Philosophers of science and social epistemologists have long analyzed the division of “epistemic,” that is, knowledge-creating, labor within scientific communities. More recently, philosophers have also begun to examine aspects of division of epistemic labor when they examine inter-individual exchange and scientific collaboration within research groups, building upon accounts of epistemic dependence or group belief and collective knowledge. To continue this line of philosophical inquiry, group-focused perspectives on the division of labor in science should be able to relate to existing, predominantly community-focused perspectives. But not only do group and community-focused perspectives apply a different “zoom” when they choose to foreground either micro or macro-phenomena, they also approach the division of epistemic labor with different conceptual emphases.
When philosophers of science and social epistemologists consider the division of labor for scientific communities, they are typically concerned with the ways in which socio-cognitive diversity at the community level serves efficiency, the elimination of unwarranted bias and epistemic risk-spreading (e.g., D’Agostino, 2009; de Langhe, 2010; Kitcher, 1993; Solomon, 2006). Philip Kitcher’s work has pioneered community- concerned approaches to this division of labor, and I therefore present his account in some more detail. Kitcher (1990, 1993) explores how a scientific community fares best when forced to choose between competing theories or methods. When two rival methods or theories are on the table, which one should the community choose? Put differently, how can a community minimize the risk of scientific failure? Kitcher proposes the division of “cognitive” labor (Kitcher, 1993, p. 344) as a means to balance the epistemic risks attached to the method or theory choices. It is not reasonable for a community of scientists, he argues, to endorse only those options which appear most promising, that is, those which can be assigned the highest probability of yielding scientific success. Instead, scientific communities perform best when they divide the labor such that different research options are pursued in competition (Kitcher, 1993,
With the help of mathematically formalized models, Kitcher discusses how such a division of labor can be achieved in a community of individual scientists who have to choose between different methods or theories to invest themselves in. If these scientists were purely “epistemically” minded, that is, if they only weighed scientific knowledge gains against the resources that the creation of such knowledge requires, then, so Kitcher argues, no broad division of labor would be achieved and the epistemic risks would be poorly distributed among community members. For individual scientists that take only strictly epistemic criteria into account, Kitcher holds, it is not rational to invest themselves in anything other than the theory or method option that has the highest probability of scientific success. Assuming that all members of the community would converge in their judgment of success probabilities, no division of labor would occur. This, however, is neither desirable from a community point of view nor actually observable in scientific practice. Scientists in communities do divide the pursuit of different theoretical or methodological routes among themselves, and they do so, Kitcher argues, because at least some of them take into account “non-epistemic” or “social” criteria when deciding which theory or method to pursue. They will consider not only the probability of scientific success, which the adoption of a particular method implies, but also the credit they could receive in the case of success. While little credit is to be gained by following the mainstream, much is to be gained in a group of vanguard outsiders. Given different preferences in balancing the risk of failure against gains in reputation, Kitcher argues, individual and community rationality can be aligned (Kitcher, 1990).
In the context of this chapter, two aspects of Kitcher’s account of the division of labor on a community basis merit further discussion—its notion of community and the role of not strictly epistemic triggers of division of labor. Kitcher’s notion of community does little to support analyses of group collaboration. Kitcher realizes that individual scientists belong to what he calls “fiefdoms” (laboratories) (Kitcher, 1990, p. 17), but he does not pursue this thought. Rather, he characterizes scientific communities as a set of individual scientists, whom he conceives of as independent, competing and rational decision-takers. Furthermore, he argues that the social arrangement of science as a credit-distributing institution, as one that addresses not just purely epistemic ambitions, triggers a desirable division of labor. However, to foreground institutional incentives easily obscures which other knowledge-concerned conditions may entail division of labor in science.
In fact, a number of philosophers have argued that the division of labor does not require non-epistemic values, credit and prestige, nor the mechanisms through which they are realized, to take effect. Instead, the uneven distribution of epistemic resources, such as experimental infrastructure, skill and expertise within a scientific community, may incite different scientists to pursue different, competing lines of research (Giere, 1988, p. 213f-). Moreover, differences in scientific background assumptions will lead individual scientists to assess the probability of epistemic success attached to theory and method choices differently (Goldman, 1999, p. 257). In addition, the epistemic values at play in scientific practice are, according to D’Agostino, sufficiently diverse and ambivalent so as to “[...] exhaust the differences that we require to support diversity in exploratory behavior and, hence, to spread risk" (D’Agostino,
2005, p. 205).
However, while community-focused approaches to the division of labor in science focus on individual decision-making, they tend not to consider the inter-individual relations that may be the trigger or result of the division of labor at the group level. The division of labor in inter-individual relations is, instead, considered by some social episte- mologists in the context of epistemic dependence between experts and relative laypeople (as, e.g., in Hardwig, 1991; Ruloff, 2003). Christopher Gauker, for example, defines the division of epistemic labor as “a social arrangement in which people benefit from the expertise that others possess regarding subjects of which they themselves do not possess an expert understanding" (Gauker, 1991, p. 303). In a similar vein, Goldberg (2011) ph rases the division of epistemic labor as a net of dependence relations among individuals within a scientific community. While I elaborate on the issue of epistemic dependence in scientific practice in Chap. 7, let me note for the time being that these dependence-focused approaches concern themselves primarily with isolated acts of inter-individual change, largely leaving aside the contexts of collaboration and the social texture in which these acts can be embedded.
New motives for the study of the division of labor at the group level and its epistemic relevance for scientific knowledge creation have recently come from accounts of research group collaboration. Yet, this strand of literature is not concerned with the division of epistemic labor per se but rather foregrounds the question as to whether collaboratively created scientific knowledge should be considered “collective knowledge" (see, e.g., Andersen, 2010; de Ridder, 2014; Matthiesen, 2006; Rolin, 2010; Wray, 2001). Much of this literature developed out of the concept of a “joint," that is, irreducibly collective “group belief," which Margaret Gilbert proposed (Gilbert, 1989) and which highlights the moral aspects of group collaboration, for example, elaborating upon the webs of commitments and dependencies that can bind group members to one another. I will discuss this literature in more detail in Chap. 9.
Many group-focused perspectives upon research collaboration display a growing interest in the nitty-gritty of group collaboration, an interest that is often pursued by means of case study. Let me briefly provide two examples to illustrate the bandwidth of the empirical phenomena that this literature has sought to deal with so far. Hanne Andersen develops her account of joint acceptance with reference to a historical case study on research on induced radioactivity that resulted from clusters of interdisciplinary collaboration in the 1930s (Andersen, 2010). In contrast, when Kent Staley and William Rehg study a case of contemporary high-energy physics, they study research collaboration on a much larger scale, involving the institutionally coordinated efforts of hundreds of scientists (Rehg & Staley, 2008). Both Andersen as well as Rehg and Staley, however, are ultimately less interested in the division of epistemic labor among collaborators than in characterizing the kind of consensus that collaborating scientists reach when they co-author publications.
Against the backdrop of this literature, in this chapter I empirically examine different patterns of division of labor in the two research groups I have studied. Investigating the nitty-gritty of group collaboration will show how far the existing notion of a division of epistemic labor, a notion that has by and large been elaborated in reference to scientific communities, can help articulating aspects of a social epistemology of research groups.