There's No Such Thing as Affirmative Duty

I.     Introduction

An individual is driving a car down a city street. Suddenly, a pedestrian steps out in front of it. What must the driver do? Under American tort law, the driver must exercise reasonable care to prevent the impending collision. Imagine, at the same time, that a bystander on the corner sees the pedestrian step into the street. What obligations does the bystander have? Under American tort law, none. The bystander has no duty to act to prevent the impending collision. This disparity is often explained in tort law as the difference between misfeasance and nonfeasance.1

One of the fundamental propositions in tort law is that misfeasance results in much more liability than nonfeasance. The general idea is that doing something—creating some kind of risk—usually generates liability, while not doing something—failing to reduce a risk one did not create or failing to confer a benefit—usually does not.2 The driver who negligently hits a pedestrian commits misfeasance and is subject to liability. A bystander who fails to rescue the pedestrian commits only nonfeasance and is not.3

This distinction is also often framed in terms of duty. In most cases, when acting, individuals have a duty to act with reasonable care to avoid risking physical harm. But to impose such a duty on those who are not acting—who merely fail to reduce a risk or confer a benefit—is to impose an “affirmative duty.”4 Affirmative duties, it is often said, are exceptional.5 The most cited illustration of this fact is that tort law does not impose a general duty to rescue in cases like that of the bystander.6

But these labels do little to draw real distinctions. The driver’s responsibility to brake is characterized as a “duty” to avoid “misfeasance,” though some would frame it instead as an “affirmative duty” not to engage in the “nonfeasance” of failing to hit the brake.7 Conversely, many obligations that the court frames as “affirmative” duties seem indistinguishable from the general “duty” to avoid “misfeasance.” A landlord has an “affirmative duty” to foresee that the level of security in its apartment building might have an effect on its tenants’ risk of being robbed or assaulted.8 But if that duty is affirmative, why is the driver’s not also? The landlord’s duty involves anticipating risk and taking precautions to reduce it, just as the driver’s does. It involves taking these precautions to avoid an unfortunate interplay with other people’s choices, just as the driver’s does. If both of these cases could be framed as affirmative duties and both could be framed as not affirmative duties, then there is confusion about the whole concept of affirmative duty.

We contend that the very idea of an affirmative duty—as actually found and typically explained in tort law, is misguided. Nor is the distinction between duty and affirmative duty a mere matter of mislabeling. Both the term “affirmative duty” and the distinction between misfeasance and nonfeasance fail to draw meaningful distinctions between the cases to which tort law applies these notions. These concepts obscure rather than clarify the issues posed.

In order to move from confusion to clarification at this point in the history and evolution of these notions, two things are required. First, the distinctions between duty and affirmative duty and between misfeasance and nonfeasance, must be deconstructed. The reasons that these distinctions do not hold up to scrutiny must be uncovered. Second, a more sensible approach to thinking about the cases that allegedly involve affirmative duty must be constructed. That reconstruction should recognize what has led the courts and commentators to employ the category of affirmative duty, but it must also develop a better way to analyze the issues the affirmative duty cases pose than the traditional categorization has permitted.

We undertake these tasks in the following manner. In Part II, we offer an overview of the traditional distinction between misfeasance and nonfeasance, the “affirmative duty” rules the distinction generates, and the categories into which these rules fall. In Part III, we deconstruct the distinction between duty and affirmative duty, by calling into question the bases upon which tort law has attempted to maintain the distinction. In Part IV, we explore the reasons, partly conceptual and partly historical, that the distinction came into being and has persisted. Finally, in Part V, we turn to reconstruction. We re-describe what is actually going on in this area of tort law and develop the considerations that should guide decisions about how to handle the kinds of cases that have traditionally been considered to involve affirmative duty.

II.     “Affirmative Duties” in Tort Law

Tort cases—as well as casebooks, treatises, and Restatements—routinely distinguish between garden-variety negligence cases and “affirmative duty” cases.9

Often the term “misfeasance” is attached to the former and “nonfeasance” to the latter.10 These commentaries treat liability for negligent conduct—for misfeasance—as a major component of tort law.11 Affirmative duties, or liability for nonfeasance, are treated as a separate topic, with separate cases that explicate this minor and somewhat curious corner of the law. For example, the Third Restatement devotes Chapter Three to “The Negligence Doctrine and Negligence Liability.”12 Within Chapter Three, Section Seven sets forth a general negligence principle:

(a)    An actor ordinarily has a duty to exercise reasonable care when the actor’s conduct creates a risk of physical harm.13

Meanwhile, Chapter Seven deals separately with “Affirmative Duties.”14 The sections within Chapter Seven deal with the general rule that there is no duty of care with respect to risks not created by an actor,15 as well as the categorical exceptions discussed below. This framing reflects the traditional distinction between misfeasance and nonfeasance.

A.     The General Rule: No Liability for Nonfeasance

The general rule is that there is no liability for nonfeasance and therefore, among other things, no duty to rescue.16 Exceptions to this rule are called “affirmative duties.”17 The general rule treats as a classic “no affirmative duty” case a bystander who fails to rescue a stranger. Whatever moral duties might exist between them, tort law does not impose a duty to rescue on the bystander.18

To be clear, our primary purpose is not to criticize the substance of the general no-duty rule. Much could be, and has been, said about that.19 “Some of the decisions,” Prosser tells us:

[H]ave been shocking in the extreme. The expert swimmer, with a boat and a rope at hand, who sees another drowning before his eyes, is not required to do anything at all about it, but may sit on the dock, smoke his cigarette, and watch the man drown.20

Similar cases occur from time to time and generally provoke outrage.21 Whether this outrage should translate into a change in the legal rule is an important question, but not our question. We want to ask whether the supposedly “exceptional” cases in which liability is imposed for nonfeasance can be distinguished from garden-variety cases of misfeasance. To do that, we must first see what those exceptional cases are.

B.     The Exceptional “Affirmative Duties”

Although the general rule is that there is no liability for nonfeasance, a substantial number of exceptions have developed. These exceptions are characterized as “affirmative duties,” and it is to these exceptions that treatises, casebooks, and the Restatements turn after identifying the general rule.22 The affirmative duties fall into the following general categories:

Special Relationships: Tort law imposes a duty of care when there is a “special relationship” between defendant A and victim B, typically, though not always, involving some form of dependence by the victim on the defendant.23 Examples of special relationships include: common carrier–passenger, innkeeper–guest, employer–employee, and school–student.24

In addition, when the defendant has a special relationship that involves a duty to control the conduct of another, tort law sometimes imposes a duty to protect third parties against risks posed by the person controlled.25 Examples include the jailer’s duty to protect potential victims from dangerous criminals and a guardian’s duty to supervise their wards so as to avoid injuring third parties. Also in this category is the duty of mental health professionals to take reasonable care to protect third parties under particularized threat of harm from their patients.26 This duty was famously recognized in the Tarasoff case, which in effect extended the category to instances in which the therapist had no pre-existing relationship with the third-party.27

Voluntary Undertakings: A party that either promises to provide a service or otherwise undertakes to do so is under a duty to exercise reasonable care toward those who benefit from the service.28

The duty applies when the party knows or should know that failure to perform could harm the promisee or a third party, if (1) failure to exercise reasonable care will leave the victim worse off than in the absence of the undertaking or (2) either the promisee or a third party relies on his taking reasonable care.29 Examples include a friend who agrees to take care of a pet and then fails to do so and a school that voluntarily supplies crossing guards but then omits doing so.30

Prior Conduct Creating Injury or Risk: Tort law imposes a duty to exercise reasonable care to prevent harm that arises from a risk resulting from the defendant’s prior conduct.31 This duty applies even if the risk results from prior nonnegligent conduct.32 For example, if a driver’s car stalls in the middle of the road, he has a duty to take reasonable care to protect oncoming cars from danger, regardless of whether the stall was the result of the driver’s negligence.33

Voluntarily Undertaken Rescue: Anyone who, despite the general no-duty rule, voluntarily undertakes a rescue, has a duty to perform it with reasonable care.34

Prevention of Aid by Others: Finally, there is a duty to take reasonable care not to prevent others from rendering aid.35 Examples include a railway carelessly running over a fire hose or blocking fire engines’ access.36 Another case pressing the category’s limits involved a restaurant refusing to allow someone to make a telephone call to the police.37

To summarize, tort law treats the general prohibition on conduct that negligently risks causing physical harm (misfeasance) differently from failing to take affirmative action to prevent harm (nonfeasance). As to nonfeasance, there is no general duty to rescue others—that is, there is no general prohibition of nonfeasance. But there are some exceptions to this rule. Tort law characterizes these as involving “affirmative duties”—duties to take reasonable steps to prevent or respond to injury. The question before us is whether the distinction between misfeasance (governed by a general duty of reasonable care) and nonfeasance (governed by a general no-duty rule plus exceptional “affirmative duties” to take reasonable care) is tenable. We now turn to that question.

III.     The Dubious Distinction Between Misfeasance and Nonfeasance

The case law and secondary sources draw firm distinctions between misfeasance and nonfeasance, and between duty and affirmative duty, in the manner we have just described. But the tenability of these distinctions is far from clear. To begin to make sense of this area of tort law, it is necessary to lay bare the ways in which these distinctions break down under scrutiny. Only after we dismantle the untenable conceptual structure of affirmative duty can its actual character be discerned.

These distinctions are problematic in both descriptive and normative respects. On a descriptive level, many cases of what tort law calls nonfeasance are difficult, if not impossible, to distinguish from misfeasance. Similarly, sometimes even misfeasance does not look like misfeasance; there are instances of what tort law calls misfeasance that are hard to distinguish from nonfeasance. On a normative level, meanwhile, the distinctions’ significance can be difficult to discern. Even in paradigm cases where the descriptive distinction between misfeasance and nonfeasance is clear—such as the case of the driver who hits the pedestrian versus the bystander who fails to rescue—the normative distinction can be elusive. This puts even greater pressure on the descriptive distinction between creating risk and failing to reduce it: if the two do not have obviously different normative features, then the descriptive distinction may be all that we have in some cases. As we argue, it is often not enough.

A.     Descriptively Difficult: The Distinction between Misfeasance and Nonfeasance

The Third Restatement opens its affirmative duties section by stating the general duty to rescue rule:

§ 37 No Duty of Care with Respect to Risks Not Created by Actor

An actor whose conduct has not created a risk of physical or emotional harm to another has no duty of care to the other unless a court determines that one of the affirmative duties provided in §§ 38-44 is applicable.38

The named sections go on to describe the affirmative duties outlined earlier.39 But first, the comments to Section 37 address the distinction between misfeasance and nonfeasance:

Misfeasance and nonfeasance have a long history as concepts used to explain the distinction between affirmatively creating risk and merely failing to prevent harm. However, this distinction can be misleading. The proper question is not whether an actor’s failure to exercise reasonable care entails the commission or omission of a specific act. Instead, it is whether the actor’s entire conduct created a risk of harm. For example, a failure to employ an automobile’s brakes or a failure to warn about a latent danger in one’s product is not a case of nonfeasance governed by the rules in this Chapter, because in these cases the entirety of the actor’s conduct (driving an automobile or selling a product) created a risk of harm. This is so even though the specific conduct alleged to be a breach of the duty of reasonable care was itself an omission.40

Thus, we learn that the driver of the car is committing misfeasance rather than nonfeasance by failing to brake, because his “entire conduct” created the risk of harm. How to determine whether “the entirety of the actor’s conduct . . . created a risk of harm,”41 however, turns out to be difficult, as illustrated in a variety of key areas.

1.     Defining “Conduct”

Consider the following example: A tenant claims that a landlord failed to install adequate locks on the exterior doors of an apartment building, which resulted in a third party entering the building and assaulting the tenant. Does the landlord have a duty toward the tenant in this instance? And if so, is it an affirmative duty or a general duty attaching to the “entirety” of the landlord’s conduct?

At common law, the landlord had no duty.42 There were many reasons for this conclusion, including that, historically, contract and property law largely determined a landlord’s duties.43 But the nonfeasance-misfeasance distinction also played a role, in that courts did not see landlords as really “doing” anything unless they had affirmatively undertaken to provide security,44 which, if they did, could bring them under the “voluntary undertakings” exception to the no-duty rule.45 Eventually, when courts did recognize a duty, they framed it as an affirmative one, an exception to the usual rule that “a private person does not have a duty to protect another from a criminal attack by a third person.”46 This is what it remains in many cases and treatises to this day.47

But to many modern lawyers and students, the landlord case looks like a straightforward application of the basic negligence principle: that actors have a duty to exercise reasonable care in their conduct.48 The landlord maintained an apartment building with faulty locks: the landlord had a duty to exercise reasonable care in his actions and, if he did not, he is subject to liability for negligence. Indeed, some cases today frame this as a general duty of care.49 Meanwhile, casebooks introduce such cases in the regular materials on the negligence principle, before students have learned about affirmative duties.50 Students have no trouble thinking of these cases in terms of the general principle imposing liability for negligent conduct. No additional legal principle seems necessary to explain the existence of a duty of reasonable care here, and these cases seem indistinguishable from the other run-of-the-mill cases of “misfeasance” that students encounter at the same time. The upshot is that this duty—while now generally recognized—manifests sometimes as an affirmative duty, sometimes as a general duty of care, despite the fact that the landlord’s “conduct” in all these cases is of the same general type.51

The problem of assaults by third parties arises not only between tenants and landlords, but also between guests and hotels, customers and businesses.52 Given that the course of allegedly negligent conduct in all of these cases is essentially the same—faulty locks, faulty lighting, faulty security—one might suppose that there would be uniformity as to whether such activities count as “conduct” for purposes of the general rule that actors must take reasonable care in conduct that risks harm to others. Not so. Hotel guests have had the easiest time establishing a duty of reasonable care on the part of hotels and hotel owners, probably because of the high duty of care innkeepers owed to guests more generally at common law.53 Even here, however, some courts and commentators believe that the duty has expanded over time.54 Courts vary as to whether it is an affirmative duty or simply a general duty of care.55 Landlords, as we have seen, had a duty of care imposed on them in the 1970s, and since then that duty has been variously described as affirmative and negative.56 Meanwhile, in recent years many courts have held that business owners owe a duty of reasonable care to protect invitees from foreseeable criminal assaults.57 This sounds at first like a straightforward application of the general negligence principle, but how courts implement it reveals a more complicated picture. Some courts have concluded that businesses only have a duty when they are aware of specific instances of harm about to occur.58 Others have concluded that businesses only have a duty if they are on notice of prior similar incidents of harm.59 Still others employ a “totality of the circumstances test,” which considers a variety of factors.60 Finally, some courts, like California, employ a balancing test, weighing the foreseeability of harm against the burden of imposing a duty.61 Thus, in many jurisdictions, seemingly simple declarations that “[l]andowners have a duty to take reasonable precautions to protect their invitees from foreseeable criminal attacks” are merely prologues to much more protracted analyses.62 These tests could be criticized for collapsing duty, breach, and proximate cause into the duty element—they essentially allow courts to conclude that there was no duty when they might really mean that a certain precaution was not cost-justified (no breach) or a certain harm was not foreseeable (no proximate cause). But what matters for present purposes is that these inquiries do not turn merely on whether the business is engaged in a course of “conduct” such that “the entirety of [its] conduct . . . created a risk of harm.”63 Instead, courts state that a duty of reasonable care exists but then define the duty question very differently from how they do in most negligence cases. These cases exist in some sort of duty limbo, embracing the language of general duty while proceeding with constraint that suggests an affirmative duty.64

The many cases of landlord, hotel, and business liability are quite similar, within each category and across categories.65 If it were straightforward to determine whether an actor were engaged in “conduct” imposing risk (and thus subject to the regular duty of reasonable care) or not (and thus in the realm of nonfeasance and affirmative duties), then all of these cases should come out the same way. Instead they differ dramatically, within categories, across categories, and over time. Depending on what court you ask and when, these cases may involve nonfeasance to which no duty attaches, nonfeasance to which an affirmative duty attaches, or misfeasance to which the regular duty of care attaches.

The same can be said for many other questions that appear in the affirmative duty section of casebooks, treatises, and the Restatement, including whether a school district is subject to liability to students assaulted by an employee hired on the basis of its untruthful recommendations,66 whether schools have a duty to protect students from physical harm and sexual assault,67 and whether a physician hired to screen for disease or abuse has a duty to do so with reasonable care.68 All of these cases could be framed as involving misfeasance on the part of the defendant—“conduct” that could implicate the regular negligence principle. Perhaps in some of them, courts would want to limit liability. They could do this through doctrines governing proximate cause—by focusing on the foreseeability of a particular risk—or through doctrines governing breach—by focusing on whether certain conduct was reasonable. They could even do this through the concept of duty, for example, by drawing a categorical line limiting liability to a particular class of victims.69 But such duty-based limitations would not need to employ the language of “affirmative” duty. Such language does nothing to elucidate whether a duty of care exists in these cases.    

2.     Conduct and Prior Risk Creation

A similar pattern emerges with the affirmative duty generated by “prior risk creation”:

When an actor’s prior conduct, even though not tortious, creates a continuing risk of physical harm of a type characteristic of the conduct, the actor has a duty to exercise reasonable care to prevent or minimize the harm.70

The difficulty with this definition is that it does not distinguish this “affirmative duty” from the general duty of reasonable care attaching to conduct. The idea of prior risk creation is simply that one must take care that one’s conduct does not unreasonably risk harming another—specifically, even if one behaved reasonably at the outset, one must continue to take reasonable care to minimize or prevent harm. This mandate is simply one manifestation of the more general duty of reasonable care in one’s conduct.

Examples illustrate this. The Third Restatement describes the following as an affirmative duty generated by prior risk creation:

While playing golf, Arnold carefully looks for others and observes no one before driving his ball. Immediately after Arnold hits the ball, Jack suddenly appears from behind a tree in the area toward which Arnold’s drive is heading. Arnold has a duty of reasonable care to Jack with regard to the risk of Jack’s being hit by the ball.71

If this is an affirmative duty, the duty of a driver to brake for a pedestrian is one as well. In both cases, at Time 1, an actor is engaging in reasonable conduct—hitting a golf ball or driving a car. At Time 2, a potential victim appears—a person on the golf course or a person in the road. The appearance of this potential victim imposes on the actor a duty to modify his course of conduct—to yell, “Fore!” or to brake or swerve the vehicle. In certain instances, it may be that the actor cannot avoid injuring the victim because the time is too short for remediation. But that merely means that in such cases, the actor has discharged his duty of reasonable care. The important point is, regardless of the outcome of the case, the actor continues to have a duty of reasonable care at Time 2 because of the course of conduct that he undertook at Time 1, no matter how reasonable that initial conduct was at the time.

Framed in these terms, these cases are indistinguishable. That is because the category of prior risk creation is defined by the same criteria that the Restatement purportedly uses to distinguish misfeasance from nonfeasance: being involved in a course of conduct that imposes risks on others.72 If engaging in conduct that imposes risks on others characterizes misfeasance, then it cannot also be the basis for an exceptional duty in a case of nonfeasance. Prior risk creation does not create an affirmative duty: it just is the basis for a general duty of reasonable care.73

“Prior risk creation” threatens to transform most negligence cases into affirmative duty cases. Many standard negligence cases assume that the defendant had a duty to anticipate that his conduct might impose risks on others and to take reasonable precautions to prevent this.74 If this constitutes an “affirmative” duty of anticipatory risk reduction—if it essentially turns every case into a rescue case—then there is no distinction left between negative and affirmative duties. And this is not a small problem, but a major one, because so many garden-variety cases of misfeasance could be seen as involving non-negligent prior risk creation. For example, whether a landlord should face liability for not providing a smooth rope on a dumbwaiter,75 or companies for not better insulating electrical wires,76 or an airline for not better supervising the boarding process to prevent one passenger from dropping a bag on another’s head,77 or a contractor for leaving a heavy radiator by an open hoistway where a third party could knock it in,78 or a gun owner for not taking utmost care to prevent a third party from shooting the plaintiff,79 or a business for not taking additional precautions to prevent someone from slipping on debris,80 or indeed, a tugboat operator for not equipping its tugs with radios81—all of these are treated as though they involve potential misfeasance, but easily could be framed as involving “prior risk creation” and therefore an affirmative duty.82 Meanwhile, failing to take reasonable care to protect oncoming drivers from one’s stalled vehicle,83 or failing to yell “Fore!” when someone unexpectedly steps into the path of one’s swing84—these are purportedly affirmative duties, but they hardly seem different from the duty to take reasonable care when a pedestrian steps in front of one’s vehicle.

3.     Special Relationships

Another confused area is the affirmative duty generated by “special relationships.”85 At common law, the category applied to a small number of relationships of dependency: carrier-passenger, innkeeper-guest, jailer-ward, and so forth.86 Over time, the category has proved so malleable that it is more a conclusion about whether a duty exists than a reason for imposing a duty. The landlord-tenant relationship, as we have seen, now generates certain duties that it did not at common law. Spousal duties and parent-child duties have evolved substantially, as have those of employer-employee, psychologist-patient, and school-pupil.87 It is not clear whether a change in the nature of these relationships has caused the courts to reevaluate them, or the courts and society have evolved in a way that has accorded changed legal significance to the largely-unchanged relationships. In any case, to the extent that courts give reasons for concluding that these relationships generate duties, those reasons, taken to their logical conclusions, might upend the general rule against a duty of rescue. Cases such as Tarasoff v. University of California, which found that a mental health professional has a duty toward a third-party victim at risk from a patient,88 and Farwell v. Keaton, which found a duty from one social companion to another,89 rest on precepts that could extend duties of care much further. It is not always clear that “special relationships” is a category that does actual work rather than simply placing a name on whatever line the courts are currently attempting to hold on affirmative duties.

Other examples of special relationships offer more reason for recognizing a duty, but in a way that should make us suspicious about whether these are really affirmative duties. Imagine again that a pedestrian steps into the road and is seen by a bystander. Imagine that this time the pedestrian is a very young child, and the bystander is his parent. Everyone would agree that this bystander-parent has a duty to take reasonable care to rescue his child, even though a bystander-stranger would have no such duty. But this is not simply because they have a “special relationship”: the child also has a special relationship with the parent but would be under no duty to rescue him were the tables turned. The parent has a duty because we think the parent is in some way responsible for the child being in this position. The child is in a position of dependency on the parent, and the parent has a duty to supervise the child. This is not so different from the notion that the parent, like the conventionally-negligent driver, through the entirety of his conduct, has contributed to this risk to the child and thus has a duty to prevent harm. In other words, it sounds like the general duty of reasonable care in one’s actions. Much the same can be said of many other special relationships, such as guardian-ward, jailor-prisoner, and so on. It is not clear that these are “affirmative” duties in the same way that a duty of rescue placed on a bystander-stranger would be. Instead these are parties whose courses of conduct happen to involve supervision of others.

The upshot is that a lot of what tort law terms nonfeasance looks curiously like conventional misfeasance, and vice versa. The reasons given for recognizing affirmative duties do not clarify the distinction between misfeasance and nonfeasance. They instead point to confusion between the two, between affirmative duty and just plain duty.

B.     Normative Attempts to Distinguish Misfeasance and Nonfeasance

The difficult task of distinguishing misfeasance from nonfeasance as a descriptive matter is compounded by the fact that the normative difference between the two can seem elusive. If there were a strong normative distinction, then that distinction could guide the analysis, and the question of what counts as “conduct” might matter less. Tort law and tort law scholars have offered several explanations to demarcate and justify the distinction between misfeasance and nonfeasance. When scrutinized, however, these explanations fall apart.

Let us return to the paradigm case: a pedestrian steps out into a street. A driver is approaching. A bystander stands on the sidewalk, within reach of the pedestrian. The driver has a duty to take reasonable steps to avoid hitting the pedestrian. If the driver fails to exercise reasonable care and for that reason hits him, she will be subject to liability (though the pedestrian’s own liability for contributory or comparative negligence may come into play). This is a classic case of misfeasance—one that is at home in Section 7 of the Third Restatement.90 The bystander, however, has no duty to take reasonable steps to pull the potential victim back to the sidewalk, no matter how close he is or how easy it would be. This is a classic case of nonfeasance, one to which no duty of rescue attaches per Section 37 of the Third Restatement.91

The driver and the bystander are far more similar than different. Neither has a causal role in the victim’s choice to step into the road. Neither has a special relationship with him. Neither has undertaken to provide a service to the victim through his or her activity. Both are minding their own business and pursuing their own projects when the victim steps into the road. Both, we stipulate, are in a position to take steps to reduce the risk of harm to the victim. To achieve that end, both would have to take what in any other context would be described as affirmative action: the driver would have to brake or swerve, and the bystander would have to reach out his hand and pull the victim back. In other words, both must interrupt their own (thus far permissible and non-negligent) projects and aims in order to mitigate the effects of the pedestrian’s carelessness or mistake. Yet only the driver is under a duty to take action, while the bystander may continue pursuing his own projects, without regard to the victim’s peril. Why is the driver not similarly allowed to continue pursuing his own projects? And if the driver is obligated to take care, why isn’t the bystander?

Tort law’s answer is that if the driver does not take care to avoid the pedestrian, she will become the instrument of the pedestrian’s harm. Through no fault of the driver, her activity has become a source of physical risk to the pedestrian, and she assumes a duty accordingly.92 Meanwhile, the bystander, though perhaps hard-hearted if he chooses not to act, is not himself imposing a physical risk of harm on the pedestrian. This is the descriptive, physical core of the distinction between misfeasance and nonfeasance. Yet it ought to have a strong normative valence in order to overcome all the features that the driver and bystander have in common. In addition, if the distinction had such a strong normative valence, perhaps that would be useful in cases that are more difficult descriptively. As we saw in the last Section, sometimes it can be difficult to decide who is engaged in risk-creating conduct and who is not. If the driver’s conduct imposes a risk, does the landlord’s? Are people who impose prior risks on others engaging in misfeasance or nonfeasance? The purely physical distinction is not always so easy to make. If a straightforward case offers a clear normative distinction, then that might guide analysis of harder cases.

The common law offers several potential distinctions between misfeasance and nonfeasance. None of them, however, satisfactorily explains the enormous legal distinction between the two.

1.     Malicious Intent and Other Culpable Mental States

One point from a criminal law perspective is that, even if the driver starts out permissibly pursuing her projects, when the pedestrian steps out in front of her, the driver is faced with a choice: try to avoid him or resolve to keep going. Assuming that the driver is not merely inattentive, to keep driving toward the pedestrian suggests, at the least, recklessness about imposing grievous bodily injury, if not knowledge or specific intent. A driver who does not take steps to avoid the pedestrian is one who is guilty of a criminal act with a criminal state of mind. Meanwhile, the same is not true of the bystander.

This may all be true under existing law, but it begs the question. Even if the act of driving toward someone recklessly (or worse) is wrong, the whole question is whether the bystander has committed an analogous wrong. Suppose that the bystander, too, apprehends the choice between reaching out to pull the pedestrian back onto the curb or continuing to pursue his own projects. His mens rea can easily be as culpable as that of the driver. The bystander can stand by with recklessness, knowledge, or even specific intent and will not be liable for the victim’s injury or death. The distinction between driver and bystander does not come down to their state of mind. It comes down, once again, to whether they are acting in a way that creates a duty.

Thus, reframing this situation from a case of negligence into a case of potential manslaughter or murder does not supply the normative distinction we seek. We are still left with the distinction between the driver, whose conduct is seen as imposing a risk, and the bystander, whose conduct is not. The actors may be as ill-intentioned as they like, but the question is whether they have a duty toward the pedestrian in the first place.

2.     But-For Causation

Another unavailing framing is to say that, because the driver is imposing a physical risk on the pedestrian, the driver would be a but-for cause of the pedestrian’s harm while the bystander would not. This again is wrong and begs the question. If the bystander had a duty to try to pull the pedestrian to safety, we would have no trouble saying that his failure to do so was a but-for cause of any harm. If, for example, the pedestrian was a blind person accompanied by a guide, it would be obvious that the guide, in failing to pull the victim back to the curb, was a cause-in-fact of the pedestrian’s injuries. If our conclusion is different about the bystander, that is not because factual causation is any less present in that case. It is because our conclusions about the scope of duty point in another direction—but those conclusions are exactly what is under examination.

3.     But-For Causation and Background Risk

Perhaps the point most frequently made to distinguish the driver from the bystander is that, absent the bystander, the pedestrian would still be in the same perilous position, whereas if the driver were not there, the pedestrian would be better off. As Prosser and Keeton put it, “The reason for the distinction may be said to lie in the fact that by ‘misfeasance’ the defendant has created a new risk of harm to the plaintiff, while by ‘nonfeasance’ he has at least made his situation no worse.”93 According to the Third Restatement, because misfeasance requires a course of conduct imposing a risk, the fundamental distinction between misfeasance and nonfeasance is “whether the same risk of harm would have existed even if the actor had not engaged in the conduct.”94 Indeed, the Third Restatement goes so far as to suggest that that is the real question in distinguishing misfeasance from nonfeasance and that the exceptional “affirmative duties” are useful mostly because they are proxies for the most common variations on that question.95

Thus, for some influential commentaries, the real question is whether in the defendant’s absence the victim would have been exposed to the same risk.96 The Restatement would presumably conclude that, absent the bystander, the pedestrian would still have been in peril, but absent the driver, he would not. Therefore, the driver is under the regular duty of reasonable care, but the bystander is not.

This may sound like a simple restating of the observation that the driver is an instrumentality imposing a physical risk on the pedestrian while the bystander is not. But it is different, and commentators should not be so quick to treat them as the same. An example relied on by the Third Restatement illustrates this point:

[A] retail store that operates in a dangerous and isolated neighborhood might be characterized as creating a risk of criminal activity to patrons. If that characterization were accepted, § 7 would impose a duty of reasonable care to provide security for patrons and employees on the site. However, determining whether the retail store has created a risk of criminal activity requires consideration of what would have happened if the store had not been in operation. Would the patron have been subject to an equivalent risk of attack elsewhere? Or would the patron have forgone late-night shopping if the store had not been there?97

The suggestion here is that, if the same level of risk would have existed in the absence of the store, then the store did not “create” the risk and cannot be held liable for misfeasance. A duty of reasonable care under Section 7 does not exist if the store imposes a level of risk identical to the background level of risk that would exist in the store’s absence.

Leave aside the difficulties of answering this question—the Restatement acknowledges those.98 Is it even the right question? If the question about the retail store is whether it elevates the background level of risk the patron would otherwise experience, then is not the question the same about the driver? What if this is a busy road and, in the absence of the driver’s car, another car would be in exactly the same place? What if this is a multilane highway, and exactly the same peril awaits the pedestrian in the next lane? In the driver’s case, it seems obvious that these details should not matter. But they are exactly the kind of details that proponents claim do matter in cases like that of the retail store. Both approaches cannot be correct.

Moreover, if the operative distinction really is whether the peril would exist in the defendant’s absence, that factual premise is subject to manipulation. We see this problem in some explications of the famous decision in H.R. Moch Co. v. Rensselaer Water Co.99 In Moch, the plaintiff’s warehouse was destroyed after a fire spread to it.100 Defendant waterworks had contracted with the city of Rensselaer to provide water, including water for fire hydrants, and then failed to provide sufficient water at the time of the fire.101 The court concluded that defendant waterworks did not owe plaintiff a duty that could generate liability for plaintiff’s losses.102 One way this result has been justified is to argue that plaintiff was no worse off than he would have been in the absence of defendant: “Had the defendant water company never existed—had there been no water promised to the city—plaintiff’s building would have burned just the same.”103

But this approach is question-begging. Whether the plaintiff would have been worse off depends entirely on what one selects as the relevant baseline. Presumably had the city not contracted with defendant for its water needs, it would have contracted with another party. Perhaps plaintiff would have been better off in that case, because that other party would have provided sufficient water. Is that the relevant point of comparison? Those justifying Moch seem to assume that the relevant counterfactual is a city with no water service. But why is that the relevant comparison? It is not even a plausible one. This explanation does not provide a neutral basis for determining when a duty of care exists and when it does not or for distinguishing misfeasance from mere nonfeasance.104

All in all, considering the background level of risk seems an enormously manipulable way to define duty. Moreover, it seems substantively misguided. If it seems normatively correct that the driver owes a duty of care, this approach does not explain why. To the contrary, it leads in the wrong direction. If it seems obvious that the driver owes a duty of care, then what matters is that it was this driver who imposed the risk, regardless of what risk would have existed in her absence. We are back to the deceptively simple observation that the defendant’s conduct imposed a risk on the victim. But-for causation and background risk have failed to articulate why that matters.

4.     Creating a Risk versus Conferring a Benefit

A related normative gloss sometimes placed on the misfeasance-nonfeasance distinction is that between creating a risk and conferring a benefit. This is related to the notion of background risk: Prosser and Keeton conclude that nonfeasance “has at least made [the] situation no worse, and has merely failed to benefit [the victim] by interfering in his affairs.”105 Cardozo in Moch relied on the risk/benefit distinction to define duty:

If conduct has gone forward to such a stage that [inaction] would commonly result, not negatively merely in withholding a benefit, but positively or actively in working an injury, there exists a relation out of which arises a duty to go forward. . . . The query always is whether the putative wrongdoer has advanced to such a point as to have launched a force or instrument of harm, or has stopped where inaction is at most a refusal to become an instrument for good.106

Commentators have long associated nonfeasance with declining to confer a benefit: Francis Bohlen in 1908 defined misfeasance as “active misconduct working positive injury to others” and nonfeasance as “passive [inaction], a failure to take positive steps to benefit others, or to protect them from harm not created by any wrongful act of the defendant.”107

On an analytical level, this distinction is a useful one. It maps neatly onto the structure of rights that Wesley Hohfeld outlined a few years after Bohlen wrote.108 In Hohfeldian terminology, when an actor owes a duty to a victim, the victim has a correlative claim-right against the actor: a claim that the actor behave consistently with his duty (in this case, to act with reasonable care).109 But where an actor does not owe a duty to the victim, the actor enjoys a Hohfeldian privilege.110 The actor may choose as he pleases to engage in conduct or not, and the victim has no right to demand performance.111 In this sense, Bohlen and the other tort commentators are correct: someone who does not have a duty of care enjoys a privilege to choose to bestow a supererogatory benefit.

The shortcoming of this analysis is precisely that it is a conclusion, not a justification. It does not tell us when a defendant is under a duty and when he is merely failing to confer a benefit or exercise a privilege. It gives us a label for a conclusion but does not help us reach it. Unsurprisingly, notions of what constitutes a mere benefit have changed over time. Bohlen concluded that some hard cases were combinations of duties and privileges, and his claims about them are puzzling from a modern perspective.112 In Moch, Cardozo concluded that failing to provide water service was merely a denial of a benefit.113 Like many other approaches, the risk-benefit distinction does not provide a normative framework for defining when a duty exists.

5.     Characteristic Risks

Another way to frame the distinction between the two cases is to consider the “characteristic risks” of the activities in question. The Third Restatement uses this language in relation to prior risk creation: prior risk creation imposes a duty on the defendant for characteristic risks, not extraneous risks.114 Hitting someone with a golf ball is a characteristic risk of golfing, while falling down on the bleachers is not a characteristic risk of going to a basketball game, such that a person who invites a friend to the game has no duty to him if he falls.115

The idea of characteristic risk may give substance to the idea that someone’s “conduct” imposed a risk. Rather than focus on the purely physical question of whether someone is acting or not (which becomes complicated in situations like that of the landlord), we ask whether there is a nexus between the party’s activities and the risk that results. This may help to finesse the question of what counts as “conduct” imposing risk. It is possible that the “conduct” in question could be an omission or inaction: the important question is whether the risk that resulted was characteristic of that conduct.

There is something to this idea of characteristic risk, and we shall return to it in later sections.116 Note for now, however, how much it depends on cultural understandings of the risk of a given activity. It shifts the focus away from the physical question of whether someone’s acts or omissions count as “conduct” and puts the focus instead on whether we as a society would view the harm that occurred as a characteristic risk of the defendant’s acts or omissions. This might work well in some cases. In the case of the driver, one might conclude that the risk of hitting persons, animals, and objects that enter one’s path is a characteristic risk of driving. Because it is a characteristic risk of an activity that the driver freely chose to engage in, the driver assumes a duty with regard to that risk. This addresses one form of the objection to a duty to rescue: that imposing liability for failing to rescue would in effect commandeer a party to aid others, thus violating her liberty interest—her right to mind her own business.117 Even if the driver was minding her own business at the point at which the pedestrian stepped out, and even if the driver generally has a liberty interest in pursuing her own projects, she surrendered some degree of that interest when she decided to drive, because she chose to assume the obligations generated by the characteristic risks of her activity. Meanwhile, the risk of being present when someone steps out into the street is not a characteristic risk of walking.

Thus, a different basis for a distinction might be that looking out for wayward pedestrians is simply part of the activity of driving, while pulling strangers back to the curb is not part of the activity of walking. Note, however, that this risks begging the question. It is all to the good that this is the rule for driving, but the notion of “walking” could be revised to include taking minimal care to rescue other pedestrians from risks posed by drivers—and that might be all to the good as well. Perhaps it is sufficient to say, descriptively speaking, that we have an intuition that walking does not involve that duty. But our account would be more complete if it could explain why it does not.

6.     A Return to the Descriptive

We return to what can most easily be said about the driver and the bystander: the driver’s car imposes a physical risk of harm on the victim, whereas the bystander is not himself imposing a risk of physical impact or injury. Otherwise, however, the driver and the bystander are in much the same position in relation to the victim. We are at the crux of the difference between misfeasance and nonfeasance, and our best conclusion is that one actor engages in course of conduct imposing a risk of physical harm, while the other does not. We might further be able to say that the driver has to respond because hitting someone is a characteristic risk of driving, while the bystander need not act, because failing to look out for fellow pedestrians is not viewed as a characteristic risk of walking. This observation may simply restate our intuition that driving involves a duty while walking does not.

Moral philosophers have done a great deal of work to generate and analyze intuitions in cases such as these.118 Our task here has been not to repeat that work, but to test various justifications that have been put forward in the tort context for treating the driver and bystander differently. Our conclusion is that these justifications do not help very much. We are back to the law’s principle that “conduct” that imposes a risk also imposes a duty, while mere inaction does not.

There is much more that could be said in favor of and against that principle. One thing, of course, is that it may not offer strong or satisfying normative guidance. Even in the easy case of the driver and the bystander, as we have seen, the distinction can feel slight. In harder cases it is even less satisfying. To take just one example, there are cases involving potential duties not to prevent others from performing a rescue.119 This category involves bystanders who unwittingly get in the way of others’ ability to render aid to victims. Examples involve railroads running over and choking a fire hose or blocking fire engines’ access to a fire.120 Like both the driver and the bystander, the railroad is merely going about its business. Like both the driver and the bystander, the railroad had nothing to do with the underlying choices or bad fortune that have placed the plaintiff in peril. Unlike the driver, the railroad is not an instrumentality that directly imposed risk on or injured the victim. The railroad is more like the bystander, going about its business, not involved in the underlying risk. Both of them could take minimal steps that would substantially reduce the risk to the plaintiff, but must they?

With regard to the railroads, courts have often, though not always, said yes.121 Perhaps there is some variation because either conclusion seems acceptable but not compelled. The railroad is not involved in whatever risk imperiled the victim. True, the railroad is involved in “conduct”—rolling over a hose, blocking a street—and that conduct ultimately disserves the victim. But so does the “conduct” of the bystander’s merely standing on the street corner. Is the mundane physical phenomenon of becoming an obstacle to another’s avoiding harm sufficient to impose a duty? It is as though the bystander, in where he was standing on the curb, was in the way of the pedestrian’s stepping back and avoiding the moving car. Would that be sufficient to generate a duty? There is sometimes little basis for concluding that such differences of sheer physical involvement are enough to put different actors in different legal categories—to impose a duty in some cases and not in others and to call some duties “affirmative” and some just regular duties of care. Yet that is what tort law does.

Perhaps more importantly for present purposes, however satisfactory or unsatisfactory these distinctions are normatively, they are of little help in the vast number of cases in which it is hard to define what counts as a course of conduct imposing risk. In tort law, these include all the landlord and business premises cases, the prior risk creation cases, and the special relationship cases discussed earlier.122 These are just examples of the areas where this problem arises. The railroad cases about preventing aid are difficult in part for the same reason—because they put pressure on the definition of conduct imposing risk. Is being an obstacle to others’ remedial action sufficiently conduct-like to generate a duty? Tort law demands answers to such questions and with few guideposts other than sheer physical involvement.

Our conclusion at this point should be obvious: If misfeasance and nonfeasance resemble each other, and the bases for these distinctions cannot be discerned from a careful analysis of the supposed differences between the two, then there is not much left of the distinction, and not much left of the category of affirmative duty that the distinction generates. For practical purposes, there is no such thing as “affirmative duty.”

We are left, then, with two questions: how did tort law get itself into this situation, and what can be done about it?

IV.     How We Got Where We Are

In the preceding Part we analyzed the tenability of the distinctions tort law draws between misfeasance and nonfeasance and between duty and affirmative duty. To completely understand the reasons these distinctions cannot be maintained, and in order to have an adequate basis for reconstructing this area of tort law, however, we need an additional piece of the puzzle. Without an understanding of the historical and doctrinal pathways that have led to these distinctions, a reformulation cannot be successful. To know where we should go, we must know where we have been.

The incoherence that we uncovered in Part III is the result of a series of developments that have led the law of liability for breach of affirmative duty to its current predicament. We divide these developments into three categories. First, the limited scope of common law liability even for negligent misfeasance influenced the law’s willingness to rest on a general rule that there is no liability for nonfeasance. Second, because there were often adequate and less radical doctrinal alternatives to wholesale rethinking of this area of the law, pressure to engage in rethinking never built up. Finally, because there was conceptual imprecision and confusion even within the category of affirmative duty, the dubious character of the category itself never clearly emerged.

A.     The Limited Jurisdiction of Tort

Two related features of the limited jurisdiction of tort law during the formative era of negligence liability influenced the development of the affirmative duty concept: there was a backdrop of limited conventional duty in tort, and because of this backdrop of limited duty in tort, other areas of the law were seen to be the governing sources of authority regarding the scope of duties that might otherwise have been more carefully scrutinized in tort cases.

1.     A Backdrop of Limited Duty in Tort

When negligence first emerged as a distinct cause of action in tort, there were severe limits on its scope.123 There was no general duty to exercise reasonable care, and not even a general duty to exercise reasonable care to avoid causing foreseeable physical harm.124 Various well-known limits based on the parties’ status, the particular activity that caused harm, and the gravitational pull of other areas of law, were the source of limits on duty.125 For example, landowners had limited duties to trespassers and licensees; different rules governed liability for injuries caused by operating a railroad and riding a horse; and contract rather than tort law was understood to govern liability for injuries caused by negligently-made products.126 Duties arising out of nonfeasance, on the part of innkeepers and common carriers, for example, were even less common.127

In this world, in which ordinary tort liability for misfeasance was limited and circumscribed at the least, and there were very few duties arising out of nonfeasance, the reason that there was no duty would often have been overdetermined. Some cases that belonged in the no-duty category then ended up in the no-affirmative duty category. Our earlier landlord hypothetical is an example.

In addition, the difference between having a circumscribed duty of care but not having breached it, on the one hand, and having no duty to confer a benefit, on the other hand, was incompletely developed. The assertion in a judicial opinion that there was no negligence could mean that the defendant had exercised reasonable care, or that there was no liability even if the defendant had not exercised reasonable care. It was not always clear, for example, whether a decision that the defendant was “not bound” to do something addressed duty or breach of duty.128 When it is a given that there is no tort liability in a particular situation, it takes but small steps to go from saying that a prominent precedent had held that the defendant had no duty, to saying that it held that the defendant breached no duty, to saying that it held that the defendant was not negligent, to saying that it held that the defendant had committed no misfeasance. And if a party had not committed misfeasance, then the basis for imposing liability would have to be the defendant’s nonfeasance, and that was not permitted.

Finally, because the courts were operating against the common law backdrop of few “affirmative” duties, imposing any such duty would have been exceptional.129 Wholesale creation of new affirmative duties, or even the articulation of a general principle explaining their basis, would therefore have been out of order. The articulation and creation of occasional new duties was instead accomplished piecemeal, seemingly exception-by-exception, an approach that needs little or no general theory behind it in order to proceed. This approach reflected the courts’ reluctance to see affirmative duty whole, so to speak, and thereby hindered their ability to see the incoherence of the doctrinal structure they were creating.

2.     Jurisdiction Outside of Tort

When modern negligence law emerged, there was not only limited liability in tort. In addition, other areas of law were understood to govern certain kinds of endeavors.130 Determination of a landlord’s duties to his tenant could be understood as governed by property law rather than tort law, for example. To say that the landlord had no duty to exercise reasonable care toward the guest was not simply to say that there was no duty, but that the landlord’s obligations were not the province of tort law at all, just as the duties of utility companies supplying water under contract might have been understood to be the province of contract rather tort.131 Because tort law did not unequivocally govern these situations, an explanation based on the difference between misfeasance and nonfeasance was in a sense superfluous. Because the inquiry did not have to get as far as this, it was less rigorously examined than it might otherwise have been.

B.     Seemingly Adequate Doctrinal Alternatives to Wholesale Rethinking

In a second category of developments that have led to the current predicament are the availability of adequate independent grounds for denying liability when the courts wished to do so, and the existence of a set of overlapping and vague exceptions to the no-affirmative-duty rule. These factors supported fashioning a limited categories of affirmative duty, rather than fostering careful examination of the notion of affirmative duty itself.

1.     Independent Alternative Grounds for Denying Liability

Some of the cases in which the courts held that there was no affirmative duty involved primary wrongdoing by third parties, in which narrow or rigid proximate cause limitations, such as the last wrongdoer rule, would have precluded liability.132 Because a landowner typically had no liability to a business invitee for injuries caused by a third party wrongdoer, on the ground that any negligence by the landlord was not a proximate cause of the injuries,133 addressing whether the landlord had committed misfeasance as opposed to nonfeasance in failing to provide adequate security would have been beside the point. If a party moved to dismiss or for summary judgment on proximate cause grounds, it was enough for a court to grant that motion without exploring the hypothetical question whether, if causation were considered proximate, liability could then be imposed on the ground that the landlord had committed misfeasance, or instead was guilty of nonfeasance only and therefore had no duty of care. Thus, the procedural and doctrinal posture of some cases would have rendered moot some of the issues that could have led to more searching examination of supposed differences between duty and affirmative duty.

2.     Overlapping and Vague Exceptions that Accepted the Distinction

There are a number of the cases in which the courts found the imposition of liability an attractive fit, or that it could be fitted, into the developing categorical exceptions for voluntary undertakings and special relationships that we discussed earlier. Because there was liability in such cases from early on, there was less need in subsequent cases to confront the nature of the distinction between misfeasance and nonfeasance. Instead, as time went on, an increasing number of relationships and undertakings were encompassed within categories that, at least to some degree, overlapped.

 Prosser, for example, treated the idea of voluntary obligations and special relationships as related.134 He said that the undertakings cases were the first exceptions to the no-duty rule, and that they were recognized for “those engaged in ‘public’ callings, who, by holding themselves out to the public, were regarded as having undertaken a duty to give service, for the breach of which they were liable.”135 The classic categories at common law were common carriers and innkeepers.136 Next came cases involving contractual undertakings, which sounded in the contract doctrine of assumpsit but were eventually regarded as offering a basis for tort liability for injuries to victims who were not party to the contract.137 Prosser argued that the “public callings” cases and contract cases eventually gave rise to the larger category of “special relationships.”138

Despite their potential overlap, the category of voluntarily undertaken obligations is treated today as having separate import from special relationships. The Third Restatement covers duties based on special relationships in sections 40 and 41139 and duties based on undertaking in sections 42 and 43.140 Under sections 42 and 43, the upshot is that assuming an obligation may create a duty of care, regardless of whether it creates a so-called “special relationship.”141

The Third Restatement, meanwhile, identifies prior conduct creating present risk as a category of affirmative duty distinct from the existence of a special relationship or a voluntary undertaking.142 But Prosser had characterized this as merely another example of a special relationship, wherein A’s imposition of risk on A creates a special obligation on A to exercise reasonable care to protect B from that risk.143 He further stated that the rule originally applied only to negligently-created risk:

Where the original danger is created by innocent conduct, involving no fault on the part of the defendant, it was formerly the rule that no such duty arose; but this appears to have given way to a recognition of a duty to take action, both where the prior innocent conduct has created an unreasonable risk of harm to the plaintiff, and where it has already injured him.144

Today, tort law recognizes a duty of care when the defendant’s prior conduct created a risk of harm to the plaintiff, whether or not the prior conduct was negligent.145

In short, the categories of affirmative duty are both ambiguous and vague. The overlapping availability and fuzziness of these categories does not necessarily explain the emergence of the misfeasance-nonfeasance distinction in the first place. But it does suggest one reason that the distinction has not met with much critical reflection. Courts confronted with the no-affirmative-duty rule and a sympathetic case were likely able to find several available exceptions into which a case fit, and this sort of availability would have obviated the need for critical reflection about the coherence of the misfeasance-nonfeasance distinction. In short, at some point the categories of affirmative duty had become sufficiently capacious and overlapping that the courts were not forced to either deny liability or to rethink the entire project of distinguishing between duty and affirmative duty in order to avoid denying liability. This relieved some of the intellectual pressure that might otherwise have existed and that could have led to a rethinking of the very idea of affirmative duty.

C.     Imprecision and Confusion about Key Concepts

The last, and in many ways most important, reason for the predicament in which the law of affirmative duty finds itself is the imprecision, and sometimes outright confusion, that has afflicted three key concepts: misfeasance, rescue, and prior risk-creating conduct.

1.     The Misused Concept of Misfeasance

Even apart from the shakiness of the distinction between nonfeasance and misfeasance, ambiguity surrounds the central concept of misfeasance. The prefix “mis” implies that there must be something mistaken or wrongful about an act in order for it to fall into the category of conduct that is a prerequisite to the imposition of liability. A negligent railroad has committed misfeasance, but an uninvolved third party who does not rescue the railroad’s victim has committed only nonfeasance. The traditional law dictionary definition of the term “misfeasance,” for example, is “[a] misdeed or trespass . . . . The improper performance of some act which a man may lawfully do.”146 In the years when the concept of affirmative duty was being developed, cases in which the defendant had created a risk to the plaintiff but had not done so negligently, and then had failed to rescue the plaintiff, seemed not to involve misfeasance, because the original risk was not wrongfully created. It followed that if liability were imposed, it had to be considered liability for nonfeasance.
We think that is the source of the concept of affirmative duty.
147 In the case of the landlord or landowner that failed to protect a tenant or customer against danger posed by a third party, for example, the landlord and landowner were seen not to have committed any misfeasance.148 The failure to maintain proper security would therefore be seen as an alleged failure to confer the benefit of security on the tenant or customer. Similarly, a driver whose vehicle non-negligently broke down and blocked a road was seen not to have committed any misfeasance.149 The failure to warn oncoming motorists of the blockage would therefore have to be seen as a failure to confer a benefit on such motorists.

Consequently, in such cases any liability imposed on the defendant would have to be seen as based on breach of an affirmative duty to confer a benefit of some sort. Such duties were exceptional, and had to be the subject of a special justification that did not threaten to swallow up the rule that ordinarily there is no liability for nonfeasance. Thus, there came about the development of separate categories of affirmative duty with their own separate justifications.

But all of this depended on a misconception. The purported reason for the strong rule that there was no liability in negligence for failure to confer a benefit—often encapsulated in the rule that there is no duty to rescue—was that the rule protected each individual’s liberty interest. There was no liability for nonfeasance, because a party guilty only of nonfeasance has not surrendered his liberty to be left alone; he was minding his own business and, under that circumstance, tort law could not justifiably commandeer him to come to the aid of another.150

On any sensible philosophical or moral way of thinking about the waiver of liberty, however, waiver could be accomplished not only by wrongdoing, but also by a whole range of actions or agreements that involve no wrongdoing—that is, without any misfeasance. There are and should be many ways to surrender one’s right to mind one’s own business, many ways to waive the liberty not to come to the aid of others. Negligently creating risk to others would certainly be one such way; but non-negligently creating risk to others, or involving oneself with others, would be other ways. In fact, there are three relevant categories here: not only misfeasance, which waives the liberty to be left alone, and nonfeasance, which the courts assumed did not, but also what might be called “feasance”—non-negligent, waiver-generating conduct.

Thus, the notion that there if there was no misfeasance, then the basis of liability, if it were imposed, automatically had to be nonfeasance, ignored the ways that the waiver of liberty could be accomplished without either misfeasance or nonfeasance—it ignored “feasance.” As long as the defendant’s conduct had created a risk, the prohibition on imposing liability for nonfeasance, the bar against commandeering the uninvolved defendant to come to the aid of a plaintiff in danger, would not have been violated. But the courts never saw this.151

And to this day, that misconception, and the confusion to which it leads, continue. The Third Restatement itself is unclear about what counts as an affirmative duty, and therefore about the difference between misfeasance and feasance. As we indicated earlier, the Restatement first states that, with the exception of the affirmative duties that it subsequently identifies, there is no duty of care with respect to risks that the defendant did not create.152 But then some of the supposedly affirmative duties it identifies do involve risks that the defendant created, while some do not. For example, the affirmative duty to minimize risks created by a defendant’s prior non-negligent conduct clearly involves risks the defendant created, as does an affirmative duty to exercise reasonable care to continue a voluntary undertaking to provide services if the failure to provide the services increases risk beyond the level that pre-existed provision of the services.

None of this means that nineteenth and early-twentieth century courts would have thought that imposing liability in the absence of misfeasance was a good idea or that they would have done so if they had not been operating under a misconception. It means only that the courts had a ready, though misconceived, basis for not even reaching that question. Imposing liability in the absence of misfeasance appeared (incorrectly, in our view) to constitute imposing liability for nonfeasance, and (with a few exceptions) tort law simply did not do that.  

2.     Imprecision about the Relation Between Nonfeasance and Rescue

Tort law has always been crystal clear in asserting that there is a distinction between duty and affirmative duty. And it has been equally clear that there is no duty to rescue on the part of a pure bystander. But in creating affirmative duty exceptions to these two propositions, tort law has not always been clear about exactly where “rescue” fits in this scheme. The result has been that the notion that there is no duty to rescue may sometimes have had a restraining effect on the development and explanation of affirmative duties. Are any of the affirmative duties, except the duty to exercise reasonable care when one undertaking a rescue, a “rescue”?

Historically the courts have refrained from calling any of the affirmative duties they have recognized a “duty to rescue.”153 The Restatement also refrains in the same way. Although the term “rescue” sometimes would fit the affirmative duties under particular circumstances, sometimes it would not. Uncertainty about when the notion of rescue does and when it does not apply to a putative or actual affirmative duty creates a number of ambiguities about the status of the affirmative duties. First, although the question is often said to be whether there is a duty to rescue, clearly this is a literal overstatement. The duty to rescue, even when there is such a duty, is a duty to exercise reasonable care to undertake, and in undertaking, a rescue.154 That is, liability for breach of a duty to rescue is liability for negligence; it is not strict liability.155 References to a duty to rescue or the absence of a duty, without more, perpetuate this ambiguity.

Second, rescue can be understood either narrowly or broadly. The narrow conception of rescue is that it involves attempting to eliminate or to reduce an imminent risk of harm. In this narrow sense, rescue occurs only in cases of emergency. The rule that there is no duty to rescue may be taken to apply only to this narrow conception, and therefore not to rule out any affirmative duties that might apply to situations that do not constitute emergencies. If holding a landlord liable for failing to prevent intruders from entering a building and assaulting tenants does not constitute liability for the landlord’s failure to rescue, then the no-duty-to-rescue rule does not address this issue. If the landlord is nonetheless not liable, it must be for reasons that are not based on this rule. As we have seen already, the basis is the idea that such liability would be impermissibly imposed for nonfeasance, but the relation between the rescue rule and the rule precluding liability for nonfeasance is not always made clear.

Third, a broader conception could be that rescue does not require that harm be imminent; that it is not limited to emergencies. Then many, arguably even all of the affirmative duties, could be said to involve rescue, of a sort. This conception would place a lot of pressure on the distinction between conventional negligence and affirmative duty; however, because the conventional duty of care also often involves minimizing or protecting another against a risk of harm whose source is not the defendant’s negligence—that is, rescuing another from such risks. The driver in our continuing hypothetical is in effect held liable for failing to exercise reasonable care to rescue the pedestrian from the risk that arises because of the pedestrian’s stepping into the street, but that term is never applied to such situations. Consequently, because the term rescue is not used to describe the defendant’s obligation in many of the affirmative duty cases, the ambiguity of the concept of rescue continues to confound efforts at understandable description of the landscape of affirmative duty.

3.     The Problematics of the Prior Risk-Creating Conduct Exception

By the time of the Second Restatement, there were some cases imposing liability that could not be shoehorned into the existing affirmative-duty categories.156 They involved non-negligent risk-creating conduct that either endangered or had already injured the plaintiff.157 In limited situations these were said to trigger an affirmative duty.158 The Second Restatement provided that there was a duty to prevent a risk “from taking effect” when an actor had created an unreasonable physical risk of harm, if the actor subsequently realized that it had done so, and even if the actor was not negligent in creating the risk.159 In addition, if an actor’s non-negligent conduct actually caused bodily harm to another and made the other “both helpless and in danger of further harm,” the actor had “a duty to exercise reasonable care to prevent . . . further harm.”160 In contrast, the Third Restatement provides that there is a duty to exercise reasonable care to prevent or minimize harm to another when one’s prior conduct has created a continuing risk to another that is “characteristic of” that conduct.161

Thus, the Second Restatement imposed an affirmative duty with respect to risk alone only when the risk (though not the actor’s conduct) was unreasonable and the actor subsequently realized he had created the risk.162 Once an actor had actually caused harm, however, there was affirmative duty as long as the party injured was helpless and in danger or further injury. The Third Restatement does not require that a risk have created an unreasonable danger or that the actor realize he has created the risk, nor does it require that an already-injured victim be helpless.163 In this respect the Third Restatement goes farther than the Second Restatement. But in one important respect the Third Restatement stops short of the Second Restatement: for an actor to have an affirmative duty (to prevent harm or to rescue) there must be a continuing risk of harm and that risk must be “characteristic of” the actor’s conduct.164

The Third Restatement provides no explanation for its requirement of characteristic risk, and the Reporters’ Note that follows cites only a single case, which declined to impose an affirmative duty, and did not use the phrase “characteristic risk.”165 By eliminating the Second Restatement’s requirement that the plaintiff have suffered bodily harm and be helpless, the Third Restatement permits imposition of a much broader duty to warn or otherwise prevent injury to those whom the defendant’s prior, non-negligent conduct has merely placed at risk of being harmed.

We have two points to make about this category. First, it all but recognizes that the distinction between misfeasance and nonfeasance is inapt. Prior non-negligent risk-creating conduct is not misfeasance. Such conduct is nonetheless a sufficient condition for imposing a duty to exercise reasonable care to prevent the risk from materializing—it is what we earlier called “feasance.” Indeed, the Third Restatement recognizes this without quite acknowledging that this exception does not merely threaten to swallow up the rule, but actually swallows the rule whole.166

Second, it takes only a moment’s reflection to recognize that virtually all conduct creates some risk of harm. If every party who created a continuing risk of any sort had a duty subsequently to exercise reasonable care to prevent or minimize the risk of harm in question, then there would be little limit on the scope of this duty. Every actor would have a duty to exercise reasonable care to rescue those who were put at risk by the actor’s conduct. Only the pure bystander would be exempt, and there would be little point to retaining the category of affirmative duty.

The characteristic-risk requirement is the Section’s means of attempting to limit this otherwise potentially unlimited duty.167 Whether it effectively accomplishes this is doubtful. The Restatement notes that the requirement is akin to the scope of liability limitations that are usually addressed under the rubric of proximate cause in conventional negligence cases.168 But the analogy is inapposite, for the following reason. When the defendant’s risk-creating conduct is negligent, the proximate cause requirement places a limit on the scope of liability by referencing the risks that made the defendant’s conduct negligent.169 That limit is appropriate because ordinarily there is no substantial corrective justice or deterrence goal served by imposing liability for harm caused by risks that did not render the defendant’s conduct negligent: since the defendant did not commit a wrong in relation to such risks it cannot be expected to take precautions that will reduce them.

In contrast, because the defendant in a Section 39 prior risk-creation case has committed no prior negligence, that method of limiting the scope of liability is unavailable as a basis for determining what does and does not count as a characteristic risk.170 There is no prior, posited set of risks for which the defendant is liable, and therefore no prior set of risks that can be said to be “characteristic” of the defendant’s conduct. Instead, the term “characteristic risk” seems tautological, a mere label that—like the supposed distinction between imposing a risk and conferring a benefit—has no independent content.

At this point, we think, the conventional duty cat is almost completely out of the affirmative duty bag. There is a set of duties to exercise reasonable care to prevent harm, the risk of which the party with the duty did not negligently create. In some instances that party non-negligently created the risk, but in other instances that party did not create the risk at all. The distinction between misfeasance and nonfeasance cannot explain these duties, but neither can they all be explained as exceptions to the rule that there is no liability for nonfeasance. Are these duties simply separate islands in a sea of no duty—not, it turns out, linked together by the notion of affirmative duty? Or do these duties have something in common, even if it is not the notion of affirmative duty?

V.     Toward a Reconstruction

In this Part, we first move toward a positive reconstruction of the affirmative duty doctrine, and then address the normative considerations that are relevant to this reconstruction. We contend that, along with the misleading distinction between misfeasance and nonfeasance, the old doctrinal influences that generated the whole apparatus of affirmative duty—the ancient limited jurisdiction of tort law, the availability of easy alternatives that helped avoid rethinking this notion, and the confusions associated with its key concepts—should be set aside. We provide a far more straightforward, and much less strained, descriptive and explanatory account of the state of the law in this area. Turning to the normative side of the issue, and with this kind of account at hand, the courts and commentators will have a much more sensible and realistic set of tools for determining the circumstances under which there should be affirmative duties to “rescue” others from the risks that they face.

A.     Positive Reconstruction: The Affirmative Duties Involve Conventional Negligence

The obvious lesson of our analysis is that the distinction between misfeasance and nonfeasance does not track or explain the differences between the cases that are understood to fall under conventional duty and those that fall under affirmative duty. The very category of affirmative duty is therefore unhelpful. The affirmative duty category is partly composed of cases in which the defendant has created a risk to the plaintiff. The category is also partly composed of cases in which the defendant has not created the risk. In these cases the defendant has in some other way involved himself with the plaintiff, by promise or conduct, in a manner that requires the exercise of care to protect the plaintiff from an independent risk. Clearly, the difference between misfeasance and nonfeasance does not link all these cases together.

If there were something else that linked these cases together—if they had some other common characteristic—then the category of affirmative duty might be worth retaining. The one thing that appears to link them together—is the idea that in none of these cases the defendant’s negligence is the source of the initial risk to the plaintiff. But this idea does not hold up to scrutiny.

The reason this idea does not hold up to scrutiny is that, as our earlier analysis showed, the distinction between creating a risk and failing to confer a benefit, making someone worse off than before and failing to make someone better off than before, is tenable only by holding background conditions constant, and assuming a starting point against which being worse off and being better off can be compared. The distinction depends on making assumptions about the background conditions that seem to be natural or automatically in force, and the background conditions that a party is seen to have created or to be responsible for.

In an earlier tort law world these distinctions were easy to maintain, because potential defendants’ responsibility for background conditions was much less contestable than it is today. Fate and inevitability were much more prominent explanations for misfortune then than in the modern world, in which responsibility for harm is much more easily attributed to the actions of individuals and entities.171 For example, in the 19th Century, dangerous conditions in the workplace were simply a fact of life, a risk assumed as a matter of law;172 and severe, below-freezing weather was not something that the makers of equipment were bound to anticipate.173 In a world in which most misfortune was not understood to be caused by human action or inaction, individuals appeared not to have caused very many of the risks that others face. But the modern development of new affirmative duties has brought the importance of background conditions into the foreground. Now that there is a great deal more conventional duty to exercise reasonable care, what previously looked like affirmative duties can be more easily understood to be conventional duties. With this recognition, we can begin to reconceptualize the supposed affirmative duties.

1.     Prior Risk-Creating Conduct

The category of cases most in need of reconceptualization involves prior risk-creating conduct. We think that, because of the courts’ cautiousness, many of the cases that are sometimes classified as voluntary-undertakings or special relationships actually fall into this category. Landlords, shopping centers, and public utilities (and in the past as well as today, innkeepers and common carriers) create the very conditions under which risk to the individuals with whom they interact may arise. Those conditions are not already in existence, neutrally or naturally supplied. Rather, the conditions would not exist if these entities had not created them. A third-party bystander may sensibly be thought to “confer a benefit” on a hiker by rescuing him from a falling tree. But landlords, shopping centers, and public utilities are not bystanders: rather, they have created the forests in question.

For this reason, it is a strained formulation to say that a landlord who has failed to install locks to guard tenants against intruders has failed to confer a benefit on the tenants, that a shopping center has failed to confer the benefit of good lighting and security in parking lots on its customers, or that a water company that supplies inadequate pressure for fire-fighting has failed to confer that benefit on a homeowner whose house is destroyed by fire. It is much more sensible, realistic, and accurate to understand these cases to be asking whether the defendants have, or should have, a conventional duty to tenants and customers to exercise reasonable care to protect them from the risks in question, subject of course to a conventional proximate cause requirement.

In contrast, the characteristic-risk test introduced by the Third Restatement—and that, as we noted earlier, has rarely applied—risks leading to decisions made as a matter of law when they should properly be considered questions of fact. The notion of characteristic risk looks like what in other contexts would be regarded as a mixed question of fact and law, analogous in some unspecified way to proximate cause. But in our experience, the vast majority of cases in which proximate cause and similar mixed questions of fact and law are a genuine issue, the decision is left to the jury as a question of fact. The only legal rules that relate to the issue are reflected in fairly general jury instructions. The courts’ role is only to set the outer boundaries on jury discretion, by making infrequent rulings that, for example, there was not (or less frequently, there was) proximate cause, as a matter of law.174

In contrast, whether the defendant had a duty (affirmative or otherwise) is a question of law.175 Therefore, cases involving a potential affirmative duty for characteristic risk under Restatement § 39—if that were to govern—would likely be regarded as primarily posing a question for the court: did the defendant have an affirmative duty under the facts of this case? But the characteristic-risk question, like proximate cause, is more nearly a question of degree than a question of kind. Courts have the tools for policing the outer boundaries of fact-finding discretion to decide such questions of degree (using the could-reasonable-people-disagree test, or something like it), but the kind of reasoning that explains decisions about questions of law is more difficult to apply in deciding questions of degree as a matter of law. Only drawing the distinction between characteristic and non-characteristic risk at a high level of generality—in a way analogous to deciding that virtually all blasting is an abnormally dangerous activity resulting in strict liability, and that virtually all damage caused by blasting is “characteristic” of that activity—could solve the problem.

Putting affirmative duty and characteristic risk aside more easily solves the problem. The prior-risk-creation cases can more simply and more accurately be seen as asking whether the defendant exercised reasonable care under all the circumstances. On such an understanding, the key questions come to the forefront, and the difficulties that would be encountered by interpreting them through the lens of characteristic risk can be avoided. Obviously, this might not always be desirable on the merits. But courts have other ways to limit liability—including breach, proximate cause, and policy limitations on duty—besides characterizing the duty question as something it is not.

2.     Relationships

The cases that remain all involve relationships of one sort or another, some of them characterized by special dependence. Students are dependent on schools and teachers, those in custody are dependent on their guardians, employees are to some extent dependent on employers, and people in danger are dependent on those who attempt to rescue them. Sometimes the parties depended on in these situations have actually created the conditions that make risk possible, examples include landlords, shopping centers, and public utilities.

But even when this is not the case, almost always, the parties depended on have voluntarily or knowingly entered into these relationships, and impliedly promised to take care of the dependent parties. Therefore, it would be a strained formulation to say that, in failing to exercise reasonable care to protect a party who depends on them, they have failed to confer a benefit on that party. It is no stretch to say that it is negligent (i.e. misfeasance) for a school not to exercise reasonable care to protect its students from outside risks or for an employer not to obtain medical assistance for a sick or injured employee in the workplace. Thus, it is much more sensible, realistic, and accurate to say that if the defendants in these relationship cases have failed to exercise reasonable care to provide such protection, they have behaved negligently, and to leave it at that.

3.     Prevention of Aid and Warnings to Others

Cases involving the prevention of aid and the duty to warn others of dangers sometimes do not fit these explanations. Many of the defendants in the prevention-of-aid cases were responsible for the background conditions that created the risks in question (thus rendering the cases explainable on this basis), but some were not. For example, the railroad that blocked emergency access in Scruggs was responsible for the location of its tracks and thus for the risk at issue,176 but the defendant that refused a rescuer access to its telephone in Soldano was not responsible for the conditions that gave rise to the need of rescue.177 Similarly, the defendant psychiatrist in Tarasoff had no relationship with the murdered victim of his patient.178 In another case, Podias v. Mairs, two passengers of an intoxicated driver were subjected to a duty of care toward the plaintiff motorcyclist when the driver hit the motorcyclist and they all left him on the highway, where he was run over by another vehicle.179 The defendants had no prior relationship with the plaintiff, and their only role in the accident was being passengers in the vehicle that hit him.180

Some of these cases might simply be considered outliers. Soldano, for example, was decided at the height of California’s tort liability expansionism and has since received criticism from the California judiciary itself.181 But a case as prominent as Tarasoff cannot be dismissed as an outlier. We can think of it as an instance of what in the language of law and economics would be called imposing liability on the cheapest cost avoider.182 But we might also call it “benign commandeering.” Like the innkeepers and common carriers of old, it may be that we impose special altruistic responsibilities on health care professionals and places of public accommodation: the defendant in Soldano, after all, was not a private party with a telephone, but a bartender, in effect an innkeeper.183

B.     Normative Reconstruction: Doing What Comes Naturally

The positive reconstruction we have just undertaken was an effort to make sense out of the law governing the set of cases that are said to fall into the category of affirmative duty, without making use of that concept. Our explanations for the outcomes of these cases, however, were not meant to be evaluative of the outcomes, but only to offer alternatives to the doctrinal rationales that are usually given for the outcomes. We did not mean in that undertaking to say anything directly about how to evaluate these outcomes from the standpoint of principle or policy. We now turn to that task.

The burden of our analysis has been that most of the so-called affirmative duty cases actually are conventional negligence cases of a particular sort: those in which the defendant is not exclusively responsible for creating the danger to the plaintiff. It follows that, since these are for the most part conventional negligence cases, the same considerations of principle and policy that ordinarily inform decisions about whether the defendant has a duty to the plaintiff should govern them.

Of course, these considerations of principle and policy are contestable and are contested.184 Ongoing debates about whether instrumental or deontic standards are the appropriate basis for determining the scope of duties in tort are just as relevant, or irrelevant, to the putative affirmative duty cases as they are to other negligence cases.185 We happen to think of ourselves as having a foot in both camps. In our view, tort law both is and should be a mixed system. Both instrumental and deontic considerations tend to lie behind tort duties, and often both types of considerations should inform decisions about these duties.

But our position on these issues is largely beside the point. Our normative contribution here derives from the lessons that can be learned from our positive reconstruction. It is only a slight exaggeration to say that the affirmative duty emperor has no clothes. A complicated doctrinal structure, based largely on the misleading distinction between misfeasance and nonfeasance, has stood in the way of clear thinking about the cases that have been understood to involve affirmative duty.

We are under no illusion that the concept of affirmative duty will now be dismantled, abolished, and assimilated straightforwardly into conventional negligence doctrine. The courts in general, and tort law in particular, do not work that way. Rather, new wine is poured into old bottles. Doctrines that were previously understood to function in a particular way, and to serve a particular purpose, are retained in name but are refashioned to function in a different way and to serve a different purpose. That is what happened to proximate cause, which was once largely and somewhat woodenly about the nearness of cause and result in time and space and slowly became about foreseeability and harm-within-the risk.186 That is also what happened to custom, compliance with which was once a firm safe harbor for defendants, and slowly became at most a thumb on the scale in trials where its main function often is just is to educate jurors about activities with which they have no personal experience or familiarity.187

We therefore anticipate, and have no insuperable problem with, retention of the existing categories of affirmative duty. But we hope that the courts will come to recognize that these categories are simply descriptions of situations in which the particular type of conduct in which the defendant engaged is relevant to—though not automatically dispositive of—the scope of the duty, if any, the defendant should bear. And we hope that, when the courts feel the need to limit the scope of liability, they will understand that they can do so in the language of duty, rather than affirmative duty.

Thus, in cases involving prior risk-creating conduct by the defendant, the focus should not be on whether particular conduct by the defendant created the particular risk that materialized in harm to the plaintiff. That will often be a sufficient condition, but it should not always be a necessary one. The function of the prior risk-creating conduct requirement should simply be to identify a level of involvement by the defendant sufficient to implicate it as potentially responsible for the plaintiff’s safety. For example, it should not be necessary to liability for a shopping center to have notice of the possibility that its customers are at risk of being assaulted in a parking lot before it is subject to liability for negligently-provided security.188 The question should be whether the shopping center’s design and security measures were reasonable or negligent under the circumstances. The existence of the shopping center itself should be seen as sufficient risk-creating conduct to warrant posing the issue and determining the scope of duty. That should not be ruled out by an inability to identify particular risk-creating conduct by the shopping center.

Similarly, once the duties imposed on those who voluntarily undertake to provide services that they know reduce the risk of physical harm to others are understood not to involve affirmative duty, a much more realistic approach should be possible. The Third Restatement appears to take the position that (subject to certain other requirements) there is always a duty in such situations, and it hinges liability on whether the defendant has exercised reasonable care.189 If the courts were less mesmerized by the notion that providing services generates an affirmative duty, however, they might better recognize that one size does not fit all in this area. The situations vary a good deal, and the differences cannot be handled with the featureless generality that the standard of reasonable care can always be sensitive to the differences.

For example, parents who act occasionally as school crossing guards may have a greater right to discontinue their services abruptly than permanent, organized volunteers; duty may properly vary accordingly, regardless of others’ reliance. A dog-sitter who provides free services might be or might not be subject to the same duties as a paid dog-sitter, despite the fact that what constitutes reasonable care in dog-sitting does not necessarily vary. These duty determinations should be at least partly about the expectations of the parties rather than the content of the objective standard of reasonable care. And the considerations that bear on the duty owed by utilities such as water companies to those who benefit from their provision of water to firefighters may turn on a variety of considerations that are irrelevant to other volunteers. Thus, although we do not necessarily agree with the outcome in the much-discussed Moch case, we agree with Judge Cardozo that insurance and risk-spreading considerations (which have little to do with the concept of affirmative duty) should figure in determinations of duty in such contexts.190

As for the third major category of affirmative duty, in most special relationship cases that do not already involve unrecognized prior risk-creation, whether the relationship is “special” in some sense is a pointless distraction. As the number of relationships that are regarded as “special” has expanded, the basis for this category has become more tenuous. There is nothing like the same dependence involved in the relationship between an employer and employee, a landlord and tenant, or an owner of a business that holds itself open to the public and those who enter the business premises, as there is between a physician and patient. In most of these former cases there is no dependence at the outset at all. It is only after an individual is in danger that there is something that could be called dependence. But of course if that counted as dependence, everyone in danger would have a special relationship with potential rescuers and there would be affirmative duty all the way down.

The Third Restatement candidly admits that the adjective “special” is very nearly just a label for the relationships that the courts hold generate affirmative duty, rather than a test for the relationships that do and do not generate affirmative duty.191 The real issue should be whether there are sufficient elements of implied promising or estopping conduct to trigger a duty of care on the part of the defendant. The dependence of the plaintiff on the defendant is likely to be a good proxy for these conditions, but it should not be an absolute prerequisite. Friends can impliedly agree to take care of each other in certain settings, even if there is no pre-existing dependence.192 Failing to exercise reasonable care to do so should be actionable under such circumstances without tangling with the baggage of affirmative duty.

There are admittedly a few loose ends, as there are bound to be in any effort such as this. We agree with the outcome in Tarasoff and similar cases, but it cannot quite be justified on any of the conventional grounds we have mentioned. We are content to say that, in accepting their licenses, psychiatrists impliedly promise that they will warn individuals who are specifically threatened by the professionals’ patients.193 But we could live with classifying Tarasoff as a rare case of genuine affirmative duty of psychiatrists to the public. We earlier called this “benign commandeering,” imposed on mental health professionals under those limited conditions. What it amounts to is imposing liability on a cheapest cost avoider whose very involvement with potentially dangerous patients constitutes a waiver of the right not to be commandeered to aid others.

Nor do we have an entirely satisfying way of justifying the duty of one who undertakes a rescue to exercise reasonable care in doing so. It is not helpful to think of the rescuer-victim relationship as special, because there need be no relationship at all prior to the time when an attempted rescue begins. It seems to us that a rescuer impliedly promises only to do his or her best, and that in the unusual situation in which the rescuer’s best does not measure up to the objective standard of reasonable care, there should be no liability. The higher standard that tort law imposes may reflect deterrence-based values: interests in deterring incompetent rescues and providing incentives that, in a situation of multiple potential rescuers, will result in the best-qualified actor stepping forward. It may be that absence of causation will sometimes avoid liability anyway, when no other potentially more careful rescuers would have been available even if the defendant had not undertaken the rescue. However, this may reflect an instance in which the courts have been so in thrall of the distinction between misfeasance and nonfeasance, and simultaneously somewhat embarrassed by the law’s unwillingness to impose liability for failing to undertake an easy, riskless rescue, that when they encountered something that seemed to count as misfeasance, they jumped at the chance to impose liability, without thinking carefully about the justification for doing so.

Finally, even the pure no-duty-to-rescue rule for bystanders is ripe for reanalysis. The law of torts already has no difficulty infringing on a bystander landowner’s right to exclude someone in danger from his property: There is a long-established conditional privilege to trespass on or take the property of another in cases of emergency.194 That amounts to rescue by property rather than by a person. Although we might consider bodily autonomy more sacrosanct than rights of property, we are not at all sure that libertarians do, and after all, it is the liberty interest that has been the obstacle to imposing a duty to rescue on pure bystanders. So conditional privilege doctrine would be at least somewhat supportive of creating a bystander’s duty to rescue. Admittedly, ordinarily the conditional privilege simply permits a party in danger to make active use of another’s property; it requires only passivity from the owner. But this is a pretty fine line. In any event, at least some rescues involve only passivity. The Soldano decision,195 which is sometimes thought to be an extreme outlier in regard to rescue,196 can actually be understood to be only a conventional conditional privilege case, in which the defendant should have passively permitted a rescuer to use its property—a telephone. It is a very short step from this form of duty to a bystander’s duty to use his vocal chords for a split-second in order to voice a warning and thereby to perform an easy rescue.

In addition, as a practical matter, concern with the economic burden associated with a bystander duty to rescue should be minimal. Non-rescues that generate liability of an individual in the course of employment will generate vicarious liability on the part of the employer, which can bear this cost. And many, perhaps most, individuals would be covered by the liability insurance features of their homeowner’s, renter’s, and auto insurance when liability did not arise out of employment.

VI.     Conclusion

Tort law has long distinguished between misfeasance and nonfeasance, between duty and affirmative duty. This distinction has done more harm than good. Many cases can easily be described as falling on either side of the line. A claim that a landlord had a duty to take reasonable safety precautions to protect against intruders was once rejected as a wrongful attempt to impose affirmative duty, then accepted as an appropriate affirmative duty, and today appears to many indistinguishable from the general duty of reasonable care. The doctrinal categories used to explain when an affirmative duty is acceptable—special relationships, voluntary undertakings, prior risk creation—too often encompass situations that could just as easily be considered regular negligence cases. Missing is any sense of what makes the cases exceptional—why the duty is “affirmative” in the first place.

Perhaps this is because, as we have argued, even the textbook illustrations of misfeasance and nonfeasance reveal little real distinction between the two. If the difference between a bystander and a driver can be difficult to articulate, that distinction becomes all the harder to apply to a complex activity, such as maintaining a premises or conducting an enterprise. What counts as acting, and as not acting, is harder to identify than one might suppose.

It is also largely beside the point. We have suggested that the existing doctrine is best explained not by its coherence but by other factors, including the common law’s backdrop of limited liability and its relegation of some cases to doctrines other than tort, as well as the accommodating nature of the present-day law’s set of categories and exceptions. The flexibility of the doctrine may have satisfied courts while discouraging the realization that the flexibility was born of incoherence. In any case, the existing categories have allowed the law to evolve and to recognize duties in more cases, even if it has continued to call some duties affirmative without much justification.

We have also begun the project of reconstructing existing law on a firmer conceptual footing. Each category of affirmative duty may be reconsidered to identify what matters in determining when a duty of reasonable care exists. That is not, as it turns out, “special relationships” or the other categories, at least not in and of themselves. Instead, these categories contain within them other factors that help to define the scope of liability.

The common law prefers evolution to revolution. The idea of affirmative duty is not likely to go anywhere, nor is the distinction between misfeasance and nonfeasance. But even if the law does not change, our understanding of these concepts can.


  1. [1]. See, e.g., John M. Adler, Relying upon the Reasonableness of Strangers: Some Observations About the Current State of Common Law Affirmative Duties to Aid or Protect Others, 1991 Wis. L. Rev. 867, 872 (“The common law’s reluctance to require one to render aid to a stranger rests upon the distinction between misfeasance and nonfeasance ... .”).

  2. [2]. See, e.g., Francis H. Bohlen, The Moral Duty to Aid Others as a Basis of Tort Liability, 56 U. Pa. L. Rev. & Am. L. Reg. 217, 219 (1908). Bohlen explains the dichotomy:

  3. [3]. One could argue that the driver’s action should be framed as inaction—failing to brake—and thus should count as nonfeasance as well. Indeed, later on, this Article will seek to complicate the distinction between the two. But tort law has typically viewed the driver as committing misfeasance. See, e.g., Harold F. McNiece & John V. Thornton, Affirmative Duties in Tort, 58 Yale L.J. 1272, 1272–73 (1949) (calling such instances “pseudo-nonfeasance”); Ernest J. Weinrib, The Case for a Duty to Rescue, 90 Yale L.J. 247, 253–54 (1980) (identifying omissions that constitute misfeasance).

  4. [4]. See infra note 9 and accompanying text.

  5. [5]. See, e.g., Kenneth S. Abraham, The Forms and Functions of Tort Law 260 (5th ed. 2017) (stating that the general rule is that there is no affirmative duty but noting that in cases involving special relationships or when engaging in certain activities an affirmative duty may be created).

  6. [6]. See, e.g., Yania v. Bigan, 155 A.2d 343, 346 (Pa. 1959) (holding that when one saw a co-worker drowning there was “no legal ... obligation or duty to go to his rescue”).

  7. [7]. For discussion of the flaws in the latter way of framing the issue see Weinrib, supra note 3, at 253.

  8. [8]. See Richard A. Epstein & Catherine M. Sharkey, Cases and Materials on Torts 519 (11th ed. 2016) (placing cases imposing liability in this situation in a Section entitled “Affirmative Duties”).

  9. [9]. 2 Dan B. Dobbs et al., The Law of Torts § 251 (2d ed. 2011) (“Where the defendant does not create or continue a risk of harm, the general rule, subject to certain qualifications, is that he does not owe an affirmative duty to protect, aid, or rescue the plaintiff.”); Richard A. Epstein & Catherine M. Sharkey, Cases and Materials on Torts 509 (10th ed. 2012) (distinguishing between torts based on misfeasance and nonfeasance); W. Page Keeton et al., Prosser and Keeton on the Law of Torts § 56, at 373 (5th ed. 1984) (“Hence there arose very early a difference ... between ‘misfeasance’ and ‘nonfeasance’–that is to say, between active misconduct working positive injury to others and passive inaction or a failure to take steps to protect them from harm.”); Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 37 (Am. Law Inst. 2012) (“An actor whose conduct has not created a risk of physical or emotional harm to another has no duty of care to the other unless a court determines that one of the affirmative duties provided in §§ 38–44 is applicable.”); Restatement (Second) of Torts §§ 281, 314 (Am. Law Inst. 1965) (defining in the former the elements of a cause of action based on negligence, and in the later the instances where affirmative duties may lead to causes of action); Restatement (First) of Torts §§ 281, 314 (Am. Law Inst. 1934) (defining in the former the elements of a cause of action based on negligence, and in the later the instances where affirmative duties may lead to causes of action); Baccus v. AmeriPride Servs., Inc., 179 P.3d 309, 312–13 (Idaho 2008) (noting that since the case involves an alleged duty to affirmatively place mats to prevent slipping, the case is one about an affirmative duty rather than regular negligence); Coghlan v. Beta Theta Pi Fraternity, 987 P.2d 300, 311 (Idaho 1999) (“‘[O]ne owes the duty to every person in our society to use reasonable care to avoid injury to the other person in any situation in which it could be reasonably anticipated or foreseen that a failure to use such care might result in such injury... .’ Ordinarily, there is no affirmative duty to act, assist, or protect someone else.” (citations omitted)).

  10. [10]. See, e.g., Restatement (Second) of Torts § 314 cmt. c (“The origin of the rule [of no duty to rescue] lay in the early common law distinction between action and inaction, or ‘misfeasance’ and ‘non-feasance.’ In the early law one who injured another by a positive affirmative act was held liable without any great regard even for his fault. But the courts were far too much occupied with the more flagrant forms of misbehavior to be greatly concerned with one who merely did nothing, even though another might suffer serious harm because of his omission to act. Hence liability for non-feasance was slow to receive any recognition in the law.”).

  11. [11]. Id.

  12. [12]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm §§ 7–19.

  13. [13]. Id. § 7.

  14. [14]. Id. §§ 37–44; see id. § 37 (“An actor whose conduct has not created a risk of physical or emotional harm to another has no duty of care to the other unless a court determines that one of the affirmative duties provided in §§ 38–44 is applicable.”).

  15. [15]. See id. §§ 37–44.

  16. [16]. See Abraham, supra note 5, at 261.

  17. [17]. Id. at 260–61.

  18. [18]. Restatement (Second) of Torts § 314 (Am. Law Inst. 1965) (“The fact that the actor realizes or should realize that action on his part is necessary for another’s aid or protection does not of itself impose upon him a duty to take such action.”).

  19. [19]. See, e.g., Weinrib, supra note 3, at 251 (discussing the arguments in favor of and against a duty to rescue and arguing for “a duty of easy rescue”).

  20. [20]. Keeton et al., supra note 9, § 56, at 375.

  21. [21]. See, e.g., Niraj Chokshi, Teenagers Recorded a Drowning Man and Laughed, N.Y. Times
    (July 21, 2017), (describing how a group of teenagers recorded a person drowning on their cell phones and laughed at the person while doing nothing to assist him or call for help).

  22. [22]. See, e.g., Epstein & Sharkey, supra note 9, at 509–78; 2 Dobbs et al., supra note 9, § 405 (stating no-rescue rule and then turning to exceptions); Keeton et al., supra note 9, § 56 (same); Restatement (Third) of Torts: Liab. for Physical and Emotional Harm §§ 37–44 (Am. Law Inst. 2012) (same).

  23. [23]. Keeton et al., supra note 9, § 56, at 383–85; Restatement (Second) of Torts § 314A.

  24. [24]. Keeton et al., supra note 9, § 56, at 383–84 (citing additional duties such as employer
    –employee, jailer–prisoner, school–pupil, landowner–trapped trespasser, parent–child, duties between spouses, and duties among drinking companions). Other duties among co-equals, such as the duty between drinking companions recognized in Farwell v. Keaton, are outliers. Farwell v. Keaton, 240 N.W.2d 217, 222 (Mich. 1976).

  25. [25]. Keeton et al., supra note 9, § 56, at 383. The same relationship may exist where the defendant has a duty to control property, such that injuries caused by that property may create liability in the defendant. Restatement (Second) of Torts § 314 cmt. a.

  26. [26]. Tarasoff v. Regents of Univ. of Cal., 551 P.2d 334, 340 (Cal. 1976).

  27. [27]. Id. (“When a therapist determines ... that his patient presents a serious danger of violence to another, he incurs an obligation to use reasonable care to protect the intended victim against such danger.”).

  28. [28]. This duty may run to the promisee or to a third-party victim. See Restatement (Third) of Torts: Liab. for Physical and Emotional Harm §§ 42–43 (Am. Law Inst. 2012).

  29. [29]. Id.

  30. [30]. Id. § 42 cmts. f & g.

  31. [31]. Id. § 39.

  32. [32]. Id.

  33. [33]. See, e.g., Montgomery v. Nat’l Convoy & Trucking Co., 195 S.E. 247, 250–52 (S.C. 1938).

  34. [34]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 44.

  35. [35]. Keeton et al., supra note 9, § 56, at 382.

  36. [36]. Id. at 382 nn.95–96; see, e.g., Felter v. Del. & Hudson R.R. Corp., 19 F. Supp. 852, 853
    –54 (M.D. Pa. 1937); Hanlon Drydock & Shipbuilding Co. v. S. Pac. Co., 268 P. 385, 386 (Cal. Dist. Ct. App. 1928); Eclipse Lumber Co. v. Davis, 195 N.W. 337, 337–39 (Iowa 1923); Metallic Compression Casting Co. v. Fitchburg R.R. Co., 109 Mass. 277, 277–78 (1872); Luedeke v. Chi. & Nw. Ry. Co., 231 N.W. 695, 696 (Neb. 1930); Globe Malleable Iron & Steel Co. v. N.Y. Cent.
    & Hudson River R.R. Co., 124 N.E. 109, 110–11 (N.Y. 1919); Phenix Ins. Co. of Brooklyn v. N.Y. Cent. & Hudson River R.R. Co., 90 N.E. 1164, 1164 (N.Y. 1909).

  37. [37]. Soldano v. O’Daniels, 190 Cal. Rptr. 310, 312–13 (Ct. App. 1983).

  38. [38]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 37.

  39. [39]. See supra Section II.B.

  40. [40]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 37 cmt. c (first and second emphasis added).

  41. [41]. Id.

  42. [42]. See, e.g., Kline v. 1500 Massachusetts Ave. Apartment Corp., 439 F.2d 477, 481 (D.C. Cir. 1970) (describing common law rule of no duty); Martin v. Usher, 371 N.E.2d 69, 70 (Ill. App. Ct. 1977) (no duty); Teall v. Harlow, 176 N.E. 533, 535 (Mass. 1931) (no duty); Tracy A. Bateman & Susan Thomas, Landlord’s Liability for Failure to Protect Tenant From Criminal Acts of Third Person, 43 A.L.R.5th 207, 207 (1996) (“Traditionally, courts have held that a landlord had no duty to protect tenants from criminal activity on the premises.”).

  43. [43]. See, e.g., Sargent v. Ross, 308 A.2d 528, 531 (N.H. 1973) (stating that liability was limited to situations where “the injury is attributable to (1) a hidden danger in the premises of which the landlord but not the tenant is aware, (2) premises leased for public use, (3) premises retained under the landlord’s control, such as common stairways, or (4) premises negligently repaired by the landlord”). Needless to say, these categories were limited to physical injury to the tenant caused by the condition of the premises. The expansion of the duties of landlords in the 1970s included both duties to protect against third party harms and broader duties to protect tenants and visitors from physical injuries. See, e.g., Kline, 439 F.2d at 481–84, 487 (“The risk of criminal assault and robbery on any tenant was clearly predictable, a risk of which the appellee landlord had specific notice, a risk which became reality with increasing frequency, and this risk materialized on the very premises peculiarly under the control, and therefore the protection, of the landlord to the injury of the appellant tenant.”); Putnam v. Stout, 345 N.E.2d 319, 324–26 (N.Y. 1976) (“Similarly, in the case of harm occurring to third parties who have come upon property with the invitation or license of the occupier, and often with the knowledge and consent of the landowner, consideration must be given to protecting these persons from injury, rather than adhering to technical, out-moded rules of contract.”).

  44. [44]. Johnson v. Wayne Manor Apartments, 837 F. Supp. 705, 707 (E.D. Pa. 1993) (stating that a landlord had “no duty, in the absence of a promise, to protect ... tenant[s] from criminal activity”); Restatement (Second) of Torts §§ 314, 323, 324A (Am. Law Inst. 1965) (treating landlord duties of care as a matter of affirmative duty, which did not exist in the absence of voluntary undertaking).

  45. [45]. Johnson, 837 F. Supp. at 707.

  46. [46]. Kline, 439 F.2d at 481 (setting the foundation for this jurisprudence).

  47. [47]. See, e.g., Restatement (Third) of Torts: Liab. for Physical and Emotional Harm
    § 40 cmt. m (Am. Law Inst. 2012) (“[T]he rationale for imposing a duty on landlords is similar to the rationale for other special relationships in this Section.”); Marc A. Franklin et al., Tort Law & Alternatives: Cases & Materials 200–02 (10th ed. 2016) (describing landlord duties within section on affirmative duties); see also Miller ex rel. Miller v. Tabor West Inv. Co., LLC, 196 P.3d 1049, 1055 (Or. Ct. App. 2008) (describing the duty of a landlord as an affirmative duty).

  48. [48]. See Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 7.

  49. [49]. See, e.g., Durden v. United States, 736 F.3d 296, 302 (4th Cir. 2013) (“In North Carolina, ‘a landlord has a duty to exercise reasonable care to protect his tenants from third-party criminal acts that occur on the premises if such acts are foreseeable.’” (quoting Davenport v. D.M. Rental Properties, Inc., 718 S.E.2d 188, 189–90 (N.C. Ct. App. 2011))); Karim v. 89th Jamaica Realty Co., L.P., 127 A.D.3d 1030, 1030 (N.Y. App. Div. 2015) (“[L]andlords have a duty to take reasonable precautions to protect tenants and visitors from foreseeable harm, including foreseeable criminal conduct by third parties.”).

  50. [50]. See, e.g., Franklin et al., supra note 47, at 50 (discussing McCarty v. Pheasant Run, Inc., 826 F.2d 1554 (7th Cir. 1987), which found for the defendant when the victim was assaulted in her room at the defendant hotel, owing to unlocked sliding glass door).

  51. [51]. See, e.g., 34 Causes of Action 2d Cause of Action Against Landlord for Failure to Protect Tenants Against Criminal Acts §13 (2007) (describing four different potential bases for a landlord’s duty, including “a general duty to” protect against foreseeable harm, a special relationship created by a contract, an implied or express “warranty of habitability,” or a voluntary undertaking of responsibility).

  52. [52]. See, e.g., George L. Blum, Liability of Hotel or Motel Operator for Injury to Guest Resulting from Assault by Third Party, 17 A.L.R.6th 453, § 3 (2006) (“One allegation commonly present in negligence complaints brought against hotels or motels by guests injured in attacks by intruders is that the locks on the doors or windows of the guestrooms were inadequate to afford the guests sufficient protection.”); Bateman & Thomas, supra note 42, at 207 (“[T]here has been increased litigation of this issue [landlord duties] in recent years, with a growing trend in favor of liability in appropriate circumstances.”); 42 Am. Jur. Proof of Facts 2d Landowner’s Failure to Provide Adequate Security § 1 (1985) (“The question of this liability arises in a variety of commercial relationships, including, among others, landlord and tenant, merchant and patron, carrier and passenger, and innkeeper and guest.”).

  53. [53]. See, e.g., Annotation, Liability of Hotel or Motel Operator for Injury to Guest Resulting from Assault by Third Party, 28 A.L.R.4th 80, 84, §2(a) (1984) (“Throughout this nation’s history, the overnight traveler has at times been the victim of assaults and robberies perpetuated by highwaymen and their modern-day counterparts. For that and other reasons, courts have long held innkeepers liable, under appropriate circumstances, for injuries to their guests resulting from attacks by third parties.”).

  54. [54]. See McCarty, 826 F.2d at 1555 (“The high crime rate in the United States has interacted with expanding notions of tort liability to make suits charging hotel owners with negligence in failing to protect their guests from criminal attacks increasingly common.”); Morrison v. MGM Grand Hotel, 570 F. Supp. 1449, 1450–51 (D. Nev. 1983) (describing an affirmative duty on the part of a hotel).

  55. [55]. Compare Morrison, 570 F. Supp. at 1450–51 (describing an affirmative duty on the part of a hotel), and Paliometros v. Loyola, 932 A.2d 128, 134 (Pa. Super. Ct. 2007) (describing duty of hotel and employee as an affirmative duty), with Finch v. HRI Lodging, Inc., 152 So. 3d 1039, 1044 (La. Ct. App. 2014) (describing duty simply as a “duty to take reasonable precautions”), Alqasim v. Capitol City Hotel Inv’rs, LLC, 989 So. 2d 488, 491–92 (Miss. Ct. App. 2008) (describing a duty to take reasonable care), and Bass v. Gopal, Inc., 680 S.E.2d 917, 920 (S.C. Ct. App. 2009) (same).

  56. [56]. See supra notes 43–46.

  57. [57]. See, e.g., Sturbridge Partners, Ltd. v. Walker, 482 S.E.2d 339, 341 (Ga. 1997); Delta Tau Delta v. Johnson, 712 N.E.2d 968, 973 (Ind. 1999), abrogated by Rogers v. Martin, 63 N.E.3d 316 (Ind. 2016); Doe v. Gunny’s Ltd. P’ship, 593 N.W.2d 284, 289 (Neb. 1999); Clohesy v. Food Circus Supermarkets, Inc., 694 A.2d 1017, 1021 (N.J. 1997); McClung v. Delta Square Ltd. P’ship, 937 S.W.2d 891, 898–99 (Tenn. 1996); Timberwalk Apartments, Partners, Inc. v. Cain, 972 S.W.2d 749, 756 (Tex. 1998); Nivens v. 7-11 Hoagy’s Corner, 943 P.2d 286, 292–93 (Wash. 1997) (en banc); see also Posecai v. Wal-Mart Stores, Inc., 752 So. 2d 762, 766 (La. 1999) (collecting cases).

  58. [58]. See, e.g., Krier v. Safeway Stores 46, Inc., 943 P.2d 405, 413–15 (Wyo. 1997) (discussing the different theories courts and jurisdictions have considered when deciding what duties are owed by businesses).

  59. [59]. See, e.g., Sturbridge Partners, 482 S.E.2d at 342 (discussing the foreseeability of risk of harm from criminal activity); Timberwalk Apartments, 972 S.W.2d at 755–56 (stating that one who has control of a premises has a duty to secure those premises when criminal activity or general danger is foreseeable).

  60. [60]. See, e.g., Maguire v. Hilton Hotels Corp., 899 P.2d 393, 399–400 (Haw. 1995); Seibert v. Vic Regnier Builders, Inc., 856 P.2d 1332, 1339 (Kan. 1993); Whittaker v. Saraceno,
    635 N.E.2d 1185, 1188 (Mass. 1994); Clohesy, 694 A.2d at 1027.

  61. [61]. Ann M. v. Pac. Plaza Shopping Ctr., 863 P.2d 207, 214–15 (Cal. 1993).

  62. [62]. Paragon Family Rest. v. Bartolini, 799 N.E.2d 1048, 1052 (Ind. 2003) (proceeding to employ the totality of the circumstances test).

  63. [63]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 37 cmt. c (Am. Law Inst. 2012).

  64. [64]. Indeed, treatises and casebooks generally treat cases involving business owners as “affirmative duty” cases. See, e.g., id. § 37; Franklin et al., supra note 47, at 201–08.

  65. [65]. See, e.g., 42 Am. Jur. Proof of Facts 2d Landowner’s Failure to Provide Adequate Security
    § 1 (2018) (“The reasoning undergirding this decision [Kline] would support the claims of plaintiffs injured in other commercial settings. As an increasing number of courts have recognized, the invitee in any commercial setting should be able to expect that he or she will not be assaulted while on the premises.”).

  66. [66]. Randi W. v. Muroc Joint Unified Sch. Dist., 929 P.2d 582, 584–95 (Cal. 1997).

  67. [67]. A.W. v. Lancaster Cty. Sch. Dist. 0001, 784 N.W.2d 907, 911–21 (Neb. 2010).

  68. [68]. See, e.g., Reed v. Bojarski, 764 A.2d 433, 436–46 (N.J. 2001); Draper v. Westerfield,
    181 S.W.3d 283, 286–92 (Tenn. 2005).

  69. [69]. Strauss v. Belle Realty Co., 482 N.E.2d 34, 35–38 (N.Y. 1985) (limiting liability for injuries caused by New York City blackout to those in privity with the power company).

  70. [70]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 39 (Am. Law Inst. 2012).

  71. [71]. Id. § 39 cmt. c, illus. 1.

  72. [72]. Id. § 37 cmt. c (“The proper question [to distinguish misfeasance from nonfeasance] is ... . whether the actor’s entire conduct created a risk of harm.”).

  73. [73]. To its credit, the Third Restatement acknowledges, with some degree of understatement, that “[t]his Section [on prior risk creation] imposes a duty that might be subsumed under the general duty of reasonable care in § 7.” Id. § 39 cmt. d. It goes on to try to distinguish § 7 cases from prior risk creation cases by claiming that, in the latter, “the actor’s
    risk-creating conduct has ceased, but the risk to the other person continues.” Id. Not only is this interpretation not clear on the face of § 39 itself, but it also fails to provide any clearer guidance or offer any real conceptual distinction. For example, the comment states:

  74. [74]. 2 Dobbs et al., supra note 9, §§ 120, 127 (noting the many possibilities for negligence cases and the reasonable care standard that applies to them); Keeton et al., supra note 9, § 53, at 358 (explaining that in ordinary negligence cases, there is a duty to foresee and avoid potential conduct that would cause an unreasonable risk of harm to others).

  75. [75]. Levine v. Russell Blaine Co., 7 N.E.2d 673, 673 (N.Y. 1937) (discussed in Franklin et al., supra note 47, at 73).

  76. [76]. Adams, 125 N.E. at 93; Braun, 94 N.E. at 207.

  77. [77]. Brosnahan v. W. Air Lines, Inc., 892 F.2d 730, 731 (8th Cir. 1989) (discussed in Franklin et al., supra note 47, at 70).

  78. [78]. De Haen v. Rockwood Sprinkler Co. of Mass., 179 N.E. 764, 765 (N.Y. 1932) (discussed in Franklin et al., supra note 47, at 83).

  79. [79]. Wood v. Groh, 7 P.3d 1163, 1166–67 (Kan. 2000) (discussed in Franklin et al., supra note 47, at 71).

  80. [80]. Negri v. Stop and Shop, Inc., 480 N.E.2d 740, 741 (N.Y. 1985) (discussed in Franklin et al., supra note 47, at 88); Gordon v. Am. Museum of Nat. History, 492 N.E.2d 774, 774 (N.Y. 1986) (discussed in Franklin et al., supra note 47, at 89).

  81. [81]. T.J. Hooper v. N. Barge Corp. (The T.J. Hopper), 60 F.2d 737, 740 (2d Cir. 1932).

  82. [82]. Indeed, they might also be understood as involving “voluntary undertakings” that generated a duty of reasonable care. Although space does not admit of a thorough exploration of this other category, like “prior risk creation” it risks collapsing the distinction between an affirmative duty and a general duty of reasonable care.

  83. [83]. See, e.g., Montgomery v. Nat’l Convoy & Trucking Co., 195 S.E.2d 247, 252–53 (S.C. 1938).

  84. [84]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 39 cmt. c, illus. 1 (Am. Law Inst. 2012).

  85. [85]. See, e.g., id. §§ 40–41.

  86. [86]. See Keeton et al., supra note 9, § 56, at 376.

  87. [87]. See id. at 383–85.

  88. [88]. Tarasoff v. Regents of Univ. of Cal., 551 P.2d 334, 340 (Cal. 1976).

  89. [89]. Farwell v. Keaton, 240 N.W.2d 217, 220 (Mich. 1976).

  90. [90]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 7.

  91. [91]. Id. § 37.

  92. [92]. See id. § 7 cmt. a, § 37 (describing how those who create “risks to others have a duty to exercise reasonable care to avoid causing physical harm”).

  93. [93]. Keeton et al., supra note 9, § 56, at 373; see also Jean Elting Rowe & Theodore Silver, The Jurisprudence of Action and Inaction in the Law of Tort: Solving the Puzzle of Nonfeasance and Misfeasance from the Fifteenth Through the Twentieth Centuries, 33 Duq. L. Rev. 807, 853–54 (1995) (invoking this question about but-for causation to determine what constitutes nonfeasance).

  94. [94]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 37 cmt. c.

  95. [95]. Id. (“In the absence of the rules provided in this Chapter, it is sometimes difficult to determine whether an actor’s conduct created a risk of harm. It would be necessary to explore, hypothetically, whether the same risk of harm would have existed even if the actor had not engaged in the conduct... . Fortunately, specific rules addressing the duty question exist for many of the common patterns in which these difficult cases arise and are contained in §§ 39-44. In the absence of such a rule, the factfinder would have to determine whether an actor’s conduct created a risk of harm as a predicate for determining whether a duty exists under § 7 or whether a duty, if any, must be found in this Chapter.”).

  96. [96]. See, e.g., Keeton et al., supra note 9, § 56, at 375; Rowe & Silver, supra note 93, at 853–54.

  97. [97]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 37 cmt. c.

  98. [98]. See id. This is exactly why the Restatement thinks the affirmative duty categories are useful proxies, though given how malleable we have seen them to be, it is unclear how they are an improvement.

  99. [99]. H.R. Moch Co. v. Rensselaer Water Co., 159 N.E. 896 (N.Y. 1928).

  100. [100]. Id. at 896.

  101. [101]. Id. at 896–97.

  102. [102]. Id. at 899.

  103. [103]. Rowe & Silver, supra note 93, at 853–54; see also Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 37 cmt. c (Am. Law Inst. 2012) (“In the absence of the rules provided in this Chapter, it is sometimes difficult to determine whether an actor’s conduct created a risk of harm. It would be necessary to explore, hypothetically, whether the same risk of harm would have existed even if the actor had not engaged in the conduct.”).

  104. [104]. Admittedly the Restatement should not be read as a statute would be read, but its black letter wording is carefully crafted, debated, and vetted. And § 42(a) requires that the provider’s withdrawal increase risk beyond what “existed” prior to the provision of the service, not beyond what “would have existed.” Id. § 42(a). This is perplexing. For example, suppose that you voluntarily act as a crossing guard near a school but abruptly cease doing so. Because the level of safety that existed prior to your undertaking was that there was no crossing guard, you appear not to be subject to liability. But a Restatement Illustration indicates that you are subject to liability for terminating your services. Cf. id. § 42 cmt. f, illus. 3.

  105. [105]. Keeton et al., supra note 9, § 56, at 373; see also 2 Dobbs et al., supra note 9, § 406 (distinguishing nonfeasance from instances where inaction is part of a larger action, such as not braking appropriately while driving a car).

  106. [106]. H.R. Moch Co. v. Rensselaer Water Co., 159 N.E. 896, 898 (N.Y. 1928).

  107. [107]. Bohlen, supra note 2, at 219.

  108. [108]. Wesley Newcomb Hohfeld, Fundamental Legal Conceptions as Applied in Judicial Reasoning, 26 Yale L.J. 710, 710 (1917).

  109. [109]. Id. at 718.

  110. [110]. Id. at 710, 743.

  111. [111]. See id. at 710–11, 714–15, 742–44 (illustrating how “duty” and “privilege” are jural opposites).

  112. [112]. Bohlen, supra note 2, at 220.

  113. [113]. H.R. Moch Co. v. Rensselaer Water Co., 159 N.E. 896, 899 (N.Y. 1928) (“The failure in such circumstances to furnish an adequate supply of water is at most the denial of a benefit. It is not the commission of a wrong.”).

  114. [114]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 39 cmt. c (Am. Law Inst. 2012).

  115. [115]. Id.

  116. [116]. See infra Sections III.C.3, IV.A.2.

  117. [117]. A libertarian reply could be that the basis of the distinction does not depend on whether a party is minding his own business, but whether his actions infringe on another party’s liberty by causing him harm. If that has not occurred, then there is no basis for imposing liability for harm the other party suffers. See Richard A. Epstein, Torts 288 (1999) (“[T]he object of the law is to maximize the sphere of individual autonomy and choice, which it does by minimizing the level of coercive legal interactions imposed upon unsuspecting bystanders.”); Weinrib, supra note 3, at 249–50 (setting out the view that causation is the only basis for imposition of liability, which cannot be found in typical duty to rescue cases). As we argue, however, this position presupposes a view of causation that is highly contestable.

  118. [118]. The literature on the Trolley Problem and related issues is voluminous. For examples, see generally Judith Jarvis Thomson, Comment, The Trolley Problem, 94 Yale L.J. 1395 (1985).

  119. [119]. See H.R. Moch Co. v. Rensselaer Water Co., 159 N.E. 896, 898 (N.Y. 1928); Keeton et al., supra note 9, § 56, at 382; see supra note 35 and accompanying text.

  120. [120]. See, e.g., Felter v. Delaware & Hudson R.R. Corp., 19 F. Supp. 852, 853 (M.D. Pa. 1937); Louisville & Nashville R.R. Co. v Scruggs & Echols, 49 So. 399, 400 (Ala. 1909); Hanlon Drydock & Shipbuilding Co. v. S. Pac. Co., 268 P. 385, 386 (Cal. Dist. Ct. App. 1928); Eclipse Lumber Co. v. Davis, 195 N.W. 337, 337 (Iowa 1923); Metallic Compression Casting Co. v. Fitchburg R.R. Co., 109 Mass. 277, 277­–78 (1872); Luedeke v. Chi. & N.W. Ry. Co., 231 N.W. 695, 696 (Neb. 1930); Globe Malleable Iron & Steel Co. v. N.Y. Cent. & Hudson River R.R. Co., 124 N.E. 109, 110 (N.Y. 1919); Phenix Ins. Co. of Brooklyn v. N.Y. Cent. & Hudson R. R. Co., 90 N.E. 1164, 1164 (N.Y. 1909).

  121. [121]. See supra Section III.A.

  122. [122]. See supra Section III.A.

  123. [123]. Robert L. Rabin, The Historical Development of the Fault Principle: A Reinterpretation, 15 Ga. L. Rev. 925, 933 (1981).

  124. [124]. Id. at 946.

  125. [125]. See id. at 945–48.

  126. [126]. See id.

  127. [127]. G. Edward White, Tort Law in America: An Intellectual History 14–15 (2003).

  128. [128]. See, e.g., Blyth v. Birmingham Waterworks Co. (1856) 156 Eng. Rep. 1047, 1049 (Eng.) (holding that defendant was “not bound to keep the plugs clear” without indicating whether this was because the defendant had no duty to do so or had a duty but that reasonable care did not require keeping plugs clear).

  129. [129]. See Keeton et al., supra note 9, § 56, at 373 (“In the early common law . . . . the courts were far too much occupied with the more flagrant forms of misbehavior to be greatly concerned with one who merely did nothing, even though another might suffer harm because of his omission to act. Hence there arose very early a difference, still deeply rooted in the law of negligence, between ‘misfeasance’ and ‘nonfeasance’—that is to say, between active misconduct working positive injury to others and passive inaction or a failure to take steps to protect them from harm.”).

  130. [130]. See Rabin, supra note 123, at 945.

  131. [131]. H.R. Moch Co. v. Rensselaer Water Co., 159 N.E. 896, 897 (N.Y. 1928) (explaining that an affirmative legal duty does not exist, but a breach of contract may have occurred); Isbell v. Commercial Inv. Assocs., Inc., 644 S.E.2d 72, 74 (Va. 2007) (“Neither does any contractual duty undertaken by a landlord to repair leased premises under a tenant’s control render the landlord liable in tort ... .”).

  132. [132]. See, e.g., Restatement (Third) of Torts: Liab. for Physical and Emotional Harm
    § 34 (Am. Law Inst. 2012); D’Avolio v. Prado, 277 A.D.2d 877, 877–78 (N.Y. App. Div. 2000) (mem.); Melendez ex rel. Melendez v. City of Phila., 466 A.2d 1060, 1063 (Pa. Super. Ct. 1983).

  133. [133]. Graham v. Atl. Richfield Co., 848 S.W.2d 747, 750–51 (Tex. App. 1993) (noting that a landowner will not be liable for harms to invitees by third parties so long as the third party acts as a superseding cause of the harm and that third party action was not foreseeable by the landowner).

  134. [134]. Keeton et al., supra note 9, § 56, at 373–74.

  135. [135]. Id. at 373.

  136. [136]. See id.

  137. [137]. Id. at 373–75.

  138. [138]. See id. at 383–85.

  139. [139]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm §§ 40–41 (Am. Law Inst. 2012).

  140. [140]. Id. §§ 42–43.

  141. [141]. Id.

  142. [142]. Id. § 39.

  143. [143]. See Keeton et al., supra note 9, § 56, at 377.

  144. [144]. Id. (footnotes omitted).

  145. [145]. See, e.g., Restatement (Third) of Torts: Liab. for Physical and Emotional Harm
    § 39 (“Duty Based on Prior Conduct Creating a Risk of Physical Harm”).

  146. [146]. Misfeasance, Black’s Law Dictionary (1st ed. 1891). This definition, rather than a more current one, seems more appropriate to use in this context. It is true that the label applied to this conduct is not “malfeasance,” which would even more clearly connote wrongfulness. But malfeasance usually connotes intentional wrongdoing or evil conduct. Misfeasance is therefore a peculiar term to apply to innocent, non-negligent conduct, especially given its traditional definition.

  147. [147]. The earliest mention of “affirmative duty” in a law review article that we have been able to find is John H. Wigmore, The Tripartite Division of Torts, 8 Harv. L. Rev. 200, 202 (1894). There were only three other such references before 1910.

  148. [148]. See, e.g., Karim v. 89th Jamaica Realty Co., L.P., 127 A.D.3d 1030, 1030 (N.Y. App. Div. 2015) (“[L]andlords have a duty to take reasonable precautions to protect tenants and visitors from foreseeable harm, including foreseeable criminal conduct by third parties.”); Timberwalk Apartments, Partners, Inc. v. Cain, 972 S.W.2d 749, 756–57 (Tex. 1998) (stating that one who has control of a premises has a duty to secure those premises when criminal activity or general danger is foreseeable).

  149. [149]. See, e.g., Montgomery v. Nat’l Convoy & Trucking Co., 195 S.E.2d 247, 253 (S.C. 1938) (holding trucking company liable for failing to warn approaching vehicles of the blocked highway).

  150. [150]. See Epstein, supra note 117, at 287–88.

  151. [151]. See, e.g., Yania v. Bigan, 155 A.2d 343, 345 (Pa. 1959) (holding that a defendant who encouraged plaintiff to jump into a body of water had no duty to rescue because he had not pushed him in).

  152. [152]. Restatement (First) of Torts §§ 281, 314 (Am. Law Inst. 1934).

  153. [153]. The strongest confirmation of this practice is that the Restatement Third identifies separate affirmative duties that it does not characterize as duties to “rescue.” See Restatement (Third) of Torts: Liab. for Physical and Emotional Harm §§ 40–43 (Am. Law Inst. 2012).

  154. [154]. See Restatement (Second) of Torts § 323 (Am. Law Inst. 1965).

  155. [155]. See id.

  156. [156]. See Keeton et al., supra note 9, § 53, at 359 nn.23–24.

  157. [157]. See id.

  158. [158]. See Restatement (Second) of Torts § 437.

  159. [159]. Id.

  160. [160]. Id. § 322.

  161. [161]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 39 (Am. Law Inst. 2012).

  162. [162]. Restatement (Second) of Torts §§ 322, 437.

  163. [163]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 39.

  164. [164]. Id.

  165. [165]. Id. § 39 Reporters’ Note cmt. c (citing Minahan v. W. Wash. Fair Ass’n, 73 P.3d 1019 (Wash. Ct. App. 2003)).

  166. [166]. The Restatement’s comment provides relevant explanation:

  167. [167]. Id. § 39 cmt. c.

  168. [168]. Id.

  169. [169]. Id. § 29.

  170. [170]. See id. § 39. The phrase “characteristic risk” appears to be taken from Section 20 of the Restatement, imposing strict liability for the characteristic risks of an abnormally dangerous activity—that is, the risks that make the activity abnormally dangerous. See id. § 20 cmt. d.

  171. [171]. See John Fabian Witt, The Accidental Republic: Crippled Workingmen, Destitute Widows, and the Remaking of American Law 43 (2004); Kenneth S. Abraham & G. Edward White, The Transformation of the Civil Trial and the Emergence of American Tort Law, 59 Ariz. L. Rev. 431, 439–41 (2017).

  172. [172]. See, e.g., Titus v. Bradford, Bordell & Kinzua R.R. Co., 20 A. 517, 518 (Pa. 1890) (holding that custom is the “unbending test of negligence” and that an employee assumes the risk of dangerous conditions in the workplace).

  173. [173]. See, e.g., Blyth v. Birmingham Waterworks Co. (1856) 156 Eng. Rep. 1047, 1047–48 (Eng.) (holding that the defendants were not negligent in designing or failing to maintain a valve that leaked due to extreme frost).

  174. [174]. See Abraham, supra note 5, at 149–50 (discussing the limited number of “rules” that govern proximate cause).

  175. [175]. Id. at 259.

  176. [176]. Louisville & Nashville R.R. Co. v. Scruggs & Echols, 49 So. 399, 400–01 (Ala. 1909). In any event, the court denied liability in that case, though courts have imposed liability in some other similar cases. See supra note 120 and accompanying text.

  177. [177]. See Soldano v. O’Daniels, 190 Cal. Rptr. 310, 317–18 (Ct. App. 1983).

  178. [178]. See Tarasoff v. Regents of the Univ. of Cal., 551 P.2d 334, 343–44 (Cal. 1976).

  179. [179]. Podias v. Mairs, 926 A.2d 859, 862, 867 (N.J. Super. Ct. App. Div. 2007).

  180. [180]. See id.

  181. [181]. See Eric J. v. Betty M., 90 Cal. Rptr. 2d 549, 559–60 (Ct. App. 1999).

  182. [182]. See generally Guido Calabresi, The Costs of Accidents: A Legal and Economic Analysis (1970) (introducing and applying the concept of the cheapest cost avoider).

  183. [183]. See Soldano v. O’Daniels, 190 Cal. Rptr. 310, 312 (Ct. App. 1983).

  184. [184]. See, e.g., Kenneth S. Abraham, Strict Liability in Negligence, 61 DePaul L. Rev. 271, 298–301 (2012) (arguing that tort law should be understood as a mixed system pursuing both deontic (rights-based) and instrumental goals); John C.P. Goldberg & Benjamin C. Zipursky, Torts as Wrongs, 88 Tex. L. Rev. 917, 918–19 (2010) (developing a deontic conception of tort law).

  185. [185]. See, e.g., Abraham, supra note 184, at 298–301; Goldberg & Zipursky, supra note 184, at 918–19.

  186. [186]. See, e.g., Restatement (Third) of Torts: Liab. for Physical and Emotional Harm
    § 29 cmt. a (Am. Law Inst. 2012) (discussing history and current view of limitations on liability for tortious conduct).

  187. [187]. See, e.g., Kenneth S. Abraham, Custom, Noncustomary Practice, and Negligence, 109 Colum. L. Rev. 1784, 1802 (2009).

  188. [188]. See, e.g., Morgan v. Bucks Assocs., 428 F. Supp. 546, 551 (E.D. Pa. 1977) (arguing that knowledge of past criminal activity in a parking lot can trigger a duty to provide reasonable security of the premises to protect third parties from potential harms); Foster v. Winston-Salem Joint Venture, 274 S.E.2d 265, 267 (N.C. Ct. App. 1981), aff’d in part, rev’d in part, 281 S.E.2d 26 (N.C. Ct. App. 1981) (holding that the criminal activity in the mall’s parking lot was not enough to give rise to a duty in the instant case, implying that were the criminal activity higher and had the mall had notice, it would have had a duty to provide reasonable security).

  189. [189]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 42.

  190. [190]. See H.R. Moch Co. v. Rensselaer Water Co., 159 N.E. 896, 898 (N.Y. 1928).

  191. [191]. Restatement (Third) of Torts: Liab. for Physical and Emotional Harm § 40 cmt. h (“The term ‘special relationship’ has no independent significance. It merely signifies that courts recognize an affirmative duty arising out of the relationship where otherwise no duty would exist ... .”).

  192. [192]. Cf. Farwell v. Keaton, 240 N.W.2d 217, 222 (Mich. 1976) (“Farwell and Siegrist were companions on a social venture. Implicit in such a common undertaking is the understanding that one will render assistance to the other when he is in peril if he can do so without endangering himself.”).

  193. [193]. See Tarasoff v. Regents of Univ. of Cal., 551 P.2d 334, 340–41 (Cal. 1976).

  194. [194]. See, e.g., Vincent v. Lake Erie Transp. Co., 124 N.W. 221, 222 (Minn. 1910); Ploof v. Putnam, 71 A. 188, 189 (Vt. 1908).

  195. [195]. Soldano v. O’Daniels, 190 Cal. Rptr. 310 (Ct. App. 1983).

  196. [196]. See Epstein & Sharkey, supra note 8, at 485–86 (“Soldano has sometimes received a rocky reception in other jurisdictions.”).


David and Mary Harrison Distinguished Professor of Law, University of Virginia School of Law.


Vice Dean and David H. Ibbeken ‘71 Research Professor of Law, University of Virginia School of Law.

The authors would like to thank Vincent Blasi, Kimberly Kessler Ferzan, and Gregg Strauss for helpful comments and Andrew Miller for excellent research assistance.