Political Philosophy of Science

Scientists often aim to produce knowledge that benefits humanity as a whole, but the conduct and dissemination of some kinds of scientific research can also impose catastrophic risks to human health, the environment, or social justice. A number of my current research projects, including a major grant from the NSF, consider the responsibilities of scientists given that (i) science is a collective enterprise, (ii) the conduct of science often imposes different risks and benefits on different populations, (ii) there is pervasise disagreement about the value of these risks and benefits (even amongst people similarly situated).

• Creating a Value Literate Culture in Science (w/ Scott Tanona & J.T Laverty)

  NSF Funded Project (2019-2022)

  Abstract

  This project explores an underexamined cultural aspect of Responsible Conduct in Research (RCR) -- the Goals and Values of Science (GVS) that inform everyday scientific practice. Examples of such practices include identifying a research topic, designing a research plan that will yield robust evidence, and selecting and recruiting research participants. By contrast, RCR training courses typically focus on narrow, rule-based prohibitions, an approach which has been shown to have limited scope and impact. This project is important because it moves beyond the limits of institutionalized, rule-based ethical proscriptions and into the heart of what scientists think and value. Moving in this direction has potential to open a new line of research between philosophy of science with educational research on RCR.

• The Social Risks of Science (w/ Scott Tanona)

  Hasting Center Report (2020), 50(6): 27-38.

  Abstract

  Many instances of scientific research impose risks, not just on participants and scientists, but also on third-parties. This class of risks – what we might term “social risks” – unifies a range of problems previously treated as distinct phenomena, including: so-called bystander risks, biosafety concerns arising from “gain of function” research, the misuse of the results of “dual-use” research, and the harm caused by inductive risks. The standard approach to these problems has been to extend two familiar principles from human subjects research regulations - “social benefit” and “informed consent”. We argue, however, that these moral principles will be difficult to satisfy in the context of widely-distributed social risks about which affected parties may reasonably disagree. We propose that framing these risks as political, rather than moral, problems, may offer another way. Instead of thinking about these risks as a problem of interpersonal consent, we may formulate solutions to the social benefit and consent requirements which are: (i) practically implementable for very large populations, (ii) sensitive to the fact of disagreement, and (iii) focused on the role of institutions and incentives.

• Agent-Based Models of Restrictions on Dual-Use Research (w/ Elliott Wagner)

  British Journal of Philosophy of Science (2019)

  Abstract

  Scientific research that could cause grave harm, either through accident or intentional malevolence, is known as dual-use research. Recent high-profile cases of dual-use research in the life sciences have led to debate about the extent to which restrictions on the conduct and dissemination of such research may impede scientific progress. We adapt formal models of scientific networks to systematically explore the effects that different regulatory schemes may have on a community’s ability to learn about the world. Our results suggest that, contrary to common wisdom, some restrictions on the conduct and dissemination of dual-use research do not inhibit scientific progress and may actually aid communities in learning.

Algorithmic Decision-Making

Algorithmic decision-making has likely improved the accuracy and efficiency of risk judgements. They are cheaper, faster and less prone to cognitive biases than human decision makers. Yet despite this, there is increasing concern that these algorithms may be systematically biased against certain classes of people. In part because of these concerns, the developers of machine algorithms have started to grapple with the complex tradeoffs involved in detecting and eliminating bias in their creations. Unfortunately, whilst computer scientists have made some strides in identifying appropriate measures of bias, there has been little engagement by philosophers with the precise wrongs done by biased algorithms. I am currently working on remedying the gap between statistical and philosophical understandings of bias and discrimination.

• Exploring Creative Design at the Human-technology Frontier Through the Emerging Artist-Technologist Occupation (Equity team co-lead)

  NSF Funded Project (2020-2022)

  Abstract

  The project involves a diverse group of researchers and practitioners with relevant expertise in the arts, AI, AR/VR, engineering, learning sciences, education, business, entrepreneurship, career education, ethics, and disability in an exploratory study of how specific applications of AR/VR and AI may affect artist-technologists? creative process, professional identity, overall work experience, preparation, and related equity and inclusion issues. The initial focus of the project is on musicians and composers, and will expand to other artistic endeavors. At the core of the study are exploratory conversations with key stakeholders (conducted in the tradition of customer discovery interviews promoted by NSF's Innovation Corps program), followed by three task forces drawing implications and identifying issues worthy of further research for enhancing future artist-technologists? design process at the human-technology frontier (future technology focus), effectively preparing future artist-technologists to make best use of AI and AR/VR in their work (future workers focus), and increasing access to this occupation by underrepresented populations (future work focus), respectively. A series of team learning experiences and touch-point events informs this collaborative work, while also supporting the project team in developing the needed cross-disciplinary expertise and strategies.

• Measuring Fairness in an Unfair World

  Proceedings of AAAI-ACM Conference on Artificial Intelligence, Ethics, and Society (AIES’20) (2020)

  PDF
  Abstract

  Computer scientists have made great strides in characterizing different measures of algorithmic fairness, and showing that certain measures of fairness cannot be jointly satisfied. In this paper, I argue that the three most popular families of measures – unconditional independence, target-conditional independence and classification-conditional independence – make assumptions that are unsustainable in the context of an unjust world. I begin by introducing the measures and the implicit idealizations they make about the underlying causal structure of the contexts in which they are deployed. I then discuss how these idealizations fall apart in the context of historical injustice, ongoing unmodeled oppression, and the permissibility of using sensitive attributes to rectify injustice. In the final section, I suggest an alternative framework for measuring fairness in the context of existing injustice: distributive fairness.

• Measuring the Biases that Matter: The Ethical and Causal Foundations for Measures of Fairness in Algorithms (w/ Bruce Glymour)

  Proceedings of the ACM Conference on Fairness, Accountability and Transparency (ACM-FAT'19) (2019)

  PDF
  Abstract

  Measures of algorithmic bias can be roughly classified into four categories, separated by the conditional independencies to which they are sensitive. First, measures of “procedural bias” diagnose bias when the score returned by an algorithm is probabilistically dependent on a ‘sensitive’ class variable, like race or sex. Second, measures of “outcome bias” are sensitive to any conditional dependence between ‘sensitive’ variables and the outcome for each subject, e.g. parole granted or loan denied. Third, measures of “behavior-relative error bias” are sensitive to any probabilistic dependence between class variables and the algorithmic score, conditional on target behaviors, like recidivism or loan default. Fourth, measures of “score-relative error bias” are sensitive to a probabilistic dependency between class and behavior, conditional on score. Several recent discussions have demonstrated a tradeoff between these different measures of algorithmic bias, and at least one recent paper has suggested conditions under which tradeoffs may be minimized. In this paper we use the machinery of causal graphical models to show that, under standard assumptions, the underlying causal and definitional relations among variables forces some tradeoffs. We delineate a number of normative considerations that are encoded in different measures of bias, with reference to the philosophical literature on the wrongfulness of disparate treatment and disparate impact. While both kinds of error bias are nominally motivated by concern to avoid disparate impact, we argue that consideration of their behavior given alternative causal structures shows that these measures are better understood as complicated and unreliable measures of procedural biases (i.e. disparate treatment). Moreover, while procedural bias is indicative of disparate treatment, we show that the particular measure of procedural bias we ought to adopt is dependent on the specific account of the wrongfulness of disparate treatment. Finally, given the failures of condition-relative measures of error bias, we suggest that error bias proper is best measured unconditionally by score based measures of accuracy, such as the Brier score.

Health Security and Emergencies

The concept of "health security" has a central role in contemporary global health governance, and yet its precise content and value is under-specified. Of particular concern is whether the value of health security justifies the allocation of resources to protect populations from "novel" risks to their health (such as fast-moving zoonoses, pandemic influenza and bioterrorism), as opposed to alleviating ongoing, serious and well-characterised threats (such as obesity, HIV/AIDS or antibiotic resistance). I am currently extending my conceptual work on health security to a number of applied problems, including: (i) the criteria for the declaration of a Public Health Emergency, and (ii) the justification for universal catstrophic health insurance.

• "Healthcare and Planning"

  Draft in Progress

  Abstract

  Reliable access to healthcare enables individuals to rationally form and revise their life plans, and this provides an additional reason to ensure universal access to catastrophic healthcare. I argue for this conclusion in four parts. First, I introduce the concept of security in the context of individual health: roughly understood as the probability of enjoying at least that many years of health-adjusted life. Second, I argue that the security of initial increments of healthy life (i.e. additional years of minimally decent health) is centrally important to our capacity to rationally form and pursue our life plans.

• Health Security and Risk Aversion

  Bioethics (2016), 30(7): 479-489

  PDF
  Abstract

  Health security has become a popular way of justifying efforts to control catastrophic threats to public health. Unfortunately, there has been little analysis of the concept of health security nor of the relationship between health security and other potential aims of public health policy. In this paper I develop an account of health security as an aversion to risky policy options. I explore three reasons for thinking risk avoidance is a distinctly worthwhile aim of public health policy: (i) that security is intrinsically valuable, (ii) that it is necessary for social planning and (iii) that it is an appropriate response to decision-making in contexts of very limited information. Striking the right balance between securing and maximizing population health thus requires a substantive, and hitherto unrecognized, value judgment. Finally, I critically evaluate the current health security agenda in light of this new account of the concept and its relationship to the other aims of public health policy.

• Obesity, Liberty, and Public Health Emergencies (w/ Angus Dawson & Heather Draper)

  The Hastings Center Report (2014), 44(6): 26-35

  Abstract

  Widespread obesity poses a serious challenge to health outcomes in the developed world and is a growing problem in the developing world. There has been a raft of proposals to combat the challenge of obesity, including restrictions on the nature of food advertising, the content of prepared meals, and the size of sodas; taxes on saturated fat and on calories; and mandated “healthy‐options” on restaurant menus. Many of these interventions seem to have a greater impact on rates of obesity than does simply providing information about health risks and healthier lifestyles. The more interventionist policy options have, however, been implemented only slowly, in large part because of criticisms that they are unjustified infringements on the liberty of consumers. Food industry groups, free‐market think tanks, and the popular press regard measures that incentivize or penalize particular food and lifestyle choices as unjustifiable state regulation of purely self‐regarding behavior.
  To counteract the liberty‐oriented position, those who favor a more interventionist role for the state have recently argued for labeling obesity as a public health emergency. By labeling obesity as a public health emergency, policy‐makers could override concerns about individual liberty in order to pursue more interventionist policies designed to guide consumer choices toward healthier lifestyles. In this paper, we argue that, contrary to initial appearances, obesity possesses some of the morally relevant features of public health emergencies, though we do not argue that it actually constitutes one.

Climate Change and the Environment

• Justifying a Consumer Right to Know

  Draft in Progress

  Abstract

  In this paper, I analyze four possible justifications for a consumer right to know. I begin with an analysis of the right to know, showing that it implies that sellers have a positive duty to verify and actively disclose the presence of the relevant property within their products. I then survey four candidate justifications for this duty: (i) promoting consumer autonomy, (ii) promoting the wellbeing of third-parties, (iii) respecting consumer autonomy, and (iv) promoting consumer autonomy. I identify the general features of the properties each justification would require sellers to disclose, and the manner of disclosure which would best serve the relevant values. I also describe difficulties for each justification, showing that active disclosure of a property may often undermine the values which purportedly justify the consumer’s right to know. Ultimately, I argue that the most promising reason to require sellers to actively disclose a property is if doing so would promote consumer wellbeing.

• Against the Autonomy Argument for Mandatory GMO Labelling

  Public Affairs Quarterly (2018), 32(2): 85-117

  Abstract

  Many argue that consumers possess a “right to know” when products contain ingredients derived from genetically modified organisms, on the grounds that it would protect consumer autonomy. In this paper, I critically evaluate that claim. I begin by providing a version of the “consumer autonomy” argument, showing that its success relies on ambiguities in the notion of autonomy. I then distinguish four approaches to autonomy and articulate the circumstances under which they would support active disclosure of a product property. I argue that none of these conceptions would support active disclosure of the presence of ingredients derived from genetically modified organisms.

• Climate-related insecurity, loss and damage

  Ethics, Policy & Environment (2017), 20(2): 184-194

  PDF
  Abstract

  The harms of climate change are deeply uncertain. Though climate change will render most individuals more vulnerable to harm, many individuals will not actually suffer climate-related harms (or will suffer only minor harms). In this paper, I argue that vulnerability to harms is itself a harm, because it undermines our enjoyment of the good of security. After some brief remarks on the concept of security, I give three reasons for thinking that depriving an individual of the security of basic goods harms them: (i) it has a strong contingent connection to fear and anxiety, (ii) it directly undermines their ability to make reasonable plans, and (iii) security may reasonably be valued as an end in itself. This suggests that one of our goals in determining climate policy ought to be mechanisms which ensure the security of basic goods.

The Concept and Value of Security

Security is a very politically powerful word, yet there is a dearth of philosophical investigation into the concept of security, and its value. In my research I have given an account of what it means to enjoy a good securely, and why this might be a weighty consideration in moral decision-making. I argue that the prudential value of security is explained by both the affective dimension of security and the effect of grave risk on our capacity to form and pursue our plans. I argue that this can help us make sense of the rationality of risk aversion, non-maximising consumption, and prioritisation of resources to emergencies.

• "Planning with Security"

  Draft in Progress

  Abstract

  In this paper I argue that the security of basic goods enables agents like us to engage in rational deliberation. I define the security of a good as the probability of realizing that good regardless of which act an agent performs within their set of alternatives. If a good is sufficiently secure then we are practically justified in treating that good as a background condition in our deliberation. This is useful, in so far as it allows us to treat that good as independent of our present deliberation, but it is especially useful if that good is a necessary means to some future course of action. Possessing the practical justification for treating some goods and future actions as independent of our current actions enables agents like us to engage in approximately rational instrumental reasoning.

• The Contribution of Security to Well-being
  

  Journal of Ethics & Social Philosophy (forthcoming)

  Abstract

  Do unknown and unrealized risks of harm diminish an individual’s well-being? The traditional answer is no: that the security of prudential goods benefits an individual only instrumentally or by virtue of a subjective affect of security. Recent work has argued, however, that the security of prudential goods non-instrumentally benefits an individual regardless of whether or not they enjoy subjective security. In this paper, I critically examine three claims about the way in which unknown and unrealized risks of harm might diminish individual well-being: (i) it frustrates a desire to be secure, (ii) it frustrates the enjoyment of modally-robust goods, and (iii) it undermines the ability to make reasonable plans. Ultimately, I argue that all three of these hypotheses are mistaken, but that they deepen our understanding of the ways in which subjective security is an important constituent of individual well-being.

• The Concept of Security, Liberty, Fear and the State
  

  Security: Dialogue across disciplines (2015), ed. P. Bourbeau. (Cambridge University Press): 22-44

  PDF
  Abstract

  In this chapter I seek to reignite philosophical interest in security by uncovering some of the ways in which the concept has been both understood and misunderstood. I begin by exploring the scarce historical understandings of security within the Western philosophical canon, from the Epicureans through Hobbes and on to contemporary political philosophy, identifying the key themes that arise within the literature. I then provide an account of the structure of the concept of security, which lays bare its relationship to contemporary debates on the distinction between “natural” and “social” threats, on the political significance of fear, and on the nature of rights. Finally, I identify four key problems in moral and political philosophy – the balance between state and individual security, between liberty and security; the status of moral principles during emergencies; and the trade-off between privacy and public safety – where a finer grained understanding of security can benefit normative theorizing.

• The Concept of Security
  

  Ethics and Security Aspects of Infectious Disease Control: Interdisciplinary Perspectives (2012), ed. M.J. Selgelid & C. Enemark. (Ashgate): 7-26

  PDF
  Abstract

  This chapter provides a philosophical analysis of the different meanings of “security” and, by so doing, identifies some key features of the concept of security. I begin by establishing a number of qualities which this chapter’s conceptual analysis should ideally possess. I then make an important distinction between security as a practice and security as a state of being, and argue that more attention should be paid to the latter if our goal is to interrogate the justifiability of using security practices in the context of infectious disease emergencies. The latter half of the chapter investigates three common features of contemporary accounts of security: (1) the referent object, (2) the conditions that object must satisfy to be secure, and (3) the distinction between the objective and subjective realisation of those conditions. I argue that accounts of the meaning of “security” identify a referent object and a set of conditions which must be reliable for that referent. I conclude by suggesting that a deeper engagement by moral philosophers with the concept and value of security (as a state of being) is required if we are to get very far in evaluating the justifiability of treating infectious disease emergencies as security issues.