This article presents a fairness principle for evaluating decision-making based on predictions: a decision rule is unfair when the individuals directly impacted by the decisions who are equal with respect to the features that justify inequalities in outcomes do not have the same statistical prospects of being benefited or harmed by them, irrespective of their socially salient morally arbitrary traits. The principle can be used to evaluate prediction-based decision-making from the point of view of a wide range of antecedently specified (...) substantive views about justice in outcome distributions. (shrink)

With their intensive use of data to classify and price risk, insurers have often been confronted with data-related issues of fairness and discrimination. This paper provides a comparative review of discrimination issues raised by traditional statistics versus machine learning in the context of insurance. We first examine historical contestations of insurance classification, showing that it was organized along three types of bias: pure stereotypes, non-causal correlations, or causal effects that a society chooses to protect against, are thus the main sources (...) of dispute. The lens of this typology then allows us to look anew at the potential biases in insurance pricing implied by big data and machine learning, showing that despite utopic claims, social stereotypes continue to plague data, thus threaten to unconsciously reproduce these discriminations in insurance. To counter these effects, algorithmic fairness attempts to define mathematical indicators of non-bias. We argue that this may prove insufficient, since as it assumes the existence of specific protected groups, which could only be made visible through public debate and contestation. These are less likely if the right to explanation is realized through personalized algorithms, which could reinforce the individualized perception of the social that blocks rather than encourages collective mobilization. (shrink)

Journal of Political Philosophy, EarlyView.

Volume 20, Issue 2, February 2020, Page 20-22.

The article by MacKay and Saylor (2020) claims that the principle of fair subject selection yields conflicting imperatives (e.g. in the case of pregnant women) and should be understood as “a bundle of four distinct sub-principles” (i.e. fair inclusion, burden sharing, opportunity, distribution of third-party risks), each having conflicting normative recommendations (MacKay and Saylor 2020). The authors also offer guidance as to how we should navigate between subprinciples that may conflict with each other. The problem is a crucial one since (...) fair subject selection is one of the principles regulating clinical research that produces generalizable knowledge with the potential of improving people’s health. Therefore, there may be cases where the way in which participants are selected directly influences the generalizability (or its lack) of the clinically relevant knowledge and its value (if any) to different groups. In my commentary article, written from the philosophical perspective, I notice a number of interrelated problems which I believe have not been discussed thoroughly in the target article: (1) the precise way in which health care priority setting should influence the content of health research priority setting and fair inclusion principles; (2) the distinction between group and individual benefits and burdens from clinical research; (3) the reference class problem in medical research. (shrink)

Richard Arneson argues that Fair Equality of Opportunity should be rejected, since it is not only too weak and too strong, but also problematically meritocratic. The paper aims to defend FEO, and argues that it is not too weak, since, pace Arneson, it does apply to the problem of stunted ambition. The argument from meritocracy is shown to be based on a conflation of different senses of meritocracy. Finally, it is shown that FEO, correctly interpreted, gives intuitive answers to the (...) examples put forward to bolster the too strong charge. It is concluded that Arneson’s refutation of FEO fails. (shrink)

A frequent caveat in online dating profiles – “No fats, femmes, or Asians” – caused an LGBT activist to complain about the bias against Asians in the American gay community, which he called “racial looksism”. In response, he was asked that, if he himself would not date a fat person, why he should find others not dating Asians so upsetting. This response embodies a popular attitude that personal preferences or tastes are simply personal matters – they are not subject to (...) moral evaluation. In this paper, I argue, against this popular attitude, that a personal preference like racial looksism is indeed wrong. A preference like racial looksism is wrong because it is an overgeneralization that disrespects individuality by treating people as exchangeable tokens of one type, and such disrespect denies its objects appreciation that their dignity entitles them to. As it turns out, there is on my account a relevant moral difference between racial looksism and simple looksism. (shrink)

Violence risk assessment tools are increasingly used within criminal justice and forensic psychiatry, however there is little relevant, reliable and unbiased data regarding their predictive accuracy. We argue that such data are needed to (i) prevent excessive reliance on risk assessment scores, (ii) allow matching of different risk assessment tools to different contexts of application, (iii) protect against problematic forms of discrimination and stigmatisation, and (iv) ensure that contentious demographic variables are not prematurely removed from risk assessment tools.

Philosophical reflection on racial profiling tends to take one of two forms. The first sees it as an example of ‘statistical discrimination,’ (SD), raising the question of when, if ever, probabilistic generalisations about group behaviour or characteristics can be used to judge particular individuals.(Applbaum 2014; Harcourt 2004; Hellman, 2014; Risse and Zeckhauser 2004; Risse 2007; Lippert-Rasmussen 2006; Lippert-Rasmussen 2007; Lippert-Rasmussen 2014) . This approach treats racial profiling as one example amongst many others of a general problem in egalitarian political philosophy, (...) occasioned by the fact that treating people as equals does not always require, or permit, us to treat them the same. The second form is concerned with how racial profiling illuminates the nature, justification, and reproduction of hierarchies of power and privilege based on skin colour and morphology. This form of reflection on racial profiling is therefore less about the justification for judging people based on the characteristics of the group to which they (appear to) belong, and more concerned with the specific ways in which the association of racialized minorities – and, in particular, black people – with crime, contributes to, and reflects, racial inequality, and oppression.(Kennedy 1998; Zack, 2015; Lever, 2005; Lever 2007). Both approaches to profiling have much to recommend them and, taken together, they form an essential component of the political philosophy of race. The statistical approach has the merits of linking racial profiling, as practice, to a body of other practices that generate and justify inequalities based on factors other than race, but it typically offers little by way of insight into the role of racial profiling itself in sustaining racial inequality and injustice. The racial construction approach, for obvious reasons, is rather better at the latter task, but its insights tend to come at the price of a broader understanding of the ways in which inequality is reproduced and justified, or of the ethical dilemmas raised by our competing claims to security. As we will see, insights from both approaches can be synthesized to clarify what, if anything, is wrong with racial profiling and what broader conclusions for equality and security follow from the study of profiling. (shrink)

There are many objections to statistical discrimination in general and racial profiling in particular. One objection appeals to the idea that people have a right to be treated as individuals. Statistical discrimination violates this right because, presumably, it involves treating people simply on the basis of statistical facts about groups to which they belong while ignoring non-statistical evidence about them. While there is something to this objection—there are objectionable ways of treating others that seem aptly described as failing to treat (...) them as individuals—it needs to be articulated carefully. First, most people accept that many forms of statistical discrimination are morally unproblematic, let alone morally justified all things considered. Second, even treating people on the basis of putative non-statistical evidence relies on generalizations. Once we construe treating someone as an individual in a way that respects this fact, it becomes apparent: (1) that statistical discrimination is compatible with treating people as individuals, and (2) that one may fail to treat people as individuals even without engaging in statistical discrimination. Finally, there are situations involving the expression of messages of inclusion where we think it is good, morally speaking, that we are not treated as individuals. (shrink)

This chapter concerns the normativity of the concepts of paternalism and libertarian paternalism. The first concept is central in evaluating public health policy, but its meaning is controversial. The second concept is equally controversial and has received much attention recently. It may or may not shape the future evaluation of public health policy. In order to facilitate honest and fruitful debate, I consider three approaches to these concepts, in terms of their normativity. Concepts, I claim, may be considered nonnormative, normatively (...) charged, or normative in that they involve more complex relationships between values or duties. While the last approach is often best, other approaches may be appropriate depending on the context and purpose of discussion. The chapter’s conceptual investigation is illustrated by application to two public health policies: a tax on the consumption of fat and the encouragement of health-promoting food displays in restaurants and supermarkets. (shrink)

Here, we provide an ethical analysis of discrimination in private insurance to guide the application of non-discriminatory algorithms for risk prediction in the insurance context. This addresses the need for ethical guidance of data-science experts, business managers, and regulators, proposing a framework of moral reasoning behind the choice of fairness goals for prediction-based decisions in the insurance domain. The reference to private insurance as a business practice is essential in our approach, because the consequences of discrimination and predictive inaccuracy in (...) underwriting are different from those of using predictive algorithms in other sectors. Here we focus on the trade-off in the extent to which one can pursue indirect non-discrimination versus predictive accuracy. The moral assessment of this trade-off is related to the context of application—to the consequences of inaccurate risk predictions in the insurance domain. (shrink)

This article examines one kind of conscientious refusal: the refusal of healthcare professionals to treat sexual dysfunction in individuals with a history of sexual offending. According to what I call the orthodoxy, such refusal is invariably impermissible, whereas at least one other kind of conscientious refusal—refusal to offer abortion services—is not. I seek to put pressure on the orthodoxy by (1) motivating the view that either both kinds of conscientious refusal are permissible or neither is, and (2) critiquing two attempts (...) to buttress it. (shrink)

Discrimination is a central moral and legal concept. However, it is also a contested one. Particularly, accounts of the wrongness of discrimination often rely on controversial and particular assumptions. In this paper, I argue that a theory of discrimination that relies on premises that are very general (rather than unique to the concept of discrimination) and widely accepted provides a plausible (exhaustive) account of the concept of wrongful discrimination. According to the combined theory, wrongful discrimination consists of allocating a benefit (...) that is not supported by a morally significant fact (a valid reason), or in a way that involves distributive injustice, or both. (shrink)

The Oxford Handbook of Philosophy and Race provides up-to-date explanation and analyses by leading scholars of contemporary issues in African American philosophy and philosophy of race. These original essays encompass the major topics and approaches in this emerging philosophical subfield that supports demographic inclusion and diversity while at the same time strengthening the conceptual arsenal of social and political philosophy. Over the course of the volume's ten topic-based sections, ideas about race held by Locke, Hume, Kant, Hegel, and Nietzsche are (...) supplemented by suppressed thought from the African diaspora, early twentieth-century African American perspectives and Native-, Asian-, and Latin-, American views. The contributors bring philosophical analysis to bear on the status of racial divisions as categories of humanity in the biological sciences, as well as within contemporary criticism and conceptual analysis. Essays present the special applications of American philosophy and continental philosophy to ideas of race as methodological alternatives to more analytic approaches. As a collection of analyses and assessments of 'race' in the real world, the volume pays trenchant and relevant attention to historical and contemporary racism and what it means to say that 'race' and racial identities are socially constructed. The essays analyze contemporary social issues including the importance of racial difference and identity in education, public health, medicine, IQ and other standardized tests, and sports. Additionally, the essays consider the societal limitations and structures provided by public policy and law. As a critical theory, the volume compares the study of race to feminism. Historical and contemporary, academic and popular, racisms pertaining to male and female gender receive special consideration throughout the volume. While this comprehensive collection may have the effect of a textbook, each of the original essays is a fresh and authentic development of important present thought. (shrink)

Algorithmic predictions are promising for insurance companies to develop personalized risk models for determining premiums. In this context, issues of fairness, discrimination, and social injustice might arise: Algorithms for estimating the risk based on personal data may be biased towards specific social groups, leading to systematic disadvantages for those groups. Personalized premiums may thus lead to discrimination and social injustice. It is well known from many application fields that such biases occur frequently and naturally when prediction models are applied to (...) people unless special efforts are made to avoid them. Insurance is no exception. In this paper, we provide a thorough analysis of algorithmic fairness in the case of insurance premiums. We ask what “fairness” might mean in this context and how the fairness of a premium system can be measured. For this, we apply the established fairness frameworks of the fair machine learning literature to the case of insurance premiums and show which of the existing fairness criteria can be applied to assess the fairness of insurance premiums. We argue that two of the often-discussed group fairness criteria, independence (also called statistical parity or demographic parity) and separation (also known as equalized odds), are not normatively appropriate for insurance premiums. Instead, we propose the sufficiency criterion (also known as well-calibration) as a morally defensible alternative that allows us to test for systematic biases in premiums towards certain groups based on the risk they bring to the pool. In addition, we clarify the connection between group fairness and different degrees of personalization. Our findings enable insurers to assess the fairness properties of their risk models, helping them avoid reputation damage resulting from potentially unfair and discriminatory premium systems. (shrink)

Statistical discrimination is a form of discrimination that uses statistical inferences about the groups to which individuals belong as grounds for treating them differently. It remains unclear what, if anything, makes statistical discrimination wrong. My thesis argues that statistical discrimination is wrong because, and insofar as, it contributes to existing social injustice. After an introduction to the issues in section 1, section 2 clarifies the concept of statistical discrimination and its differences with non-statistical discrimination. Section 3 discusses different accounts that (...) seek to explain when and why statistical discrimination is wrong. I examine two approaches, one of which regards the wrong of statistical discrimination as part of discrimination in general, while the other conceives of the wrong as distinctive to statistical discrimination itself. I argue the former approach is better. Among different accounts within the approach, the context-based consequentialist account that explains the wrong in relation to existing social injustice is most promising. Section 4 uses racial profiling as an example to illustrate how the account can help explain this hard case of statistical discrimination, calling upon an argument developed by Benjamin Eidelson. (shrink)

This paper tries to explain why racial profiling involves a serious injustice and to do so in a way that avoids the problems of existing philosophical accounts. An initially plausible view maintains that racial profiling is pro tanto wrong in and of itself by violating a constraint on fair treatment that is generally violated by acts of statistical discrimination based on ascribed characteristics. However, consideration of other cases involving statistical discrimination suggests that violating a constraint of this kind may not (...) be an especially serious wrong in and of itself. To fully capture the significant wrong that occurs when racial profiling is targeted at black Americans or other similarly situated groups, it is argued that we should appeal to the idea that this basic injustice is exacerbated when it forms part of a larger pattern of similar actions that collectively realize a state of cumulative injustice. (shrink)

It might seem that racial profiling by doctors raised few of the same concerns as racial profiling by police, immigration, or airport security. This paper argues that the similarities are greater than first appear. The inappropriate use of racial generalizations by doctors may be as harmful and insulting as their use by law enforcement officials. Indeed, the former may be more problematic in compromising an ideal of individualized treatment that is more applicable to doctors than to police. Yet doctors, unlike (...) police, should not attempt to ignore race altogether. Race and ethnicity are associated with the geography of disease, with social and cultural factors relevant to disease, and, to some extent, with genetic predispositions to disease. Moreover, greater attention to the distinctive health conditions of certain racial and ethnic groups is necessary to make up for past neglect. But it will be a tricky business to steer a middle course between a naïve, reckless, or dogmatic color blindness and a stereotype-driven overreliance on race and ethnicity. In trying to steer such a course, the doctor can only hope for the day when a combination of more individualized genomic medicine and greater social equality will make it possible for her to pay less attention to race and ethnicity without detriment to her patients' health. (shrink)