1. The general conception of duty
In his great judgment in Donoghue v. Stevenson1 Lord Atkin, before articulating the neighbor principle, noted a deficiency in the judicial treatment of duties of care in negligence. Only rarely had the common law formulated “statements of general application defining the relations between the parties giving rise to a duty of care.”2 Instead, proceeding on a case-by-case basis, the courts had concerned themselves with the particular kind of relationship before them on any occasion and had therefore focused on the specific status of one or other of the parties, “whether manufacturer, salesman or landlord, customer, tenant, stranger, and so on.”3 Thus, the only way to determine whether a duty of care existed was to see whether the case could be referred to some “particular species”4 that the law had already recognized. He continued:
And yet the duty which is common to all cases where liability is established must logically be based upon some element common to all the cases in which it is found to exist. … [T]here must be, and is, some general conception of relations giving rise to a duty of care, of which the particular cases found in the books are but instances.5
The general conception of the duty of care—its theoretical basis, its structural constituents, its more recent disintegration back into particular duties, and the need to recapture what a general conception of duty implies—is the subject of the present chapter.
Lord Atkin regarded the existence of a general conception as a necessity (“there must be, and is, some general conception”). Although he said little about the nature of this necessity, he presumably had in mind something like this: the common law, by its own internal logic and dynamism, cannot treat the particular instances of duty as a chaotic miscellany of disparate and independent norms.7 Duties of care are constituents of a normative system. The coherence of such a system requires that all duties of care should be thematically unified through the same underlying principle. The general conception of duty reflects the common aspect that each particular duty must have if it is to be systematically related to every other particular duty. The necessity to which Lord Atkin refers is a juridical one: the general conception of duty is an implication of the internal coherence required by the law’s systematic nature.
In recent years the sense of juridical necessity apparent in Lord Atkin’s judgment has waned. Lord Atkin’s own formulation of the general conception in terms of a duty to avoid foreseeable harm to a neighbor, path-breaking as it was, has been recognized not to provide a practical test. Moreover, the very idea of a general conception is sometimes thought to be superfluous given the casuistic nature of common law reasoning.8 Even in courts that accept the need for an overarching framework for the duty of care, the general conception takes the form of multi-staged formulae that are verbally comprehensive without necessarily being juridically coherent.9 The widely accepted idea that duty is a matter of “policy” (whatever that might mean) has led to a distaste for the abstract practical reasoning that undergirds a general conception of duty.
The result has been a disintegration of duty in two related senses. First, because each kind of duty reflects the particular set of policies deemed appropriate to it or the particular constellation of casuistic considerations from which it emerges, the preoccupation with particular species of duties, which Lord Atkin decried, has returned. Duties are differentiated, usually according to the kinds of harms with which they deal (physical injury, economic loss, psychiatric injury, and so on), without attention to (in Lord Atkin’s words) “the element common to all cases in which [a duty] is found to exist” and therefore without awareness of the strands that weave all duties of care into a coherent system in which each duty illuminates, and is illuminated by, all the others. Second, not only has the whole ensemble of duties disintegrated into a collection of particular kinds of duties, but the very idea of a duty has disintegrated. Instead of the duty of care being an internally coherent normative category, it has been fragmented into the separate factors that determine the duty’s ground and limits. Hence, the reasoning in support of a duty is marked by ad hoc compromises among these separate factors rather than by an elucidation, in the context of a particular case, of the conception under which the defendant’s act and the plaintiff’s injury form a unified normative sequence. These two senses of disintegration go together, because the absence of coherence within the notion of duty renders otiose the necessity for coherence that Lord Atkin postulated among particular duties.
This disintegration of duty has undermined the most notable achievement of negligence law in the twentieth century. However indeterminate Lord Atkin’s own formulation of the general conception of duty, other leading negligence cases also took up and advanced the striving for coherence to which he gave voice. In contrast, the more recent disintegration of duty manifests a failure to develop tort law in a normatively coherent way. If I am right about this, a principal task for negligence law in its ongoing evolution is to grope its way back to the conception of coherence that was implicit in Lord Atkin’s celebrated judgment.
In this chapter, I want to discuss these developments from the standpoint of corrective justice. This standpoint is especially germane, because corrective justice is the theoretical notion that accounts for whatever coherence private law might have. Corrective justice ties coherence to the justifications that inform private law’s characteristic concepts. Legal doctrine is viewed as coherent only to the extent that its underlying justifications are coherent. These justifications, in turn, are coherent only if their force as reasons is congruent with—rather than artificially truncated by—the structure of the relationship between the parties. Because liability relates the defendant as doer of the injustice to the plaintiff as sufferer of the same injustice, a finding of liability is coherent when the reasons for considering the defendant to have done an injustice are the same as the reasons for considering the plaintiff to have suffered that injustice. As was argued in chapter 1, the normative basis of such coherence is a right of the plaintiff that imposes a correlative duty on the defendant.
Seen in the light of corrective justice, Lord Atkin’s comments can be understood as follows. The notion that the law of negligence is a coherent system of justification precludes particular duties of care from being regarded as isolated norms. Rather, the particular duties are systemically related to one another because they participate in a shared general conception of duty. To play its role as a necessary determinant of coherence across particular duties, the general conception of duty must itself be an expression of the very coherence that it imparts. For “the relation between the parties” (as Lord Atkin calls it)10 to be coherent, that relation has to be regarded as correlatively structured through the defendant’s doing and the plaintiff’s suffering of the same wrong. This in turn has two complementary aspects. On the one hand, the general conception of duty must be part of an integrated ensemble of concepts that allows the sequence from the defendant’s doing of the negligent act to the plaintiff’s suffering of the injury to be understood as a normatively coherent unit in which the injustice done and the injustice suffered are the same. On the other hand, the general conception of duty considered in itself must be formed from normative elements that reflect the correlativity of the defendant’s duty to the plaintiff’s right. When these two aspects are present, the general conception of duty then embodies the correlative structure of justification that renders any particular duty coherent both with the other members of the ensemble of negligence concepts and with other particular duties. There “must be” such a general conception (as Lord Atkin insists) because otherwise the law of negligence would be incoherent—a possibility that he assumes the law cannot acknowledge.
Taking corrective justice as the theoretical notion underlying Lord Atkin’s insistence on a general conception of duty, this chapter discusses the disintegration of duty in the following steps. Section 2 is concerned with the first of the aspects of coherence mentioned above—that is, with how duty fits with other negligence concepts (failure to exercise reasonable care, factual causation, and proximate cause) to connect the defendant’s act to the plaintiff’s injury in a normatively coherent way. Accordingly, the section outlines the role of the landmark cases of the twentieth century in treating the injustice done by the defendant as identical to the injustice suffered by the plaintiff. Section 3 deals with the second aspect of coherence; it sets out the internal structure of the duty of care—that is, what its constituents must be if it is to reflect a coherent conception of the doing and suffering of a wrong. In so doing, the section sets out the general conception of the duty of care that juridical coherence requires. The duty of care will thereby be exhibited as having a definite structure toward which the legal reasoning of any particular case ought to be oriented. The presence of such a structure suggests that it is mere dogmatism to maintain, as tort scholars often do, that duty is nothing but “a shorthand statement of a conclusion, rather than an aid to analysis.”11 Section 4 examines the two-stage test for negligence that has been used in Canada and elsewhere. My argument is that this internally fractured test, as applied by the Supreme Court of Canada, provides a conspicuous example of the disintegration of duty and, accordingly, is inadequate for the development of a coherent jurisprudence of negligence. Finally, section 5 discusses the meaning and relevance of the much-invoked “policy” for the determination of the duty of care. The disintegration of duty is the consequence of thinking that duty is a matter of policy, and that policy, in turn, refers to the various independent goals that liability might serve. On this view each particular kind of duty represents the balance of goals, in themselves diverse and competing, that is peculiar to it. However, another notion of policy refers to the exercise of practical judgment in elucidating what the general conception of duty might mean in particular circumstances. The general conception provides not (as has often been assumed) a test of duty, but a structure of thinking that is actualized in legal reasoning through the casuistic assessment of facts or comparison of cases or through the elucidation of its particular normative features in the overall context of a legal system that values coherence. This second notion of policy is, I suggest, not only compatible with but indeed required by the general conception of duty.
2. The place of duty among the negligence concepts
How, then, does negligence doctrine treat the defendant and the plaintiff as the correlatively situated sufferer and doer of the same wrong? Negligence concerns the plaintiff’s being wrongfully injured through the defendant’s creation of an unreasonable risk. If negligence liability is to be a coherent normative phenomenon, the injury and the risk-creation have to be understood as the constituents of a single wrong that is elucidated through an integrated ensemble of legal concepts. In this way, the differing experiences of the parties as doer and sufferer and the temporal gap between the doing and the suffering are subsumed under a unified set of normative categories that render the wrong done identical to the wrong suffered.
Coherence requires that the injustice relate act to injury and vice versa. Precluded are definitions of the injustice between the parties in terms that pertain to one of them alone. As negligence law recognizes, the injustice does not consist merely in the unreasonably created risk considered in itself; that would one-sidedly focus on the defendant’s wrongful action and entail liability for unreasonable risk-creation even without damage. Nor does it consist in the harmful effect considered in itself; that would one-sidedly focus on the plaintiff’s injury and entail strict liability, rather than negligence.12 Nor, again, does the injustice consist in the combination of wrongful action and injury each considered one-sidedly; that would mean that, although the defendant has committed a wrong and the plaintiff has suffered one, these are two different wrongs, each resting on its own foundation. Rather, the injustice embraces the entire span from the act to injury: the defendant’s act is viewed as a wrong because of its potential to cause this kind of injury, and the plaintiff’s injury is viewed as a wrong because its potential occurrence is a reason for considering the defendant’s act wrongful. Then the sequence from act to injury forms the single wrong that the defendant can be said to have done and the plaintiff to have suffered.
To be coherent, tort doctrine elaborates legal concepts that treat the defendant’s act and its effect on the plaintiff as an integrated sequence in which there is a single injustice that is the same for both parties. In legal terms, this sequence begins with the defendant’s breach of the standard of reasonable care and ends in the factual causation of injury. However, the sequence can be regarded as integrated only if its two termini operate not as atomistic elements that the law simply adds together, but as constituents of liability that, for purposes of tort law, each derive their significance from the other. Hence, the unreasonableness of the risk created by the defendant must lie in the possible occurrence of the kind of injury that the plaintiff suffered. This way of relating the negligent act to the injury makes the injustice of unreasonable risk-creation the same for both parties.
The signal achievement of negligence law in the twentieth century was to develop the concepts of negligence analysis in a way that coherently links the unreasonable risk to the harm suffered. Duty and proximate cause are crucial components in this linkage. These concepts connect wrongdoing and injury by describing the wrongful risk in terms of the range of the potential victims and consequences through which the risk is to be understood as wrongful. Duty connects the defendant as a wrongdoer to the plaintiff as a member of the class of persons wrongfully put at risk. Similarly, proximate cause connects the defendant’s negligence to the plaintiff’s suffering of the kind of injury or accident the risk of which rendered the defendant’s act wrongful. Because both duty and proximate cause are requirements of liability, the defendant is not held legally responsible for the materialization of a harm that is not within the set of possibilities that supply a reason for exercising due care. When, however, the injury suffered by the plaintiff is to a member of the class of persons that the defendant wrongfully put at risk and is the kind of injury or accident that due care is supposed to avoid, then the wrongfulness of both the defendant’s action and the plaintiff’s injury is referable to the same sort of risk. Under those circumstances, the sequence from the defendant’s creation of an unreasonable risk to the materialization of that risk in injury to the plaintiff constitutes the same injustice for both parties.
The leading twentieth-century cases on duty and proximate cause gave legal expression to this conception of negligence liability. Three developments were particularly noteworthy. The first, of course, was Donoghue v. Stevenson13 (and its American predecessor MacPherson v. Buick Motor Co.).14 In place of a fragmented collection of duties that varied according to the particular social and contractual relationships between the parties, Donoghue established that the duty of care flowed from the defendant’s risk-creating action as such and from its reasonably foreseeable effect on those who ought to have been within the defendant’s contemplation. The general conception of the duty of care formulated in this case thereby connected the defendant as the creator of an unreasonable risk and the plaintiff as a person whose endangerment made the risk unreasonable.
A more explicit and complete development of the same idea appeared in Cardozo’s judgment in Palsgraf v. Long Island Railroad Co.15 Cardozo held that there could be no liability unless the defendant’s breach of duty was a wrong in relation to the plaintiff. Because in that case the defendant’s conduct was not wrongful toward the plaintiff (although it was arguably wrongful toward someone else), the defendant was held not to be under a duty with respect to the plaintiff’s loss.
Cardozo’s explicit purpose was to construe the wrongfulness of negligence in a way that was specifically appropriate to tort law. He noted “the shifting meanings of such words as ‘wrong’ and ‘wrongful.’”16 What the plaintiff had to show was a wrong that was “a violation of her own right, and not merely a wrong to some one else, nor conduct ‘wrongful’ because unsocial but not ‘a wrong’ to any one.”17 To detach the notion of wrong from a beach of duty owing to the plaintiff “is to ignore the fundamental difference between tort and crime.”18 For Cardozo, tort liability—in contrast to other kinds of societal judgments about culpability—featured a wrong done by the defendant in relation to the plaintiff. Accordingly, the very reason for thinking that the defendant acted wrongfully also had to be the reason for thinking that the injury suffered by the plaintiff was wrongful. The duty of care in negligence law was to be understood as rendering the normative significance of the wrong identical for both parties. “Negligence … is thus a term of relation.”19
Cardozo’s outstanding contribution was to align the relational nature of tortious wrongdoing with the relational significance of unreasonable risk. In the words of the opinion, “The risk reasonably to be perceived defines the duty to be obeyed, and risk imports relation.”20 As a way of referring to the harmful potentialities inherent in a given act, risk is not intelligible in abstraction from a set of perils and a set of persons imperiled. A negligent act releases a set of harmful possibilities that due care should have avoided. The plaintiff cannot recover unless the injury she suffered actualizes a possibility within this set. The imperiling of the foreseeably affected class of persons is the reason for considering the defendant’s act negligent; it also must be the reason for thinking that the plaintiff has been wronged. Only if the plaintiff is among that class does the reason for thinking of the defendant’s action as wrongful pertain to the plaintiff. Because in Palsgraf the prospect of the plaintiff’s injury was not a reason for considering the defendant’s action negligent, the defendant was not under a duty toward her.
Cardozo’s treatment of duty forges a normative connection between the defendant’s action and the plaintiff’s injury. The connection is normative, because it is based on the reason for considering an act to be wrongful. Given that the structure of the relationship between the parties is one of correlativity, that reason must simultaneously provide a ground both for holding the defendant liable and for holding the plaintiff entitled to recover. Under Cardozo’s analysis of the relational quality of unreasonable risk, the duty of care makes the same reason pertain to both parties.
In contrast, Andrews’s dissenting judgment in Palsgraf connects the parties in a historical but not a normative way. The connection is merely historical, because the element of fault required for the defendant’s liability to the plaintiff is satisfied by the fact that the defendant’s negligence is the historical antecedent of the plaintiff’s injury. In Andrews’s view, the injury suffered can be the basis of the plaintiff’s recovery even though the wrong was relative to a third party rather than the defendant. Unanswered is the question of why the merely historical connection between the defendant’s negligent act and the plaintiff’s injury should justify liability on the basis of fault. It is no answer to say that this negligence caused the injury; that answer, by transforming cause into the determinant of the plaintiff’s recovery, should also apply to causally effective action that is innocent (and thus as nonculpable relative to the victim as the defendant’s conduct in Palsgraf). Negligence liability would then collapse into strict liability.
Another difficulty flows from Andrews’s postion. Ever since the devastating critique of strict liability by Oliver Wendell Holmes, no one has been able to sustain the position that liability should be based on causation.21 The particular problem that Holmes identified is that causation is unable to generate its own limits, so as to preclude a regression to causes however distant. Andrews is not insensitive to this problem. He insists that for the defendant to be liable, the negligence must be not merely a cause, but the proximate cause of the damage. However, having rejected the notion that the parties are normatively connected (and thus liability is limited) through the idea of unreasonable risk, he has no coherent conceptual framework within which judgments about the proximateness of causation can be situated. In Andrews’s own account the determination of proximate cause turns out to be an arbitrary exercise in practical politics, intelligible more through rhetoric and metaphor than through legal analysis.22
A more coherent conception of proximate cause than the one Andrews put forward emerged from yet another major development of tort jurisprudence in the twentieth century, the Privy Council’s decision in the Wagon Mound case.23 The case held the requirement of proximate cause unsatisfied when the defendant negligently exposed the plaintiff to the risk of one kind of injury, but the plaintiff suffered an injury of a different kind. This decision made proximate cause run parallel to Cardozo’s conception of duty.24 Just as Palsgraf required that the plaintiff be within the class of persons unreasonably imperiled, so Wagon Mound required that the injury or accident be of the sort that renders the defendant’s risk-creation unreasonable. Thus, with respect both to the person injured and to the injury or accident, the harm that occurred actualized the possibilities for danger that it was negligent for the defendant to have created. Duty and proximate cause both functioned to make the wrongfulness of what the defendant did the same as the wrongfulness of what the plaintiff suffered.
When this complex of leading cases is considered as a whole, the main categories of negligence liability—duty, breach of duty, proximate cause, and factual cause—form a coherent set that traces the sequence from the defendant’s negligent act to the plaintiff’s injury. Breach of duty and factual causation are the termini of this sequence, with the former referring to the defendant’s creation of unreasonable risk and the latter to the materialization of risk in injury to the plaintiff. Duty and proximate cause integrate these termini into a normatively coherent unit by characterizing in terms of the same unreasonable risk the wrongfulness of what the defendant did and of what the plaintiff suffered. The result is that the duty of care is a member of an interconnected ensemble of concepts through which the parties to a finding of negligence can be understood as doer and sufferer of the same injustice.
3. Right and correlative duty
Having situated the duty of care within the ensemble of concepts that allows negligence liability to be understood as a normatively coherent phenomenon, I want in this section to focus more specifically on the duty itself. What are its constituents and how do they give expression, in accordance with corrective justice, to the correlativity of right and duty? The answer to this question indicates the internal structure of the duty of care, thus revealing (in Lord Atkin’s words) “the general conception of relations giving rise to a duty of care, of which the particular cases found in the books are but instances.”25
Cardozo’s judgment denying liability in Palsgraf provides the most explicit judicial elucidation of the correlativity of right and duty in the negligence context. Cardozo observes that liability requires that “the plaintiff sues in her own right for a wrong personal to her, and not as the vicarious beneficiary of a breach of duty to another. A different conclusion would involve us, and swiftly too, in a maze of contradictions.”26 In Palsgraf the defendant arguably created an unreasonable risk to the third party’s package but was being sued for a wrongful infringement of the plaintiff’s right to bodily integrity. The right that the defendant unreasonably put at risk was therefore different from the right whose wrongful infringement was the basis of the plaintiff’s complaint. The duty to the person carrying the package was thus not correlative to the plaintiff’s right in her bodily integrity. Liability to the plaintiff would mean that the defendant was being held liable for the infringement of a right that its employee did not wrongfully imperil, so that the award of damages would then not represent the wrong that the defendant did. The contradiction is that under these circumstances liability to the plaintiff would be inconsistent with the nature of tort law as a mechanism for obligating defendants to make reparation for the rights that they have wrongfully injured.
As Cardozo points out, Palsgraf featured a particularly striking example of “the maze of contradictions.” Palsgraf was a case in which the defendant’s breach of duty and the plaintiff’s complaint involved a diversity both in the nature of the rights (property and personal integrity) and in the holders of the rights (the passenger carrying the package and the injured plaintiff). Cardozo correctly observes that the same contradiction would obtain even if the rights were of the same order but the person foreseeably endangered was different from the person actually injured.27 He also suggests, as Wagon Mound (No.1) later decided, that the contradiction would also be present if the discrepancy between the defendant’s breach and the plaintiff’s injury were to rights of different orders.28 In all these instances there would be no liability because the plaintiff would have been injured with respect to a right that was not the basis of the duty that the defendant breached.
When negligence law is conceived in terms of the correlativity of right and duty, the issue of the duty of care is composed of two constituents. First, the interest of the plaintiff that is protected against the defendant’s conduct must have the status of a right as against the defendant. Second, the duty breached must be correlative to that right. These two constituents are the complementary aspects of a unified conception of the duty of care, because a right both implies and is required by the correlative structure of liability. A right implies correlativity because a right always entails the existence of a corresponding duty. A right is required by correlativity, because (along with its corresponding duty) it is the only normative concept that has the correlative structure inherent in a regime of liability. Thus, the notions of right and correlative duty together form a unified general conception of the duty of care.
Cardozo’s treatment of duty in the Palsgraf case gives paradigmatic legal expression to this conception of the duty of care. For Cardozo, the principal issue presented by the case was that of correlativity. Because in that case the defendant’s conduct was not wrongful toward the plaintiff (although it was arguably wrongful toward someone else), the defendant was held not to be under a duty with respect to the plaintiff’s loss. Although Cardozo’s focus was on the absence of correlativity between the duty breached and the injury, Cardozo also insists that the duty breached has to be correlative to a right: “Negligence is not actionable unless it involves the invasion of a legally protected interest, the violation of a right.”29 Because “the commission of a wrong imports the violation of a right,”30 the plaintiff is precluded from recovering unless the defendant’s conduct is a wrong in relation to that right. Hence, “[w]hat the plaintiff must show is a wrong to herself, i.e., a violation of her own right.”31 That the plaintiff had a right to her bodily security was not disputed in the case, but the defendant’s action was not wrongful relative to that right. Thus, Cardozo affirms that the existence of a right in the plaintiff is presupposed in the requirement that the duty breached by the defendant be a wrong relative to her.
Accordingly, Cardozo’s opinion presents the two interrelated functions of the inquiry into the defendant’s duty. The first function is to establish whether the plaintiff’s damaged interest has the status of a right, because it is only to a right that the defendant’s duty can be correlative. The second function is to establish whether correlativity obtains in the case at hand—that is, whether the defendant breached a duty correlative to that right by creating an unreasonable risk to persons such as the plaintiff. When these two functions are brought together, the question of duty produces a structure of inquiry geared to ascertaining whether the parties can plausibly be regarded as the doer and sufferer of the same injustice. Action by the defendant that is incompatible with the plaintiff’s right marks the injustice of the defendant’s conduct, and that injustice is the same for both parties.