hold someone responsible in the sense of “attributability.” According to Darwall, “responsibility as attributability” is essentially a claim about “esteem.” For Darwall, “Attributions of virtue and vice concern how estimable someone is; they call on attitudes of esteem and disesteem. To make a claim about someone's ‘responsibility as attributability’ is to assess how an action should affect our esteem or disesteem of the agent” (Darwall 2016, 17). These sorts of attributability claims played a key role in the #MeToo movement. The remedies that accusers, policy makers, and community members seek in these scenarios are often focused on removing social markers that note a person's high esteem or social status—be they titles, positions, scholarly citations, or public markers of philanthropic contributions; these are the kinds of claims we make when our general trust in another person has been undermined, when we can no longer hold them in the same kind of “esteem” as we did previously.

Although Manne's ameliorative definition gives us the tools to better recognize how misogyny works to enforce patriarchal norms, the shift toward an entirely deontic analysis that underlies it makes it harder to make social and political use of our deontic moral judgments. When we have discovered that someone has raped, we want to be able to “attribute” the action to them, to be able to hold them in less esteem, and to have that revised judgment reflected in public. But the only way to make this sort of move, or for that matter, to begin to consider what it means for us to attribute a deontic wrong to someone, is to develop a clearer sense of what our deontic judgments mean for our aretaic ones.²⁰ It is to this relationship between deontological and virtue-ethical moral assessments that I now turn.

A Way Forward?

Toggles assume virtue ethics and deontology are two separate modes of moral analysis, but we need not treat them as such. Rather, developing a clearer account of how deontic norms and the virtues ought to relate to one another—both in efforts to call one another to account, as well as in other areas of moral discourse—may offer us useful tools to reopen the moral conversations that “toggling” shuts down. I want to be careful not to overstate my claim here: these ideas may offer us useful tools. Whether these tools will prove as useful as they seem to be, and indeed, whether they can be used effectively given the other kinds of social and political challenges we face remains to be seen.

Our first step is to dislodge one key psychological assumption that lies at the root of some virtue-ethical theorizing and that seems to be pervasive in contemporary discourse: that deontological moral thinking is only useful and appropriate in situations where we are punitively (and often, therefore, coercively) seeking to hold another person accountable for violating a moral norm.²¹ Recent empirical work in moral psychology has demonstrated the prevalence of this assumption. Joshua Knobe demonstrated that people are more willing to describe an action as “intentional” if they took it to be morally wrong; conversely, an intentionally morally good action was less likely to be considered “attributable” to the agent (Knobe 2010). In his analysis of Knobe's observations, Darwall suggests that Knobe's subjects apply a deontic lens when analyzing morally negative actions and an aretaic one when analyzing morally positive ones. Darwall explains this by suggesting that, in the negative case, people might have been afraid of letting the wrongdoer “off the hook” (Darwall 2016, 25). But Knobe's results may also be understood as products of a different moral psychological assumption: many associate the use of a deontic moral lens with behavior that is morally coded negative. To think deontically is to “accuse” or to construct a list of prohibited actions. In contrast, to think aretaically is to offer a substantive account of “character,” a term that is already coded positive in contemporary parlance. These associations then make shifting the conversation to a discussion of “character” a natural move in response to a deontic moral claim, since character analysis is so often focused on positive traits, thereby laying the groundwork for toggling. In order to avoid toggling, we will have to work to dislodge the assumption that deploying deontic moral language is always about “finding someone guilty.”²²

Here I want to suggest that deontic norms might have (at least) three other relevant functions. First, a claim that someone violated a deontic norm might be articulated with an eye to positively influencing a person's character, or to pointing out a wrong of which the moral actor might not be fully cognizant. This kind of rebuke is not designed primarily to “catch” someone doing something wrong, but instead to help shape future behavior both of the original actor, and of others who hear the rebuke. In this way, rebuke gives living voice to a set of rules that might otherwise be found only in codes, or simply ignored. Understanding rebuke in these terms helps us develop one potentially useful response to a toggle from deontic to virtue-ethical reasons—when someone responds by shifting a discussion about a specific action to a set of virtue-ethical reasons, the rebuker can insist that she was talking about character all along, and that character is determined by the morally significant actions a person takes. In order to be a good person, this rebuker asserts, one must not be the kind of person who performs this kind of illicit action; the two modes of analysis cannot be separated. One cannot be a “good person”²³ and a “rapist,” at least not the sense of a “person who has raped others.”

It is important not to let our picture get too rosy here—it is unfair to assume that Miller meant to help Turner “become a better person,” or that those who came forward during the #MeToo movement did so out of concern for their harassers’ and abusers’ ability to “live the good life.” But what Miller, and others like her, did do was to help the broader moral community understand that behavior like Turner's is not part of the “good life” as she understands it, and that her response to him, rather than “cruel” or “vindictive,” is a picture of the “good life.”²⁴ Although she may not have been able to shape Turner's character, she may have shaped her audience's.

In this sense, those who lamented that the “rules had changed” as a result of the #MeToo movement misidentified the moral change that had really been inaugurated (but has by no means been completed) by the movement: the legal rules prohibiting sexual harassment had been on the books in the United States since the passage of the Civil Rights Act in 1964, but it had remained possible for someone to repeatedly violate these deontic norms and still be held in high esteem—in many cases, these individuals were seen as “leaders” in their fields and were allowed to continue to hold positions of significant power even after they had been credibly accused.²⁵ Having violated these deontic norms did not significantly affect many people's aretaic assessments of them. It was not the rules that changed as a result of the #MeToo movement, but instead, our sense of what it meant to live well and be worthy of esteem and all of the benefits (economic, social, political, professional) that a high aretaic assessment enables one to accrue. This shift in the general aretaic consensus took place in part because of efforts to hold people responsible for violating existing deontic norms that had previously been all but ignored.²⁶

Second, a discourse of deontic norms can be a tool for cultivating the virtues. In her treatment of epistemic injustice, Miranda Fricker suggests that one way to combat epistemic injustice is to cultivate a virtue she calls “reflexive critical openness.” Fricker's description of how this virtue is cultivated provides a helpful example of how deontic norms can be useful tools for cultivating the virtues. In one of her core examples, Fricker imagines a teenager “whose testimonial sensibility has contracted the defect of not taking seriously what old people say,” but who “finds himself one day struck by the veracity of his grandfather's stories of the war” (Fricker 2003, 162). In this case, “active critical reflection on his habits of hearer response will first produce some sort of corrective policy external to the hearer's sensibility. (Perhaps this teenager disciplines himself when in conversation with the elderly, ‘Don't be dismissive. . .’)” (163). If we read Fricker's language carefully, we can see that she describes the teenager's exercise of this virtue in deontic terms. For the teenager in the example, exercising the virtue of “reflexive critical openness” involves using a new deontic norm, a “corrective policy,” that originally works outside of the person's baseline character or second nature (which Fricker calls “sensibility”); this deontic norm is used to “discipline” the teenager in order to help him give appropriate epistemic credibility to the elderly. Eventually, though, applying this new “policy” ends up shaping the teenager's character. “Given time, and all being well,” Fricker writes, “such a corrective policy will become internalized as an integral part of sensibility, so that it comes to be implicit in his newly conditioned perception of elderly speakers” (163). Fricker assumes that this willingness to be open to the kind of initial revision of the person's epistemic sensibility is itself a virtue: this may be true, but these “revised policies” need not always be supplied internally, as a product of a virtue itself; they can be encountered externally, through rebuke, study, or conversation, and then integrated into a person's psychology in the way that Fricker describes.

Third, in addition to helping shape an individual's psychology, deontic norms can also play a role in critiquing the virtue concepts deployed in public discourse. We can see this kind of work in action in Miller's victim impact statement. In one telling passage, Miller said that “It is deeply offensive that he would try and dilute rape with a suggestion of ‘promiscuity.’ By definition rape is the absence of promiscuity, rape is the absence of consent, and it perturbs me deeply that he can't even see that distinction” (Baker 2016). Here, Miller quietly but forcefully contests a set of implicit assumptions that often accompany the deontic concept “rape.” Often, the deontic norm “one ought not rape another person” travels with a quiet virtue-ethical rider, which assumes that if the victim is “promiscuous” (a virtue trait that applies both to the victim's conduct in the incident, and to the community of actors who inhabit a given milieu) then the norm does not apply or does not apply with the same moral force. Here, Miller critiques the use of this virtue-ethical rider, reminding her listeners that the only salient feature of the victim that we need to consider in order to understand whether the deontic concept “rape” applies is the person's lack of consent.

Miller offers her own redefinition of the concept of “promiscuity.” Although many in her audience, and in American culture more generally, take this to be characterized by a set of cultural markers, including clothing, physical behavior, and even posture, Miller suggests that people need to take this virtue concept (in most cases, though not all, it is treated as a vice) as requiring a person's genuine consent to have sex—something that cannot be signaled only with the things often taken as marks of promiscuity. In this way, Miller critiques both deontic and virtue-ethical concepts by analyzing how they relate to one another.

This kind of critique can happen in two distinct ways: first, we can identify character traits that respond to social realities produced by frequent violation of deontic norms. Lisa Tessman's category of “burdened virtues” is instructive here: Tessman argues that a “burdened virtue” is a trait that leads to flourishing only under conditions of systemic oppression; absent these conditions, though, the trait would not be conducive to flourishing generally (Tessman 2005). However, a discourse of deontic norms can be useful to identifying which virtues are “burdened”—if a trait leads to flourishing because it enables someone to respond to frequent deontic violations, then it is a “burdened virtue.” In addition, a discourse of deontic norms can help us describe a moral universe where the burdened virtue is no longer useful; this will help us understand that the trait is not a genuine virtue and should be cultivated only instrumentally. Treating a trait this way may also help make us more willing to jettison it when it is no longer serving its instrumental purpose. Perhaps more important, we also need to investigate the ways that traits that appear to simply be virtues appear so only because we are ignoring the persistent deontic violations that make it possible to develop and to value these traits.

Second, these deontic norms also enable us to express moral hopes for a world in which some forms of structural oppression no longer exist—many contemporary virtue theorists argue that the moral gap between a person's desires and the moral law puts the agent in a dangerous and frustrating psychological position.²⁷ However, this gap can also be psychologically helpful—deontological moral language can help us use the gap between our lived moral realities and those we imagine and work for to express our moral hopes; when the moral world in which we live is perfused with moral wrongness and vice, deontic norms enable us to express an aspiration for a world governed by different norms. In this way, interactions with deontic norms need not always be punitive; sometimes they can be aspirational. Once expressed this way, these deontic norms can be the subject of discussion and critique, and, perhaps most important, we can begin the character work necessary to be the kinds of people who can eventually live by them.

To engage in this kind of critical work, we would have to be willing to give up on the goal of achieving the kind of moral “self-evidence” that many contemporary virtue theories prize. Although these theories take it as obvious what the virtutes are, either through an analysis of nature or of historical and literary sources, the examples analyzed here should make us cautious about identifying any given set of conduct as “self-evidently” virtuous.²⁸ Any given set of behaviors may be “burdened virtues,” which would not be present or would not be deemed virtuous in the absence of systemic oppression that does significant harm to many. We will not, then, be able to do as G. E. M. Anscombe suggests and replace a conversation about “moral wrongness” with a conversation about whether conduct is “unjust” or “unchaste,” since those concepts are themselves deeply integrated with, and open to, critique using the general concept of “moral wrongness” and the specific moral rules and obligations we associate with it (Anscombe 1958, 9).²⁹

Allowing deontic norms to perform these functions will require us to treat deontic norms not only as a “law code” whose primary use is juridical or punitive; rather, these deontic norms, whether expressed textually or using other cultural media, need to become a tool for shared conversation and study.³⁰ Developing a more frequent and sustained mode of engagement with deontic norms in this way educates us to think of interacting with deontic norms as a normal part of cultural life, rather than something that occurs only in an accusatory, juridical context. This may, in turn, make us less defensive when others claim that we have violated a deontic norm.

To summarize my findings, then, I propose the following structure for the relationship between deontic norms and the virtues:

1. 1) A discourse about deontic norms can be used to describe a vision of the “good life” for an individual or a society by helping us identify which virtues are most important, and by helping us articulate what constitutes those virtues. When someone calls us to account for what they take to be a violation of a moral norm, they are, in effect, teaching us about how it would be good for us (and for those who are affected by our actions) to live. Similarly, when someone articulates a set of norms that they hope a society or group could adopt, they articulate a vision of how we ought to live, even if that vision seems impossible to realize under current conditions.

2. 2) Deontological and virtue-ethical descriptions of morality can usefully critique one another, thereby helping us sharpen and improve our accounts of what it means to live well. Building on Tessman's analysis, this includes helping us revise previously misunderstood virtues, as well as distinguishing between virtues that are conducive to human flourishing and those that do so only because of morally damaging forms of structural oppression, or even because of moral wrongs committed by those who seek them.

3. 3) A diverse range of interactions with deontic norms can shape our character. Character is shaped by adopting and acting in accordance with deontic norms, but conversation about and study of deontic norms can also be character-forming.

4. 4) Understanding and pursuing the virtues can help us become the kinds of people who can live according to the aspirational deontic normative structures described in (1). This can include calling someone to moral account by offering a negative aretaic assessment of them.

5. 5) The virtues can help us recognize and revise poorly formed or poorly enforced deontic norms; this includes revising norms that, although they seemed to make it possible to live well, actually made a good life possible only for a small range of people.

On this view, deontic and virtue-ethical moral analyses are more closely related than deontologists and virtue ethicists have tended to assert, at least explicitly. At the same time though, this approach rejects certain moves that connect the virtues and deontic norms because these moves misunderstand, or, more nefariously, seek to undermine, the proper relationship between the virtues and deontic norms. Under this framework, an appeal to virtue is not allowed to defuse or undermine a claim about deontic norms. This is because the process described in (3) will be successful only if we recognize that a deontic claim is indeed also a claim about some moral or psychological work we take the person to need to do. When we say that someone has violated a deontic norm, we ask them not only to make amends for the damage done (to respond to the juridical use of the deontic norm), but also not to commit the violation again. To accomplish the latter, the person will have to change their character in some way: they will have to become the kind of person who does not commit these kinds of wrongs. To do this, they will either have to become more “continent” about these actions, or they will have to do the more significant character work of ridding themselves of the desire to perform them. My moral theory is agnostic about which of those paths they choose, provided that whichever strategy they choose leads them to stop performing the actions and otherwise does not negatively affect their behavior in other areas.

This view requires us to adopt a different moral psychology than is assumed in most other moral theories. On the view that I am advocating for here, there is very little room for the “innocent mistake” or “blip” on a person's moral record. Instead, on this moral picture even these “innocent mistakes,” which seem to take place without any particular premeditation or thought, nonetheless can represent the kinds of moral agents we have become; this renders the “toggle” from deontic to virtue-ethical modes of analysis unproductive. It does not make sense to argue that a person is both “a good guy” and a “rapist,” since a person who rapes displays a moral disregard for others that is incompatible with being a good person, regardless of his “prior record.” This may seem obvious for actions of a sufficiently large moral magnitude, but the view I am advocating here suggests that even much “smaller” wrongs have some meaningful basis in our character; unaddressed, they too are incompatible with a fully good character. Even if these behaviors do not match up with the rest of our (hopefully better) moral conduct, they are nonetheless morally significant and demand our moral attention; similarly, we must seek to rectify a long period of vicious behavior, even if it was also accompanied by significant morally good actions in other areas. This moral psychological picture assumes that a person's moral life can be quite fragmented—one can be virtuous in one area, and nonetheless behave poorly, or violate deontic norms, in another. Starting from this place of fragmentation, deontic norms offer us a set of tools to help us work toward the kind of moral psychological unity and clarity that many contemporary virtue theorists assume is a baseline in moral life.³¹ Thus, on my view, it does not make sense to respond to a moral claim by saying “that wasn't me,” because, in fact, it really was “you” who performed the action. To the extent that we want to express a moral disavowal of what we have done in the past, we need instead to express our sense of alienation from our own actions in aspirational terms: “that wasn't the me that I want to be.”³²

Coda: Toggling and Moral Plasticity

Toggles, especially ones from deontic to virtue-ethical modes of analysis, also reflect a set of assumptions about moral plasticity. Part of the reason that a person chooses to say, “that wasn't me” rather than “that wasn't the me that I want to be” is that the second formulation makes sense only under conditions where substantive moral change is possible, where I can turn from someone who has harmed others into “the me that I want to be.” However, if one denies the possibility of moral change (or, more frighteningly, denies that such a change would be desirable in the first place), then it makes more sense to simply write off any given instance as morally insignificant, especially if it departs from a previous pattern of behavior.³³

Some have responded to recent efforts to “call out” systemic racism and misogyny with the claim that, in the words of New York Times columnist Bret Stephens, “the Woke left” is only interested in a moral discourse in which “Absolution is off the table” (Stephens 2021). But it may well be that the assumption that “absolution is off the table” comes more from those who respond to accusations of misogyny or racism than from those who make them. Absolution is indeed off the table in an analysis where character is more or less fixed and where “who a person is” (whether “golden boy,” “druggie,” “slut,” or “criminal”) has been decided before any given action, and its potentially deleterious consequences, are analyzed, and before there has been any effort to make genuine amends for past wrongs, or to address the social and political forces that often play a role in them.³⁴ This fixed sense of “who a person is” is reinforced by “grouped” virtues that are assumed to go together. Putting something like “absolution” (or, I might prefer to say, the possibility of becoming the kind of person deserving forgiveness) back on the table requires first accepting a moral psychological picture in which moral improvement and moral decay are both possible. To do this, we will have to be willing to weaken our associations between commonly “grouped” virtues, to let people to surprise us, to exceed our moral expectations, as well as to disappoint them. A strong account of deontic norms that is deeply connected to who we are as people is a necessary component of that work.