The approach to expert communities and political representation of non-experts in Harry Collins and Robert Evans’ elective modernism reflects the conviction that experts are not representative of ordinary citizens. I use an analysis of aspects of representation and the argument from inductive risk to argue that experts can be seen as representative of (some) non-experts, when we understand representation as resemblance based on shared social perspectives and acknowledge the inevitable involvement of such perspectives in decisions under inductive risk. This, in turn, has implications for some of the proposals about practices and institutions made in elective modernism.
The aim of the paper is to contribute to the discussion of scientific expertise and democracy by engaging with one vision proposed for them—Harry Collins and Robert Evans’ “elective modernism” (Collins and Evans 2017, p. vii). In particular, the paper focuses on some issues related to expert communities and representation of non-experts in decision-making concerning science and technology. Elective modernism’s approach reflects the idea that members of expert communities are non-representative of citizens in general: in the case of expert groups, “[s]uch groups may speak ‘to’ the members of the wider society but cannot claim to speak ‘for’ them” (Evans and Plows 2007, p. 829, emphasis in the original). I argue that the discussion of representativeness, or the lack thereof, requires considering different meanings of representation. In particular, I argue that once several aspects of representation are discerned, on one of these views of representation experts can be seen as speaking for (some) non-experts and this has implications for several aspects of elective modernism. In developing this argument, I draw on some work in philosophy of science and at the end of the paper I briefly discuss how the argument, in turn, contributes to work on inductive risk.
I begin by summarizing Collins and Evans’ account of expertise (2002) and the related discussion of the political phase of decision-making that Evans and Alexandra Plows (2007) propose as a necessary addition. I also describe some further developments of Collins and Evans’ ideas (2017). In the following section, I introduce Mark Brown’s (2009) work on different aspects of democratic representation. I use Brown’s account to argue that some social groups may in an important sense be represented in practices and decisions of expert communities. In section 4, I give an overview of several arguments that discuss inductive risk and the role of values in science, aiming to show that a particular type of these decisions is ubiquitous and unavoidable in scientific research. In section 5, I discuss the consequences of these arguments for elective modernism. Collins and Evans (2017) conclude that there is some agreement between elective modernism and Heather Douglas’ (2009) version of the argument about values in science and that on several points of disagreement elective modernism has advantages. Contrary to this conclusion, I argue that the argument from inductive risk and the work on different senses of representation, when brought together, have more serious implications for elective modernism. Section 6 discusses some of the more specific consequences for the proposals made in elective modernism, with a particular focus on the issues related to transparency in expert communities. In the penultimate section, I describe how my argument adds another consideration to the debate on transparency—an ongoing debate in the discussion of inductive risk and values.
2. Expertise and Democratic Society According to Elective Modernism
In their introduction of the “Third Wave of Science Studies” (2002), Collins and Evans focus on the question of decision-making when issues have both a scientific-technological side and wider public relevance—for example, whether to prefer nuclear power to coal or vice versa (2002, p. 236). They suggest that there are two problems raised by what they call “technical decision-making in the public domain” (2002, p. 235)—later, Collins and Evans refer to it as “technological decision-making” (2017, p. vii). One is “the Problem of Legitimacy” (2002, p. 237), or the problem of ensuring the political legitimacy of decisions.
According to Collins and Evans, the earlier science studies have shown convincingly that these decisions should not be left to technical experts alone. However, there is also the second problem, “the Problem of Extension” (2002, p. 237), or the problem of showing how extensive inclusion in decision-making should be and, importantly, where it should stop. The aim of the “Third Wave of Science Studies,” or the “Studies of Expertise and Experience” (2002, p. 236), is to provide a principled solution to this second problem. Collins and Evans (2002, pp. 261–262) propose to distinguish between the technical phase and the political phase of technological decision-making. Against the background of this distinction, Collins and Evans make a normative claim about expertise. It is expertise that is to serve as the basis for inclusion in the technical phase.
Crucially, Collins and Evans describe experts’ decision-making as properly distanced from social and political influences. Politics may influence science—it inevitably does, in fact, and this is an important lesson of the earlier science studies (the Second Wave). However, one must not appeal to political values in the technical phase explicitly or make it easier for these values to influence experts.
What Shapin’s and similar studies show is that politics of this sort may influence science, but not that it is a legitimate input to scientific decision-making. … [O]ne would never set out to design scientific or political institutions to enhance the influence of “big-P” politics on the content of such an esoteric science: one would do quite the opposite. (Collins and Evans 2002, p. 245, italics in the original)
Esoteric sciences provide a model for thinking about the Problem of Extension. Collins and Evans suggest that in their case it is generally recognized that only the group of experts—the “core-set” (2002, p. 242)—who are directly involved in research on the issue should have a say on related scientific-technological matters. According to Collins and Evans, the general recognition of this is a fact about our society; rejecting it means rejecting our type of society.
[I]f one takes a really esoteric scientific controversy such as that over the detection of gravitational waves, … then members of Western society know, without having to agonize, that anyone who is not a recognized physicist with a great deal of equipment or special theoretical knowledge will not be, and should not be, counted as a member of the set of decision-makers in respect of the scientific knowledge itself. (Collins and Evans 2002, p. 242, italics in the original)
At the same time, Collins and Evans argue that expertise is not limited to “certified” (2002, p. 237) experts. Laypersons without scientific or technical qualifications may sometimes possess “experience-based” expertise (2002, p. 238). Again, this is an important lesson of some studies in the Second Wave, such as Brian Wynne’s work on Cumbrian sheep farmers after Chernobyl (see, e.g., Collins and Evans 2002, pp. 255–256). These experience-based experts, too, should be involved—not because they are citizens but because they are experts. In the absence of relevant expertise, one has no right to participate in the technical phase of decision-making—neither being a citizen in general nor having a stake in the issue in particular gives such a right.
Evans and Plows (2007) provide an important addition to the Third Wave by discussing the issues of participation in technological decision-making for non-experts. In particular, Evans and Plows focus on the problem of representation. Crucially, they argue that experts—both certified and non-certified—cannot legitimately represent lay perspectives. They stress the importance of acknowledging “the way in which holding an expertise that is not ubiquitous undermines claims to be speaking for ‘citizens in general’” (2007, p. 829; emphasis in the original). What experts represent is their own expert communities, or specific epistemic groups.
Accordingly, for technological decision-making to remain democratically legitimate and accountable, participation for non-experts specifically must be guaranteed. This participation is to take place in the political phase, which includes making decisions about framing, planning, and use of new technologies and ensuring democratic oversight over them.
Specifically, Evans and Plows discuss citizen juries as a format that can be helpful for providing representation to non-experts. Citizen representatives on such juries are defined through their disinterestedness, understood as the lack of expert knowledge about the issue in question. Evans and Plows argue that such a lack is necessary for being capable of representing the ordinary public. As Evans and Plows stress, “it is only those who are non-experts with regard to the science in question who can authentically represent the lay perspective implied in calls for the democratization of science” (2007, p. 829). Toward the end of their argument, they reinforce this point: “if the intention is to put science before lay people, then it is only those who stand outside the committed knowledge cultures of both the scientific and activist communities who can operationalize a genuinely civic epistemology” (2007, p. 845, italics in the original).
In addition to disinterestedness, in discussion of citizen juries in practice Evans and Plows draw attention to the importance of diversity among those chosen as representatives. For example, a 2004 citizen jury on “designer babies” they describe (2007, pp. 842–843), aimed to have a group that was representative in terms of gender and socio-economic status and also, in terms of including both participants who had children and those who did not, as well as participants with genetic conditions that in future might be targeted by screening. As Evans and Plows note, selecting participants so as to have “a representative sample of lay participants” whose judgments can be seen as a “a legitimate representation of the concerns of lay citizens” is one of the basic assumptions of such participatory events (2007, p. 845).
One further addition in elective modernism is the proposal for a novel expert institution. The purpose of the “Owls” (Collins and Evans 2017, p. 86) is to ensure that decision-making in the political phase can be fully informed by expert knowledge without being dominated by experts. This institution is to consist of persons who are capable of characterizing the state of debate on a particular scientific issue—not to proclaim what is the truth about the issue but to identify what is the position of the relevant expert community. After the Owls have done their job, policy-makers may take the experts’ position into account or ignore it; the only requirement is not to mischaracterize it. For example, policy-makers must not claim that more evidence is needed to settle a controversial issue when experts have in fact reached a strong consensus on it.
Since my overview of the Third Wave and elective modernism involves several works with different authors, it is helpful to clarify the subject of the subsequent argument. As described in the introduction, my argument aims to discuss some of the issues related to expert communities and representation of non-experts in science- and technology-related decision-making according to elective modernism. Collins and Evans (2017) present the most general account of elective modernism. However, it is the paper by Evans and Plows (2007) that discusses the question of representation in most detail. In what follows, I treat elective modernism as building on the argument by Evans and Plows (2007) and incorporating the same view on representation. The justification for doing so is Collins and Evans’ characterization of elective modernism as integrating several earlier papers in the Third Wave. In particular, Collins and Evans (2017, p. 18) refer to Evans and Plows’ (2007) work when discussing the further development of their discussion of the technical and the political phases.
In the remainder of this section I outline my interpretation of how representation is understood in elective modernism, aiming to make it clear what position my argument engages with. I suggest that in the arguments by Collins and Evans (2002) and Evans and Plows (2007), there can be distinguished several strands in the understanding of representation.1 On one view, representation is understood in the sense of ensuring representation for “the wider society” (Collins and Evans 2002, p. 272), reflecting diversity that exists among the citizenry. The discussion of representativeness in citizen juries by Evans and Plows (2007) can be seen as the most detailed expression of this understanding. Collins and Evans’ (2017, p. 115) warning that it is a mistake to see specific groups (such as some sections of the middle class) as representing the entire society may be interpreted as reflecting a similar understanding. Expert groups can be said to be non-representative on this view of representation.
Alongside this understanding, the notion of similarity, or the lack of it, also plays a role in discussing representation in the Third Wave. An indication of this understanding may be seen in Collins and Evans’ comment on the Cambridge Experimental Review Board (CERB), a 1976 citizen panel on recombinant DNA research:
Of course, it is possible to question the sense in which the members of the CERB ‘represented’ the population of Cambridge. Clearly they were unusual in that they were selected for, and then chose to be involved in, a complex scientific and technical controversy. (2002, p. 294, emphasis in the original)
As a number of the quotes provided earlier show, Evans and Plows (2007) stress that it takes a non-expert to represent a non-expert perspective authentically. Having non-ubiquitous expertise disqualifies one from that—possessing expertise undermines one’s ability to speak for those not in possession of expertise. I take this to mean that being dissimilar from others in a relevant respect makes one unsuitable to represent them, or non-representative of them.
Conversely, it appears that a relevant kind of similarity enables one to represent those with whom similarity is shared. The actual instances of citizen juries described by Evans and Plows (2007) raise the question of what it means for these small events to represent the diversity of the wider society. At first sight, the citizen jury on “designer babies” can be said to capture some relevant aspects of this diversity by including participants both with and without children and both with and without genetic conditions. However, this event, by design, only included participants between 16 and 19 years of age. If one considers age a relevant dimension of diversity, this jury was not fully representative of the wider society. (If one thinks of diversity in terms of experience, interests, values, and perspectives rather than age, one may still worry that a narrow age selection limits the diversity of experience, interests, values, and perspectives represented.) Yet, this and similar events are animated by the idea that the participants selected are capable of representing the “lay perspective.” I suggest that thinking about representatives and constituents as similar in some respect that matters is a part of this idea.
Thus, I see similarity as an element in elective modernism’s understanding of representation. The presence or the absence of expertise is, in turn, the grounds on which similarity is to be evaluated in the case of experts and non-experts. I acknowledge that expertise matters—there is a substantive dissimilarity between the two groups. At the same time, I aim to offer a further discussion of the notion of similarity and to argue that there is a different epistemically and politically relevant dimension on which experts may be said to be similar to (some) non-experts and representative of them in this sense. As the first step of my argument, I discuss what it means to be representative and how this applies to experts in elective modernism.
3. Considering Elective Modernism: What Does It Mean for an Expert Community to be Representative?
For discussing the notion of representativeness in connection with experts, I rely on Brown’s discussion of “elements” (Brown 2009, p. 201) of representation in his analysis of science, representation, and democracy (2009, pp. 201–237). Brown distinguishes five such dimensions of representation: authorization, accountability, participation, deliberation, and resemblance. Authorization means being authorized as a representative, most commonly through being elected (like a member of parliament) or through being appointed to a role (like a member of a scientific advisory board). Accountability has the meaning of being held accountable—for example, through being re-elected (and thus rewarded) or failing to be re-elected (and thus punished). In addition, it has a meaning of giving an account, or justifying one’s actions as a representative. Participation refers to forms of participation and venues for participation that contribute to democratic representation. An example is using lobbying or demonstrations when fighting to be recognized as someone who should be represented in policy-making. Deliberation describes the ways deliberation plays a role in democratic representation. For instance, deliberation may be seen as a way to identify reflective interests of constituents as opposed to their impulsive desires. Finally, resemblance involves some similarity, or some form of identification with representatives on the part of constituents.
In what follows, I propose a two-step analysis of elective modernism in terms of different elements of representation. First, I suggest that at first sight, such a discussion of core-sets supports the claim that experts are non-representative of non-experts. Second, I argue that despite that, further analysis focusing on one of the aspects of representation enables showing how in one important sense, (some) non-experts may be represented in expert communities.
Considering expert communities as described by Collins and Evans (2002), Evans and Plows (2007) and, following them, Collins and Evans (2017) in the light of different elements of representation, one can conclude that they do not realize any of them. Members of core-sets are not authorized by non-experts. One can be neither elected nor appointed to a core-set (as Collins and Evans stress, becoming an expert is only possible through successful socialization in the expert community, e.g., 2017, p. 14). Accordingly, members of core-sets are not held to account the way elected and appointed officials are. The division of functions between the technical and the political phases means that participation for non-experts is limited to the latter. In a similar way, the political phase is where democratic deliberation is expected to take place. So, expert communities are by design not meant to realize any of these two dimensions of representation. Finally, the issue of resemblance is where the lack of representativeness on the part of experts seems the clearest: experts have expertise and non-experts do not. As described in the previous section, this is also an element of representation that Evans and Plows (2007) explicitly discuss.
So far, Brown’s analysis of representation appears to be highly compatible with the understanding of representation in elective modernism. Collins and Evans describe it like that. In their discussion of related developments, Collins and Evans express their agreement with Brown’s analysis and his insistence that citizens need access to a variety of different forms of representation (Collins and Evans 2017, pp. 111–112). Nevertheless, Collins and Evans stress that expert advice is to be provided by experts, suggesting that expert communities are not a location to cultivate these opportunities for representation. As Collins and Evans state, “Beyond the insistence that expert advice should be provided by experts, elective modernism largely supports contemporary developments in political theory around deliberative systems, in which it is argued that deliberation and representation occur in multiple forms and locations” (2017, p. 131).
I do not question the claim of experts’ lack of representativeness on the first four dimensions of representation. Instead, I suggest that it is precisely the notion of resemblance that enables using Brown’s account to show in what sense experts can be representative of (some) non-experts.
As previously described, Brown begins the discussion of resemblance with the general notions of similarity and self-identification. However, Brown (2009, pp. 228–229) points out that thinking about similarity as a demographic match between constituents and representatives—for example, in characteristics such as race and gender—has limitations. In particular, this approach assumes that it is unproblematic to identify which of the multiple demographic groups one belongs to is the most relevant for representation. It also assumes that members of such a group all share the same interests and preferences. Instead, Brown proceeds to discuss resemblance using Iris Young’s notion of “social perspective” (Brown 2009, pp. 229–230). Social perspective is defined as based in some shared experience and giving rise to shared concerns. Elsewhere, discussing similarities between social and professional perspectives, Brown explains that “[i]n either case, perspectives encompass the questions, concerns, knowledge, and worldviews of particular social and professional groups” (2009, p. 234). However, there is no one-to-one match between social perspectives and wider social categories such as gender. Similarly, a perspective does not have a fixed content—social perspectives may evolve. For Brown (2009, p. 230), this open-endedness is what makes social perspectives especially suitable when deliberation is valued.
For my argument, the most important feature of social perspectives is their ability to serve as the basis for resemblance between individuals or groups that may be dissimilar in other ways. In particular, I see this understanding of resemblance as showing a possibility that experts may be representative of (some) non-experts. If an expert community exhibits a particular social perspective in its decisions and choices, it can be said to represent other social groups who share this social perspective. Conversely, this expert community can be said to be non-representative of other social groups who have different social perspectives. Representation in this case is understood as resemblance based on shared social perspectives. Importantly, this is a kind of resemblance that can coexist with the lack of resemblance on other grounds—for example, the lack of resemblance on the basis of some demographic characteristics but also the lack of resemblance due to the fact that one group possesses expertise and the other does not.
On the view of resemblance I propose here, the claim about resemblance is a descriptive claim about two groups each holding the same social perspective. This claim may be true even if these two groups do not perceive a connection based on such a shared perspective. In the case of expert communities, it is possible that choices and decisions involving specific social perspectives may never be explicated as such. In such cases, representation or the lack of it may remain unnoticed by both purported representatives in the expert community and their purported constituents.
Due to this descriptive character, my approach differs from Brown’s own discussion of resemblance. Brown argues that in thinking about resemblance it is important to recognize the active and deliberative role of constituents in developing self-identification with representatives. As Brown concludes, “A sense of resemblance may be intimated by shared group identity, or by statistical representativeness, but it is animated by ideas. It can be sustained only through deliberation and participation by the represented” (2009, p. 231). Thus, according to Brown, the descriptive claim of resemblance I discuss is just an “intimation” of resemblance as an element of representation. (However, I still follow Brown in focusing on social perspectives rather than statistical representativeness.)
I acknowledge the minimal character of the notion of resemblance I use (I briefly return to discussing Brown’s richer notion in section 6). Nevertheless, I argue that such a descriptive claim of representation as resemblance based on a shared social perspective allows seeing an important aspect of the activities of expert communities. When we analyze decisions made in the course of research from the point of view of the social perspectives involved, it may be possible to say which social perspectives are taken into account and which are not. To the degree that various social groups identify with these social perspectives, they too may be said to be either represented or not in the relevant expert community.
As described earlier, the notion of similarity constitutes an element of the discussion of representation in elective modernism: experts lack resemblance on the dimension that matters—expertise—and due to that cannot represent non-experts on this view of representation. The aim of this section is to take the first step for establishing that there is another relevant dimension on which experts can be said to share resemblance with (some) non-experts and thus be representative of them in this sense.
By way of introduction for the second step of the argument, I conclude this section by describing a potential objection to its first step. One could object that the claim just made is something of a trick to make expert community fit the definition of being representative even if it is not so in any meaningful sense. In particular, one could argue that this claim lacks significance if there are no epistemically and practically significant decisions and choices in experts’ communities that are suitable for analysis in terms of social perspectives. In the next two sections, I answer to this objection. In order to do so, I use the argument from inductive risk.
4. The Argument from Inductive Risk
Contemporary work on inductive risk often refers to the earlier argument by Richard Rudner (1953). According to Rudner, risk is unavoidable when making decisions about the sufficient amount of evidence and the sufficient degree of certainty for accepting a hypothesis. How much is sufficient depends on how bad it would be to make an error when making this judgment. Potentially serious consequences of the error require high standards of evidence. Evaluating the badness of consequences, in turn, is an ethical task that reflects one’s values.
Other philosophers have extended this argument to show how evaluation of potential errors plays a role in decisions throughout the research process. For example, Douglas (2000) analyzes several questions involving inductive risk in the course of toxicology research on laboratory animals. These questions include the classification of tissue samples (should ambiguous samples be classified as healthy or affected?) and the choice of the extrapolation model (should one use a threshold or a no-threshold model?). In a similar vein, Torsten Wilholt (2009) discusses inductive risk related to the question of what strain of laboratory animals to choose (different strains may have different sensitivity to specific chemicals). Depending on such choices, either false negatives or false positives become more likely. As a result, the chemical is likely to seem either less or more dangerous than it is. Depending on that, it becomes more likely that it is either under-regulated or over-regulated. Thus, choices made throughout the research process reflect the preference to avoid a specific type of error.
Another extension of the argument from inductive risk is the analysis of ways in which institutions (rather than individuals) deal with this risk. For example, Jacob Stegenga (2017) argues that setting the standards for the approval of new medicines reflects a particular “inductive risk calculus.” The institution evaluating medicines needs to balance two risks: the risk of approving unsafe or ineffective medicines, or the risk of “unwarranted drug approvals” (2017, p. 22), and the risk of rejecting safe and effective medicines, or the risk of “unwarranted drug rejections” (2017, p. 22).
According to the argument from inductive risk, certain decisions that are not just practically unavoidable but also epistemically necessary for research are not dictated by evidence—they require a judgment that reflects ethical and social values. The argument showing this can then be used to support proposals for institutional innovations for expert advice in democratic society.
For example, for Douglas (2009, pp. 66–86) the starting point is the general moral responsibility to avoid recklessness and negligence in one’s actions. In the case of scientific experts, actions include decisions made under the conditions of inductive risk in the course of research. Thus, meeting moral responsibilities for an expert requires consideration of values involved in these decisions.
Douglas suggests that this reflection upon values is where the public may play a helpful role: “At the very least, scientists need not do all this work alone” (2009, p. 157). Douglas (2009, pp. 156–174) describes two approaches to involving the public: the analytic-deliberative process and the consensus conference. The analytic-deliberative approach is suitable when there are specific local issues and the stakeholders are in a position to work with the researchers directly. The role of lay participants is to help to shape the guiding assumptions of the analysis, specifying acceptable levels of uncertainty. In the cases where issues are global and potential stakeholders too numerous, communication between the public and researchers cannot be as immediate. In such cases, Douglas recommends a consensus conference. A group of lay citizens broadly representative of the general public are gathered together to be informed about the issue, to deliberate, and to produce a consensus statement identifying values that are to guide subsequent research.
Despite similarity between participants of Douglas’ (2009) consensus conferences and Evans and Plows’ (2007) citizen juries, I suggest that Douglas’ proposal is considerably different from elective modernism’s. According to Douglas, experts are to consider social values in their choices as experts and lay participants are to play an active role in shaping experts’ epistemic choices. Close integration of experts’ and non-experts’ contributions is especially visible in the analytic-deliberative process where experts and stakeholders work together to shape choices of the technical phase. And even though participants in consensus conferences are not involved with experts directly, their recommendation can also be said to reach into the technical phase. In contrast, in elective modernism non-experts’ contributions are limited to the more political questions such as “the development of the policies and regulatory frameworks within which technical debates are permitted to take place” and the development of “a strategy for action that sets out what should be done” (Evans and Plows 2007, p. 833).
In the next section, I describe how Collins and Evans perceive Douglas’ argument and respond to it. I then lay out my claim that the argument from inductive risk, combined with a discussion of different senses of representation, is more consequential than Collins and Evans acknowledge.
5. Considering Elective Modernism: The Significance of the Argument from Inductive Risk
Collins and Evans (2017, pp. 119–128) discuss Douglas’ argument among other developments they see as relevant. There are several distinct issues Collins and Evans address in their response. First, they state that they fully agree with the claim about the ineliminable role of values in science (2017, p. 120). This is something demonstrated by the earlier science studies and thus it is a part of the Third Wave as well. Second, Collins and Evans discuss the practical consequences of the respective accounts of expert advice in democracy. They conclude that the comparison shows advantages of their own approach. As they argue, their Owls enable better input for decision-making in the political phase than analytic-deliberative processes or researchers’ own efforts could (2017, pp. 124–126). At the same time, policy-makers have the option to ignore scientific consensus in their decision-making. As a result, elective modernism steers clearer of technocracy than Douglas’ proposal (Collins and Evans 2017, pp. 127–128).
Finally, the crucial difference Collins and Evans bring forward is their commitment to the values of science and their protection from “diluting” (2017, p. 126) by the general social values (which is especially problematic if there are concerns about the current values of society). As a consequence, there is no expectation for scientists to be good citizens, as Douglas requires. In particular, it means that one should not expect experts to consider social values in their activities as experts:
[O]ne cannot and should not expect every scientist who finds themselves working in some small core-set whose work has social implications to consult citizens and stakeholders about the appropriate balance of risks and benefits before submitting a paper for publication. (Collins and Evans 2017, p. 124)
When Collins and Evans first discussed the place of the earlier science studies in the Third Wave (2002), they stated that showing the role of politics in science does not mean that politics should be pursued explicitly there. I suggest that this conviction underlies their response to the argument from inductive risk. This argument may be said to clarify the operation of “intrinsic” (Collins and Evans 2002, p. 245) politics in science. As a matter of principle, elective modernism chooses to keep it intrinsic—as they explain,
Intrinsic politics means that experts try to resolve their differences whilst adhering to the values we have just described. This means disputes are resolved by reference to observation and theory and without resorting to arguments about the consequences or desirability of particular outcomes; to do this would be to make the political concerns extrinsic and to stop acting “scientifically.” (Collins and Evans 2017, p. 60; emphasis in the original)
Even though the paragraph above is not written in response to Douglas’ argument, “resorting to arguments about the consequences … of particular outcomes” may be taken to characterize the argument from inductive risk as well. According to the argument from inductive risk, it is precisely the reflection on consequences of possible errors that should shape epistemic decisions made throughout the research process.
Collins and Evans conclude that elective modernism agrees with Douglas’ about the inevitable presence of values in science but offers a different, and better, approach for organizing expert advice in the light of that. In what follows, I aim to show how the argument from inductive risk can be of greater relevance for thinking about elective modernism.
As described earlier, elective modernism treats experts as non-representative of non-experts; the lack of relevant similarity is a part of this non-representativeness. In section 3, I argued that there is an important sense in which experts can resemble (some) non-experts and be representative of them—the sense of representation as resemblance on the basis of shared social perspectives. I suggest approaching the argument from inductive risk through the lens of this sense of representation. On the view I propose, the crucial contribution of the argument from inductive risk is the demonstration that certain value-laden choices are not only ubiquitous in science. Even more importantly, they can be fittingly described in terms of social perspectives that serve as the basis for resemblance between expert and (some) non-expert groups.
In section 3, I described an abstract scenario: if certain social perspectives can be demonstrated in decisions and choices of an expert community, non-experts who share these social perspectives can be said to be represented in this community on the basis of resemblance. I now suggest interpreting the previously described arguments exploring inductive risk as showing how this scenario is ubiquitous in research practice. In what follows, I describe Stegenga’s (2017) analysis of institutional inductive risk calculus in greater detail. I use it to show how the notions of social perspective and inductive risk are relevant for understanding the issue of representation in connection with expert communities.
As previously described, Stegenga discusses evaluation of new medicines in terms of balancing two risks—the risk of unwarranted rejections and the risk of unwarranted approvals. He uses this framework to analyze the regulatory practices of the Food and Drug Administration (FDA) in the United States. Stegenga argues that these practices tend to decrease the former risk at the cost of increasing the latter. At the core of Stegenga’s argument is the discussion of clinical trials, a precondition for a new medicine approval being the evidence from at least two phase 3 randomized controlled trials with a statistically significant positive result (2017, pp. 20–21). According to Stegenga (2017, pp. 23–28), there is considerable freedom in organizing clinical trials: setting the inclusion and exclusion criteria for participants, selecting the outcomes to measure and to analyze, deciding how long trials should last, selecting which data and analyses to publish afterwards etc. Stegenga argues that the ways in which this freedom is typically used make it more likely that the positive effects of a medicine are overestimated and the side effects are underestimated. In other words, the organization of clinical trials can be said to reflect the perspective that 1) underestimating the positive effect of a new medicine is the worse risk and decreasing it is worth increasing the risk of overestimating it; 2) overestimating the side effects of a new medicine is the worse risk and decreasing it is worth increasing the risk of underestimating it. To use the notion Stegenga employs, this is the perspective according to which the risk of unwarranted rejections is the greater of the two evils. And, according to Stegenga, the perspective just described is currently very common in the expert community that organizes clinical trials or accepts the results of these trials as evidence for evaluation of new medicines.
The perspective that sees unwarranted rejections as the greater problem may be further analyzed in terms of relevant value considerations. Stegenga mentions some of them briefly (see, e.g., Stegenga 2017, pp. 17, 22). For example, one may worry about the harms to patients who would not have access to a safe and effective medicine or the financial harms to the company that would not be allowed to sell the medicine and recuperate the cost of development. The harms caused by unwarranted approvals—the negative effects of unsafe or ineffective medicines on patients or the financial harms of paying for ineffective or unsafe medicines—are seen from the perspective shaped by these considerations as more tolerable in comparison.
Importantly, this social perspective and the related value considerations are not limited to experts. Various groups of non-experts may hold a similar perspective on the comparative badness of unwarranted rejections compared to unwarranted approvals.2 I suggest that these non-experts can be said to be represented in the research and regulatory expert community that works on testing and evaluation of new medicines. On the other hand, non-experts who have a different perspective—who consider unwarranted approvals the greater evil—are not currently well represented in this community of experts.
I suggest that in this particular case, representation for non-experts within the expert community on the basis of resemblance through shared social perspectives is patchy. Drawing on Stegenga’s (2017) analysis it is possible to say that the social perspective according to which new medicines require evidence-based evaluation is represented in the relevant expert community in quite a well-established way. It is explicitly required that a new medicine be evaluated by the FDA on the basis of evidence from phase 3 randomized controlled trials. At the same time, according to Stegenga, the organization of many clinical trials reflects the perspective according to which it is desirable to decrease the risk of unwarranted rejections (at the cost of an increased risk of unwarranted approvals). Again, this perspective may be said to be represented quite firmly—Stegenga describes such an approach to organizing clinical trials as typical. However, this perspective is not as explicit. Thus, one could say that some non-experts are represented in this expert community on the basis of a shared perspective quite permanently, others less so, and some not at all. It is plausible that this is typical for expert communities in general.
Thus, I do not claim that this form of representation in expert communities is sufficient for saying that non-experts are adequately represented in science- and technology-related decision-making. On the contrary, I have shown how it only realizes one sense of representation and, due to its patchiness and mostly hidden nature, does so with considerable limitations and potential unfairness.3
One could object to the claim about the representativeness of expert communities in this sense by pointing out that they only fit one out of Brown’s five dimensions of representation—in other words, they are non-representative overall. I suggest that such an objection is problematic if one follows the approach of distinguishing different senses of representation to begin with. As Brown argues, the recognition that representation includes different elements leads to the recognition that different institutions may realize different senses of representation. This is why a variety of options when it comes to representation of citizens is necessary (see, e.g., Brown 2009, p. 237). I agree with Brown (and, as noted before, Collins and Evans also agree with him) that enabling such a variety is important. At the same time, I suggest that it is also important to analyze how a specific institution is involved in representation even if this is limited to one sense of representation—as, according to my argument, it is the case with expert communities.
To conclude, in this section I used Stegenga’s analysis to show that the argument from inductive risk can be used to demonstrate how expert communities are a location where representation of (some) non-experts happens, when representation is understood as resemblance on the basis of shared social perspectives. Thus, I argued that taking the argument from inductive risk seriously should have an impact on how one thinks about expert communities in connection with non-experts’ representation. In the following section I describe what this means, in turn, for some of the institutions and practices proposed in elective modernism.
6. Considering Elective Modernism: Implications for Practices and Institutions
Elective modernism offers a vision for expertise in democratic society, including some proposals for specific institutions and principles of their operation. The aim of this section is to show that the argument I have developed makes it desirable to rethink some aspects of this vision. I do not suggest that it should be abandoned. Yet, I argue that ignoring the implications of the argument would make elective modernism less appealing. In the first part of this section I defend this claim. In the second part, I discuss how some specific proposals of elective modernism may be adjusted in the light of it.
I have argued so far that there is an understanding of representation on which experts are representative of (some) non-experts. This claim is compatible with the claim that experts are non-representative on other views of representation. Yet, I suggest that there is a danger in making a claim about non-representativeness without describing clearly the aspect of representation in question. The danger is that as a result, expert communities may be treated as non-representative black boxes. Since they are believed to represent their own expert communities only, the question of whom expert communities consist and what it may mean for representation may not receive sufficient attention.
I suggest that once it is recognized that experts can represent (some) non-experts on the basis of social perspectives, such black-boxing becomes problematic for someone who wants to avoid technocracy and to ensure more effective and fairer representation for non-experts in the political phase. According to the argument from inductive risk, social perspectives are not introduced when making political decisions on science- and technology-related issues—they are already involved in technical decisions in the conditions of inductive risk. Elective modernism accepts this as something shown by the earlier science studies. Yet, I suggest that elective modernism underestimates the consequences of that. If the choice is made to keep these decisions “intrinsic” in Collins and Evans’ terms, there are reasons to be concerned about its negative effects when products of the technical phase are brought for consideration by citizens in the political phase.
One aspect of the problem concerns the quality of deliberation. If it is plausible that being aware of the social perspectives that played a role in the technical phase could justifiably change the course or the outcome of deliberation in the political phase, leaving this information hidden can be said to be problematic. For example, on the basis of Stegenga’s (2017) analysis it is possible to suggest that the awareness of the relevant expert bodies’ inductive risk calculus could—and should—have an impact on political decisions related to new medicines. Failing to explicate this calculus therefore risks having a negative impact on the quality of these decisions. One could question the significance of this by pointing out that elective modernism already guarantees favorable conditions for decision-making in the political phase by stressing that decision-makers are always free to reject the experts’ consensus. Even so, I suggest that having access to relevant information about the social perspectives involved in the shaping of this consensus would put decision-makers in a better position for considering this choice.
Another aspect of the problem may be described in terms of fairness: the social perspectives that happen to be present in the relevant expert community could end up having an unfairly more powerful influence on the outcomes of decision-making compared to alternative social perspectives. For example, such a concern arises if all relevant social perspectives count equally in the political phase but some of them have already played a role in shaping the technical phase. Stegenga’s (2017) analysis provides an illustration of this possibility. It is plausible that some non-experts are concerned about unwarranted rejections and others about unwarranted approvals. However, it is the first of these perspectives that plays a role in shaping the technical phase of the testing and evaluation of new medicines as well. As a result, it could unfairly have a greater influence on the shaping of decision-making overall.
I suggest that the necessary first step for addressing both aspects of the problem is ensuring greater transparency about the representation of social perspectives in expert communities. Transparency about social perspectives involved in the choices, decisions, practices, and regulations of expert communities in the technical phase is relevant for effective and fair representation of non-experts in the political phase.
Depending on the case, further steps may be desirable. It is conceivable that in some cases there may be no alternative social perspective—the social perspective identified in the practices of an expert community is held by everyone. More likely, in many cases the realization that there are alternative perspectives may require decisions about what institutions and procedures are needed to give them fairer representation. For example, Stegenga (2017) makes a case for more demanding regulation of new medicines—thus proposing to give the perspective concerned with unwarranted approvals a greater prominence. Stegenga discusses some of the options for doing so, with a focus on regulatory changes—stricter requirements for clinical trials so that they are less likely to overestimate benefits and underestimate harms (2017, pp. 28–31). In other cases, the response may focus on the political phase of decision-making instead.
Thus, the response to a clearer of view of social perspectives in a given expert community is likely to be case-specific. In all cases, however, increasing transparency about such perspectives remains crucial. In the remainder of this section I discuss what recognizing the desirability of transparency means for several aspects of elective modernism.
First, this recognition creates some tension within elective modernism’s view on the appropriate functioning of expert communities. On the one hand, a greater clarity about the social perspectives involved may be enabled simply by the wider inclusion of non-certified experts that elective modernism recommends. As Evans and Plows point out, “It is at least arguable that including a wider range of experts within the technical phase will actually promote critical reflection about hidden values” (2007, p. 834). On the other hand, as previously noted, elective modernism prescribes to keep politics intrinsic in the technical phase. Relatedly, when discussing Douglas’ proposals, Collins and Evans stress that experts as experts do not, and should not, engage in discussion of social values or consequences of their actions with non-experts. It is conceivable that elective modernism aims to discourage an increase in the influence of social values on experts’ activities while allowing experts to reflect and to explicate social values already present in their work. However, this possibility appears to sit at least somewhat uneasily with elective modernism’ principle of preventing the politics of the technical phase turning extrinsic. Accordingly, I suggest that taking the desirability of transparency seriously means loosening this principle without abandoning it—in particular, so that experts will be expected to reflect openly on social values (or social perspectives) involved in their work.
Second, accepting the requirement of transparency supports certain changes in the tasks set for Owls. These, as described by Collins and Evans (2017, pp. 84–95), focus on identifying “the substance and the strength” (2017, p. 85) of the expert consensus on an issue. Due to the expertise they possess, the Owls are also in an excellent position to identify the social perspectives involved in the consensus of a specific expert community. However, while explicating such social perspectives may be compatible with the Owls’ mandate, it is not described explicitly as a part of it. I suggest that the commitment to transparency requires making this explication a part of what the Owls are expected to do.
In addition to that, since the Owls are an expert community as well, the requirement of transparency about inductive risk and relevant social perspectives also applies to them. Therefore, the Owls need to be transparent about the social perspectives reflected in their own ways of dealing with inductive risk. Again, this recommendation creates a certain tension with some expectations about experts’ practices in elective modernism I discussed earlier. I suggest that similar considerations apply.
Finally, a greater transparency about social perspectives may make expert communities more visible as a location where (some) of non-experts are in one sense represented. Brown describes such a possibility in the case of citizen juries and other deliberative forums: inclusion of perspectives in these forums “may also evoke a sense of being symbolically represented among citizens who imaginatively identify with those perspectives” (2009, p. 237). A similar process may be possible in the case of expert communities, once potential constituents who hold certain social perspectives become aware of the presence of the same perspectives in an expert community.4 I suggest that this possibility for self-identification may be compatible with elective modernism as long as it is acknowledged that elective modernism sets certain limitations for what follows from it. Since the inclusion to expert communities is based on expertise, there is no option to increase diversity of perspectives within these communities by inviting non-experts. In other words, in elective modernism expert communities may become more visible in connection with the issues of non-experts’ representation but there is no way to change these communities from the point of view of the perspectives represented other than the inclusion of previously unrecognized experts.
Such increased visibility is associated with certain risks—most importantly, the possibility that those who do not self-identify with the social perspectives in question may trust the particular expert community less. (Although it is possible to suggest, as Douglas does in the argument I describe in the next section, that transparency may ultimately be better for trust.) Nevertheless, recognizing expert communities as a location for (some) non-experts’ representation is essential for having a fuller picture of different aspects of non-experts’ representation and for enabling better decision-making in the political phase.
To conclude, in this section I argued that taking the argument from inductive risk seriously should influence how one envisages several aspects of the operation of expert communities in elective modernism. Before concluding the paper, in the next section I briefly consider what this discussion of elective modernism may in turn add to the discussion of inductive risk.
7. Another Look at the Argument from Inductive Risk: Why to be Transparent
In discussing inductive risk, there are several possible justifications for greater transparency about values (or social perspectives) involved in judgments under inductive risk. One argument for open consideration of the values involved was already made by Rudner:
But for the scientist to close his eyes to the fact that scientific method intrinsically requires the making of value decisions, for him to push out of his consciousness the fact that he does make them, can in no way bring him closer to the ideal of objectivity. To refuse to pay attention to the value decisions which must be made, to make them intuitively, unconsciously, haphazardly, is to leave an essential aspect of scientific method scientifically out of control. (Rudner 1953, p. 6, italics in the original)
As discussed earlier when summarizing Douglas’ argument, Douglas proposes another justification for openness about values: doing so is helpful for dealing with the moral responsibility that researchers carry. Transparency can help careful consideration of the consequences of possible errors, which is required for acting responsibly.
Douglas (2009, pp. 148–154) also adds a further consideration to support the necessity of explicit discussion of the values involved. She argues that doing so may help trustworthiness of science and its ability to fulfil its social functions such as scientific advice. Explicating values may help to make controversies more tractable, as it allows one to distinguish disagreements about facts from disagreements about values, to specify steps that are required for the resolution, and to identify instances of junk science where values take the place of evidence.
To these three justifications (from improving objectivity, moral responsibility, and trustworthiness), this paper adds the fourth—improving representation in technological decision-making for non-experts. For that, as I argued in the previous section, transparency about social perspectives involved in the practices and decisions of expert community is important. An advantage of this justification is that it is relevant for approaches that stress the commitment to democracy (as elective modernism does) even if they reject the notion of the moral responsibility of individual researchers (as elective modernism does).
Discussions about the desirability of value transparency and possibilities and limitations of achieving it in practice are a part of ongoing discussion of inductive risk (see, e.g., Elliott and Richards 2017, pp. 272–273). I suggest that considering issues related to representation of non-expert citizens in technological decision-making—and elective modernism provides a good starting point for considering them—can add another argument for the desirability of greater transparency.
To conclude, I used the analyses of elements of democratic representation and inductive risk to discuss elective modernism’s approach to expert communities and representation of non-experts in science- and technology-related decision-making. I suggested that in discussing whether experts are representative of non-experts, it is important to distinguish different meanings of representation. In particular, some social groups may be said to be represented within expert communities when representation is understood as resemblance on the basis of shared social perspectives. The argument from inductive risk shows, in turn, that decisions and choices that can be fittingly characterized in terms of social perspectives constitute a large and unavoidable element in experts’ epistemic activities. I then argued that the acknowledgement of the sense in which experts may already be representative of (some) non-experts supports the recommendation of greater transparency about relevant perspectives in expert communities in order to enable better representation for non-experts outside of expert communities. Finally, I described how greater transparency means certain adjustments for some of the practices and institutions proposed in elective modernism.
In presenting the argument, I intended to demonstrate how work in philosophy of science may be of a greater impact for elective modernism than Collins and Evans themselves acknowledge in their responses to Brown and Douglas. At the same time, as the penultimate section demonstrated, analyzing elective modernism can be helpful for advancing reflection on the topic of transparency in discussions of inductive risk.
I am grateful to the anonymous reviewer who attracted my attention to the necessity to address this issue.
Understanding what social groups hold specific social perspectives on a particular issue and how these perspectives relate to major categories such as race, gender, or disability is an important question both in this case and for discussing social perspectives in general. In this particular case, it is plausible that persons belonging to the same gender (as one example) may hold different perspectives and that persons of different genders may hold the same perspective. As mentioned earlier, such lack of one-to-one relation is something Brown stresses when discussing social perspectives. Answering this question, however, requires empirical work that I am not able to undertake as a part of this paper.
The absence of representation for a perspective reflecting valid concerns of some citizens is one general reason to consider the situation unfair. It is plausible that an analysis of connections between social perspectives, values, and wider social categories may show more specific reasons to think of unfairness. For example, according to Stegenga, participants chosen for clinical trials tend to be younger and healthier and this may mean that typical (older and less healthy) patients experience fewer benefits and more harms than a clinical trial demonstrated (2017, p. 24). A perspective concerned with overestimation of benefits and underestimation of harms may thus reflect a concern that the current approach to organizing clinical trials is ageist. However, as I have noted previously, a necessarily empirical exploration of such connections remains beyond the scope of the paper.
I am grateful to the anonymous reviewer who prompted me to discuss this possibility.
This study was supported by the Estonian Research Council (PUT 732 and PRG 462), the Estonian Ministry of Education and Research (IUT 20-5), and by the European Union European Regional Development Fund (Centre of Excellence in Estonian Studies). I am grateful to the anonymous reviewers for helpful feedback. Earlier versions of the paper were presented at the conference of the Baltic Association of the History and Philosophy of Science in 2017, at a seminar hosted by TINT (Helsinki) in 2017, and at the conference “From φ-science to practical realism: an international conference in honor of Rein Vihalemm (1938–2015)” in 2019. I am grateful to the audiences for helpful feedback.