Social Group Moral Encroachment

Abstract

According to moral encroachers, the moral stakes of a belief partly determine how much evidence we need for the belief to count as knowledge. This view concerns the beliefs of individual believers. In this paper, I argue for a social group version of moral encroachment: dominant groups, such as white people or men, need to have more evidence than the marginalised in order for some of their beliefs to constitute knowledge. I argue for this claim in three steps. First, I spell out the group moral stakes involved – the harms dominant knowers cause the marginalised and the knowledge economy. Second, I show off the theoretical benefits of having the notion of social group moral encroachment at our disposal: it can be an invaluable tool for decolonial and feminist epistemologists if they want to avoid the relativistic ring to their views (a ring that tends to put off many well-wishers). Finally, I start on a positive account of social group moral encroachment by addressing a potential puzzle and responding to objections.

1. Introduction

Suppose that a mostly white audience is watching a TV series in which a selection committee is deciding on who should fill the open post at a prestigious philosophy department. A Black and a white candidate – Dr Ncube and Dr Smith – are tying for first place. Their qualifications are evenly matched; their CVs look near identical and equally impressive; there is only one difference between them (apart from their race): although both have summa cum laude degrees, Smith's is from an Ivy League university, while Ncube's is from a historically disadvantaged one. As the audience sees the CVs and watches the qualifications discussed by the selection committee, it forms the belief that Smith is most likely the better philosopher of the two, based on this one difference. Call this case Ncube vs Smith.

Now let's pause the series before any hiring decisions have been made and double click on this belief. It has the following two features: (i) it is held by members of a socially dominant group (white people), whose dominance is sustained by unjust structures; and (ii) as I will argue, the structural moral stakes are high: if the belief is false, it will not only wrong and harm Ncube, but will perpetuate the structural racist patterns of economic and epistemic oppression that are responsible for his not having gone to an Ivy League university in the first place.

In this paper, I argue that beliefs with these two features motivate developing a group version of so-called moral encroachment – the idea that the moral stakes of a belief affect how much evidence is enough for the belief to constitute knowledge (e.g., Basu 2019a, 2019b; Bolinger 2020b; Moss 2018; Schroeder 2018). In Ncube vs Smith, because of the dominant status of the group to which the belief belongs and the potential structural harms that beliefs like this perpetrate, the evidence on the basis of which the audience forms this belief is insufficient to justify the belief.

I proceed as follows. In section 2, I outline the somewhat disparate bunch of views that travel under the banner of ‘moral encroachment’, and pin down the one that is relevant to my argument. In sections 3 and 4, I motivate a group version of this view and discuss the moral stakes that are involved in it. In section 5, I argue that unstructured groups like dominant groups can have collective beliefs. In section 6, I motivate further the need for a notion of social group moral encroachment by appeal to a major theoretical benefit of having this notion: it can help liberatory epistemologies shake off some of their relativism, a relativism that is incompatible with the absolute imperative to end oppression as well as responsible for much of the resistance to these epistemologies. In sections 7 and 8, I start on a rough sketch of social group moral encroachment by addressing a potential puzzle and some objections.

2. Moral encroachment

The basic idea of moral encroachment is simple enough: moral considerations can affect whether a belief is justified or counts as knowledge. Here is the classical example that motivates this thought:

In the summer of 1995, historian John Hope Franklin – author of From Slavery to Freedom – received a call from the White House informing him that President Clinton planned to present him with the Presidential Medal of Freedom, the nation's highest civilian honor. On the night before the award ceremony, Franklin hosted a dinner for a small group of friends at the Cosmos Club, a Washington DC social organization of which he was a member. He writes: “It was during our stroll through the club that a …woman called me out, presented me with her coat check, and ordered me to bring her coat”. (Gendler 2011: 35)

The intuition that moral encroachers have about this example is this.¹ Most members of the Cosmos Club are white, and all staff are Black. So, the statistical evidence strongly supports the woman's belief that Franklin is staff. Nonetheless, there is something morally wrong with her forming the belief, and this moral wrongness undermines the belief's epistemic status – its justification (Basu 2019a), eligibility for knowledge (Moss 2018), or rational permissibility (Bolinger 2020a). Here is one way of capturing this thought a bit more precisely (e.g., Moss 2018):

Basic Idea

If B¹p and B²p are beliefs with the same content and the same evidential support but differ in epistemic status due to some moral consideration, then we have a case of moral encroachment.

Although the basic idea is simple enough, the literature on moral encroachment has by now generated many subtle variations on this theme. It is beyond the scope of the present paper to review all the nuances, but a rough guide should help situate the view on which I model social group moral encroachment. Using Renée Bolinger's (2020b) helpful map, moral encroachment views can be distinguished along three axes (amongst others²).

The first concerns the question of whether the relevant moral considerations attach to the belief itself, on the one hand, or to the way it was formed and/ or how it ‘orient[s] actions’, on the other (Bolinger 2020b: 14–16).³ Proponents of the former view – ‘Radical Encroachment’ (ibid.; Fritz 2020: 3054) – think that the mere fact of holding a belief with a certain content wrongs the object of the belief (in this case Franklin), regardless of how the belief was formed or how it would affect the believer's actions. In contrast, what Bolinger (2020b: 19) calls ‘Modest Encroachment’ holds that the wrongness attaches to the way the belief was produced or affects action.⁴ The view on which I model social group moral encroachment is of this latter, modest, kind.⁵

A second question that divides moral encroachers is whether moral considerations operate directly as epistemic reasons or whether they merely set the threshold for how much evidence is enough for a belief to be justified and/ or classify as knowledge.⁶ On the former view – ‘Hard Encroachment’ (ibid.: 12) – the moral wrongness of believing Franklin to be an attendant counts as a reason against the woman's holding the belief in exactly the same, epistemic, sense as the statistical evidence counts in favour of it; and the moral epistemic reason trumps the statistical. In contrast, on the latter view – ‘Moderate⁷ Encroachment’ (ibid.) – moral considerations cannot act as epistemic reasons. Only evidential reasons can be epistemic on this view; but moral considerations help determine how much evidence is enough for justification. In our example, the moral wrongness raises the bar on evidential support, such that mere statistical evidence, no matter how strong, is insufficient to justify the woman's belief that Franklin is staff. The group view I will defend here will be modelled on this latter, moderate, encroachment.

A third question that can divide moral encroachers is whether moral considerations only raise the evidential threshold or can also lower it. On the former view – ‘Cautious Encroachment’ (ibid.) – moral considerations can't make it easier to obtain justification; while on the latter – ‘Robust Encroachment’ (ibid.) – they can. The group encroachment that I will defend is of this latter, robust, kind.

The moral encroachment view that is modest, moderate, and robust, then goes something like this:

Moral Encroachment

B¹p and B²p are beliefs with the same content and the same evidential support but differ in epistemic status due to the differing moral stakes involved in them, where:

1. i. The epistemic status is affected because of the way the belief was produced or orients action (Modest).

2. ii. The moral stakes do not act as epistemic reasons for/ against the belief but affect the threshold for how much evidence is enough for the belief to be justified (Moderate).

3. iii. The evidential threshold can be raised or lowered by the moral stakes (Robust).

This view seems to me the most defensible version of moral encroachment in light of other examples in the literature that are supposed to motivate moral encroachment yet have been found unconvincing.⁸ Consider the following case that Rima Basu and Mark Schroeder offer:

suppose that you have struggled with an alcohol problem for many years, but have been sober for eight months. Tonight you attend a departmental reception for a visiting colloquium speaker, and are proud of withstanding the temptation to have a drink. But when you get home, your spouse smells the wine that the colloquium speaker spilled on your sleeve while gesticulating to make a point, and you can see from her eyes that that she thinks you have fallen off of the wagon. (Basu and Schroeder 2019: 182)

Basu and Schroeder argue that your partner's belief has wronged you. But this verdict is only plausible, to my mind, if we adopt Modest, Moderate, and Robust. First, take Modest. If we were radical encroachers, we'd have to hold that your partner wrongs you just by holding the belief that you've fallen off the wagon. But given that by stipulation you are an alcoholic, this seems implausible. It seems far more plausible to think, instead, that given your relationship and the high stakes of distrusting you, the wrong lies in her forming the belief so automatically, without giving you a chance and investigating further, i.e., in the way she adopted the belief rather than the belief itself. This is just what Modest says.

Next, take Moderate. If we are hard encroachers instead of moderate ones, we'd be forced to argue that the trust your partner owes you directly militates against her adopting the belief that you have been drinking. But clearly, and given that you are an alcoholic, it is far more plausible to think that what she owes you is to collect more evidence. At the very least she could have asked you if you'd been drinking before forming the belief! In other words, the moral stakes have raised the bar for how much evidence she should have before her belief counts as justified. This is just what Moderate says.

Finally, take Robust. If we only allowed moral considerations to raise the threshold for evidence, then we'd be forced to say that the trust your partner owes you doesn't lower the evidence-threshold for the belief ‘They have not gone off the wagon’. This, although technically compatible with the injunction to raise the bar for not believing that you have been drinking – since that injunction can mean that you should suspend belief – is still a less natural position to hold towards one's partner than simply believing they have not been drinking. This is what Robust says.

These rough and ready thoughts should hopefully be enough to give an idea of both the kind of moral encroachment I have in mind and that it is at least in some varieties a plausible view. But the unpersuaded are welcome to hear the main claim of the paper as a conditional: if some form of Moral Encroachment is correct, then we'd do well to also think about this phenomenon at the social group level. Arguing for this claim as I do here will mean effectively adding to the map another two questions on which a moral encroacher must decide to refine their position: what kind of agent is the owner of the belief that is being encroached (group vs individual); and what kind of moral stakes are involved in it (structural vs non-structural).

3. Social group moral encroachment

Here is an intuitive argument from individual to social group moral encroachment. First, notice that once we accept individual moral encroachment, we should (trivially) accept some version of group moral encroachment on the assumption that groups can have beliefs (more of which in section 5).⁹ The current argument, however, doesn't concern any old group, but social groups in particular. Why should we think their beliefs are subject to moral encroachment?

The short answer is: because some groups cause significant moral harms and wrongs, and they do so through certain beliefs that are typically held by members of this group. Take white people in Apartheid South Africa or Jim Crow US. Many of the moral harms that such groups caused were through their, and their members', actions. But some very important harms were caused through their beliefs – that Black citizens are inferior to white; that they don't deserve the same kind of schooling due to their lower intelligence; that they should not mix with white citizens in order not to corrupt them, and so on. These beliefs oriented the believers' actions in familiarly vicious and harmful ways.¹⁰

If this is right, and we accept individual moral encroachment of the kind I just outlined, then it would make sense to think that the moral risks of these beliefs raise the bar for how much evidence is enough to justify them. And if we accept this, we accept that there is such a thing as social group moral encroachment. Its distinguishing marks are that the relevant beliefs are held by a dominant group and that because of the dominance of this group, the beliefs' moral harms are structural. Conversely, if we don't accept the possibility of social group moral encroachment, it becomes difficult to see how we can hold individual perpetrators responsible, since given the structural nature of these wrongs and the racist environment in which they find themselves, individual perpetrators may simply have misleading evidence for their beliefs. Social group moral encroachment helps us overcome this unpalatable consequence.¹¹

To keep it analogous to the individual case, let me characterise it as follows:

Social Group Moral Encroachment

B¹p and B²p are group beliefs, with the same content and the same evidential support, but differ in epistemic status due to the fact that the groups are situated differently and to the consequent structural moral stakes involved in the beliefs, where:

1. i. The epistemic status is affected because of the way the group belief was produced or orients the group's actions (Modest).

2. ii. The structural moral stakes do not act as epistemic reasons for/ against the belief, but affect the threshold for how much evidence is enough for the belief to be justified (Moderate).

3. iii. The evidential threshold can be raised or lowered by the structural moral stakes (Robust).

I have italicised the additions to the original individual version of moral encroachment in order to highlight the distinctive features of social group moral encroachment – that the moral stakes are structural and the beliefs involved are group beliefs. This is also how I described the original Ncube vs Smith case: beliefs like these were held by dominant groups and the moral stakes concerned the sustenance of the very structures that allowed for the group's dominance.

But these distinctive features also raise complications for the simple argument I just offered in favour of social group moral encroachment. First, we need to articulate the exact moral stakes involved in ways that distinguish them from traditional moral encroachment. Second, even if we accept that highly structured groups like a government can have collective beliefs, we aren't forced to accept this for very loose groups of the kind I have in mind here. I address these complications in the next two sections, respectively.

4. The structural moral stakes in group encroachment

Let us return to Ncube vs Smith and be a little more specific about the beliefs that the audience holds. Two beliefs were implicitly mentioned in the introduction:

1. (1) Smith is most likely a better philosopher than Ncube.

2. (2) The reason for (1) is that Smith has an Ivy League degree while Ncube's is from a previously disadvantaged university.

Note that the inference in (2) rests on a strong statistical generalisation: Ivy League universities are typically more likely to produce better philosophers than previously disadvantaged institutions. But I will now argue that the moral stakes involved in believing (1) are too high for the belief to be justified by such evidence alone. Just as in the individual, Cosmos Club case, statistical evidence in such high stakes situations is never good enough on its own; we need other kinds of evidence as well.¹²

What exactly are these moral stakes? Start with thinking of the inevitable decision that will follow the belief if its owner is the selection committee: Smith will be hired. This decision is clearly unjust on distributive justice grounds: by stipulation, the candidates are otherwise completely evenly matched, so the just thing to do would be to hire Ncube given how badly represented marginalised philosophers are. So, one of the moral stakes of beliefs of this kind is that they orient action in ways that block the right distributive justice outcome. After all, who could argue with hiring the best philosopher? This blockage serves to enforce white privilege and structures that systematically favour white people: if we keep hiring Ivy League white candidates, our universities will always have a self-enforcing bias in favour of such candidates. This shapes knowledge production in obviously unjust and epistemically vicious ways by enforcing continued exclusion of the marginalised and their epistemic resources.

But what does our white audience have to do with this? After all, they are just watching; they have no say in the hiring decision. I have chosen the audience as the right owner of the belief precisely for this reason – to focus our minds on the belief itself without the background noise of intuitions about the specific decision that will be made on its basis. So, how is this decision relevant to the audience and why should we think of their belief, which is irrelevant to the decision, as particularly problematic?

The answer is that when members of a dominant group hold certain beliefs, even when they are not directly involved in a particular decision, the beliefs cause harm. This is because such beliefs – when held by a dominant group – enforce narratives that tend to maintain the dominant group's unfair advantages and dominance. One obvious such narrative is the one involved in what Charles Mills (2007) calls ‘white ignorance’ and Gaile Pohlhaus (2012) more generally dubs ‘wilful hermeneutical ignorance’. These are forms of active and pernicious ignorance: they help the dominant person maintain their sense of innocence while enjoying the plentiful fruit of their non-innocence – a long history of slavery, colonialism, and continuing exploitation. They do so through foregrounding the importance of individual achievement and erasing race (with its attendant socio-economic inequalities) from explanations of who ends up with what education. Such narratives are essential to sustaining these very inequalities.

Once this sort of narrative flourishes in a group, it doesn't matter who is the de facto owner of any particular belief of this kind. As long as a good chunk of the group hold such beliefs, the damage to both persons and the knowledge economy will be done. This is guaranteed simply in virtue of the vicious cycle such narratives generate: first, we call a particular kind of person the best philosopher and a particular kind of institution the best institution (they just ‘happen’ to be both largely white and male); then we use that philosopher and that kind of institution to set the criteria for what counts as the best institution and the best philosopher (they also ‘happen’ to be largely white), and on it goes.

Notice that the point here isn't that the belief itself was formed out of white ignorance. Had this been so, we would not have a case of moral encroachment, since the belief would be unjustified for standard reasons – it would have been formed due to bias. Moreover, by saying that Smith and Ncube are evenly matched in their qualifications, I have specifically described the case in a way that rules out any standard epistemic injustices that are typically responsible for the low Black representation in our discipline, such as for instance testimonial injustice of various kinds (Davis 2021). Such injustices would also result in standard lack of justification, since they would distort in biased ways the evidence that Ncube and Smith are equally well qualified. In contrast, the belief, as I have described it, is formed on what would otherwise be good grounds (strong statistical evidence), but because of its consequences when held by a dominant group – maintaining unjust structures – those grounds are not good enough for justification in this case.¹³

Let me stress that the envisaged harms are dependent on the dominant group's holding the belief in question, where both the dominance and ‘groupiness’¹⁴ are essential, since it is these that enable mere beliefs to have such dramatic structural consequences. Neither the oppressed as a group nor individuals of any group on their own have such powers. One might worry, however, both about the groupiness of dominant groups and about the plausibility of crediting them with beliefs. In the next section, I address these worries by way of earning the label ‘group belief’ for the beliefs involved in social group moral encroachment.

5. Dominant group beliefs

By ‘group’, social ontologists mean an entity that is structured and agent-like in the sense that it can be attributed quasi-actions and responsibility for those actions. How exactly we understand such agency is a delicate matter.¹⁵ But at all events, corporations, governments, and other bodies with a constitution are the least problematic quasi-agents of this kind (French 2020). One way of cashing out their agency and responsibility is in terms of the group's individual members (e.g., List and Pettit 2011); another is in terms of its operative members (e.g., Pauer-Studer forthcoming).

Social epistemologists have extended these thoughts to epistemic agency. Although it may sound odd to attribute belief to a group – after all, beliefs are mental states, and a group doesn't have those – there are good reasons for crediting at least very structured groups with beliefs.¹⁶ Jennifer Lackey (2021) offers the most intuitively compelling one to my mind: groups of this kind can lie; but lying that p entails that you believe not-p. Think back to the 2015 Volkswagen scandal, for instance, when Volkswagen was accused of lying about the carbon-emissions controls it fitted into its cars. If Volkswagen lied, then it believed (indeed knew) otherwise.

How exactly we understand such group belief is again a disputed matter. So-called summativists think that a group belief is simply a function of (a certain portion of) the members' beliefs; while non-summativists think that a group can have a belief qua group, independently of what its members believe in their private capacity.¹⁷ One persuasive argument in favour of non-summativism invokes the so-called doctrinal paradox (List and Pettit 2002). It shows that if we try to aggregate the judgements of individual members, we may get an irrational collective judgement (as a result of an invalid inference), while the judgements of each individual remain rational (since they are the conclusions of valid inferences). Thus, a group belief cannot be the same as that of its members' beliefs. Although I find this argument convincing, I will not insist on either view here. As long as it is accepted that a group can have beliefs, the argument of this paper is safe.

But even with these concessions in place, we still don't have enough for my argument that the moral risks of certain dominant groups' beliefs raise the justification-threshold for these beliefs. The problem is that all of these thoughts only get us to claims about the beliefs of structured groups such as corporations or governments. But the dominant groups that Social Group Moral Encroachment concerns aren't structured like this. Such groups do not have spokespeople; nor is membership in them endorsed in the way membership in a corporation or a government is. I have argued elsewhere (Mitova 2022) that despite these apparent infelicities, social identity groups can, in fact, be credited with enough agency for beliefs (and reasons for beliefs) to be attributable to them. I will not rehearse the arguments here, but just give two intuitive considerations in favour of this claim.

First, the reason why some social identity groups can be attributed enough agency to be properly called groups is because of the active ignorance they cultivate and the active harms to others for which they are responsible. I have already discussed both in section 3. Let me say a little more on the active ignorance point. When Mills introduces the notion of white ignorance, he treats it as a form of ‘motivated irrationality’ where the motivation is ‘white group interests’ (Mills 2007: 34, my italics). The ‘motivated’ here captures the appropriateness of thinking of this ignorance as intentional; and the ‘group’ bolsters the point that the intention attaches to the group. Similarly, when José Medina extends the discussion of white ignorance more generally to ‘the epistemic vices that privileged subjects develop to protect themselves’ (2013: 33), he is clear that what is at stake is privileged subjects as a group. Thus, he describes one of these vices, intellectual laziness, as ‘socially produced and carefully orchestrated lack of curiosity’ (ibid.: 31, my italics). That is, both authors think of these negative epistemic states as intentional and as attaching to privileged groups. (This is not to say that individuals aren't to be held responsible, but just to acknowledge the collective and structural dimension to the injustices that are perpetrated through such active ignorance.) If this is right, then it is not as implausible to treat dominant groups as groups proper and, by extension, to credit them with group belief.

Second, these thoughts also point to a more direct argument for why it is appropriate to credit such groups with group belief: if we are happy to attribute ignorance to a group – which many social epistemologists and epistemologists of ignorance are (e.g., Alcoff 2017) – then we are committed to the possibility of such groups holding beliefs. We do not call flowers or dogs ignorant; we only call ignorant those who can overcome their ignorance through acquiring correct beliefs or knowledge.¹⁸ Thus, crediting a group with group ignorance entails that this group can have group belief.

So, we have at least prima facie reasons to think that if we accept moral encroachment, we should accept a group analogue, as characterised by the conditions in Social Group Moral Encroachment. This characterisation, of course, needs much polishing and precisifying. I start on this work in sections 7 and 8. But before I do so, I show that doing this work is worthwhile, by gesturing at the theoretical usefulness of having this group notion of moral encroachment.

6. Theoretical benefits

In this section, I argue that the notion of social group moral encroachment is a useful tool for liberatory epistemologies, such as decolonial and feminist epistemology. Such epistemologies often sound unnecessarily relativistic, a relativism that both puts off well-wishers and sits uneasily with the absolute imperative to end oppression that lies at the heart of liberatory projects. I thus take it that a tool that can help them avoid such relativity wears its value on its sleeve.

Let me just mention two ways in which relativity arises – implicitly, in the rationale for epistemic liberation that such epistemologists offer, and explicitly, in their rejection of an objective conception of knowledge.

The first is most obvious in decolonial theorists' motivating the necessity to decolonise by appeal to the epistemicide that colonialism has committed and then arguing that the only way to redress it is by treating all knowledge systems as (in some undefined sense) epistemically equal.¹⁹ There are many statements of this kind of view in the decolonial literature. Here is just one from the debate on decolonising academia in South Africa:

in advocating for the reversal of epistemicide, we necessarily seek to place indigenous knowledge systems of the conquered peoples of South Africa on the same level of parity with other epistemological paradigms in order to achieve both formal and substantive equality. (Lebakeng et al. 2006: 76, my italics)

The thinking is that one of the important things of which colonialism has robbed the colonised is their epistemic authority – the right to say what counts as knowledge, what count as interesting research problems, and generally the right to theorise the world from their own socio-geographic location (Mitova 2021). Thus, restoring this epistemic authority involves crediting the knowledge systems of the colonised with (at least) the same epistemic status as those of the dominant Global North.

This move is also seen in the rationale for other liberatory epistemologies, such as feminist ones, where scholars argue that oppressed communities of all kinds are suffering ‘contributory injustice’ (Dotson 2012) – the unjust marginalisation of the epistemic resources that the oppressed have developed to make sense of their experiences and to combat their oppression. Such resources don't get the right uptake in the mainstream, dominant, knowledge economy, hence blocking the oppressed from contributing to, and influencing, knowledge production in ways that would end their oppression. Thus, one of the requirements of all epistemic liberation is to undermine the self-arrogated and unearned hegemonic epistemic authority of the dominant knowers and to restore equal epistemic authority to colonised and other oppressed knowledge systems.

The second source of relativism for liberatory epistemologies lies in their rejection of what Walter Mignolo (2009) calls ‘the Zero Point’ – the Global North's (Cartesian) conception of knowledge as the property of a disembodied, dislocated subject. Such a conception, both decolonial and feminist theorists argue, purports to be neutral and objective, but simply reflects the socio-political location of the Global North (for decolonial theorists, e.g., Mignolo 2009) and the white cis straight male (for feminists, e.g., Code 1993). Thus, both urge a move from thinking about ‘knowledge’ in the singular to ‘knowledges’ i.e., a ‘[r]ecognition of various forms of knowledge and knowing’ (Ndlovu-Gatsheni 2018: 5).

Both of these rationales for restoring equality of epistemic authority to the marginalised sound obviously attractive until we notice that there will be cases in which different knowledge systems or sets of epistemic resources will dictate competing diagnoses of, and solutions to, the same problem. Thus, for example the diagnosis of a western medical doctor and a traditional healer will at least sometimes be in tension with each other, and their proposed cures incompatible. If we insist that all knowledge systems have equal epistemic authority, we are committed to the view that in conflict cases of this kind, there is no fact of the matter which diagnosis is correct or which cure will fix the problem.²⁰

This kind of relativism is at least prima facie incompatible with liberatory epistemologies' commitment to the absolute, non-relativistic wrongness of colonial, sexist, racist, and other forms of, oppression. Moreover, liberatory epistemologists are often concerned to establish the objectivity of the conception of knowledge that they propose as an alternative to the oppressive one (Code 1993; Harding 1991). One source of such arguments is the idea that the oppressed are in possession of two sets of epistemic resources – one to understand their own oppressive experiences and one to survive in the world of the oppressor, in which they are forced to live (Pohlhaus 2012). This, the objectivity thought goes, gives them a more comprehensive, and hence objective, view of reality.

Suppose we find it compelling that a strong liberatory epistemology should make room for the recovery of the epistemic authority of the oppressed without thereby lapsing into relativism. Then the current notion of social group moral encroachment can help us walk this tightrope. On the one hand, the question of whether a particular group knows a proposition (of a certain kind) can never be answered without thinking of the structural stakes involved, i.e., how that group's belief in that proposition affects the structural injustices in which we are imbedded. On the other hand, this is not done at the sacrifice of objectivity: both the notion of evidence featuring in this conception of knowledge and the structural moral considerations that determine how much evidence is enough are thoroughly objective. There is a fact of the matter, in other words, as to what counts as evidence for what, how grave the moral consequences of holding a belief are, and how much they push up or down the threshold for how much evidence is enough for knowledge.

Thus, the group notion of moral encroachment I have proposed here has the theoretical advantage of allowing both for liberatory concerns to shape our conception of knowledge and for this conception to be objective.

7. A puzzle

These thoughts hopefully suggest that the notion of social group moral encroachment proposed here is worth taking seriously and developing. In this and the next section, I start on a rough sketch by first addressing a potential puzzle (this section), and then some objections (section 8).

The puzzle is this: the proposal entails that the marginalised need less evidence in order to know certain propositions than do the dominant. (As always, the relevant propositions are the ones that, when believed by dominant groups have high structural moral costs.) This might seem puzzling for at least two reasons.

First, as already discussed, the marginalised have more kinds of epistemic resources at their disposal than the dominant. In particular, they have a whole set of resources that dominant knowers don't. The marginalised are, thus, in a position of experts with respect to the one set of resources and, hence, in a position of experts with respect to the very propositions at issue here. Given how much we rely on experts, social group moral encroachment should imply that they need more, not less, evidence: the extra epistemic responsibilities experts bear should raise the evidence-bar, making them need more evidence than the non-expert dominant knowers.

Second, it might be thought that the proposal should imply that the marginalised need more evidence in virtue of the even greater damage that their believing certain propositions would cause given that they are experts. Think of internalised racist or sexist beliefs. Presumably, given their higher epistemic authority on such beliefs, when the marginalised hold them, they would entrench unjust structures even more deeply than when the oppressor holds them. But if so, this danger would raise – not lower – the moral stakes, hence raising the evidence threshold for the marginalised holding such beliefs.

These ways of articulating the puzzle are related, since they both rely on the greater epistemic authority of the marginalised. But they are also clearly distinct: the first works from the way this authority raises the evidence bar directly; the second from how it raises it indirectly, through greater structural damage. Since each requires a different solution, let me give them different names – Expert and Damage, respectively. Treating them independently will also have the advantage that if someone is unpersuaded by one version but persuaded by the other, we have a solution to their doubts.

Let me start with three unpromising ways of defusing the puzzle in either form. The first is to deny the explanandum – that social group moral encroachment entails that the marginalised need less evidence than the dominant. The reason we shouldn't deny this is that there is no way of doing so without losing the distinguishing mark of social group moral encroachment – the fact that it essentially involves differently situated groups. Denying the explanandum would amount to denying this, since it would mean implicitly denying that the relevant contrast class is that of the marginalised.

The second unpromising way of addressing the puzzle is by denying the expert-status to the marginalised with respect to the relevant group of propositions.²¹ Doing so would rob them of legitimate epistemic authority in ways familiar from colonisation, patriarchy, and other oppressive systems. Finally, we can't, of course, deny the distinguishing mark of moral encroachment of any kind – that when the moral stakes are higher the evidence threshold gets raised.

With these unpromising moves out of the way, here is how we resolve the puzzle above in each of its two forms. The way to defuse Expert is by reminding ourselves that moral considerations always work as pro tanto reasons that are weighed against each other in order to determine what is all things considered the right thing to do. The way moral stakes affect the evidence threshold is no different. There will be moral considerations that speak in favour of raising the bar, and others that speak in favour of lowering it. The possibility of such lowering was already written into Social Group Moral Encroachment, in virtue of the third condition, Robust.

The case in hand is just like this. While the expertise of the marginalised indeed raises the evidential threshold, another consideration pushes it down: the risk of placing a higher epistemic burden on the marginalised in a world in which they are already constantly asked to explain and justify the nature of their oppression to the supposedly well-meaning dominant group who are ‘just asking a question’ or ‘making a well-intentioned effort to learn’ (Berenstain 2016). This phenomenon and the ways in which it perpetuates epistemic and other oppressions is well documented under the label of epistemic exploitation (ibid.). We don't need to go into it in huge detail to get the plausibility of this way of defusing Expert: some structural moral considerations push the bar up, but others push it down. So, there is nothing puzzling about my proposal's commitment to the marginalised needing less evidence than the dominant.²²

An additional advantage of this way of solving the puzzle is that it overcomes an independent objection to encroachment arguments in the literature. For instance, Julius Schönherr and Javiera Perez Gomez (2022) have recently argued that pragmatic encroachment has the consequence of making it harder for victims of epistemic injustice to know, because the risk of prejudice against such victims raises the threshold for the justification of their beliefs.²³ If my argument here works, this is not in fact a consequence of encroachment views. The risk of prejudice is only one consideration to weigh against others; and erosions of the victim's epistemic authority, in general, in fact lower the bar on the victim.

But what about the Damage version of the puzzle? This, recall, said that my proposal should imply that the evidence threshold is higher, not lower, given how much more structural damage the same beliefs would cause if held by the marginalised. The reply to this version of the puzzle is even simpler than the reply to the last: we should deny that the marginalised can indeed cause more structural damage than dominant knowers. Indeed, it seems highly plausible to me that the marginalised can't cause any structural damage, let alone more. For marginalisation and oppression are, by definition, states of structural powerlessness, i.e., states in which what you do doesn't have structural effects.²⁴ As Iris Marion Young puts it, ‘in the most general sense, all oppressed people suffer some inhibition of their ability to develop and exercise their capacities and express their needs, thoughts, and feelings’ (Young 1990: 40).

But couldn't the marginalised overcome this state, hence putting themselves in a position of structural power? Yes, of course, that is the aim of liberatory movements. But this aim will not be achieved if the marginalised hold beliefs that endorse and propagate their marginalisation. At best, then, the marginalised have the power to slow down the end of marginalisation by adopting such beliefs, but that is not as great as the harm of actively maintaining and deepening it. Thus, once again, there is nothing puzzling about the implication of my proposal that the marginalised need less evidence for certain propositions than do the dominant.

Of course, when we talk about more or less evidence, and raising and lowering the bar, this can't be taken literally in a quantitative sense. For one of the lessons of standpoint theory – a lesson that the argument above has drawn on – is that differently situated groups have different kinds of evidence in the same context. So, talk of raising and lowering the bar should be read as a demand either for more or better evidence and a permission to get away with lesser quality or less evidence.

8. Objections

If these thoughts are on the right track, we should take seriously social group moral encroachment as a theoretical option. In order to further refine this rough sketch let me consider some objections to it. The objections fall into two clusters: the first targets the plausibility of explaining cases like Ncube vs Smith in terms of moral encroachment (section 8.1); the second the collectivity of the encroachment (section 8.2).²⁵

8.1. Moral encroachment alternatives

The first cluster of worries says that we can explain cases like Ncube vs Smith in terms of standard epistemic statuses such as justification and defeat. Now, I said in section 2 that sceptics about moral encroachment should hear the conclusion of this paper as a conditional (if moral encroachment is plausible then so is social group moral encroachment). Nonetheless, one wouldn't want to defend such a conditional in the first place if it turned out that the antecedent is a no hoper. Hence the pressure to consider this cluster of objections.

The most prominent alternative to explaining the target cases is to argue that the beliefs in question lack justification not because of moral encroachment considerations but because brute statistical evidence is never good enough for justification (e.g., Silva 2023, Gardiner 2018). The idea is that such evidence, no matter how highly probabilifying, does not justify belief about individuals. If it did, the argument goes, it would suffice for conviction in court, but of course we cannot convict on such evidence. What we need is something else, like a causal connection between the individual and the statistical generalisation (e.g., Enoch et al. 2012).²⁶ If so, the worry here goes, the audience in Ncube vs Smith is unjustified in believing Smith the better philosopher, for purely epistemic reasons and not because of moral encroachment.

Let me start my reply by citing the kind of case used in the literature to motivate the insufficiency of statistical evidence for justification:

Prison Yard. One hundred prisoners exercise in the prison yard. Ninety-nine prisoners together initiate a premeditated attack on a guard. Security footage reveals one prisoner standing against the wall refusing to participate. There is no evidence indicating who refused to participate. The prison officials decide that since for each prisoner it is 99% likely they are guilty, they have adequate evidence to successfully prosecute individual prisoners for assault. (Nesson 1979, cited in Gardiner 2018: 2)

What exactly is wrong with cases like this? The most popular view is that they treat a person as a mere statistic, thus violating the person's right to be treated as an individual (Jorgensen 2021). But how does this moral consideration count as an epistemic reason for not believing that prisoner number 42 (say) is guilty? The only plausible way, I submit, is if we think that moral considerations affect justification in some way, either directly or indirectly by raising the justification threshold. But if so, then the apt explanation of such cases is, after all, moral encroachment (either hard or moderate, respectively). Hence, the worry fails to point to an encroachment-independent explanation of standard cases against brute statistical generalisations. And if so, it fails to work against the moral encroachment explanation in Ncube vs Smith.

A second alternative explanation to the moral encroachment one is that the justification of the beliefs in Smith vs Ncube is defeated in standard ways that make unnecessary appeal to moral encroachment. Jennifer Lackey's (1999) notion of normative defeat seems perfect for making the case here. This kind of defeater is a normative consideration that the subject should have had in their possession that undermines the belief's justification. According to this view, the audience should have looked for more evidence, investigated further, and so on, before forming the belief. The fact that they didn't, undermines the justificatory status of their beliefs.²⁷

Let me start the response by noting that the moral encroacher agrees that the audience should have investigated further. What we disagree about with the current objector is why. The objector thinks for purely epistemic reasons; the moral encroacher says for encroachment-based epistemic reasons. An easy test for who is right is whether the objector maintains their verdict once we eliminate the moral stakes. If she does, we'll know that the appropriate explanation is in terms of a normative defeater.

So, let's add one detail to Ncube vs Smith. Suppose that it is well known around the world that an overwhelming majority of Ivy League graduates carry green handkerchiefs in their pocket, while most previously disadvantaged university graduates carry red ones. And suppose that while the audience in Smith vs Ncube is watching the series, it also forms the belief that Smith is most likely carrying a green handkerchief at the interview. Do the audience have a normative defeater for this belief? I don't see how they could: they are just watching this series; all the evidence points to this belief; why wouldn't they be justified? But if so, the only reason why we think there is a defeater in the original Ncube vs Smith is because of the moral stakes there. Hence, moral encroachment is still the right explanation for the original Ncube vs Smith.

8.2. Worries about social groups and their beliefs

Hopefully these responses at least go some way towards quieting doubts about the appropriateness of moral encroachment explanations in the kind of case that I have tried to isolate here. But they still leave room for doubts concerning the collectivity of this encroachment.

First, one might question that loose social groups such as white people can have beliefs whose justification we can meaningfully discuss. I tried earlier to motivate this possibility by appealing to the harms such groups' beliefs can cause and the fact that we talk about such groups being actively ignorant. Both presuppose, I argued, genuine groupiness on the part of such groups as well the possibility of genuine collective belief. But one might still worry about this possibility by appeal to Lackey's (2021) notion of ‘base fragility’. This is a state in which ‘the bases of the individually-held beliefs [of the members of the group] are wildly conflicting’ (Lackey 2021: 46), hence blocking belief ascription to the group:

For even if every member of a group believes that p, they might do so for wildly conflicting reasons, which renders the resulting state unfit for epistemic evaluation and for future deliberation in relation to group action. This, then, prevents the state from being a group belief. (ibid., 47)

Let me start my response by noting that this kind of objection applies only to what I earlier called the summative view of group belief – where we take the group to have a belief only if a certain number of its members hold this belief. As I mentioned earlier, I prefer the non-summative view. But since I have not defended it here and have claimed to remain neutral on this debate, I must address this worry.

My response, then, is that the relevant class of beliefs held by the dominant group are far from base fragile. On the contrary, such beliefs are held for very similar reasons by members of the group – reasons stemming from the kinds of racist narratives and evidence-erasures that I discussed in section 4. Thus, even on a summative view of belief, the relevant group of beliefs are not base fragile.

This response, however, only works by focusing on a very specific class of the group's beliefs, and opens up the way for a further worry: how exactly are we individuating these differently situated social groups, such that this class of their beliefs both turns out to be shared by enough members of the group (so we allow the view to be consistent with both summativism and non-summativism) and isn't held for ‘wildly conflicting reasons’ (so we avoid base-fragility)? We can't do so by appeal to the beliefs themselves or related worldview, on pain of circularity. But nor can we appeal to the basic social identity defining the group, such as race, since – as Mills (2007: 22) warned – things like white ignorance are neither exclusive to white people nor are had by all white people. A further complication to this question of individuation is that members of the same social group can be differently situated within the same group both in terms of other aspects of their intersectional identities (think of a white woman vs a white man in the 60's) and have different epistemic resources (think of Clarence Thomas vs Patricia Hill-Collins).

The reply to this worry is that the individuation of these groups must be a nuanced, multiple-step affair, the first step of which is to pin down the group's identity by the kinds of harms it perpetrates or is a victim of. Further steps would take into account more dynamic and intersectional aspects of these group identities. These steps will be complex and will set equally complex constraints on the moral stakes involved in social group beliefs and the corresponding raising and lowering of the evidence threshold. But I hope to have at least made it plausible here that undertaking this complex work is both worthwhile and necessary.

9. Conclusion

I have argued in this paper that if we think that individuals' beliefs are subject to moral encroachment, we should think that so are certain classes of the beliefs of social groups. I first situated the proposal in the existing moral encroachment literature, settling on modest, moderate, and robust moral encroachment as the right kind for social groups. I then discussed the structural moral stakes involved in certain classes of dominant groups' beliefs: they harm both the epistemic environment and its marginalised inhabitants. I then provided an additional argument for this view by appealing to the way it can benefit liberatory epistemologies. A potential puzzle that the view generates is that marginalised knowers need less evidence even though they in fact have better evidence than dominant knowers. Dispelling this puzzle helped not only strengthen the proposed view but also showed how existing arguments that encroachment views foster epistemic injustice are misguided. Finally, I fended off objections to both the moral encroachment aspect and the groupy aspect of the proposed social group moral encroachment.

These are no more than the first touches of a sketch of this view and the arguments motivating it are by no means meant to be conclusive. That said, I hope to have shown here that this is, at the very least, a promising direction for future research.²⁸