Jennifer Lackey’s The Epistemology of Groups is a wonderful book, full of new insights and arguments which will shape research in this area in the decades to come. Here, I want to focus on one strand of this stimulating book, namely her novel account of group justified belief. Plausibly, whether a group justifiably believes that p is affected by the group’s evidence. For, one’s evidence can affect whether one’s belief is justified. Interestingly, Lackey doesn’t explicitly present an account of group evidence even though, as we will see, her account of group justified belief contains a condition relating to the evidence of its members. Roughly, she holds that a group justifiably believes p only if the total combined evidence of its members sufficiently supports p. It’s not immediately obvious why the status of a group’s belief as justified is hostage to what is supported by the combined evidence of its members, for it is controversial whether a group’s evidence consists in the sum total of the evidence of its members. In the following discussion, I will consider how Lackey might support this evidential condition within her account of group justified belief.

I start by presenting Lackey’s account of group justified belief and the key puzzle cases which she uses to motivate it. I then consider and reject two ways in which she could support her evidential condition: (1) a summative account of group evidence and (2) a certain ‘transmission’ principle. I end by suggesting an alternative account of group justified belief to Lackey’s.

Lackey presents a novel account of group justified belief which contains three main conditions: (1) a significant percentage of the key members of the group justifiably believe p; (2) they do so on bases which can be coherently combined; and (3) the total combined evidence that members do have and ought to have had sufficiently supports p. Given condition 1, she rejects classic non-summative accounts of group justified belief which allow that a group can justifiably believe that p even if none of its members do. Furthermore, given conditions 2 and 3, her account is not a summative account on which whether a group justifiably believes that p is a simple function of whether its members do. In more detail, she presents the account as follows.

A group, G, justifiably believes p if and only if:

  1. 1.

    A significant percentage of the operative members of G (a) justifiably believe p and (b) are such that adding together the bases of their justified beliefs that p yields a belief set that is coherent.

  2. 2.

    Full disclosure of the evidence relevant to p accompanied by rational deliberation about that evidence among members of G in accordance with their individual and group epistemic normative requirements would not result in further evidence that, when added to the bases of G’s members’ beliefs that p, yields a total belief set that fails to make sufficiently probable that p (Lackey 2021: 97).

A key part of Lackey’s defence of this account consists in her intuitive judgements about her museum guard cases (chapter 2). In each of these cases, most of a group of museum guards justifiably believe a certain claim but do so on different bases. In her ‘conflicting bases case’, or CBC, each of a group of 100 museum guards justifiably believes that a theft of a famous painting is being planned by one of the guards, or g. However, they do so on bases which cannot be coherently combined. For instance, guards 1–20 justifiably believe this on the basis that only guard Albert is planning a theft whereas guards 21–40 justifiably believe this on the basis that only guard Bernard is planning a theft. Lackey thinks that it’s intuitive that, in CBC, the group of museum guards does not justifiably believe g. Her account of group justified belief secures this result via condition 1b which requires that the bases of the members' beliefs can be coherently combined.

In a variant museum guard case, her ‘non-conflicting bases case’ (NBC), each of a group of 100 museum guards justifiably believes that a man was responsible for the theft of a famous painting (or m) where the bases of these beliefs can be coherently combined, but the total evidence of the guards when combined doesn’t justify the belief that a man was responsible for the theft. In particular, Lackey imagines that for each of the relevant bases, some guard has a defeater. Lackey thinks that it’s intuitive that in NBC, the group of museum guards does not justifiably believe m. She secures this result via condition 2 in her account of group justified belief which requires that the combined evidence of the members sufficiently supports m.

In what follows, I want to focus on whether we should accept the suggestion that a group justifiably believes p only if the combined evidence of its members sufficiently supports p (call that the ‘collective evidence condition’).Footnote 1 It’s not obvious that the collective evidence condition is correct even if we accept that whether a group justifiably believes p depends on the group’s evidence, for there are a variety of different accounts of group evidence and not all support the idea that a group’s evidence consists in the combined evidence of its members.

To see that, let’s start by considering potential summative accounts on which a group’s evidence is a function of the evidence of its members. According to pooled evidence, p is part of the evidence of a group if and only if it is part of the evidence of some member of the group. By contrast, according to shared evidence, p is part of the evidence of a group if and only if it is part of the evidence of most of the members of the group. Both of these summative accounts contrast with non-summative accounts of group evidence which deny that a group’s evidence is a function of the evidence of its members. For instance, Hedden (2019) identifies a group’s evidence with its knowledge, where he embraces a non-summative account of group knowledge. On such a non-summative account, it is possible that a group knows p even though no member does and that a group fails to know p even though every member does (e.g. Gilbert, List and Pettit). Thus, on Hedden’s account of group evidence, it’s possible that a group has p as part of its evidence even though no member does and that a group fails to have p as part of its evidence even though every member does. Notice that of these three accounts of group evidence, only the pooled account supports the claim that a group’s evidence consists in the combined evidence of all of its members. By contrast, on the shared account, evidence which is not shared by enough members is not part of the group’s evidence. And, on Hedden’s non-summative account, a proposition p could be part of the evidence of some member of the group but not part of the evidence of the group itself if p is not known by the group.

Even though pooled evidence would support the collective evidence condition and Lackey’s judgements about the museum guard cases, it’s hard for Lackey to appeal to pooled evidence to support her position both because doing so would fit badly with some of her other claims in the book and because the view is independently problematic. Lackey makes claims about knowledge that suggest that she would reject the pooled account of group evidence. In particular, in considering what she calls the ‘collective knowledge doctrine’, Lackey rejects the idea that a group’s knowledge is the sum total of the knowledge of its members (2021, chapter 3). But, that suggests that she would reject an analogous claim about evidence, namely that a group’s evidence is the sum total of the evidence of its members.

Furthermore, pooled evidence is open to significant objections (Brown, 2022). First, on a non-factive conception of evidence, pooled evidence would have the result that groups routinely have inconsistent evidence sets, for if one member of a group has p as part of her evidence and another member has not-p, then by pooled evidence, the group’s evidence includes p and not-p. Second, pooled evidence ascribes all of the evidence of members to the group. But, that’s surely implausible. For instance, if I’ve just signed a contract to move to a different institution but haven’t told anyone yet, then that’s part of my evidence but not my department’s. Third, pooled evidence is inconsistent with the widely endorsed doxastic constraint on which p is part of a subject’s evidence only if the subject bears the appropriate doxastic relationship to it. For that p is part of the evidence of a member of a group doesn’t entail that the group bears the required doxastic relationship to p, whether on a summative or non-summative approach. To illustrate, consider Williamson’s identification of evidence and knowledge and assume that p is known by some member of a group and so part of her evidence. By the pooled account of group evidence, it follows that p is part of the group’s evidence. But, that some member of a group knows p doesn’t entail that the group knows that p whether on a summative or non-summative approach to group knowledge. Thus, the pooled account is inconsistent with E = K.

Instead of appealing to pooled evidence, Lackey might attempt to defend the collective evidence condition in a different way. The broad idea of this strategy would be to argue that even if a group’s evidence cannot be equated with the total evidence of its members, the members play a fundamental role in the epistemology of a group’s belief as they are its source. Thus, it might be argued that the epistemic standing of a group’s belief that p depends on the epistemic position of its members with respect to p. This would be analogous to the suggestion in the epistemology of testimony that the epistemic standing of a hearer’s belief that p formed on the say-so of a speaker depends on the epistemic standing of the source of her belief, namely the speaker. In particular, some suggest that the following transmission principle governs testimony: a hearer gains justified belief that p on the basis of a speaker’s testimony that p only if the speaker justifiably believes p (call that ‘testimonial transmission’). This testimonial principle might inspire a ‘transmission’ principle governing the conditions under which a group can acquire justified belief that p from the opinions of its members:

Transmission (members to group): G justifiably believes that p on the basis of the opinions of its members only if the combined evidence of the members provides justification to believe that p.

If we apply this transmission principle to CBC and NBC, then it would support Lackey’s judgement that the group of museum guards doesn’t justifiably believe the relevant claim, for the combined evidence of the members doesn’t provide justification to believe the relevant claim. Furthermore, Lackey could use this transmission principle to support the collective evidence principle.

However, use of such a transmission principle to support the collective evidence principle faces the objection that transmission principles have been widely criticised in the testimony literature including by Lackey. Many authors accept that a hearer can come to justifiably believe p on the basis of the testimony of a speaker who doesn’t either justifiably believe p or even have justification to believe p. Thus, they allow that one’s epistemic position regarding p can be better than the epistemic position of the source of one’s belief that p. If that’s right in the case of testimony, it suggests that the epistemic position of a group concerning p can be better than the epistemic position of the source of its belief, namely its members, and so Transmission (members to group) is false.

Transmission principles in the epistemology of testimony have been rejected by appeal to a range of cases in which it is intuitive that a hearer can gain justified belief that p from a speaker’s testimony that p even though the speaker lacks justified belief that p. In one of Lackey’s examples, ‘creationist teacher’, pupils gain justified belief, and indeed knowledge, of certain evolutionary principles from the teaching of a creationist teacher who doesn’t believe the evolutionary principles she teaches. Lackey argues that as long as the creationist teacher is a reliable testifier about evolution and the students have no defeater for her testimony, they can gain justified belief and knowledge from her testimony (2008:48). In another of her examples, a hearer gains justified belief that p from the reliable testimony of a speaker who asserts p even though the speaker doesn’t have justification to believe p because she has a defeater for p that she doesn’t communicate via the testimony. For instance, Lackey imagines that Millicent possesses normal reliable vision and believes on this basis that p despite having misleading higher-order evidence that her vision is not reliable (she has been participating in certain scientific experiments, and the scientist involved has told her that her vision is not reliable due to an intervention that is part of the experiment). Millicent then testifies to her friend, Bradley, but without communicating the defeater. Bradley has no other defeater for p and has excellent evidence to trust Millicent, and thus, he gains justified belief that p from Millicent even though Millicent herself lacks justification to believe p given her undefeated defeater for p (59–61). In the light of these counterexamples, Lackey rejects Testimonial Transmission. Furthermore, Testimonial Transmission is rejected by a range of approaches to testimony including reliabilist approaches, reductionist approaches, and hybrids of reductionism and reliabilism.Footnote 2 And it has been rejected not only in the case of individual testimony but also in cases in which a hearer gains a belief from the testimony of several testifiers (e.g. Leonard, 2018).

In the light of these challenges to testimonial transmission principles, it would be difficult for Lackey to defend the collective evidence condition and her judgements about CBC and NBC by appeal to Transmission (Member to Group). For her own views in the epistemology of testimony allow that a hearer can gain justified belief that p from a source—another speaker—even where that source does not justifiably believe p, or even have justification to believe p. Similarly, it seems that she should allow that a group can justifiably believe p from some source—here the opinions of its members—even if their combined evidence would not yield justification to believe p.

Indeed, considering the epistemology of testimony has the potential to provide a very different style of account of group justified belief than Lackey’s, for if the members of a group are considered to be distinct agents from the group itself, then the members can testify to the group, and the group can testify to its members. Thus, we might consider group belief formed by aggregating the expressed opinions of members as illuminated by the epistemology of testimony (Brown Forthcoming). Of course, the epistemology of testimony is itself controversial. But on one broadly reliabilist account of the epistemology of testimony, a speaker gains justified belief of that p from the testimony of a speaker that p if and only if the speaker is a reliable producer of testimony, the hearer is a reliable consumer of testimony, and the consumer lacks a defeater for p. As illustrated by the example of Millicent/Bradley, this can be the case even if the speaker does not themselves justifiably believe p and the speaker’s evidence does not support p. So if we apply this testimonial model to group belief formed by aggregating the opinions of members, we would end up with a view on which a group could justifiably believe p by aggregating the opinions of its members even though none of the individual members justifiably believe p and even though their evidence when combined does not provide justification to believe p. Instead, all that would be required is that most of the individual members are reliable producers of testimony, the group is a reliable consumer of testimony, and the group lacks a defeater for p. Thus, we could end up with an account on which the conditions for group justified belief are very different from those in Lackey’s account.

Setting aside this tentative positive suggestion for an alternative account of group justified belief, the main purpose of this discussion has been to raise a question about the evidential condition in Lackey’s account of group justified belief and in particular the suggestion that a group justifiably believes p only if the combined evidence of the members of the group sufficiently supports p. I have argued that this condition requires defence given that, plausibly, whether a group’s belief is justified depends on the group’s evidence, but it’s controversial how to understand group evidence. I’ve considered but rejected two ways in which Lackey might support this evidential condition: by appeal to a pooled account of group evidence or a certain ‘transmission’ principle. In the light of these problems, I ended by sketching an alternative account of group justified belief based on the epistemology of testimony.