The Failure of Mixed-Motives Jurisprudence

Introduction

How should we judge people who act for both good and bad motives? For example, in 1971, an Ohio school district fired Fred Doyle from his job as a public school teacher. The district had some good reasons: Doyle fought with coworkers and made obscene gestures at students.1 However, the district also admitted to retaliating against Doyle for talking to the media about matters of public importance.2 Given the school district’s mixed motives, did it break the law in firing Doyle?

In Mt. Healthy City School District Board of Education v Doyle,3 the Supreme Court’s answer was this: judge people by their good motives if their good motives would have been enough; judge people by their bad motives only if they would have acted differently but for the bad motives. Mt. Healthy therefore announced a “But-For standard” for mixed motives.4

Given the complexity of human nature, courts must often evaluate mixed-motives cases,5 and Mt. Healthy’s But-For standard has proven distinctly influential. Eight-thousand five-hundred federal court opinions, including sixty-five Supreme Court opinions, cite Mt. Healthy.6 Its But-For standard is used to evaluate mixed motives in numerous areas of the law, such as employment and labor law;7 securities market manipulation claims;8 landlord-tenant disputes;9 Fourteenth Amendment Equal Protection claims, such as those demanding racially integrated juries;10 and First Amendment suits demanding viewpoint neutrality11 or government workers’ expressive rights.12 So great is its influence that many courts conflate this particular standard with the broader category of standards of which it is a member: a mixed-motive analysis just is a but-for analysis.13

The But-For standard is also the one that scholars invoke, almost instinctually, in law reform projects.14 With limited exceptions,15 the But-For standard has been widely praised or recommended.16

It is understandable why the But-For standard might seem attractive. Such a standard prevents overcompensation of plaintiffs who would have suffered anyway because the standard ignores bad motives that made no difference to anyone’s conduct; no harm, no foul. The But-For standard also prevents liability for mere stray thoughts and limits intrusive judicial second-guessing. Humans are fallible, and the law must make some allowance for our impure thoughts, so long as they are causally inert. If Fred Doyle was a danger to his colleagues and students, the But-For standard lets the school district get him out of the classroom.

Despite the superficial plausibility of such justifications, this Article argues that the But-For standard does not deserve its laurels. To the contrary, widespread acceptance of the But-For standard represents a great failure of the jurisprudence of mixed motives.

Part of my case is a matter of intellectual history. The most familiar arguments for the But-For standard are of such recent and disreputable provenance that they do not deserve the deference that comes with long use.17 Another part of my case is to point out how the But-For standard has normatively unattractive implications: it does not guide courts with difficult decisions and allows them to conceal their real reasoning.18 It also silences important voices, preventing the airing of important grievances, because the operation of a But-For standard ends up focusing much more on the exculpatory good motives than the inculpatory bad ones.19

Yet the main problem with the But-For standard, and the main thrust of this Article, is that the But-For standard does not actually serve the goals its proponents care about. It seems to prevent windfalls when in fact all it does is allocate windfalls.20 Relatedly, the But-For standard seems to protect individuals from excessive judicial meddling, but other mixed-motives standards make far better safe harbors.21 A judge interested in reducing windfalls or intrusion should instead select an alternative motive rule.

With no strong policy argument to support the But-For standard, what remains is a bare concern for causation. The intuition is that, if a given motive is not a but-for cause of an action, then the motive cannot matter. This reasoning simply misunderstands black letter tort law on causation. In cases of multiple sufficient causation (like two raging fires converging on the same cabin), the law regards several factors as causes even though no one is a but-for cause.22 Any analysis that likens motives to causes would not limit itself to a But-For standard, if only because tort law does not limit itself to but-for causes.23

More fundamentally, arguments from causation often involve a mistaken inference: if causation is important, then motive-causation is important. Yet a commitment to imposing liability only when a defendant caused a plaintiff’s injury does not entail that every aspect of the defendant’s action must have played a causal role. Batterers are liable for battery if they cause nonconsensual injury, but lack of consent need not have made any causal contribution. As a whole, actions with motive as an element may need to cause injury, but we rarely see arguments for why motives must be held to causal standards.

Given the problems with the But-For standard, why has it proven so influential? Bad standards persist, and bad arguments go unexamined, when people are not aware that there are alternatives. Previous research in this area was premised on the idea that material advances in our motives jurisprudence would be possible once we learned how to talk and think clearly about motives.24 This Article is the first normative application of that agenda, identifying an important judicial standard and then testing it against alternatives and objections.

This Article is structured in five Parts. Part I introduces the But-For standard and mixed-motives analysis generally. It provides a vocabulary for describing mixed motives and a way to think about them. That Part is an applied summary of previous scholarship, so those readers familiar with the recent literature can pass swiftly through it. Nevertheless, the goal of Part I is not just preparation. Rather, learning how to talk and think about motive standards is itself a tacit attack on the primacy of the But-For standard; noticing that the law has many options for motive standards loosens hold of the But-For standard as essential to any particular area of law. If there are many options, one must ask what justifies any given option.

The next three Parts examine candidate justifications. Part II considers whether the But-For standard comports especially well with the demands of causation analysis. I argue that proper causation analysis of motive would permit far more suits than does the But-For standard. The only interesting question is whether we should care about causation at all. I trace the origins of motive-causation analysis in the law and argue that concern for causation is best understood as a proxy for other policy concerns, which can be better addressed directly (as the next two Parts actually do).

Part III addresses arguments concerned with defendants’ freedom of action. For example, a stringent motive test could grind the government to a halt whenever bigoted officials are elected; a But-For standard permits actions vital to national security even if the official happens to entertain unconstitutionally bigoted motives for the action. Yet this argument trades on two errors. First, it ignores alternative motive rules that provide even better assurance to defendants (without necessarily compromising the interests of plaintiffs). Second, it conflates liability and remedy questions. It is often possible to permit the defendant’s action while still recognizing the injury to its victim.

Part IV considers the interests of plaintiffs. A liberal motive standard might allow overcompensation, for example, by allowing incompetent employees to be reinstated because of the boss’s stray thoughts. While a But-For standard can prevent a plaintiff’s windfall, it goes too far and imparts a windfall on defendants. Part IV considers other motive standards that minimize, rather than allocate, any windfall.

While the bulk of this Article criticizes the But-For standard indirectly as resting on a troubled foundation, Part V criticizes it directly as lacking properties that are plainly desirable in a mixed-motive rule. Holding the But-For standard to a series of intuitive requirements both shows its inadequacy and suggests the sorts of requirements that a proper motive approach must meet. Part V is therefore the foundation of a general normative theory of motives and serves to preview developments forthcoming in later research.

It remains possible that the But-For standard is justified in some cases. Perhaps there are areas of law for which all alternative standards are problematic or impossible. A final verdict on the But-For standard would require a deep dive into each area of application, a comprehensive normative framework for motive analysis, and a head-to-head comparison of the But-For standard against its most promising competitor—too much for one article.25 For now, I merely argue that, for any area of law, courts and lawmakers should take seriously the possibility that a But-For standard is not an appropriate way to address mixed-motives claims. In making my case, I cleave most closely to the law of employment discrimination because it is the area with the most development and influence on mixed-motives jurisprudence in recent years.26

I. An Introduction to the But-For Standard

What is a But-For standard in a mixed-motives case? How would we know if it has been satisfied? How does it differ from other approaches to motive analysis?

One way to describe it is with words, precisely defining the content of this standard. The Uniformed Services Employment and Reemployment Rights Act27 uses a But-For standard to evaluate workplace discrimination against veterans. It bars adverse treatment “on the basis of” their military service.28 It then goes on to say that such actions violate the law “unless the employer can prove that the action would have been taken in the absence of” the protected status.29 The statute defines a motive-based offense and then provides a safe harbor in the shape of the But-For standard.

In other areas, such as federal regulation of securities market manipulation, the governing statute is silent on motive standards, so it must be found in case law: “The Court holds that in order to impose liability for an open market transaction, the SEC must prove that but for the manipulative intent, the defendant would not have conducted the transaction.”30 That rule emerges from a case in which a defendant bought a terrific quantity of stock at the end of a trading day both as a prudent hedge for some maturing liabilities and to eliminate those liabilities by monkeying with the stock price on which the liabilities were based.31

These different formulations—a statutory protection for veterans, a common law standard for traders—use different words. They may differ in other important ways, too, such as who has the burden of proving or disproving a motive that meets the But-For standard. For discrimination against veterans, the defendant bears the final burden of showing that bad motive was not a but-for factor in the decision; for market manipulation, the government bears the burden of showing that it was a but-for factor. Still, both tests share the notion that the defendant can prevail if the lawful motive would have independently motivated the act such that the illicit motive appears to be superfluous to the outcome.

Another way to think about mixed motives is numerically, assigning values to represent the relative importance or intensity of various motives, which can then be depicted graphically. We could talk about some sort of motivational “strength,” the magnitude of which can be given a number. When there are just two motives, we can graph the pairing as Cartesian coordinates. A previous article offered a way of describing motives and motive standards, which I restate briefly here.32

Consider four individuals (bosses) who each take an equivalent given action (say, firing employee). For each boss, A-Motive denotes her acceptable reasons for the action, such as the fact that employee is a very poor worker, and B-Motive denotes unacceptable motives for the firing, such as antidisability animus. Figure 1 depicts the motive combinations attributable to each individual: Alpha (A: 1.3, B: 0.7), Beta (A: 0.3, B: 1.5), Delta (A: 0.4, B: 0.9), and Gamma (A: 1.4, B: 1.6).

Figure 1:  Graphing Motives

This graphical placement makes it easy to quickly see individuals’ motive combinations. Alpha fired the employee with mixed motives, but her acceptable motives were strong. Indeed, they were so strong (A > 1), that they would have independently motivated her to action. Conversely, her bad motives, if they were isolated from any other motives, would not have actually led Alpha to fire the employee. The animus would have lingered without expression, and the employee would have kept her job.

Beta has the opposite sort of pairing. She acted for mixed motives too, but her acceptable motives were quite weak and would not have amounted to a firing on their own. Conversely, her bad or B-Motives needed no help to spur her to action. She would have fired the employee for animus alone.

Delta has mixed motives, neither of which are compelling, but which are collectively sufficient to spur action. If either motive were weakened or eliminated, she would not have acted. And Gamma has two very robust operative motivations. If either motive fell away in full, she would still have plenty of reason from the other wellspring for action.

Which of these individuals violates the Americans with Disabilities Act33 (ADA)? The answer is Delta and Beta. The test under the ADA is a But-For standard,34 which asks: Would the defendant have acted differently but for the illicit motive? For Delta and Beta, the answer is “yes.” Without the B-Motive, they would not have had sufficient motivation for action. Conversely, Alpha and Gamma would have remained motivated for action even if the B-Motive were eliminated.

This description is true not just of these four individuals but of all motivational combinations similar to theirs. Figure 2 displays shaded regions where the law finds a violation of the But-For standard. It leaves a blank triangle at the lower left to show where there is no liability because there is no action—the motives, even when summed, do not add up to enough (1) to motivate action.

Figure 2:  But-For Standard

Now let us consider other liability standards. A Primary-Motive standard would assess motives based on their strength relative to one another. Notice that both Alpha and Gamma have a substantial degree of A-Motive, but only Gamma also has a strong B-Motive. Indeed, Gamma’s B-Motive is stronger than her A-Motive. We could assign liability to Gamma on that basis, as well as Delta and Beta. The intuition here is that individuals are charged with whatever motive was foremost, regardless of whether it was big or small in some absolute sense. Figure 3 displays this possibility.

Figure 3:  Primary Motive

The Primary-Motive standard is the normal approach for challenges of malicious prosecution,35 corporate law’s business judgment rule,36 racial gerrymandering,37 and most tax matters.38

Another option is a Sole-Motive standard, which would find for the defendant only if B-Motive were the solitary motive urging action, and the defendant would prevail upon showing any A-Motive at all. If such a standard governed employment discrimination, all four individuals would avoid penalty for their firing. Figure 4 depicts this.

Figure 4:  Sole Motive

This Sole-Motive standard is used to evaluate legislative enactments of retroactive punishment,39 racial profiling by police,40 intentional interference with business,41 “spite walls” and other uses of property intended to frustrate neighbors,42 and the tax question of whether a transaction is a “sham” without economic substance.43

The reciprocal of the Sole-Motive standard is the Any-Motive standard, which rewards the plaintiff whenever bad motives were at all part of the defendant’s decision even if acceptable motives were very significant. We depict this proplaintiff standard in Figure 5.

Figure 5:  Any-Motive

The Any-Motive standard is used, inter alia, to evaluate kickback payments to doctors,44 persecutor motives in asylum cases,45 core discrimination claims,46 many Batson challenges to biased jury selection,47 and interference with employee retirement benefits under ERISA.48

There are numerous motive rules available, and the fields of legal practice draw widely from them. So it is fair to ask what justifies the But-For standard over any other workable approach in a given context. In the next three Parts, we consider the main arguments in favor of the But-For standard, beginning with causation.

II. Causation

Should we use theories of causation to evaluate motives? In other words, should we acquit defendants of a motive-based offense if their bad motives (though present) were not a but-for cause of their act? It may seem obvious that we must. The central idea is that a defendant’s illicit motive should justify a remedy only if that motive actually caused harm. If the same actions would have taken place regardless of anyone’s bad motives, then what difference did the motives play at all? And if they made no causal difference, how can it be legitimate to regulate them?

Recent US Supreme Court opinions give the impression that motive-causation is essential as a matter of precedent, background principles of law, and statutory law. These decisions adopted a But-For standard with essentially no policy justification.49 Instead, the Court located a causation requirement and deduced a But-For standard from it.

This Part argues that the turn to causation as a basis for the But-For standard is unfounded. Part II.A argues that we should not overstate the degree to which the law requires causation; many motive inquiries are not causal. More importantly, even if motives must be causal in order to count, there is no reason to think that this would lead to a But-For standard. Part II.B explains why sensible causation analysis would impose no such requirement. There is, accordingly, no argument from causation in support of a But-For standard for mixed motives.

A. Is Causation Required?

Whatever motive-causation would require, it is not clear that causation is itself required.50 This Section identifies and criticizes the rationales sometimes given for incorporating causation into a mixed-motives standard. Recent Supreme Court decisions articulate three such rationales for their brand of motive-causation.51

First, the Court has pointed to its own past decisions, applying seemingly similar causation standards.52

Second, the Court has derived the standard from the background of “textbook tort law.”53 On a page punctuated with citations to a torts treatise and the Restatement of Torts (First,
Second, and Third),54 the Court in University of Texas Southwestern Medical Center v Nassar55 declared that such tort principles “are the default rules [Congress] is presumed to have incorporated, absent an indication to the contrary in the statute itself.”56

Third, the Court has located the requirement in statutory text. For example, the Age Discrimination in Employment Act57 (ADEA) protects employees from adverse workplace treatment “because of such individual’s age.”58 Justice Clarence Thomas cited three dictionaries and one case to deduce that “because of” means “causation.”59 From causation, but-for causation followed almost deductively.60

The following Sections argue against requiring causation on the foregoing bases.

1. The Court’s own decisions.

Despite the recent enthusiasm for motive-causation, it is not an ancient and venerable doctrine. As recently as 1972, the Supreme Court wrote that causal analysis in a mixed-motives case is a “temptation . . . best rejected, and we reject it here.”61 Causation entered mixed-motives analysis in 1978 with Mt. Healthy,62 which derived motive-causation from a rather cryptic analogy to cases involving neither motives nor causation: “In other areas of constitutional law, this Court has found it necessary to formulate a test of causation which distinguishes between a result caused by a constitutional violation and one not so caused. We think those are instructive in formulating the test to be applied here.”63

Those “other areas” were all “tainted evidence” cases, in which criminal convictions were upheld despite coerced or otherwise problematic confessions.64

It is fine to draw analogies to other areas built on causation. Unfortunately, none of the cited cases so much as discusses causation at all, or uses the words “causation” or “cause” (apart from terms of art like “probable cause”), let alone draws on causation to justify its holding.65 Causation thus insinuated itself into motive jurisprudence only recently and on mysterious terms.66

Motive-causation analysis’s recent and peculiar emergence tarnishes any effort to link motive-causation to enduring background principles of law. Motive-causation was invented just recently, so it cannot be a self-evident entailment of deep tort principles, for example. Nevertheless, several recent Supreme Court cases have dressed up motive-causation with respectable vestments. We have seen that these arguments are ahistorical, but the following Section shows how they are also illogical.

2. Background principles of law.

In Nassar, Justice Anthony Kennedy’s majority opinion imposed a causation requirement on mixed-motive retaliation cases on the basis of background tort principles. He begins the opinion by asserting such principles, without citation:

When the law grants persons the right to compensation for injury from wrongful conduct, there must be some demonstrated connection, some link, between the injury sustained and the wrong alleged. The requisite relation between prohibited conduct and compensable injury is governed by the principles of causation, a subject most often arising in elaborating the law of torts.67

These principles of causation are “the background against which Congress legislated in enacting Title VII, and these are the default rules it is presumed to have incorporated, absent an indication to the contrary in the statute itself.”68 A “standard requirement” arising from these background principles is “[c]ausation in fact—i.e., proof that the defendant’s conduct did in fact cause the plaintiff’s injury.”69 Therefore, any law pertaining to injury and compensation must meet tort standards of causation.

There are three problems with deriving a causation standard from background principles of tort law. First, by their own terms, such arguments do not reach many areas of the law in which the But-For test is deployed. While employment discrimination is arguably a statutory tort focused on compensating injured employees,70 other areas of the law are not torts and are not focused squarely on compensation. For example, the Equal Protection Clause forbids prosecutors from striking potential jurors based on their race or sex.71 When motives are mixed, the majority of federal courts adopt a But-For standard to evaluate prosecutor motives.72 Often, they derive this standard by assuming the relevance of causation, either without argument or solely by citing other cases’ invocation of causal principles.73 Yet Batson challenges to discriminatory jury selection are not torts, and they are not primarily concerned with injury and compensation, so it is an error to directly or derivatively import a causation requirement on the basis of background tort principles.74

Second, while tort principles may presumptively govern common law torts, statutory torts do not automatically partake of common law tort principles.75 Professor Jonathan Cardi and others have argued that statutory torts incorporate common law principles only if the statute explicitly says so. If something is required of torts at common law, it is not presumptively applicable to statutory torts such as employment discrimination or market manipulation.76

Third, and most importantly, it is an error to think that background principles of tort law teach that motives must be causal. Tort principles certainly do require that the defendant’s action or negligence as a whole caused the plaintiff’s injury, but they do not entail that every element or aspect of the conduct injured the plaintiff. To infer from the fact that tort law cares about causation that motive must be causal is to commit a fallacy of division,77 akin to inferring that hydrogen is wet because water is wet.78 Torts actions as a whole must satisfy several elements, one of which is causation, but it does not then follow that each element of a claim (such as motive) must satisfy a causation standard.

To see the error in this inference, consider a parallel inference drawing on the tort of battery. The elements of battery are (1) intent, (2) act, (3) causation, and (4) contact that is (5) harmful or offensive and (6) without consent or privilege. How do these elements interact? These elements are partially interdependent. For example, the (2) act must play a (3) causal role. But they are not fully interdependent. No one would suggest that (6) lack of privilege must play a (3) causal role. It is no defense to a proven case of battery that the defendant would have acted identically even if the plaintiff had given her permission for the harmful contact, and hence lack of consent or privilege was causally inert. No one thinks, for that matter, that the defendant’s (2) intent must have been without (6) consent or privilege.

These remarks accord with courts’ actual treatment of mental causation in tort cases. Courts are generally clear about the independence of the mental elements from such requirements.79 They do not seem to impose causation requirements on mental elements.80 Thus, there is little in tort law to support requiring motive-causation.81

To summarize, it’s clear that bad motives must stand in some relationship to a challenged corporate firing or congressional enactment in order for the plaintiff to prevail, but it does not follow from background principles of tort law that the relationship must be causal. We should be even more skeptical as we move from common law to statutory torts (for example, employment retaliation), and from statutory torts to nontorts (for example, jury selection).

3. Statutory language.

Whatever the background principles of law, it is clear that a statute can stipulate to courts that they should hold motives to a causal standard. Do statutes actually do this? To the contrary, no statute uses the word “causation” in setting out a motive-based rule.82

For example, the ADEA prohibits discrimination “because of . . . age.”83 Title VII of the Civil Rights Act of 1964 likewise prohibits discrimination “because of . . . race, color, religion, sex, or national origin.”84

Courts have often read “because of” language as requiring a causal link between the discriminatory motive and harm.85 Many scholars have also accepted that “because of” is causal language.86 Yet there are good reasons to resist such interpretation. There are noncausal ways to read apparently causal statutes, and the common body of mixed-motives jurisprudence is threatened if courts lean heavily on fine gradations of word choice. The remainder of this Section is devoted to those two objections to a causal construal of ambiguous statutory language.

To begin, “because of” is obviously an ambiguous phrase. It can indicate a causation, but we also use it in all sorts of noncausative contexts, such as whenever we seek to explain.87 Some explanations are causal,88 but not all. Some explanations are interpretive,89 constructive,90 pragmatic,91 or teleological.92 Any of these are semantically compatible with statutory language, as I demonstrate in the following paragraphs.

For example, we might tutor a confused math student with the phrase “X is 7 and not −7 because of this operation on the previous line.” This use of “because of” is not causal in the physicalist sense typically invoked by contemporary tort theory. That is true even if we insert more explicitly causal language: “This operation on the previous line caused X to be 7 and not −7.” Causal metaphors are tempting even in math problems, but we can be sure that whatever mathematical law “caused” X to be 7 is quite unlike whatever “caused” a boss to fire someone.

We also say “because of” in settings calling for characterization. Imagine that we agree that an amount is deductible if spent “because of” business and that I ask about whether my Wall Street Journal subscription is deductible. I am not proposing causal inquiry, such as whether “business” made me subscribe. I may know well what made me subscribe—some mixture of intrinsic interest in the stories and the sense that it will help me with work, etc. The crucial question is whether that list of reasons counts as business.93 Shall we characterize this as a business expense or not? Characterization is ubiquitous in law, and it would be unsurprising if an antidiscrimination statute used the language of characterization.94

Even if “because of” entailed causation under the ADEA and Title VII, it would not follow that all statutes, with or without reminiscent language, are similarly fixated on motive-causation. Other discrimination statutes use different language,95 and some use no plausibly causal language. For example, the prohibition on discrimination in contracting simply provides, “All citizens of the United States shall have the same right . . . as is enjoyed by white citizens.”96 This language seems to award certain entitlements or require certain outcomes without special regard to causal influence of any particular impediment.

Other laws associated with a But-For standard make no textual reference that can be used to identify the status of causation. For example, market manipulation is prosecuted under § 10(b) of the Exchange Act of 1934,97 which reads, in relevant part, “It shall be unlawful for any person . . . [t]o use or employ, in connection with the purchase or sale of any security . . . any manipulative or deceptive device or contrivance.”98 Courts confronting mixed-motives market manipulation will find no explicit reference to mental states or causation. Likewise, free expression and equal protection claims in constitutional law, which are subject to the But-For standard, draw on constitutional amendments lacking any singing endorsement of causation.99 There is no “because of” language to hint at causation. If statutory text is the cradle of causation requirements, a great many laws—addressing discrimination and otherwise—might be exempt.

Even when a legal rule is apparently stated in terms of causation, we should not be too quick to assume that its actual content is causal. The test for whether an employer is liable for the death or injury of a traveling employee depends on whether the travel was within the course of employment, which in turn seems to invoke causation: “The test as to whether a given trip is within the course of employment where both business and personal motives are involved is whether the business motive was a concurrent cause of the trip.”100 But the conditions for establishing a “concurrent cause” are really about moral responsibility, knowledge, and endorsement rather than actual causal influence: “There must at least be some action on the part of the employer to connect the trip to employment, some sponsorship, some approval, some employer action must be present.”101 If endorsement is present, the “causation” standard is met even if the employer “does not sponsor or encourage the trip.”102 The purportedly causal motive standard can be satisfied if an employer approves of a trip but does not request or urge it. The causal label drops out (in favor of a responsibility characterization) once the test is elaborated and applied. A reader looking at only the Westlaw headnote would wrongly overstate the law’s commitment to causal analysis.

The point of the foregoing is not to show conclusively that the better interpretation of a given statutory text or judicial rule precludes a causation requirement in any given case.103 Rather it is to show that causation is not singularly plausible and that any statutory derivation of causation must be settled either through policy arguments or through extended textual analysis intended to clarify the potentially ambiguous meaning.

And when that analysis is done, the lesson may well be bounded to that area of law; even if, say, employment discrimination law calls for causation, it would be a mistake for disputes arising under different statutes—say, market manipulation or constitutional law—to partake of the other domain’s preference for causation. A certain kind of fixation with statutory language would tend to split the law’s motive jurisprudence at a time when comparison and cross-substantive learning would seem desirable. The price of a text-born motive-causation requirement is a balkanized motive jurisprudence, in which the lessons of one legal controversy have little to teach the others.

B. What Does Causation Require?

To whatever degree that general principles or statutory text compel plaintiffs to prove motive-causation, they simply do not entail a But-For standard. It is true that but-for causation is the first and central test for causation in tort law.104 However, an item failing the but-for test can still be a cause.

This is clearest in cases with multiple sufficient causes.105 If two raging fires both strike a house, neither is a but-for cause—the house would be destroyed by the other fire if the first one did not exist—but liability can attach to either fire starter because each individual’s negligence was sufficient to cause the injury.106 Tort law regards each one as a cause of the terrifying blaze to which it contributes.

If we analogize a defendant’s motives as potential causes of the action, then mixed motives are analogous to torts with multiple causes. A termination resulting from a bad reason (hostility toward the employee-plaintiff’s race) and also a good one (hostility toward her absenteeism) is akin to a fire resulting from a tortious cause (the defendant’s match) and a nontortious one (a lightning bolt).107 Neither is the but-for cause of the destruction because either motive might be subtracted and the other would be adequate to spur action, but both causes satisfy the causation requirement.

Tort law therefore endorses causation both in the same places as the But-For standard (Quadrants I and IV) as well as the region of overdetermination, Quadrant II. The test that would implement conventional tort causation is not a But-For standard; it is the Causal-Motive standard, which finds for the plaintiff except in Quadrant III. Such a liability zone is depicted in Figure 6.108

Figure 6:  Causal Motive

Even in Quadrant III, the absent upper left-hand corner, we should note that the Restatement is agnostic as to causation. There is a reasonable case to be made that causation should be found even in Quadrant III, which is to say, in every region.

Recall that, in Quadrant III, A-Motive is sufficient on its own to motivate action, and B-Motive is neither independently sufficient nor is it necessary given the strength of the A-Motive. Under such parameters, it is perhaps not intuitive to regard the B-Motive as a cause; the B-Motive is like a match that is tossed on a blazing inferno. And yet the Restatement does not rule out tiny matches,109 and it does not rule out this most expansive conception of multiple causation.

The argument is that small causes, though dwarfed by a larger cause, do still have a causal impact. They assure the outcome against some set of background facts—namely a partial slice of the presently robust alternative cause. They are therefore necessary elements of a sufficient set—which, according to scholars such as Professor Richard Wright, is what we mean by “cause.”110 The tiny match is a necessary member of a causally sufficient set: the portion of the inferno that was not quite big enough to destroy the house is made big enough by the match.

To see this “portion” notion in action, consider an action that was amply motivated by an acceptable motive (1.5), with an additional and independently insufficient bad motive also operating (0.5). This is akin to a raging inferno into which one tosses some matches, which would normally be inconsequential. For exposition, we can describe the intensity of the blaze as 1.5, where 1 is sufficient to reach and destroy the house. The matches contribute an additional 0.5, by themselves far too little to destroy the house.

The entire fire is analogous to a defendant’s total motivational state. The large fire is analogous to the defendant’s large acceptable motive, sufficient to cause the harmful fire or firing. The small portion of the fire emanating from the match is analogous to the putatively superfluous bad motive, by itself too small to change anything.

Figure 7:  Motive-Portion

One way to think about causation here is to see that, in the universe of causes, where “1” is the minimum amount to be effective, the acceptable motive or large fire is more than sufficient. There is a causal set sufficient to meet or exceed “1” that draws exclusively on the larger cause (be it an acceptable motive or a fire). We can name this designation “Sufficient Set 1” because it contains enough to cause the observed result. This set leaves some other material unnecessary, including a third of the primary cause as well as all of the secondary cause. On this construal, the matches or B-Motive are truly incidental.

But why draw the lines in that way? We could just as easily bracket the set defined by “Sufficient Set 2.” Here, the bad motive (or the matches) forms a vital part of the motive set (or fire) that would lead to the harm. The one-half it contributes is fully half of the sufficient set, the other half of which is composed of one-third of the blaze or acceptable motive. And on the sidelines, unnecessary for the sufficient set, is some additional portion of the acceptable motive (or large fire). There are two perfectly sensible descriptions of the fire or firing, one in which the small factor was pivotal.

If we were so inclined, we could think of the bad motive pairing with a portion of the acceptable motive and succeeding in completing a causal set with that portion. If we accept the fungible, partial causal pairings, we can find causation anywhere that the sum of good and bad motives combines as sufficient to motivate action. Sufficient Set 2 is also sufficient to destroy the house or motivate the action, and it construes a large part of the large motive as unnecessary rather than a bit of both motives.

We can superimpose this combination idea onto our familiar Cartesian depiction of motives. Figure 8 shows how we can think of a small portion of the acceptable motive (here, one-half, or one-third of the total) combining with the bad motive to cross the threshold into Quadrant IV. The remaining acceptable motive can be thought of as causally irrelevant if we wish to apply that label to the factors that could be removed without changing the result.

The ability to pair portions of each motive dramatically increases the causal significance of even lesser motives. Operationalizing this approach means adding Quadrant III to the Causal-Motive test, which results in an Any-Motive test. Figure 9 depicts a rule drawn from this version of tort causation, which would apply to any combination of motives so long as some illicit motive is present. On such a construal, causation is established in all mixed-motives cases.

Figure 8:  Any-Motive

Whether this is a sound approach is plainly controversial. Can fires really be subdivided into portions?111 Can motives?112 Are there really hard lines, a sliver below which the house is preserved or the employee retained? These are bold positions to stake out, so it is understandable why the Restatement noted the possibility and neither endorsed nor rejected it. Still, regardless of whether one is persuaded that the Any-Motive standard is a better fit for causation than the Causal-Motive standard, it should be clear that either is a better fit than the But-For standard.113

While reasonable minds can differ over the wisdom of allowing motive-causation with less than a But-For standard, it is baffling that so many commentators assume that a But-For standard draws any support from principles of causation. The US Supreme Court is one body that has alternated in its awareness of this problem.

Despite unanimously accepting a causation requirement in employment discrimination mixed-motives cases, in half a dozen mixed-motives cases, the Supreme Court has nevertheless split over and over on what exactly that requirement entails. In Price Waterhouse v Hopkins,114 the plurality accepted multiple causation,115 with the dissent insisting on only but-for causation.116 All subsequent decisions have flipped. Therefore, some courts interpreting employment discrimination cases may be bound to implement causation as but-for causation. But future courts should take heed of how little constraint a proper causation analysis actually imposes on mixed-motives analysis. 117

* * *

A causal argument for the But-For standard finds little support in history, statutory text, or tort principles. Despite all the problems, but-for causation arguments remain in circulation. Why? It may be because causation was only a stand-in for certain policy arguments, such as freedom from intrusion or avoidance of windfalls. We turn now to these policy arguments to see whether there is anything worth vindicating through motives analysis.

III. Freedom

The But-For standard is supposed to be the antidote to two closely related worries about motives analysis: excessive constraint and excessive intrusiveness. Responding to both, the But-For standard is meant to ensure defendants a measure of freedom.

Yet to whatever degree the law seeks to create safe harbors, it should do so in a way that is logical. By contrast, a But-For motive standard is not needed to prevent the worst impositions on freedom, nor is it successful in preventing the ordinary ones.

A. Blocked Action

Analyzing motives means that some otherwise legitimate actions might be blocked because of the mental state of the decisionmaker.118 For example, a city with a severe budgetary crisis might be unable to close costly amenities if its city council at the time happened to harbor additional (bigoted) reasons to close the amenities.119 This Section first introduces several such examples and how they have worried scholars and jurists. I then explain why these worries need not be decisive: it is possible to calibrate remedies to avoid blocking action, and a serious commitment to this objection would prove too much—concern for blocked action leads not to the But-For standard but to an abandonment of motive jurisprudence altogether.

But first: a majority of the justices in Palmer v Thompson120 took the problem of blocked action quite seriously.121 Cities must have the power to close pools that they cannot afford even if city leaders’ ulterior motives are not admirable.122 Although the Palmer Court rejected motive analysis altogether, it is easy to consider the But-For standard as a compromise, guaranteeing that defendants can always do what they know is right even if tainted by some bad motives too.

For a contemporary application, consider the current presidential administration’s Travel Ban. One week after taking office, President Donald J. Trump signed Executive Order 13769,123 which halted entry by individuals from seven predominantly Muslim countries124 and constrained admission of refugees.125

These were arguably mixed-motive orders. On the one hand, The Court of Appeals for the Fourth Circuit said of the second order that it “drips with religious intolerance, animus, and discrimination.”126 On the other hand, the orders offered numerous national security rationales.127 As a result of challenges under, inter alia, the Free Exercise Clause of the First Amendment, federal district courts issued nationwide preliminary injunctions against both executive orders, which their respective courts of appeals affirmed.128

If we accept a simplified version of these facts,129 we confront a mixed-motives case with a serious blocked action problem. If issued by another president, the orders might save lives, but they are illegal because of the animus of this president.

Numerous commentators have raised this worry. For example, Professor Eric Posner has argued that motive-based challenges endanger Americans:

[T]his extra layer of judicial review will burden Trump even when he acts with a national security justification—like when Obama ordered extra vetting of Iraqi refugees in 2011. While a court might not ultimately block Trump from acting, a [temporary restraining order] will always be a real possibility, which means that a necessary measure for our safety may be blocked. We may have a new regime of heightened judicial review in national security cases because courts believe the president is a bigot. This has never happened in all of American history. The country is less safe as a consequence.130

Given a bigoted president, an expansive motive standard makes every security-related act potentially invalid, regardless of importance.131

Posner does not quite call for a But-For standard, but it is understandable why others would think such a standard would defuse some of the trouble of blocked action.132 If the president can show a meaningful good faith national security motive for an action, then the But-For standard would protect the action. Thus, we need not wait for a less bigoted president in order to protect national security.

As another example, Professor Richard Fallon argues for some caution in motive analysis because “few would judge it tolerable for courts to strike down a law prohibiting murder if historical examination revealed that most members of the legislature voted for it solely for the constitutionally forbidden purpose of enforcing one of God’s commandments. The consequences would be too draconian.”133

Yet the effects are draconian only if laws are actually struck down and they stay struck down. The main fallacy with the blocked action argument is that it conflates liability with a particular remedy: permanent injunction of an action or overturning of a statute. But an effective motive jurisprudence can calibrate its remedy to avoid blocking vital actions.

First, a motive-based remedy could be temporally bounded to accommodate later changes in motive.134 A legislature whose law is blocked because of its bad motive can repass the law when its motives are demonstrably pure. A defendant that takes heed of the legal effect of motive will have good reason to adjust its motives ahead of time to avoid judicial trouble.

Can defendants easily change their motives to dynamically minimize the cost of blocked action? In many cases, the answer is yes. Legislators can abstain from votes when their motives are unlawful. For an aggregate entity, it is often quite feasible to reshuffle personnel to achieve an acceptable motive structure.

And knowing that these shuffles might take place gives individuals a reason to look within and attempt the hard work of addressing their own motives. If a boss’s bias creates liability risks for her firm anytime she takes action, then her motives are a career liability. If a legislator must abstain on key votes, lest the law be invalidated, her constituents may take notice. If the law puts a price on bad motives, individuals may do whatever they can to adjust their motives or else be replaced by those who have done so.

Nor need the remedy be even a temporary injunction or invalidation of an action. When we wish to avoid constraint but still think that the behavior warrants regulation or liability, we can apply a comparatively small penalty: paying a fine, an award of attorneys’ fees, or even just a declaratory judgment. A defendant may deem this consequence sufficiently small that she will still take the action in question.

We do this in the law frequently, setting penalties at a level that many defendants will be willing to pay.135 The penalty imposed can be used just as a benchmark to separate low-value acts from high-value acts, or it can be used to actually fund compensation to victims. This can be true even when wealth is not the thing transferred. A largely dignitary harm might lead to a largely dignitary remedy. Granting the plaintiff a declaratory judgment and attorneys’ fees may be important for encouraging suits, empowering individuals, vindicating rights, and expressing the law’s solidarity.

When the plaintiff’s dignity and the defendant’s freedom are both important, we can protect a measure of both.136 We already grant attorneys’ fees and related remedies to Title VII discrimination victims even when discrimination was not a but-for cause of their termination. What about a similar remedy for the travel ban orders? Individuals excluded from entry under the orders would receive financial compensation for their jarring time holed up at an airport in recognition of (and to discourage) the bigotry involved in creating the orders. But Order 2 stands, and (by hypothesis) the important national security goals are still effectuated.137

Relatedly, a tough remedy (say, invalidation) can allow the putatively important action while encouraging defendants to take steps to affirm the dignity of victims. For example, an administration that makes anti-Muslim statements faces credibility problems in asserting that its travel ban is not meant to discriminate based on religion; its case gets better if officials apologize about or repudiate past statements. These reparative steps arguably do some good for the victims of the ban, whose victimization might otherwise go unanswered by officials in the process of unblocking the action.138

Another nonblocking remedy is for the motive inquiry to end not in a determination of liability (or the striking down of a law) but simply a shift in judicial posture from prodefendant to proplaintiff in subsequent stages of litigation. For example, courts normally abstain from interfering with the decisions of corporate directors.139 However, when directors’ motives are shown to be sufficiently tainted, things look different. The tainted decisions are not per se invalid, nor are tainted directors automatically liable; they simply lose the protection of the abstention doctrine, the business judgment rule. The litigation continues, now with the board having a burden to prove to the court that its conduct was entirely fair.140 If the conduct was entirely fair, and so was objectively the right action, the court will not block it or censure the board. Thus, a motive inquiry can proceed—under a standard other than the But-For standard—without any risk of blocking action.

Likewise, many constitutional motive inquiries lead not to invalidation but instead to additional inquiries. Acts flunking a motive test may still be tolerated if the state can defend them under a strict scrutiny standard. This is true of Dormant Commerce Clause challenges to protectionist motives,141 Due Process Clause challenges to laws motivated to place an “undue burden” on abortion rights,142 Free Exercise Clause challenges to efforts to “infringe upon or restrict practices because of their religious motivation,”143 Establishment Clause challenges to promoting religion,144 Equal Protection Clause challenges to racial discrimination,145 and other areas.146 For all these, the motive inquiry is only the first step. The action will not be blocked if it is narrowly tailored to serve a compelling government interest.147

All of this is to say that we don’t need to altogether block a defendant’s conduct if she loses on a motive-based test. In many areas of law, we allow losers on a subjective test to defend themselves with facts supporting the objective appropriateness of their actions. Defendants don’t like having to defend themselves on these terms, nor do plaintiffs like having to go through one more step toward success, but such a compromise position is plainly available to resolve some of our most pressing concerns about blocked action.

If a president’s executive order is vital for national security, the Department of Justice can defend it by establishing the objective importance of the law on a nondeferential standard. Considerations of expertise, secrecy, and deference make this inappropriate in cases of unimpeached motives, but quite fitting when the defendant-executive has revealed problematic judgment. As Fallon explains, the point of motive or intent analysis

is not whether the legislature would have reached the same result absent an affirmatively forbidden purpose, but whether a court should defer to the legislature’s judgment that a challenged statute comports with constitutional norms. With regard to that question, the legislature’s breach of its deliberative obligations should bring deference to an end and, for reasons stated above, should provoke elevated judicial scrutiny.148

In this light, it is significant that each court reviewing the Executive Orders stressed the paucity of evidence of the actual security justification for the orders.149

B. Intrusiveness

Even if the law’s remedies are appropriately tailored to avoid blocking good actions, the mere act of scrutinizing a defendant’s motives has its costs. When the defendant is a legislature or state actor, asking a defendant to prove its good motives may be an unacceptable intrusion on the independence of another branch or department.150 Legislators have privilege, and it is a worrisome thing to build doctrines that will frequently run up against that principle.151 Nongovernmental institutions like corporations may retain rights to autonomous private ordering, reflecting America’s commitment to employer autonomy and discretion in labor markets.152 Employers need not worry about having to defend actions performed with “mere discriminatory thoughts.”153 They are safe to make decisions without backseat drivers. Most generally, natural persons should not have to explain their every thought.154

A But-For standard seems to protect such freedom because it protects defendants who can document a good motive, perhaps before dirty laundry is aired. Yet the standard does a poor job at creating a safe harbor against intrusion relative to other motive standards. Most obviously, the Sole-Motive test is a better test for preserving the defendant’s freedom because it exonerates defendants every time a legitimate motive is present. Relative to that benchmark, a But-For standard leaves defendants highly vulnerable.

If the Sole-Motive test is rejected, it must be because the law is seeking to balance the importance of both legitimate and illicit motives. When the law takes stock of two motives, it often does so by way of a Primary-Motive test. This test also allows some plaintiffs to prevail while still giving assurance to many defendants. Holding the But-For standard up against the Primary-Motive standard, the latter does a far better job in protecting defendants and giving them appropriate assurance.

To compare the Primary-Motive standard and the But-For standard, let us consider some properties that a test would have if it were to give appropriate assurance to defendants, protecting their A-Motivated acts against intrusion. First, the test would generally not penalize increases in A-Motive because A-Motive is what the test is meant to protect. Second, a suitably protective test would give greater legal confidence to the wide swaths of relatively common and innocent cases. It should provide protection in questionable cases too, but privilege should weaken nearer its limit than its core. Third, the test should provide greatest protection when the risk of judicial error is highest; conversely, when the evidence will clearly bear out the defendant’s innocence, there is less need for a stridently prodefendant motive rule.

The But-For standard flunks each of these conditions.

First, between two individuals who differ only in their level of A-Motive, the one with higher levels of A-Motive should not be treated more harshly. Yet a But-For Standard does just this, and does so more often than competing tests.

Consider Figure 9, which shows the liability region for a But-For standard.

Figure 9:  A-Risk

An individual with a low level of A-Motive (Low-A) often escapes liability—she simply hasn’t taken any intentional action on the basis of A-Motive, and B-Motive was not enough to complete the motivation package. But an otherwise identical individual who nevertheless has higher levels of A-Motive could cross the threshold into the realm of hybrid cases. In fact, this remains a potential path to liability unless B-Motive increases enough to put the combined motivation into the zone of sole determination in Quadrant I. Anywhere in the starred region of Figure 10, the stars move up higher and closer to legal risk if A-Motive rises.

Figure 10:  But-For A-Risk

 

 

By contrast, a primary purpose test suffers this defect to a much lesser degree, as is clear from Figure 11.155

Figure 11:  Primary A-Risk

The Primary-Motive test endangers a rising A-Motive only if B-Motive is greater than 0.5. If B-Motive is less than 0.5, no increase in A-Motive can result in liability because A-Motive will remain primary in all cases in which action is taken.

Now compare the tests in terms of when they provide the most assurance to defendants that their lawful exercise of A-Motive will not be interrupted. Recall that a logical test should provide ample protection in contexts applicable to most defendants and the most innocent defendants; it would be perverse if a rule meant to protect A-Motive (while balancing some concern for B-Motive) were strongest right before defendants who are actually culpable and weakest in the heartland. Yet that is just what the But-For standard achieves.

Consider the zones of relative protection for the two rules. Figure 12 depicts the But-For standard with a rectangular region, encompassing much of Quadrants II and III, shaded with vertical and horizontal lines.

Figure 12:  But-For Risk

This additional shading represents a realm of risk. In that realm, the But-For standard test does not produce liability, but risk remains that slightly erroneous fact-finding might lead to liability. The case is close enough that a factfinder or prosecutor is at a meaningful risk of misreading the motives. As is clear, the zone of risk disappears once A-Motive is high enough, but it does not vary with B-Motive. There is no risk that overstating B-Motive could lead to liability, but there is always a risk that understating A could lead to liability.

Contrast this to Figure 13, which shows the risky region for the Primary-Motive standard in stars.

Figure 13:  Primary-Motive Risk

The Primary-Motive standard also leaves the defendant uncertain in some cases, but under different circumstances. Unless both A-Motive is high and B-Motive is low, the defendant might be misconstrued as having a problematic motive combination.

Now let us compare these two regions of risk. To see how they differ, let us eliminate the realms of actual liability; these plainly matter but have already been considered. Eliminate also the places where danger is the same. The remaining (and thus, different) regions are depicted in Figure 14.

Figure 14:  Difference in Risk

Both motive rules have distinctive realms in which the risk of liability is nontrivial, but it looks like the But-For standard creates a normatively more troubling realm of risk. For that test, the risky region is where B-Motive is low and A-Motive is moderate. Presumably this combination is rather common. For any given person, our starting guess should be that she has a moderate level of good reason for her acts and only a shadow of bad motive. Such people are at risk with the But-For standard because they are in the realm of risk that covers many of the most common and most innocent cases.

It is also a region where evidentiary checks are less likely to control the risk. Because neither motive is high, there is likely to be a less than complete record by which the defendant can build her defensive case. At the margin, we would prefer that potential errors arise in areas where they can be dispelled by careful evaluation of the evidence rather than places where there is little evidence either way.

The Primary-Motive standard also leaves a region of risk, but it is a preferable region in which to impose risk. In the starred region, A-Motive is high and B-Motive is medium-to-high. This is probably a better region for risk from the ordinary defendant’s point of view. Most people don’t have high levels of B-Motive, making this region of protection useful to many people and making the region of risk less frightening to most people. And when A-Motive and B-Motive are both present, there is likely to be enough evidence of both for the defendant to assess and make her case.

The point is not that the But-For standard should be replaced by the Primary-Motive standard: perhaps the latter standard is itself inferior to a third option in light of the goals and constraints for a given legal inquiry. For now, the point is simply that the But-For standard has withstood criticism largely because it has never been subject to comparison against its relevant competitors. Nor is it self-evident that a motive rule should protect the defendant’s autonomy and freedom from intrusion; but if the law seeks to do so, it should do so in logical fashion rather than failing on its own terms.

IV. Overcompensation

Courts are reluctant to compensate plaintiffs without seeing that the defendant’s motives led to new harm. Otherwise, the plaintiff may be rewarded for misfortunes the law would normally have deemed noncompensable.156 For example, the Mt. Healthy Court was concerned that a judgment for Doyle might make the First Amendment a vehicle for grabbing job security. As an untenured teacher with serious professional deficiencies, Doyle was at risk. He then did something “dramatic and perhaps abrasive” but also vaguely political.157 The Court worried that

a decision not to rehire could place an employee in a better position as a result of the exercise of constitutionally protected conduct than he would have occupied had he done nothing.158

Overcompensation can seemingly be avoided with a But-For standard.159 If the employee was sure to be fired anyway on the merits as the employer saw them, then the bad motives are arguably harmless, so a But-For standard successfully bars recovery.

Intuitive though it may be, there are reasons for doubt. Some of these problems are connected with the particular facts and arguments in Mt. Healthy, the case in which compensation was first cited as a justification for a But-For standard. Other problems are general difficulties with controlling windfalls through the But-For standard.

A. Overcompensation from Mt. Healthy

Mt. Healthy has become a touchstone for numerous areas of law, including employment and First Amendment. It is therefore sensible to ask whether Mt. Healthy’s reasoning can be sensibly extended to these other domains or was even sound on its own terms.

1. Extrapolating compensation arguments to nonstrategic contexts.

It is not clear that the compensation policies implicated by Mt. Healthy sensibly apply in other domains, such as First Amendment disputes. That is because the overcompensation argument in Mt. Healthy is largely motivated by concern for strategic action: employees may seek the protection of motive-based laws when termination seems imminent.

The strategic-overcompensation argument for a But-For standard arises in the context of a particular type of public school dispute, in which we worry that undeserving employees may opportunistically fabricate First Amendment suits to protect their jobs. Yet the But-For standard is also invoked when school districts eliminate books from school library for both good pedagogical reasons and in order to censor ideas.160 Is there any risk that authors of badly written books will fabricate First Amendment claims, perhaps by organizing political protests intended to stoke the ire of library boards, in order to secure a potential windfall? Courts should be cautious citing Mt. Healthy in areas in which the plaintiff is unlikely to act strategically. More generally, concern for harm and proportionate compensation is not essential in all areas of the law.161

2. The uncertain relationship between compensation and the But-For standard in Mt. Healthy.

Even if the reasoning of Mt. Healthy can be safely applied to other areas of law, it is not entirely clear how that reasoning linked overcompensation to the But-For standard. Mt. Healthy’s invocation of overcompensation as an argument for a But-For standard is mysterious because the Court does not explain the basis of its idea that overcompensation fits only with the But-For standard, nor does it defend it. Indeed, what little we know about the idea’s genesis should caution against linking compensation with a But-For standard.

Mt. Healthy did not offer any authority for connecting compensation with the But-For standard—the linkage occurs without explanation or argument—but reviewing the litigant submissions gives some hint as to its further provenance. In 1970, a student case note offered the following compensation argument:

It is conceivable that a teacher might openly engage in activities which, although controversial, are constitutionally protected, and thereby make it difficult for a school board to even attempt to terminate his employment despite the presence of unquestionably valid grounds for nonrenewal. This is expecially [sic] true in situations where school officials would be unable to candidly deny that the teacher’s “extra-curricular activities” played no part in their decision to deny reemployment. Therefore, it seems imperative that the improper motivation be the primary cause of nonrenewal before judicial relief is granted.162

The student note plainly raised the overcompensation argument on grounds essentially identical to Mt. Healthy. Although the Court does not cite the student note, the school district quoted this passage in full in its brief, suggesting that the Court’s overcompensation idea was inspired by the anonymous note.163

The reason that history matters here is that the first breath of the compensation argument for a But-For standard does not itself support a But-For standard. Instead, the student author urged a different test altogether. He or she would require improper motive to be the “primary cause” of the termination.164 This clearly tracks one of the But-For standard’s main competitors—the Primary-Motive standard.

In this, the school district was happy to assent, arguing that the Mt. Healthy case “exemplifies the test suggested by a commentator.”165 Thus, while the Mt. Healthy Court endorses a But-For standard on the basis of a windfall argument, both the student note and the party brief that first introduced the windfall argument thought it led elsewhere than a But-For standard.

The point of this discussion is not that the student author has privileged authority in drawing conclusions from his or her premises nor that the Primary-Motive test is superior to the But-For standard—though we shall see, in Part IV.B.3, that it does have some powerful merits in this context. However, it can be liberating to see that neither the inventor of the argument nor its proponent in litigation thought that it leads to a But-For standard, suggesting that actual arguments are required to make the case that the test follows from the objection. No such arguments appear in Mt. Healthy or elsewhere.

B. Controlling Windfalls

The foregoing Section disputes the applicability of Mt. Healthy’s compensation reasoning, while this Section now turns to the merits of that reason. In some cases, apparent overcompensation is really just compensation. In other cases, a plaintiff-friendly motive rule does indeed overcompensate, but a portion of overcompensation (or a windfall of some sort) may be inevitable, and so the best a standard can do is minimize and allocate overcompensation—a task for which the But-For standard is poorly suited.

1. Denying windfalls.

In many cases, it is questionable whether we should regard the plaintiff as overcompensated, even if the defendant would certainly have taken the same action if only acceptable motives were considered.166 In particular, when motives constitute part of a larger action, a given act can take on harmful significance in light of inappropriate motives.167 When this occurs, the motive changes something, making it worse, even though it may not change anything physically.

The transformative power of a mental state is familiar to lawyers.168 For example, a journalist’s untrue and provocative story about a public official is “pulp” unless paired with bad motives, in which case it becomes “defamation.”169 A boss’s firing someone for no reason is “arbitrary,” but firing for race is “discrimination.”170 With and without the motive, many facts are the same: the same number of people read the story, the same period of unemployment may be endured.171 Yet something is also different and seemingly more harmful.172

Discrimination in particular has drawn ample commentary to locate the core of its wrongness.173 Is it that it reenacts and participates in practices of oppression?174 That it insults the individual?175 We may debate why motive changes those acts, and whether those transformations ought to count as compensable harms in a given case,176 but this important discussion matter is simply short-circuited when a But-For standard is adopted.

2. Accepting windfalls.

In some cases, we may agree that there is overcompensation but also that it is a feature, not a bug. If plaintiffs face disproportionate challenges in pursuing certain claims, then supracompensatory payouts may be necessary to optimize their incentives.177 Overcompensation also has a place in many schemes of private enforcement. An unemployed teacher may just want his job back, but his suit helps discipline government actors against First Amendment violations. A passed-over accountant may just want her promotion, but her suit spurs employers to end bias in the workplace.178 Getting the right number of these suits to discipline defendants may entail giving plaintiffs more than they lost.

This general principle is widely accepted in both private and public law,179 and it has likewise been suggested as an appropriate consideration for the scope of a motive standard.180

3. Apportioning windfalls.

In many mixed-motives cases, it may be that the law must give someone a better deal than she deserves, it is just a question of whom. In denying a “windfall” to plaintiffs, the But-For standard grants a “windfall” to the defendant.181 Otherwise compensable and wrongful acts avoid liability when the defendant locates an additional acceptable motive.182 In a discrimination case, a blatant and invidious instance of deliberate bigotry escapes justice if the defendant can produce a sufficient work-related reason to fire the employee. In a market manipulation case, a member of a notorious rat pack of traders can escape liability for her actual and intended market abuse if she can produce an additional investment rationale for the abusive trade. The market injury occurs, by design, but goes uncompensated because the defendant also “liked the stock.”183

In all cases in Quadrant II, where both motives are individually sufficient, the plaintiff appears overcompensated for recovering despite the defendant’s bad motive, but the defendant also appears to enjoy good fortune despite her actionable mental state. The But-For standard apportions all windfalls to the defendant, who escapes liability for an otherwise actionable act-motive pair. The But-For standard does not actually reduce the magnitude or frequency of windfalls.

In fact, there is no liability rule that can stop windfalls.184 The best we can do is apportion the windfall to reduce its magnitude or improve the justice or efficiency of its distribution. We might think that it is appropriate to award the windfall to the party whose conduct came closest to blameless, as measured by motives. Then there would still be a plausibly unjust windfall, but it would be the better of the two options.

If deeply bigoted employers paid a windfall to their employees, but slightly bigoted employers retained the windfall of immunity, it might strike us that windfall is being apportioned in keeping with desert. Likewise, if employers pay a windfall when their acceptable motive is quite weak, but they retain the windfall when their acceptable motive loomed quite large for them, then it would give the windfall more often to the employers for whom acceptable motives are more important.185 Figure 15 visually depicts this intuition by using dotted lines to depict the strength of the A-Motive and dashed lines to show the B-Motive. X is an individual with much more significant B-Motive than A-Motive. Holding her liable gives her no credit for her A-Motive, potentially compensating a plaintiff in an amount roughly indicated by the vertical dotted line, but this is far less than the windfall X would obtain if exempted from liability: she would escape any consequences for her much greater B-Motive (indicated by the horizontal dashed line). In other words, better that the shorter vertical line be allocated as windfall than the longer horizontal line. Examination of Y’s motives yields the opposite conclusion. The Primary-Motive test sorts individuals accordingly, apportioning the windfall wherever it is smallest.

Figure 15:  Apportioning Windfalls

This method of limiting and allocating windfalls alleviates the most uncomfortable cases. Imagine a boss fires an employee who literally never works; even if the boss’s B-Motive was large enough to independently drive the firing, the plausibly enormous A-Motive makes it seem unjust for the employee to recover despite her dereliction. Conversely, an unrepentant and brazenly racist boss who fires someone with a string of bigoted epithets probably ought to lose a Title VII suit even though the employee missed enough work to be fired.186 Yes, the employee is treated better than other absentee employees fired that day, but this seems like the less unjust windfall.187

This alternative approach, which allocates the windfall in light of the relative strength of the relevant motives, should prove particularly attractive to those who remain invested in analogies to tort law and theory. When the defendant’s negligence was only one cause of the plaintiff’s injury, there is an analogous windfall to allocate: either a culpable defendant escapes because of a concurrent cause, or a plaintiff recovers from the defendant despite other individuals (including, perhaps, the plaintiff herself) having contributed to the injury. The majority rule is to give the windfall to whomever was least responsible for the outcome: if the plaintiff (or a third party) was most at fault, the plaintiff does not recover from the defendant; if the defendant was the most responsible, then the plaintiff recovers in full.188

The logic of using a Primary-Motive test in apportioning windfalls is not a novel idea.189 Recall that this solution predates the windfall problem itself. Though the windfall problem was born in Mt. Healthy, its embryo was probably a 1970 student note,190 which was cited in Mt. Healthy’s appellate brief.191 That note urged adoption of a Primary-Motive standard to deal with windfalls.192

If we regard it as a windfall whenever the law gives a party something it would not have gotten had the other motive been considered, then windfalls are inevitable. Yet the But-For standard is far from the only or best way of apportioning these windfalls. The Primary-Motive standard also apportions windfalls, and with seemingly better justice and efficiency effects.

The point is not that the Primary-Motive standard is the best test in every case or in all respects.193 It is just that a failure to consider an alternative standard, such as the Primary-Motive standard, shows the incompleteness of windfall arguments in favor of the But-For standard.

V. Other Problems with the But-For Standard

While the foregoing Part criticizes the positive case for the But-For standard, it criticizes the But-For standard itself only obliquely by pointing out that it lacks the justification commonly attributed to it. This Part goes further, identifying problems with the standard that should give pause even to those who passed through Part III unpersuaded.

A. Stability

The But-For standard is unstable and threatens collapse sub rosa into one of the other standards because it is highly sensitive to the court’s definition of the action to be analyzed. Courts can scale up or scale down the granularity of analysis of the action. The standard therefore offers little guidance or constraint to courts looking for a principled way to address motive analysis.

Defining the action granularly can help a plaintiff. A highly granular analysis asks whether any aspect of the action changed as a result of bad motive. For example, did the defendant act at a slightly different time as a result of the bad motive? Then we have a different harm that can be attributed to the motive, and the But-For standard is in some sense satisfied—even though in another sense, the conduct was bound to happen anyway, just a moment sooner or later. And when acts are examined pixel-by-pixel, even tiny levels of bad motive might satisfy a But-For standard. And therefore, the But-For standard reaches into an Any-Motive standard.194

A low granularity test might acknowledge and disregard certain allegedly different acts and results traceable to the motive so long as the gravamen of the act or harm remained the same as would have occurred anyway. By taking a wider view, the bad motive’s effects blur away. Only a deeply bad motive, the sort that would satisfy a Primary-Motive test for example, can predictably survive a wider standard.

For example, consider a case in which a broker was accused of market manipulation.195 Assume that she had ample legitimate reasons to buy the stock in question for her client so that bad motives may not have been a but-for cause of her disruptive purchase.196 A court seemingly could not convict such defendants on a But-For standard.197 Nevertheless, the defendant might have “engaged in trades at particular times, and in particular amounts, because of . . . [a] manipulative scheme.”198 In other words, she might have bought the stock even without a bad motive, but she might not have bought a particular share at a particular moment but for the bad motive. It is easy to bypass a But-For standard in market manipulation cases because “that theory loses its applicability if the prohibited intent alters the trade in any material respect (e.g. by changing the time at which the trade would otherwise have been executed).”199

Courts use similar divide-and-inquire strategies in other contexts, such as redistricting cases,200 in which the battle is often fought by defining the scope of the inquiry. In Miller v Johnson,201 the Court located bad motives in a mixed-motives redistricting case by zeroing in on the rationale for particular stretches of district boundary. “The State admitted that it ‘would not have added those portions of Effingham and Chatham Counties . . . [to the] Eleventh Congressional District but for the need to include additional black population.’”202 Indeed, for some spots, “race was clearly the sole objective behind its creation.”203 The Court recently explicitly endorsed the granular approach in Alabama Legislative Black Caucus v Alabama,204 overruling a district court that had judged the motives for the districting plan “as a whole.”205

While it is legitimate to question how finely to slice up actions for any given inquiry,206 we should be cautious about unprincipled efforts to slice actions in such a way as to circumvent the results of a given motive standard.

For one thing, this general strategy of slicing downward faces serious conceptual problems.207 More seriously, the handicap of meeting the But-For standard for only a sliver of the action may entail reduced remedies—in the present action and others. Imagine that a market manipulation plaintiff indeed bought 1 percent more stock because of her bad motive, or a legislature indeed extended a legislative district by one block because of racial motives. Arguably, the defendant should provide a remedy—the loss on the incremental stock purchase, or a redrawing of the key block—and no more. This is true even if bad motives were prevalent throughout their trading and districting. Our hunger for a more general remedy will go unsated if we conceive of the action as a series of actions, only one of which triggers the law’s essential conditions.

For someone skeptical of the But-For standard, it may be a comfort to see that courts avail themselves of an escape valve. Yet it would be better to be explicit about the rule, both in order to assure similar rule conformity from court to court and also to give notice to litigating parties as a matter of legality or process. Therefore, it would be better to outright endorse one of the alternatives that courts covertly adopt.

Other tests are not as fully pliable as the But-For test because they deliver results that do not strike courts as at odds with the just outcome. A court that wishes to grant some remedy to the plaintiff—small or large—may find greater flexibility with an Any-Motive standard, for example. The defendant is liable on that standard in all mixed-motives cases, with the court able to subsequently condition the size and sort of remedy.

Also, other tests are generally less responsive to the reference class and so cannot be gamed by plausibly redefining the granularity of the action. The Any-Motive standard, for example, is met whenever bad motives played any role in the action. This test can be dodged only by defining the action so narrowly that the bad motive was part of some unrelated conduct. For example, a boss who conceded to describing an employee with racial epithets prior to firing her might concede that race played a role in her decision-making, but just as to talking about the defendant with the HR department, not as to employment decisions. Then the bad motive is isolated to some discrete and largely separate action. This is hardly a plausible defense.

The Primary-Motive test is likewise less sensitive to reference class adjustment. If the primary motive for a complex action is B-Motive, there is a fair chance that the primary motive for any given part of it was also primarily motived by B-Motive. For example, if the legislature’s primary motive in drawing district boundaries was to disenfranchise racial minorities, any contested boundary line is very likely motivated by that same impulse. To be sure, a court could zoom in on a particular district boundary and note that this line had no illicit motive; but that just concentrates the troublesome motives at the lines that were most objectionable anyway. On a Primary-Motive test, a defendant who successfully lobbies a court to subdivide the action stands to lose as much as is gained.

B. Silence

The But-For standard is insensitive to, and tends to suppress, inquiries that really ought to be encouraged. This is because the But-For standard is frequently invoked in cases in which the bad motive is plainly vital to the inquiry and yet allows the litigation to proceed without ever examining the bad motive.

The But-For standard turns only on acceptable motives. If acceptable motive is less than 1 (meaning it is independently insufficient), any resultant action will incur liability.208 That is because some bad motive must have been present to establish sufficient motivation for the observed action; without that bad motive, the action would not have occurred given the low level of other motive. Conversely, if acceptable motive is greater than 1, liability is avoided regardless of the level of bad motive. Because the level of A-Motive ends up solely deciding the result, B-Motive is of only derivative importance.

This emphasis on acceptable motive relative to bad motive could have undesirable effects on trial practice. A defendant who concedes the presence of bad motives would convert the trial into a referendum on her acceptable motives. If these are greater than 1, the defendant wins and the bad motives are irrelevant. Regardless of who bears the burden of establishing the level of acceptable motive, this fixation on acceptable motives de-emphasizes the bad motives in the adjudicatory process.

Indeed, a court could reasonably exclude evidence bearing on bad motive. For example, the plaintiff in an employment discrimination suit may wish to introduce invective-filled emails of the defendant’s rampant bias.209 With bias admitted already, this evidence might be excludible as not probative of any element in dispute (and, of course, more than a little prejudicial).210

At most, such evidence might be admissible to discredit the defendant’s proffered acceptable motives. Yet the point of mixed-motive analysis is that defendants sometimes have two motives, neither of which is pretextual. If courts take seriously the possibility of mixed motives in a given case, savvy defendants may deprive plaintiffs and the public of the airing of the very facts central to their injury. Though the law is meant to address bad motives, 211 the trial may not concern them.

Showing a particularly high level of bad motive cannot help satisfy a But-For standard once the presence of bad motive has been accepted at all. Yet if we wish to have a public airing of the bad motives, we should use a test that invokes bad motive as an actual element.

Conclusion

In the foregoing Sections, I criticized reasons commonly cited in favor of the But-For standard in motive analysis. Causation, freedom, and compensation all fail to establish the viability of the But-For motive standard, at least relative to alternative standards. Each also entered the law through less than commendable reasoning.

Some may feel that these criticisms were more effective for some areas of law than others and that there are even domains in which the But-For standard remains justified. Moreover, even if these criticisms were successful, it does not preclude the possibility that other arguments in favor of the standard could be offered. Still, whatever ground was given up in the foregoing pages should serve to caution against a return to the But-For standard in any area. For many, the intuitive appeal of the But-For standard seems immediate and plausible and does not turn on fine details, such as precise statutory language or the availability of liability rules as penalties. When a complex and dry theory is disproven in a variety of contexts, it can be right to save it with more complexity; when a simple and appealing theory is disproven in a variety of contexts, it can be a sign that its appeal was illusory.

So why is the But-For standard the presumed standard? In part, it reflects an inchoate recognition that several other tests amount to single-motive inquiries.212 And in part, this merely reflects a lack of awareness of other possible motive rules. One cannot disambiguate if one doesn’t know that there is ambiguity.

Yet its core support must be discomfort with motive analysis. Arguments that might have slayed the whole enterprise instead shackled it. I predict that support for the But-For standard will not wane until we have achieved appropriate comfort with, or ended, the motive analysis enterprise. Therefore, in a future project, I will engage the question of why and how the law uses motives.

This Article showed that other standards—the Any-Motive standard, the Primary-Motive standard, and others—each has something to offer and deserves serious consideration. Deciding which, if any, should replace the But-For standard in a given area of law requires considering what that area of law is trying to achieve through the use of motives. Differing mixtures of practical and normative considerations will lead to different motive standards in many areas, but there is reason to be optimistic about a principled and systematic approach filling the space left by the But-For standard—a coming triumph of mixed-motives jurisprudence.

• 1Mt. Healthy City School District Board of Education v Doyle, 429 US 274, 281–82 (1977). Doyle’s suit was filed under 42 USC § 1983, which authorizes individuals to sue for damages or injunctive relief in response to violations of their constitutional rights. Note that the obscene gesture would probably not upset contemporary sensibilities. See Brief for Respondent Fred Doyle, Mt. Healthy City School District Board of Education v Doyle, No 75-1278, *7 (US filed Aug 2, 1976) (available on Westlaw at 1976 WL 181614) (Doyle Brief) (stating that the meaning of Doyle’s gesture depended on geography: in Texas, it would be recognized as the “Hook ’em Horns” gesture associated with the University of Texas athletics program; in Ohio, it had a “somewhat different connotation also related to bulls”).
• 2Mt. Healthy, 429 US at 282–83 (noting that, in its decision to not rehire Doyle, the school district mentioned that Doyle discussed the school’s proposed dress code with a radio personality).
• 3429 US 274 (1977).
• 4Id at 287 (holding that the school district is not liable if it can prove “that it would have reached the same decision as to [Doyle’s] reemployment even in the absence of the protected conduct”).
• 5As of October 12, 2018, searching Westlaw for federal opinions containing the words “mixed motive(s)” returns over ten thousand cases, and the Supreme Court decides a mixed-motives case essentially every other year.
• 6Almost five thousand cases (including eighteen Supreme Court opinions) cite the Westlaw headnote that links to this test.
• 7See, for example, University of Texas Southwestern Medical Center v Nassar, 570 US 338, 343 (2013); Gross v FBL Financial Services, Inc, 557 US 167, 180 (2009).
• 8See, for example, Securities and Exchange Commission v Masri, 523 F Supp 2d 361, 372 (SDNY 2007).
• 9See, for example, Mass Gen Laws Ann ch 186, §§ 17–18.
• 10See, for example, Price Waterhouse v Hopkins, 490 US 228, 240 (1989):

  In determining whether a particular factor was a but-for cause of a given event, we begin by assuming that that factor was present at the time of the event, and then ask whether, even if that factor had been absent, the event nevertheless would have transpired in the same way.

• 11See, for example, Board of Education, Island Trees Union Free School District No 26 v Pico, 457 US 853, 871 n 22 (1982) (“By ‘decisive factor’ we mean a ‘substantial factor’ in the absence of which the opposite decision would have been reached.”), citing Mt. Healthy, 429 US at 287.
• 12See, for example, Hartman v Moore, 547 US 250, 260 (2006) (applying the But-For standard to retaliatory prosecution against a lobbyist); Givhan v Western Line Consolidated School District, 439 US 410, 417 (1979) (remanding to determine whether the petitioner’s exercise of First Amendment rights was a but-for cause of the school district’s failure to rehire her rather than “the ‘primary’” reason).
• 13See, for example, State v Ornelas, 330 P3d 1085, 1092–93 (Idaho App 2014) (referring to the But-For test, with burden shifting, as the “mixed-motives analysis” as contrasted to other approaches to mixed-motives jury strikes); Ndonyi v Mukasey, 541 F3d 702, 710–11 (7th Cir 2008) (criticizing the Board of Immigration Appeals in an asylum case because it “completely ignored the doctrine of mixed motives—the opinion does not analyze whether Ndonyi’s oppressors were partially motivated by politics or religion”).
• 14Scholars have urged its adoption in numerous fields. See, for example, Daniel J. Hemel and Eric A. Posner, Presidential Obstruction of Justice, 106 Cal L Rev 1277, 1319 (2018). These endorsements come often, sometimes without considering alternatives. See, for example, Laurence H. Tribe, The Mystery of Motive, Private and Public: Some Notes Inspired by the Problems of Hate Crime and Animal Sacrifice, 1993 S Ct Rev 1, 32 n 77 (assuming without argument the applicability of a “But-For” test). And sometimes after considering alternatives. See, for example, Paul Brest, Palmer v. Thompson: An Approach to the Problem of Unconstitutional Legislative Motive, 1971 S Ct Rev 95, 119 (rejecting the “‘sole or dominant’ motivation” standard in favor of asking whether illicit motive “determined the outcome of the decision”), quoting Palmer v Thompson, 403 US 217, 224–25 (1971). See also Jed Rubenfeld, The First Amendment’s Purpose, 53 Stan L Rev 767, 777 (2001) (“Because a content-based regulation makes communication itself an element of the prohibited act, the communication effected by the act must be a but-for cause of whatever wrong or harm the regulation is supposed to prevent.”).
• 15The But-For standard has met stiff resistance from employment law scholars. See, for example, Martin J. Katz, The Fundamental Incoherence of Title VII: Making Sense of Causation in Disparate Treatment Law, 94 Georgetown L J 489, 515–27 (2006). Other seeming critics actually endorse the But-For standard, subject to modification. For example, Professor Mark Brodin would retain a But-For test at least in Title VII cases seeking a reinstatement and back pay but would simply place the burden of proof on the defendant. See Mark S. Brodin, The Standard of Causation in the Mixed-Motive Title VII Action: A Social Policy Perspective, 82 Colum L Rev 292, 323–24 (1982). This is not a rejection of the But-For standard so much as it is a request to accommodate the But-For standard.
• 16See, for example, Anjum Gupta, The New Nexus, 85 U Colo L Rev 377, 457 (2014) (proposing a But-For standard in asylum claims).
• 17See Parts II.A.1, IV.A.2.
• 18See Part V.A.
• 19See Part V.B.
• 20See Part IV.B.
• 21See Part III.
• 22See, for example, Kingston v Chicago & Northwest Railway, 211 NW 913, 915 (Wis 1927).
• 23See Part II.B.
• 24See, for example, Andrew Verstein, The Jurisprudence of Mixed Motives, 127 Yale L J 1106, 1112–13 (2018).
• 25This Article does not address whether motive analysis, rather than some entirely objective test, is appropriate in a given domain. I take the law as it is, which sometimes involves motive analysis. Nor does my criticism of the But-For standard culminate in a clear, issue-by-issue endorsement of an alternative test. That task would require introducing and defending a whole normative jurisprudence of mixed motives, which is the task for the next normative paper to follow this one. For now, I often hold up other motive standards as alternatives to show that substituting another motive standard would often lead to better results than the But-For standard. This is both to show problems with the But-For standard and to stoke the imagination: if we see that it is easy to list alternatives that outshine the But-For standard, we may become receptive to yet other possibilities.
• 26See, for example, Nassar, 570 US at 343; Gross, 557 US at 179.
• 27Pub L No 103-353, 108 Stat 3149 (1994), codified as amended at 38 USC § 4301 et seq.
• 2838 USC § 4311(a). Its scope is actually wider than just prior military service, covering also current and future military service.
• 2938 USC § 4311(c)(1).
• 30Securities and Exchange Commission v Masri, 523 F Supp 2d 361, 372
  (SDNY 2007).
• 31In Masri, a trader bought two hundred thousand shares of a Mexican television station in the last minutes of the trading day, which represented the great majority of the day’s purchase. Id at 362–65. The trader had a clear financial reason to spike the market price: he was contractually committed to buy 860,000 shares at $5 if the price finished below that point that day. By pushing the price up, he averted an overpayment loss of more than $100,000. On the other hand, the trader also had nonmanipulative reasons to buy the stock: he had a looming obligation to deliver half a million shares a few months hence, and he deemed the present price low enough to warrant early action. So here we have a case in which two different agendas might suggest the same conduct: buy as much as you can carry.
• 32See generally Verstein, 127 Yale L J 1106 (cited in note 24).
• 33Pub L No 101-336, 104 Stat 327 (1990), codified at 42 USC § 12101 et seq.
• 34See Serwatka v Rockwell Automation, Inc, 591 F3d 957, 962 (7th Cir 2010) (“[A] plaintiff complaining of discriminatory discharge under the ADA must show that his or her employer would not have fired him but for his actual or perceived disability; proof of mixed motives will not suffice.”); Gentry v East West Partners Club Management Co Inc, 816 F3d 228, 235 (4th Cir 2016) (“The only remaining question is whether the ADA’s text calls for a ‘but-for’ causation standard. We hold that it does.”). For other areas of law using the But-For standard, see notes 7–13 and accompanying text.
• 35Restatement (Second) of Torts § 668, cmt c (1977).
• 36Directors’ decisions are immunized from shareholder challenge if their motives were primarily “loyal” even if they had some personal interest in the decision. See Alan R. Palmiter, Reshaping the Corporate Fiduciary Model: A Director’s Duty of Independence, 67 Tex L Rev 1351, 1389 n 151 (1989) (describing mixed motives in corporate law). See, for example, Panter v Marshall Field & Co, 646 F2d 271, 304 (7th Cir 1981); Johnson v Trueblood, 629 F2d 287, 293 (3d Cir 1980); Cheff v Mathes, 199 A2d 548, 554 (Del 1964).
• 37See, for example, Bush v Vera, 517 US 952, 959 (1996); Miller v Johnson, 515 US 900, 916 (1995). Racially motivated gerrymanders are subject to strict scrutiny, which is often regarded as “‘strict’ in theory and fatal in fact.” Gerald Gunther, The Supreme Court, 1971 Term—Foreword: In Search of Evolving Doctrine on a Changing Court: A Model for a Newer Equal Protection, 86 Harv L Rev 1, 8 (1972). But see generally Adam Winkler, Fatal in Theory and Strict in Fact: An Empirical Analysis of Strict Scrutiny in the Federal Courts, 59 Vand L Rev 793 (2006) (examining the results of strict scrutiny challenges).
• 38See William A. Klein, The Deductibility of Transportation Expenses of a Combination Business and Pleasure Trip—A Conceptual Analysis, 18 Stan L Rev 1099, 1104 (1966); Walter J. Blum, Motive, Intent, and Purpose in Federal Income Taxation, 34 U Chi L Rev 485, 508 (1967). See also, for example, Mohn v United States, 2001 WL 1399366, *5 (ED Mich) (business expenses); Olk v United States, 536 F2d 876, 879 (9th Cir 1976) (gifts); Bank of Palm Beach & Trust Co v United States, 476 F2d 1343, 1350 (Ct Cl 1973) (death benefits); United States v Generes, 405 US 93, 104 (1972) (bad business debts).
• 39See Jane Harris Aiken, Ex post Facto in the Civil Context: Unbridled Punishment, 81 Ky L J 323, 353–59 (1992). See, for example, Wiley v Bowen, 824 F2d 1120, 1122 (DC Cir 1987); Flemming v Nestor, 363 US 603, 612 (1960). See also US Const Art I, § 9, cl 3 (barring ex post facto federal laws). The general ban on punitive law derives in part from this Bill of Attainder Clause. Id.
• 40See William M. Carter Jr, Whren’s Flawed Assumptions Regarding Race, History, and Unconscious Bias, 66 Case W Res L Rev 947, 953 (2016). See, for example, United States v Avery, 137 F3d 343, 358 (6th Cir 1997). A similar test is applied for probable cause of searches generally. See, for example, Commonwealth v Adams, 605 A2d 311, 313 (Pa 1992).
• 41See Dan B. Dobbs, Paul T. Hayden, and Ellen M. Bublick, The Law of Torts § 625 (West 2d ed 2011).
• 42See, for example, United States v 480.00 Acres of Land, 557 F3d 1297, 1311 (11th Cir 2009) (takings judged by the “primary purpose” standard); Holbrook v Morrison, 100 NE 1111, 1111 (Mass 1913) (selling property to putatively undesirable owners); Rideout v Knox, 19 NE 390, 391 (Mass 1889) (spite wall); Ronald J. Krotoszynski Jr, Expropriatory Intent: Defining the Proper Boundaries of Substantive Due Process and the Takings Clause, 80 NC L Rev 713, 733–34 (2002) (proposing that courts should ask whether the primary purpose of a taking is expropriatory rather than regulatory).
• 43See, for example, Goldstein v Commissioner of Internal Revenue, 364 F2d 734, 741 (2d Cir 1966).
• 4442 USC § 1320a-7b(b)(2)(B) (criminalizing payment “to induce” purchase of items or services for which Medicare or Medicaid will ultimately make payment). See United States v Narco Freedom, Inc, 95 F Supp 3d 747, 753 (SDNY 2015) (discussing the statute).
• 45Compare Singh v Gonzalez, 406 F3d 191, 197 (3d Cir 2005) (“[A]n applicant must show that the persecution was motivated, at least in part, by one of the protected characteristics.”) (emphasis omitted), with Gebremichael v Immigration and Naturalization Service, 10 F3d 28, 35 (1st Cir 1993) (finding that the enumerated ground must be at the “root of persecution”). The former is the majority rule. See, for example, Lopez-Soto v Ashcroft, 383 F3d 228, 236 (4th Cir 2004); Marku v Ashcroft, 380 F3d 982 (6th Cir 2004); Girma v Immigration and Naturalization Service, 283 F3d 664, 667 (5th Cir 2002); Singh v Ilchert, 63 F3d 1501, 1509 (9th Cir 1995); Osorio v Immigration and Naturalization Service, 18 F3d 1017, 1028 (2d Cir 1994). See also Marisa Silenzi Cianciarulo, Terrorism and Asylum Seekers: Why the Real ID Act Is a False Promise, 43 Harv J Legis 101, 115–20 (2006) (discussing mixed motives).
• 46See Verstein, 127 Yale L J at 1114 n 42 (cited in note 24).
• 47See, for example, State v Ornelas, 330 P3d 1085, 1092 (Idaho App 2014) (noting that “most states have adopted what is the [sic] referred to as the per se approach”); Robinson v United States, 890 A2d 674, 680 (DC 2006); McCormick v State, 803 NE2d 1108, 1112–13 (Ind 2004); State v Lucas, 18 P3d 160, 163 (Ariz App 2001); State v Shuler, 545 SE2d 805, 811 (SC 2001); Payton v Kearse, 495 SE2d 205, 210 (SC 1998); State v King, 572 NW2d 530, 535 (Wis App 1997); Rector v State, 444 SE2d 862, 865 (Ga App 1994); Owens v State, 531 S2d 22, 23–24 (Ala Crim App 1987).
• 48Employee Retirement Income Security Act, Pub L No 93-406, 88 Stat 829 (1974). See, for example, Teumer v General Motors Corp, 34 F3d 542, 550 (7th Cir 1994) (liable if illicit motive “contributed toward the employer’s decision”); Titsch v Reliance Group Inc, 548 F Supp 983, 985 (SDNY 1982) (contrasting “mere consequence” with “a motivating factor”); Pickering v USX Corp, 809 F Supp 1501, 1532 (D Utah 1992) (“[A]n employee need only prove that the desire to defeat pension eligibility was a ‘motivating’ or ‘determinative,’ factor behind the challenged conduct.”), quoting Gavalik v Continental Can Co, 812 F2d 834, 860 (3d Cir 1987). See also Schlenz v United Airlines, Inc, 678 F Supp 230, 234–36 (ND Cal 1988) (rejecting a sole-motive standard); Nemeth v Clark Equipment Co, 677 F Supp 899, 903 (WD Mich 1987) (rejecting a sole-motive standard). See generally Christina A. Smith, Note, The Road to Retirement—Paved with Good Intentions but Dotted with Potholes of Untold Liability: ERISA Section 510, Mixed Motives and Title VII, 81 Minn L Rev 735 (1997).
• 49In University of Texas Southwestern Medical Center v Nassar, 570 US 338 (2013), there is one policy argument: “The proper interpretation and implementation of [the statute] and its causation standard have central importance to the fair and responsible allocation of resources in the judicial and litigation systems. This is of particular significance because claims of retaliation are being made with ever-increasing frequency.” Id at 343, 358. Interestingly, critics of causation in employment law have long focused on the distribution of costs associated with litigation. See Cynthia L. Estlund, Wrongful Discharge Protections in an At-Will World, 74 Tex L Rev 1655, 1673 (1996). Merely culling lawsuits is of course not a good thing unless more is said about the quality of the lawsuits defeated and the costs and benefits of addressing them some other way.
• 50Several scholars are skeptical that motives should ever be evaluated for their causal relationships. See generally H.L.A. Hart and Tony Honoré, Causation in the Law (Oxford 2d ed 1985); Paul J. Gudel, Beyond Causation: The Interpretation of Action and the Mixed Motives Problem in Employment Discrimination Law, 70 Tex L Rev 17 (1991) (rejecting that motives can be causal). Professor Martin Katz recognizes such skepticism but nevertheless proceeds as though motive can be causal, which he takes courts to accept. Katz, 94 Georgetown L J at 495 n 17 (cited in note 15). But see Richard Thompson Ford, Bias in the Air: Rethinking Employment Discrimination Law, 66 Stan L Rev 1381, 1406 (2014) (“[W]e have realized at least since the time of David Hume that causation is less a physical fact than a philosophical dilemma.”).
• 51Mt. Healthy is often cited, but so is Price Waterhouse v Hopkins, in which causation was accepted by all four opinions. Price Waterhouse v Hopkins, 490 US 228, 237 (1989) (Brennan) (plurality); id at 259 (White concurring); id at 262 (O’Connor concurring); id at 281 (Kennedy dissenting). The plurality opinion differed from the others only in its insistence that causation need not be but-for causation. Id at 241–42 (Brennan) (plurality). Note that, despite rejecting but-for causation, the plurality still would have denied a plaintiff any remedy if a But-For standard is not met. Id at 249 (Brennan) (plurality) (“A court that finds for a plaintiff under this standard has effectively concluded that an illegitimate motive was a ‘but-for’ cause of the employment decision.”). Evidentiary problems mattered, but only with respect to the burden shifting procedure, not the ultimate But-For standard. Id at 271–75 (O’Connor concurring).
• 52See Nassar, 570 US at 346–51.
• 53Id at 347. The second sentence of the opinion explains: “The requisite relation between prohibited conduct and compensable injury is governed by the principles of causation, a subject most often arising in elaborating the law of torts.” Id at 342.
• 54Id at 346–47.
• 55570 US 338 (2013).
• 56Id at 347.
• 57Pub L No 90-202, 81 Stat 602 (1967), codified at 29 USC § 621 et seq.
• 5829 USC § 623(a)(1). Similar language supported causation analysis in Price Waterhouse. That case’s statute—42 USC § 2000e-2(a)(2)—likewise prohibited adverse employment action “because of” prohibited factors, and all but one justice explicitly asserted a causal relationship between motive and act. Price Waterhouse, 490 US at 241 (Brennan) (plurality); id at 262 (O’Connor concurring); id at 281 (Kennedy dissenting).
• 59Gross v FBL Financial Services, Inc, 557 US 167, 176 (2009).
• 60Id at 177 (“It follows, then, that under § 623(a)(1), the plaintiff retains the burden of persuasion to establish that age was the ‘but-for’ cause of the employer’s adverse action.”).
• 61United States v Generes, 405 US 93, 105 (1972).
• 62Nassar, 570 US at 360, and Gross, 557 US at 176–77, adopt a But-For standard familiar from Price Waterhouse. In Price Waterhouse, causation was the only argument accepted by all four opinions, and all four opinions cited Mt. Healthy. Price Waterhouse, 490 US at 248–49 (Brennan) (plurality); id at 258–60 (White concurring); id at 277 (O’Connor concurring); id at 289–90 (Kennedy dissenting).

  Village of Arlington Heights v Metropolitan Housing Development Corp, 429 US 252 (1977), also deserves some credit for introducing motive-causation. In an oft-cited footnote, the Court wrote:

  [If] the same decision would have resulted even had the impermissible purpose not been considered . . . the complaining party in a case of this kind no longer fairly could attribute the injury complained of to improper consideration of a discriminatory purpose. In such circumstances, there would be no justification for judicial interference with the challenged decision.

  Id at 270–71 n 21.

  Although this passage does not use the word “causation,” later courts have cited it to justify their causal approach to mixed motives. See, for example, Price Waterhouse, 490 US at 249 (“A court that finds for a plaintiff under this standard has effectively concluded that an illegitimate motive was a ‘but-for’ cause of the employment decision.”), citing Arlington Heights, 429 US at 270–71 n 21. Still, sharing credit between Arlington Heights and Mt. Healthy would do little to extend motive-causation’s ancestry: these two cases were decided on the same day. Note that Nassar and Gross differ from earlier But-For standard cases in their allocation of the burden of proof. Price Waterhouse permitted the burden to shift to the defendants, while Nassar and Gross did not. This article focuses only on the motive standard, not on who bears the burden of proving it. See Verstein, 127 Yale L J at 1134 n 99 (cited in note 24).

• 63Mt. Healthy, 429 US at 286.
• 64See, for example, Parker v North Carolina, 397 US 790, 796 (1970); Wong Sun v United States, 371 US 471, 491 (1963); Lyons v Oklahoma, 322 US 596, 607 (1944). The Court cited cases overlooking improperly obtained confessions or guilty pleas for analogy despite noting the differences between that context and Doyle’s claim: “While the type of causation on which the taint cases turn may differ somewhat from that which we apply here, those cases do suggest that the proper test to apply in the present context. . . .” Mt. Healthy, 429 US at 287.
• 65One of the three cases (Parker) cites to a case (Nardone) that does invoke causal metaphors. Parker, 397 US at 796, citing Nardone v United States, 308 US 338, 341 (1939) (“Sophisticated argument may prove a causal connection between information obtained through illicit wire-tapping and the Government’s proof.”). Later courts have likewise drawn on causation as an analogy. In re M.S., 896 P2d 1365, 1385 n 1 (Cal 1995) (Kennard concurring) (“Although the analogy is admittedly not exact, it is sufficiently close to be helpful, at least in the absence of another and superior method of analysis.”). But none of those causation passages is quoted or cited in Mt. Healthy or even Parker. And even Nardone does not derive from causation a clear endorsement of a But-For standard. Nardone, 308 US at 341 (“A sensible way of dealing with such a situation . . . ought to be within the reach of experienced trial judges.”). So motive-causation was born in an analogy to cases that address neither motive nor causation.

  One could have pushed the roots further back still: Why was causation cited in even one tainted evidence case? It is possible that causation’s appeal has sometimes operated more subliminally than rationally. One of the evidence cases cited in Mt. Healthy mentioned “probable cause” twenty-four times. See generally Wong Sun, 371 US 471. Likewise, all Batson decisions are rife with references to “for cause” strikes. Is it possible that the mere earliest germ of causation analysis was simple homophony?

• 66Nor is there discussion of causation in Mt. Healthy’s lower court opinions, appellate briefs, or oral argument. The closest reference I can find to causation is as follows:

  Where, for example, an employer is shown to have engaged in a pattern and practice of racial discrimination in hiring, the impermissible factor of race can be said to have played a part each time a black employee was denied a job.

  . . .

  In the instant case the showing is stronger: here the finding of causation does not rest upon an inference from a pattern of conduct, but rather upon the employer’s own admission that he relied upon the impermissible factor.

  Doyle Brief at *43 & n 20 (cited in note 1). See also Michael Wells, Three Arguments against Mt. Healthy: Tort Theory, Constitutional Torts, and Freedom of Speech, 51 Mercer L Rev 583, 599 (2000) (citation omitted):

  Despite the importance of causation, Mt. Healthy addresses the issue in just three pages at the end of an opinion devoted mainly to other matters. No doubt three pages would be sufficient, if they were used wisely. However, the Court in Mt. Healthy offered only fragments of arguments rather than cogent reasoning. The result is an unconvincing hodgepodge that leaves unanswered the difficult questions raised by the causation issue.

• 67Nassar, 570 US at 342.
• 68Id at 347.
• 69Id at 346.
• 70But see United States v Burke, 504 US 229, 241 (1992) (holding that Title VII claims are not tort-like for the purpose of excluding damages awarded for them from gross income under IRS regulations).
• 71See Batson v Kentucky, 476 US 79, 89 (1986); J.E.B. v Alabama, 511 US 127, 129 (1994) (holding “that gender, like race, is an unconstitutional proxy for juror competence and impartiality”).
• 72See, for example, Gattis v Snyder, 278 F3d 222, 234–35 (3d Cir 2002); Weaver v Bowersox, 241 F3d 1024, 1032 (8th Cir 2001); King v Moore, 196 F3d 1327, 1335 (11th Cir 1999); Wallace v Morrison, 87 F3d 1271, 1274–75 (11th Cir 1996); United States v Tokars, 95 F3d 1520, 1533 (11th Cir 1996); United States v Darden, 70 F3d 1507, 1531 (8th Cir 1995); Jones v Plaster, 57 F3d 417, 421 (4th Cir 1995); Howard v Senkowski, 986 F2d 24, 26 (2d Cir 1993).
• 73See, for example, Kesser v Cambra, 465 F3d 351, 372 (9th Cir 2006), citing Mt. Healthy, 429 US at 287; Arlington Heights, 429 US at 270 n 21.
• 74Other areas of law have rejected the But-For standard in large part because the court rejected the tutelage of tort doctrine. See, for example, Generes, 405 US at 105.
• 75See W. Jonathan Cardi, The Role of Negligence Duty Analysis in Employment Discrimination Cases, 75 Ohio St L J 1129, 1160–61 (2014) (“[I]f courts are indeed engaging in common-law tort duty analysis, such practice is improper under the law governing statutory torts.”); Jon O. Newman, Suing the Lawbreakers: Proposal to Strengthen the Section 1983 Damage Remedy for Law Enforcers’ Misconduct, 87 Yale L J 447, 461 & n 59 (1978) (arguing that “[c]ommon law notions . . . simply have no place” in statutory civil rights litigation).
• 76Securities law scholars are familiar with the Supreme Court’s ambivalent approach developing 10b-5 claims upon common law tort theory. See, for example, Stoneridge Investment Partners, LLC v Scientific-Atlantic, Inc, 552 US 148, 162 (2007) (“Section 10(b) does not incorporate common-law fraud into federal law.”). See also generally Jill E. Fisch, Cause for Concern: Causation and Federal Securities Fraud, 94 Iowa L Rev 811 (2009). Section 11 of the Securities Act does not require causation. See Robert Belton, Causation in Employment Discrimination Law, 34 Wayne L Rev 1235, 1242–43 n 27 (1988).
• 77Oxford English Dictionary 693 (Clarendon 2d ed 1989) (defining “fallacy of division, the fallacy that whatever is true of a whole must be true of any part or member of that whole”).
• 78See John R. Searle, Minds, Brains, and Science 22 (Harvard 1984).
• 79See Fire Insurance Exchange v Altieri, 1 Cal Rptr 2d 360, 364 (Cal App 1991), quoting J.C. Penney Casualty Insurance Co v M.K., 804 P2d 689, 699 (Cal 1991):

  The focus is on the harmful nature of the insured’s act, not on his or her subjective intent to cause harm. “Motive is irrelevant for purposes of [law barring insurance coverage for intentional torts]. Motive is relevant only to the different question of whether the conduct was wrongful, thereby giving rise to liability.”

  See also Atlantic Lloyd’s Insurance Co of Texas v Scott Wetzel Services, Inc, 1997 WL 211537, *5 (Tex App) (“Intent to cause injury is irrelevant when determining whether the injury was caused by accident.”).

• 80The closest I can find are stray sentences, such as the following: “Additionally, the defendant’s conduct must not only be intentional, but must also be ‘improper in motive or means, causing harm to the plaintiff.’” Premier Technical Sales, Inc v Digital Equipment Corp, 11 F Supp 2d 1156, 1166 (ND Cal 1998), citing Draghetti v Chmielewski, 626 NE2d 862 (Mass 1994). Such dicta is far from convincing proof that mental states must causally impact the harm.
• 81It is true that tort law may require that the tortious aspect of a plaintiff’s conduct must contribute to an injury. Restatement (Third) of Torts: Liability for Physical and Emotional Harm § 26, cmt g. For example, a hunter is not liable for negligently handing a loaded rifle to a child if the child’s subsequent injury comes only from dropping the gun on her toe. Restatement (Third) of Torts: Liability for Physical and Emotional Harm § 29, cmt d, illus 3 (2010). Such a defendant was clearly a cause of the injury, but he escapes liability because the injury (a broken toe) bore no causal relationship to his negligent act (leaving bullets in the gun). It might seem that causally inert wrongfulness cannot give rise to liability.

  Alas, we have traced tort principles upriver from causation into another tributary entirely. The rifle injury example is drawn from a Restatement section on limitations to scope of liability, what was once called “proximate cause.” The current Restatement discourages use of the term “proximate cause” as a source of confusion. Restatement (Third) of Torts: Liability for Physical and Emotional Harm § 27, cmt b (2010). The idea that the wrongfulness of the act must create the harm is not part of “cause-in-fact.” It is not part of the empirical sort of causation required by Justice Kennedy in Nassar and that we have been exploring. It is instead part of the normative, policy-driven, world of assignment-of-responsibility. To state that individuals can sometimes escape liability for harms they cause, even when they have acted wrongfully, is merely to highlight the importance of identifying the factors that make salient one wrongful act but not another. And that is the task for some inquiry other than causation.

• 82See Gudel, 70 Tex L Rev at 92 (cited in note 50). Nor does the legislative history of important statutes decisively establish motive-causation. See Heather K. Gerken, Note, Understanding Mixed Motives Claims under the Civil Rights Act of 1991: An Analysis of Intentional Discrimination Claims Based on Sex-Stereotyped Interview Questions, 91 Mich L Rev 1824, 1839 (1993). The conference report accompanying the Civil Rights Act explains that “any discrimination that is actually shown to play a role in a contested employment decision may be the subject of liability. Conduct or statements are relevant under this test only if the plaintiff shows a nexus between the conduct or statements and the employment decision at issue.” HR Rep No 101-856, 101st Cong, 2d Sess 18–19 (1990). If an employment action can be “because of” motive when that motive plays a role or has a nexus with the outcome, it suggests that causation is not the only way to establish an appropriate link. See Black’s Law Dictionary 1205 (Thomson Reuters 10th ed 2014) (defining “nexus” as “[a] connection or link, often a causal one”).
• 8329 USC § 623(a). See also 42 USC § 2000ff-1(a) (barring discrimination “because of genetic information”); 42 USC §§ 3604(a), 3605 (barring discrimination “because of” protected statuses in connection with housing or financing of real estate).
• 8442 USC § 2000e-2(a).
• 85See, for example, Price Waterhouse, 490 US at 241 (Brennan) (plurality) (arguing that “because of” need not be “divest[ed] of causal significance”); id at 260 (White concurring); id at 262 (O’Connor concurring) (arguing that the plain language of Title VII requires causation); id at 281–82 (Kennedy dissenting).
• 86See, for example, Noah D. Zatz, The Many Meanings of “Because of”: A Comment on Inclusive Communities Project, 68 Stan L Rev Online 68, 72 (2015) (“The phrase ‘because of’ refers directly to causal concepts, not mental ones.”); Cardi, 75 Ohio St L J at 1135 (cited in note 75) (accepting that “because of” means causation); David S. Schwartz, When Is Sex Because of Sex? The Causation Problem in Sexual Harassment Law, 150 U Pa L Rev 1697, 1709 (2002) (“The phrase ‘because of’ implies a causal connection.”).
• 87See Whitley R.P. Kaufman, Motive, Intention, and Morality in the Criminal Law, 28 Crim Just Rev 317, 323 (2003). Courts engaged in motive analysis frequently emphasize “explanation” rather than causation. See, for example, Shaw v Reno, 509 US 630, 643–44 (1993).
• 88See generally Wesley C. Salmon, Scientific Explanation and the Causal Structure of the World (Princeton 1984) (arguing that explanation amounts to description of causes). See also Carl G. Hempel, Aspects of Scientific Explanation and Other Essays in the Philosophy of Science 331–497 (Free Press 1965).
• 89See Gudel, 70 Tex L Rev at 92 (cited in note 50).
• 90See Bas C. van Fraassen, The Scientific Image 97–153 (Clarendon 1980). On a constructivist view, explanations must be empirically adequate. A constructivist judge may find a firing to be “because of” motive if a motive-based account was consistent with all of the observed data. This would suggest a far more permissive standard than a But-For standard—probably an Any-Motive standard.
• 91See generally Peter Achinstein, The Nature of Explanation (Oxford 1983). On Professor Peter Achinstein’s account, something counts as an explanation if it both is an assertion of some true state of affairs and also succeeds in helping a particular individual understand why that state of affairs in fact answers some relevant question. For examples, see Peter Achinstein, The Pragmatic Character of Explanation, 1984 Proceedings of the Biennial Meeting of the Philosophy of Science Association 275, 282. Thus, “because of race” on such a pragmatic account might be understood to emphasize the comprehensibility of race as an explanation to particular listeners (The victim? The factfinder?) rather than from some objective, depersonalized perspective. See also id at 286.
• 92See generally Martha Craven Nussbaum, Aristotle on Teleological Explanation, in Martha Craven Nussbaum, ed, Aristotle’s De Mout Animalium 59–99 (Princeton 1978). A teleological explanation cites the goal-directed nature of a thing. For example, why did a lion rush at a flock of sheep? A nonteleological explanation may cite antecedent physical changes (a muscle spasm here and there), but it may be simpler and more useful to say that the nature of a lion is a kind of boldness directed toward its survival and that this is why the lion charged. A teleological discrimination statute might use “because of” to denote instances in which the most general and predictive explanation of the conduct is race. For example, when someone fires someone for race (among other reasons) and this is the sort of reason this boss generally and predictably accepts, then we have explained the firing as because of race in the teleological sense. Conversely, even when race may have been a significant factor, it might not trigger a teleological explanation if it is deeply out of keeping with the form of discrimination. If a lion did does something peculiar like sing, we could not “explain” the lion’s singing by reference to the nature and telos of a lion. Likewise, even if race played a causal role in a firing, if this was quite an exceptional fluke, then the firing might not be best explained by the telos of a racist boss so much as by the fluke. The foregoing, which is my interpretation of Nussbaum on Aristotle, would direct the statutory prohibition most focally on durable and general structures of discrimination, which plausibly reflects the law’s goal of changing and penalizing enduring oppressive structures.
• 93See Blum, 34 U Chi L Rev at 502 (cited in note 38) (arguing that an objective standard of “related to” comes to dominate much of motives analysis).
• 94It is easy to mix these things up. See Schwartz, 150 U Pa L Rev at 1710 (2002) (cited in note 87) (“[C]ausation in discrimination cases asks whether the harm to the plaintiff was discriminatory in nature.”). This is far from obvious. Whether something was “X in nature” is a different question from whether “X caused Y.”
• 95For example, Title VIII, the federal Fair Housing Act of 1968 § 800, Pub L No 90-284, 82 Stat 81, uses three different causal language formulations: “because of” in 42 USC §§ 3604(a), (b), (d), 3605, 3631(a); “based on” in 42 USC § 3604(c); and “on account of” in 42 USC §§ 3606, 3617, 3631(b). The ADA Amendments Act (ADAAA) § 5, Pub L No 110-325, 122 Stat 3553 (2008), codified at 42 USC § 12101 et seq, amended the ADA to prohibit discrimination “on the basis of” disability rather than “because of” disability.
• 9642 USC § 1982.
• 9748 Stat 881, codified at 15 USC § 78a et seq.
• 9815 USC § 78j. Manipulation is also prohibited under § 9, which gives little more instruction. 15 USC § 78i.
• 99US Const Amend I (“Congress shall make no law . . . abridging the freedom of speech.”); US Const Amend XIV (“No State shall . . . deny any person within its jurisdiction the equal protection of the laws.”).
• 100Matter of Tally v Newberry Co, 30 AD2d 898, 898 (NY App 1968).
• 101Id at 899. See also Sturgeon Electric Co v Industrial Commission of Arizona, 2015 WL 871813, *4 (Ariz App).
• 102Tally, 30 AD2d at 899.
• 103It is also possible that ambiguous language such as “because of” should be read capaciously to allow both causal and noncausal claims. For example, numerous scholars have urged a causation-based approach to discrimination law in order to help plaintiffs overcome problems of proving a defendant’s bad motive. See, for example, Linda Hamilton Krieger, The Content of Our Categories: A Cognitive Bias Approach to Discrimination and Equal Employment Opportunity, 47 Stan L Rev 1161, 1242–43 (1995). Yet a pure causation-based approach would disadvantage plaintiffs who can show motive but cannot easily show causation. Those sympathetic to plaintiffs may therefore be sympathetic to reading “because of” inclusively in order to capture appropriate links, be they causal or noncausal.
• 104Restatement (Third) of Torts: Liability for Physical and Emotional Harm § 26 (2010) (stating that but-for causes are causes in fact).
• 105See Restatement (Third) of Torts: Liability for Physical and Emotional Harm § 26 (2010) (necessary cause is a cause). See also id at § 27 (multiple sufficient cause is a factual cause); Restatement of Torts § 432(2) (1934) (“If two forces are actively operating . . . and each of itself is sufficient . . . the actor’s negligence may be held by the jury to be a substantial factor.”). See also Katz, 94 Georgetown L J at 494 (cited in note 15) (proposing the “necessity-or-sufficiency” test for motive in employment discrimination).
• 106Restatement (Third) of Torts: Liability for Physical and Emotional Harm § 27 (2010). The first and second Restatements used the term “substantial factor” to cover these overdetermined cases. See, for example, Restatement of Torts § 432(2) (1934). See also Barbara A. Spellman and Alexandra Kincannon, The Relation between Counterfactual (“But For”) and Causal Reasoning: Experimental Findings and Implications for Jurors’ Decisions, 64 Law & Contemp Probs 241, 254 (2001) (finding that experimental subjects agreed that multiple sufficient causes were each a cause).
• 107Or perhaps this analogy is closer if the defendant is in fact the source of both innocent and culpable causal channels—a negligent match dropping and a nonnegligent one.
• 108See Katz, 94 Georgetown L J at 494 (cited in note 15) (advocating for what amounts to a Causal-Motive test). I have called this the “Causal-Motive” standard in previous work precisely because it jives with a conservative reading of black letter tort causation. See Verstein, 127 Yale L J at 1154–59 (cited in note 24).
• 109See Restatement (Third) of Torts: Liability for Physical and Emotional Harm § 26, cmt j (2010) (eliminating discretion for factfinder to find no factual causation on grounds that putative cause was not sufficiently substantial); id at § 27, cmt b (same for multiple causes). The Restatement preserves the possibility that an actor should be exempt from liability when its contribution was “only a trivial contribution” but that exemption operates not by way of causation analysis. The actor (and its conduct) remains a cause. The exemption is effected by way of the scope-of-liability analysis. Restatement (Third) of Torts: Liability for Physical and Emotional Harm § 36 (2010).
• 110Richard W. Wright, Causation in Tort Law, 73 Cal L Rev 1735, 1788–1803 (1985).
• 111I have used fires as my example to track the Restatement. But the Restatement also considers cases that may seem more intuitively divisible than a blaze. Consider a case of poisoning. Victim’s dinner has been dosed with slow-acting poison. Her three food items each contains 0.5 grams of Poison A. Her wine contains 0.5 grams of Poison B. Anyone who receives a full gram of poison dies, and Victim dies. Did Poison B cause Victim’s death? Arguably, the answer is no: there was Poison A enough for the job, which is to say that there was a causally sufficient set that excludes Poison B and regards it as superfluous. Then again, it is just as true to say that Poison B paired with a dose of Poison A to kill Victim, rendering two doses of Poison A irrelevant. There is a causally sufficient set to which Poison B is essential. There is no principled way to say which of these two descriptions is the “true” causal description, so Poison B must be regarded as a cause of the death. Analogously, Motive-B must be regarded as a cause of the motivated action.
• 112Motives are complex, but it should be intuitive that they are at least partially subdividable. An employer’s A-Motive likely grows by some amount when the employee is absent from work or tenders poor service. The total A-Motive is in some sense the product of discrete experiences. When a mixed-motives employer has ample A-Motive, we might also say that some portion of those negative experiences generated a level of justified hostility that combined with a low level of B-Motive to cause the action and that the additional justified hostility was causally superfluous.
• 113This observation comports with ordinary usage. A recent paper used surveys to determine what Americans thought “causation” entailed. See generally James A. Macleod, Ordinary Causation: A Study in Experimental Statutory Experimentation, 94 Ind L J *4 (forthcoming 2019) (on file with author):

  [S]ome alternative causation standards [such as the Causal-Motive and Any-Motive standards] rejected by the Court come much closer to tracking ordinary usage (and, independently, assessments of moral blameworthiness) than does the Court’s “but-for” test. . . . Specifically, the “substantial factor” standard [akin to the Causal-Motive standard] for causation comes much closer to tracking common sense and statutory causality attribution than does the “but-for” test.

  While there may be limits to the utility of such data, it does go to confirm that the court’s preference for but-for causation is not justified by ordinary usage.

• 114490 US 228 (1989).
• 115Id at 242–45 (Brennan) (plurality); id at 260 (White concurring).
• 116Id at 281 (Kennedy dissenting); id at 262 (O’Connor concurring).
• 117For example, Professor Deborah DeMott has worked to avoid use of But-For causation in fiduciary litigation. Deborah A. DeMott, Causation in the Fiduciary Realm, 91 BU L Rev 851, 865–66 (2011) (explaining that a fraud victim’s reliance establishes causation “although ‘his reliance on the truth of the fraudulent misrepresentation [was not] the sole or even the predominant or decisive factor in influencing his conduct.’”), citing Restatement (Second) of Torts § 546, cmt b (1977).
• 118This is a specific legal example of a general problem arising out of moral philosophy’s “double effect” literature. See generally Judith Jarvis Thomson, Physician-Assisted Suicide: Two Moral Arguments, 109 Ethics 497 (1999). See also T.M. Scanlon, Moral
  Dimensions: Permissibility, Meaning, and Blame 19, 20–21 (Belknap 2008). Professor Micah Schwartzman calls the blocked action problem the “permissibility objection,” which he then criticizes with respect to the travel ban discussed in this Section. See generally Micah Schwartzman, Official Intentions and Political Legitimacy: The Case of the Travel Ban (University of Virginia School of Law Public Law and Legal Theory Research Paper No 2018-22, Apr 2018).
• 119See Palmer v Thompson, 403 US 217 (1971). Certain facts about the case make it hard to take Palmer’s blocked action problem at face value. Id at 229 (Blackmun concurring) (noting that “[n]o other municipal recreational facility in the city of Jackson has been discontinued”).
• 120403 US 217 (1971).
• 121Id at 227; id at 228 (Burger concurring); id at 230 (Blackmun concurring) (finding petitioner’s response “disturbing”). Similar concerns were important in Mt. Healthy, 429 US at 286 (“A borderline or marginal candidate . . . ought not to be able . . . to prevent his employer from assessing his performance record and reaching a decision not to rehire on the basis of that record.”) (emphases added).
• 122Palmer, 403 US at 227 (“[C]ities could be forced . . . to construct or refurbish swimming pools which they choose not to operate for any reason, sound or unsound.”). Justice Harry Blackmun likewise feared that the city would “be ‘locked in’ with its pools for an indefinite time in the future, despite financial loss of whatever amount, just because at one time the pools of Jackson had been segregated.” Id at 230 (Blackmun concurring). Counsel for the plaintiff agreed with the lock-in argument. Id, citing Transcript of Oral Argument at 43–44 (“A. If the question is, are they locked in forever because of racial problems which cause a rise in economic difficulties in operating the pool, my answer is that they would be locked in.”).

  Justice Burger took a forward-looking perspective, anticipating how other cities may respond: “[Any holding that] constitutionally ‘locks in’ the public sponsor so that [the service or facility] may not be dropped . . . would plainly discourage the expansion and enlargement of needed services in the long run.” Id at 228 (Burger concurring) (citations omitted).

• 123Protecting the Nation from Foreign Terrorist Entry into the United States, 82 Fed Reg 8977 (2017) (Order 1). The executive orders were issued pursuant to 8 USC § 1182(f).
• 124Order 1 at § 3(c).
• 125Id at § 5(a)–(d). In the face of litigation, the president later enacted a second order, placing similar restrictions on immigration but with greater discussion of the national security objectives of the policy. Executive Order 13780, Protecting the Nation from Foreign Terrorist Entry into the United States, 82 Fed Reg 13209 (2017) (Order 2).
• 126International Refugee Assistance Project v Trump, 857 F3d 554, 572 (4th Cir 2017). The orders were presaged by numerous statements—by Trump and others—reflecting a desire to target Muslims for restricted entry. Id at 594–95.
• 127The orders purported to justify the ban by citing problems with the past and present visa regime, Order 1 at § 1, 82 Fed Reg at 8977, a need for immediate review of our immigration policies, Order 1 at § 3(a), (b), (h), 82 Fed Reg at 8978, and the prevalence of “hostile attitudes” and “violent ideologies” in the subject countries, Order 1 at § 1. See also Order 2 at § 1 (d), 82 Fed Reg at 13210 (“Each of these countries is a state sponsor of terrorism, has been significantly compromised by terrorist organizations, or contains active conflict zones.”).
• 128Washington v Trump, 2017 WL 462040, *2 (WD Wash), affd, 847 F3d 1151, 1167 (9th Cir 2017) (per curiam) (regarding Order 1); International Refugee Assistance Project v Trump, 241 F Supp 3d 539, affd in part, revd in part, 857 F3d 554 (4th Cir 2017), vacd and remd, 138 S Ct 353 (2017) (regarding Order 2). The Supreme Court recently affirmed the newest version of the travel ban. Trump v Hawaii, 138 S Ct 2392 (2018).
• 129The real travel bans elicited doubts about the sincerity of purported national security rationales. At the District Court of Maryland, plaintiffs cited a Homeland Security report and an affidavit by former senior security officials arguing that the executive order served no security function and would even undermine security. International
  Refugee Assistance Project, 241 F Supp 3d at 548, 561–62. Likewise, continued statements by Trump seemed to undermine his own security claims. See, for example, @realDonaldTrump (Twitter, June 5, 2013), archived at http://perma.cc/R58L-STMN; Matt Zapotosky, Trump’s Latest Tweets Will Probably Hurt Effort to Restore Travel Ban (Wash Post, June 5, 2017), archived at http://perma.cc/B3PF-JVLY. Still, there are those who would defend the orders on national security grounds. See, for example, Hans A. von Spakovsky, Trump Travel Ban Decision: Politics, Not the Law, in the 4th Circuit (Fox News Opinion, May 28, 2017), archived at http://perma.cc/Z2CH-LV68.
• 130Eric Posner, The Presidency Shrinks (EricPosner.com, Feb 9, 2017), archived at http://perma.cc/TMV8-CX3D. For a law professor with an even stronger stance on the matter, see Eugene Kontorovich, Opinion: The 9th Circuit’s Dangerous and Unprecedented Use of Campaign Statements to Block Presidential Policy (Wash Post, Feb 9, 2017), archived at http://perma.cc/3EEE-7WSV:

  The states’ argument is in essence that Trump is a bigot, and thus his winning presidential campaign in fact impeaches him from exercising key constitutional and statutory powers, such as administering the immigration laws. This would mean that Trump is automatically disbarred, from the moment of his inauguration, of exercising certain presidential powers.

• 131But see Aziz v Trump, 234 F Supp 3d 724, 737 (ED Va 2017) (denying that the court’s ruling would “render every policy that the president makes related to Muslim-majority countries open to challenge”). Indeed, the Trump v Hawaii dissenters’ opinions do not just assess the evidence of animus differently than the majority: they use different motive rules. Justice Sonia Sotomayor seems to apply a Primary-Motive standard. Trump v Hawaii, 138 S Ct at 2438 (Sotomayor dissenting):

  Rather, the dispositive and narrow question here is whether a reasonable observer, presented with all “openly available data,” the text and “historical context” of the Proclamation, and the “specific sequence of events” leading to it, would conclude that the primary purpose of the Proclamation is to disfavor
  Islam and its adherents by excluding them from the country.

  Justice Breyer presumes an Any-Motive standard. Id at 2429 (Breyer dissenting):

  If its promulgation or content was significantly affected by religious animus against Muslims, it would violate the relevant statute or the First Amendment itself. . . . If, however, its sole ratio decidendi was one of national security, then it would be unlikely to violate either the statute or the Constitution. Which is it?

  The applicable motive standard is largely dispositive of the result.

• 132See, for example, Vikram David Amar and Alan E. Brownstein, The Complexities of a “Motive” Analysis in Challenging President Trump’s Executive Order regarding Entry to the United States (Justia, Mar 24, 2017), archived at http://perma.cc/2BRP-3CQL (“If any established impermissible intent ended up not being a ‘but for’ cause of the executive order, then it should not be a basis of invalidation.”).
• 133Richard H. Fallon Jr, Constitutionally Forbidden Legislative Intent, 130 Harv L Rev 523, 531 (2016).
• 134By contrast, Professor Brandon Garrett accepts that “government actors may be forbidden from carrying out their policies, and for quite some time” in order to restore the state’s legitimacy. Brandon L. Garrett, Unconstitutionally Illegitimate Discrimination, 105 Va L Rev *5 (forthcoming 2019), archived at http://perma.cc/DR9D-59WU.
• 135One familiar variant on this approach is to set penalty levels as “liability rules,” which compensate the victim, as opposed to the more stringent “property rules,” which are protected with penalties too costly for defendants to rationally accept. See generally Guido Calabresi and A. Douglas Melamed, Property Rules, Liability Rules, and Inalienability: One View of the Cathedral, 85 Harv L Rev 1089 (1972). See also Richard A. Posner, Economic Analysis of Law 57 (Little, Brown 1972); Robert L. Birmingham, Breach of Contract, Damage Measures, and Economic Efficiency, 24 Rutgers L Rev 273, 292 (1970) (introducing the notion that the law should encourage efficient breach).
• 136For example, Garrett would block badly motivated government acts in order to safeguard the state’s legitimacy. See Garrett, 105 Va L Rev at *28–39 (cited in note 135) Perhaps the state can restore its legitimacy merely by penalizing the acts of some defendants without outright preventing their conduct.
• 137I do not wish to be glib about this suggestion. There are reasons to resist a regime in which government actors may violate the Constitution whenever they are willing to spend the taxpayers’ money afterward. But this tentative suggestion does not presume that it is actually okay to violate core rights in exchange for a token payment; it is instead an effort to imagine new imperfect solutions in tough cases when there may be no perfect solutions.
• 138See International Refugee Assistance Project v Trump, 883 F3d 233, 268 (4th Cir 2018) (“President Trump could have removed the taint of his prior troubling statements; for a start he could have ceased publicly disparaging Muslims.”) See also Schwartzman, Official Intentions at *22–23 (cited in note 119). I do not wish to overstate this argument. The actual travel ban was affirmed, though officials made only limited efforts to hide or undo evidence of B-Motives. Trump v Hawaii, 138 S Ct at 2439 (Sotomayor dissenting) (noting absence of disavowal).
• 139See, for example, In re GSC, Inc, 453 Bankr 132, 173 (SDNY 2011).
• 140See, for example, In re MFW Shareholders Litigation, 67 A3d 496, 500 (Del Chanc 2013).
• 141See, for example, Wyoming v Oklahoma, 502 US 437, 454 (1992) (“This ‘negative’ aspect of the Commerce Clause prohibits economic protectionism—that is, regulatory measures designed to benefit in-state economic interests by burdening out-of-state competitors.”), quoting New Energy Co of Indiana v Limbach, 486 US 269, 273–74 (1988). See also City of Philadelphia v New Jersey, 437 US 617, 628 (1978) (condemning measures enacted “to slow or freeze the flow of commerce for protectionist reasons”).
• 142See, for example, Planned Parenthood v Casey, 505 US 833, 877 (1992) (plurality).
• 143Church of the Lukumi Babalu Aye, Inc v City of Hialeah, 508 US 520, 533 (1993).
• 144See, for example, Wallace v Jaffree, 472 US 38, 43, 56 (1985); Lemon v Kurtzman, 403 US 602, 612 (1971).
• 145See, for example, Washington v Davis, 426 US 229, 240 (1976).
• 146See, for example, Miller v Johnson, 515 US 900, 920 (1995) (race-based redistricting).
• 147See, for example, Church of the Lukumi Babalu Aye, 508 US at 533. It is often said that “‘strict’ in theory [is] fatal in fact.” Gunther, 86 Harv L Rev at 8 (cited in note 37). But see Winkler, 59 Vand L Rev at 796 (cited in note 37) (“30 percent of all applications of strict scrutiny [in federal courts] . . . result in the challenged law being upheld.”).
• 148Fallon, 130 Harv L Rev at 579 (cited in note 134). See also Richard L. Hasen, Bad Legislative Intent, 2006 Wis L Rev 843, 850.
• 149Washington v Trump, 847 F3d at 1168; International Refugee Assistance Project, 241 F Supp 3d at 562–63, affd, 857 F3d 554.
• 150See Caleb Nelson, Judicial Review of Legislative Purpose, 83 NYU L Rev 1784, 1809–10 (2008) (noting that courts have been more willing to scrutinize city counsel qua administrator than qua legislature).
• 151US Const Art I, § 6, cl 1 (providing that “for any Speech or Debate in either House, [Senators or Representatives] shall not be questioned in any other Place”). This clause is to protect open deliberation and preserve the legislative branch’s independence. See United States v Brewster, 408 US 501, 525 (1972); Tribe, 1993 Sup Ct Rev at 16 n 34 (cited in note 14) (noting the contribution of the Speech and Debate Clause). See also Brest, 1971 S Ct Rev at 128–29 (cited in note 14) (discussing intrusiveness concerns in Palmer); Village of Arlington Heights v Metropolitan Housing Development Corp, 429 US 252, 268 (1977) (“In some extraordinary instances the members [of Congress] might be called to the stand at trial to testify concerning the purpose of the official action, although even then such testimony frequently will be barred by privilege.”).
• 152See Price Waterhouse, 490 US at 242 (Brennan) (plurality) (stressing that one of Congress’s key concerns in passing Title VII was the “preservation of an employer’s remaining freedom of choice”). Congress sought to retain employment at will in its construction of the Title VII discrimination regime. Id at 244–45. See also id at 239 (“Title VII eliminates certain bases for distinguishing among employees while otherwise preserving employers’ freedom of choice. This balance between employee rights and employer prerogatives turns out to be decisive.”). In forty-nine states, an employer is free to fire an employee for any reason or no reason—other than those reasons specifically barred by antidiscrimination law. The drafters of Title VII considered and rejected language that seemed to impinge on employment at will. See id at 239 n 4. But see Cardi, 75 Ohio St L J at 1145–50 (cited in note 76) (criticizing deference to employment at will).
• 153But see Price Waterhouse, 490 US at 262 (O’Connor concurring) (rejecting a “mere discriminatory thoughts” standard).
• 154See generally Gabriel S. Mendlow, Thoughts, Crimes, and Thought Crimes, 118 Mich L Rev (forthcoming 2020) (on file with author); Gabriel S. Mendlow, Why Is It Wrong to Punish Thought?, 127 Yale L J 2342 (2018).
• 155A Sole-Motive test never punishes a defendant for increasing quantities of A-Motive because a safe harbor covers all mixed-motives actions.
• 156See Price Waterhouse, 490 US at 249 (Brennan) (plurality). See also Geller v Markham, 481 F Supp 835, 841 (D Conn 1979):

  If relief were to be afforded every time age was considered, the effects would go well beyond the remedial designs of the drafters of the ADEA. Back pay would be awarded to those who never had a chance for the job. . . . While doing so would provide a strong deterrent against age discrimination, it would be a deterrent far in excess of the limited deterrent Congress intended.

  See also Texas v Lesage, 528 US 18, 21 (1999) (per curiam) (“Simply put, where a plaintiff challenges a discrete governmental decision as being based on an impermissible criterion and it is undisputed that the government would have made the same decision regardless, there is no cognizable injury warranting relief under § 1983.”).

• 157See Mt. Healthy, 429 US at 285.
• 158Id. The compensation argument can also be viewed dynamically as providing opportunities for strategic action. An employee “ought not to be able, by engaging in such conduct, to prevent his employer from assessing his performance record and reaching a decision not to rehire on the basis of that record.” Id at 286.
• 159See id at 287 (adopting a “same decision” test—that is, a “But-For” test).
• 160See Board of Education, Island Trees Union Free School District No 26 v Pico, 457 US 853, 871–72 (1982).
• 161Batson challenges to discriminatory jury strikes use a But-For standard, though it is strange to say that these challenges are about compensating the movant rather than vindicating the movant’s right to an impartial jury.

  Whether a payment is taxed as income depends on the motive of the payer. If the payer had a subjective charitable motive, it is not taxable. If a recipient does not report income, and the giver turned out to have a noncharitable (that is, business) motive, would we say that the IRS was harmed? It seems more natural to say that the taxpayer did not pay enough, without reference to harm or compensation.

• 162Note, Refusal to Rehire a Nontenure Teacher for a Constitutionally Impermissible Reason, 1970 Wis L Rev 162, 169, discussing Jones v Hopper, 410 F2d 1323 (10th Cir 1969) (emphasis added).
• 163Brief for Petitioner, Mt. Healthy School District Board of Education v Doyle, No 75-1278, *15 (US filed June 3, 1976) (Mt. Healthy Petitioner’s Brief).
• 164Note, 1970 Wis L Rev at 169 (cited in note 163).
• 165Mt. Healthy Petitioner’s Brief at *15 (cited in note 164).
• 166See Katz, 94 Georgetown L J at 524 (cited in note 15) (arguing that we should not regard such cases as involving a windfall). See also Part IV.A.2; Note, 1970 Wis L Rev at 169 (cited in note 163).
• 167See Shaw v Reno, 509 US 630, 647 (1993) (“[R]eapportionment is one area in which appearances do matter.”).
• 168See Lazar v Merchants’ National Properties, Inc, 45 Misc 2d 235, 518 (NY Sup 1964) (describing how “[t]he purpose, maliciously, to inflict injury” can “transform a lawful act into an actionable wrong”) (emphasis added), affd, 23 AD2d 630 (NY App 1965). See also generally F.B. Ames, How Far an Act May Be a Tort Because of the Wrongful Motive of the Actor, 18 Harv L Rev 411 (1905).
• 169See generally Thomas Edward Powell II, Comment, The Truth Will Not Set You Free in Nebraska: Actual Malice and Nebraska’s “Truth plus Motive” Defense, 72 Neb L Rev 1236 (1993). See also New York Times Co v Sullivan, 376 US 254, 283 (1964) (requiring actual malice to overcome constitutional privilege in stories about public figures). Note that the New York Times v Sullivan “actual malice” standard for First Amendment protection has been interpreted to only indirectly incorporate motives analysis insofar as spiteful motives may indicate actual malice. See Harte-Hanks Communications, Inc v Connaughton, 491 US 657, 667–68 (1989). We could count it as a weak proxy usage. See Part II.A.2–3. Related to “actual malice” is “express malice,” which is motive as constituent. Express malice is necessary to overcome qualified privilege in common law defamation. See Nodar v Galbreath, 462 So 2d 803, 806 (Fla 1984).
• 170See Price Waterhouse, 490 US at 265 (O’Connor concurring).
• 171Id (“This Court’s decisions under the Equal Protection Clause have long recognized that whatever the final outcome of a decisional process, the inclusion of race or sex as a consideration within it harms both society and the individual.”).
• 172See Christine M. Korsgaard, Acting for a Reason, in Jonathan Dancy and Constantine Sandis, eds, Philosophy of Action: An Anthology 207, 210 (Wiley-Blackwell 2015). For Kant, every action is composed of an act and a motive, and so it is impossible to evaluate any action without clarifying the motive. But see Marcia W. Baron, Kantian Ethics Almost without Apology 152 (Cornell 1995) (applying a Primary-Motive standard to Kantian analysis of mixed motives).
• 173Even when no harms are alleged, ambiguous facts are patterned through motive. Boris I. Bittker, Income Tax Deductions, Credits, and Subsidies for Personal Expenditures, 16 J L & Econ 193, 203–04 (1973).
• 174See generally Charles R. Lawrence III, The Id, the Ego, and Equal Protection: Reckoning with Unconscious Racism, 39 Stan L Rev 317 (1987) (describing the symbolic content and context of discrimination). See also generally Michael Kent Curtis, Race as a Tool in the Struggle for Political Mastery: North Carolina’s “Redemption” Revisited 1870–1905 and 2011–2013, 33 L & Inequality 53 (2015) (describing repeating patterns of race in oppression).
• 175See Oliver Wendell Holmes Jr, The Common Law 3 (Little, Brown 1881) (“[E]ven a dog distinguishes between being stumbled over and being kicked.”). The dignitary harm is greater in a kick (or, if you will, a battery) than a stumble. The pairing of the physical act with the mental state arguably enacts the kicker’s exemption from the moral community and denies the victim’s bodily autonomy.
• 176Tort compensation for dignitary harms is unusual, but it does appear that this is a realm in which Congress envisioned it and courts have supplied it. For example, Supreme Court precedent does not allow employers to avoid liability altogether by showing that the employee would have been fired anyway if the evidence supporting termination was acquired only after termination. McKennon v Nashville Banner Publishing Co, 513 US 352, 360 (1995). The liability that remains, even after showing that the employee would have suffered the same result anyway, may recognize some lingering dignitary harm. This conclusion is supported by the difference in treatment between discrimination suits and other employment suits lacking the dignitary patina with respect to after-acquired evidence. Most other actions allow robust use of such evidence, in many cases to bar all remedies. See generally Stephen J. Humes, After Acquired Evidence of Employee’s Misconduct as Barring or Limiting Recovery in Action for Wrongful Discharge, 34 ALR5th 699 (1995) (compiling case law on after acquired evidence). Other areas of the law are likewise resistant to recovery for abstract, unconventional harms. See, for example, Memphis Community School District v Stachura, 477 US 299, 310 (1986) (holding that abstract “value” or “importance” of constitutional rights are not a permissible element of compensatory damages in an action under the Civil Rights Act of 1871).
• 177The employer alone has access to its own mental states. And employers may be more systematic in documenting their A-Motives than B-Motives. Human resources departments can log employee problems with great specificity but are unlikely to canvass for B-Motives. Ex post overcompensation may yet be appropriate ex ante compensation when those challenges are considered.
• 178See Mark C. Weber, Beyond Price Waterhouse v. Hopkins: A New Approach to Mixed Motive Discrimination, 68 NC L Rev 495, 532 (1990).
• 179See Richard H. Fallon Jr and Daniel J. Meltzer, New Law, Non-retroactivity, and Constitutional Remedies, 104 Harv L Rev 1731, 1736 (1991); Wells, 51 Mercer L Rev at 588–89 (cited in note 66).
• 180See Weber, 68 NC L Rev at 531 (cited in note 179) (“Imposing remedies even when other causes would have had the same result increases deterrence by raising the likelihood of sanction.”).
• 181See Katz, 94 Georgetown L J at 542 (cited in note 15).
• 182See, for example, Bibbs v Block, 778 F2d 1318, 1327 (8th Cir 1985) (Lay concurring) (“[T]he employer should not be able to exculpate its proven invidious discriminatory practices by having a second chance to show that racial considerations did not affect the decision’s outcome.”).
• 183United States v Mulheren, 938 F2d 364, 367 (2d Cir 1991).
• 184Nor can rules adjusting remedies reduce windfalls. For example, if we deem a defendant only 50 percent responsible when both she and another individual independently contribute to a fire, we can discount the defendant’s liability by half. Relative to full recovery, the plaintiff’s windfall is reduced. But relative to full recovery, the defendant’s windfall is increased.
• 185Similar comparisons could be applied in other domains. Consider the mixed-motives market manipulator. We might consider it better to convict an inveterate market sharper despite her relatively minor investment motive while exonerating a mostly scrupulous investor who succumbs in part to a manipulative motive for action.
• 186See Gudel, 70 Tex L Rev at 108 (cited in note 50).
• 187This is roughly analogous to tort law’s treatment of comparative negligence. See note 189. This approach might also improve incentives: Fear of paying a windfall gives an incentive to reduce bigoted attitudes in the employer. Fear of losing the windfall gives the employee a reason to be utterly professional, to reduce the legitimate motive for termination. Both actors gain an incentive to improve her conduct because neither has a safe harbor.
• 188The majority rule is some form of modified comparative negligence. Many states implement this rule by also reducing recovery for partially responsible plaintiffs. See Restatement (Third) of Torts: Apportionment of Liability § 7 (2000) (providing that, when plaintiff is partially responsible for injury, damages can be reduced). Such a modification of recovery is also possible under the Primary-Motive rule.

  The approach preferred by academics is to allow the plaintiff to recover regardless of its level of relative responsibility but to simply reduce recovery accordingly. This is the Pure Comparative Fault regime. Such an approach could be adopted in employment too—allowing recovery under Any-Motive but then reducing it according to some assessment of responsibility, including the presence of A-Motive.

• 189Professor Mark Weber offers a similar suggestion by incorporating disproportionate hardship. Weber, 68 NC L Rev at 521 (cited in note 179). If an employer would be forced to rehire an employee despite her utter lack of qualifications, then the hardship of a reinstatement remedy is disproportionate.
• 190Note, 1970 Wis L Rev at 169 (cited in note 163), discussing Jones, 410 F2d 1323.
• 191Mt. Healthy Petitioner’s Brief at *15 (cited in note 164).
• 192Note, 1970 Wis L Rev at 169 (cited in note 163), discussing Jones, 410 F2d 1323.
• 193A Primary-Motive test finds for the defendant when the good motive was primary but the bad motive was nevertheless necessary to motivate the action. It therefore acquits in some cases in which the plaintiff’s protected status was a but-for factor in her adverse treatment. That may be an unacceptable result in some areas of the law. If so, other options can be considered. In other work, I will discuss at greater length a compound rule, which I call the Pivotal-Motive rule, that finds for the plaintiff whenever bad motive was primary or necessary. Thus, under the Pivotal-Motive rule, the plaintiff prevails should she satisfy either a But-For standard or a Primary-Motive standard.
• 194On the proper degree of conceptual separation, see generally Lee Anne Fennell, Accidents and Aggregates, 59 Wm & Mary L Rev 2371 (2018). See also generally Lee Anne Fennell, Slices and Lumps (John M. Olin Law & Economics Working Paper No 395 2d Series, Mar 2008), archived at http://perma.cc/C2AM-T35B.
• 195This example is based on Securities and Exchange Commission v Kwak, 2008 WL 2705417 (D Conn). Two brokers were accused of manipulating the price of securities bought for themselves and their clients.
• 196See, for example, Joe Nocera, Chasing Small Fry, S.E.C. Let Madoff Get Away (NY Times, June 26, 2009), archived at http://perma.cc/9T54-U9ST (describing facts of Kwak, including how the defendants viewed the underlying company as “the road to riches” and never sold its stock, even at its highest price).
• 197See, for example, Defendant Stephen J. Wilson’s Reply in Support of Renewed Motion for Judgment as a Matter of Law, Securities and Exchange Commission v Competitive Technologies, 2008 WL 62818, *9–10 (D Conn filed Jan 17, 2008) (Wilson Brief) (making this argument and citing Securities and Exchange Commission v Masri, 523 F Supp 2d 361, 372–73 (SDNY 2007)).
• 198Kwak, 2008 WL 410427 at *4 n 10 (emphasis altered). See Andrew Verstein, Benchmark Manipulation, 56 BC L Rev 215, 263–64 (2015) (advocating this approach).
• 199Kwak, 2008 WL 410427 at *4 n 10.
• 200Redistricting cases are probably not using But-For standards. See generally Verstein, 127 Yale L J 1106 (cited in note 24). But the jurisprudence is not fully settled and certain strands of the But-For standard often emerge. One fixture of redistricting law has been the shifting role of alternative maps. In Easley v Cromartie, 532 US 234, 249 (2001), the Court seemed to endorse a requirement that plaintiffs must present an alternative district map in order to vindicate their claims against the challenged map. If no alternative map can be produced, then it may suggest that the legislature could not have done any better by fixing its motives.

  The more recent Cooper v Harris case rejected that requirement, clarifying that such a map would be helpful but is not the only path to success. Cooper, 137 S Ct 1455, 1472–74 (2017). The dissent clung tightly to the map requirement. Id at 1486–87 (Alito dissenting). One way of explaining the passion about this map issue is that it is a fault line for the motive test. On a Primary-Motive test, absence or presence of an alternative map is only one datum among many that may indicate the relative motives of the legislature; if it was easy to produce an alternative map, it will lend credence to the idea that some bad motive may have led it away from that map. However, the absence of such a map would not defeat a primary motive case because a district would be invalid (even if best in every other way) if offered for the wrong motives. Conversely, on a But-For standard, the possibility of alternative action is rather important. If the legislature’s only legal option was the district as formed, it would defeat the claim that bad motives had made any difference. The clash over maps is therefore a sub rosa clash over the motives standard.

• 201515 US 900 (1995).
• 202Id at 918. Richard Briffault, Race and Representation after Miller v. Johnson, 1995 U Chi Legal F 23, 51, takes Miller to be a “but-for” test.
• 203Johnson v Miller, 864 F Supp 1354, 1377 (SD Ga 1994) (emphasis added), affd, 515 US 900 (1995).
• 204135 S Ct 1257, 1265–66 (2015).
• 205Alabama Legislative Black Caucus v Alabama, 989 F Supp 2d 1227, 1287, 1292, 1293 (MD Ala 2013). “Equal opportunity harasser” cases serve as another example of how high granularity can convert a But-For standard into an Any-Motive standard. Suppose a boss tends to verbally abuse any employee in her clutches, but she customizes the abuse to match features of each victim. In one sense, the victim’s sex (or race, etc.) made no difference because the employer was sure to misbehave regardless. On the other hand, the particular form of the misconduct would certainly have been different but for the plaintiff’s protected status.
• 206See generally Ariel Porat and Eric A. Posner, Aggregation and Law, 122 Yale L J 2 (2012). See also Fennell, 59 Wm & Mary L Rev 2371 (cited in note 195).
• 207Numerous luminaries of torts have sought a similar strategy to avoid problems of overdetermination there. For example, while two fires might each have galloped toward the plaintiff’s house, the defendant’s fire still made a difference. The building wouldn’t have been destroyed “by two fires” but for the defendant’s fire; plus the defendant’s fire might have burned it a little earlier (or later) than the natural fire would have caused. While this solution may sometimes be satisfactory, it threatens to consume as much as the fire it would contain. As Professor Wright points out, every feature of the universe was a but-for cause on some sufficiently precise description. That America is spelled with a single “R” is a but-for cause of “Fires destroyed your house and America is spelled with a single ‘R.’” Efforts to granularly define actions raise particularism problems that must be addressed, but results-oriented courts may not satisfactorily do so. See generally Wright, 73 Cal L Rev 1735 (cited in note 111).
• 208See Verstein, 127 Yale L J at 1125–27 (cited in note 24).
• 209At present, different types of discrimination get to the But-For standard through different initial allocations of proof. It is beyond the scope of this Article to evaluate the proper use of burden shifting, though that is clearly very important to a final theory of motive in the law.
• 210Whether a court would actually exclude the evidence is another question. On the one hand, evidence is admissible to the degree it bears on a material fact regardless of whether the fact is in dispute. FRE 401, Notes of the Advisory Committee ¶ 7 (“The fact to which the evidence is directed need not be in dispute.”). On the other hand, inflammatory information that goes to prove only what is already conceded might be excludable as unfairly prejudicial. FRE 403. The silencing tendency of the But-For standard might also be expressed elsewhere, such as in courts’ widespread tendency to exclude evidence of “stray remarks” of discrimination. See Jessica A. Clarke, Explicit Bias, 113 Nw U L Rev 505, 524 (2019).
• 211See, for example, Price Waterhouse, 490 US at 265 (O’Connor concurring) (“There is no doubt that Congress considered reliance on gender or race in making employment decisions an evil in itself.”).
• 212See Part III.B. The Primary-Purpose test is a true mixed-motives test, perhaps even more so than the But-For test because the latter can be satisfied by looking at A-Motive alone. However, its telos is a single-motive inquiry. To ask which of two motives predominates is similar to asking “which, if you had to choose among these two, was the defendant’s motive.” We would use the same test if our goal was investigation of pretext (“She says she acted for motive A, but that sounds like a fig leaf given how badly she wanted B”). The operation of this inquiry is similar too. Courts weighing motives for predominance try to reconcile various facts with various motives in order to test the competing theories that one motive or the other was the sole real motive. Thus, Primary-Motive analysis can collapse into pretext analysis, which assumes that the defendant has only one motive in reality, which the court attempts to divine. See, for example, McDonnell Douglas Corp v Green, 411 US 792, 804–07 (1973).