Jump to ContentJump to Main Navigation
The Epistemic Life of GroupsEssays in the Epistemology of Collectives$

Michael S. Brady and Miranda Fricker

Print publication date: 2016

Print ISBN-13: 9780198759645

Published to Oxford Scholarship Online: April 2016

DOI: 10.1093/acprof:oso/9780198759645.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 21 June 2018

Changing Our Mind

Changing Our Mind

(p.111) 6 Changing Our Mind
The Epistemic Life of Groups

Glen Pettigrove

Oxford University Press

Abstract and Keywords

If groups are to prove capable of possessing and increasing their knowledge, then they will need to be capable of revising their beliefs. While the nature of collective knowledge has begun to receive more attention in recent years, surprisingly little work has been done on the process of belief revision that makes it possible. And that which has been done has focused on a very particular kind of community and a very particular kind of knowledge, namely the scientific community and scientific knowledge. This paper aims to extend the analysis of collective belief revision to a different sort of community and a different sort of knowledge, namely, moral communities and moral knowledge. The chapter will suggest that, whatever one’s preferred account of what groups know, we need a richer account of belief revision to support it than has been offered thus far. And it will propose one such alternative

Keywords:   belief, revision, collective, moral, communities, knowledge


For creatures like us, the journey to knowledge follows an oddly circuitous route. We seldom proceed by steady steps in a single direction. Rather, we follow hunches, make intuitive leaps, wander down blind alleys, backtrack, and suddenly stumble upon a more promising path. So if we are to prove capable of increasing our store of knowledge, one of the things we must learn is to revise our beliefs. The same holds true for the groups of which we are a part. If they are to prove capable of possessing and increasing their knowledge, they too will need to be capable of revising their beliefs. While the nature of collective knowledge has begun to receive more attention in recent years, surprisingly little work has been done on the process of belief revision that makes it possible. And that which has been done has focused on a very particular kind of community and a very particular kind of knowledge, namely the scientific community and scientific knowledge. My aim is to extend the analysis of collective belief revision to a different sort of community and a different sort of knowledge, namely, moral communities and moral knowledge. This chapter will suggest that, whatever one’s preferred account of what groups know, we need a richer account of belief revision to support it than has been offered thus far. And it will propose one such alternative.

1. Current Accounts of Collective Knowledge and Collective Belief Revision

Over the past twenty-five years Margaret Gilbert has been developing a distinctive account of collective agency (1989, 1996, 2000, 2013). In the course of so doing she has defended not only claims about collective agents and their actions (p.112) but also claims about collective beliefs and collective emotions. According to her account of collective belief: ‘There is a collective belief that p if some persons are jointly committed to believe as a body that p’ (2000, 39). Persons are jointly committed to a belief if ‘each of the parties has expressed his or her personal willingness to be party to it in conditions of common knowledge’, that is, in conditions where each of the parties committing to the belief ‘has expressed his or her personal willingness to be a party to the joint commitment’ (2000, 40). Under such conditions, she argues, it is appropriate to say: ‘We believe that p.’1

Starting from an account like Gilbert’s of (a) collective belief that p, one might add the familiar conditions of (b) the group’s warrant to believe that p, and (c) the truth of p, in order to generate an account of group knowledge that p which parallels accounts of individual knowledge.2 Of course, one might challenge such an account for reasons that have been well rehearsed in contemporary epistemology. One might worry, for instance, that conditions (a)–(c) fail to distinguish between cases of good epistemic luck of the sort Gettier (1963) highlighted and cases of ‘genuine’ knowledge. Or one might prefer a virtue-epistemic account of knowledge, according to which: ‘Knowledge is a state of cognitive contact with reality arising out of acts of intellectual virtue’ (Zagzebski 1996, 270). Or one might challenge the account on the grounds that it fails to be sufficiently attentive to the fact that our ascriptions of knowledge are context sensitive, such that we hold agents to quite different epistemic standards in different conversational contexts (DeRose 2000). More will be said about knowledge and other epistemic states, like understanding, in section 2; however, for the most part the discussion will set these familiar disputes among epistemologists to one side. This is not because they are uninteresting or unimportant. On the contrary, those working on collective epistemology would do well to pay more attention to the implications of the conversational contexts in which we ascribe knowledge to groups.3 (p.113) And whether or not one can offer a satisfactory virtue-epistemic account of knowledge,4 the notion of an epistemically virtuous collective is a topic that deserves further attention. Rather, these issues are set aside because there is an important set of questions regarding what is involved in the revision of group beliefs that will remain, no matter how these disputes about the best definition of knowledge are ultimately settled. Consequently, I shall focus on issues relating to group belief revision.

As Gilbert notes, collective beliefs can be remarkably difficult to change. To begin with, there is the basic challenge of changing the mind of not just one but many members of the group. How many members’ minds must be changed is an interesting question. Gilbert herself has expressed reservations about ascribing mental states to a group if less than half of the group members experience it (2002, 133); whereas Raimo Tuomela has argued that there may be cases where a group believes something ‘that possibly no single member finds privately acceptable (think of a case of voting in which no one’s first choice is elected)’ (2007, 137). While I prefer a criterion that would permit the ascription of a belief to a group even in (some) cases where fewer than half of the group’s members believe it (in what Tuomela calls the I-mode; see Pettigrove and Parsons 2012), my central argument does not depend upon the outcome of this debate.

There is also the socio-epistemic challenge of changing people’s minds in a context in which it is believed that many other people accept the belief that one is thinking about setting aside. We are profoundly influenced by the fact that other people believe some proposition, p (Cialdini 2001). Furthermore, we ought to be influenced by this fact (Zagzebski 2012, 52–74). So the belief that a number of other people believe p can give p a high level of credence within a population even when that population has been presented with strong evidence to the contrary.

Third, there is a distinctive set of normative issues raised by collective belief revision. By committing to a collective belief, an individual places herself under an obligation to the other members of the collective ‘to constitute—as far as is possible—a body that believes that p’ (Gilbert 2000, 41). This will include acting in ways that are consistent with the belief, such as asserting it in appropriate contexts and not calling it or beliefs that it necessarily presupposes or entails into question. There is some scope for an individual’s personal view to diverge from the view of the group of which she is a part. Nevertheless, if she expresses her (p.114) disagreement it is incumbent on her to flag the fact that she is only speaking as an individual, rather than as a representative of the group. And when she does so, it raises questions regarding her fidelity to the group and she may risk losing her standing within it (Gilbert 2000, 41).

Finally, there is the structural challenge of getting a change of mind from the individual to the collective level. The structure of some collective agents may be such that this requires little more than that a significant number of those who make up the collective change their minds and are aware that others have too. Within other collectives, however, the change may require that key office-holders change their mind, or that the change comes to be reflected in authoritative documents that officially represent the mind of the group.

Gilbert offers the following as a way of characterizing the conditions that, if met, would constitute collective belief revision. ‘For a body of which I am a member to change its beliefs requires something akin to an agreement to stop believing that p together and to start believing that q instead’ (Gilbert 2000, 46). This characterization provides a useful starting point for the analysis of group belief revision. However, I shall argue that the account she has offered will need to be modified before it can adequately cover a range of cases of group belief revision that ought to be central to any account of group belief revision, but especially to Gilbert’s account.

Gilbert builds her discussion of group belief revision around an imagined scientific community. I shall work with a different community that in many ways better exemplifies the conditions of group belief that she identifies, namely, the Presbyterian Church (USA). One reason for focusing on the PC(USA) is that, unlike most academic communities, which tend to have a rather loose structure and are fairly reticent about committing themselves to collective beliefs, the PC(USA) explicitly presents itself as a collective that believes a number of things. The current Constitution of the Presbyterian Church (USA) is composed of two parts: The Book of Confessions and The Book of Order. In relation to the first of these parts, the constitution states: ‘In these statements the church declares to its members and to the world who and what it is, what it believes, and what it resolves to do. These statements identify the church as a community of people known by its convictions as well as its actions’ (Presbyterian Church 2013, F-2.01). These confessional documents often begin with ‘We believe…’, and are sprinkled throughout with similar assertions such as ‘We confess and acknowledge…’ and ‘We declare…’.

Furthermore, these collective beliefs have arisen out of a process that satisfies Gilbert’s conditions for constituting a joint commitment. Presbyterians have a representative form of government that is built around regional units called (p.115) presbyteries, which are composed of the clergy in a particular region plus lay representatives from each of the churches in that region. The standard process whereby something makes it into the constitution of the church is as follows. Initially a member of a presbytery presents a proposal to their presbytery for consideration. After the presbytery has discussed the proposal a vote is taken. If a majority in the presbytery favour the proposal, it is sent to the national governing unit, the General Assembly, for consideration.5 The General Assembly, which is composed of clergy and lay representatives from around the United States, discusses the matter and then votes upon it. If a majority within the Assembly approve, then the proposal is sent back down to the presbyteries—this time all of the presbyteries—for their consideration. If a majority of presbyteries approve the proposal,6 it is sent to the General Assembly once more for final ratification. By the time a proposal has made it through this process, the parties to the collective belief have expressed their ‘willingness to be party to it in conditions of common knowledge’ (Gilbert 2000, 40).

Multi-generational collectives pose a special challenge, insofar as one might be concerned about the ability of past members to commit current and future members to believe something. But here too, the Presbyterian Church (USA) is well situated to deal with the challenge. In the case of the confessions in the first part of the constitution, the beliefs that have been endorsed by the church are widely disseminated. Selections from the confessions are often incorporated within the liturgy of Sunday morning church services. They play an important role in the theological education of both laity and clergy, and when individuals within the church are set aside for leadership roles, they agree to be guided by them. Hence, if any group satisfies Gilbert’s condition for constituting a body that believes that p, it is the Presbyterian Church (USA).

Not only does the PC(USA) satisfy Gilbert’s conditions for collective belief, it also provides a useful test case for an account of collective belief revision. The church has long identified itself with the motto: Ecclesia reformata, semper reformanda—‘the church reformed, always reforming’ (Presbyterian Church 2013, F-2.02). This reforming spirit has expressed itself at various points in the church’s history in, among other ways, the revision of parts of its constitution. By far the most dramatic of these revisions took place in 1967. At a structural level, there were two principal changes. The church went from having a single (p.116) confessional document—the Westminster Confession of Faith of 1646—to having a book of confessions containing historic documents from multiple time periods and locations. The second structural change was the addition of a newly written confession, the Confession of 1967.7

These structural changes brought with them important changes in content as well. And some of the changes look as though they fit Gilbert’s account quite nicely. For example, prior to 1967 leaders within the Presbyterian Church were asked to affirm ‘that the Westminster standards contained the system of doctrine taught in Scripture’.8 The standards in question were those produced by a committee set up by the English Parliament that met in Westminster Abbey from 1643 to 1646, which were adopted by the Church of Scotland in 1647 and subsequently endorsed by the English and Scottish parliaments. As one might expect, the moral and political worldview of American Presbyterians in the 1960s was somewhat different than the worldview of English and Scottish theologians and parliamentarians at the time of the English Civil War. The Westminster Divines, for example, insisted that it was the duty of the civil magistrate to suppress ‘all Blasphemies and Heresies’ and prevent ‘all corruptions and abuses in Worship and Discipline’ (Leith 1982, 220), and it is clear that they considered the Roman Catholic Church to be among those whose teachings should be suppressed.9 Well before the 1960s American Presbyterians had removed references to the suppressing of heresies, and so on, from the magistrate’s list of duties and had inserted in its stead, ‘no law of any commonwealth should interfere with, let, or hinder, the due exercise thereof, among the voluntary members of any denomination of Christians, according to their own profession and belief’ (Presbyterian Church 1999, 6.129). The Confession of 1967 takes the matter a step further, asserting that: ‘The Christian finds parallels between other religions and his own and must approach all religions with openness and respect’ (Presbyterian Church 1999, 9.42).

If one were to represent this shift in Gilbert’s terms, one might use p1 and p2 to represent the propositions, ‘Christians should reject the teachings of Catholics, Hindus, Muslims, etc. as false’ and ‘The civil magistrate should suppress them’ and q1 and q2 to represent the propositions, ‘The civil magistrate should not suppress the teachings of Catholics, Hindus, Muslims, etc.’ and ‘Christians should be open to and respectful of the truths they might disclose’. One might (p.117) then say that at some point between 1647 and 1967 the PC(USA) agreed to stop believing together that p1 and p2 and to start believing that q1 and q2 instead. However, even if Gilbert’s account of collective belief revision works for some cases, it does not work for others. And even some of the cases it appears to fit can be more perspicuously described. Or so I shall argue. However, before I present that argument it will be useful to get some other conceptual tools in place first.

2. From Epistemology to Collective Moral Epistemology

There are a number of different things that have marched under the banner of ‘knowledge’. One of these was mentioned above, according to which an agent with knowledge of p, (a) believes that p, (b) is warranted in believing that p, and (c) p is true.10 Call this propositional knowledge-that. Jonathan Kvanvig (1992) and others have criticized formulations like this, arguing that while they fit some instances of knowledge they fail to capture others. One problem, Kvanvig argues, is that they are too focused on propositions, which are treated as discrete atomic units that are isomorphic with the sentences of a natural language. Many instances of knowledge do not have such propositions as their objects. He gives as an example the knowledge someone has after standing on the edge of the Grand Canyon and gazing at it for a few minutes. If asked afterward whether the canyon is longer than it is wide or deeper than an elephant is tall, the observer would confidently answer, ‘Yes’. But this is not because those two propositions were among the bundle of discrete atoms of knowledge that he acquired while standing at the canyon’s edge and which he is now carrying around in his head. The object of knowledge that he acquired was something more holistic, which he is able to carve up in various ways when he finds it useful to do so (Kvanvig 1992, 180–2). For convenience let us call this holistic knowledge-that.

In recent years a number of people have suggested that, in addition to the kinds of knowledge with which contemporary epistemologists have primarily been concerned, there are also other truth-related states that at earlier points in the history of western philosophy were the main concern of epistemologists. Knowledge so construed, which Linda Zagzebski calls ‘understanding’, is a matter of seeing or grasping relations between things, concepts, meanings, or propositions. (p.118) It is exemplified by the scientist who not only possesses knowledge of discrete facts but grasps the relationships that hold between them, fitting them into an overarching theoretical framework, seeing them as parts of a larger whole. It is manifested by the skilled reader who does not merely follow the basic plot of a novel, for example, but recognizes themes that run through the work, or through the author’s wider corpus, or through the western literary tradition, or that connect the work with other events in the world as she interprets the novel (Zagzebski 2001, 237; Baehr 2011, 146–8). And it is disclosed in the philosopher’s recognition of relations of entailment, necessity, and possibility and the mathematician’s comprehension of a proof.11

Each of the kinds of knowledge mentioned above can be seen in contexts where the object of concern is moral knowledge. An agent can have knowledge-that of moral propositions as well as holistic moral knowledge-that. She can also have moral understanding, grasping moral concepts, meanings, and relations that might hold between them. ‘Moral understanding’, Zagzebski observes, ‘includes seeing the connection between moral reasons and moral judgments, and perhaps also the connection between certain emotions and moral judgments. Understanding permits us to see how to extend a moral judgment to different situations, and to see how distinct moral judgments relate to each other’ (2012, 175). However, even if one prefers to work with a narrower conception of knowledge that excludes epistemic states like understanding, Allan Gibbard has suggested that one need not be troubled by talk of moral knowledge. There may be differences between moral knowledge and knowledge of other kinds, but whatever their differences, it is reasonable to think their rough outlines will be similar enough that we can treat them as essentially of a piece.12

To this point the examples we have discussed have been examples of things known by individuals. However, it would seem that we can draw the same (p.119) distinctions with respect to collective knowledge that we do with respect to individual knowledge. We can refer to collective propositional knowledge-that, collective holistic knowledge-that, and collective understanding, as well as to collective moral knowledge of each of these types. One might worry that the ascription of knowledge to collectives requires one to buy contentious metaphysical assumptions about collective mental states. However, this does not follow from references to collective knowledge or collective beliefs any more than it follows from a theorist’s use of folk concepts like belief, thought, and desire when talking about individuals that she is committed to a metaphysical framework in which these are distinct pieces of mental furniture that possess causal powers similar to the powers of material objects. We can ascribe collective knowledge of the above-mentioned types without taking a stand on whether collective knowledge and collective belief are (a) novel emergent properties that supervene on properties of the individuals who make up the collective, or (b) a shorthand way of referring to qualities that are ultimately reducible to properties of individuals.

Consequently, the move from a discussion of individual knowledge to a discussion of collective knowledge is less metaphysically contentious than it might seem. And making such a move can draw our attention to epistemic qualities of groups that might otherwise go unnoticed. One such quality is the surprising reliability of the average judgment of certain sorts of groups on a wide range of questions. Francis Galton (1907) drew attention to group reliability in the early twentieth century when he noticed that the statistical average of the judgments of people at an English fair who estimated the weight of an ox came within 1 per cent of the actual weight of the animal. This result had been anticipated in the eighteenth century by Nicolas de Condorcet’s jury theorem, and similar results have been replicated in cases involving estimating the temperature in a classroom, producing a rank-ordering of the sizes of very similar piles of buckshot, predicting the winners of baseball playoffs and American presidential elections, and even in choosing the best move in a game of chess (Sunstein 2006, 21–43).

Another domain that comes more clearly into view when we attend to the epistemic attributes of collectives is the body of knowledge that is commonly referred to as tradition. A tradition is a way of making sense of ourselves and our world that is shared with and inherited from others: ‘What we receive from the past are, in effect, beliefs, persuasions, convictions; that is, ways of “holding for true”…’ (Ricoeur 1988, 222–3), ‘ways of seeing and of speaking about the world’, ways of acting and interacting (Williams 1983, 12, 20–1; MacIntyre 1983, 221–2). Our tradition reflects the wisdom (and sometimes the folly) of our forebears and (p.120) shapes our sense of who we are, who we aspire to become, what we are doing, and what we ought to be doing. It is preserved in rituals, texts, habits, social structures, architectural styles, and countless other forms and features of life.13 Born as we are into a tradition-saturated world, we imbibe it from our earliest moments, as we learn to think, speak, and act. ‘We are carried along by it before we are in a position of judging it, or of condemning it’ (Ricoeur (1988, 223). It makes and makes-sense-of our characteristic ways of going on, and it is revealed both in what we think about self-consciously and in what we take for granted when thinking (Gadamer 1989, 276–7; Taylor 2004, 23–30).

3. Traditions and Collective Belief Revision

With these distinctions in place, we are now in a better position to see why Gilbert’s account of collective belief revision is inadequate as it stands. The first reason is its focus on belief as a state whose object is a proposition.14 If we can have holistic knowledge-that and if the holistic qualities of our epistemic experiences can be shared by a group of which we are a part, then an account of belief revision that characterizes the object of collective belief as a proposition will be incomplete, insofar as it leaves out an entire category of beliefs, namely those involved in collective holistic knowledge-that.15

There is good reason to think that if individual believers have such non-propositional belief contents, collective believers could, too. To revert to Kvanvig’s Grand Canyon example, one could imagine a group of tourists standing at the edge of the canyon forming a collective belief on the basis of their shared experience. If one requires joint commitment as a condition for collective belief, then to some degree their shared experience may need to be put into words. But these words might themselves simply be a placeholder for the not yet articulated—and perhaps not fully articulable—non-propositional belief content. (p.121) For example, one of them might say, ‘The Grand Canyon is big’, and the others might agree to share this belief, without any of them thinking that the proposition ‘The Grand Canyon is big’ captures the full content of their shared belief. Indeed, the expression of shared belief need not wear the guise of a proposition at all. Standing at the canyon’s edge, one of them might simply say, ‘Wow!’ And the others might echo, ‘Yeah, Wow!’ Against the backdrop of a standing commitment within the group to share beliefs based on shared experiences,16 these expressions of wonder could be sufficient to give rise to a shared belief with holistic content. Within the context of a religious community that gathers regularly for prayer, meditation, and worship, whose members share liturgically mediated religious experiences, and who are jointly committed to sharing religious beliefs, we should expect shared beliefs with holistic content to be fairly common.

Hilary Putnam draws attention to another reason we might expect such communities to share beliefs with holistic content. Often practical reasoning involves imagining what it would be like to pursue the different courses of action that are available to us at a particular moment. The mountain climber who is considering two different possible routes to the top imagines what each route will involve and uses this process to direct his next step.17 The quandary the imaginer faces may also run deeper than this, perhaps calling into question who she is or who she should become. The person in mid-career who is considering a change of vocation will imagine the alternative career path and compare it with what she imagines her future will look like if she remains in her current vocation.18 Our imaginative engagement with these possibilities can enable us to see which course we should take. Putnam observes:

this sort of reasoning need not at all be reducible to any kind of linear proposition-by-proposition reasoning…Of course, saying that this is not linear propositional reasoning is not to deny that after he has imagined ‘what would happen if…’ he can put the relevant considerations into words. It is to say that he need not have stored the relevant information in the form of words.

(Putnam 1978, 86)

(p.122) Preaching often makes use of precisely this sort of imaginative engagement, as is illustrated in Frederick Buechner’s classic sermon, ‘A Sprig of Hope’:

Let us start with the story itself; more particularly, let us start with the moment when God first spoke to Noah; more particularly, let us start with Noah’s face at that moment when God first spoke to him. When somebody speaks to you, you turn your face to look in the direction that the voice comes from; but if the voice comes from no direction at all, or the voice comes from within and comes wordlessly, and more powerfully for being wordless, then in a sense you stop looking at anything at all…Your face goes vacant because for the moment you have vacated it and are living somewhere beneath your face, wherever it is that the voice comes from. So it was maybe with Noah’s face when he heard the words that he heard, or when he heard what he heard translated clumsily into words: that the earth was corrupt in God’s sight, filled with violence and pain and unlove—that the earth was doomed.

It was presumably nothing that Noah had not known already, nothing that any man who has ever lived on this earth with his eyes open has not known. But because it came upon him, sudden and strong, he had to face it more squarely than other people usually do, and it rose up in him like a pain in his own belly. And then maybe, like Kierkegaard’s Abraham, Noah asked whether it was God who was speaking or only the pain in his belly; whether it was a vision of the glory of the world as it first emerged from the hand of the Creator that led him to the knowledge of how far the world had fallen, or whether it was just his pathetic human longing for a glory that had never been and would never be. If that was his question, perhaps a flicker of bewilderment passed across his vacant face—the lines between his eyes deepening, his mouth going loose, a little stupid. A penny for your thoughts, old Noah.

But then came the crux of the thing, because the voice that was either God’s voice or an undigested matzoh ball shifted from the indicative of doom to the imperative of command, and it told him that although the world was doomed, he Noah, had a commission to perform that would have much to do with the saving of the world.

(Buechner 1994, 228–9)

The preacher invites the gathered community into a story, often a story that plays an important role in structuring and preserving this community’s sense of who it is and how it ought to live, a tradition-shaping story. Engagement with the story enables members of the community to reflect on their own past experiences and to anticipate a range of responses to future experiences.19 It encourages them to notice certain features of their circumstances that might otherwise have gone unnoticed and offers them alternative ways of making sense out of those and (p.123) other aspects of their lives. This sort of imaginative engagement through story can function like the shared experiences of the group standing at the edge of the Grand Canyon. It can provide the object of their joint belief. And when the story invites them to reflect on future decisions the group might make—as Martin Luther King, Jr.’s ‘I Have a Dream’ address did for many churches—the group’s imaginative engagement might function like that of the mountain climber deliberating between two routes or the person in mid-career deliberating about vocational paths.

To accommodate these sorts of cases, we might modify Gilbert’s account by adding variables (p+ and q+) to represent the sorts of objects that beliefs take in cases of holistic knowledge-that:

A group g revises its belief that p or p+ iff the members of g (or the operative members of g) agree to stop believing that p or p+ and start believing that q or q+.20

Such an account will be better suited to manage a wider range of cases. However, it does not yet leave room for the kind of belief revision involved in acquiring a greater depth of understanding. To begin with a familiar example, consider what is involved in coming to understand modus ponens. A student might believe p, and p→q (indeed she might even believe q), but fail to see that q follows from the conjunction of the other two propositions. The student who finally comes to understand modus ponens has a different belief set than she did beforehand. But she has neither rejected a belief she previously held nor simply added another proposition to her prior set of beliefs (Carroll 1895). Rather, she has come to understand something new about the relationship between the objects of beliefs she already accepted. And if her prior beliefs that p, p→q, and q were warranted and true, we might say she has come to understand what she already knew.

A similar sort of drawing together and making new connections between distinct and perhaps already familiar beliefs can be seen in the context of moral (p.124) and theological understanding, as well. For example, the Confession of 1967 used the theme of reconciliation—borrowed from a passage in Paul’s second letter to the Corinthians—to reorganize and reinterpret a number of the church’s historic doctrines in a way that highlighted their relevance to social issues at the forefront of American consciousness in the 1960s. The result was a call to social action that was rooted in a distinctive understanding of who God is, who the church is, and what the church is called to do that differed in a number of respects from the vision articulated in the Westminster Confession. The Westminster Divines did not feel the need to address questions of racial justice, and for more than two centuries many of those churches who used some version of the Westminster Confession as their statement of faith deemed it compatible with a system whereby members of one race enslaved those of another.21 By contrast, the Confession of 1967 asserts, ‘the church labors for the abolition of all racial discrimination and ministers to those injured by it. Congregations, individuals, or groups of Christians who exclude, dominate, or patronize their fellowmen, however subtly, resist the Spirit of God and bring contempt on the faith which they profess’ (Presbyterian Church 1999, 9.44). Similarly, even though they are drawing on shared theological resources, the Westminster Divines would have been surprised by the critique of nationalism expressed in the Confession of 1967, as well as by the claim that: ‘The church, in its own life, is called to practice the forgiveness of enemies and to commend to the nations as practical politics the search for cooperation and peace’ (Presbyterian Church 1999, 9.45). To accommodate the possibility of such revisions in collective belief, we need to add what we might call an understanding operator (u) to our account and remove the necessity of ceasing to believe that p. If we let u(Rpq) stand for an understanding of a relationship that exists between p and q, then:

A group g revises its belief that p or p+ iff the members of g (or the operative members of g), who previously believed that p or p+, agree to believe something new, such as (i) that not-p, or (ii) that q or q+, or (iii) that p or p+ and u(Rpq).

Revisions of belief that involve changes in moral understanding characteristically involve another sort of change that it would be useful to capture in our account of collective belief revision, namely, changes in conception. Our most primitive evaluative judgments are built around encounters with exemplars (Zagzebski 2004, 40–50). Some of these exemplars are ordinary people who happen to be (p.125) among our narrow circle of associates at an early stage in our lives. Others may stand out in ways that come to be recognized in a much wider community. Let us imagine some such individual who has been rightly picked out by the community as an especially admirable person. In referring to what is admirable about this person, the members of the community use some word—for example, just or loving—to designate the trait in question. They may also discuss the nature of the property they have picked out. Since their ability to pick out the exemplar differs from their ability to describe what is admirable about her, it would not be surprising if this discussion were quite illuminating in some respects, even though it was obfuscating in others. Subsequent speakers, perhaps even subsequent generations of speakers, may be left to work out a more faithful or complete account of the nature of the quality first identified by the community. This process may be complicated by the way in which earlier speakers attempted to describe the nature of the quality, since some of what they built into the description may mislead. Furthermore, given the way our concepts shape the world we see, once the concepts we are working with have been influenced by prior descriptions, we are likely to see the exemplar in a distinctive way that reflects details of these descriptions.

Something like this story seems plausible as an account of the origins of many of the moral judgments preserved within a tradition. And as W. B. Gallie noted in his seminal paper on the topic, it is also an ideal recipe for the development of essentially contested concepts, that is, ‘concepts the proper use of which inevitably involves endless disputes about their proper uses on the part of their users’ (1955–6, 169). When combined, (a) the complex nature of the quality in question, (b) the equally complex range of the contexts in which it might or might not be appropriately manifested, and (c) the different vantages from which observers might perceive and make sense of the quality, can easily generate competing specifications of the quality in question—what John Rawls called rival conceptions of the concept or quality.22

Gallie himself contended that many of our ethical concepts are essentially contested. Whether one thinks this is correct as a claim about secular ethics or not, it is exceedingly plausible in the case of theological ethics. Theological ethics is concerned with how one should live in the light of what one takes to be a divine reality. Christian theological ethics, which is the variant most relevant to the example we have been discussing, is concerned with how one should live ‘in the (p.126) presence of a gracious God’ (Tracy 1996, 52), whose nature was revealed in the person, work, and teachings of Jesus of Nazareth (Presbyterian Church 1999, 9.07–15). Thus, Christian theological ethics combines the features of the historic exemplar story above with ‘a mystery beyond the reach of man’s mind’ (Presbyterian Church 1999, 9.15). This means that communities like the PC(USA) will find themselves continually drawn back to important accounts of what early Christians heard and saw of Jesus and his ministry. But at the same time, there are at least three reasons why they should find it challenging to understand what was heard and seen. First, the initial observers had difficulty understanding the person they saw and heard.23 Second, the way in which the twenty-first century community interprets those early texts is mediated by two thousand years of interpretation and ritual which themselves originated in and were shaped by remarkably different cultural and conceptual horizons. Third, the reality being disclosed, in the light of which the community seeks to live, is by its very nature supposed to be beyond the bounds of the community’s comprehension.24 Consequently, we should expect the moral reflections of communities like the PC(USA) to be marked by an interesting tension between a conservative and an innovative impulse, between recovering and discovering, remembering and revising.25

One of the more significant changes expressed in the Confession of 1967 concerned precisely this tension. It involved rival conceptions of the inspiration and authority of scripture. The conception advanced by the Westminster Confession claimed that the books of the (Protestant) Old and New Testaments were ‘immediately inspired by God, and by his singular care and providence kept pure in all ages’ (Presbyterian Church 1999, 6.008). Under the influence of several generations of Princeton theologians in the nineteenth century, this conception was taken to entail, among other things, ‘the inerrancy of the Bible’. And starting (p.127) in 1910, endorsement of this inerrantist conception was made compulsory for all candidates for ordination (Rogers 1991, 205). The Confession of 1967’s comments on Scripture, by contrast, are rooted in an alternative conception of inspiration and authority.

The Scriptures, given under the guidance of the Holy Spirit, are nevertheless the words of men, conditioned by the language, thought forms, and literary fashions of the places and times at which they were written. They reflect views of life, history, and the cosmos which were then current. The church, therefore, has an obligation to approach the Scriptures with literary and historical understanding.

(Presbyterian Church 1999, 9.29)

According to this latter conception, the inspiration and authority of Scripture was compatible with it also being a human—sometimes all too human—document that reflected the human author’s imperfect, albeit instructive, understanding of divine reality. And this conception carried with it new possibilities for thinking about what it means for the actions of today’s PC(USA) to be guided by these ancient texts.

This alternative conception of authority was further supported by the expansion of the Presbyterian constitution to include a larger number of confessional documents. For a considerable portion of its history, candidates for ministry in the church had to indicate their agreement with the Westminster Confession. If they disagreed with the Westminster Divines on any point, they had to identify that fact and defend their position before the Presbytery. However, the expansion of the ‘subordinate standards’ contained in the Book of Confessions in 1967 reflected a decisive shift in perspective. Some of the documents that were added, like the Nicene Creed and the Apostles’ Creed, were already routinely used in Presbyterian liturgy. Others, like the Scots Confession and the Heidelberg Catechism, had played an important role in the church’s past and continued to play an important role in other contemporary denominations in the Reformed tradition, but had fallen out of use in most Presbyterian Churches. Still others, in particular the Declaration of Barmen and the Confession of 1967, were twentieth-century documents explicitly written as a response to the needs of their current social and political moment. The most interesting change introduced by this expansion of the church’s constitution arose out of the multiplicity of voices that were brought to stand alongside those of the Westminster Divines. This represented a shift in thinking that had already taken place in the church, which the new structure was instituted to reflect as much as to create. It represented a different conception of theology as well as of the kind of guidance offered by confessional documents. The inclusion of multiple documents from different places and times encouraged a greater diversity of theological perspectives. It also (p.128) drew attention to the contextual nature of theology, inviting church members to think of theology not so much as a timeless statement of eternal truths to be memorized and repeated but rather as a tool for engaging with particular questions that arise in particular situations. At one point in the church’s history, these were genuine innovations. But the most natural way to describe what was taking place in 1967 is not in terms of the rejection of one belief and the endorsement of another. Rather, what happened was that the institutional structure was modified to reflect an underlying set of beliefs that were already in place.

Given the nature of moral communities like the PC(USA), we should expect a number of instances of belief revision to involve a shift from one way of conceptualizing some feature of who they are and who they are meant to be to another. In many cases the concepts will remain fixed but the conceptions will change. To accommodate such changes, it will be useful to introduce another feature to our account to represent conceptions. Let c1 and c2 stand for rival conceptions. We can then modify our account of collective belief revision as follows:

A group g revises its belief that p or p+ iff the members of g (or the operative members of g), who previously believed that p or p+, one of whose concepts is construed in the c1 way, agree to believe something new, such as (i) that not-p, or (ii) that q or q+, or (iii) that p or p+ and u(Rpq), or (iv) that p or p+, one of whose concepts is construed in the c2 way.

This account highlights a number of aspects of collective belief revision that Gilbert’s original account either lumped together or excluded entirely. However, for all we have said, there may be other instances of collective belief revision that our modified account fails to accommodate. The easiest way to leave room for this possibility is to replace the biconditional in the preceding variant with a simple conditional. This would yield:

A group g revises its belief that p or p+ if the members of g (or the operative members of g), who previously believed that p or p+, one of whose concepts is construed in the c1 way, agree to believe something new, such as (i) that not-p, or (ii) that q or q+, or (iii) that p or p+ and u(Rpq), or (iv) that p or p+, one of whose concepts is construed in the c2 way.


Contemporary discussions of collective actions, emotions, beliefs, and knowledge have benefited greatly from Margaret Gilbert’s work on these topics. Her discussion of group belief revision, likewise, has drawn attention to an important but under-theorized aspect of collective life. However, perhaps because it focused on an imagined scientific community, her discussion failed to draw attention to a (p.129) number of features of belief revision that play a significant role in the lives of many collectives. By contrast, the constitutional changes made by the Presbyterian Church during the middle part of the last century provide a nice illustration of revisions of group belief that are not readily covered by Gilbert’s account. For the most part the lack of fit is due to the fact that the revisions were not (or not simply) a matter of agreeing to reject one proposition and replace it with another. Instead, they involved changes in conception, understanding, or holistic knowledge.

As should already be clear from the preceding discussion, the need for a revised account is not limited to religious communities like the PC(USA). It will apply to a wide range of communities, but especially to those whose identities are built around normative commitments. Julia Annas has observed that ‘each generation alters its predecessor’s conception of some virtues, while others fall out of favour altogether’ (2011, 22).26 These generational shifts will affect community groups and charitable organizations like Rotary Clubs and Oxfam. It will be relevant to the beliefs and decisions of school or university trustees whose actions and membership are shaped around a founding trust deed. And it will have a bearing on a number of more loosely structured moral communities. To account for the kinds of revisions that take place in these moral communities will require a more diverse account of group belief revision than the one Gilbert provided. The account developed in section 3 goes a considerable way toward addressing this need. But it is only the beginning of what will need to be a much longer conversation. Sooner or later it, too, will need to be revised.27 (p.130)


(1) Similarly, see Tuomela (2007, 135). While there are differences between Tuomela’s and Gilbert’s views, they will not make a difference to the argument developed here. One might worry that such an account is committed to voluntarism about belief. However, the individual beliefs on which the existence of the collective belief (in part) depends need not presuppose voluntarism, even if the collective beliefs to which they give rise are voluntarist. It is also worth noting that there is an important difference between the conditions for the creation of a belief and the conditions for the maintenance of that belief. There is more room for an exercise of will in the maintenance of a belief even in individuals than there is (on most accounts) in the acquisition of a belief.

(2) See e.g. Tuomela (2007), 136.

(3) Deborah Tollefsen is a nice example of someone who is attempting to do precisely this. She has criticized Gilbert’s and Tuomela’s accounts of group beliefs on the ground that ‘our attributions of intentional states to organizations are often, if not always, made in ignorance of the intentional states of the members (even the operative members)’ (2002, 396–7). One advantage of the communities and beliefs on which I shall focus is that we are not ignorant of the relevant intentional states of the group members. They have articulated their individual beliefs quite clearly at a number of stages in the process of formulating the group’s belief. So the argument I develop will not depend upon the success of Tollefsen’s challenge.

(4) Jason Baehr (2011) mounts a compelling case against the attempt to define knowledge in virtue-epistemic terms.

(5) Occasionally the recommendation comes from a committee set up by the General Assembly and it skips the initial phase of being presented to a presbytery. But thereafter the process is the same.

(6) Some decisions require a two-thirds majority, others only a simple majority, i.e. one more than 50%.

(7) Although it was the most dramatic, it was also one of the most widely supported changes, with nearly 90% of presbyteries endorsing the Confession of 1967 (Loetscher 1983, 162).

(8) Willis (1988), 119. For other changes made to the ‘subscription formula’ in 1967, see Loetscher (1983), 164–5.

(9) For example, the Pope is referred to as the Antichrist (Leith 1982, 222).

(10) An agent is warranted in believing a proposition p when an agent’s getting to the truth about p isn’t merely the result of certain kinds of luck. This description is not meant to take a side in current debates about what constitutes knowledge. The aim here is simply to get on the table a range of things that might go by the name of knowledge.

(11) Zagzebski (2012), 175. Stephen Grimm (2006, 517–18) has objected that what Zagzebski calls understanding is not as different from the kind of knowledge with which contemporary epistemologists have been concerned as she thinks. However, his argument is built around confusing what Miranda Fricker (1998, 164–6) calls ‘the philosopher’s question’ with what she calls ‘the historian’s question’. Grimm starts with the claim that ‘our understanding of natural phenomena’ is ‘arguably the paradigm case of understanding’. And the illustration that follows shows that what he has in mind is identifying the causal explanation of a particular natural occurrence (the historian’s question). However, Zagzebski is concerned with a much more diverse range of paradigmatic cases: music, art, literature, mathematical proofs, conceptual relations, metaphysical relations, and philosophical and scientific theories. And these other cases do not share important features with the understanding of natural phenomena on which Grimm’s argument depends. Consequently, Zagzebski’s account of understanding is not (or not obviously) subject to Grimm’s objection.

(12) He argues that both the moral realist and the expressivist can accept such a claim (Gibbard 2002, 224–6).

(13) Compare what we have described as a ‘tradition’ with what Tuomela refers to as a group’s ‘ethos: groupish ‘thinking and acting involves the existence of an ethos, namely, some constitutive goals, values, standards, and norms and we-mode thinking and acting relative to them’ (2007, 146). In most cases a group’s ethos will consist in a subset of the goals, values, standards, and norms that (partly) make up the group’s tradition.

(14) Tuomela likewise focuses his discussion on propositional knowledge-that and on beliefs that take propositions as their objects (see 2007, 135–6).

(15) Some readers might prefer to account for the phenomenon Kvanvig highlights by appealing to a dispositional account of belief, rather than positing beliefs with holistic content. A dispositional account is quite attractive for those who wish to draw attention to collective beliefs, since collective dispositions might seem less metaphysically spooky than collective mental states. Nevertheless, dispositional accounts of belief still face a number of unanswered questions. So at present it is not clear than a dispositional account is preferable to the holistic one offered above.

(16) Oliver O’Donovan claims that such a commitment to being in agreement is definitive of the church (2008, 30–4).

(17) Putnam (1978), 85–7. It was precisely the need to make room for such cases and the knowledge they might bring that prompted Christopher Peacocke (2003) to revise his account of concepts by introducing the idea of ‘implicit conceptions’.

(18) As Charles Taylor points out, these imaginative engagements may involve quite a bit more than simply weighing future goods that one expects will accompany each of these career paths. The options may also involve competing conceptions of who one is, what the decision amounts to, and what really matters (1985, 26–7).

(19) Oliver O’Donovan claims: ‘We focus our attention on the good presented to us by approaching it through narrative and projecting it through hope’ (2008, 99). He argues that this is a distinctive feature of being a temporally extended agent that experiences the present ‘as set “between” past and future’. Nevertheless, he suggests, Christian theology provides this common activity with a distinctive flavour and rationale for members of the Christian community.

(20) This is the first in a sequence of proposed definitions of group belief revision. Two features of the definitions in this sequence are worth noting at the outset. First, the sequence illustrates what it is describing, namely, a process of belief revision. Second, it departs from the mainstream of logicians working on belief revision who take their bearings from Alchourrón, Gärdenfors, and Makinson (1985) and Gärdenfors (2008). A trivial departure is that I am using ‘belief revision’ to cover both the kind of change generated by attempting to resolve a contradiction in one’s belief set (which they call ‘belief revision) and the kind of change involved in adding a new belief to an existing set of beliefs (which they call ‘belief expansion’). The more important departure is that these logicians have been attempting to model belief revision using propositional logic. However, it is unclear that a move from p+ to q+ can be modelled perspicuously using propositional logic. The same is true for the changes in conception discussed below. Modelling these kinds of changes will require the use of at least predicate logic. It is hoped that these definitions will identify the kinds of factors logicians will need to accommodate when developing more nuanced models of belief revision.

(21) In addition to being endorsed by Presbyterians and (briefly) Anglicans, the Westminster Confession ‘was adopted with modifications by Congregationalists in England and New England, and it was the basis of the Baptist creeds, the London Confession, 1677, 1688, and the Philadelphia Confession of Faith, 1742’ (Leith 1982, 192–3).

(22) Rawls (1999), 427. For further discussion of the nature of essentially contested concepts and the challenges that might be faced by the claim that justice, for example, is such a concept, see Christine Swanton (1985).

(23) One of the more endearing literary features of the gospel narratives is the way in which they convey the disciples’ deep and persisting confusion regarding who Jesus was and what he was doing. Far from portraying the first generation of Jesus-followers as having all the answers, the gospels show them still not getting it, even after three years of intense tutoring.

(24) In a similar vein, John Cottingham observes, ‘it is the task of religious discourse to strain at the limits of the sayable’ (2003, 8).

(25) Rowan Williams, the recently retired Archbishop of Canterbury, explores the importance of this tension in a number of his publications. For example, in his work on Arius he argues: ‘The doctrinal debate of the fourth century is thus in considerable measure about how the Church is to become intellectually self-aware and to move from a “theology of repetition” to something more exploratory and constructive’ (1987, 235); and in ‘What Is Catholic Orthodoxy?’ he argues that one of the distinguishing features of orthodoxy is that, in the very act of seeking to be faithful to its source it also ‘always subverts its own finality—as a system of words or pictures’ (1983, 19). For a nice overview of Williams’s work on this theme, see Benjamin Myers (2009).

(26) Similarly, MacIntyre suggests: ‘[A]ll reasoning takes place within the context of some traditional mode of thought, transcending through criticism and invention the limitations of what had hitherto been reasoned in that tradition…A living tradition…is an historically extended, socially embodied argument, and an argument precisely in part about the goods which constitute that tradition’ (1983, 222).

(27) This chapter began in a conversation with Dan Speak and I am grateful for the many hours he spent discussing the topic and commenting on multiple drafts. The chapter has also benefited from feedback provided by Garrett Cullity, Antony Eagle, Patrick Girard, and Rod Girle.