While Kevin Elliott's book does not actually answer the question posed in the title, it does give philosophers of science plenty to think about. In particular, the book poses crucial questions concerning how the social practices of science should function in a democratic society. The book does not give ironclad answers to the questions it poses, but it will provide plenty of grist to those willing to take up the challenge.
The book's title and central case study derive from the issue of hormesis, the theory that at low doses, some (or perhaps most?) toxic chemicals reduce effects to rates below what a control organism would experience. If this is true, then for some chemicals, low doses might actually be better for us than no exposure at all. But, as Elliott details the case for hormesis and examines in detail the evidence provided for it, he shows how the uncontroversial claim that some chemicals exhibit hormetic effects gets expanded to most chemicals, which then gets expanded to the idea that hormesis is the dominant dose-response model, which is further expanded to claims about how to reform chemical regulation. Elliott nicely disambiguates these claims and, by examining the evidence for them, shows how different judgments at different points in the performance and assessment of the relevant studies lead to different conclusions about what to make of the phenomena.
What to make of this pervasive problem of judgment? Elliott turns to the values-in-science literature, describing evenhandedly the current terrain of that debate. He then makes the limited and readily defensible claim that in cases like the hormesis case, regardless of what you make of the distinction between epistemic and nonepistemic values and given either arguments from general underdetermination or arguments concerning the significance of error, social and ethical values should not be excluded at any of the four key points of judgment he uses to describe his case: (1) the choice of projects and the design of studies, (2) the development of terminology and categories, (3) the interpretation and evaluation of studies, and (4) the application of research. Elliott's consideration of not just the usual locations of 1, 3, and 4 but also 2, the way in which language choices shape our knowledge, is a welcome addition to the values-in-science debate, as considerations of language choices have been mostly the province of the philosophy of social science. Elliott shows that philosophers concerned with the natural sciences need to think about these choices too.
But Elliott's analysis of values and hormesis is only a jumping off point, from which Elliott turns to terrain not usually tackled by philosophers of science: How should we structure the social practices of science, given this lesson on values in science? He addresses three questions/possibilities for social institutional reform in science: (1) how to deal with conflicts of interest; (2) how to use deliberative processes, including deliberation that could incorporate members of the public, in science advising; and (3) what should constitute ethical behavior from our experts. These are crucial questions for contemporary scientific practice; questions with which social epistemology has yet to fully grapple.
Although Elliott is asking excellent questions, his answers are ultimately less satisfying. This is in part because his account of values in science is based on such a wide argumentative base that it results in a weak norm for such values: social and ethical values should not be excluded. But this watered-down claim leaves little basis for saying when such values might impair judgment as opposed to properly inform judgment. And without this normative scaffolding, his assessments of values in science do not have the teeth they might.
Consider the question of conflict of interest. The underlying worry with such conflicts is that they will impair a scientist's judgment (or generate deceptive behavior), but Elliott cannot say clearly when a value commitment impairs a judgment. Instead, he reviews the disturbing evidence that funding sources distort study outcomes (83–88) and defines conflict narrowly, in terms of massive amounts of funding. Whether such a narrow definition captures the underlying epistemic worry is not clear. And even with this narrow definition, he argues that neither mere disclosure nor management of such conflicts will suffice (92–102). Divestiture is taken off the table as impractical (although sometimes he thinks it might be pursued; 91–92). More “independent” funding is called for (although what is to count as independent is not clear). Trial registries might remediate one source of distortion in the evidential field (102–3). But it is in deliberative practices that Elliott places the most emphasis as a solution for conflict of interest (104–5).
His discussion of the potential for deliberation, however, undermines its potential usefulness as a solution to conflict-of-interest problems. In his arguments for deliberation, Elliott is focused on not just deliberations among scientists but deliberative forums (e.g., consensus conference, citizen juries, and citizen advisory panels) that include members of the general public (110–15). Elliott realizes that the specific design of a deliberative process is crucial, so rather than argue for a one-size-fits-all approach, he outlines a process for thinking through which deliberative processes might be appropriate in any given context. This is useful guidance, but it presumes actors not corrupted by conflicts of interest and experts behaving ethically. This is made clear with Elliott's acknowledgment, later in the book when he applies his analysis to the cases of multiple chemical sensitivity syndrome and endocrine disruption, that deliberative processes can be distorted by misframing and conflicts of interest. He describes a case in which a deliberative process failed, in part because of the participation of scientists perceived to have a conflict of interest (169) and in part because of disciplinary disputes among scientists. It seems that deliberation cannot be a solution to conflict-of-interest issues but rather depends (in part) on their already being solved.
For the problem of what ethical guidelines we should have for experts, or an ethics of expertise, Elliott draws an analogy with informed consent standards in biomedical ethics. There is something intuitive about this analogy: presumably we want our decision makers, be they regulators, politicians, or the public, to be able to choose with informed autonomy. But the analogy with the doctor-patient relationship is strained in important ways: when an expert speaks, it is often unclear who the audience is (unlike a doctor speaking to a patient), so how one tailors the information so that it gets across and bolsters rather than hampers autonomous decision making is less than clear. Doctors can also presumably assess the extent to which the patient is understanding the key points, whereas experts, without being in contact with their audience, have a harder time checking on the comprehension of their audience(s). Perhaps clear deliberative contexts where experts can be placed in close contact with a well-defined audience could help, but Elliott does not explore this possibility.
Indeed, this is the key disappointment of Elliott's examination of these three areas: he does not draw the connections between them. Some are just waiting to be drawn, and others reveal apparent tensions in his analysis. For example, while he argues that disclosure of conflicts of interest is ineffective in chapter 4, he argues that it is a crucial part of ethical behavior for an expert in chapter 6. Where deliberation is presented initially as a solution to conflict-of-interest problems, it seems rather to depend on their solution. The extent to which having a well-defined deliberative context might alter the ethics of expertise remains an open (and interesting) question. Finally, one wonders whether deliberative processes are a resource for addressing the (mis)framing issues that are part of the challenge of the ethics of expertise or whether deliberation itself is so dependent on framing issues that it is of little assistance (116, 146). Indeed, it may be that issues of framing cannot be addressed without stronger normativity in the analysis of the role for values.
Despite these flaws, the book is a welcome addition to the field. It is one of the few (the only?) books that attempts to work across debates concerning values in science and social epistemology, to develop insights not just about science in policy but also policy for science. This kind of finely textured analysis of the specific social institutions around scientific practice is desperately needed. Elliott's book opens this terrain up for examination, debate, and hopefully, eventual improvement.
