Non-summative Belief and Joint Commitment
Following Gilbert’s approach, collective knowledge is a form of group belief, characterized as a “non-summative" belief held by a “plural subject" that is constituted through the “joint commitment" of individuals (Gilbert, 1989,1994, 2000, 2013). Let me elaborate on the conceptual elements of group belief one after another and provide a concise formulation of the condition to which Gilbert’s group belief applies. My goal, here, is to probe the extent to which Gilbert’s group beliefapplies and accounts for the two cases of collaborative knowledge creation in the research groups that I observed.
Gilbert’s group belief is based on the distinction between summative and non-summative collective beliefs. A summative account holds that groups of individuals can have the belief that p insofar as all or most of their individual members have the belief that p. In other words, summative collective beliefs are reducible to individual beliefs. Non- summative beliefs, in contrast, are irreducibly collective beliefs. For these non-summative collective beliefs, Gilbert argues, it is neither a necessary nor a sufficient condition that all or most (or, in fact, any) group members individually hold the collective belief.
The observation that underlies Gilbert’s non-summative collective belief is that there are cases in which individuals adopt beliefs only as a group. She argues that it can be plausible (and, for that matter, warranted) for two different groups with identical individual members to have conflicting group beliefs. She presents the example of a library committee and a food committee at a residential college which are identical in terms of membership. She argues that it is plausible for the two committees to differ in their group beliefs. All group members might individually have the opinion that college food is too high in carbohydrate; yet, while the food committee comes to adopt the group belief that college members are consuming too much starch, the library committee as a group need not hold this belief (Gilbert, 1989, p. 273; for an elaboration of this example see Schmitt, 1994, p. 261). These examples suggest that what is decisive for group belief are institutional constraints under which a group is forced to act collectively.
According to Gilbert, for a group to hold the non-summative, irreducibly collective belief thatp, individual members must jointly express their individual willingness to let a proposition p stand as the belief of their collective—a condition that she describes as “joint commitment" of all group members (Gilbert, 1994, p. 245; Gilbert, 2000, p. 40). A joint commitment, then, constitutes a “plural subject" which holds the proposition in question as a collective belief. Two or more individuals form a plural subject holding a collective belief if, and only if, they are jointly committed to holding this belief “as a body" (Gilbert, 1994, p. 244).
What a non-summative group belief thus demands from group members is not to adopt the group belief as individual belief, but to decide jointly with all other group members to let a certain proposition stand as the belief of their group. As members of a plural subject, individuals’ actions are constrained. Their joint commitment imposes mutual obligations upon them. Once committed to others, individuals are entitled to expect other jointly committed individuals not to undermine their collective belief. Hence, a joint commitment prevents individual members from rescinding collective beliefs, since unilateral resignation will be negatively sanctioned by the collective (see, e.g., Gilbert, 1994, p. 238 or Gilbert, 2000, p. 44).
Drawing on the paradigm of knowledge as a form of belief, Gilbert’s group belief can be reformulated as collective (scientific) knowledge, a line of thought which, in fact, a number of social epistemologists have pursued (Andersen, 2010; Bouvier, 2004, 2010; Rolin, 2010; Tollefsen, 2006). Let me label this notion of collective knowledge the joint commitment account (JC) and provide a concise formulation of the condition under which this account applies:
According to the JC account, a necessary condition for collective knowledge to obtain is that a group of individuals is jointly committed to allowing a proposition to stand as the knowledge of their group.
The question that I will discuss in the remainder of this section is how far, and with what analytic benefit, the JC account can be applied to the observed cases of collaborative knowledge creation. This discussion will be rather critical, and I will argue that to conceive of collaborative knowledge creation on Gilbert’s terms creates a misleading image of scientific practice.
Gilbert’s notion of group belief and its reformulation as collective knowledge have done much to challenge epistemic individualism, reviving social epistemology’s debate about the collective character of science. The conceptual appeal of her group belief is the “slack" that this notion cuts between individual and collective believing. Group beliefs, she argues, need not be reducible to individual beliefs. This notion of group beliefs seems to explain the socio-epistemic phenomena of scientific knowledge creation, such as an alleged conservatism in the adherence to scientific paradigms (Gilbert, 2000; Rolin, 2008; Wray, 2001). However, as Fagan (2011) critically points out, Gilbert’s notion of group belief does not explain these phenomena particularly well. According to Fagan, accounts based on Gilbert’s notion posit empirically unconfirmed explananda and fail to “identify positive counterparts (or extensions) [...] which inferentially connect to features of collective belief stemming from joint commitment" (Fagan, 2011, p. 263). Joint commitment, Fagan holds, is an idle wheel (and what it seeks to explain, Fagan (2012) suggests, is better done through a web of interpersonal commitment).
My discussion of the JC account and its applicability bears parallels to Fagan’s critique. Like Fagan, I have considerable doubts as to whether the JC account resonates with, let alone explains, empirical observations of scientific practice. To explicate these doubts in the following, I first revisit some of the—non-scientific—examples Gilbert proposes to illustrate regarding the plausibility of joint commitment. These examples, I argue, seem to suggest additional conditions for JC to obtain. Thereafter I spell out in more detail what the JC account of collective knowledge implies for collaborative scientific practice and discuss how far these implications relate to the empirical data I gathered.
It is important to note that when Gilbert develops her notion of group belief, she illustrates her conceptual discussion with a series of small, ad hoc examples—most of which refer to everyday situations in which a small number of people decide to endorse a statement or attitude jointly as a group. One example, the food and library committee, has been mentioned above. Other recurrent examples include a poetry discussion group which arrives at a group interpretation (e.g., in Gilbert, 2013, p. 169) and the spontaneous act of two people walking together (e.g., in Gilbert, 2013, p. 24). The range of Gilbert’s examples suggests wide applicability. But in many of the examples she uses, there is an immediate pressure to decide and act on an issue as a group—while it is, at the same time, legitimate for group members to entertain private thoughts that do not coincide with the jointly accepted group view and that individual group members feel no pressure to act upon. It seems therefore as if her examples tacitly imply additional conditions necessary for JC to obtain. And it begs the question as to whether these conditions are met in collaborative scientific practice.
Gilbert and other philosophers, however, have in fact applied the notion of joint commitment to science. In the epistemological analysis of science, Gilbert’s group belief, rephrased as collective knowledge, has been ascribed to different kinds of collectives in science, to specialist communities as well as to research groups. When Gilbert herself brings her account of collective belief to bear on the explanation of scientific change, she conceptualizes disciplinary peer communities as collectives bound by joint commitment to a scientific paradigm: “[...] I shall assume that, by and large, scientific communities do have scientific beliefs of their own” (Gilbert 2000; for a modified account of community-borne collective knowledge see Rolin 2008). According to Gilbert, the seemingly sudden character of scientific change and its absence over long periods of “conservative,” normal science cannot be explained in terms of individual scientists. Rather, she suggests that members of a specialist community should be understood as being jointly committed to a scientific paradigm. This, she argues, explains the conservativeness of science, as individual members would face negative sanctions if they unilaterally dissented from the reigning paradigm. Moreover, it explains the sudden presence of rapid scientific change because if joint commitment is violated, then it breaks in its entirety. Peer communities, however, are not my focal interest.
In contrast to Gilbert, Wray (2007) argues that large peer communities cannot hold collective knowledge because they are unable to strike a joint commitment. According to Wray, only research groups, that is, small groups of interdependent collaborators, can do so. Yet with this argument, Wray posits an additional condition for JC to obtain—the condition of “organic solidarity,” a feature of group members’ relations with one another that reflects their state of mutual dependence, a state caused by division of labor. While Wray does not go very far in clarifying how organic solidarity should be linked to the collective possession of knowledge, I will seek to explicate a link between mutual interdependence and collective knowledge in the next section.
For my own discussion of whether the JC account of collective knowledge applies to research groups and their collaborative scientific practice, and particularly to the collaborative authoring of research publications, I confine myself to my comparative case study. To support the JC account of collective knowledge, I need to mobilize
Observations of the first type (i) should make it plausible that the JC account explains, in fact, a phenomenon of empirically observable practice. Observations of the second type (ii) should make it plausible that the JC account is compliant with actual scientific practice, that it aligns with the normative notions that are ingrained in scientific practice. Thus, while (i) reflects a concern for the explanatory value of JC, (ii) reflects a concern for its empirical applicability (see also Fagan, 2011).
Regarding (i), the data I have been able to gather do not testify to empirical phenomena that would be particularly well explained through JC. According to the JC account, a phenomenon that calls for joint commitment as an explanation would be a collectively held belief that need not be identical with those beliefs held individually by single members of the collective in question. Such a phenomenon, if it is a feature of scientific practice, should be especially observable in the case of interdisciplinary research, that is, research where individual scientists with different research backgrounds, perspectives and interests collaborate. However, in my fieldwork and my interviewing, I have not found evidence of such a phenomenon.
Instead, what the scientists I observed conveyed to me was an ethos of individual judgment and conviction—if you are not personally convinced of something, you should not commit yourself to it. Collaborative scientific practices should mobilize, not sideline, individual judgment so as to forestall the premature dissemination of badly corroborated knowledge claims. As Laura, junior physicist in the planetary science group, observes about collaborative authoring: “[As a contributing author] I might not be able to refer exactly to the details, but I’ve checked it through and there is nothing I am majorly unhappy about"—implying that if she or other authors were unhappy about a piece of research, she should bring it up with the first author who should “try to address their concerns and make sure that the article is written in such a way that people agree" (Laura, interview, group1).
If your scientific collaborators are not convinced of a research result, you should try to convince them, explaining your reasoning and helping them understand. As Laurits, senior geologist in the planetary science group puts it: “you have to convince other people about the stuff you know best" (Laurits, interview, groupl). In their interviews, both Laurits and Adam, one of the group’s physicists, describe an incident of disagreement which led an outside collaborator to opt out of a paper. “[I]t’s sad when that happens," Adam says, “but it’s kind of rare. It doesn’t happen a lot—not to us, anyway. And it doesn’t happen in-house." Because “[...] if we disagree with the conclusions of [collaborative experiments], then we would—we do some more measurements until we, we’re sure" (Adam, interview, groupl). I have described this practice of collaboration as “dialoging," a practice that values “explanatory responsiveness," in Chap. 8.
Notwithstanding these dialoging practices, supporters of the JC account may still apply the notion of joint commitment to the collaborative creation of scientific knowledge if they declare individual understanding and conviction to be irrelevant, or secondary, to scientific practice. Supporters of the JC account could argue that while group members may have individual understanding and conviction, they are still bound by a non-summative group belief. But at least for the cases of collaborative scientific practice I observed, such an interpretation appears forced and the notion of joint commitment becomes, as Fagan puts it, “otiose" (Fagan, 2011, p. 269).
Regarding (ii), the data I have been able to gather do not indicate that scientific practice complies with implications of JC. According to the JC account, an indication of joint commitment would be the willingness of individual group members to endorse a view as their group view, a willingness that would have to be expressed by all involved individuals jointly. Another indication of joint commitment would be the negative sanctioning of group members who break a joint commitment. My data, however, do not lend support to any such things. In my fieldwork, I found that the professional freedom of individual group members to pursue their own interests and ideas plays a fundamental role in the functioning of the two research groups observed (see Chaps. 4 and 5). Supporters of the JC account might argue that jointly committed group members have deeply internalized the constraints that their joint commitment places upon their professional performance. But that is not what the joint commitment account, which emphasizes precisely the difference of collective belief and individual believing, is saying. Supporters of the JC account may furthermore point out that especially younger researchers are usually confined in their professional freedom and embedded in hierarchical relations. Yet Gilbert’s notion of joint commitment does not capture hierarchical decision processes either. In hierarchical research groups, where senior scientists lead junior scientists, it stands to reason that the beliefs of senior scientists influence the individually held views that the junior scientists are endorsing (or are supposed to endorse). This, again, is not what a JC account is saying.
So despite the fruitful debate that Gilbert’s group belief and the JC account of collective knowledge have spurred, their usefulness for an epistemological analysis of collaborative knowledge creation in scientific practice is not apparent. My empirical data provide no evidence suggesting that the JC account resonates with collaborative scientific practice, and that its application would generate explanatory benefits. Worse, in Gilbert’s formulation, the JC account undercuts the practices of dialoging and explanatory responsiveness, blackening the processes that underlie, as my case study shows, the creation of scientific knowledge. As it fails to account for these processes, Gilbert’s collective belief, rephrased as collective scientific knowledge, risks promoting forms of collaborative scientific knowledge creation that rely on intransparent and truncated collective deliberation and judgment (as, e.g., in Beatty, 2006; see also Beatty & Moore, 2010).
Given that the JC account of collective knowledge does not apply well to the practices of research collaboration I have studied, I will turn to the epistemic dependence (ED) account of collective knowledge in the following section. As I will show, there is, in fact, a robust sense in which the collaborative practices I have studied yield collective knowledge.