Introduction to Special Issue of Social Epistemology on “Collective Knowledge and Collective Knowers”

We freely attribute doxastic and epistemic features to groups. Perusing some recent headlines, one finds such statements as “Outsourcing giant Infosis believes it’s the new model,” “Pakistan believes consensus on IPI gas tariff likely,” “Scientists believe photos show evidence of flowing water on Mars,” “Privacy board knows the number of Americans targeted by NSA,” “Michelin knows how to keep its cool,” “France knows Nascar,” and “NBC knows what it wants.”

Of course, one might worry that such talk is too much like the following headlines: “Ethernet knows where it is going” and “Nature knows the parents children need.” These attributions are clearly using “knows” in a metaphorical sense and are, thus, not properly the concern of epistemology. Questions such as “Did the ethernet come to such knowledge via a reliable process?,” “Does nature have propositional or non‐propositional knowledge about how to assign parents to children?,” or “Is the ethernet certain or does it just assign a high probability to its reaching its destination?” are clearly out of place. Such questions about group knowledge and other group epistemic features, however, are not so obviously out of place.

Indeed, given that our epistemic lives are increasingly dependent on complex social systems of investigation and dissemination of knowledge, not only is such epistemic investigation of social groups possible, it is essential. The premise of this special issue of Social Epistemology is that we can use the tools of epistemology to understand the nature of group knowledge and other group epistemic features. In taking social groups seriously as epistemic agents, we are advocating a novel (although not unprecedented) approach. The standing assumptions of analytic epistemology have been that knowledge is justified true belief (plus Gettier conditions) and that the epistemic subject—the one who knows—is an individual. To the extent that philosophers have acknowledged social aspects of knowledge, the focus has largely been on the social factors that influence or determine what and how the individual knows. ¹ Those who have discussed cultures or societies “knowing” typically mean something like commonly held or culturally endorsed beliefs, leaving out the traditional justification and truth criteria of knowledge. ² Others have characterized the knowledge of the society in terms of its collected discoveries as recorded in books and other cultural objects, leaving out the belief criterion. ³

Less frequently has the focus been on the question of whether social groups themselves may be knowers in something like the traditional sense of subjects with justified true beliefs. Groundwork has now been laid by those working on collective intentionality, ⁴ collective belief, ⁵ group rationality, ⁶ and distributed cognition. ⁷ This work provides us with valuable tools we can use to address the question of how far we can maintain the traditional conception of knowledge while admitting corporations, collectives, and other social groups as knowers. The purpose of this special issue is to evaluate the progress made so far on this question and to push forward the philosophical exploration of collective knowledge.

The articles included here make valuable contributions to a number of topics within contemporary epistemology, such as epistemic virtue, epistemic responsibility, testimony, epistemic value, and philosophy of science. Here I can only highlight a few main threads that tie together the various contributions. The first thread focuses on the issues surrounding the concepts for group belief and acceptance. The second thread is concerned with questions of epistemic warrant and those features of social groups that make them epistemically trustworthy. The third thread can be found in the focus on groups of scientists as possibly forming group epistemic agents. Before I proceed to discuss these threads in more detail, it will be helpful to frame the issues by briefly canvassing the recent debate on group belief.

I Social Epistemology and the Rejectionist/Believer Debate

The articles here all take what Goldman (2006) calls the “classical” approach to social epistemology; that is, they are all concerned to some extent with questions of truth, epistemic rationality, and justification. According to Goldman’s original formulation, “Social epistemology would identify and evaluate social processes by which epistemic subjects interact with other agents who exert causal influence on their beliefs.” On this early characterization the only epistemic subjects are individuals, and the social aspect of the theory is captured by the impact that other people have on the beliefs of the individual. While Goldman now includes questions of what we might call “group epistemology” in his characterization of social epistemology, he leaves it an open question whether group knowledge is a legitimate area of investigation for the social epistemologist. Goldman asks, “Should social epistemology pursue its agenda by focusing, in whole or in part, on group knowledge?” He answers that, “Obviously, this depends on whether groups or collectivities are legitimate bearers of epistemic states like knowledge or justified belief.” Much of the previous work in group epistemology has focused on the question of whether groups can be legitimate bearers of belief.

As noted above, we often make statements about what groups know or believe. Some have argued that all such statements are merely shorthand for a much longer statement, which would list all or most of the members and their beliefs. This is what has been called the “summative” account of group properties. On the summative view of group features or properties, to say that a group has some feature y is simply to say all or most members have feature y. Indeed, some statements of the form “Group g believes y” are shorthand in this way. For example, it seems plausible to say that a statement like “Scientists believe photos show evidence of flowing water on Mars” simply means that some number of scientists individually believe this. The question is whether all statements about group epistemic states or features can be understood in this summative way. A number of authors, including many in this special issue, have rejected the claim that group belief statements can always be understood summatively.

Nevertheless, even if one rejects summativism, one may still argue that groups do not literally “believe” propositions. The question of whether groups believe or merely accept propositions has been the focus of the “rejectionist/believer” debate. ⁸ This debate hinges on the question of whether group views or positions are best understood as cases of belief, or as cases of acceptance. Table 1 summarizes a number of key properties of belief as opposed to acceptance that are appealed to by “rejectionists” in their arguments that group views belong in the acceptance, not the belief, category.

Table 1 Key Properties of Belief and Acceptance

Belief	Acceptance
1. Involuntary	1. Voluntary
2. Aim at truth	2. Aim at utility
3. Shaped by evidence	3. Shaped by pragmatic considerations
4. Subject to the ideal of integration	4. Not subject to the ideal of integration
5. Context‐independent	5. Context‐dependent
6. Come in degrees	6. Categorical

Rejectionists argue that group views have the characteristics on the right‐hand side, and are thus cases of acceptance, not belief. Believers, on the other hand, argue that either group views do have the characteristics listed on the left, or ordinary beliefs can have one or more of the features listed on the right. ⁹ As we will see, a number of articles in the present volume discuss the rejectionist/believer debate, but from the perspective of epistemology. This perspective allows them to move beyond the question of whether groups can “really” have beliefs to the more epistemologically interesting questions of group knowledge, justification, reliability, and epistemic value.

II Knowledge, Belief and Acceptance

Knowledge has been divided into “know how” or non‐propositional knowledge and “know that” or propositional knowledge (Ryle 1971). ¹⁰ If groups can be knowers, it may be reasonable to ask whether they can have either sorts of knowledge. While most of the articles in the present issue focus on the questions surrounding group propositional knowledge, Axel Seemann’s article proposes joint awareness as a kind of non‐propositional group knowledge. According to Seemann, “joint awareness” is a sort of know‐how that enables persons to take a joint perspective on the world and thereby act together. Our capacity to act together on group epistemic projects is foundational for our ability to have group propositional knowledge. Thus, Seemann’s article points out the importance for those who work on collective epistemology and collective intentionality to explore how social cognition grounds our capacities to engage in joint cognition.

The majority of articles that focus on propositional group knowledge, for example the articles by Don Fallis, Kent Staley, Brad Wray, and Raul Hakli, accept at least the possibility of non‐summative group knowledge. Angelo Corlett, however, defends a summative account of group knowledge based on a summative account of group belief. Corlett argues that acceptance by a group, the most common condition for non‐summative group belief, is neither necessary nor sufficient for a group to have a belief. According to Corlett’s “sophisticated summativism,” group belief is indexed to the number of members who believe and to the strength of their beliefs. He argues that his view is superior to acceptance‐based views, because it allows us to see the internal complexities of getting to a shared understanding and agreement on a particular proposition.

Brad Wray and Raul Hakli both take a non‐summativist approach and argue along rejectionist lines that groups accept, rather than believe, propositions. However, both authors move us beyond this debate by focusing more directly on epistemic concerns. They each argue that knowledge may be understood as justified true acceptance, rather than justified true belief. Hakli’s argument distinguishes two ways in which one can know based on whether one’s views are formed involuntarily (and are thus beliefs) or voluntarily (and are thus acceptances). Adult human beings may know in either way, he argues, but groups can only know in the second acceptance‐based way. Thus, both Wray and Hakli argue, one may hold that groups can know even while one denies that they have beliefs.

It is often argued that groups cannot have properly epistemically grounded beliefs (or acceptances), because pragmatic considerations feature too prominently in the formation of non‐summative group views. Given that group acceptances are voluntary, they may be guided by pragmatic goals, rather than by epistemic reasons. ¹¹ Anthonie Meijers, for example, argues that, “If only epistemic reasons for believing are taken into account, it is impossible that there is a difference in content between what I as an individual believe and what I believe as a member of the group…” (2003, 379). Don Fallis and Kent Staley both argue that pragmatic concerns are not necessarily anti‐epistemic, and, thus, should not disqualify groups as epistemic agents or group views as candidates for knowledge. Indeed, both Fallis and Staley claim that pragmatic concerns help to structure our epistemic goals in unavoidable and positive ways.

Nevertheless, in order for group views to be of concern to the epistemologist, one might argue that the primary goal of the group’s view‐forming activities must be epistemic. Thus, the question of whether groups can be knowers will depend on whether groups can have epistemic goals. In his article, Fallis focuses on this question; he asks what group epistemic goals might be and how group epistemic goals relate to the epistemic goals of their members. Fallis argues that groups may have epistemic goals and that these goals are not always simply the sum of the epistemic goals of the members.

III Group Justification and Epistemic Virtue

Of course, in order to have knowledge it is not sufficient that groups have epistemic goals, they must also pursue these goals in appropriate ways. Staley, for example, argues that we need to ask “whether the group’s statements are constrained by the group’s own judgments of the relevant evidence, and whether the group functions so as to produce unbiased collective assessments of the evidence.” A number of the articles in the present issue focus on these questions.

In his discussion of collective justification, for example, Corlett discusses the conditions necessary for collective epistemic responsibility. Interestingly, Corlett takes voluntarism as central to epistemic responsibility, in contrast to many authors who have argued that voluntarism is anti‐epistemic. ¹² Indeed, for Corlett all knowledge requires acceptance, rather than involuntary belief, because only on conditions of epistemic responsibility, which requires voluntary acceptance, are beliefs justified. In Corlett’s view, groups can be warranted in their beliefs only if they possess the epistemic virtues of open‐mindedness and self‐reflection. According to Corlett, “It is only by possessing these sorts of epistemic virtues that a group can deem itself sufficiently trustworthy to be considered one who believes according to reliable belief‐forming processes.”

Reza Lahroodi makes the consideration of the nature of such group epistemic virtues the focus of his piece. Lahroodi defends a non‐summative account of group epistemic virtue. He argues that for a group to have some epistemic virtue, such as open‐mindedness, it is neither necessary nor sufficient for the individual members to have this virtue. According to Lahroodi, group epistemic virtue is a genuine group trait. Lahroodi disagrees with Corlett and with Margaret Gilbert, however, when he rejects second‐order acceptance as necessary for group traits such as epistemic virtues.

We are concerned with the reliability and epistemic virtue of groups at least partly because groups will act on their beliefs. One type of action groups might take is to issue statements or other forms of testimony. Deborah Tollefsen asks us consider how our standards for trusting the testimony of groups relates to our standards for trusting the testimony of individuals. She points out that when a group testifies this is not equivalent to the testimony of the individual members. She furthermore argues that, given the differences between group and individual testimony, one popular account of how we can be justified in trusting testimony—non‐reductionism—does not apply to accepting group testimony. She argues that we can, however, have good reductionist grounds for determining which group testimony to accept. We can look, for example, to a group’s trustworthiness as evidenced by its track record. Tollefsen points out that the track record of the group will, again, not be equivalent to the track records of the individual members. Thus, for a group to be “trustworthy” is a group property not equivalent to the trustworthiness of the individual members.

IV Group Epistemology and Science

As we have seen, whether group views amount to knowledge largely depends on the goals of the group and on their epistemic practices. Thus, we would expect that the most likely candidates for group knowledge will be those groups whose goals are primarily epistemic and whose practices are shaped by a concern with truth, evidence, and rationality. Groups of scientists working together on shared projects are paradigmatic examples of what we might call “epistemic groups,” and a number of the articles included here focus on them.

It has become increasingly clear that traditional individualistic epistemology is inadequate to account for the complexity of scientific knowledge production in an age of large scientific research groups. Articles by Wray and by Staley explore how the concept of collective knowledge can be a useful way to understand how science works. Ronald Giere uses the paradigm of distributed cognition to understand the activities of scientists, but stops short of attributing epistemic agency to distributed cognitive systems. Giere argues that “A distributed cognitive system is a system that produces cognitive outputs … There is, however, no need to endow the system as a whole with other attributes of human cognitive agents.” One may make a similar argument with regard to socially distributed cognitive systems such as epistemic groups. Giere argues that talking about collective epistemic agents is unhelpful, and we should rather seek to understand how individual agents working together produce cognitive outputs.

Wray disagrees, at least with regards to the utility of endowing “attributes of human cognitive agents” to groups of such agents. He argues that such attributions may help us understand the activities of groups of scientists. According to Wray, “Collective knowing may be the only way to get at some of the knowledge we now take for granted … [which] could never have been discovered without the efforts of plural subjects, agents formed by people working in groups with intentions that are irreducibly the intentions of the group.” Wray goes on to explore the question of which scientific groups we can sensibly call “knowers.” He argues that only groups with what Durkheim called “organic solidarity”—that is, those within which there is some division of cognitive labor—can be aptly described as having collective knowledge. Thus, he argues, research groups, but not scientific subfields, can be collective epistemic agents that “know.”

Conclusion

The framing question for this special issue was whether groups can be “knowers.” While many of the papers in this collection address this question, all of the authors move beyond this question to consider the more complex nature of the epistemic capacities of groups and how those capacities relate to those of the members who form them. The articles collected here point the way to further exploration of a number of questions within group epistemology: Does group justification differ from individual justification, and in what ways? What sorts of group structures promote the production of knowledge? What is the role of systems of communication and information storage in large epistemic groups? How can we evaluate group testimony? What are the explanatory benefits of attributing knowledge to groups as well as individuals? How do our individual capacities for social cognition enable our participation in groups with epistemic capacities?

My hope is that the articles in this special issue inspire epistemologists to explore these and other questions that arise within group epistemology.

Notes

[1] See Goldman (1999).

[2] See Berger and Luckman (1966).

[3] See Popper (1972).

[4] See, for example, Searle (1995), Tuomela (1995), Gilbert (1989), and Bratman (1999).

[5] See, for example, Gilbert (1987), Tuomela (1995), Schmitt (1994), Pettit (2003), and Tollefsen (2000a, 2000b).

[6] See, for example, Sugden (1993) and Graham (2002)

[7] See Giere (2002).

[8] Those engaged in this debate include Heimir Geirsson (2004), Margaret Gilbert (1987/1996, 1989, 1994, 1996, 2000, 2002, 2004), Raul Hakli (2006), Christopher McMahon (2003), Anthonie Meijers (2002, 2003), Gerhard Preyer (2003), Frederick Schmitt (1994), Deborah Tollefsen (2002a, 2002b, 2003), Raimo Tuomela (1992, 1995, 2000, 2004), and K. Brad Wray (2001, 2003).

[9] Raul Hakli’s article most directly engages with the previous literature on group belief; his article is a good introduction to the issues involved.

[10] This distinction has recently been questioned in Stanley and Williamson (2001). For a critical discussion of the Stanley and Williamson argument with regard to how we can learn the skills necessary for collaborative research, see Thagard (2006) and Fallis (2006).

[11] For an argument that groups’ views, although based on acts of acceptance, may not always be voluntary, see Mathiesen (2006).

[12] See Hakli (2006) for a discussion.