Irreducibly Collective Justification
When de Ridder (2014) formulates an account of collective scientific knowledge, he locates the collective quality of such knowledge in its justification. Scientific standards, he argues, demand a particularly qualified justification, a kind which individual scientists cannot always guarantee. If scientific justification for a piece of knowledge can only be guaranteed by a team of scientists, then scientific knowledge is collective—provided that the team in question consists of scientists who are mutually dependent upon one another and whose mutual dependence is cognitively necessary. Let me elaborate upon the details of this account.
Scientific justification, according to de Ridder, is a particularly demanding form of belief justification. The belief of a subject S has scientific justification “[...] only if it is properly based on a properly performed and objectively reliable process of scientific inquiry, the purpose of which was to gather evidence for the truth of p, and S understands this to be so" (de Ridder, 2014, p. 45). A scientific justification thus understood comprises a procedural requirement (i.e., the belief has to be properly based on an adequate research process) and a reflective requirement (i.e., S understands this to be so). While the procedural requirement need not be fulfilled by S and could be fulfilled by another epistemic subject, S has to fulfill the reflective requirement.
Both requirements are coupled to one another. For S to fulfill properly the reflective requirement, S has to be capable of obtaining first-hand information about the fulfillment of the procedural requirement. If S would necessarily have to rely upon the testimony of others for the information that the procedural requirement is fulfilled, de Ridder argues, S does not meet the reflective requirement: “[...] in relying on testimony- based justification, she [the subject] doesn’t really understand in a direct way that the process of inquiry on which her belief is (ultimately) based is properly performed and objectively reliable, and that the evidence it produces indeed supports p [i.e., the belief whose scientific justification is in question]" (de Ridder, 2014, p. 48).
Because research is a collaborative endeavor, in many cases only a group of individual scientists is able to fulfill both procedural and reflective requirement. Yet not all collaboratively created scientific knowledge, de Ridder argues, is irreducibly collective. He confines irreducibly collective knowledge to cases of mutual and “cognitively necessary epistemic dependence" (de Ridder, 2014, p. 46). Epistemic dependence is cognitively necessary when the dependent individual is limited in expertise and when these limits are the motivation for engaging in a dependence relation. An individual that is cognitively dependent upon collaborators, de Ridder continues, is not able to fulfill the reflective requirement of scientific justification.
De Ridder distinguishes cognitively necessary epistemic dependence from dependence that is merely “practically necessary" (de Ridder, 2014, p. 46), a form of dependence that arises when scientists delegate parts of the research process to others, not because they lack the expertise, but because they lack the material resources to perform them. Sometimes, de Ridder observes, “the evidence needed to substantiate a conclusion is too much for any individual to collect and process. It is not that doing so requires skills or expertise that no single individual has; it is purely a matter of time" (de Ridder, 2014, p. 46). Yet in such cases, de Ridder argues, individual scientists can access the evidence after it has been created, provided it is properly conserved. Given the necessary expertise, practically necessary dependence does not hinder scientists in understanding how a piece of research is properly based upon an adequate process of inquiry. Hence, de Ridder argues, practically dependent scientists can fulfill the reflective requirement and have scientific justification individually.
When scientific justification can only be had collectively, that is, when collective scientific knowledge obtains, an individual can have collective scientific knowledge “derivatively" (de Ridder, 2014, p. 48). Individuals, de Ridder points out, can appropriate collective scientific knowledge into their individual knowing on the basis of testimony. Testimonial justification, however, falls short of scientific justification.
For the purpose of this chapter, let me label de Ridder’s notion of collective scientific knowledge the epistemic dependence account (ED). On the ED account, in de Ridder’s formulation, collective scientific knowledge applies under the following condition:
According to the ED account, a necessary condition for collective scientific knowledge to obtain is mutual, cognitively necessary epistemic dependence among a group of individuals that allows only the group to satisfy the requirements of scientific justification for a proposition p.
This condition for collective scientific knowledge can be translated into the terminology that I introduced in Chap. 7. There, I developed the notions of opaque and translucent epistemic dependence. While opaque dependence is cognitively necessary, translucent dependence is not. I have also developed notions of first and second-order reasons, a distinction that can be linked to de Ridder’s argument as well. De Ridder argues that for a subject to have scientific justification for p means to have sufficient “non-testimonial evidence" (de Ridder, 2014, p. 48) to assume that p is properly based on an adequate process of inquiry. Put differently, first- order reasons can constitute a scientific justification as they pertain to immediate, that is, non-testimonial, evidence forp. Second-order reasons, in contrast, concern the trustworthiness of a speaker and justify reliance upon a speaker’s testimony thatp. Second-order reasons may be admissible in the justification of individual knowing, but they cannot constitute a scientific justification. Interpreting de Ridder’s ED account with this terminology, we can reformulate the condition for collective scientific knowledge as follows:
According to the ED* account, a necessary condition for collective scientific knowledge to obtain is mutually opaque epistemic dependence among a group of individuals that allows only the group to provide those first-order reasons that the scientific justification for a propositionp requires.
For a discussion of whether the ED* account of collective knowledge applies to research groups and their collaborative scientific practice, and particularly to the collaborative authoring of research publications, I will again confine myself to my comparative case study. To support the ED* account of collective knowledge, I need to mobilize
- (i) observations of empirical phenomena that are well accounted for by ED* and/or
- (ii) observations of empirical phenomena indicating that scientific practice complies with implications of ED*
Drawing on analyses in previous chapters, I argue that the ED* account explains, in fact, an empirically observable phenomenon. It explains the predominance of collaborative authorship, itself a reflection of the division of labor in research groups. Group members collaborate with one another on a day-to-day basis, and I have characterized their collaborative relationships as help, delegation and deference, and I have argued that these relationships give rise to epistemic dependence, some of it opaque epistemic dependence. Yet, ED* implies that in the absence of mutually opaque epistemic dependence, collaboratively created scientific knowledge is not collective—an implication that I will take issue with in the remainder of this section.
The ED* account follows de Ridder’s ED account in limiting collective knowledge to cases of mutually opaque (or, as de Ridder phrases it, “cognitively necessary") epistemic dependence. De Ridder maintains that:
For cases where teamwork is a practical necessity, it could be argued that
once the evidence is all collected, properly stored, and made accessible,
individuals can have it non-testimonially. (de Ridder, 2014, p. 47)
I want to highlight two issues with this argument, one analytic and one concerning its relevance for the analysis of scientific practice. First, even when teamwork is a mere “practical necessity" and evidence is “made accessible," an individual scientist cannot have scientific justification entirely non-testimonially. In contrast to de Ridder, I maintain that even for a scientist with the necessary expertise, mere access to stored evidence is not sufficient to be certain enough about the way in which a piece of evidence is based on experimental inquiry. If a scientist has not eye- witnessed the experiment, she cannot be sure that the experiment has been performed correctly and its results reported truthfully. An unexpected result may catch her eye and raise skepticism, but an expected yet inferior result, accompanied by proper documentation, may not. Hence, if she has not eye-witnessed the experiments, even the scientist with the necessary expertise will have to rely upon the trustworthiness of her collaborator— that is, in justifying her belief in the scientific quality of the evidence in question, the scientist will have to rely upon second-order reasons. She alone cannot provide a scientific justification based on the experimental evidence provided by a collaborator. Therefore, contrary to de Ridder, I argue that collective knowledge does apply here.
Second, even when evidence could be had non-testimonially, my empirical observations suggest that, in actual practice, scientists often choose to let their epistemic dependence stand. They do not, whenever possible, eye-witness collaborators’ experiments or reproduce their experimental results—time is too scarce, and experimental resources too expensive. This holds particularly for the molecular biology laboratory where single experimental steps can quickly cost several hundred, if not several thousand, Euros. So even though the laboratory is a mono-disciplinary research group where epistemic dependence is often translucent, scientists do not necessarily seek to minimize dependence and reliance upon testimony.
When I asked Johan, the lab leader, whether he controls or checks the scientists in his group, he replies: “I mean, as a mentor, right, you do have an influence on people as a mentor." In the interview, he also finds it important to assure me that for the research in his lab he knows exactly what the “criteria for good scientific conduct" are. He furthermore mentions that he has a “thorough" understanding of the journal articles that he co-authors as a lab leader and collaborator (Johan, interview, group2). He does not claim, however, that he typically makes a point of eye-witnessing or reproducing experimental procedures. And neither do those lab members who allot more of their time to work at the lab bench. When I ask Alex, who has just finished his dissertation, whether he repeats experimental steps that a close collaborator of his performed, he answers:
I mean for sure one is controlling each other as well within a group. I think that’s also important, because there is always and is always gonna be people doing mistakes, if they are made just by chance or if it was really willing to do the mistake, it’s always gonna happen, because there is so much pressure, right. So, it’s important that you also internally have kind of a control system. However, this is more I would say on a base like random control and so on, yeah. And generally I mean it's not a big issue I would say within the group, because you know the experimental set-up, you follow them in group meetings and so on, so you know what they’re doing. (Alex, interview, group2)
Collaborators, as Alex sees it, do control one another. But they do not do so all the time since, he continues, “you gonna easily figure out if they’re constructing data" (Alex, interview, group2). Much more difficult to detect than fraud and plain failure are experimental results that could have been better, for example, more precise or comprehensive, that maybe could have yielded serendipitous insights—if experimental routines had been handled with more skill, care or patience:
[...] at the end of the day you have to convince other people that what you have done in your experiments is right, it’s — you know, that the experiments have been done correctly and so on, and there is no sloppiness in there and so on. (Alex, interview, group2)
Even adequately stored data need not show “sloppiness.” And even scientists with the necessary expertise often choose to remain epistemically dependent upon their collaborators, not fulfilling the reflective requirement of scientific justification.
A robust notion of collective scientific knowledge should reflect these realities of scientific practice—the shortness of material resources and time, the pervasiveness of dependence and trust. Collective scientific knowledge should account for a broad range of de facto situations. It should not put too much emphasis on the available, yet unrealized possibility of epistemic independence. Therefore I suggest an account of collective scientific dependence, ED**, that is not limited to opaque epistemic dependence but encompasses cases of translucent epistemic dependence as well:
According to the ED** account, a necessary condition for collective scientific knowledge to obtain is mutual epistemic dependence among a group of individuals that allows only the group to provide those first-order reasons that the scientific justification for a proposition p requires.
One may object to the scope of the ED** account, arguing that according to it most collaboratively created scientific knowledge should be considered collective—a conclusion that many social epistemologists are uneasy about. De Ridder, for example, anticipates the objection that his ED account would “overgeneralize,” an objection he defends himself against by emphasizing how limited the conditions for collective scientific knowledge are if mutual cognitively necessary epistemic dependence is required (de Ridder, 2014, p. 51).
But why does a broad notion of collective scientific knowledge appear objectionable? Because it challenges individualism, a stance with a strong foothold in much of epistemology (as noted in, e.g., Fricker, 2006; Grasswick, 2004; Kusch, 2002). Epistemological individualism focuses on the individual as a primary epistemic subject. What is more, as a value, individualism transports the ideal of epistemic self-reliance, that is, the ideal of individual knowing that would not need to resort to the testimony of others. De Ridder defends this ideal, not sweepingly but discreetly, when he argues that scientists whose epistemic dependence is practically necessary could be epistemically independent—that they are, in fact, often not is an empirical insight about actual scientific practice that he is either not aware of or deems irrelevant.
With ED** I propose an account that recognizes the outcome of much of collaborative scientific practice as collective scientific knowledge, a position that I suggest calling “inter-individualism." While I provide a sense in which collaboratively created scientific knowledge should be considered collective, my approach foregrounds the individual epis- temic efforts that scientists undertake in collaborative scientific practice, highlighting the inter-individual exchanges that scientific collaboration consists of.