Book Discussion: Lackey on Epistemology of Groups, Part I

We are excited to host the following discussion of Jennifer Lackey’s 2020 book The Epistemology of Groups (Oxford).

The discussion is in two parts: Part I, posted here, addresses group attitudes, while Part II, to follow on Wednesday, deals with group speech.

This discussion is the work of Brian Ball, Head of Faculty in Philosophy at New College of the Humanities at Northeastern University in London, with MA students Louise Bickerstaff and Henry Wang (Part I: Group Attitudes), and Will Francis and Florita Gudeikaite (Part II: Group Speech).

Part I: Group Attitudes

By Brian Ball, Louise Bickerstaff, and Henry Wang

Introduction

The first three chapters of Lackey’s book are concerned with group belief, justification, and knowledge respectively. Lackey is a realist about all of these ‘attitudes’ (as we’ll call them), but repudiates views on which they are subject to voluntary control – such as ‘orthodox’ joint acceptance accounts, on which groups believe things by virtue of (certain) members overtly endorsing them (with knock on effects for justification and knowledge). In our view, this complaint is essentially correct. And Lackey’s criticisms of what she calls ‘summative’ views are also strong. Her own positive proposals, however, are less compelling in our estimation.

Summativism and (Post-Gettier) Reductionism

Let’s begin with Lackey’s discussion of group belief. Lackey is concerned to argue against ‘summative account[s] – according to which a group’s believing that p can be understood in terms of the individual members of the group believing that p.’ (p.20) And she contrasts these with ‘non-summative account[s] of group belief – according to which a group’s believing that p is irreducible to some or all of its members believing that p.’ (p.24) It is not entirely clear, from these characterizations, whether non-summativism is intended to be (as one might expect) simply the negation of summativism – this depends on what ‘understood in terms of’ and ‘irreducible to’ mean (amongst other things) – but Lackey herself, somewhat surprisingly, offers a view (considered presently) which she regards as neither summative nor non-summative. Given the centrality of the distinction to her discussion, however, we would have welcomed greater precision here.

Turning to Lackey’s positive proposal, she offers the

Group Agent Account: A group, G, believes that p if and only if: (1) there is a significant percentage of G’s operative members who believe that p, and (2) are such that adding together the bases of their beliefs that p yields a belief set that is not substantively incoherent. (p.48)

Let’s set aside clause 2 here, to focus on 1. The notion of an operative member of a group that figures in it is borrowed from Tuomela, who says: ‘The operative members in… cases of… group beliefs are those… [individual] belief-formers by virtue of whom… beliefs are attributed to groups.’ (Tuomela, ‘Group Beliefs’, 1992, p.288) One immediate concern – setting aside nuances surrounding the distinction between belief and its (appropriate) attribution – is that appealing to this definition in an account of group belief is circular. 

Equally, the notion of a ‘significant percentage’ of the operative members of a group is vague – and we suspect that any attempt to make it precise (specifying a percentage) will result in errors in at least some contexts. This suggests that we resolve whatever context-sensitivity is involved by recourse to a prior understanding of the notion of group belief. Moreover, this – and the previously mentioned circularity – speaks to a wider issue raised by Lackey’s approach – that her method of articulating reductive definitions, identifying counterexamples and modifying accounts accordingly belongs squarely to the ‘post-Gettier’ paradigm of the late 20th century. It would be better, in our view, to simply deploy the notion of group belief in theorizing, articulating necessary connections with other notions, as appropriate, rather than trying to define it (in terms of necessary and sufficient conditions).

The Horizontal and the Vertical; or, Rationality and Aggregation

We turn now to a second cluster of issues suggested by Lackey’s discussion. Attitudes don’t (typically) occur in isolation. They (and their contents) stand in rational relations: some beliefs (the states or their contents) might (collectively) be coherent or incoherent; they might (evidentially) support, or undermine/defeat, another belief; and so on. Indeed, some consider such rational relations to be constitutive of attitudes of these kinds.

When it comes to group attitudes, the following question arises: should such rational relations, which yield holistic constraints on the attribution of attitudes, be taken into account before any aggregation of individual attitudes occurs (the ‘vertical’ approach), or after, so that they can be considered directly at the level of the group (the ‘horizontal’ approach)? (We can think of the aggregation process as having a vertical aspect to it, progressing from the low level of individuals to the high level of groups, while the global constraints have a horizontal character, remaining at a single level. The two approaches are then named after the last of the two steps in the process of determining group attitudes.)

Lackey seems to us to be rightly wary of approaches of both kinds. In her discussion of group belief, she insists that a group believes something only if some (operative) members do, and crucially that when the grounds of these members’ beliefs (assuming they have such) are aggregated, the resulting set of grounds is not incoherent. In short, she rejects a straightforwardly vertical approach to (the determination of) group belief (by individual members’ beliefs). Her resistance to the vertical approach is perhaps even more pronounced in relation to justified group belief: she argues that it leads to a ‘paradox of justification’, on which a group can end up with incoherent (and yet) justified beliefs; and it also yields the (unpalatable) consequence that (contrary to the epistemological consensus at the individual level) groups can have justified beliefs for which there are defeaters. At the same time, Lackey’s discussion of the doctrine of collective knowledge – pursued in connection with police groups, where there is US legal precedent – suggests that purely horizontal approaches are also problematic: if some group members know one thing (e.g. that this person matches the description of the suspect), and others know another (e.g. that if this person matches the description, then it is permissible to arrest them), it does not follow (contrary to at least some versions of the horizontal approach), that the group automatically knows what follows (e.g. that it is permissible to arrest this person); rather, on Lackey’s view, the group must take active measures to ensure that these rational relations between what are, prima facie, its attitudes are somehow consummated (e.g. through communication). The group can’t know something, despite no member knowing it, just in virtue of the rational relations available horizontally at the group level.

To repeat, these conclusions strike us as sound – and yet the argument for them might have been made more directly, by appeal to a standard epistemological distinction that Lackey herself notes, namely that between doxastic and propositional justification. Take the case of collective knowledge, understood horizontally: if one individual – and therefore the group – knows q, and another (hence the group) knows if q then p, it follows that p is propositionally justified for the group; but unless the group draws the inference and concludes that p, it may not so much as believe p, let alone have a (doxastically) justified belief that p – and even if it does believe p, unless its belief is caused (or causally sustained) by something that propositionally justifies it, that belief won’t be doxastically justified (or count as knowledge).

Social Knowledge

Lackey also discusses the notion of social knowledge, as developed by Bird (2013), which characterises group knowing as functionally analogous to individual knowing, rather than as supervenient on or reducible to it. Under this account, a group can know something without any individual member knowing, so long as members have reliable access to a relevant information deposit (e.g. a scientific journal). That group knowledge can occur without member knowledge puts Bird’s view in conflict – via the principle that knowledge implies justification – with Lackey’s own account of justified group belief, which requires that ‘some of the operative members of a group have the relevant justified beliefs themselves’ (p.111). 

Lackey suggests that Bird’s view leads to unacceptable epistemological consequences. In particular, she alleges that on the social knowledge view, it is permissible for a group to act whilst lacking knowledge, thus severing ‘the crucial [normative] connection between knowledge and action’ (p.137). However, Bird himself states that it would be ‘an abuse for institutions to make assertions that are not socially known to be true by that institution’ (Bird, 2013: 34). If Lackey can support her accusation, then Bird’s theory appears to be internally inconsistent.

Let’s examine Lackey’s attack against Bird’s social knowledge in detail. Lackey imagines a case in which each member of the scientific community is ignorant of an old paper published establishing that p, but that nevertheless the community as a whole acts in ways that are appropriate only if p – for instance, ‘approving cancer drugs’ (p.120) that depend for their safe use on p. By the principle that it is permissible to act on the basis of knowledge, it seems that Bird predicts that such acts are permissible; yet intuitively, they are not. 

Lackey then proceeds to turn the screw a little further. In an embellishment of the original case, she now imagines that the various members of the scientific community come to believe that not p, albeit incorrectly. It seems, then, that the group believes that not p in this case, and that this group belief is a defeater for its alleged knowledge that p – contrary to the claim that the group knows p (due to its earlier discovery).

In our view, however, these criticisms of Bird’s view fail. First, there is no reason why Bird must predict that the scientific community’s approval of the drug is permissible in the original case. Crucially, even though the community knows that p on his view, it does not base its action on this knowledge in an appropriate way – for example, it does not (we assume) first consult the journal article establishing that p when approving the cancer drug (otherwise its action would be perfectly in order). So there is a perfectly good account available to Bird of why the action is impermissible that does not conflict with his claim that the group knows. And second, there is no reason why Bird must hold to the view that even in the embellished case, in which a defeater is present, the scientific community knows that p. Why should he not allow, in line with epistemological orthodoxy, that adding a psychological defeater reverses an otherwise plausible verdict on a case? Lackey argues that the scientific community doesn’t believe both that p and that not p, since this exemplifies “the height of irrationality” (p.126). But sometimes individuals are irrational in this way – so why shouldn’t groups be?