Author Information: Line Edslev Andersen, Vrije Universiteit Brussel, email@example.com
Andersen, Line Edslev. “Community Beliefs and Scientific Change: Response to Gilbert.” Social Epistemology Review and Reply Collective 6, no. 10 (2017): 37-46.
The PDF of the article gives specific page numbers. Shortlink: http://wp.me/p1Bfg0-3LJ
Please refer to:
Image credit: NASA Goddard Space Flight Center, via flickr
Margaret Gilbert (2017) has provided an engaging response to my paper on her account of joint commitment and scientific change (Andersen 2017). Based on Donald MacKenzie’s (1999) sociohistory of a famous mathematical proof, my paper offered an argument against her account of why a scientist’s outsider status can be effective in enabling scientific change (Gilbert 2000). On her account, scientists have collective beliefs in the sense of joint commitments to particular beliefs. The term ‘collective belief’ is used in this sense in the present paper. When a group of scientists are jointly committed to some belief, they are obligated not to call it into question. According to Gilbert, this makes joint commitments work as a brake on scientific change and gives outsiders an important role in science. Since outsiders to a given scientific community are party to no or relatively few joint commitments of that community, they are less constrained by them. For this reason, outsiders play a central role in bringing about scientific change.
I argued that Gilbert’s account is inherently difficult to test because it requires data that are hard to interpret. At the same time, I pointed out that we have available a simpler explanation of why a scientist’s outsider status can be effective in enabling scientific change: During their education and training, scientists learn to see things in certain ways. If solving some problem requires one to look at things in a different way, scientists with a different educational background will have an advantage. I have become aware that Melinda Fagan (2011, 255-256) also compares these two explanations.
Gilbert’s response to my paper has two main parts. In the first part (Gilbert 2017, 46-48), she argues for the role of collective beliefs in science by considering the role of collective beliefs in everyday life and in the context of education. I discuss her argument in the next section. In the other part (48-49), she suggests that collective beliefs of scientific communities can help explain why some personal beliefs become deeply entrenched in the minds of scientists. I agree with this. Deborah Tollefsen has made the related suggestion that one could respond to my argument by claiming that the degree to which some personal beliefs are entrenched in the minds of scientists cannot be explained without collective beliefs of scientific communities. This is an interesting suggestion. Here, however, I take a simpler approach to the question of whether scientific communities have collective beliefs.
As mentioned, the remainder of this paper begins with a discussion of Gilbert’s argument (section 1). I then address the question of whether (section 2) and to what extent (section 3) scientific communities in particular (as opposed to, for example, research teams) have collective beliefs. On this basis, I assess the potential of collective beliefs to work as a brake on scientific change (section 4).
Collective Beliefs in Science
On Gilbert’s account, a collective belief is a joint commitment to believe some proposition p (e.g., Gilbert 1987). A joint commitment to believe p is the commitment of a group as one body to believe p; i.e., it is the commitment of a group to emulate, by virtue of the actions of all, a single believer of p. Group members are thus “to speak and act as if they are of ‘one mind’ on the subject” (Gilbert 2017, 46). According to Gilbert, this implies that the joint commitment to believe p is persistent in the sense that it can only be rescinded with the concurrence of all the parties (Gilbert 2014, 118). Each of them has an obligation towards the others to act in accordance with the joint commitment to believe p and not, for example, express contrary beliefs. If someone violates the joint commitment, the others gain the standing to rebuke her and may even ostracize her (Gilbert 2000, 40). By virtue of these features, collective beliefs can act as powerful behavioral constraints and work as a brake on scientific change on Gilbert’s account. Speaking about scientists’ collective beliefs, she thus writes that they can have as far-reaching consequences as “inhibiting one from pursuing spontaneous doubt about the group view, inclining one to ignore evidence that suggests the falsity of that view, and so on” (Gilbert 2000, 44-45).
While collective beliefs can be very persistent, they are quite easily formed. A joint commitment to believe p can be formed without all or most or even any of the group members personally believing p. What matters is that they have expressed their personal willingness to let p stand as a belief of the group—if only tacitly—and this is common knowledge between them. On Gilbert’s (2017, 46-48) account, this happens all the time in everyday life and in the context of education. One of the examples she gives in her response to my paper is that of two people having an informal conversation. One of them says “What a lovely day!” and the other responds “Yes, indeed!” This establishes the belief that it is a lovely day as a collective belief the two have. Gilbert reasons that, if this is all it takes for a collective belief to form, they must play a role in science as well: “If collective beliefs are prevalent in human life generally, and if, in particular, they are the predictable outcome of conversations and discussions on whatever topic, we can expect many collective beliefs to be established among scientists in the various specialties as they talk about their work in small and large groups” (Gilbert 2017, 47).
I agree with Gilbert that, if collective beliefs play this type of role in everyday life, they must play a role in science. I am also convinced by her claim that joint commitments are generally ubiquitous. However, when Gilbert describes how easily collective beliefs are formed in everyday life and in the context of education, she gives examples of smaller groups forming collective beliefs, such as the group of people attending a meeting of a large literary society or a student and a teacher having an interchange. By contrast, when she, I, and others discuss the potential of collective beliefs to work as a brake on scientific change, we are referring to collective beliefs of whole communities. This is relevant, since there seems to be a difference between how easily collective beliefs of smaller groups in science (such as research teams) and scientific communities are established. In fact, I argue below that it is at least rare for scientific communities to form collective beliefs. This is where I disagree with Gilbert.
As the previous work on the potential of collective beliefs to work as a brake on scientific change, the present paper focuses on collective beliefs of whole communities. I thus leave open the possibility that collective beliefs of smaller groups in science can work as a brake of scientific change. However, Hanne Andersen and I have examined the instability of joint commitments of smaller groups and argue that they can rather easily be dissolved (Andersen and Andersen 2017). This limits the potential of collective beliefs of smaller groups in science to work as brakes on scientific change.
The Existence of Community Beliefs
Having explained Gilbert’s account of collective belief, I will now examine the question of whether communities in science have collective beliefs. A consensus established at a consensus development conference is a good candidate for being a collective belief of a scientific community. Paul Thagard (1998a, b, 1999) attended the 1994 consensus conference on methods of diagnosing and treating ulcers as part of his work on the bacterial theory of ulcers. In a later paper, Thagard (2010, 280) addresses Gilbert’s account of collective belief, stating that collective beliefs of scientific communities strike him as “rather rare,” but that he believes consensus conferences establish such collective beliefs. Consensus conferences are themselves rare and in most disciplines non-existent, but in medical research, Thagard explains, “the need for a consensus is much more acute, since hypotheses such as the bacterial theory of ulcers have direct consequences for the treatment of patients” (1998b, 335).
The consensus conference Thagard attended was conducted by the U.S. National Institutes of Health. He describes the purpose of their consensus conferences to be “to produce consensus statements on important and controversial statements in medicine” that are useful to the public and health professionals (1998b, 335). A consensus statement is prepared by a panel of experts after deliberation. Most likely the members of the panel do not all personally agree with everything in the statement given the controversial nature of the subject, but the statement expresses the view that they have agreed to let stand as the view of the panel. In this paper, I assume that when members of a group agree to let a view stand as the view of the group, a joint commitment is involved, so the view of the panel involves a joint commitment. It is, in other words, a collective belief.
The question I am interested in here is whether a consensus statement sometimes expresses not only the collective belief of the consensus development panel, but the collective belief of a whole community of scientists. Let us consider the consensus conference Thagard attended. The consensus development panel was chosen to represent a community—an appropriately delineated medical research community—in the following sense. Its members were chosen by the planning committee whose chair is required to be an authority, “a knowledgeable and prestigious medical figure,” and to be neutral in the sense of ‘not identified with strong advocacy of the conference topic or with relevant research’ (Thagard 1998b, 336). The fourteen people on the consensus development panel were chosen for various kinds of expertise and for their neutrality in the stated sense. Finally, the statement of the panel was based on presentations at the public consensus conference by 22 researchers representing different points of view; contributions from conference attendees during open discussion periods; and closed deliberations within the panel. Sometimes, although apparently not in this case, a draft statement is published online for public comment (e.g., NN 2013, 1).
The first page of the statement tells us that it “provides a ‘snapshot in time’ of the state of knowledge on the conference topic,” implying that these are early times and work remains to be done (NN 1994). This proviso limits the potential of the collective belief expressed in the statement to work as a brake on scientific change, for it must limit the ability of the collective belief to incline scientists to ignore evidence that suggests the falsity of the belief. But the proviso does not lessen the potential of the collective belief for being the collective belief of a whole community. It seems to me plausible to say that the members of the community in question in 1994 expressed their willingness (most of them tacitly) to let a belief stand as the belief of the community at that point in time. This is due to the relative neutrality of the panel, the diversity of the speakers, and the fact that members of the community are given the opportunity to have their voice heard.
A similar point can be made about certain group views that are established in a similar way, but are about something else. I have in mind certain codes for responsible conduct of research. For example, the European Mathematical Society (EMS) introduced a Code of Practice in 2012 (NN 2013, 12). This code may be said to express the view of the EMS in a way similar to how the 1994 consensus statement may be said to express the view of a community of medical researchers. The Code of Practice was prepared by the Ethics Committee of the EMS and approved by the EMS council, which in total consists of about 100 member-elected “delegates from all of the national societies which are members of the EMS” and “delegates representing the individual members of the Society” (www.euro-math-soc.eu/governance). The code will apparently be considered for revision every three years in light of comments received by the chair of the ethics committee from members of the EMS.
While I have addressed the question of whether communities of scientists have collective beliefs, Wray (2007) addresses the broader question of whether they have beliefs in a non-summative sense. When a group believes p in a summative sense it just amounts to all or most of the group members personally believing p. Wray argues that scientific communities, as opposed to research teams, are not capable of having beliefs in a non-summative sense. He uses Emile Durkheim’s distinction between societies characterized by organic solidarity and societies characterized by mechanical solidarity (Wray 2007, 341-342). Wray writes that a group or community is characterized by organic solidarity when its members “depend upon the proper functioning of the other members” (Wray 2007, 342), as the parts of an organism depend on each other, and are organized so as to advance a goal. Groups that are not bound together by organic solidarity, are bound together by similar thoughts and attitudes; by mechanical solidarity. Wray argues that a group must be cohesive in the sense of being characterized by organic solidarity to be capable of having beliefs in a non-summative sense and that scientific specialty communities and the scientific community as a whole are not cohesive in this sense.
If consensus conferences produce group beliefs that can properly be described as beliefs of whole communities, they are strictly speaking inconsistent with Wray’s account. These beliefs would then be produced by community acts characterized by organic solidarity, but such acts seem to be exceptional and do not speak against the claim that communities in science are generally characterized by mechanical solidarity. On Gilbert’s account, the non-summative group beliefs Wray discusses imply joint commitments. Wray’s account is neutral on this question. His is an argument that communities in science do not form non-summative beliefs in general. It thus implies that they do not form joint commitments to beliefs. In the next section, I give an additional argument for this particular conclusion.
The Frequency of Community Beliefs
In their everyday practice, scientists often express a view of a research team they are part of, for example in publications, in conference presentations, or in conversation with other scientists. They less frequently express a view of one of the communities they belong to, regardless of how we conceive ‘community view’ here. It seems to be rather rare that the typical scientist is prompted to say, “We as a scientific community believe…” That the motivation to express group views is relatively low at the community level may suggest that the motivation to actively establish group views at the community level is relatively low as well. However, the following argument does not depend on it.
The argument focuses on collective scientific beliefs of communities, since these are the ones that work as brakes on scientific change on Gilbert’s account. In the next section, I return to the topic of community beliefs about responsible conduct of research. So let us consider a Gilbertian scientific community belief p that has just been formed. p would have to be somehow broadly relevant in the community; the community members must, after all, be aware of a joint commitment to believe p for there to be one such. Furthermore, in order to agree to let a belief stand as the belief of the group, the group members must have some motivation to do so. They must do so as a means to realizing a goal (see Wray 2001). Sometimes the members of a community will be motivated to let a belief stand as the belief of the community although they personally have very different beliefs on the matter. For example, in the above case of the consensus conference, there was a need to present a community belief to the public and health professionals. But this is rare.
If the proposition p is broadly relevant in the community and the community belief that p is “unforced” in the sense that it has not been formed quickly under external pressure from the public or others, experts will have discussed and tested whether p until there is broad agreement among them. If the experts broadly agree that p is well established by the evidence, the other community members are likely to believe p because the experts do. Hence, at the time of being established, the collective belief p will reflect what a large majority of the community members personally believe, except in rare cases similar to the consensus conference example considered above. This fits well with Gilbert’s (2000) account of collective beliefs and scientific change. The negative potential of collective beliefs, as she describes it, is not associated with their being established in spite of recalcitrant evidence, but by their being maintained in spite of recalcitrant evidence discovered later.
This raises the question of how members can be motivated to jointly commit to a belief they already broadly share. There is already a community belief that p (albeit in a summative sense) that can be presented as such to the public and others. But there may be a reason internal to the community for making the joint commitment. Kristina Rolin (2008) raises the general question of how the members of a community are motivated to jointly commit to beliefs. She argues that community members are motivated to jointly commit to background assumptions because individuals can then use these assumptions and remain epistemically responsible even when they do not have the expertise to defend them if they are challenged. They can do so because the joint commitments obligate the relevant experts in the community to defend the assumptions if they are appropriately challenged. As implied by the above, I am unconvinced by Rolin’s premise that all the members of a community would be prepared to jointly commit to the same mere assumption, especially given the obligations and constraints this implies on Gilbert’s account. It is unclear how they would determine which background assumptions to commit to.
That it is unclear if and how community members would be motivated to jointly commit to believe p is a serious challenge to the claim that scientific communities form unforced collective beliefs. I believe the challenge may well be insurmountable. But even if we assume that community members are motivated to (and do) form unforced collective beliefs, we have a problem if we want to establish that such community beliefs work as a brake on scientific change. Recall that p is broadly relevant in the community in addition to being widely believed by the community members to be well established by the evidence. Hence, p expresses the sort of view that would make its way into textbooks for students or young researchers entering the subdiscipline or be used widely in further research. If the unforced collective beliefs of scientific communities have the characteristics of being broadly relevant and widely believed, at least for a while, it will be hard to test whether they work as a brake on scientific change. For much relies on such a belief, so it will be unpleasant for community members if recalcitrant evidence turns up regardless of whether they are jointly committed to the belief. Recalcitrant evidence is thus likely to be met with some skepticism or resistance whether a joint commitment is in place or not. Hence, if joint commitments make it harder than it would already be to abandon such views, it will be hard to detect.
Fagan (2011) defends a similar conclusion. She criticizes certain explanatory arguments that scientific groups have collective beliefs: It is not the case, she argues, that collective beliefs can explain certain phenomena in science—the inertia of science and the stability of groups in science—that cannot be explained just as well by other means.
Community Beliefs: Brakes on Scientific Change?
If my argument is correct, joint commitments of communities in science are rare and without much potential for working as brakes on scientific change. Collective beliefs developed at consensus conferences appear to sometimes be collective beliefs of whole communities. But, as explained above, their potential for working as a brake on science is limited by the fact that they are very rare and that at least some consensus statements come with the proviso that this is our view at this point in time. Codes for responsible conduct of research may also be good candidates for collective beliefs of communities. But if a community has a collective belief on what are responsible research practices, this limits the potential of any collective scientific belief of that community to work as a brake on scientific change. If evidence recalcitrant to the collective scientific belief turns up, the members of the community are forced to violate one of two joint commitments. By ignoring the recalcitrant evidence, they violate their collective belief about responsible research practices, by doing the opposite, they violate their collective scientific belief. In such cases, we have no argument why collective beliefs work as a brake of scientific change by making scientists ignore recalcitrant evidence, unless we can argue that the collective scientific beliefs of a given community are somehow harder to violate than the other joint commitments of the community. Hanne Andersen and I (2017) argue that there are cases in which participants can, due to changes in circumstances, violate a joint commitment without risking rebuke and that this is a major source of instability of collective beliefs and joint commitments in general. We would expect tension between collective beliefs due to changes in circumstances to be another major source of instability of collective beliefs.
If we instead assume that collective beliefs of communities are as ubiquitous as Gilbert claims, norms in general are also collective beliefs (Gilbert 1999). But then norms in the scientific community and scientific subcommunities, such as the norm of sharing counterevidence with one’s colleagues, are also collective beliefs. If members of a scientific community discover evidence that goes against one of their collective scientific beliefs, they are thus forced to violate either the collective scientific belief or a collective belief about responsible conduct of research. The relevant collective belief(s) about responsible conduct of research may be held by the community in question, the scientific community as a whole, or both. Hence, unless it can be argued that the collective scientific beliefs of a given community are harder to violate than other collective beliefs the community members are party to, it is not clear that collective scientific beliefs, even if ubiquitous, will work as a brake on scientific change.
I would like to end by thanking Gilbert for her inspiring work, which I continue to explore.
Acknowledgements: The author thanks K. Brad Wray for helpful feedback on an earlier draft.
Andersen, Hanne. “Joint Acceptance and Scientific Change: A Case Study.” Episteme 7, no. 3 (2010): 248–265.
Andersen, Line Edslev. “Outsiders Enabling Scientific Change: Learning from the Sociohistory of a Mathematical Proof.” Social Epistemology 31, no. 2 (2017): 184–91.
Andersen, Line Edslev, and Hanne Andersen. “The Stability and Instability of Joint Commitment.” Submitted. 2017b.
Bird, Alexander. “Social Knowing: The Social Sense of ‘Scientific Knowlegde’.” Philosophical Perspectives 24, no. 1 (2010): 23–56.
Bouvier, Alban. “Individual Belief and Collective Beliefs in Science and Philosophy: The Plural Subject and the Polyphonic Subject Accounts.” Philosophy of the Social Sciences 34, no. 3 (2004): 382–407.
Cheon, Hyundeuk. “In What Sense is Scientific Knowledge Collective Knowledge?” Philosophy of the Social Sciences 44, no. 4 (2014): 407–423.
de Ridder, Jeroen. “Epistemic Dependence and Collective Scientific Knowledge.” Synthese 191, no. 1 (2014): 37–53.
Dragos, Chris. “Which Groups Have Scientific Knowledge? Wray vs. Rolin.” Social Epistemology 30, no. 5–6 (2016a): 611–623.
Dragos, Chris. “Justified Group Belief in Science.” Social Epistemology Review and Reply Collective 5, no. 9 (2016b): 6–12.
Fagan, Melinda Bonnie. “Is There Collective Scientific Knowledge? Arguments from Explanation.” The Philosophical Quarterly 61, no. 243 (2011): 247–269.
Gilbert, Margaret. “Modelling Collective Belief.” Synthese 73, no. 1 (1987): 185–204.
Gilbert, Margaret. “Social Rules: Some Problems for Hart’s Account, and an Alternative Proposal.” Law and Philosophy 18, no. 2 (1999): 141–171.
Gilbert, Margaret. “Collective Belief and Scientific Change.” In Sociality and Responsibility, edited by Margaret Gilbert, 37–49. Lanham: Rowman & Littlefield, 2000.
Gilbert, Margaret. Joint Commitment: How We Make the Social World. Oxford: Oxford University Press, 2014.
Gilbert, Margaret. “Scientists Are People Too: Comment on Andersen.” Social Epistemology Review and Reply Collective 6, no. 5 (2017): 45–49.
MacKenzie, Donald. “Slaying the Kraken: The Sociohistory of a Mathematical Proof.” Social Studies of Science 29, no. 1 (1999): 7–60.
NN. “Helicobacter pylori in Peptic Ulcer Disease.” NIH Consensus Statement 12, no. 1 (Feb. 7–9, 1994): 1–22. https://consensus.nih.gov/1994/1994HelicobacterPyloriUlcer094PDF.pdf.
NN, “Diagnosing Gestational Diabetes Mellitus.” NIH Consensus Statement 29, no. 1 (March 4–6, 2013): 1–30. https://consensus.nih.gov/2013/docs/Gestational_Diabetes_Mellitus508.pdf.
NN. “Code of Practice.” Newsletter of the European Mathematical Society 87 (March 2013): 12–15.
Rolin, Kristina. “Science as Collective Knowledge.” Cognitive Systems Research 9, no. 1–2 (2008): 115–124.
Staley, Kent W. “Evidential Collaborations: Epistemic and Pragmatic Considerations in ‘Group Belief’.” Social Epistemology 21, no. 3 (2007): 321–35.
Thagard, Paul. “Ulcers and Bacteria I: Discovery and Acceptance.” Stud. Hist. Phil. Biol. & Biomed. Sci. 29, no. 1 (1998a): 107–136.
Thagard, Paul. “Ulcers and Bacteria II: Instruments, Experiments, and Social Interactions.” Stud. Hist. Phil. Biol. & Biomed. Sci. 29, no. 2 (1998b): 317–342.
Thagard, Paul. How Scientists Explain Disease. Princeton: Princeton University Press, 1999.
Thagard, Paul. “Explaining Economic Crises: Are There Collective Representations?” Episteme 7, no. 3 (2010): 266–283.
Tollefsen, Deborah, and Rick Dale. “Naturalizing Joint Action: A Process-Based Approach.” Philosophical Psychology 25, no. 3 (2012): 385-407.
Tossut, Silvia. “Which Groups Have Scientific Knowledge? A Reply to Chris Dragos.” Social Epistemology Review and Reply Collective 5, no. 7 (2016): 18–21.
Weatherall, James Owen, and Margaret Gilbert. “Collective Belief, Kuhn, and the String Theory Community.” In The Epistemic Life of Groups: Essays in the Epistemology of Collectives, edited by Michael S. Brady and Miranda Fricker, 191–217. Oxford: Oxford University Press, 2016.
Wray, K. Brad. “Collective Belief and Acceptance.” Synthese 129, no. 3 (2001): 319–333.
Wray, K. Brad. “Who Has Scientific Knowledge?” Social Epistemology 21, no. 3 (2007): 335–345.
Wray, K. Brad. “Collective Knowledge and Collective Justification.” Social Epistemology Review and Reply Collective 5, no. 8 (2016): 24–27.
Wray, K. Brad. “The Impact of Collaboration on the Epistemic Cultures of Science.” In Scientific Collaboration and Collective Knowledge, edited by Thomas Boyer-Kassem, Conor Mayo-Wilson, and Michael Weisberg. Forthcoming from Oxford University Press, 2017.
 This is not necessarily a fully appropriate term. There has been some debate on whether groups can be jointly committed to beliefs or whether they can merely be jointly committed to accept claims, a debate started by K. Brad Wray (2001). This paper is neutral on this question.
 For convenience, I speak of this as the Kuhnian explanation in the paper, but the paper is not intended as a comparison of Gilbert’s account with Kuhn’s account. I do not mean to argue that we should choose Kuhn’s whole theory of scientific change over Gilbert’s theory of scientific change. This is not clearly stated in the paper. I thank Deborah Tollefsen for pointing this out to me. I do think the question, addressed in Weatherall and Gilbert 2016, of how the work of Gilbert relates to that of Kuhn is an important one.
 In her paper, they are compared as alternative explanations of the effects of the dogma of reproductive biology that there is no cell renewal in the ovary.
 This question was also examined by Rolin 2008. Wray (2007) started a discussion of the general ability of scientific communities to hold views (Rolin 2008; Cheon 2014; Dragos 2016a, b; Tossut 2016; Wray 2016).
 For case studies that support the view that smaller groups in science form collective beliefs, see Bouvier 2004; Staley 2007; and Andersen 2010. For a promising approach to test whether joint commitments exist, see Tollefsen and Dale 2012, which gives an account of how empirical research in cognitive science is important to understanding the nature of shared intention.
 Gilbert herself has been focusing on the persistence of joint commitments and written very little about the sense in which they lack persistence, but acknowledges that this is an important topic (Gilbert 2014, 32).
 Gilbert’s (2000, 47) first paper on the role of collective beliefs in science was prompted by this work.
 This work has recently been taken over by others (www.consensus.nih.gov).
 One of the phenomena Gilbert tries to make sense of with her account of joint commitment are cases where people have inconsistent beliefs on some matter and nonetheless let a view stand as the view of the group. Kent Staley (2007) shows that the members of a research team can do (and do) this in epistemically rational ways (see also Wray 2017, 118–119). His argument applies equally well to other groups of scientists.
 By contrast, Bird 2010, 10, and de Ridder 2014, 41, state that there is no mechanism of Gilbertian community view formation in science.
 Hence, I disagree with Hyundeuk Cheon (2014) who argues that Gilbert and Wray speak about two different types of collective belief. On my interpretation, Gilbert and Wray are examining different questions about non-summative belief rather than different types of collective belief: Wray examines what kinds of groups have beliefs in a non-summative sense, while Gilbert examines how groups have beliefs in a non-summative sense (and argues that they do so by virtue of joint commitments).
 I mentioned above that scientists rather infrequently express community views, in whatever sense, but they do so in textbooks.
 Fagan argues that the existence of collective beliefs of communities is thus hard to test from their consequences. In the previous section, I made a case for the existence of collective beliefs of scientific communities by looking at how consensus is established at consensus conferences. But this is also hard in the case of “non-forced” collective beliefs of communities. It is not clear where we have to look to observe the process by which they are established. They are likely not established at a single event. It is also harder to see whether a group belief is a collective belief when it, judging from the personal beliefs of the group members at the time of its establishment, could just as well be a mere summative belief.