The Yale Law Journal

October 2015

Against Immutability

Jessica A. Clarke

abstract. Courts often hold that antidiscrimination law protects “immutable” characteristics, like sex and race. In a series of recent cases, gay rights advocates have persuaded courts to expand the concept of immutability to include not just those traits an individual cannot change, but also those considered too important for anyone to be asked to change. Sexual orientation and religion are paradigmatic examples. This Article critically examines this new concept of immutability, asking whether it is fundamentally different from the old one and how it might apply to characteristics on the borders of employment discrimination law’s protection, such as obesity, pregnancy, and criminal records. It argues that the new immutability does not avoid the old version’s troublesome judgments about which traits are morally blameworthy and introduces new difficulties by requiring problematic judgments about which traits are important. Ultimately, immutability considerations of both the old and new varieties distract from the aim of employment discrimination law: targeting unreasonable and systemic forms of bias.

author. Associate Professor and Vance Opperman Research Scholar, University of Minnesota Law School. I am grateful to Bradley Areheart, Stephen Befort, Ann Burkhart, June Carbone, Mary Anne Case, Carol Chomsky, Antony Duff, Elizabeth Emens, Allan Erbsen, Joseph Fishkin, Michele Goodwin, Jill Hasday, Kristin Hickman, Claire Hill, Neha Jain, Erin Keyes, Heidi Kitrosser, Bert Kritzer, Nancy Levit, Brett McDonnell, William McGeveran, Isabel Medina, Stephen Meili, Amy Monahan, Perry Moriearty, Rebecca Morrow, JaneAnne Murray, David Noll, Shu-Yi Oei, Hari Osofsky, Stephen Rich, Christopher Roberts, Jessica Roberts, Veronica Root, Vicki Schultz, Daniel Schwarcz, Francis Shen, and participants at workshops at the University of Minnesota Law School, Loyola University New Orleans College of Law, the Vulnerability and the Human Condition Initiative, the Colloquium on Scholarship in Employment and Labor Law, and the Law and Society Annual Meeting for helpful feedback on previous drafts. My thanks to Cresston Gackle, Soren Lagaard, Inga Nelson, Katharine Saphner, Leah Tabbert, and the University of Minnesota Law Library for superb research assistance.


Why is it illegal to discriminate on the basis of certain traits, like race or sex, but not others, like experience or beauty? One answer that has been offered in the context of the constitutional guarantee of equal protection is that certain human traits are immutable, meaning they were not chosen. This concept has long endured the scholarly criticism that it is “both over- and underinclusive.”1 For example, it is permissible to discriminate on the basis of intelligence, which some say is innate, but not religion, which some say can be changed. In response to the argument that sexual orientation might be changed and is therefore undeserving of protection, gay rights advocates have persuaded many courts, perhaps even the Supreme Court, to adopt a different understanding of immutable characteristics.2 Many courts now ask “not whether a characteristic is strictly unchangeable, but whether the characteristic is a core trait or condition that one cannot or should not be required to abandon.”3 Or, as another judge put it, “‘immutability’ may describe those traits that are so central to a person’s identity that it would be abhorrent for government to penalize a person for refusing to change them, regardless of how easy that change might be physically.”4

The success of the revised version of immutability in the courts has given new life to a concept once thought dead and led scholars to apply the insight to other identities or traits that are not currently protected by antidiscrimination law.5 Scholars have been optimistic about the so-called “new immutability”6 for its potential to expand those aspects of identity covered by antidiscrimination law.7

This Article offers the first sustained challenge to the new immutability.8 Despite the extensive attention the theory has received in judicial opinions and legal scholarship,9 no work has critically considered its broader implications for the development of antidiscrimination law. The evolution of immutability has important implications for antidiscrimination doctrine, as well as debates among the public, legislatures, and employers over whether to prohibit discrimination on the basis of various traits and identities. To assess the theory’s potential and limits, this Article examines how arguments based in the revised version of immutability might play out with respect to characteristics on the borders of employment discrimination law’s protection. It concludes that, while the new immutability has had success in constitutional litigation for LGBT rights, it is a questionable strategy for reconceptualizing the broader project of equality law. As a normative matter, the new immutability obscures critical questions about why some characteristics ought to be treated equally, offering only the empty assertion that they are fundamental to personhood. Rather than replacing the old theory of immutability, which entails problematic moral judgments about individual responsibility, the new version reinvigorates the ideology behind the old. As a strategic matter, the new immutability may backfire for groups advocating that new forms of bias be prohibited, because it creates line-drawing problems and justifies only limited forms of protection.

This Article is concerned with the migration of the new immutability from equal protection cases to new contexts, particularly the various statutes prohibiting employment discrimination.10 While this Article suggests reasons to be skeptical of the new immutability in general, its intervention is not focused on equal protection doctrine or the same-sex marriage cases. Instead, it focuses on employment discrimination, not only because of the economic importance and profound social significance of the workplace,11 but also because employment discrimination law has shown remarkable willingness to extend legal protections to new traits.12

As the role of immutability in the Supreme Court’s equal protection jurisprudence has diminished, the concept has continued to have a strong influence and enduring explanatory force in employment discrimination law.13 While immutability is but one among many factors in equal protection doctrine,14 it often plays a determinative role in employment discrimination disputes. Courts use the old concept of immutability to limit the reach of employment discrimination statutes, narrowly construing what counts as discrimination based on characteristics such as race, sex, and disability.15 The old immutability’s pervasive influence on employment discrimination law suggests that the new version might have obvious applications there as well.16 This Article considers how the new immutability might play out in controversies over whether the law forbids employment discrimination based on obesity,17 pregnancy,18 and criminal records.19 Despite plausible statutory arguments for covering these types of discrimination, courts often refuse to extend protection.20

In these contexts, the old immutability’s argument that these traits were chosen lies at the heart of courts’ refusals to extend familiar forms of antidiscrimination protection.21 The old immutability assumed that certain traits, like race and sex, were not blameworthy on account of being “accidents of birth.”22 The corollary is that traits for which an individual is accountable, in some sense, are appropriate bases for discrimination.23 This reasoning may be premised on the moral intuition that discrimination against those who are blameworthy is fair. Or it may rest on the unstated assumption that the law should create incentives for good (or efficient) behavior by allowing discrimination on the basis of certain bad (or costly) choices. For instance, discrimination based on weight often goes unredressed by the law because obesity is commonly thought to be a mutable trait that may be prevented or ameliorated through adjustments to lifestyle and diet.24 Judges may refuse to require that employers accommodate pregnancy because they believe that women who make private reproductive choices ought to bear the costs.25 Employment discrimination on the basis of criminal records is thought to be fair as a collateral consequence of conviction.26 The judgments underlying these views are often harsh, intrusive, and stigmatizing. Yet these moral and economic judgments lie below the surface of policy and legal doctrine and are rarely interrogated or theorized. Is obesity more morally blameworthy than heart disease, which is protected from discrimination despite having behavioral components? Should pregnant women alone bear the costs of pregnancy? Should those with criminal records be shunned from all employment opportunities?27

The new immutability is no better than the old on these questions. It fails to provide a theoretically satisfying basis for understanding which characteristics deserve protection and invites normatively problematic judgments that are at odds with the purposes of antidiscrimination law. While the old immutability assumed that certain traits might be blameworthy because they were chosen, the new immutability’s appeal to “personal identity” masks underlying moral assessments about which traits, while entailing some degree of choice, ought not be blameworthy.28 These estimations may be unfair and irrelevant to employment. But by softening the edges of immutability theory to render it more appealing, the new immutability shields problematic judgments from scrutiny. Moreover, the new immutability’s focus on valued traits leaves out many stigmatized identities—identities that might have the strongest claims to protection precisely because judgments based on them are superficial and perpetuate systemic subordination.29 For example, many people would dispute that weight is a central part of identity, and most people would prefer to change their weight if they could.30 (And many would dispute that even sexuality, sex, and race are, or ought to be, central to personhood.31) Even worse, to argue a trait is fundamental to personality is to bolster the argument that it cannot change.32 The suggestion that criminal records are central to personality would lend support to employer beliefs that automatic exclusion of all those with criminal records will help avoid workplace crime.33 In this way, the new immutability reinforces stereotypes of the sort that antidiscrimination law is intended to disrupt.

Nor does the new immutability clear a path toward legal protection for new characteristics. The new immutability protects traits that are fundamental to a person’s identity.34 But defining what makes a trait fundamental is not easy, giving rise to judicial anxiety that protecting new identities might lead down a slippery slope to protecting all variations in personality.35 For example, judicial opinions on whether obesity is a protected disability demonstrate that courts are likely to resist extending protection for weight if the question is framed as a right to personality, because, these courts reason, every aspect of an individual’s appearance might be said to be central to personality.36 Additionally, the protection offered by the new immutability may be sparse. The new immutability draws on the ideas of liberty and privacy, but protections for liberty and privacy are often limited to rights against state interference, rather than the full set of antidiscrimination remedies.37 Finally, the new immutability invites intractable conflicts among groups asserting that certain choices are fundamental to their identities, such as between women seeking insurance coverage for contraception and employers whose religious beliefs do not countenance non-procreative sex.38

While the new immutability may have been a useful doctrinal workaround for courts seeking to expand equal protection to sexual orientation, it is not a fruitful way to reimagine the law of equality in every context. Asking whether a characteristic is immutable, in either the new or old sense, focuses attention on the victims of discrimination and their blameworthy or costly choices, rather than the systemic effects of biases that are not required for the workplace to function. Immutability is a poor fit for employment discrimination law because it measures a person in the abstract, not that person’s qualifications for a particular job.39 It also fails to call attention to how certain biases, when compounded, can result in caste-like social structures, leading to wholesale disadvantages or constrained opportunities based on identity.40

This Article proceeds in four parts. Part I describes the revised theory of immutability, discussing its origins in equal protection jurisprudence, its evolution in recent gay rights cases, and its emergence in employment discrimination law. Part II raises several normative and tactical objections to the revised immutability. It argues that even when revised, immutability is a harsh and intrusive moral theory. The new immutability’s protections for “personhood” exclude the most stigmatized, and its underlying premises reinforce stereotypes. Practically, the new immutability fails to give courts a principled basis for distinguishing between those traits that deserve protection and those that do not. It cannot justify transformative interventions into discriminatory social practices, and it invites conflicting equality claims. Part III applies these objections to current controversies in employment discrimination law. It discusses how the old immutability limits the law’s reach in the weight, pregnancy, and ex-offender discrimination contexts, and how the new immutability also fails to capture the wrong of these forms of discrimination. Part IV analyzes two alternatives to immutability arguments in the employment discrimination sphere: specifically, universal approaches that seek to enhance fairness for all workers, and targeted approaches that address systemic and superficial barriers to opportunity. It argues for targeted, incremental expansion of employment discrimination law, along with explicit scrutiny of the moral judgments behind immutability arguments of any stripe.

I. the revised theory of immutability

This Part discusses the original reasoning behind the immutability factor in equal protection doctrine, describes objections to that theory, and explains how the immutability factor has been transformed by lower courts in gay rights cases. This Article’s primary intervention, however, is not aimed at equal protection doctrine. Rather, this Article aims to demonstrate problems with the theory of immutability as a general test of what traits ought to be protected against discrimination, by looking at its applications in a specific context: employment discrimination law. It examines the equal protection cases to excavate the justifications for the assumption that immutable traits deserve protection, to connect those arguments to moral theories about egalitarianism, and to critique those arguments as applied to specific controversies in employment discrimination law. While this analysis may also suggest directions for constitutional law, this Article leaves those questions for another day. This Part will describe the revised immutability and how it attempts to address the principal objections to the old.

A. Two Concepts of Immutability

1. Protection from Chance

This Part will discuss the “old” concept of immutability as chance, luck, or an “accident of birth,” as that idea arose in the Supreme Court’s equal protection jurisprudence. Other scholarly treatments of this subject have examined immutable traits defined as those characteristics that cannot be changed.41 This account examines another definition of immutable traits: characteristics for which an individual is not responsible. It will then connect that concept of immutability with the moral theories of egalitarianism that might support it, an exercise that reveals a number of objections to the old immutability. This Part will discuss those objections and conclude by describing the demise of the old immutability in Supreme Court jurisprudence.

To begin, the Supreme Court has mentioned immutability as one of several factors that might be relevant to the question of whether a legislative classification based on a particular trait deserves heightened scrutiny by the courts.42 For instance, the Court has referred to immutability alongside “visibility”—whether a group “exhibit[s] obvious . . . or distinguishing characteristics that define [it] as a discrete group.”43 The Court has considered other independent factors as well, including whether the class has “experienced a ‘history of purposeful unequal treatment,’”44 whether it has “been subjected to unique disabilities on the basis of stereotyped characteristics not truly indicative of [its] abilities,”45 and whether it is “in need of extraordinary protection from the majoritarian political process.”46

The Supreme Court has referred to immutable traits simply as those that their “possessors are powerless to escape or set aside.”47 But the concept of immutability is deeply rooted in notions of individual responsibility, referring not just to traits that cannot be changed, but also traits that were never chosen.48 The term “immutable characteristic” first appears in the Supreme Court’s equal protection jurisprudence in Frontiero v. Richardson, a case striking down policies that discriminated on the basis of sex by providing more benefits to the wives of male military servicemembers than to the husbands of female servicemembers.49 A plurality of the Court reasoned:

[S]ince sex, like race and national origin, is an immutable characteristic determined solely by the accident of birth, the imposition of special disabilities upon the members of a particular sex because of their sex would seem to violate “the basic concept of our system that legal burdens should bear some relationship to individual responsibility . . . .”50

To support this conclusion, the Frontiero plurality cited Weber v. Aetna Casualty & Surety Co., a case with facts that starkly illustrate the unfairness of accidents of birth.51 In Weber,death benefits under a workers’ compensation scheme had been denied to a deceased man’s children because those children were “illegitimate.”52 The Court held:

The status of illegitimacy has expressed through the ages society’s condemnation of irresponsible liaisons beyond the bonds of marriage. But visiting this condemnation on the head of an infant is illogical and unjust. Moreover, imposing disabilities on the illegitimate child is contrary to the basic concept of our system that legal burdens should bear some relationship to individual responsibility or wrongdoing. Obviously, no child is responsible for his birth and penalizing the illegitimate child is an ineffectual—as well as an unjust—way of deterring the parent.53

Immutability is therefore not confined to biological traits; as this legitimacy example demonstrates, social categories too may be assigned at birth.54

To be sure, the Court has not held that immutability is necessary55 or sufficient56 to turn a classification suspect. But even when the Court has refused to hold that a group with an immutable characteristic is a “suspect class,” it has been solicitous in its application of rational basis review toward the rights of those deemed innocent.57 In Plyler v. Doe, the Court struck down a Texas statute excluding undocumented immigrant children from public education.58 While the Court acknowledged that “undocumented status” is not “an absolutely immutable characteristic since it is the product of conscious, indeed unlawful, action,” it nevertheless concluded that the law in question “imposes its discriminatory burden on the basis of a legal characteristic over which children can have little control.”59 This type of unfairness “suggests the kind of ‘class or caste’ treatment that the Fourteenth Amendment was designed to abolish.”60 In City of Cleburne v. Cleburne Living Center, the Court confronted a Texas city’s ordinance denying a permit for the operation of “a group home for the mentally retarded.”61 The Court held that the group did not have suspect class status, but went on to quote John Hart Ely: “Surely one has to feel sorry for a person disabled by something he or she can’t do anything about . . . .”62 The Court struck down the ordinance.63

Immutable characteristics, defined as “accidents of birth,” are suspect to the Court because they bear no relationship to individual responsibility.64 As a practical matter, the law is unlikely to deter private conduct by discriminating on the basis of accidents of birth because their bearers did not choose, or may be powerless to change, these immutable traits.65 The term “accident,” then, is used in the moral sense, as the opposite of a matter for which one is accountable.66

This moral concept has powerful intuitive appeal. It is linked to notions of childhood innocence, eliciting empathy for those who were blameless in their misfortunes and evoking disdain for those who obtained their privilege without merit. The phrase “accident of birth” has a long philosophical pedigree, and was an important theme in the writing of John Stuart Mill on sex and race equality in the nineteenth century.67 Mill wrote:

If it be said that . . . virtue is itself the greatest good and vice the greatest evil, then these at least ought to be dispensed to all according to what they have done to deserve them; instead of which, every kind of moral depravity is entailed upon multitudes by the fatality of birth; through the fault of their parents, of society, or of uncontrollable circumstances, certainly through no fault of their own.68

More recent philosophical work has developed this intuition—that individuals should not be responsible for chance occurrences—into a theory of distributive justice called “luck egalitarianism.”69 An extreme version of this theory would justify widespread redistribution.70 A more limited version undergirds the notion of immutability operative in equal protection doctrine—that the Constitution calls for scrutiny of government classifications that burden those whose fault is only in their stars.71

Luck egalitarianism is subject to a number of criticisms that might also be applied to the use of immutability as a criterion for antidiscrimination protection. These include problems distinguishing between chance and choice, as well as the theory’s harshness, its intrusiveness, and its stigmatizing effects.

Problems Distinguishing Between Chance and Choice. First, critics argue that luck egalitarianism rests on a thin notion of choice, as distinguished from chance. Luck egalitarians struggle to draw a principled line between merit and luck: talents, aptitudes, and even the motivation to work hard may be just as much accidents of birth as race or sex.72 One need not believe free will is an illusion to agree that what may seem a free choice from a privileged perspective may seem predetermined by socioeconomic circumstances from a disadvantaged one. What appears merit may be luck, as privileged families transmit myriad advantages to their children.73 And the “irresponsible” choices on which luck egalitarians would rest moral responsibility may not, upon closer examination, turn out to be “informed, voluntary, uncoerced, and deliberated upon.”74 Moreover, there are good reasons to question impulses to assign individual agency for certain traits but not others. Social science research demonstrates spontaneous “blame validation” effects in which observers tend to over-ascribe volition and causation to individuals they have already implicitly judged as morally culpable.75

Harshness. A second criticism is that luck egalitarianism is harsh. On a pure version of the theory, an uninsured driver who is at fault for an accident should not be granted emergency medical assistance.76 If the driver survives and is disabled, society has no obligation to accommodate her disability.77 The theory distinguishes between “the deserving and the undeserving disadvantaged,” and abandons the latter, even if her circumstances are catastrophic.78 It is not concerned with providing second chances, opportunities to correct mistakes, or paths to redemption.

It is no answer to these criticisms to say that some traits, like intelligence, are often relevant to legitimate purposes while others, like race, are usually not. As Ely put it: “At that point there’s not much left of the immutability theory, is there?”79 Equal protection jurisprudence generally asks the question of a trait’s relationship to a governmental objective only after deciding the extent to which the classification is suspect and deserving of special scrutiny.80 Likewise, employment discrimination law generally asks whether a trait was related to the job, but this inquiry is only necessary if the plaintiff has shown that the employer discriminated on some prohibited basis.81 The relevance of the trait to a particular purpose does all the work in the analysis, and immutability has no bearing, unless we imagine that irrelevant traits are more suspect when they are immutable, and less suspect when they are mutable. But to endorse the idea that the state might discriminate on the basis of irrelevant traits just because they are mutable is to justify all forms of “state-sponsored cultural conformity and assimilationism.”82 Thus, immutability arguments allow those in power to require stifling conformity to conventional norms.

Intrusiveness. This observation points to a third objection to luck egalitarianism: intrusiveness. Luck egalitarianism requires “moralizing judgments of individual choices” that “interfere[] with citizens’ privacy and liberty.”83 “[I]n order to lay a claim to some important benefit, people are forced to obey other people’s judgments of what uses they should have made of their opportunities. . . .”84 This intrusiveness objection resonates with the concerns of scholars critical of immutability as a forward-looking concept (as inability to change), rather than a backward-looking one (as lack of responsibility). Kenji Yoshino has argued that the immutability factor reflects an “assimilationist bias,” allowing government policies to create incentives for those who can “change or conceal their defining trait” to conform to mainstream expectations.85 According to Yoshino, one problem with the immutability factor is that it transforms the “descriptive claim that a group can assimilate . . . into the prescriptive claim that it should assimilate without much intervening investigation by the courts into the legitimacy of the legislation.”86 For instance, “Jews generally can change or conceal their religion, while blacks generally cannot change or conceal their race. This surely does not make anti-Semitic legislation more legitimate than racist legislation.”87 This example demonstrates how immutability arguments deflect attention from questions about the extent to which religious coercion is a legitimate pursuit for governments or employers. Likewise, immutability arguments focused on past responsibility deflect attention from questions about whether those in power have legitmate reasons for imposing moralizing judgments on citizens or employees.

Stigma. A fourth criticism responds to the argument that law might appropriately disincentivize or deter mutable, but not immutable, traits. Whether discrimination is an effective means of incentivizing behavior is an empirical question, but it is rarely examined as such.88 Discrimination may be intended to change behavior by shaming individuals possessing a certain trait.89 Whether shaming is generally effective in shaping behavior, as an empirical matter, is disputed and depends on context.90 In discrimination contexts, there are reasons to doubt whether adverse treatment of those with mutable traits would be effective in shaping behavior. Rather than giving people incentives to take personal responsibility, an immutability requirement may instead put people to the disempowering task of proving they were victims of circumstances beyond their control.91 For example, an individual staking a claim to disability benefits must tell a story of misfortune that convinces adjudicators it is impossible for her to work through no fault of her own. And in the process, she must disclaim whatever abilities, competences, and hopes of returning to work she might have.92 Additionally, an immutability requirement stigmatizes some traits for which an individual certainly bears some responsibility, leading those individuals to dissemble about their status, conceal the trait, or avoid seeking needed assistance.93

Apart from these utilitarian considerations, many egalitarians oppose shaming practices as being “characteristic of hierarchical relationships.”94 Shaming penalties have historically been employed to reaffirm class relationships and reinforce the shamed person’s subordinate status.95 Martha Nussbaum has distinguished shame, which is about actors, from guilt, which is about acts, arguing that “whereas shame focuses on defect or imperfection, and thus on some aspect of the very being of the person who feels it, guilt focuses on an action (or a wish to act), but need not extend to the entirety of the agent . . . .”96 While the criminal law may appropriately determine guilt and assign punishment, shaming is a form of “mob justice” in which the public punishes offenders in ways that are “not the impartial, deliberative, neutral justice that a liberal-democratic society typically prizes.”97 Stigmatizing practices go beyond expressing disapproval of a particular act or behavior to impose a “spoiled identity” on their targets.98 One feature of stigma is that people “tend to impute a wide range of imperfections on the basis of the original one.”99 Thus, “[t]he harm of stigma is that a single perceived characteristic is seen as ‘disqualifying’ the whole person, excluding him or her from membership in the community that calls itself the ‘normals.’”100 As a result of this exclusion, empathy for the stigmatized group breaks down.101 Those targeted by shaming practices often internalize stigma, coming to believe themselves to be deficient.102 Antidiscrimination law has long been concerned with practices that assign stigmatized identities to individuals based on certain traits or behaviors, and then shame and disparage those individuals as less deserving of equal respect or consideration.103

In the 1980s and 1990s, a number of courts confronted the question of whether sexual orientation classifications violate the Equal Protection Clause. Most upheld these classifications based, at least in part, on the logic that homosexuality was “behavioral” rather than immutable.104 During this time, the concept of immutability was under attack by scholars and judges.105 The courts, including the U.S. Supreme Court, began citing academic criticisms of immutability106 and omitting it as an express consideration in equal protection analysis.107 Over the last two decades, the Supreme Court’s equal protection cases have seldom mentioned immutability.108 A notable exception is the brief reference to sexual orientation’s immutability in Obergefell v. Hodges, the Supreme Court’s landmark 2015 same-sex marriage opinion.109

2. Protection for Choice

Despite the Supreme Court’s move away from immutability in the decades prior to Obergefell, that concept enjoyed a substantial resurgence in the equal protection jurisprudence of lower federal and state courts.110 These courts reformulated the immutability factor in same-sex marriage cases. This Part will describe those cases and discuss how this new concept of immutability attempts to respond to the objections to the old.

It is important to note that the new immutability is “new” only in the sense that its rise to prominence is recent. As early as 1981, Douglas Laycock argued that “some characteristics should be treated as immutable because of fundamental interests in not changing them,” such as sex and religion.111 The new immutability began gaining jurisprudential traction in 1989, when Judge William Norris advanced the theory in his concurring opinion in Watkins v. U.S. Army, a Ninth Circuit case challenging the military’s refusal to re-enlist a soldier because he was gay.112 A similar definition of immutability developed concurrently in asylum law.113 This Article follows recent scholarship in referring to this definition as the “new immutability,”114 although it has also been termed “personhood”115 or “soft immutability.”116

Judge Norris’s opinion resisted the simplistic distinction between chance and choice. It demonstrated how traits thought to be products of chance may be subject to control, and how traits thought to be products of choice may be experienced as inevitable. The opinion rejected an understanding of “immutability in the sense that members of the class must be physically unable to change or mask the trait defining their class.”117 While acknowledging that “[i]t may be that some heterosexuals and homosexuals can change their sexual orientation through extensive therapy, neurosurgery or shock treatment,” Judge Norris concluded this did not make sexual orientation “mutable.”118 As examples of how traits considered immutable might be changed, the opinion pointed to sex change through surgery, naturalization of aliens, legitimation of children, and racial passing.119 Changing such traits is not impossible, but it “would involve great difficulty, such as requiring a major physical change or a traumatic change of identity.”120 In support, the opinion cited research from disciplines such as psychiatry.121 To drive home the point that some aspects of identity, while not “strictly immutable,” might be “effectively immutable,” the opinion asked “whether heterosexuals feel capable of changing their sexual orientation.”122 As another court explained, such traits “may be altered only at the expense of significant damage to the individual’s sense of self.”123

But ultimately, neither scientific proof of the difficulty of change nor the trauma of conversion efforts is what makes a trait immutable.124 Judge Norris explained that immutable traits are

those traits that are so central to a person’s identity that it would be abhorrent for government to penalize a person for refusing to change them, regardless of how easy that change might be physically. Racial discrimination, for example, would not suddenly become constitutional if medical science developed an easy, cheap, and painless method of changing one’s skin pigment.125

This analogy to racial discrimination only works if the reader believes, like Judge Norris, that the various sexual orientations are all on equal moral footing, just like racial categories are. Judge Norris’s opinion acknowledges that in determining which traits to protect, constitutional law must distinguish between fair and unfair bases for discrimination: “After all, discrimination exists against some groups because the animus is warranted—no one could seriously argue that burglars form a suspect class.”126 The opinion is not clear on why discrimination based on sexual orientation was unfair, unlike discrimination against burglars, but it likely rested on Judge Norris’s conclusion that sexual orientation status did not necessarily entail any sort of illegal conduct.127 To be sure, Bowers v. Hardwick, a 1986 Supreme Court decision upholding a state statute criminalizing homosexual sex,was still good law at the time Judge Norris was writing.128Thus, the new immutability had a narrow reach: invalidating a military rule that defined sexual orientation as a “status,” but not any rules prohibiting same-sex sodomy, marriage, or other “conduct.”129

The new immutability is inflected with ideas about privacy and liberty, by contrast to the intrusiveness of the old theory. It finds inspiration in Justice Blackmun’s dissent in Bowers, which drew on cases protecting the right to privacy to argue that “[w]e protect those rights . . . because they form so central a part of an individual’s life” and are “significant” ways “that individuals define themselves.”130 Lawrence v. Texas overruled Bowers in 2003.131 In light of Lawrence, the Connecticut Supreme Court expanded the theory of the new immutability in Kerrigan v. Commissioner of Public Health.132 Kerrigan holds that sexual orientation is immutable because the Constitution protects the right of “homosexual adults to engage in intimate, consensual conduct” as an “integral part of human freedom.”133 Quoting the U.S. Supreme Court’s decision in Paris Adult Theatre I v. Slaton, the Connecticut Supreme Court held that “[s]exual intimacy is ‘a sensitive, key relationship of human existence, central to . . . the development of human personality . . . .’”134 In this formulation, immutability has mutated into an argument about choice—“a person’s fundamental right to self-determination.”135 Cases on the new immutability hold that, if a trait is not innocent in the sense of being an accident of birth, it must be innocent in the sense of being an “integral part of human freedom.”136

Obergefell reflects this new understanding of immutability, although that case did not explicitly define immutability, nor did it specify the role immutability might have played in its analysis.137 In Obergefell, the U.S. Supreme Court cited a brief by amici, led by the American Psychological Association (APA), in support of the claim that “psychiatrists and others” have recognized that sexual orientation is “immutable.”138 The APA Brief did not argue that sexual orientation is never chosen or impossible to change.139 Instead, it explained that sexual orientation “defines the universe of persons with whom one is likely to find the satisfying and fulfilling relationships that, for many individuals, comprise an essential component of personal identity.”140

For the old immutability, a normative conclusion flows from a descriptive premise: because an individual never chose or cannot change a trait, that trait should not be the basis for discrimination. The new immutability substitutes a normative judgment for that descriptive premise: a certain trait should not be the basis for discrimination because it is a normatively acceptable, protected exercise of individual liberty or expression of personality.

B. The Synthesis

Courts describe the immutability test as a synthesis of the old and new formulations: a trait “that either is beyond the power of an individual to change or is so fundamental to individual identity or conscience that it ought not be required to be changed.”141 On this theory, those traits that should be protected by antidiscrimination law include those that are immutable in both the old and new senses. This Article refers to this synthesis as the revised immutability. Its analysis is directed at the revised immutability as a normative theory of what traits should be covered by antidiscrimination law.

The revised theory makes immutability more palatable to those who objected to the old version. The old immutability was subject to the criticisms that it rested on a thin delineation between chance and choice, and that it was harsh, intrusive, and stigmatizing.142 By integrating the new immutability, the revised theory might be understood as abandoning the fraught distinction between chance and choice. It refocuses the doctrine to consider the costs of change for an individual’s sense of self. The revised theory is not as harsh as the old version, in that it expands protection beyond those deemed blameless for possessing disfavored traits to include those who have made protected choices to adopt particular traits. It addresses the intrusiveness objection by shielding a private sphere of liberty for the formation of personality, in which government intervention is prohibited. And it counters stigma by allowing those claiming discrimination to do so without disavowing their own agency and pride in determining the content of their characters. The synthesis of the old and new versions of immutability protects both those whose misfortunes result from accidents of birth and those who seek freedom from discrimination to define their own personalities.

C. Potential for Migration

The revised immutability has the potential to carry over from opinions on equal protection doctrine to broader normative debates regarding what traits should be protected against discrimination. Such a migration could impact judicial decisions, public arguments, and legislative actions in areas like employment discrimination law.

My argument is not that there is a direct doctrinal pathway through which immutability considerations must move from equal protection to employment discrimination law. Employment discrimination law differs from equal protection doctrine in many important ways. While the Equal Protection Clause constrains public entities and actors,143 employment discrimination law constrains many public and private employers.144 Although courts define which classifications are suspect for purposes of equal protection, legislatures specify suspect classifications for purposes of employment discrimination law.145 The array of traits protected by statute is far broader than that of those protected by the Constitution.146 Additionally, employers themselves are also important sources of discrimination rules and norms.147

Despite these differences, the concept of immutability from the equal protection context plays a role in employment discrimination law in ways both direct and indirect.148 Even though the term “immutability” does not appear in any employment discrimination statute, courts have borrowed immutability concepts to answer definitional questions about the scope of statutory prohibitions on discrimination.149 For example, courts have held that employer policies that allow women, but not men, to wear their hair long do not discriminate on the basis of sex because hair length is not an immutable trait.150 Immutability’s endurance may be a result of the concept’s intuitive appeal. Or it may be due to the “gravitational pull” that constitutional principles often exert on statutory interpretation.151 Some cases suggest that the Court is inclined to interpret equal protection doctrine and employment discrimination statutes in a unified manner,152 a practice likely to impact lower court decisions. This “doctrinal migration” may be subtle as well as formal, reflected in how rules operate if not in their specific dictates.153

Outside the courts, immutability-related ideas influence debates among legislators, scholars, employers, and members of the public regarding which traits should be prohibited bases for discrimination.154 Identity-based social movements seeking legal redress frequently articulate their aspirations in “constitutional terms” and are influenced by the “rhetoric, strategies, and norms” of constitutional law.155 This Article is concerned with the persuasive power of the revised immutability as a principle to explain what traits should be protected by equality law broadly, and by employment discrimination law specifically.156 In analyzing the prospects for the addition of new groups to antidiscrimination legislation, Elizabeth Emens has offered a descriptive model that includes the revised immutability as its first criterion.157 Sharona Hoffman goes further, arguing that the revised immutability both does and should provide a unifying principle for determining which traits employment discrimination law protects.158 For example, while race, color, sex, age, national origin, genetics, and many disabilities are commonly considered accidents of birth,159 religion, sexual orientation, transgender identity, marital status, parental status, cigarette smoking, and political affiliation might be considered crucial to an individual’s right to self-definition.160 Many argue that certain characteristics associated with the old immutability (for example, race) may include traits that are immutable in the new sense (for example, an African American woman who wears her hair in cornrows), and therefore deserve protection too.161

II. objections to the revised immutability

Despite its advantages over the old immutability, there are reasons to object to the revised immutability’s suitability as a unified theory of protected traits. This Part will sketch out those objections. It draws empirical support from recent gay rights controversies.162 I explore these objections with an eye toward how the revised immutability might function as a normative theory of employment discrimination law.163

Some of these concerns are normative: first, that the revised immutability masks questionable moral judgments about the blameworthiness of traits and reinvigorates the old theory of immutability; second, that it inserts a highly disputable notion of “personhood” into the doctrine that omits many traits that are stigmatized or irrelevant to any government or employer purpose; and third, that it reinforces stereotypes about the identities it protects.

Other interrelated concerns are strategic: first, that the revised immutability is unlikely to succeed with courts or legislatures because of the difficulty of drawing a principled line between those characteristics that are essential to personhood and those that are not; second, that even if it does succeed, rights based on the new immutability alone will be more limited than other antidiscrimination rights because they will be perceived as negative liberties against state or employer interference, rather than positive demands for equal access; and third, that groups concerned with maintaining traditional social norms may argue their convictions are just as “immutable,” requiring compromises that would not be palatable in other contexts, like race discrimination.

After this Part describes the theoretical and empirical bases for these objections, Part III will discuss how these critiques might specifically apply in controversies over protection for traits currently at the borders of employment discrimination law: obesity, pregnancy, and criminal records.

A. Masking Moralizing Judgments

As discussed, the old immutability often rests on untenable, harsh, intrusive, and stigmatizing judgments about the traits for which individuals should receive blame.164 These judgments are often not just moral, but moralizing: that is, based on superficial assumptions or made without consideration of their broader implications.165 Cases revising the theory of immutability have not abandoned the core idea of immutability as blamelessness, or the moralizing judgments inextricable from the concept. Rather, these cases have buried immutability under a notion of personhood that protects certain choices already deemed morally acceptable from discrimination. Thus, the revised theory merely mitigates and obscures the untenability, harshness, intrusiveness, and stigma objections to the old.166 Moreover, the new immutability fails to explain why certain characteristics ought to be protected while others should not. Partial corrections to the theory of immutability run the risk of masking the moral judgments that are its necessary prerequisites, and making a flawed concept more palatable.

The revised immutability may become a Trojan horse: an appealing conceptual package that allows the harsh, intrusive, and stigmatizing judgments that corrupted the old immutability to sneak back into doctrine. As an initial matter, not all courts that revise the immutability factor understand the gravamen of an immutable trait to be its importance for an individual’s self-determination.167 Rather, some have merely softened the definition—from “unchanging” to “difficult to change.”168 In Baskin v. Bogan, for example, the Seventh Circuit reformulated the immutability factor to mean, if not unchangeable traits, then “at least tenacious” characteristics, including attributes that are “biological, such as skin color, or a deep psychological commitment, as religious belief often is.”169 Similarly, the Iowa Supreme Court has made the definition of immutability a question of “degree”: “[t]he degree to which an individual controls, or cannot avoid, the acquisition of the defining trait, and the relative ease or difficulty with which a trait can be changed.”170 This “difficult to change” standard acknowledges the trouble with the old immutability’s vexed notion of choice. It recognizes that certain psychological commitments may be just as difficult to overcome as, for example, skin color. This standard is somewhat less harsh, intrusive, and stigmatizing than the old, in that it does not require individuals to make Herculean (or Sisyphean) efforts to change their personalities.

But the “difficult to change” definition of immutability remains firmly rooted in notions of “individual responsibility,” requiring the same moralizing judgments as the old immutability.171 The Seventh Circuit distinguished “tenacious” traits from “characteristics that are easy for a person to change, such as the length of his or her fingernails.”172 The Iowa Supreme Court spelled out the reasoning: the relative difficulty of change “may separate truly victimized individuals from those who have invited discrimination by changing themselves so as to be identified with the group.”173 These courts did not stray far from the original theory of immutability; they simply “transpos[ed] the site of immutability from the body to the personality.”174 This rule requires harsh, intrusive, and stigmatizing judgments about who is “truly” victimized, based on whether a victim might have been able to change, hide, or downplay a disfavored characteristic.175 The “difficult to change” definition may not prohibit, for example, discrimination against a woman for dressing in ways associated with masculinity.176 This interpretation of the immutability factor may, in effect, be no different than the old version. Indeed, after softening the definition of immutability, the Seventh Circuit nevertheless found that sexual orientation met the test of the old immutability as a characteristic that is “probably . . . in-born.”177 Courts following the Seventh Circuit and running with this logic may simply apply the old immutability.178

By contrast to this “difficult to change” standard, other courts have defined the new immutability to emphasize that self-determination with respect to personality is a fundamental liberty, and the state ought not discriminate among choices individuals make within that sphere.179 Yet this definition is disingenuous, obscuring moral judgments about what types of variation in personality society should tolerate. Many aspects of “personality”—like risk taking, aggression, addiction, and impulse control—generate behaviors society regulates, particularly at the workplace. The concept of personality alone does not identify the appropriate limits of discrimination.

Sexual orientation provides an example. It is doubtful whether arguments from immutability, old or new, ultimately persuade courts to apply heightened scrutiny to sexual orientation classifications.180 As Michael Boucai has noted, courts that accept arguments about mutability also tend to accept other dubious rationales for denying protection.181 The argument that sexual orientation is important to personality does not rebut claims that social policy should discourage same-sex sexual behavior or encourage those few on the fence to choose heterosexuality.182

Thus, in Obergefell, the Court summarized evolving views on sexual orientation with the statement that “sexual orientation is both a normal expression of human sexuality and immutable.”183 The normative point is essential. An intermediate step in the new immutability’s argument is that there is no principled basis for treating certain gay relationships differently than straight ones, since they are both “normal.”184 “Normal” here does not simply mean commonplace; it means the opposite of pathological.185 Implicit is the idea that private, consensual, adult sexuality—whether gay or straight—is not harmful, shameful, or morally depraved. Rather, it is productive, fundamental, and essential to family life.186 Obergefell used the word “immutable” to describe not only sexual orientation, but also the “profound” features of the institution of marriage, which the Court held would not be disrupted by the inclusion of same-sex couples.187

The new immutability does not protect all intimate liberties; it is limited to certain forms of happy domesticity that are akin to traditional, heterosexual marriages.188 For example, the U.S. Supreme Court decisions that the Connecticut Supreme Court cited in support of the new immutability, Paris Adult Theatre I v. Slaton and Lawrence v. Texas,protect a traditional form of private, consensual, adult sexuality.189 Slaton was a case about hardcore pornography.190 That case endorsed a traditional view of sexual morality as “central to family life,”191 and as a virtue that had to be protected from “the tide of commercialized obscenity” even in the absence of empirical proof that commercial obscenity caused any harm.192 If Slaton protects a certain form of romanticized sexuality from the threat of the market, Lawrence protects it from the threat of the state. Critical to the success of this argument has been gay rights advocacy focused on convincing courts and the public that same-sex relationships might resemble an idealized form of traditional marriage as an “enduring” bond with “transcendent dimensions.”193 This reasoning may protect everyone’s right to enter into “transcendent” relationships, but it does not require that straight- and cross-sex relationships receive equal treatment. Thus, the right to intimate liberty alone may not bar a state from penalizing crimes like statutory rape more harshly when they involve same-sex sexual conduct.194

The concept of liberty or choice in the new immutability does not explain why a trait is within the sphere of morally blameless choices. Other arguments must fill that void. In both the old and new forms, the argument from immutability rests on generalized notions of individual responsibility, merit, incentives, and just desert. Thus, even in its revised form, the new immutability responds only partially to the objections to the old. The new immutability remains based on an inchoate sense of egalitarianism that rests on a problematic definition of responsibility, with harsh, intrusive, and potentially stigmatizing consequences. These considerations are masked behind the superficially appealing invocation of the need to protect traits “fundamental to a person’s identity,” now repeated as a mantra by courts.195

B. Excluding Inessential and Stigmatized Traits

The old immutability protected those with traits considered pre-determined; it left out those deemed accountable for the traits that form the grounds for discrimination against them. The new immutability protects a subset of the accountable: those for whom a particular, chosen characteristic is essential to who they are as individuals and a matter of pride rather than shame or regret. This formula omits inessential yet chosen conditions, such as stigmatized conditions, that might otherwise be the focus of antidiscrimination protection.

The various formulations of the new immutability assume that the trait to be protected is essential to its bearer’s identity, such that if it were to change, she would no longer be the same person. Courts describe such traits as “integral,”196 “core,”197 or “fundamental to one’s identity.”198 As one court put it, “[T]o discriminate against individuals who accept their given sexual orientation and refuse to alter that orientation to conform to societal norms does significant violence to a central and defining character of those individuals.”199 Dissenting in Bowers v. Hardwick in 1986,Justice Blackmun wrote that “[h]omosexual orientation may well form part of the very fiber of an individual’s personality.”200 Janet Halley described this definition of gay identity as “‘a personhood definition,’ in which the class of homosexuals is defined by a form of personality shared by its members.”201

This concept of personhood, as a set of core, stable traits, is both historically and culturally contingent.202 In the context of sexual orientation, for example, scholars have argued the personhood definition is by no means universal. Rather, it relies on a contested Freudian narrative of self-discovery through revelation of repressed sexuality.203 As Michel Foucault famously argued, this narrative is particularly Western and arose in a specific historical moment.204 It was nineteenth-century medicine that first gave rise to the idea of homosexuality as a type of identity distinct from heterosexuality.205 In its contemporary manifestations, the personhood idea of homosexuality

entirely fails to represent those pro-gay constituencies that deny the centrality of a particularized homosexual orientation to their psychic makeup, whether because they identify as bisexual, because they seek to de-emphasize the gender parameters of sexuality, because they are experimental about sexuality, or because they experience sexuality not as serious self-expressiveness but as play, drag, and ironic self-reflexivity.206

Sonia Katyal has detailed how the concept of “gay personhood” is both historically and culturally contingent, “[f]or at the heart of the fabled closet lies a predominantly Western assumption that a gay, lesbian, or bisexual identity is a major determinant in the lives of all individuals.”207 Katyal has described alternative cultural understandings that do not link sexual preferences with identity or that place sexual orientation in the background, with class, occupation, gender, or race as foreground determinants of sexuality.208

This concept of personhood also relies on a romanticized story of self-discovery and public disclosure through which the authentic self is actualized. In early cases formulating the new immutability, some courts cited a 1985 Harvard Law Review Note209 that argued for consideration of whether a trait plays an “important role” in “personhood” defined as “self-perception, group affiliation, and identification by others.”210 As evidence that sexual orientation is a crucial component of personhood, the Note looked to “personal testimonies in gay literature and . . . the well-known phenomenon of ‘coming out,’ or publicly acknowledging one’s gay identity,” as well as the existence of “gay communities” in the form of “bars, bookstores, newspapers, political lobbies, legal rights committees, and gay neighborhoods.”211 The legal success of sexual orientation’s personhood argument was possible only as a result of a gay rights movement that emphasized pride and “coming out” as counters to shame, secrecy, and self-loathing.212

This template for legal change is unlikely to succeed, however, for two types of traits that are traditionally covered by antidiscrimination law: those that individuals would prefer to disclaim as constitutive of their authentic selves, and those traits that individuals would prefer to change due to shame or stigma. For many multiracial individuals, for example, racial identity might be experienced as manipulable and changing over time,213 or unimportant to one’s fundamental sense of self.214 Many forms of disability, too, might fall through the cracks of the revised immutability, as conditions subject to control and yet seldom celebrated as features of identity. Sexually-transmitted diseases are an example. During debate over the Americans with Disabilities Act (ADA) in 1989, Senator Jesse Helms objected that the Act would not allow an employer to “set up any moral standards for his business by asking someone if he is HIV positive, even though 85 percent of those people are engaged in activities that most Americans find abhorrent.”215 Senator Ted Kennedy took a different tack, submitting a statement from the National Commission on AIDS that called for “understanding” and “compassion.”216 Compassion won the day.217 The internalized stigma was a reason for protection.

These gaps in the revised immutability’s coverage subvert some of antidiscrimination’s most commonly accepted premises. The argument that an identity characteristic deserves protection because it is essential or important is in tension with one of the basic impulses underlying antidiscrimination law: that individuals should be judged according to their qualifications rather than extraneous identity traits such as race, sex, and disability.218 On this theory, such traits are forbidden grounds for discrimination not because they are important, but because they are not.

Moreover, leaving out many stigmatized traits undermines antidiscrimination law’s goal of eradicating disparaging forms of subordination.219 Antidiscrimination law targets stigmatizing practices that undermine an individual’s sense of self-worth. Martin Luther King, Jr. famously wrote of the “degenerating sense of ‘nobodiness’” imposed by racial subordination.220 Stigmatized individuals may believe it futile to invest in their own futures, creating a self-fulfilling prophecy in which discrimination discourages those who find themselves defined by the stigmatized trait from pursuing education or employment opportunities. That effect of discrimination is later cast as the cause of the group’s lack of advancement.221 Thus, the revised immutability leaves out those who most acutely feel the sting of stigmatizing bias: those who have internalized a stigma and feel shame and responsibility for their conditions.

C. Reinforcing Stereotypes

Arguments that certain traits are fundamental to personality quite easily slide into the assumption that those traits are unchanging and correlate with a set of behaviors. Thus, these arguments reinforce stereotyping of identities, another of the foundational harms that discrimination law is designed to address.

The personhood argument, as a rhetorical strategy, complements and supports the assumption that certain conduct necessarily follows from status.222 “By conceiving of the conduct that it purports to protect as ‘essential to the individual’s identity,’ personhood inadvertently reintroduces . . . the very premise of the invidious uses of state power it seeks to overcome.”223 In this way, the new immutability is not far from cases predating Lawrence v. Texas that understood homosexuality as a result of deviant sexual acts that formed the foundation of a person’s character.224 In those cases, individuals might have been classified as gay because, for example, they identified as such225 or acknowledged involvement in a gay organization.226 On the basis of that classification, courts assumed these individuals had engaged in then-illegal sodomy and had the propensity to do so in the future, even without proof of any past sodomy.227 With this understanding of sexual orientation as “essential, foundational, and inescapable,” courts denied equal protection.228 Arguments based on the new immutability are thus inextricably tied to a set of assumptions about the stability of personality types and propensity for certain behavior. The assumption that personalities can be “typed” and that individuals will conform to their types is the definition of stereotyping.

Yet antidiscrimination law condemns stereotyping.229 A stereotype, in antidiscrimination law, is “any imperfect proxy” or “overbroad generalization.”230 Importantly, the problem of stereotyping is distinct from the problem of stigma and subordination. Mary Anne Case provides a hypothetical to explain:

Imagine, for example, a society with two castes, not upper and lower, not Brahmin and untouchable, but priest and warrior. The two castes are equal in status, but radically different in role. Those born into the priest caste are limited to the role of priest even if they would rather fight than pray, and vice versa.231

Such a caste system, while not subordinating, would implicate the Constitution’s guarantee of equal liberty.232 However, the injury of stereotyping, if conceptualized as a constraint on an individual’s opportunities, is not limited to rigid roles assigned at birth, during childhood, or at some other stage of relative innocence.233 Suspect stereotypes may attach at myriad stages of life, such as stereotypes about what it means to be a “mother” or “father,”234 “disabled,”235 or “old.”236

Suspect stereotypes include even generalizations that are accurate to some degree.237 Statistically sound generalizations about identity groups—for example, that most women prefer cooperative to competitive environments—may be suspect for a number of reasons.238 Because some identity-based generalizations, including many of those based on sex, are “used for purposes going far beyond the predictive capacity of the generalization,” they “can be said to be, in general, suspect.”239 Sex-based generalizations are this suspect sort of stereotype, having historically been wielded as questionable proxies for various competencies from strength to intelligence.240 Additionally, generalizations may be self-fulfilling prophecies. For example, women’s preferences for cooperation and men’s for competition may result from systemic stereotypes or patterns of discrimination rewarding conformity with the stereotype, with the resulting conformity later held up as evidence of the “truth” of the generalization.241 Equality law endeavors to address this type of harm, while arguments premised on the new immutability reinforce it.

D. Creating Line-Drawing Problems

Apart from these normative objections, the new immutability raises a number of tactical or strategic concerns for social movements seeking to extend antidiscrimination protection. The first is a line-drawing problem. The expanded concept of immutability does not have any limiting principle. While the old immutability seemed to be restricted to involuntary traits, there are no readily apparent parameters to limit which chosen traits are essential to personhood. This difficulty is likely to create judicial and public resistance to arguments premised on the new immutability.

Line-drawing problems are often articulated as hypothetical consequences of proposed theories, through the metaphors of floodgates and slippery slopes.242 Line-drawing concerns have purchase in public debates243 and in courtrooms.244 This is particularly true in equal protection cases, in which “[e]very new characteristic the courts recognize as warranting greater protection threatens to open the floodgates to a new wave of groups asking for protection based on that characteristic.”245 For example, the Supreme Court has refused to extend suspect class treatment to “the mentally retarded,” for lack of a “principled way” to distinguish this group from “the aging, the disabled, the mentally ill, and the infirm.”246 Even where statutes specify protected classifications, texts cannot provide answers to every potential question of application, and courts must use common law reasoning to decide which traits fall within the ambit of a statute’s protection. Common law reasoning requires the identification of principles that do not produce results that jurists consider counterintuitive, opposed to statutory purpose, or otherwise absurd.247

The new immutability does not help in this effort. Jed Rubenfeld has replied to personhood arguments by asking, “Where is our self-definition not at stake? Virtually every action a person takes could arguably be said to be an element of his self-definition.”248 The claim that the right to privacy should protect decisions crucial to personhood might be qualified with an exception for decisions that cause harm to others.249 But “[t]he minute someone starts defending her actions against a storm of protest with the claim that she is only affecting herself, we may be certain that the opposite is true.”250 Few actions affect only the actor, and those that generate litigation are always those where another person perceives her own interests to be at stake.251 Rights to new forms of personhood threaten social interests in saving traditional norms from disruption by iconoclasts.252 The question then becomes one of balancing relative rights, rather than recognizing absolute rights to personhood.253

E. Limiting Rights to Privacy and Recognition

The new immutability also creates a risk that rights based on the doctrine will be less robust than those based on other antidiscrimination theories. This is due to two dynamics. First, arguments based on the new immutability generally employ privacy-like reasoning: an individual has the right to make certain fundamental decisions without interference from government or employers. Privacy rights are traditionally understood as negative liberties, not positive rights that require transformative change. Second, rights premised on the new immutability may be understood as claims to cultural recognition of identities, which can work at cross purposes, politically, with claims to resource redistribution. As a result of these dynamics, rights grounded on the new immutability alone may prohibit only overt discrimination. These rights may not garner the full panoply of protections under employment discrimination statutes, which require employers to make structural changes to create more inclusive workplaces.

Rights based on the new immutability may be limited because the new immutability draws its notion of personhood from privacy doctrine.254 Privacy rights are generally understood as defending individual choices against outside influences.255 The concept of the individual evoked here is “an autonomous core—an essential self identifiable after the residue of influence has been subtracted.”256 Rights to privacy thus protect an individual’s autonomous core from intrusions by the state, society, or the market. They do not envision the autonomous self as formed through the interaction of the individual with outside forces (or at least, not in any beneficial way). Thus, they do not see any positive role for the law, society, or the market in changing existing arrangements so as to create the conditions under which individuals might better achieve autonomy in self-determination.257 To give an example: privacy rights to abortion limit the ability of the state to ban that procedure, but do not require that the state make it accessible.258 By contrast, employment discrimination statutes often require affirmative changes to the structure of the workplace to combat bias: these statutes challenge employer practices that subtly perpetuate hierarchies259 by outlawing certain practices with a disparate impact on minority groups,260 requiring changes to job requirements or accommodations for certain employees,261 and allowing affirmative action as a remedy.262

The new immutability is also hindered by its emphasis on cultural respect for outsider identities over redistribution of resources to the marginalized. It identifies the problem of discrimination as the lack of respect for an individual’s “sense of self”263 or choices as to “self-determination.”264 The harm of discrimination is thus characterized in psychological or cultural terms, rather than economic or material ones. Political theorist Nancy Fraser distinguishes between the politics of recognition and the politics of redistribution.265 Struggles for recognition seek to achieve social revaluation of disrespected identities, while efforts to achieve redistribution seek more equitable allotment of material resources.266 Fraser argues that claims for recognition often displace claims for redistribution.267 Advocates of recognition often ignore redistribution, overlooking the links between the two, seeing inequality as “free-floating” through “demeaning representations” rather than “socially grounded” in “institutionalized significations and norms.”268 For example, they may not see how “heterosexist norms which delegitimate homosexuality” were connected to a social-welfare system that denied resources to gay men and lesbians by prohibiting them from marrying.269

The success of the gay rights movement might partly be explained by its emphasis on the recognition of a private right to gay identity, rather than the redistributive effects of expanding the right to marriage.270 Thus, in United States v. Windsor, the Court mentioned not only the financial benefits but also the financial obligations that same-sex marriage would entail, such as receiving less federal financial aid for a child’s education due to a same-sex spouse’s income.271 The emphasis on the costs of marriage suggests that marriage does not always entail financial benefits for families, and thus denies that recognition of same-sex marriage necessarily redistributes resources to same-sex couples. The harm of marriage inequality is thought of primarily as misrecognition rather than unfair distribution of resources.272 Conceiving the problem thus means that redistribution is rarely seen as a remedy to heterosexism. It may also explain why disparate impact claims and affirmative action remedies are often excluded from legislative proposals to forbid sexual orientation discrimination in employment.273 Other movements following this recognition model are likely to run up against the same limits.

F. Inviting Conflict

Grounding a right on the new immutability may invite conflict from those whose moral opposition to the trait in question is also immutable. For instance, in the sexual orientation context, “[t]he intolerant heterosexual can claim, on personhood’s own logic, that critical to his identity is not only his own heterosexuality but also his decision to live in a homogeneously heterosexual community.”274 When a right is based on personhood, such conflicts are inevitable and create indeterminate legal results.275

Religious constituencies seeking protection for “immutable” convictions (e.g., the immorality of same-sex relationships) may find support from the new immutability. Some states that allow same-sex marriage also have “marriage conscience protection” provisions, giving those with religious objections to same-sex marriage the right to discriminate in employment, housing, and other domains.276 Douglas NeJaime argues that the basis for these exemptions is a “movement/countermovement” dynamic in which the Christian Right and LGBT advocates both claim minority group status.277 Each side casts the other as the oppressor, “seeking to use the force of the state to stamp out belief systems with which they disagree.”278 In the words of one religious liberty scholar: “Religious minorities and sexual minorities . . . make essentially parallel and mutually reinforcing claims against the larger society.”279

This understanding is also reflected in legislative proposals to add sexual orientation to the list of prohibited bases for employment discrimination. Although religious employers generally must abide by laws forbidding discrimination on the basis of race, sex, national origin, age, and disability,280 they may discriminate on the basis of religion.281 The 2013 draft Employment Non-Discrimination Act (ENDA) would have put sexual orientation in the same class as religion, extending broad discretion to religious employers to discriminate on the basis of sexual orientation.282 As a result, several LGBT-rights organizations withdrew their support for ENDA, arguing that its religious exemption “essentially says that anti-LGBT discrimination is different—more acceptable and legitimate—than discrimination against individuals based on their race or sex.”283

NeJaime argues that this problem occurs because same-sex relationships are inappropriately understood as conduct, rather than status. Thus, gay rights are cast as liberties, rather than questions of equality.284 But the revised immutability does not provide a workaround for gay rights advocates. Rather, it supports the case for religious exemptions. One advocate of religious exemptions acknowledges that “for both same-sex couples and religious objectors ‘conduct is fundamental to their identity.’”285 The issue is framed as two minorities battling over conflicting fundamental choices: the convictions of conscience of religious minorities versus the committed intimate associations of same-sex couples. When the values at stake sound in the same register, accommodation seems more reasonable.286 If the right to same-sex marriage were conceptualized on a different order—for example, like the right to interracial marriage—the case for religious exemptions would lose traction.287

Thus, the new immutability is of limited persuasive value. Moreover, to the extent that legislatures and courts accept traits as immutable aspects of personhood, they may afford lesser protection to those traits than other attributes covered by antidiscrimination law.

III. applying the revised immutability to new contexts

This Part will apply the revised immutability to current controversies in discrimination contexts at the borders of existing legal protections, including employment discrimination on the basis of characteristics that the old immutability is thought to omit: weight, pregnancy, and criminal records. It argues that, due to the objections raised in the previous Part, the new immutability fails to capture the wrong of these forms of discrimination and may be counterproductive as a strategy for advancing protection. Weight is unlikely to meet the test of the new immutability because many do not regard it as important and it is often the source of stigma. Additionally, courts resist the argument that weight is fundamental to personality because the same could be said of most aspects of appearance. The argument that pregnancy is a choice that is integral to identity risks perpetuating stereotypes about women’s roles, has limited persuasive force due to the difficulty of explaining why pregnancy is a more important choice than other life pursuits, may result in limited rights to privacy rather than support, and invites conflicts with religious and business interests. The new immutability does not offer any obvious arguments against the use of criminal background checks to screen out job applicants, and may even promote discrimination against ex-offenders by suggesting that criminal conduct is a fundamental personality trait.

A. Weight

According to survey data, weight is one of the most prevalent bases for discrimination; weight discrimination occurs in interpersonal, employment, healthcare, and educational settings.288 Yet weight is a category at the borders of antidiscrimination protection.289 Typical objections to protecting against weight discrimination echo concerns underlying the old immutability: that those who are accountable for their own status should not be protected,290 and that discrimination might send “pedagogical” signals of disapproval of excess weight.291 Tellingly, one opponent of weight discrimination legislation objected to prohibiting discrimination “based merely on ‘weight’—as if weight were immutable and worthy of protected status on par with an individual’s race or sex.”292

Similar concerns have prompted courts to interpret existing antidiscrimination statutes narrowly to exclude weight discrimination. The ADA, which prohibits discrimination based on disability, is a potential source of legal redress for weight discrimination.293 The ADA defines a disability as “a physical or mental impairment that substantially limits one or more major life activities.”294 But even if a condition is not substantially limiting, a person may qualify as “disabled” under the statute if she is “regarded as having such as impairment.”295 In 2008, Congress passed the Americans with Disabilities Act Amendments Act of 2008 (ADAAA), which clarifies that to be “regarded as” having a disability under the ADA, a plaintiff must prove only that she was perceived as having a “physical or mental impairment” that is not “transitory and minor.”296 Impairments are defined broadly to include “[a]ny physiological disorder or condition, cosmetic disfigurement, or anatomical loss affecting one or more of the following body systems: neurological; musculoskeletal; special sense organs; respiratory, including speech organs; cardiovascular; reproductive; digestive; genitourinary; hemic and lymphatic; skin; and endocrine . . . .”297

The issue of whether obesity counts as an “impairment” was disputed both before and after the ADAAA. The EEOC has offered the guidance that impairments do not “include physical characteristics such as . . . weight . . . that are within ‘normal’ range and are not the result of a physiological disorder.”298 In litigation, the EEOC has taken the position that obesity is an impairment when either (1) the plaintiff’s weight falls outside the “‘normal range” or (2) the plaintiff has proof that her weight has a physiological basis.299 Some courts, however, have held that regardless of how far a plaintiff’s weight deviates from the norm, it is not a disability unless it has an underlying “physiological cause.”300

The “physiological cause” requirement is best explained by the logic of the old immutability. The intuition is that weight discrimination is unfair only if a plaintiff’s weight is a function of her biology and a result of chance, not choice.301 While science increasingly suggests “hormonal factors, metabolism, and genetics are all factors that predetermine one’s weight and impede attempts to lose it,”302 the causal force of such factors in any particular individual’s obesity can be difficult to prove.303 Perhaps sensing a problem with the harshness, intrusiveness, and stigmatizing nature of a rule requiring a plaintiff to provide expert medical testimony showing she is not to blame for her weight, judges do not always state the premise of the physiological cause requirement explicitly. Rather, they grope for other arguments to explain their results. In EEOC v. Watkins Motor Lines, the Sixth Circuit held that the line between true impairments and all other “‘abnormal’ . . . physical characteristics” had to be drawn by a physiological basis requirement.304 The court hypothesized that covering obesity without a physiological cause would lead down the slippery slope to providing coverage of, “for example, someone extremely tall or grossly short.”305 But the court did not explain how the physiology requirement draws a meaningful line. Height is indisputably a function of physiology, and the ADA often applies in such contexts.306

Some courts go so far as to hold that obesity is not “regarded as” an impairment absent evidence the employer believed the plaintiff’s weight was caused by a physiological condition.307 In Spiegel v. Schulmann, Spiegel, a former karate instructor, sued his employer, Schulmann, alleging discrimination on the basis of weight.308 The district court held that, as per Spiegel’s complaint, his weight was not regarded as an impairment by his employer:

Schulmann explained to Spiegel at considerable length his views about overweight people. Defendant Schulmann told him that the fact that he was fat demonstrated that he had no self-esteem and was a weak person. As such, Schulmann thought Spiegel could not be a proper role model for others. It was clear that it was not Schulmann’s view that Spiegel was physically unable to teach karate, at least at the beginner level. Rather, it was simply his view that fat people are essentially undisciplined and weak, and therefore cannot be in a role in which others are supposed to look up to and respect them.309

The court dismissed the complaint for failure to allege that the employer “believed that Spiegel’s weight condition was the symptom of a physiological disorder.”310 Thus, an employer’s stereotyped views about the causes and correlates of obesity inoculated his discrimination from legal attack.311

What would weight discrimination look like if considered under the rubric of the new immutability rather than the old? The new immutability may ease the requirement that a trait be unchangeable, and ask instead whether the change would be very difficult or traumatic.312 Legislatures, the public, and judges are more likely to conclude that obesity meets this requirement.313 But it is less likely weight would be considered “highly resistant to change” for workers penalized for exceeding weight requirements by small amounts,314 such as flight attendants315 or NFL cheerleaders.316

A trait’s resistance to change alone does not make it immutable in the new sense.317 In addition, the trait must be central to personality, or its expression a protected liberty. In a provocatively titled article, Yofi Tirosh calls this The Right To Be Fat, and argues that “American law’s current constitutional commitments to liberty, autonomy, and human dignity entail that it legally recognize the right to be of any body size.”318 This argument also has an antipaternalistic bent—that even if being overweight is a poor choice, it is one reserved for each person to make, without interference from the government and employers.319

Yet obesity is a condition that is regarded as both inessential and stigmatized, falling through the cracks of the revised immutability.320 Body weight seems an unlikely candidate for inclusion in a list of traits thought central, fundamental, or integral to identity.321 Those seeking to lose weight are particularly unlikely to advocate for this characterization.322 As Tirosh describes in ethnographic detail, “[m]ost people perceiving themselves as fat experience the center of gravity of their identity in their imagined, post-transformation future.”323 They feel that excess weight is blocking them from realizing their true selves, not that weight is constitutive of their authentic identities. To say that weight is fundamental to personhood would strike many as deeply offensive, reinforcing stereotypes that weight entails certain personality traits. Many advocates of “fat acceptance” are committed to a paradigm that considers weight as a characteristic that is, and should be, irrelevant to individual identity.324 Some of these advocates expressly seek to deemphasize the preoccupation with weight, arguing size is a poor measure of personality, health, and beauty.325

The concepts of personhood and liberty are unlikely to persuade in this context because they lack limiting principles. Courts see expansion of discrimination law to cover obesity as eroding employer prerogatives to discriminate on the basis of all other bodily traits and practices that are meaningful to individuals, such as muscle tone, smoking, alcohol and drug use, piercings, tattoos, dress, and grooming.326 Tudyman v. United Airlines is one example.327The plaintiff, an “avid body build[er],” was denied a job as a flight attendant because he was fifteen pounds over the airline’s maximum weight for flight attendants of his height and sex, due to his “low percentage of body fat and high percentage of muscle.”328 Under United’s policy, an employee who exceeded the weight maximum would be given an opportunity to lose weight or would be terminated.329 The policy allowed the airline’s medical staff to make exceptions, but they declined to make one for William Tudyman, “as his condition was voluntary and self-imposed.”330 The Tudyman court found that Congress had no intention of covering this sort of condition as an impairment: “What plaintiff is really suing for is his right to be both a body builder and a flight attendant, a right that [the statute] was not intended to protect.”331

In another case, Powell v. Gentiva Health Services, Inc., the plaintiff, who worked selling hospice services, argued that she had been terminated because her employer regarded her as obese.332 As evidence, Powell pointed to a comment by her supervisor during a performance review that Powell’s “dress and appropriateness was not up to par, and that she wasn’t even going to discuss the weight issue at this time.”333 Assuming the supervisor “viewed Powell’s appearance as a whole (clothing, accessories, weight) as negatively affecting her sales performance,” the court held that the statement was nevertheless insufficient to prove the supervisor “perceived Powell’s weight to constitute a physical impairment.”334 The court offered the following analogy to illustrate its point:

[S]uppose plaintiff wore her hair in a neon green mohawk. Such an unconventional hairstyle choice might be viewed as unprofessional, and might well impede her efforts to sell hospice services . . . but it obviously is not a physical impairment. The same goes for weight. An overweight sales representative may have difficulty making sales if the prospective customer perceives her appearance to be unprofessional, but that does not render her weight a “physical or mental impairment” within any rational definition of the phrase.335

Employment discrimination law seeks to balance employer prerogatives against the rights of employees to avoid invidious discrimination. Without more, arguments based in claims to “personhood” and “liberty” seem to open the floodgates to eliminate too much employer discretion.

In sum, harsh, intrusive, and stigmatizing judgments about individual responsibility for weight operate in the background of judicial opinions and public discussions regarding obesity discrimination. But the new immutability does not provide the conceptual resources to overcome these judgments. Rather, it leaves out weight as an inessential and stigmatized condition. Moreover, the argument that weight is fundamental to personality is likely to find resistance because the same argument can be made of appearance in general.

B. Pregnancy

Immutability arguments also limit antidiscrimination protection for pregnancy. In a 1976 decision, General Electric Co. v. Gilbert, the Supreme Court distinguished pregnancy discrimination from prohibited sex discrimination, based in part on the argument that pregnancy is a choice.336 Gilbert was a Title VII challenge to an employer’s exclusion of benefits for pregnancy-related disabilities from its disability insurance plan.337 The Court did not equate this exclusion with sex discrimination, because “[w]hile the [group of pregnant women] is exclusively female, [the group of nonpregnant persons] includes members of both sexes.”338 The Court further explained:

[A] distinction which on its face is not sex related [such as a pregnancy-based distinction] might nonetheless violate the Equal Protection Clause if it were in fact a subterfuge to accomplish a forbidden discrimination. But we have here no question of excluding a disease or disability comparable in all other respects to covered diseases or disabilities and yet confined to the members of one race or sex. Pregnancy is, of course, confined to women, but it is in other ways significantly different from the typical covered disease or disability. The District Court found that it is not a “disease” at all, and is often a voluntarily undertaken and desired condition.339

Thus, because pregnancy is “voluntarily undertaken and desired,” it is not immutable in the old sense and may properly be excluded from an employer’s health plan.340 The law seldom construes pregnancy as an injury resulting from misfortune, the exception being in cases of sexual assault.341 This reasoning rests on the normative judgment that “pregnancy’s character as a choice legitimate[s] the attribution of the costs of reproduction to the private family.”342

In response to Gilbert, Congress amended Title VII with the Pregnancy Discrimination Act (PDA) of 1978.343 The PDA provides:

The terms “because of sex” or “on the basis of sex” include, but are not limited to, because of or on the basis of pregnancy, childbirth, or related medical conditions; and women affected by pregnancy, childbirth, or related medical conditions shall be treated the same for all employment-related purposes . . . as other persons not so affected but similar in their ability or inability to work . . . .344

Courts have limited the reach of the PDA by reasoning that pregnancy is not immutable because it is impermanent, and therefore discrimination based on a planned or past pregnancy may not be prohibited.One court has noted, in an oft-cited passage:

Pregnancy . . . differs from most other protected personal attributes in that it is not immutable. While some effects of pregnancy linger beyond the act of giving birth, at some point the female employee is no longer “affected by pregnancy, childbirth, or related medical conditions,” for purposes of the PDA.345

Using this rationale, courts distinguish between prohibited pregnancy discrimination and permissible discrimination based on parental status.346 Along these lines, some courts interpreting the PDA have refused to include in the protected class women who plan to become pregnant.347 Others hold that employer-provided health insurance plans need not include fertility treatments.348

Lurking behind these decisions may be the unstated assumption that certain reproductive choices are voluntary decisions deserving of moral judgment. The Eighth Circuit has held that employers may refuse to cover contraception, so long as it is not covered for men (condoms) and women (birth control pills), reasoning that contraceptive services are different in kind from other forms of preventative healthcare.349 The court took this difference for granted, declining to explain any rationale for the distinction.350 The thinking may be the unstated assumption that women who choose to engage in nonprocreative sex should not ask the workplace to fund that pursuit.

Immutability arguments also undergird the idea that the workplace has no duty to accommodate the incidents of pregnancy. Despite Congress’s repudiation of Gilbert, lower courts continue to reason that “pregnancy is a voluntary condition and women bear the burdens of that choice if becoming a mother is not compatible with the rigors of the workplace.”351 Courts have interpreted the PDA to bar discrimination on the basis of stereotypes or generalizations about pregnant workers,352 but not to require that employers provide pregnant workers with accommodations—such as excusing tardiness resulting from morning sickness—absent proof that nonpregnant employees received the same accommodations.353

The ADA, unlike the PDA, expressly requires reasonable accommodation of disabilities,354 defined as impairments that substantially limit a major life activity.355 Consistent with Gilbert’s classification of pregnancy as a voluntary condition rather than an impairment,no court has ever held that pregnancy, as such, is a disability under the ADA.356 The EEOC has interpreted the term “disability” to include pregnancy-related impairments that substantially limit a major life activity, such as gestational diabetes.357 But courts have consistently denied coverage to women experiencing more “typical” pregnancy-related conditions, such as the inability to lift heavy objects,358 fatigue,359 or nausea.360 Since the ADA’s 2008 amendments, which overrode judicial interpretations of that statute as requiring that impairments be permanent or long term,361 courts have recognized more pregnancy-related conditions as disabilities.362 Nonetheless, the EEOC continues to maintain that pregnancy itself is not a disability, and the amendments did not expressly repudiate judicial interpretations of disability as excluding the “normal” incidents of pregnancy.363

In interpreting the PDA’s requirement that pregnant workers “shall be treated the same for all employment-related purposes . . . as other persons not so affected but similar in their ability to work,”364 courts draw on notions of mutability as individual responsibility. In Young v. United Parcel Service,the Supreme Court held that a fact finder might infer pregnancy discrimination if an employer denied accommodations (such as exemptions from heavy-lifting requirements) to pregnant workers while granting those accommodations to nonpregnant workers with similar limitations.365 The Court set forth a modified version of the McDonnell Douglas test366 for such cases rather than any per se rule.367

Prior to the Young decision, some courts imposed an implicit requirement in such cases: that pregnant plaintiffs demonstrate that they were just as deserving of workplace support as those nonpregnant workers who received accommodations. This requirement was evident in the distinction courts drew between, for example, pregnant workers and workers who received accommodations for on-the-job injuries. Several circuits held that an employer who provides accommodations for on-the-job, but not off-the-job, injuries need not accommodate pregnancy under the PDA.368 The language of the PDA asks only whether workers are “similar in their ability to work” without reference to the origins of their incapacities.369 But these courts saw on-the-job injuries as of a different kind than pregnancy. The workplace is thought to be, on some level, accountable for on-the-job injuries, while pregnancy is seen as a personal choice, wholly external to employment.370

For example, consider the facts in Young. In that case,the plaintiff Peggy Young worked as a driver for the United Parcel Service (UPS) in a position that required her to lift packages weighing up to seventy pounds.371 When she became pregnant, her doctor informed her that she could not lift more than twenty pounds for the first twenty weeks of her pregnancy and no more than ten pounds for the remainder.372 UPS denied Young’s request for an accommodation to do light-duty work, even though UPS offered temporary accommodations to certain other workers, including those with on-the-job injuries, those with ADA “disabilities,” and those whose driver’s certificate had been revoked by the Department of Transportation (DOT).373 But UPS argued that it did not accommodate all nonpregnant workers with disabilities.374 For example, UPS would not have accommodated a worker who suffered minor injuries “while picking up his infant child” or during “off-the-job work as a volunteer firefighter.”375 In this circumstance, formalistic reasoning failed to provide an answer to the question of whether pregnant workers had to be treated like the group of nonpregnant workers who were accommodated or if they could be treated like the group who were not.376

The Fourth Circuit assumed, without explanation, that pregnancy was different in kind from on-the-job injuries, disabilities under the ADA, or loss of DOT certification, and appropriately grouped with limitations stemming from childcare or volunteer work. The explanation may relate to judgments about pregnancy as mutable. In its recitation of the facts, the Fourth Circuit saw fit to point out that Young underwent three rounds of IVF before becoming pregnant in 2006, as if to emphasize that her pregnancy was not a chance occurrence,377 and that her employer had the beneficence to grant her leave for these procedures.378 The Fourth Circuit also noted, as “backdrop,” that the light-duty accommodations provided for other workers were part of the applicable Collective Bargaining Agreement.379 The Fourth Circuit expressed the concern that nonpregnant workers would have to bear the costs of their coworkers’ pregnancies, as though accommodation would upset the workplace bargain.380 Implicit here is a moral prioritization of contractual entitlements over pregnancy accommodations.381 Although the Supreme Court reversed the Fourth Circuit’s opinion, it did not clarify what types of reasons might be legitimate to distinguish between workers disabled by pregnancy and workers disabled by other conditions; it only precluded arguments based on the cost and inconvenience of accommodating pregnant workers.382

What if pregnancy were viewed through the lens of the new immutability as a protected choice or liberty, one essential to self-determination? The choice to become pregnant, on this theory, would be protected because becoming a parent is central to a person’s conception of her identity, like a person’s sexual orientation.383 Unlike obesity, which is almost invariably constructed negatively in contemporary U.S. law and culture, pregnancy is almost invariably constructed as “a wonderful, life-affirming, overwhelmingly good event in the life of the woman.”384 In law, culture, and media, pregnancy may be recognized as “physically taxing, occasionally painful, and frequently burdensome,” but “the experience of pregnancy remains, at the end of the day, a good thing.”385 Even pro-choice advocates describe abortion as “always a tragic situation,” necessary only “because a woman’s pregnancy may occur during a time when she is incapable of taking pleasure in its inherently wondrous nature.”386 Pregnancy as protected choice would seem, then, to be a firm basis on which to ground a right against discrimination. Moreover, most people would agree that, although it is possible for a woman to terminate her pregnancy, it would be abhorrent to ask her to do so.387 Equal protection doctrine has long recognized the right to reproduction as a fundamental liberty.388

Yet this view reinforces stereotypes.389 The trouble with personhood arguments for pregnancy is that they are “nothing other than a corollary to the insistence that motherhood, or at least the desire to be a mother, is the fundamental, inescapable, and natural backdrop of womanhood against which every woman is defined.”390 Such a framing may perpetuate sexist ideologies by imagining the plot of a woman’s life story as centering around her decision to have, or not to have, children. “[T]he idea that a woman is defining her identity by determining not to have a child is the very premise of those institutionalized sexual roles through which the subordination of women has for so long been maintained.”391

The new immutability is also of limited value here because it is likely to be characterized as supporting only negative rights against direct interference.392 The law may recognize pregnancy’s value for the pregnant woman, but refuse to require the public, employers, or other workers to “sacrifice” to accommodate her pregnancy. The new immutability’s claim for recognition casts pregnancy as special, while the claim for redistribution argues that pregnancy is not meaningfully different from other potentially disabling conditions. More specifically, the new immutability is about protecting the individual from the trauma of being asked to change a trait fundamental to her identity and the result of her private choice. By staking protection on a characterization of pregnancy as a private choice with benefits for the individual woman, the new immutability may lend support to arguments that women alone should bear the costs of pregnancy—the very arguments that employers used to lobby against the PDA.393

This logic goes hand-in-hand with a certain judgmental view of sexual morality.394 As Shari Motro has written, “[m]any people believe that sexual freedom comes with responsibility for the consequences. A woman who engages in sexual relations assumes the risk that she might conceive.”395 A woman’s “unilateral decisionmaking power” over whether to have an abortion is counterbalanced by her unilateral responsibility for the costs of pregnancy.396 Feminist and anti-racist legal scholars have long described how public assistance programs demonize women who are unable to earn enough income or find a male breadwinner to support their children, for example, by capping welfare payments to women who conceive while receiving public assistance.397

Even assuming that the argument that reproductive rights are negative liberties could be surmounted, another limitation of the new immutability in the pregnancy context is the line-drawing problem.398 Not just the state of being pregnant, but the very choice to have children could be thought central to the development of one’s personality and family life, a fundamental exercise of liberty, and a decision no one should be asked by the workplace to change. This logic requires that employer-sponsored health plans cover contraception399 and fertility treatments.400 Courts are resistant to extending discrimination law that far. In Krauel v. Iowa Methodist Medical Center, a case holding that the PDA did not require coverage of IVF, the court saw no evidence of sex discrimination, overlooking an admission by the company’s vice president that the reason the company excluded coverage was “because too many women of child-bearing age were employed by IMMC and infertility treatments result in too many multiple births, thereby creating a financial burden on the Plan.”401 Employers also express concern about disruptions caused in the workplace by women having too many children.402

Moreover, the new immutability does not draw a principled line between childbearing and childrearing for both men and women,403 nor between caring for children and other types of care, or even other valuable life pursuits workers might engage in, off the job, besides childrearing.404 Any number of off-the-job life pursuits may very well be central to a person’s identity such that the individual considers them immutable features of her personality and would experience trauma if forced to give them up. At oral argument in Young v. United Parcel Service, Justice Alito asked whether UPS would allow an accommodation “if a UPS driver fell off his all-terrain vehicle . . . on the weekend and was unable to lift.”405 The implicit comparison behind this question is between limitations brought on by pregnancy and those caused by recreational pursuits. The Fourth Circuit provided more sympathetic examples: the “employee who injured his back while picking up his infant child” and the “employee whose lifting limitation arose from her off-the-job work as a volunteer firefighter.”406 We might distinguish some life pursuits as leisure and others as socially valuable caregiving, but we would have to look outside the logic of immutability to do so.407

Additionally, religious exemptions gain traction when the right to contraception is framed as a protected choice.408 In Burwell v. Hobby Lobby Stores, Inc., the Supreme Court concluded that closely held corporations were not required to comply with the Patient Protection and Affordable Care Act’s mandate to provide health insurance coverage for certain forms of contraception thought to cause abortion, because those forms of contraception violate the sincerely-held religious beliefs of those corporations’ owners.409 Under the Religious Freedom Restoration Act of 1993 (RFRA), a sort of strict scrutiny applies to government actions that substantially burden the exercise of religious freedom.410 The action must be justified by a compelling interest and use the least restrictive means of fulfilling that interest.411 In examining the compelling interest behind the contraceptive mandate, the majority opinion focused on “the constitutional right to obtain contraceptives,” and how this right might be impaired if women were required to share the costs of those contraceptives.412 The Court “grudgingly” assumed this to be a compelling interest, without so holding.413 But the Court concluded that the government had not found the least-restrictive means for enforcing this interest, because the government itself could create a program to fund the forms of contraception objected to by the women’s employers.414

The dissent criticized the breadth of this holding, asking whether, for example, a restaurant owner who “refused to serve black patrons based on his religious beliefs opposing racial integration” might opt out of antidiscrimination laws.415 The majority responded, “The Government has a compelling interest in providing an equal opportunity to participate in the workforce without regard to race, and prohibitions on racial discrimination are precisely tailored to achieve that critical goal.”416 The distinction is difficult to understand unless one grasps the nature of the compelling interest that the Court has articulated in the Hobby Lobby case: a right to obtain contraceptives, rather than a right to gender equality.417 The government can supply contraceptives, but the government cannot remedy race discrimination in employment by paying off the victims. Race discrimination occupies a different plane of consideration.

In sum, beliefs that pregnancy is voluntary and a result of morally fraught sexual conduct limit equality law’s reach in the pregnancy discrimination context. But arguments based on the new immutability—that reproductive choices are fundamental to personhood—do not avoid these judgments. Rather, they risk perpetuating stereotypes about women’s roles, they have limited persuasive force due to the difficulty of differentiating the value of pregnancy from childrearing and other life pursuits, and even if successful, they are likely to result in limited rights and invite demands for exemptions by religious and business groups.

C. Criminal Records

Arguments about mutability, defined as personal responsibility, undergird opposition to rules that would circumscribe employers’ ability to automatically decline job applicants with criminal records. A wide range of employers conduct background checks on job applicants to screen for criminal histories.418 An industry of commercial vendors collects criminal records in databases and provides employers with computerized reports on job applicants.419 The practice makes it difficult for many people to find employment. By one estimate, one in four American adults has some sort of criminal record.420 And African Americans and Hispanics are more likely to have been arrested or incarcerated by substantial margins.421 Whether higher rates of crime among minorities could explain these disparities is disputed; consider, for example, that whites and minorities engage in similar rates of drug possession and sales, yet minorities are far more likely to be convicted of drug offenses.422 Discrimination against ex-offenders in the workplace is compounded by racial discrimination. Sociologist Devah Pager has conducted well-known audit studies in which white and black testers applied for jobs with identical resumes, varying only in that sometimes the testers were given criminal histories.423 In one study, the white applicants received half as many callbacks when saddled with a criminal history, while the black applicants received only one third as many callbacks when saddled with a criminal history.424

Automatic exclusion of job applicants with criminal records arguably violates Title VII’s prohibition on race discrimination. Title VII forbids not only intentional discrimination but also employment practices that are “fair in form, but discriminatory in operation” on the theory of disparate impact discrimination.425 To prove a prima facie case of disparate impact discrimination, a plaintiff generally presents statistics showing that a particular employment practice has an adverse impact on the basis of a protected trait.426 In its 2012 guidance on the question, the EEOC took the position that national data suffices to demonstrate that employer exclusions based on criminal records have a disparate impact based on race.427 An employer, however, may defend against a charge of disparate impact discrimination by showing that the practice is “job related for the position in question and consistent with business necessity.428 The EEOC has advised that exclusions based on arrests alone are not eligible for this defense because “[t]he fact of an arrest does not establish that criminal conduct has occurred.”429 While a conviction does establish criminal conduct, the EEOC maintains that “[i]n certain circumstances . . . there may be reasons for an employer not to rely on the conviction record alone.”430 An employer’s policy may be justified if it includes, for example, some sort of individualized assessment, considering “at least the nature of the crime, the time elapsed, and the nature of the job.”431 Apart from Title VII, eighteen states have enacted legislation known as “ban the box” which prohibits certain employers from asking about criminal records on job applications, although employers are not precluded from considering criminal history at later stages in the hiring process.432

Employers screen applicants for a variety of utilitarian reasons. For example, they screen to prevent theft and fraud in the workplace, ensure the safety of other workers or customers, or avoid liability for negligent hiring.433 But this type of employer risk management runs counter to the public policy goal of reducing recidivism by encouraging ex-offenders to find employment.434 Setting aside the normative question of whether employers should bear the costs of public policy goals, there are empirical questions about the level of risk that ex-offenders pose to employers. Support for the specific claim that all those with criminal records are more likely to commit workplacemisconduct is lacking.435 Criminal history reports are likely to be inaccurate, especially those from commercial vendors.436 Many include incorrect information, such as convictions that have been expunged by court order.437 And some employers are screening out applicants with any type of criminal history, regardless of the severity or nature of the offense,438 whether the arrest led to a conviction,439 or how long ago it occurred.440 Criminologists generally agree most people who commit crimes eventually desist, and some research suggests that the likelihood that certain individuals who have not re-offended for a number of years will commit crimes may be close to that of those without criminal records.441 Other factors, such as youth, may correlate with crime, yet employers do not argue that the correlation is a basis for screening out younger workers.442

Immutability’s logic is exerting influence here. A criminal record may be immutable in the sense of being virtually impossible to change, considering the difficulty of expungement.443 But a criminal record may be mutable in the sense that the decision to commit a crime was within an individual’s control and responsibility. Opponents of the EEOC’s enforcement policy have made clear that ex-offenders are morally responsible for their lack of employment opportunities. One theme of opposition to the EEOC’s enforcement guidance is that it represents “the illegitimate expansion of Title VII protection to former criminals.444 A Heritage Foundation report argued, “[i]t is not racial discrimination that deprives felons, black or white, of their ability to obtain employment ‘but their own decision to commit an act for which they assume the risks of detection and punishment.’”445 A letter from nine states’ Attorneys General objected to the EEOC’s guidance by arguing that “[y]our real target appears to be the perceived unfairness of judging an individual—of any race—solely by his or her past criminal behavior.”446 An early district court opinion put it more bluntly: “If Hispanics do not wish to be discriminated against because they have been convicted of theft then, they should stop stealing.”447

Opposition to the EEOC policy is also rooted in the endorsement of employers’ moral judgments about ex-offenders. These judgments are phrased in terms of ex-offenders’ inherent lack of trustworthiness as a class. In one district court case, even though the employer’s business reasons for conducting background checks were not at issue, the judge found it necessary to begin his opinion with a lengthy explanation of the “obvious,” “rational,” and “legitimate” reasons for conducting criminal history and credit background checks, to avoid hiring those who “appear to be untrustworthy and unreliable.”448 Similarly, in their letter to the EEOC opposing its enforcement guidance, the states’ Attorneys General wrote that “[a] criminal background may . . . be indicative of a lack of dependability, reliability, or trustworthiness.”449

The new immutability—with its emphasis on psychological understandings of personality, its attempts to avoid the trauma of being asked to change a fundamental trait, and its exaltation of self-determination—does not suggest any obvious arguments for prohibiting discrimination on the basis of criminal records. The new immutability does not counter the moral judgments condemning ex-offenders.450 It is difficult even to find an example of an ex-offender staking a political claim that criminal records are fundamental to personality.451 Amy Myrick has offered a sociological account of the depersonalization that individuals often experience when inquiring about expungement or sealing of criminal records.452 Rather than viewing their criminal records as fundamental to their personalities, those confronting their records “felt reduced to pieces of personal information that did not represent a holistic identity, even a deviant one.”453 Myrick’s clients offered accounts of their own identities that “were antithetical to the records’ way of representing, or misrepresenting” them.454 These accounts have two recurrent features: emphasis on “personal changes over time” and assertion of “social identities of parent, worker, and property owner” rather than criminal.455 Somewhat like those overweight individuals who see their true, essential selves as thin, ex-offenders may see their true, essential selves as law-abiding.456

In this context, the new immutability reinforces stereotypes.457 If anything, the new immutability may introduce a type of “criminal essentialism” that lends support to employer arguments for excluding those with criminal records.458 Many psychologists assume that personality is continuous and reliable—in other words, immutable—and offenders are not likely to desist from criminal behavior.459 Advocates of background checks might concede that the underlying identity, that of a “criminal,” is fundamental—if not in the old sense of being impossible to change because it is determined by social, economic, and genetic factors, then in the new sense of being highly impervious to change, requiring the traumatic experience of remorse and repentance for one’s crimes, and giving up the life of a criminal.460 Such views lend support to employer arguments that ex-offenders are untrustworthy and that blanket hiring bans are sound practices. These arguments are bolstered by the assumption that criminality is immutable as a fundamental aspect of personality.

The new immutability may gain more traction if the question were reframed as protecting not the identity of the criminal but rather the identity of the ex-offender who has desisted from crime.461 Advocates of “ban the box” policies and other protections for ex-offenders use narratives of redemption and the rhetoric of second chances to promote their cause. Consider the following example from a report by the National Employment Law Project:

Like many, Darrell Langdon struggled with addiction in his youth. Now 52 and having raised two sons as a single father, Darrell, through his strength of character, has been sober for over twenty years. Although he has moved forward in life through hard-won rehabilitation, his 25-year-old felony conviction for possession of cocaine remains.462

This vignette appeals to a notion of fundamental character, a true self that is emerging through the hard work of rehabilitation and proven over twenty-five years of sobriety, a life redeemed through the socially valuable work of child-rearing.

But this is too much of a stretch for the new immutability, which is about characteristics so fundamental no one should be asked to change them. Instead, the argument from the new immutability is that even if he could have his conviction expunged, Langdon should not have to, because his conviction was a formative and important part of his life’s journey toward redemption. The problem with this claim, however, is that it casts criminal conduct as necessary rather than regrettable, a claim unlikely to have political purchase, and one that does not fit with the narrative of crime as an unfortunate aberration from an individual’s fundamentally good character. The new immutability would understand crime as constitutive of character.

As a strategic matter, arguments based on the new immutability also fall flat with respect to criminal records for other reasons. In contrast to the pregnancy and obesity examples, where the concern is that protection would lead down a slippery slope, ex-offenders are already at the bottom of the slope, the reductio ad absurdum of expansive concepts of the protected class. It seemed obvious to Judge Norris, when he expounded on immutability, that “discrimination exists against some groups because the animus is warranted—no one could seriously argue that burglars form a suspect class.”463

Thus, the revised immutability is only likely to strengthen resistance to legal rules that might require employers to perform more careful criminal background checks.


As this Part has demonstrated, courts impose implicit immutability requirements not supported by statutory text when interpreting employment discrimination law. Immutability concerns also feature in public opinion and political debates over executive enforcement of employment discrimination law and expansion of state and local laws to cover new forms of discrimination. While discrimination on the basis of weight, pregnancy, and criminal records is often considered job-related, arguments about job qualifications are infused with moralizing judgments about persons in the abstract. These moralizing judgments are problematic for all the reasons described by critics of luck egalitarianism. They rest on untenable assumptions about chance and choice, like the idea that pregnancy alone, among all potentially disabling conditions, is within an individual’s control and therefore need not be accommodated by the workplace. They are harsh, like the argument that ex-offenders should be automatically and permanently denied entry to the job market. They are intrusive, like demands that workers conform to ideal norms regarding body size. Employers may believe they are creating incentives for better behavior (weight loss, personal responsibility for procreation, law-abiding conduct). But shaming and blaming can backfire, causing social dysfunction as their targets cope with the assignment of spoiled identities.

Rather than providing an easy rebuttal, the revised version of immutability first obscures these moralizing judgments and then raises new barriers to protection. The idea that weight, pregnancy, or a criminal record might be an essential aspect of personhood, to be romanticized and protected, fails to resonate because many experience these conditions as negative, neutral, or immaterial to personality. Rather than countering stereotypes, the new immutability reinforces them by suggesting that such traits are essential aspects of personality. The new immutability may even lend credence to arguments that traits like obesity, pregnancy, and ex-offender status are good predictors of future behavior. Courts have difficulty imagining what distinguishes these traits from all other choices that individuals may consider fundamental. To the extent it might be accepted, the new immutability rationale provides a lower tier of protection than that afforded to classifications such as race and sex. In this lower tier, it becomes more difficult to argue that civil rights obligations might override employer cost concerns or conflicting “immutable” convictions such as religious objections.

IV. beyond immutability

This Part considers alternatives to revising the theory of immutability for those interested in expanding employment discrimination law to new forms of bias. It examines two potential approaches: (1) universalizing a rule that employers only require qualifications that are both job-related and reasonable, and (2) incremental expansion of targeted antidiscrimination protection through legislative, judicial, or employer prohibitions on additional forms of systemic bias. It argues that universal approaches, while theoretically appealing, may not be politically possible in any form that would effectively address inequality. Thus, it suggests efforts toward incremental expansion of antidiscrimination categories based on analogies between old forms of bias and new that do not rely on stretching the concept of immutability. In particular, it argues for consideration of whether biases create systemic limits on equal opportunity.464

A. Universalizing a Reasonable Relationship Requirement?

An often-suggested solution to the problem of identifying suspect classifications is to universalize antidiscrimination protections.465 Workers in the United States are not universally protected against unfair treatment. The default position in the United States is at-will employment, meaning employers may hire, promote, demote, compensate, or fire employees without cause.466 Likewise, employers are not generally required to accommodate the needs of workers on such matters as job duties, scheduling, or the physical workspace.467 Employment discrimination law, among other regulatory regimes, provides exceptions to these defaults, forbidding adverse treatment on the basis of enumerated traits such as race and sex,468 and requiring accommodations for workers with disabilities or religious commitments.469 Employment discrimination law contains defenses, for example, when an employer can show that a trait such as sex is a “bona fide occupational qualification reasonably necessary to the normal operation of that particular business” (BFOQ),470 when a policy with a disparate impact is “job related for the position in question and consistent with business necessity,”471 or when an accommodation would pose an “undue hardship.”472

Targeted approaches isolate certain identity traits as bases for legal interventions into employer decision making. Instead, we might imagine a universal model of workplace protection, in which employers would need to show cause for all personnel decisions. For example, the District of Columbia’s Human Rights Act was enacted “to secure an end . . . to discrimination for any reason other than that of individual merit.”473 Similarly, some proposed legal reforms would require employers to have a reasonable basis for rejecting any request for worker accommodation.474 These approaches would not single out particular groups, traits, or biases for scrutiny; rather, they would evaluate employment decisions based on their reasonableness in terms of business needs. Such approaches may be normatively superior to targeted expansion of discrimination law, but come with significant strategic disadvantages for those concerned about inequality.

As a normative matter, universal policies may be superior to focused interventions in that they guarantee important rights to liberty, dignity, and job security, raising the baseline of protection for all workers.475 Universal policies would eliminate the difficult hurdle of proving discriminatory intent, the cause of death for many employment discrimination cases.476 And they would include all forms of discrimination not justified by job requirements or employer necessity, including much discrimination on the basis of weight, pregnancy, and criminal records. Universal approaches to workplace protection may have political traction in that they expand the potential constituency for change from minority groups to all workers, avoid the characterization of employment opportunities as zero-sum, evade the problem of fatigue with identity politics, and eliminate the fraught task of defining the beneficiary class.477

Yet there are reasons to be concerned that universal protections may not eliminate discrimination as well as targeted approaches. In other work, I have called these problems “dilution” and “assimilation.”478 Dilution occurs when a right must be narrowed to be extended to more claimants, or when the expansion of rights strains the resources of enforcers, reducing protection for those who need it most.479 For example, the proposed Model Employment Termination Act, which would prohibit discharge of employees without “good cause,” would apply only to terminations, not hiring or other aspects of employment; would limit employees to streamlined arbitral procedures rather than litigation; and would offer workers fewer remedies than employment discrimination law.480 Legislative proposals to expand accommodations beyond protected traits may limit a worker’s right to that of requesting the accommodation and receiving consideration from the employer, without imposing a duty on the employer to accept reasonable requests.481 The assessment is that compromise is required to ensure judicial and political will for such broad disruptions of employer prerogatives.482

The assimilation problem is that universal solutions often have goals other than the disruption of biases and inequality.483 Equality, as a goal, is thought to be folded into other goals, such as providing universal job stability or protecting the liberties of all workers against employer intrusion. But in the enforcement and application of universal rules, equality norms may be obscured and subverted. Imposing a rationality or reasonableness requirement for certain personnel decisions is unlikely to weed out discriminatory practices because, in many cases, discrimination may be cost effective.484 Discrimination may allow employers to use statistically sound generalizations as efficient sorting mechanisms;485 to maintain a homogenous workforce that can be managed more efficiently;486 to cater to the preferences of clients, customers, or co-workers for certain types of employees;487 and to avoid the risk that minority employees will bring litigation over workplace conditions such as harassment, denial of promotion, or termination.488 Moreover, most proposals would only protect employees from unfair termination, rather than requiring fairness in hiring or in the terms and conditions of employment.489 Some analysis of the European experience suggests that without a commitment to antidiscrimination in hiring, job protection for incumbent employees locks members of subordinated groups out of opportunities.490

Simply requiring that employers evaluate potential employees based on “merit” would not call into question discrimination based on weight, pregnancy, or criminal records, insofar as employers and judges implicitly and explicitly consider these factors demonstrative of character, commitment, and integrity. Adding the requirement that discrimination be “job-related” would better address these biases, as an employer may find it difficult to convince a court that weight, for example, has any relationship to a sedentary back-office job. However, employer arguments gain plausibility if that office is in an image-conscious industry, or if the job is a highly visible one that requires that the employee inspire confidence.491 In employment discrimination law, courts reject the argument that employers must engage in sex discrimination to cater to customer preferences for men or women in certain jobs, reasoning that employers are merely acting as conduits for societal discrimination.492 Employment discrimination law requires scrutiny of sexist stereotypes by explicitly forbidding discrimination based on sex. A simple job-relatedness rule, by contrast, would not forbid an employer from catering to such prejudices. If the law does not delineate certain widespread biases as subject to scrutiny, it may not prompt critical examination of those biases.493

A law that imposes some sort of universal requirement, while also delineating prohibited bases for discrimination and not watering down remedies or resources, may be the most normatively desirable policy outcome for those concerned about employee rights generally as well as equal opportunity. Yet there are reasons to doubt whether such an approach is a viable strategy. Despite the strategic appeal of universal solutions in the abstract, there is little political support at present for a revolution in at-will employment.494 Joseph Fishkin has described such proposals as “non-starter[s]” because they “run[] roughshod over our law’s commitment to leaving employers substantial discretion over whom to hire, promote, and fire.”495 Recent history suggests changes to workplace regulation are more likely to be incremental than revolutionary, occurring through the expansion of existing categories by courts and the addition of new ones by legislatures, at subnational as well as national levels, and by private employers as well as governments.496

B. Targeting Systemic Biases

This Part proposes incremental expansion of employment discrimination law with the goal of targeting systemic forms of bias, rather than the goal of protecting immutable traits. This project has both legal and political dimensions. Legally, it could entail judicial recognition that systemic biases—such as those based on weight, pregnancy, and criminal records—fall within the categories of disability, sex, and race discrimination. Politically, it could entail the enactment of new rules by legislative bodies or employers that would prohibit certain forms of discrimination on the basis of additional enumerated biases. Rather than revising the concept of immutability, advocates might expose the moralizing nature of judgments about “mutable” traits that are the bases of systemic forms of inequality. This Part will begin by describing a systemic bias approach and discussing its merits relative to universal protections or arguments from immutability. It will then apply this approach to discriminatory practices based on weight, pregnancy, and criminal records.

By systemic biases, I refer to discriminatory practices that are both structural and pervasive.497 Structural approaches to employment discrimination are concerned with whether institutional practices contribute to unequal opportunity, rather than the guilt or innocence of particular types of victims or perpetrators. Structural accounts of discrimination locate the causes and consequences of inequality in social and institutional practices, arrangements, and systems.498 They change the focus from individuals and their choices to how workplace structures “contribut[e] to the production or expression of bias.”499 The structural approach’s focus on the workplace itself as the cause of inequality creates an argument for legal intrusion into the prerogatives of employers.500 By contrast, immutability arguments look to whether the victims of discrimination, considered as a class or group, deserve protection.501 Because they are understood as based on bad or costly “choices” (gluttony, sex, criminal behavior), certain traits receive limited or no protection. Victims of these forms of discrimination have lost their innocence, unlike those who did not choose their race or sex, and so the analogy to race or sex fails. The problem with analogical arguments that compare groups in this manner is that they “promote[] the idea that the traits of subordinated groups, rather than the dynamics of subordination, are the normatively important thing to notice.”502 Thus, a structural frame would put this Article’s question not as which classes should be protected, but which forms of bias the law should disrupt.

Pervasive forms of bias connect to larger social systems of hierarchy and segregation and contribute to broader problems of inequality.503 The focus on pervasive biases differentiates a targeted approach from a universal one and provides a limiting principle.504 Unlike isolated instances of workplace unfairness, pervasive biases substantially limit the opportunities of affected individuals.505 For example, “victims of sex discrimination will encounter it in workplace after workplace,”506 while a man who is fired because, for example, “he reminded the employer of the employer’s hated stepfather” is unlikely to ever encounter this same unreasonable prejudice again.507 But the difference is not just that a victim of sex discrimination has diminished prospects for finding another job: the problem is also that sex discrimination reinforces larger patterns of superficial prejudice, stereotyping, and stigmatization.508

Whether a bias is pervasive in the sense of being widespread might be demonstrated through quantitative measures.509 But qualitative measures are also important; for example, patterns of bias might be pervasive because they are self-reinforcing: discouraging those affected from pursuing opportunities and limiting the options available to them, or distorting judgments about those identified as group members.510 Or these forms of inequality might be complex in that they spill across more than one domain of social life, such as in employment, education, housing, the family, or politics.511 The focus on pervasive biases accords with the aim of disrupting wholesale patterns of discrimination that assign group-based statuses to individuals so as to limit their range of opportunities.512

This systemic bias approach is both provisional and open-ended. It is provisional because it is not the aim of this Article to provide a unified theory of protected traits. Rather, this Part aims only to sketch out a potential alternative to the revised immutability that might be more normatively attractive and politically feasible.

This approach is open-ended in that, beyond asking about whether systemic biases attach to particular traits, it does not take sides in debates about the forms of bias that equality law should prohibit. As previously discussed, equality law already recognizes that biases may be prohibited because, for example, they are generally superficial (judging on a basis that is not, or need not, be required for the job),513 stigmatizing (demeaning or subordinating based on identity),514 orstereotyping (making assumptions about roles and competences based on group status).515 This Article does not attempt to resolve debates among these theories; rather, it has argued that the revised immutability subverts the goals of eliminating each of these forms of bias,516 and thus employment discrimination law requires a more inclusive theory of discrimination.

Some might argue that immutability could be stretched to encompass systemic biases. Susan Schmeiser, in her analysis of the new immutability, applauds those courts she sees as asking not whether a trait is immutable, but whether discrimination on the basis of that trait has been immutable.517 She argues for a “reading of immutability . . . that turns not on the significance of individual self-definition or the question of volition, but rather on the persistence of ‘social and legal ostracism’ as the relevant aspect of group definition.”518 For example, Schmeiser discusses an Oregon state appellate decision holding that immutability is not about “the common, class-defining characteristics, but instead the fact that such characteristics are historically regarded as defining distinct, socially-recognized groups that have been the subject of adverse social or political stereotyping or prejudice.”519 Such a definition of immutability, however, has not had wide uptake.520 Moreover, just as it is problematic to argue that human traits are immutable, so it is problematic to argue that biases are immutable. This line of argument may be self-defeating by suggesting that equality law is futile: if discriminatory assignments of identity and practices of segregation and subordination are immutable, the law cannot disrupt them.

To argue that employment discrimination law should forbid more forms of systemic bias is not to say an employer can never discriminate on these bases. Rather, to discriminate on the basis of a forbidden trait, an employer must demonstrate a business reason sufficient to meet a statutory standard or other exception.521 Absent such a defense, moralizing judgments that an employee was to blame for her own deficiency should not excuse employer actions that perpetuate systemic problems of social inequality. Employment discrimination law seeks to balance employer prerogatives against the aim of eradicating invidious forms of bias. The law is justified in intervening in employer judgments with more force where those judgments cause social problems.522 The focus on pervasive biases limits a systemic approach to those forms of bias that severely curtail opportunities, cut across social domains, or are self-perpetuating for those defined by certain traits. These forms of workplace unfairness create larger problems of inequality and make a stronger case for intervention in employer prerogatives.

It is true that more expansive antidiscrimination laws may trade off with the rigor of enforcement or the extent of remedies.523 Some might argue that immutability should serve as a proxy for measuring whether a group deserves priority in terms of antidiscrimination protection. Yet, as this Article has argued, the moral judgments underlying considerations of immutability are problematic. Rather than asking impossible questions about the relative moral desert of various groups, equality politics might focus on the harms to society when the labor market subdivides workers into rigid groupings based on superficial, stigmatizing, or stereotypical categories. Systemic problems of inequality have a stronger claim on limited enforcement resources and remedies. Moreover, stratification based on “immutable” characteristics like race may be impossible to address without attention to widespread discrimination on the basis of “mutable” characteristics, such as ex-offender status.

By contrast to the universal approach, an incremental approach to expanding antidiscrimination law may be more politically feasible. American civil rights law has often expanded by establishing analogies to and overlaps with discrimination based on race, sex, disability, and increasingly, sexual orientation.524 Analogies “can inspire empathy and understanding of harms previously unrecognized, and they may be desirable, if not necessary, in an adjudicative system based upon fidelity to precedent.”525 To suggest the power of careful analogical arguments is not to insist on the equivalence or ranking of oppressions.526 Rather than arguing that the common thread is immutable traits, in any sense of the term “immutable,” advocates might argue that biases on the basis of traits such as weight, pregnancy, and criminal records perpetuate systemic inequality, and that the arguments against protection are moralizing.

Looking at the social dynamics behind weight discrimination reveals a kinship with other forms of disability discrimination527 and substantial overlap with sex discrimination.528 Obesity is a stigmatized condition with systemic implications for employment opportunity,529 yet it often falls outside the definition of disability due to immutability concerns.530 One notable victory against weight discrimination is a 1993 First Circuit decision, Cook v. Rhode Island, Department of Mental Health, Retardation & Hospitals.531In that case, the court addressed immutability head-on and found it irrelevant. It rejected arguments that “morbid obesity” was not an impairment on account of being “caused, or at least exacerbated, by voluntary conduct.”532 This was because the statute

contains no language suggesting that its protection is linked to how an individual became impaired, or whether an individual contributed to his or her impairment. On the contrary, the Act indisputably applies to numerous conditions that may be caused or exacerbated by voluntary conduct, such as alcoholism, AIDS, diabetes, cancer resulting from cigarette smoking, heart disease resulting from excesses of various types, and the like.533

Rather than appealing to an alternate concept of immutability as a protected realm of liberty, the court highlighted the hypocrisy of treating weight differently than other conditions brought on by voluntary conduct. It instead directly confronted the social stigma surrounding weight, ending with disapproval for “a society that all too often confuses ‘slim’ with ‘beautiful’ or ‘good.’”534 The European Court of Justice has addressed weight discrimination in a similar manner.535

Likewise, failure to accommodate pregnancy and related conditions has systemic effects on the employment opportunities of women.536 “[T]he maternal wall—the barriers to employment equality faced by mothers—begins with pregnancy.”537 Although the time during which a worker is pregnant is a relatively short period in the span of her career, the effects of pregnancy discrimination have a long-lasting impact.538 This form of discrimination is self-perpetuating: employers discriminate based on the stereotype that women will be less devoted to their jobs due to family responsibilities, resulting in fewer employment opportunities for women and creating incentives for women to devote themselves to family responsibilities rather than paid work.539

With respect to pregnancy, arguments based on the intersections between reproduction and sex equality, rather than any sort of immutability, have had some success in courts. In International Union, United Automobile, Aerospace & Agricultural Implement Workers of America v. Johnson Controls, Inc., the Supreme Court struck down as unlawful sex discrimination an employer’s “fetal-protection policy” that forbade fertile women, but not fertile men, from working in certain jobs involving hazardous lead exposure.540 What doomed the company’s policy was its distinction between men and women, not its distinction on the basis of “fertility alone.”541 Unlike rights to reproductive privacy or liberty, the interest in sex equality was strong enough to override employer arguments that discrimination was necessary to avoid the risk of costly tort liability.542

Rather than allowing immutability arguments to remain submerged in sex discrimination contexts, courts might directly address them. In one district court case, Erickson v. Bartell Drug Co., an employer argued expressly that contraceptives are “voluntary” and “not truly a ‘healthcare issue.’”543 The district court called this immutability point out as “[a]n underlying theme” of the employer’s argument and rejected it, reasoning that “the availability of affordable and effective contraceptives is of great importance to the health of women and children because it can help to prevent a litany of physical, emotional, economic, and social consequences.”544

Likewise, with respect to criminal records, a focus on the harshness of immutability arguments, the racially disparate impact of screening practices, and the systemic effects of those practices has been persuasive with lawmakers. In analyzing the reasons that state legislatures enacted ban-the-box legislation, Fishkin describes how advocates directly confronted arguments regarding personal responsibility, exposing how these arguments are harsh and stigmatizing.545 For example, Philadelphia Mayor Michael Nutter emphasized that “people who ‘have paid their debt to society’ deserve ‘an opportunity to work to provide for their families and should not be discriminated against before they even have a first interview.’”546 Immutability concerns clash with the goal of

open[ing] up a wider range of life paths and opportunities not only to those who demonstrate particular merit, desert, or promise, but to everyone—including those who have done poorly and those who did not manage to do as much as one would hope with the opportunities that were available to them.547

Fishkin also concluded that the disparate impact of criminal background checks on racial minorities played an integral role, by engaging anti-racist activists and organizations such as the NAACP to work for reforms.548 When the question is the moral desert or freedom of ex-offenders as a class, it is hard to make the case for imposing the additional costs of individualized assessments on employers. The case is much stronger if the problem is envisioned as systemic inequality: a potential future in which all employers automatically exclude every ex-offender, creating a large, permanent underclass of individuals with criminal records, disproportionately people of color, unable to find any employment.549


The new immutability has been useful for courts seeking to overcome doctrinal hurdles to protection against sexual orientation discrimination. But analysis of its potential applications to employment discrimination contexts reveals that the revised immutability is deeply flawed as a way of rethinking equality law. The new immutability is focused on determining whether individuals have made choices that ought to be protected aspects of their “personhood,” rather than asking how workplace policies limit equal opportunity by perpetuating systemic biases. Such biases may include the ideas that thin is always good, criminals are always bad, and pregnancy is always special. The promise of employment discrimination law is its ability to disrupt the stereotypes, stigmatizing practices, and superficial judgments that contribute to systems of inequality. This exercise will ultimately require more empathy and understanding, not revisions of the theory of immutability.