3.1. Motivated reasoning

Perhaps the most familiar mechanism by which sexism may serve as the basis of belief is the one in play during motivated reasoning. Motivated reasoning occurs when a believer “has a desire or preference for a particular conclusion and that desire guides their reasoning in a way that facilitates their drawing that conclusion” (Ellis Reference Ellis2022: 2). At the psychological level, sexism is often taken to constitutively involve volitional, affective, and/or motivational states that, in turn, implicate various desires. These desires then go on to be involved in the sort of motivated reasoning that gives rise to sexist beliefs.

For the sake of simplicity and clarity of exposition, I will take on board a volitional account of the relevant psychological states, and account for them in terms of will.Footnote ¹⁰ Broadly, at the individual level, sexism constitutively involves ill will, or a lack of good will toward women and, derivatively, toward (some) individuals qua women. To have good will toward women, and, derivatively, individuals qua women is to desire and care about their flourishing or, perhaps more mundanely, their welfare, broadly construed to countenance considerations about rights, interests, preferences, and autonomy. To lack good will toward women is to not desire or care about their welfare, or at least not to a morally sufficient degree. To have ill will toward women is to desire and care about a deficiency in their welfare. This may include a desire to mistreat or control women, or to “keep them in their place.” Such desires are inimical to respecting women, and their autonomy, interests, preferences, and rights.

Before considering how the desires and cares constitutive of someone’s will toward women can be the basis of belief, one important clarification is in order: although ill will or lack of good will toward women may be correlated with felt contempt or animosity toward women, someone with the relevant volitional states may not have such feelings toward all women, or each individual woman he comes across. As Kate Manne (Reference Manne2017) points out, a misogynist can love his mom or wife, in full knowledge of her womanhood. On my view, that’s because the ill will or lack of good will toward women may give rise to radically different treatment (and feelings) toward “good” women and “bad” women, where – following Manne – “good” women are those who abide by their society’s patriarchal norms and “bad” women contravene those norms. More specifically, on Manne’s view, “good” women are willing givers of feminine-coded goods (e.g., nurture, affection, and sex), and they do seek out masculine-coded goods to which only men are entitled (e.g., leadership, power, and money).

Ill will or lack of good will toward women, then, may manifest in felt animosity toward only “bad women,” or individual “bad women,” who fail to abide by the relevant patriarchal norms. Such animosity – or the potential for such animosity, even when it is not triggered by “good” women – flows from ill will or a lack of good will toward women. For example, consider the desire to keep women in their place (where “their place” is defined by the ruling patriarchal order). This desire is inimical to the desires and cares that are constitutive of good will toward women, but it is frustrated only by “bad” women, and it is this frustration that spurs felt animosity toward women. The upshot is that someone may be sincerely warm (even loving!) toward “good” women, but nevertheless still harbor ill will or a lack of good will toward women, even when any felt animosity remains latent. A sexist does not necessarily “hate” (all) women.Footnote ¹¹

With that in mind, let’s turn to how desires and cares constitutive of ill will or a lack of good will toward women can serve as the direct basis of belief. Consider some paradigmatic sexist belief – say, that women are unfit for high political office. First, consider a sexist who harbors ill will toward women: perhaps he wants to mistreat or control women, or he wants to “keep them in their place.” The desires cause the sexist to favor a particular predetermined conclusion when thinking about the roles of women: recognizing that women are fit for high political office would be in tension with his desire to keep women in their place, and so this desire leads him to favor the view that women are simply unfit.Footnote ¹² Alternatively, the relevant desires can lead to sexist beliefs because those beliefs are wish-fulfilling: the sexist might believe that women are weak or fragile (and, thus, unfit for high political office) so that he can justify his efforts to control and dominate the women in his life as benevolent (Arpaly and Schroeder Reference Arpaly and Schroeder2014: 235).

Now consider a milder sexist who does not have ill will toward women: he does not hate women, nor does he desire to mistreat or control them or to “keep them in their place.” But he does want to fit in with others in his traditionalistic community – where inferiorizing views of women serve as a marker of belonging – and this desire causes him to favor gender views embraced by his community, including the belief that women are unfit for high political office. Wanting to fit in is not a morally bad desire, but it leads this sexist to adopt a belief that it is unlikely he would have if he cared more about the welfare of women.

As these examples make clear, a belief based on the desires and cares constitutive of ill will or a lack of good will is not based on evidence. In cases of sexist belief caused by ill will, the relevant desire is morally bad. In cases of sexist belief caused by a lack of good will, the relevant desire is not morally bad itself, but it causes a belief that would have been unlikely if the believer had the desires and cares constitutive of good will toward women to a morally sufficient degree.

3.2. Encoded bias

Another mechanism by which sexism can serve as the basis of belief is the one involved in encoded bias.Footnote ¹³ Encoded bias against women starts with culturally prominent descriptions of and prescriptions for women – for example, gendered stereotypes and social narratives. These stereotypes and social narratives are sexist to the extent that, together, they socially disadvantage women and, perhaps more importantly for our purposes, foster an unfavorable shared understanding of (some subset of) women.

Importantly, a stereotype or narrative may be culturally prominent even if it is not widely endorsed. Consider the stereotype that women are hyperemotional. This stereotype may be culturally dominant in a society even if most people in that society do not outright believe that women are hyperemotional – the stereotype may still be prevalent in the news, entertainment, humor, stories, and historical interpretations. Or consider the social narrative that “good” women are nurturers and caretakers. This narrative may be culturally dominant even if most people do not outright believe that women should be nurturers and caretakers.Footnote ¹⁴

When certain stereotypes or narratives are culturally prominent in a society, people in that society will regularly encode them even if they do not endorse them. For example, people may associate women with fragility, or “good” women with being nurturers and caretakers, even though they do not believe that women are fragile or should be caretakers – in fact, even if they are avowed feminists who not only have ample good will toward women but also knowledge of the gender injustices that have given rise to the stereotypes and narratives in question.Footnote ¹⁵ These encoded stereotypes and narratives can be triggered even in people who do not endorse them by encounters with women, or representations of them.

When an encoded stereotype or narrative is triggered, it can illicitly sway the belief formation process: the stereotype or narrative guides reasoning in ways that favor outcomes that align with the relevant stereotype or narrative. For example, encoded stereotypes about women lacking competence or leadership ability may cause members of a hiring committee for an executive position to believe that an application with a man’s name is more impressive than an application with a woman’s name even though both applicants are equally well qualified, or it may cause a voter to believe that there is “just something off” about a woman candidate running for president.Footnote ¹⁶ An encoded narrative about women being homemakers may cause a husband to believe that the division of domestic labor between him and his wife is equal, when in fact she does significantly more.Footnote ¹⁷ An encoded narrative that women should be accommodating may cause a university student to believe that his female professor is a “bitch” when she is simply enforcing the class rules about deadlines spelled out on the syllabus. And an encoded stereotype about women and hysteria may cause a doctor to attribute a female patient’s complaints of chest pain to anxiety, or to underestimate the severity of her pain.Footnote ¹⁸

A belief based on encoded stereotypes or narratives is not based on the evidence. An encoded culturally prominent stereotype about women being hyperemotional, for example, does not, itself, bear on the truth of the beliefs it causes – after all, culturally prominent stereotypes tend to be false and unreliable. The same thing can be said about encoded cultural narratives. Beliefs that are based in part on a stereotype or narrative that has been encoded, then, are not based on the available evidence, sufficient or not.

To sum up: according to the narrow interpretation of the causal claim, a belief is caused by sexism in a normatively significant way to the extent that it is directly based on sexism. The sexist basis of a belief may be a desire that is implicated in having ill will or a lack of good will toward women, or it may be a culturally prominent stereotype or narrative that has been encoded. Either way, the relevant component of sexism guides the belief formation process in ways that favor a particular predetermined outcome: the resulting beliefs are based on sexism and, thus, not on available evidence, rendering them epistemically irrational according to both purism and revisionism.

4.1. Stacking the evidence

Let’s start with the idea that the evidence is stacked in favor of sexist beliefs. The basic thought here is that the world itself is prejudiced (sexist) and so gives evidence for prejudiced (sexist) beliefs. And, since the world itself is sexist and it is the world that provides evidence for our beliefs, then of course there are beliefs that are both sexist and supported by the evidence (at least, supported by evidence that is sufficient to rationalize beliefs in many contexts). Consider the following passages from Basu’s work.

Although Basu’s points are about racism and racist beliefs, they apply to sexism and sexist beliefs as well: sexism causes something to be true about the world – sexism causes it to be the case, for example, that some negative or diminishing property is prominent among women – and thus causes there to be evidence that this is true, and this evidence goes on to be the basis of beliefs. Some of these beliefs are sexist. This is what it means for the world to “stack the evidence in favor of” sexist beliefs. There are a number of cases that give traction to this idea. Consider two such cases.

The beliefs in question are Steve’s belief about Jane and Stacy’s belief about Jenna. These beliefs seem sexist. But they are also supported by the sort of statistical evidence that seems sufficient to rationalize inferential beliefs about things like the weather. For example, suppose that 9 out of 10 women employed in the surgery center are nurses, and that there’s a 9 out of 10 chance that any randomly selected girl scored lower on last year’s standardized math exam than a randomly selected boy. If we know that there is a 9 out of 10 chance of rain tomorrow, it is rational for us to inferentially believe that it will probably rain tomorrow. So it seems that – if the evidential threshold that a belief needs to pass in order to be epistemically rational holds steady across contexts – then the evidence that Steve and Stacy have to support their respective beliefs about Jane and Jenna is sufficient to make those beliefs epistemically rational.

4.2. The broad interpretation

The idea that the evidence is “stacked in favor” of sexist beliefs highlights another potential way that sexism can cause sexist beliefs in a normatively relevant sense – namely, when sexism causes something to be true and, thus, causes there to be evidence supporting this truth, and this evidence goes on to be the basis of belief. As we will see in §4.3, some beliefs caused in this way are epistemically irrational according to revisionism. So, within a revisionist framework, sexism causing a belief in this way has the potential to render that belief epistemically irrational, and this potential is realized when the belief being so caused thereby ends up epistemically irrational. This gives us the broad interpretation of the causal claim: a belief is caused by sexism in a normatively significant way insofar as it is either (i) directly based on sexism or, potentially, (ii) based on evidence that is there because of sexism. Call (i) the direct causal pathway, and (ii) the indirect causal pathway.

The broad interpretation allows for the beliefs in question in The Consultant and The Teacher to be caused in a potentially normatively significant way – namely, in way (ii). That’s because it is ultimately because of sexism that the propositional content of these beliefs is true, and, thus, that there is evidence supporting them: it is because of sexism that it is true, and there is evidence, that girls tend to score worse than boys on math exams (and, thus, that Jenna probably scored lower than Joel), and that women tend to be nurses rather than surgeons (and, thus, that Jane is probably a nurse).

When it comes to the gender achievement gap in math, studies suggest that stereotypes related to gender and mathematical ability negatively affect girls’ performance in competitive testing environments.Footnote ²⁰ When it comes to the disparity between female nurses and female surgeons, there is a lack of empirical literature dedicated to this particular issue. But something like this is a plausible story: a variety of sexist historical practices (e.g., the historical exclusion of women from intuitions of higher education and fulltime careers outside of the home), social norms (e.g., that women are expected to be the caretakers, which prompts them to seek more flexible work compatible with caretaking duties), and social narratives (e.g., those that render high-paying, high-status jobs like surgeon masculine-coded and lower-paying, lower-status jobs like nurse feminine-coded) make it the case that women are disproportionately nurses rather than surgeons.Footnote ²¹ Stacy’s belief about Jenna and Steve’s belief about Jane, then, are both caused by sexism indirectly by way of the evidence. They are, thus, caused by sexism in a potentially normatively significant way. Of course, the next question is this: is the potential realized in these cases? That is, does being caused by sexism in this way render the beliefs in question epistemically deficient?

On purism, beliefs caused by sexism way (ii) are epistemically rational (supposing the evidence that is there because of sexism is sufficient to rationalize beliefs in other contexts). So purism leaves no room for the possibility that a belief being caused by sexism in way (ii) can render that belief epistemically deficient. But revisionism does have room for this possibility: the revisionist account of epistemic rationality allows for it to be the case that the beliefs in question are epistemically irrational despite being based on evidence that (we’re supposing) is sufficient to rationalize beliefs in other contexts. This account also gives revisionists the theoretical resources to explain why sexism is the source of that irrationality. In other words, within a revisionist framework, in some cases, being caused by sexism in way (ii) does render a belief epistemically irrational. It is in these cases that the potential normative significance of being caused by sexism in way (ii) is realized. To understand all this, and how it bears out in cases like The Consultant and The Teacher, we must look further at the account of epistemic rationality at the center of revisionism: moral encroachment.Footnote ²²

4.3. The broad interpretation and moral encroachment

According to popular stakes-based versions of moral encroachment, the evidential threshold that a belief needs to pass in order to be epistemically rational is determined by moral considerations. In particular, when the moral stakes are high – when having the belief is harmful or poses a significant risk of harm – the evidential threshold that belief must pass tends to be high. In other words, the believer needs better or stronger evidence than they would otherwise need in order for their belief to be epistemically rational. Given the high moral stakes, it is not rational to infer that an individual probably has some negative or diminishing property based on statistical information about that individual’s apparent social group, even if it is rational to infer that it will likely rain tomorrow given the same degree of statistical evidence about the weather.

Given moral encroachment, being caused by sexism in way (ii) ends up guaranteeing that the beliefs in question in cases like The Consultant and The Teacher are epistemically deficient within a revisionary framework. In these cases, the belief in question ascribes a negative or diminishing property to an individual woman based on statistical evidence suggesting that most women have that property. These beliefs are situated in a sexist social environment – after all, our world has been heavily shaped by sexism. And when sexism’s impact on our social environment makes it so that some negative or diminishing property is prevalent among a historically marginalized social group, this same impact also makes it especially harmful to ascribe such properties to individuals qua members of that social group. Consider what Basu writes about a case like this involving a seemingly racist belief that ascribes the property of “probably being a staff member” to a Black man in a swanky social club.

When applied to sexism, the thought is this: when sexism indirectly causes a belief about an individual that ascribes a negative or diminishing property to her based on her womanhood, this guarantees that the belief is harmful, given the sexist social environment in which the belief is situated. This, in turn, raises the evidential threshold higher than the belief can pass so long as the believer has only suggestive but inconclusive group-level evidence to support it. So, if the belief in these cases is caused by sexism in the indirect sense, then it is guaranteed not to be epistemically rational, at least in any world like ours.Footnote ²³

Importantly, according to this sort of view, it is the moral stakes related to the relevant harms – not the sheer fact that it is indirectly caused by sexism – that raise the evidential threshold for the belief in question. But the fact that the belief is indirectly caused by sexism implicates the relevant harms in a social environment like ours, thus guaranteeing that the belief is a morally high-stakes one, and thereby raising the relevant evidential threshold. So being caused by sexism in way (ii) makes it so that the beliefs in question are harmful (and, thus, high stakes) in ways that raise the evidential threshold higher than the belief can pass given the suggestive but nonconclusive evidence involved. Given this, being caused by sexism in way (ii) does render beliefs in question in cases like The Consultant and The Teacher epistemically irrational. The relevant potential, then, is realized in these cases: given the broad interpretation, Steve’s belief about Jane and Stacy’s belief about Jenna are caused by sexism in a normatively significant way within a revisionist framework.

That said, it’s important to emphasize that a belief being caused by sexism in way (ii) does not always implicate the relevant harms or render it morally high stakes like it does in The Consultant and The Teacher. So, the relevant potential is not always realized. Whether being caused by sexism in way (ii) renders a belief morally high stakes depends on the content of the belief and the way beliefs with that content interact with the believer’s social context. For an example of harmless, low-stakes beliefs that have been caused in way (ii), consider beliefs about the suffrage movement. Sexism causes certain facts to be the case and, thus, causes there to be evidence that these facts are the case. The historical record of these facts in, say, American history textbooks causes us to have beliefs about the women’s suffrage movement – such as the belief that American women did not have the right to vote until 1920. The belief that American women did not have the right to vote until 1920 in not harmful in our social environment in any relevant sense, even though it was caused by sexism in way (ii). Thus, the moral stakes for this belief are not especially high, and the evidence provided by American history textbooks is clearly sufficient to push the belief over the relevant evidential threshold. Revisionists can say that such beliefs about the suffrage movement are morally and epistemically permissible. These beliefs have been caused by sexism in a potentially normatively significant way on the broad interpretation, but this potential is not realized.

6.1. The revisionist story

Let’s start with the revisionist story. The broad interpretation seems to entail something like this: if evidence exists because of some moral problem (like sexism or other form of prejudice) in the world, then – to the extent that the beliefs that the evidence supports are harmful in part because of the way that those beliefs interact with the world as shaped by the moral problem – we epistemically and morally ought not believe what the evidence supports (i.e., unless the evidence supporting that belief is exceptionally strong, perhaps even conclusive – in any case, much stronger than is needed to rationalize beliefs in many other contexts). For example, Steve and Stacy epistemically and morally ought not to have their respective beliefs about Jane and Jenna: sexism has shaped the world in ways that not only make it true that P is common among G but also make it harmful for believers to ascribe P to J qua member of G.Footnote ²⁴ And, importantly, this holds even though the morally problematic source of the evidence does not affect the quality of the evidence qua evidence – that is, the moral problem does not disrupt the connection between the evidence and the truth of the relevant propositional content.

6.2. A purist story

Now let’s turn to what purists can say here. A good place for the purist to start is by parsing out what, exactly, it means for evidence to have a sexist provenance.

In cases like The Consultant and The Teacher, the relevant evidence is the statistical evidence about P’s commonness among G. The fact that this evidence has a sexist provenance means that it is because of sexism that it is true, and there is evidence, that girls tend to score worse on standardized math exams than boys, and that most women employees at the surgery center are nurses. Given this, perhaps the surrounding beliefs – beliefs about why the propositional content of the belief in question is true – are central to a plausible purist story. The purist can say that the fact that the evidence about an unfortunate truth has a sexist provenance bears not on whether it is epistemically permissible to believe the propositional content capturing that truth, but rather on the permissions of surrounding beliefs. More specifically, it bears on what is epistemically permissible to believe about why the proposition is true – in cases like The Consultant and The Teacher, why it is that P is common among G (and, thus, why J is probably P).Footnote ²⁵ So, on this purist story, in light of the fact that the relevant evidence has a sexist provenance, the only rational thing to believe is that P is common among G because G has been subject to a long history of sexism.

This story can make sense of the epistemic significance of the sexist provenance of the relevant evidence. It is epistemically important that our beliefs reflect reality. But true beliefs about certain facts may distort reality rather than reflect it when they are held in isolation, or otherwise not couched in true surrounding beliefs.

This should not be surprising. We all know that the meaning or significance of facts can be distorted when they are taken out of context, even when what is said is technically true. This can happen for a couple of reasons. First, a fact that is not properly contextualized may capture only a small unrepresentative bit of reality. For example, consider a media outlet that regularly reports on instances of voter fraud or shark attacks without also citing the very low rate at which they occur. Second, and most relevant here, a fact that is not properly contextualized may invite a misunderstanding of important surrounding factors. For another media example, consider the following headline from a Canadian newspaper: Trudeau devotes a quarter of his time to “personal days.”Footnote ²⁶ Although it is technically true that 25 percent of the days since Trudeau took office were personal days, important contextualizing information is missing: most of the days Trudeau took off were weekends and statutory holidays, and the 25 percent rate of personal days is well below the 34 percent rate of the average Canadian. When presented on its own, the information in the headline invites readers to think that the reason Trudeau devotes 25 percent of his time to personal days is that his priorities are out of whack, and that he is not a good steward of a prime minister’s time.

Similarly, a true, evidence-backed belief that is not properly contextualized may lead to a distorted view of reality. Consider the group-level and individual-level beliefs in The Consultant and The Teacher. When not properly contextualized, these beliefs invite a misunderstanding of why P is common among G – in particular, they invite the common misunderstanding that scoring worse than the average boy on the math exam or being a nurse rather than a surgeon is common among girls/women because of some inherent flaw or deficiency characteristic of girls/women (e.g., low competence or intelligence or skill). After all, like many historically marginalized social groups, women have long been negatively characterized in ways that are viewed as causally responsible for properties common among them. And so this mistaken explanation of why P is common among G is readily available and culturally prominent, poised to fill in an explanatory vacuum if the believer does not already understand the correct explanation. When not properly contextualized within an understanding that it is because of sexism – not some inherent characteristic deficiency or flaw – that P is common among G, having the relevant group-level and individual-level beliefs makes us susceptible to a distorted broader view of reality by putting us in a position to form false beliefs about why P is common among G.

The upshot is that true, evidence-backed beliefs about bare statistical facts about women/girls and about individuals inferred directly from those statistical facts may end up leading to a distorted broader view of the relevant slice of reality by inviting false (and harmful!) beliefs about important surrounding factors. So, if the sexist provenance of the relevant statistical evidence makes it epistemically impermissible for those who are cognitively sensitive to this provenance to have such beliefs, then the sexist provenance of the evidence is, indeed, epistemically significant. It is epistemically significant to the extent that cognitive sensitivity to it discourages and renders irrational a broader set of beliefs that distort the relevant bit of reality and promotes and rationalizes ones that reflect it. So, it seems, this purist story can explain the epistemic significance of the sexist provenance of the relevant evidence.

Although the focus so far has been on epistemic significance, I think it is worth noting that this purist story can also explain the moral significance of the sexist provenance of the relevant evidence. The idea here is that understanding why P is common among G has morally significant motivational and affective implications. The affective and motivational dimensions of understanding are discussed at length in relation to moral propositions in the literature on moral testimony – understanding why it is wrong to lie or why it is good to donate to charity can be central to (re)acting in morally appropriate ways in a variety of morally significant contexts.Footnote ²⁷ But it is widely agreed that understanding some nonmoral propositions can also be central to (re)acting in morally appropriate ways. To borrow an example from Laura Callahan (Reference Callahan2018: 444), I might need to understand the reasons why my partner is late picking up our child from daycare in order to react appropriately – if he has a legitimate excuse, it would be inappropriate to be angry at him.

Similarly, I may need to understand why some negative or diminishing property is common among girls/women in order to react appropriately to an individual girl or woman who probably instantiates that property. For example, a doctor may need to know why women’s complaints of severe pain tend to be met with skepticism – if it is because healthcare professionals are prone to illicitly view women as “hysterical” or “dramatic,” then skepticism about the level of reported pain would be inappropriate. Or an admissions committee for an elite STEM-focused charter school may need to understand why girls tend to score lower on average on math exams in competitive testing environments in order to give the dossiers of girl applicants a fair evaluation.

To put another spin on the same idea, if I misunderstand why P is prevalent among women/girls, I may be liable to (re)act in ways that are inappropriate given the truth. For example, if a teacher inaccurately believes that girls tend to score worse on standardized math exams than boys because girls are just inherently worse at abstract thinking, then she might be liable to ignore or downplay the mathematical acumen of the girls in her class.

So, if the sexist provenance of some evidence bears on the epistemic permissions of the surrounding beliefs about why P is common among G – rendering permissible only those beliefs attributing the property’s commonness to sexism, and not to something inherently deficient or negative about girls/women – then cognitive sensitivity to this fact promotes further sensitivity to morally relevant features of the social environment. This, in turn, enables and encourages (re)actions and attitudes that are appropriate in light of morally significant features of the bit of reality that is captured in part by propositional content of the relevant group-level and individual-level beliefs. The culturally prominent misunderstanding – that P is common among G because of some inherent or characteristic flaw or deficiency – is liable to give rise to actions and attitudes that are morally inappropriate.

6.3. Concluding thoughts

So now we have two different stories about the normative significance of the fact that evidence about an unfortunate truth has a sexist provenance. According to the revisionist story, in cases like The Teacher and The Consultant, the believers epistemically and morally ought not believe that J is probably P, given the sexist provenance of the supporting evidence and the related harms of so believing. According to the purist story, the sexist provenance of the relevant evidence does not rule out permissibly believing that J is probably P; rather it bears on the permissions of the surrounding beliefs, rendering it epistemically and morally important for the believers to understand that it is because of sexism that P is common among G.

I have suggested that it is significant in itself that purists have a plausible story to tell: purists can resist challenges stemming from worries that they lack something to say about something so important as the normative significance of sexism’s impact on the world that gives us evidence for our beliefs.

Even so, the question remains: which story is more plausible? To conclude, I want to consider how to go about answering this question. My goal is not to defend a particular answer favoring one story or another. Instead, I want to develop a way of arbitrating between the two stories, on their own terms.Footnote ²⁸

It will be helpful, then, to consider a case that clearly has all the normatively relevant features picked out by both stories: namely, a case in which evidence supporting an unfortunate truth has a sexist provenance, this evidence is the basis of some belief that ends up being harmful in part because of the way that the belief interacts with the sexist social environment in which it is situated, and the believer understands that it is ultimately because of sexism that the belief is true. Each story will render different verdicts about such a case, and we will have some reason to accept whichever story seems to render the most plausible verdicts. To that end, consider the following variation on The Teacher, adapted from Brinkerhoff (Reference Brinkerhoffforthcoming).

According to the revisionist story, Informed Stacy’s belief about Jenna is epistemically and morally problematic. After all, the evidence supporting Informed Stacy’s belief about Jenna has a sexist provenance, and this belief poses many of the same harms as Stacy’s belief about Jenna in the original case, in part because of the way it interacts with Stacy’s sexist social environment. So, according to the revisionist story, this makes Informed Stacy’s belief about Jenna epistemically and morally wrong – her statistical evidence is not strong enough to rationalize it.

The purist story renders different verdicts. On the purist story, Informed Stacy’s belief about Jenna may well be epistemically and morally in the clear, so long as it is based on the relevant statistical evidence that supports it. What’s more, the purist story recognizes that it is both epistemically and morally good that Informed Stacy understands why it is that girls tend to score worse on standardized math exams than boys – there would be something morally and epistemically deficient about Stacy’s broader doxastic state if she lacked this understanding while believing that P is common among G (and, thus, that J is probably P).

If it is morally and epistemically wrong for Informed Stacy to believe that Jenna likely scored worse than average despite her understanding, then we have reason to accept the revisionist story along with the broad interpretation. On this take, the sexist provenance of the facts and evidence featured in the relevant cases bears directly on the permissions of the belief about the fact itself; the understanding in which the belief is couched does not dispel its normative deficiency.

If, instead, Informed Stacy’s belief about Jenna is (or, at least, may be) morally and epistemically in the clear despite the morally high stakes that render it impermissible with a revisionist framework, then we have some reason to favor the purist story and the narrow interpretation. On this take, the sexist provenance of the facts and evidence featured in cases like The Consultant and The Teacher bears on the permissions of the surrounding beliefs rather than the belief about the fact itself.

In sum: at the heart of the disagreement between the broad and narrow interpretations of the causal claim are two different stories about the normative significance of the fact that evidence supporting some unfortunate truth has a sexist provenance. In order to decide which story to favor, it is helpful to consider variations on cases like The Consultant and The Teacher in which the belief in question is couched in an understanding of sexism’s impact on the socioeconomic environment. Which story is most plausible will depend, in part, on our verdicts about the normative status of the beliefs in question – and the surrounding beliefs – in these variations.Footnote ²⁹