The collective epistemic reasons of social-identity groups

Abstract

In this paper, I argue that certain social-identity groups—ones that involve systematic relations of power and oppression—have distinctive epistemic reasons in virtue of constituting this group. This claim, I argue further, would potentially benefit at least three bodies of scholarship—on the epistemology of groups, on collective moral responsibility, and on epistemic injustice.

1 Introduction

Can a group of people who share a social identity—such as white people, Patagonians, or women—have epistemic reasons in virtue of being this group? Current discussions of the normativity of groups tend to have three features which make this question difficult to answer. First, they focus on either very structured or very unstructured groups, and social groups are neither. Second, existing discussions are framed not in terms of reasons but in the language of responsibility, duty, obligation, and blame. And finally, they typically focus on the moral, rather than the epistemic, understanding of these notions. Thus, for instance, philosophers of groups theorise the moral responsibilities of corporations, or, on the other extreme, the moral obligations a random collection of adults on a beach acquires when a child starts drowning and the only way to save her is to do it together. Of groups made up of people who share a social identity (such as race, gender, ability) little is said, and the little that is said is in the language of moral responsibility.

This paper shifts our attention along all three of these standard axes of theorising—to a different kind of group (social-identity groups), to another normative notion (a reason), and to a different normative domain (the epistemic). I defend the claim that certain social-identity groups—those structured around systematic relations of power and oppression—have distinctive epistemic reasons in virtue of being this group.¹

By way of making clear the theoretical significance of this claim, let me mention three debates in philosophy which would benefit from it. The first is the one on collective² moral responsibility. The most obvious benefit here is that moral responsibility rests on our epistemic standing, but this epistemic standing is, in turn, fixed by the kinds of epistemic reasons we have. As the saying goes, (non-culpable) ignorance is an excuse: if a group had no reason to believe that their actions would bring harms, the group is morally blameless.

Second, the epistemology of groups would benefit from the present argument. Discussions here revolve around how we should conceive of the epistemic justification, responsibilities, and knowledge of groups (e.g. Lackey, 2021; Silva, 2019). But it is unclear how philosophers hope to resolve such questions without an account of collective epistemic reasons. Even if one does not accept the increasingly popular thought that reasons are our most basic normative notion (e.g. Parfit, 2001; Schroeder, 2021), it is uncontroversial that the epistemic justification and responsibilities of individuals hinge on the fit between their doxastic attitudes and their epistemic reasons (e.g. the evidence). It would make sense that group epistemic justification and responsibility are likewise determined by this fit.³

The final, and to my mind greatest, benefit of the current argument will be for scholarship on epistemic injustice. This scholarship is founded on the basic premise that knowledge and knowers are essentially social. One would have thought that taking this sociality seriously would have meant a lively dialogue between scholars of epistemic injustice and philosophers of groups, with each profiting from the tools of the other. Yet surprisingly, no such dialogue exists at present.⁴ This paper fills this theoretical lacuna.

Here is the plan. In Sect. 2, I zone in on the kind of group that is the focus of this paper. In Sect. 3, I dispel initial doubts about the possibility of such groups having collective epistemic reasons. In Sect. 4, I set out the necessary and sufficient conditions for a collective epistemic reason. In Sect. 5, I argue that some of the reasons of social-identity groups structured by systematic relations of power and oppression meet these conditions. In Sect. 6, I conclude by considering the implications of this claim for the three debates I just mentioned.

2 Social-identity groups

Thanks to Herlinde Pauer-Studer for making me realise that I need to say something about the ontology of social groups here as well as for pointing me to the relevant literature.

Social ontologists tend to have quite a liberal understanding of social groups. Thus, for instance, Katherine Ritchie (2020: 402) includes not only obvious ones such as race, gender, and ethnicity, but also teams and clubs.⁶ I think that this liberality is a mistake if we want to think about the normativity of social groups. For teams and clubs have all the normatively relevant features of highly organised collectives, which social-identity groups lack—voluntary membership in the group, decision procedures, spokespersons, and everything else we need for talk of responsibility and reasons to be obviously applicable. Lumping together such groups with much looser ones is treating as a normatively unified kind something that plainly is not. But I am not a social ontologist, so will not dispute these classifications. My focus here is on a subset of the social ontologist’s category of social groups—social groups as understood by critical race theorists and feminists. I have called these social-identity groups in the title to avoid misunderstanding, but since I will not discuss any other social groups, I will often let ‘social group’ stand for ‘social-identity group’ to avoid verbiage.

Social groups in this narrower sense are groups of people who share a social identity—race, gender, class, ability, sexuality, and so on. Such identity is typically understood along social constructionist lines. Thus, while most people (happily) no longer think that race is a biologically real kind, many think of it as a real social kind—it has causal powers in the social world due to the social structures in which we are embedded.⁷ On the descriptive version of this view, what constitutes a group as a race, for instance, are the common experiences its members have as a result of being given the race-label that they have been given (e.g. Mallon, 2006). On the normative version, what constitutes a group as a race is being subject to certain institutional practices and normative expectations in virtue of these labels (Haslanger, 2000; Taylor, 2013). We can say similar things about all other social identities. (Though one should be careful not to overstate the parallels. More on this shortly.)

Let me capture the kind of social group at issue here in terms of a sufficient condition:

• Social(-identity) group: G is a social group if its members share a social identity such as gender or race.

Social-identity groups, as black feminists have taught us (e.g. Crenshaw, 1991), are intersectional, so that each of us belongs to several in virtue of our intersectional social identities. This will considerably complicate the way group reasons are a source of reasons for individuals, but that is just as it should be. Given that we are creatures with such complex social identities, we should not expect that the reasons we get from these identities should follow a simple pattern.

This minimalistic, constructionist, and intersectional characterisation of social groups dovetails nicely with structuralist proposals by social ontologists. For example, Ritchie (2020) argues that what constitutes a social group (in her broader sense of the term) are networks of relations amongst its members and to other groups, which are partly constituted by social factors, and in which any point of intersection amongst relations—what she calls a ‘node’—comes with normative requirements on the occupant of this node. The account is too complex to be done justice to in such a short discussion. Suffice it to say that it is available for those who want a proper ontology of social groups. For my purposes, it will be enough to characterise such groups in terms of a subset of these relations—the ones that come with the socio-political relations of power in which our race, gender, and so on embroil us.⁸

3 Two challenges for claiming collective reasons for social-identity groups

These then are the groups which I will argue have certain collective epistemic reasons. But before I do so, let me defuse two challenges for the claim. The first is that such groups cannot have reasons of any kind because they are not agents (Sect. 3.1). The second is that they cannot have epistemic reasons, since they do not have beliefs (Sect. 3.2).

3.1 No agency for social-identity groups?

It is taken as a standard desideratum on a theory of collective responsibility that it must not violate more ordinary, individual-responsibility intuitions. One important example is the so-called agency principle—the idea that moral obligations, and hence attributions of responsibility, can only be predicated of agents (e.g. Schwenkenbecher, 2018: 111). As a result, the simplest accounts of collective (moral) responsibility and obligations concern very organised groups like corporations, to which it is relatively easy to ascribe agency (e.g. French, 2020; List & Pettit, 2011). And the most discussed puzzles of collective responsibility concern the groups that are least agent-like—random bunches of people who suddenly find themselves having to do something together (e.g. Fleisher & Šešelja, 2022; Held, 1970; Schwenkenbecher, 2018).

It would make sense to have a similar agency-desideratum on an account of group reasons.⁹ Reasons are considerations that speak in favour of an agent’s acting, believing, or feeling something. (A more precise account follows in Sect. 4.) If an entity is not an agent, talk of its reasons looks out of place. But there is a big challenge to thinking of social-identity groups as agents: such groups are too unstructured for the things that the group does qua group to be properly described as intentional actions. To add insult to injury, membership in such groups is neither typically voluntary nor does it require that the member identify with the group. A white person might, for instance, hate the idea of being white. But this is irrelevant to what makes her white—the social structures in which she is embedded and the experiences that the white label generates. It is, thus, unclear how a group made up of such members can have reasons. Race is probably the most inflexible on this score: if you do not identify with your gender-label, for instance, you can change it; in contrast, the possibility of transraciality is far more controversial. So, I will stick to racial groups as the paradigm here. If we can make the case that there are collective epistemic reasons that we have in virtue of being a certain race, we can make it for any other, more voluntary, social-identity group.

I can see three ways for the friend of social-group reasons to meet the agency-challenge. The first, and least promising, would be to accept the agency principle for responsibility and obligations, but to deny it for reasons. Such a move would drive an artificial wedge between clearly interrelated normative notions. Consequently, it would deprive the current account of much of its theoretical significance by debarring it from contributing to the more established discussions on group responsibility mentioned in Sect. 1.

The second way to meet the agency-challenge is to argue that social-identity groups are just like very unorganised groups that are faced with the need to perform a joint action. The usual way to make sense of the obligations of such groups is by some version of a two-stage duty. In the first stage, individuals have a duty to mobilise (Fleisher & Šešelja, 2022), join forces (Hindriks, 2019), or constitute an agent with others (Collins, 2013). In the second stage, the collective now-agent has a duty to perform a joint action (e.g. save a drowning child). Translating into the language of reasons, this would mean that individual members of social groups have a reason to mobilise, and the mobilised collective, which can now be ascribed agency, has the epistemic reasons which I am after. But I do not think that this proposal is going to fare much better than the first one. For it is quite puzzling (to me at least). To oversimplify, it implies that we have a reason to join forces to do something that none of us—either individually or collectively—has a reason to do.¹⁰

So, the third and best way to meet the challenge is head-on—to make the case that certain social groups are enough of an agent to give reasons-talk purchase. Let me do this in four steps. First, note that the fact that membership in a group is involuntary does not foreclose the applicability of claims about reasons. We do not voluntarily choose our individual agency either. So, lack of voluntary membership in a group cannot have the normative significance that the challenge presupposes.¹¹

The second step is to argue that, contra the challenge, we cannot deny a group intentional action on the grounds either that it is unorganised or that its members do not identify with the group.¹² One of the most remarkable historical illustrations of both these points is the ‘Underground Railroad’, the biggest network for secretly assisting slaves to move out of the US South in the eighteenth and nineteenth centuries. While there are many competing accounts of the details, most scholars nowadays agree that the Underground Railroad had no governing structure or body, making it as unorganised a group by group-theorists’ lights as a social-identity group. Moreover, many suggest that at least some of the individuals of which it consisted did not identify with the umbrella group. Here is an explicit statement of both the lack of organisation and of identification:

Many [abolitionists] who were active in the cause [of the Underground Railroad] had only a limited interest in assisting fugitive slaves. Very few of them approved of luring slaves from the South, and those who did operated on their own, without the benefit of any intricate organization, as did the slave stealers who abducted slaves for profit. (Gara, 1961: 69, my italics; see also ibid.: 91)

And another:

Blacks, who aided fugitive slaves near the Ohio River on the border with slave territory, did not consider their actions related to an organized Underground Railroad. To them it was simply an extension of their community values…(Calarco et al. 2011: 153, my italics)

But despite the lack of organisation and identification, it would be odd to suggest that the liberation of between 40,000 and 100,000 slaves (scholars disagree on the number) did not amount to a series of group intentional actions.

The third step to seeing that social groups can have (enough) agency is to note that we can appropriately ascribe intentional action to them. Think of Charles Mills’s (2007) notion of white ignorance, in which white people as a group deliberately ignore the experiences of the oppressed and our role in their oppression. Deliberately ignoring something is an intentional action par excellence. Think, too, of the severe epistemic and moral harms we cause others in virtue of being the group we are. The way these harms are caused is not some brute involuntary thing the way forest fires cause harm or a bunch of bystanders on a platform can cause me to miss the train. Social groups, unlike forest fires and bystanders, enjoy considerable benefits and privileges in virtue of having such causal powers. Moreover, they have the power to stop causing these harms—they do not need to be externally extinguished, so to speak.

Given the unspeakable and systematic harms we have caused and continue to cause as a group, as well as the continued privileges that we enjoy as a direct product of these harms, it is unclear why anyone would want to claim that the causing of such harms is unintentional.¹³ Indeed, some white people’s claims to unintentionally doing these things ring quite hollow. But if these actions are intentional and done as a group, we have credited the group with agency.

Conversely, to deny group agency to the oppressed would itself seem to be a good mechanism for perpetuating their oppression. A common way of understanding oppression is precisely as the denial to, or the undermining of, the agency of another. This scales up naturally to groups: for instance, in colonialism, what was being denied the colonised as a group was precisely agency (as well as the agency of its individual members). Thus, at least some social groups—those structured by systematic relations of power and oppression—have a sufficient claim to agency for talk of group responsibility and reasons to get purchase.

The fourth and final step in answer to the agency challenge is to note that on a plausible, mild consequentialism about reasons (and responsibility), the fact that S’s action A will cause certain harm gives S a reason to not A; and the fact that S’s action B would promote the well-being of others gives S a reason to B. Thus, it would be strange (and morally pernicious) for you to deny that white people’s privilege harming others is a reason for white people to forego this privilege as a group, or to deny that the potential benefits of depriving us of this privilege give the oppressed as a group a reason to deprive us of it.

These steps concern moral collective responsibility, agency, and reasons. The story about epistemic reasons is more complicated (as we will see in Sect. 4), but at least the present considerations clear the ground for the possibility of collective reasons for social groups despite the involuntary membership of its individual members, their lack of identification with the group, and the lack of organisation in the group.

3.2 No beliefs for social groups?

A second hurdle to my claim that social groups can have collective epistemic reasons is that one might think that such groups do not have collective beliefs to be regulated and guided by such reasons (e.g. Millar, 2021: 499). The only groups with a shot at having beliefs, the worry might go, are very organised groups. For instance, the most plausible intuitive argument for group belief of which I know is Jennifer Lackey’s (2021): such groups can lie (think of Volkswagen claiming that they did not know about the ‘faulty’ carbon emissions controls); but lying when asserting p presupposes that you believe that not p. Such an argument does not seem available to me since social-identity groups do not seem to be structured enough to be capable of lying.

The first thing to do in response to this challenge is to deny that social groups cannot lie. After all, we do talk of men having systematically deceived women about certain homely truths (especially concerning sex). We also talk of white people self-deceiving about the role of our own merit in having gotten us where we are (as opposed to the truth, which involves slavery, colonialism, and continuing racist oppression). This kind of self-deception is, indeed, an essential feature of white ignorance (Mills, 2007: 17, 34). But if both deception and self-deception are possible for social groups qua groups, then it becomes unclear why we cannot think of such groups as having collective beliefs.

The second thing to do in response to the no-belief challenge is to note that a lot of the weirdness of thinking of social-identity groups as having beliefs is shared by all groups, and comes from a very phenomenal conception of mental states. To be sure, beliefs are, by all accounts, the least phenomenologically rich of our mental states. (Think: what is it like to believe that Gaborone is the capital of Botswana?) Nonetheless, one might maintain that without some kind of phenomenology—which is clearly absent in groups qua groups—talk of any mental states is meaningless. Fortunately, this problem is neither specific to social-identity groups nor intractable. A moderate functionalism about group mental states allows us to overcome it.¹⁴ Christian List (2018), for instance, has recently argued that we should conceive of group consciousness as awareness, understood along functionalist lines, rather than as phenomenal experience. If he is right, then we have the resources of thinking of group epistemic states such as awareness, perception, and belief along the same functionalist lines (List, 2018: 301–2).

This proposal is ‘non-summativist’—it posits mental states for groups that they have qua groups. The opposing, so-called summativist, view (e.g. Lackey, 2021), requires a specified portion of the group’s individual members to hold the beliefs in question. As we will see (Sect. 6), the current argument will most probably commit us to a kind of non-summativism, but for the moment, we just need to know that options are available for making talk of social-group beliefs a lot less weird. But if social groups can have collective beliefs, then they must have collective reasons for those beliefs. The question then becomes what kind of reasons these reasons might be. I now turn to this question.

4 Collective epistemic reasons

Suppose that you are the boss of a company that makes buttons, and I am one of your factory hands. Our company has promised Levi’s a batch of sixty thousand buttons by 2 pm this Friday. The promise gives us qua company a reason to do something (produce and deliver the buttons by the promised time). It is a special reason—we, the group, have it in virtue of being this group, and it is the source of further reasons to us as individual members of the group. Thus, it is a source of reasons for me to hurry up, skip lunch, and grudge your promise to Levi’s; and a source of reasons for you to yell at me if I do not help us meet the deadline, commend me if I do, tell me that my grudge is childish, and so on. This is an example of practical reasons for a very organised or structured group and the reasons such groups give its members. The question that this paper addresses is whether the much less organised groups described in the previous section—social-identity groups—can have similar epistemic reasons. To answer this question, we need a more principled account of a reason and of collective reasons. This is what I offer in this section.

4.1 Kinds of reasons

A reason is a three-place relation: it is a reason for someone to do something. Specifying what a group reason is involves specifying the relata and the nature of the relationship between them. The literature distinguishes three kinds of reason—normative, motivating, and explanatory (e.g. Alvarez, 2010). A (pro tanto) normative reason for φ-ing is a consideration that favours φ-ing.¹⁵ A motivating reason for φ-ing is the consideration that the agent took to favour φ-ing and for which she φ-ed or tried to φ. An explanatory reason is the consideration we cite in explaining why the agent φ-ed. Here, my focus will be on normative collective reasons, with only a quick glance at motivating ones near the end (Sect. 6). So, ‘reason’ will refer to ‘normative reason’ unless otherwise specified. I will remain neutral on the ontology of such reasons; the above characterisations are of the concepts alone.

4.2 Kinds of epistemic reasons

The distinction between normative, motivating, and explanatory reasons pertains to all types of reasons—epistemic, moral, prudential, and whatever other kinds of reasons there are. Let me now zone in on epistemic reasons. They come in two flavours, to my mind:

• Doxastic reasons: reasons to believe, disbelieve, or suspend judgement about a proposition.

• Epistemic-conduct reasons: reasons for epistemic conduct, such as to adhere to epistemic norms, gather more evidence, investigate further into a topic, or cultivate epistemic virtues and quash one’s vices.

Since this way of carving up the epistemic is mine, two clarifications are in order. First, there is some debate about whether what I am calling epistemic-conduct reasons are properly conceived of as epistemic. Some philosophers argue that imperatives to inquire—and hence the corresponding reasons—are either moral or prudential, never epistemic (see, e.g. Conee & Feldman, 2004). I have argued elsewhere (Mitova, 2019) that such views get importantly wrong the normative contours of certain epistemic situations. I will not rehearse the arguments here. For those who think epistemic-conduct reasons are prudential or moral, there will be still an important insight from this paper that there is a special category of such reasons that are had by social-identity groups as such.

The second clarification is that reasons to inquire have lately been increasingly grouped under a different umbrella—the ‘zetetic’—partly in response to worries like the above (e.g. Friedman, 2019). I think that the category of epistemic-conduct reasons gives us greater theoretical latitude by allowing us to treat reasons to inquire and to do so well/ virtuously under the same rubric. But nothing turns on the terminology here, as long as we are clear that these reasons are exclusively to do with our belief-forming practices and are meant to promote distinctively epistemic goals.

4.3 Collective epistemic normative reasons

If there are such things as epistemic collective reasons for social-identity groups, they would be instances of a more general kind of epistemic collective reason—considerations that speak in favour of the group’s either believing a proposition (doxastic reasons) or performing epistemic actions (epistemic-conduct reasons), and which are such considerations at least partly in virtue of the group’s being this group.

It is at least prima facie plausible that such reasons exist for very organised groups. Thus, we say that Volkswagen had every reason to believe—indeed know—that Bosch had installed software to cheat on carbon emissions. And we blame a university for taking an unenlightened line to the Humanities, on the basis of its ignoring good evidence that the Humanities cultivate vital intellectual skills.

Let me make this prima facie plausible notion slightly more formal:

CERG: R is a (pro tanto) collective epistemic normative reason for group G to φ iff:

(1) R is a consideration that epistemically counts in favour of G’s φ-ing;

(2) φ-ing is either believing a proposition or undertaking epistemic actions such as inquiring; and

(3) the favouring relation obtains partly in virtue of G’s being a group of the kind it is.

Condition (1) is what makes such reasons normative and epistemic; (3) is what gives them their groupiness,¹⁶ so to speak. I have already discussed (2) at some length (Sect. 4.2), as well as what is meant by a normative reason in (1) (Sect. 4.1). Let me now quickly comment on the epistemic favouring relation in (1), and then say more about (3).

The first thing to note is that we need to specify in (1) that the favouring relation is epistemic, because (2) is not enough to ensure that we are talking about epistemic reasons. That is, the fact that the favouring relation features an epistemic state or conduct as one of its relata does not guarantee that the relation itself is epistemic. We can have non-epistemic reasons both for believing a certain proposition (e.g. our friendship is a non-epistemic reason for me to believe that you are a great person even if the evidence favours believing you a jerk) and for epistemic conduct (e.g. I would get a grant if I investigated further).

Hence, (1) stipulates that the favouring relation is epistemic. What does that mean? Simply that the favouring is from the epistemic point of view—doing the favoured thing (believing or undertaking epistemic actions) promotes epistemic goals such as truth and knowledge.¹⁷ The paradigm of such favouring is the evidential relation. But we can also have others. For instance, the fact that I have conclusive evidence to believe you a jerk is not an evidential consideration, but certainly favours in the epistemic way believing that you are.

Now for condition (3). It states that in some instances, this relation of epistemic favouring obtains partly in virtue of the bearer of the relation being the group it is. Why only partly? Because no relation can obtain solely in virtue of one of its relata being what it is. An epistemic favouring relation, for instance, must also obtain in virtue of the way belief and epistemic conduct relate to epistemic goals like truth and knowledge.

How should we understand the ‘in virtue of’ relation in (3) apart from its being a partial one? First, it should be the case that R would not exist without that group. But this is not what gives the reason its groupiness, because it does not go beyond re-iterating the truism that a reason is always a reason for someone (to do something).

A second, and more properly collective, dimension of the ‘in virtue of’ relation is that the reason would not exist without the group’s being a group of that kind. To invoke the more standard discussions of group responsibility, when we say that the government has an obligation to ensure equitable access to education, we are positing an obligation that obtains in virtue of G being a government. If G were our button company, its reasons would not include reasons to ensure equitable education. So, part of what it means to say that a reason is had in virtue of the group’s being that group is to do with the kind of role the group plays in relation to other groups. This chimes nicely with my earlier structuralist suggestion that we should think of groups in terms of the relation-networks in which they are embedded.

One natural way of transposing this idea to the epistemic realm is to think that the relevant group, as a group, has special epistemic reasons in virtue of the kind of role it plays because this role gives it privileged access to some evidence to which other groups do not have access. This is plausibly the kind of evidence that was at work in the Volkswagen and university examples earlier.

The final and most collective dimension of the ‘in virtue of’ relation is that the reason would not exist without the group’s being a group of that kind.¹⁸ The most common resource in the literature on collective responsibility for capturing this dependency of a reason to φ on the group is the idea that no individual member of the group can bring about the outcome at which φ-ing aims (e.g. Collins, 2019; Schwenkenbecher, 2018). Thus, no single member of the government can bring about equitable access to education, and on the plausible ought-implies-can principle, no individual has a duty to bring it about; the group alone does.¹⁹

Translating all this into reasons-speak and building it into the third condition of CERG:

• CERG(3) the favouring relation obtains partly in virtue of G’s being a group of the kind it is; i.e., it obtains because:

• CERG(3)^role the group plays a particular role in relation to other groups, giving it privileged access to reasons for φ-ing, and

• CERG(3)^team the aim of φ-ing can be attained by G but cannot be attained by individuals alone (in their capacity as individuals).²⁰

Note that I do not mean CERG(3)^role and CERG(3)^team as either necessary or sufficient conditions on the groupiness of a reason (though of course CERG(3) is). They are, nonetheless, a very plausible indication that a reason is collective. If there are such things as collective epistemic reasons for social-identity groups, then, they would need to be instances of the CERG schema.

5 Epistemic collective reasons for social groups

In this section, I argue that some social-identity groups can have both doxastic and epistemic-conduct collective reasons, by showing that both can be instances of CERG.

5.1 Collective doxastic reasons

Can a social group as such have doxastic reasons, reasons to believe a proposition in virtue of being this group? To begin with, note that a social group can have reasons to believe at least in one wrong sense. In virtue of being this group, the group must believe particular propositions that constitute the group’s ideological identity. For instance, Pizzagate fans, as a group, have a reason to believe that Democrats are engaged in child trafficking and abuse in the cellar of Comet Ping Pong pizzeria. Without these beliefs, this group would not be the group it is. But this is, clearly, not an epistemic reason to have these beliefs. Epistemic reasons for a group need to be related to epistemic goals to meet CERG(1).

I now offer three considerations in favour of collective doxastic reasons for social groups. First, if there were no such reasons, there would be no such thing as collective belief for social groups either.²¹ Second, social groups meet CERG(3)^role. Finally, they meet CERG(3)^team.

The first, and simplest, consideration is that as soon as we allow beliefs for social groups, as we did in Sect. 3.2, we effectively allow for collective doxastic reasons. For on a minimal version of the charity principle, having beliefs presupposes having reasons for those beliefs. So, having collective beliefs requires having collective reasons for those beliefs. Otherwise, one would have the ontological oddity of a collective (functional equivalent of a) mental state, with individuals’ reasons somehow grounding this collective state. This would be both normatively and psychologically weird. Motivating reasons (as discussed in Sect. 4 and a little further in Sect. 6) belong to the agent whose state they are reasons for. So, I cannot have a motivating reason for your action or belief. Similarly, it cannot be that the motivating reason for my university’s belief is somehow had by me alone, since we are different agents.

The second consideration in favour of thinking that a social group can have reasons to believe a proposition in virtue of being that group is that certain social groups have privileged access to evidence concerning some propositions. In Sect. 4.3, I suggested that having privileged evidence is one indicator of a collective doxastic reason—captured by CERG(3)^role. Thus, the university has privileged evidence about the usefulness of the Humanities, and Volkswagen had privileged evidence about its carbon emissions’ controls. I think that something similar holds for social groups. At least according to some social epistemologists, being members of certain social groups—especially the oppressed—gives the group knowledge and kinds of evidence that is unavailable, or at least not easily available, to other groups. There are two factors that make this evidence not easily available: the evidence is provided by the group’s shared lived experiences, and systematic structures of oppression make the epistemic resources of the oppressed unavailable to the oppressor.

The idea here (e.g. Dotson 2012; Fricker 2007; Pohlhaus 2012) is that our epistemic resources depend on our position within systematic structures of power. When one is in a vulnerable position, one must, as a survival strategy, attend to what the powerful are likely to notice and expect (Pohlhaus 2012: 717, 721). Thus, one needs two sets of epistemic resources—those of one’s own group and those of the powerful. In contrast, the powerful group, which only needs its own epistemic resources, remains (deliberately) ignorant of the resources of the oppressed. This enforces its dominance by excluding those resources from the mainstream knowledge economy.

This widely accepted picture favours the existence of doxastic reasons for certain social groups. Take, for instance, the concept of ‘double consciousness’ (du Bois 1903). Crudely put, and echoing the previous paragraph, this is the idea that one is forever looking at oneself through the gaze of the oppressor who has set himself as the norm of all things superior, civilised, rational, good, etc. The fact of double consciousness and the unique access that black people have to it is privileged evidence of white normativity, of the invisibility of white privilege, and of the ways in which ignoring race is likely to reinforce such normativity. If that is right, then we have an example of collective epistemic reasons of the doxastic kind—considerations that epistemically favour (CERG(1)) the social group believing some propositions (first disjunct of CERG(2)), where the favouring relation obtains in part because the group is of the kind it is (CERG(3)^role).

The final reason for thinking that social groups can have collective doxastic reasons is that some of these groups’ reasons meet CERG(3)^team. Now, at first glance, social groups may seem to breach this condition. The only (epistemic) aim with which we adopt a belief that p is to get at the truth, or knowledge, about p. And, one could argue, such an aim is never beyond the reach of the individual, except perhaps in very special circumstances.²² All individuals within the group ‘black people’ can get—and indeed most have gotten—to the truth about colour blindness. Perhaps, this has been made easier through collective inquiry. After all, one does not learn complex social truths on one’s own.²³ But it is not in principle impossible for one person to achieve this aim as it is with equitable education. Hence, social-identity groups seem to breach CERG(3)^team.

This doubt is misguided, however, for it rests on a misunderstanding of the aim that features in CERG(3)^team. The aim on which the doubt rests is that the individual is in possession of the truth about p. This indeed is attainable by individuals. But the aim relevant to CERG(3)^team is that the group is in possession of the truth about p. The relevant aim is this collective one simply because the belief for which the doxastic reason is a reason is a group belief. The aim of the group being in possession of the truth is plainly not attainable by the individual. So, appearances notwithstanding, some social-identity doxastic reasons meet CERG(3)^team.

5.2 The collective epistemic-conduct reasons of social groups

I now argue that social-identity groups can also have epistemic-conduct reasons. To begin with, note again that these groups can clearly also have pragmatic group reasons to engage in epistemic conduct. For instance, List and Pettit (2011), in the short passages where they discuss social groups, offer a pragmatic, ‘developmental rationale’ for holding such groups responsible: doing so, even though strictly inappropriate in light of their lack of agency, they argue, ‘may actually prompt the grouping to incorporate and organise against the condemned behavior’ (2011: 170). Although List and Pettit do not use the language of reasons, such a rationale clearly generates corresponding pragmatic or moral collective reasons to cultivate certain epistemic virtues and epistemic conduct. But equally clearly, such considerations will not generate collective epistemic reasons: the fact that X will promote moral or social aims does not epistemically favour collecting more evidence on a topic. CERG(1) has been breached.

But there are plenty other group considerations which do epistemically favour certain kinds of epistemic conduct for social groups and meet CERG(1). Recall the deliberate ignorance that white people engage in, which I mentioned as an example of intentional action. Clearly, white people must stop doing so and get educated about the experiences of those whom we ignore and oppress. This ‘must’ is, of course, a plain indicator of a (conclusive) reason. Moreover, the reason is both epistemic and collective. Or so I now argue.

The reason is epistemic because there is a clear epistemic imperative to stop being ignorant: white ignorance creates and widens what Miranda Fricker (2007) calls ‘hermeneutical lacunas’—gaps in our collective resources for understanding our experiences and the world. I have argued elsewhere (Mitova MS) that such gaps create epistemically risky environments, environments which deprive agents of epistemic reasons and hence knowledge, and lead to the proliferation of epistemic vice, such as the vices of the privileged (Medina, 2013). Let me quickly try to summarise a complex argument through an example. Take again double consciousness. If this concept is not part of our collective epistemic repertoire—as it arguably is not or was not until very recently—many agents cannot cite reasons featuring this concept by way of explaining and justifying their beliefs. Given asymmetries in power, when such reasons are cited, their owner is gaslighted. This has the effect of shaking the confidence of the oppressed in their reasons for holding the (justified) beliefs in question. Such doubts then act as defeaters which undermine claims to knowledge.²⁴ Avoiding such epistemically bad environments is an epistemic goal, since these environments undermine knowledge and the epistemic agency of its denizens.²⁵ Hence, the considerations which favour epistemic conduct that avoids such environments favour it epistemically. They thus meet CERG(1).

Moreover, the reasons are distinctively collective, because they meet both CERG(3)^ROLE and CERG(3)TEAM. First, the reasons depend on the group because of its role in relation to other groups—white racism and its dominance of other social-identity groups. And second, the aim of the epistemic conduct for which they are reasons is not attainable by individuals but by the group alone. This is because hermeneutical gaps are not made, nor can they be fixed, by single individuals. It takes a group, embedded in certain structural relations, to create and widen such gaps. By the same token, no single individual can close hermeneutical gaps on her own, no matter how rigorously she inquires or how assiduously she cultivates epistemic virtues.²⁶ The gaps can only be closed by groups. In the first instance, the groups that created them through the marginalisation of others’ epistemic resources can close the gaps by collectively undertaking actions to change their faulty epistemic conduct.

5.3 From group to individual

If the arguments so far worked, social-identity groups can have both doxastic and epistemic-conduct collective reasons. Let me conclude this section with a word on how such reasons transmit to group-members.

Many group theorists think that there are certain domains for which collective responsibility is prior to the responsibility of members of the group (e.g. French, 2020: 17). I think that social groups’ epistemic-conduct reasons most obviously work like this, although I must confess upfront that I have no account of precisely how the normativity-transmission works. So, in this subsection, I will just provide some intuitive thoughts.

To begin with, why should we think of collective epistemic-conduct reasons as prior to individuals’ reasons? Simply because the unjust relations that shape these reasons are structural and hence a matter of the groups involved in these relations. So, on the plausible ought-implies-can principle which drove CERG(3)^team, I cannot have reasons to change anything, unless I somehow inherit them from the group.

How does this inheritance work? Intuitively, if my group has a normative reason to change things and I can do my bit to help, then I have a reason to help. One potentially fruitful way of thinking of such reasons is along the lines of Christopher Woodard’s notion of a ‘group-based reason’ (2017). In our language of favouring here, it is a consideration that speaks in favour of an action not in virtue of the goodness of the action itself, but in virtue of the goodness of the ‘pattern’ (ibid.: 107) in which this action is playing its part. Thus, for instance, and again applying to our case, if white people have a reason as a group to actively learn more of the experiences of those whom we oppress, my individual reasons for performing certain actions in that direction (to read books, undergo training) are considerations that speak in favour of these actions in virtue of their contribution to this good thing—our group’s learning more.

While this way of putting it bears both its collective and epistemic character on its sleeve, it also might sound a bit odd, by suggesting that there is nothing good about my specific acts of investigation in themselves (since, by stipulation, group-based reasons are not had in virtue of the goodness of the action itself). But I think that the oddness dissipates if we remind ourselves that a reason can be grounded in different kinds of normative goodness. Thus, the fact that my reading books is good because of the group’s aiming at certain epistemic conduct does not preclude its being also good because it will personally improve my character, say. What matters is that all members of our group have the former, collective-inherited, reason to read the relevant books, regardless of other reasons we may or may not have for reading them.

6 The theoretical benefits of group epistemic reasons

My case for collective epistemic reasons for social-identity groups is now complete. Both doxastic and epistemic-conduct reasons meet the necessary and sufficient conditions on collective reasons. The collectivity condition, I suggested, should be understood to obtain in virtue of the group’s role in relation to other groups (CERG(3)^role) and the fact that no individual member of the group can attain the aim of the thing the reason is a reason for (CERG(3)^team)—the attainment of truth by the group (for doxastic reasons), or the preservation of epistemically healthy, risk-free environments (for epistemic-conduct reasons).

This argument obviously requires a lot more work to get us to a comprehensive account of collective epistemic reasons. But as I do not have space for that here, let me conclude by showing that such work would be worthwhile, since it would benefit at least the three areas of philosophy with which I opened this paper—collective moral responsibility, the epistemology of groups, and epistemic injustice.

First, my proposal can be helpful for the literature on collective moral responsibility in two ways. As Michael Doan (2020: 212) points out, solutions to collective problems need thinking about rather than just making themselves available in the way many philosophical discussions of group responsibility presuppose. Such thinking involves getting clear about the problem and available solutions—that is, inquiring and forming beliefs. Since determining whether a group action is morally responsible will partly involve evaluating this thinking, theorising group moral responsibility requires an account of collective reasons for inquiry and belief. I have offered the beginnings of such an account here through the CERG-style schema.

This is a benefit of my proposal concerning epistemic reasons. But there are also more general benefits to switching to the language of reasons for group moral philosophers. Our intellectual and moral lives are made up of weighing against each other many different kinds of reasons. Duties and obligations—in which current discussions are framed—provide only one kind amongst the myriads of pro tanto reasons that rich, messy life throws at us. If this is right, then in the absence of an account of collective reasons, group theorists’ toolboxes are missing some of the most important tools for making sense of the normative lives of groups. The CERG-style schema supplies this lack.

Second, as already mentioned, a major debate in the epistemology of groups concerns how we should understand groups’ beliefs and their normative status. Summativists (e.g. Lackey, 2021) think that these are a function of the beliefs of the individual members of the group. Non-summativists, in contrast, think that groups have beliefs in their own right, which do not necessarily coincide with those of its members (e.g. Brown, 2022; recall too List, 2018 discussed in Sect. 3.2).

The current proposal sits more easily with a non-summativist account of both belief and epistemic reasons, given that I have cashed out the collectivist condition in terms of the role and irreducible effort of the group in attaining epistemic aims. Admittedly, for the account to pay its full way, we will need not just the account of a collective normative reason I offered here, but also of a collective motivating one. In the epistemic context, S’s motivating reason for believing that p is the consideration on which S bases her belief that p. What determines the normative status of a belief is the fit between normative reasons and motivating reasons. Needless to say, the translation of motivating reasons to groups will be tricky. But I have at least given some tools to negotiate these difficulties: we can model collective motivating reasons on the CERG schema.²⁷

Finally, epistemic injustice theorists often talk of reasons we have as this or that group in the structural meshes of power in which we are embedded. Most obviously, this paper has both legitimated and made precise the notion of a reason that vaguely features in these discussions. Less obviously, the current proposal might be helpful for various more specific debates. Take, for example, the debate on our responsibility for our numerous implicit biases. As Miranda Fricker (2016) has recently pointed out, intuitions pull in opposite directions here. On the one hand, we feel that any prejudice against a group is blameworthy. On the other, such biases are largely inaccessible to the individual, and so seem to be cases of non-culpable ignorance, which is the paradigm of an excuse.

The way Fricker tries to defuse this tension is by appeal to Bernard Williams’s distinction between responsibility and blameworthiness, applied to the epistemic domain. Thus, she argues that we are responsible for many of our implicit biases in the same, non-blameworthy, way that (Williams’s) Oedipus is responsible but not blameworthy for killing his father and marrying his mother. Curiously, Fricker does discuss collective epistemic responsibility, yet instead of making use of the standard tools of this trade (as I have here), she chalks up many of our implicit biases to epistemic bad luck, thus rendering us non-blameworthy for them. This move invites an obvious criticism: construing the injustices which are the origin and lifeblood of implicit bias as bad luck obscures their structural and systematically pernicious character. Moreover, as we saw in Sect. 5.2, such proposals suggest that the best we can do is provide a kind of developmental rationale for discouraging implicit bias.

Looking at the problem from the lens of collective epistemic reasons can help us do better here. If we recall that they are primary in such cases, we can account for both intuitions equally elegantly without these potentially pernicious implications. First, we are epistemically blameworthy for such biases as a group, since such biases are the stuff of white ignorance. The fact that we bear responsibility as a group also accounts for why we (mistakenly) think that we are not individually blameworthy: after all, blaming it on the group seems to exonerate the individual. But (and second) all individuals with implicit biases are epistemically blameworthy in virtue of engaging in epistemic conduct that is contrary to the individual epistemic reasons that we inherit from the group (as per Sect. 5.3).²⁸ Both kinds of reasons are ultimately grounded in the imperative to cultivate an epistemically healthy, non-risky environment, free of the kinds of hermeneutical lacunas and epistemic vice which block knowledge.

If all this is on the right track, social-identity groups can have collective epistemic normative reasons for certain kinds of epistemic conduct. Such reasons are epistemic because the imperatives that undergird them are epistemic—to cultivate a healthy epistemic environment and attain knowledge. They are collective because they are had partly in virtue of the group’s being the kind of group it is. The arguments have been, admittedly, programmatic. But if they work, they have revealed an important kind of group reason. This will, hopefully, be of theoretical benefit both to group theorists and scholars of epistemic injustice.