INTRODUCTION
A very simple account of tort law is as follows:
The simple picture: Tort law repairs wrongful losses. Wrongful losses may be defined as harms caused by a right-infringing action of another. The torts of negligence, trespass, and so on would then be understood as based on different varieties of rights infringement.
For the particular tort of negligence, then, we may suggest two principles that underlie the law in accordance with this picture:
Fault idea: We have a right that we not be exposed to certain types of risk of harm.
Causation idea: If one person infringes the right of another and thereby causes harm to the right-bearer, then the infringer is under a duty to compensate the right-bearer for the harm caused.
We call these principles fault and causation not because they define those concepts or explain them for us but because they show how the two pillars of negligence law—fault and causation—work in terms of rights, harm, and compensation. While clearly very rough, these two principles provide a reasonably neat explanation of negligence law.
Moreover, one could readily extend this account to other forms of tort. For instance, in lieu of fault we might suggest a “trespass idea,” to the effect that we have a right that others not intentionally interfere with our bodies in a direct manner. The conjunction of trespass and causation would give a simple picture of the tort of trespass.
This simple picture could no doubt be criticized on many grounds. Our aim is not to engage in such criticism but to attempt to support an alternative picture. We think the alternative has some advantages in explaining or justifying various ways in which we might adopt nonstandard compensation practices. Consider first, the sorts of nonstandard cases we are interested in:
Multiple fault, single cause: In some well known cases,1
See, e.g. Summers v. Tice, 199 P.2d 1 (1948); Cook v. Lewis, S. Ct. 830 (1951).
Probabilistic cause: It may happen that a party is exposed to a risk of an injury occurring and the injury subsequently occurs, but it remain unclear whether it was caused by the negligence of the plaintiff or by other, nontortious means.2
See, e.g. Hotson v. East Berkshire Area Health Authority, 2 W.L.R. 287 (1987), rev'd 1 A.C. 750 (1987).
In cases of these sorts, application of the standard picture would give rise to results that seem morally objectionable. Not surprisingly, courts have—it would appear—striven to avoid these objectionable outcomes while remaining consistent with the body of law.3
We have already mentioned Summers v. Tice, supra note 1. Another important attempt to deal with problems of determining which of a number of similarly faulty agents caused an injury is Sindell v. Abbott Laboratories 607 P.2d 924 (1980).
Incurable Cancer: Jones is visiting a hospital and is inadvertently exposed to some radiation by medical scientists conducting an experiment. The radiation increases Jones's chance of contracting an incurable, lethal cancer by a small amount. It will not be known whether or not Jones will contract this cancer for some time. If Jones does not contract cancer, there is every reason to believe she will live a long and happy life. If she does contract the cancer, there is every reason to believe her life will be greatly shortened.4
David McCarthy, Liability and Risk, 25 PHIL. & PUB. AFF. 247 (1996).
If the courts wait until it is discovered whether Jones has contracted the cancer, then any sum awarded will—very probably—be insufficient compensation for the harm she will suffer. What could compensate for imminent death? On the other hand, the risk of such an outcome from the radiation exposure is sufficiently small that Jones might have been, before the accident, willing to expose herself to the risk in return for some compensation. Thus compensation for the risk is in effect easier to achieve than compensation for the harm which might ensue.
The difficulties in dealing with such cases emerge, in particular, from (1) epistemic burdens of establishing who caused harm; and also (2) an intrinsic difficulty in compensating for some harms post facto, such as death.
A third difficulty may be said to afflict the simple picture. Note that we do not hope to establish conclusively that this third difficulty afflicts the simple picture, but we do hope to show why you might want at least to investigate alternatives. The third difficulty is that the simple picture seems to allow luck to infect the morality of negligence law. It seems prima facie arbitrary that two equally negligent parties who differ in their luckiness may be liable for vastly different amounts. If negligence is itself wrongdoing and the parties are equally negligent, then—it might be argued—the liability of the parties ought to be similar. Consider the two shooters in Summers v. Tice. Each of them negligently shot in the direction of the plaintiff. Only one shooter was lucky enough to miss, and surely that luck should be morally irrelevant. The simple picture entails, however, that only the unlucky negligent shooter should be liable.
Put thus, the objection clearly begs the question against the simple picture. It is not easy to make the objection non-question-begging, but nonetheless we think it has enough prima facie force, especially when combined with the first two difficulties we noted, to warrant considering alternatives.
Below we give a brief exposition of an alternative account, developed by David McCarthy, which explains our compensation practices in terms of an underlying norm that holds individuals liable for impositions of risk. McCarthy's specific ideas have not been much discussed, to our knowledge, by philosophers of law. Nonetheless, similar risk–liability proposals have been discussed by legal philosophers and have generally received quite harsh criticism. The principal aim of our paper, then, is to defend McCarthy's proposal from objections raised by legal philosophers to the very idea of explaining negligence law in terms of liability for risk.
As a piece of legal interpretation, our project is unambitious, for we do not claim that the risk–liability theory is a more plausible interpretation of negligence law than another, causation-based account. Our project is nonetheless significant as a form of moral reflection on the body of the law. One way such moral reflection could proceed would be to derive from fundamental moral axioms a prescription as to what legal institutions are acceptable, without any regard for actual legal practice. Such an approach would be a fairly pure form of political or moral philosophy. An attractive alternative, however, would resemble the attempt to obtain a form of reflective equilibrium, in Rawlsian fashion, between our central legal practices and an underlying moral theory.5
JOHN RAWLS, A THEORY OF JUSTICE 40–46 (rev. ed. 1999).
THE RISK–LIABILITY THEORY
McCarthy endorses something very like the first tenet of the simple picture. He calls it the risk thesis:
Risk thesis: We each have a right that other people not impose risks of harm upon us.6
David McCarthy, Rights, Explanation, and Risks, 107 ETHICS 208 (1997).
But McCarthy ties compensation far more tightly to rights infringement. Thus:
Compensation idea: Other things being equal, if one person infringes the right of another, then, very roughly, the infringer is under a duty to compensate the bearer of the right.7
Id. at 219.
In effect, McCarthy's compensation idea removes the element of causation from negligence law, and as such, it removes an aspect that brings in the morally arbitrary effects of luck.8
That is not to say that it removes moral luck entirely. Generally, we have more control over what risks we impose on others than over what effects we bring about on others, but we still do not have perfect control over what risks we impose, and we may thus still be unlucky in imposing a high risk of harm on someone. Thanks to Peter Cane for this point.
This proposal appears bizarre, because we surely do not compensate every risk imposition. We do not even begin to think of seeking compensation for mere negligence but almost always only for actual harm suffered. Courts vary the requirement that the plaintiff bears the burden of proof for establishing causation only in rare cases such as Summers and Sindell. And they never—to our knowledge—remove the need to establish causation altogether. Nonetheless, McCarthy believes his account is roughly compatible with the conventional practice of awarding damages only where harm actually occurs. In order to see the compatibility of McCarthy's proposal with our usual practices, we need to consider what is required to compensate someone for an exposure to risk.
When an agent is exposed to a risk, she is made worse off by bearing an expected harm or made prospectively worse off. Thus such an agent must be compensated with a benefit that will make her prospectively as well off as she would have been had she not been exposed to the risk.9
McCarthy, supra note 6, at 219–220.
Direct payment rule: If X imposes a risk of harm on Y, then X is under a duty to compensate Y directly for the expected cost of the risky behavior.10
Cf. McCarthy, supra note 4, at 250. We omit the qualification McCarthy adopts in his compensation rules: that liability is incurred only if X knows or ought to know that her action will impose a risk of harm on Y. For the purposes of this paper, we remain agnostic as to whether such a qualification is required. See JUDITH JARVIS THOMSON, THE REALM OF RIGHTS 229–234 (1990), for related discussion.
So, for example, an agent exposed to a one-in-ten risk of a one-hundred-dollar loss would be paid ten dollars compensation, thereby making her prospectively as well-off as if she had never borne the risk of the hundred-dollar loss.11
McCarthy, supra note 6, at 220.
McCarthy stresses, however, that we need not think of compensation as always being an actual transfer of goods. If we are seeking to compensate mere risks of harm, we may compensate by offering merely potential transfers of goods. For example, we might compensate a risk-exposed individual by providing a guarantee to be compensated for the harm she suffers if she suffers harm as a result of bearing the risk. McCarthy calls this the “Natural Lottery rule” for compensation.12
Id. See also David Lewis, The Punishment that Leaves Something to Chance, 18 PHIL. & PUB. AFF. 53 (1989).
Natural lottery rule: If X imposes a risk of harm on Y, then if Y is harmed as a result, X is under a duty to compensate Y for the harm suffered.
Perhaps surprisingly, the natural lottery method compensates victims fully for bearing the risk of harm. Consider the possible outcomes of an imposition of risk. An agent might not suffer any harm as a result of the risk imposition, in which case nothing further is required by the rule. Alternatively, an agent might suffer harm as a result of being exposed to a risk of harm, in which case the rule directs that the agent be fully compensated for suffering that harm. Because of this guarantee of full compensation for possible harm, the natural lottery rule ensures that agents are made prospectively no worse off for having been exposed to risk. While it appears that under the natural lottery rule, an agent exposed to a risk of harm that does not manifest as an actual harm is not compensated, this is incorrect. Such an agent is fully compensated for bearing such a risk by receiving the guarantee that should harm eventuate as a result of the risk exposure, he will be fully compensated for that harm. Both the natural lottery and the direct payment rules, therefore, are legitimate means of compensating risk impositions.
As is no doubt obvious, implementing a natural lottery rule is superficially identical to the standard practice of negligence law. No actual compensatory payments are made unless harm results. But that does not establish that McCarthy's account in terms of the risk thesis and compensation idea are merely a more complicated rendition of the simple picture introduced above. Rather, McCarthy's account has a welcome degree of flexibility as to which liability rule may be used to effect the underlying norms. In addition to the natural lottery rule, a direct payments rule may be used. This is precisely the sort of rule that seems most appropriate for cases such as the one described in Incurable Cancer above.13
McCarthy, supra note 4, at 247–249.
Moreover, other liability rules are possible. One suggested by McCarthy is something like a generalization of the Summers ruling:
Risk proportionality rule: If X performs an action that will impose a risk of harm on Y, then if Y suffers that harm, X is under a duty to pay Y a share of what is required for full compensation for the harm in proportion to her contribution to the total risk of that harm that Y bore.
There are, as it happens, some very good practical reasons for not using these alternative liability rules in the majority of cases. Both the direct payments and risk proportionality rules require information about what degree of risk individual parties imposed on others. This is extremely difficult, in many cases, to obtain. The natural lottery rule, in contrast, only requires information about actual injuries and who caused them. Thus McCarthy suggests that there are reasons of fairness for adopting a natural lottery rule as our typical practice. But in general, McCarthy suggests, we may choose the liability rule that is fairest for the type of activity in question.14
Id. at 250–251.
Finally, consider a rule of our own devising: the hybrid rule holds individuals liable for harm in the same way as the direct payments rule. That is, for every risk of harm that an individual imposes, a corresponding liability arises for that risk. Unlike the direct payments rule, however, the hybrid rule does not require that the compensation be paid directly to the individuals who have been exposed to risk. Rather, it directs that compensation be paid in accordance with the natural lottery rule: to those who have been harmed.15
Breaking the nexus between liability and compensation in this fashion is most frequently associated with JULES L. COLEMAN, RISKS AND WRONGS 285–288 (1992).
Hybrid rule: If X performs an action that imposes a risk of harm on Y, then she is under a duty to make a contribution, equal to the expected harm arising from the action, to the compensation of anyone who has actually suffered a harm as a result of risks taken by others.
To illustrate: suppose there is a group of ten drivers, driving for a set period of time. During that time, each driver imposes a 5 percent risk of injury on each of the others. For simplicity, suppose that there is only one possible type of injury and that, if such an injury occurred, it would require two hundred dollars to be fully compensated. Each driver would be liable for nine payments of ten dollars: one for each of the drivers on whom she imposes a risk. Those liabilities would only be executed, however, if one of the drivers is actually injured. In that case, the injured driver is entitled to receive the full two hundred dollars, and it is a matter of administrative choice how that amount is to be derived from the liabilities of all the drivers. One could institutionalize such a scheme so as to resemble the risk-pooling arrangement advocated by Christopher Schroeder.16
Christopher H. Schroeder, Corrective Justice and Liability for Increasing Risks, 37 UCLA L. REV. 439 (1990).
Given the variety of liability rules that are compatible with the risk–liability theory, then, McCarthy has shown us that an apparently radical thesis about rights and compensation can nonetheless have moderate consequences for our institutions of liability and compensation. The risk–liability theory is not entirely conservative, however, for it suggests ways in which we might reform our compensation practices and achieve fairer outcomes in cases of probabilistic causation, multiple possible causes, and cases where it is difficult to compensate for harms after they occur.
At this point, something needs to be said about the “level” at which decisions about which liability rule is appropriate ought be made. It might be thought we are proposing that the decision of which liability rule to apply in a given case be left to a judge hearing an individual case. Decisions about the appropriateness of certain liability rules in different spheres of activity, however, need not be made on a case-by-case basis. Instead, it might be deemed appropriate to institutionalize a particular liability rule for a particular sphere of activity. So, for example, we might choose to institutionalize the risk proportionality rule for all motor vehicle accidents involving multiple vehicles while using a natural lottery rule for all single-vehicle accidents. Moreover, not only might different rules be appropriate for different spheres of activity, but it also might be appropriate to have the determination of the appropriate rule carried out at different levels for different spheres of activity. Thus, for example, it might be considered appropriate to have an institutionalized natural lottery rule for motor vehicle accidents but better to leave the choice of liability rule to the courts for all cases involving toxic agents such as the radiation in Incurable Cancer.
We remain agnostic as to precisely how liability rules are best institutionalized and which rules are appropriate for particular spheres of activity, and merely wish to stress that there is a range of possibilities available. A particularly attractive feature of the risk–liability theory, as compared to other accounts of corrective justice, is that it retains a tight connection between rights infringements and duties of compensation (as embodied in the compensation idea), but by removing the requirement of causation is compatible with a wider range of institutional practices.17
Cf. Richard W. Wright, Right, Justice, and Tort Law, in PHILOSOPHICAL FOUNDATIONS OF TORT LAW 176–180 (David G. Owen ed., 1995). Wright holds that the correlativity of rights and duties of compensation rules out risk-pooling schemes as failing to satisfy corrective justice.
The Meaning of Risk Imposition
One point needing clarification is precisely what it means to impose a risk on another. An activity that one undertakes in isolation might involve very little risk to others, but if the activity is undertaken in conjunction with certain other factors, the risk might escalate. Consider the example of a hang glider who plans to land in a field. Provided no one walks in the field, the risk to others is negligible. But if someone is walking in the field, the risk is very significant indeed. Should we consider this to be a case where the hang glider imposes a risk on walkers, or a case where walkers voluntarily impose the risk of injury on themselves? In effect, it is a question of where to set the baseline for questions of risk. Do we ask: What is the additional risk of hang gliding, given that walking happens? Or rather: What is the additional risk of walking, given that hang gliding happens?
Negligence law treats this issue by means of the conceptual apparatus of a duty of care. Questions of who, if anyone, has a duty of care to the other and of whether this duty has been breached achieve the same result as the risk–liability theory attempts to achieve by inquiring into whether a risk has been imposed. If we wish to develop the risk–liability theory in a manner so as to minimize any revisionary consequences for negligence law, we ought simply to define risk imposition in such a way as to make it equivalent to breach of duty. Effectively, the negligence standard of care would translate into the following account of risk imposition:
- If no-one was behaving carelessly, then no risk is thought to have been imposed on anyone. This is the baseline from which further comparisons are made.
- If the victim was not behaving carelessly, but others were, then all additional risk is deemed to be imposed by the careless agents.
- If the victim was behaving carelessly, some division of the additional risk between the victim and other careless agents is called for.18
The third step of this risk-determination procedure is relatively indeterminate, and this is because there is a variety of ways of treating negligence on the part of the plaintiff in current law. This complication is discussed further below.
Clearly, the first step is something of a legal fiction: all behavior is accompanied by risks in some sense. A crucial question for the advocate of the risk–liability theory is whether there is a coherent, principled conception of risk that correlates with the legal distinction between careful and careless behavior.
Is Negligence Only about Risk?
An objection to our account might be that negligence law is not solely concerned with behavior that imposes risks upon others. Sometimes it is concerned with deviations from conventional norms of practice. Behavior that deviates from such norms is held to be a breach of the duty of care without any particular regard to the question of risk.19
This objection was made by an anonymous referee.
It might have to be conceded that this is an aspect of negligence law that fits the risk–liability theory poorly. However, we are optimistic that even where the law directly addresses only the question: “Was the behavior in conformity with a conventional standard?” it is nonetheless the case that the reason this question is asked is because the relevant community is assumed to have standards that work to minimize risk. Thus, if a defendant was observed to deviate from standard practice and in so doing caused an injury, if the defendant could succeed in showing that the deviation in fact reduced the risk to the plaintiff, this would make at least a morally compelling case for a judge to find for the defendant.
Moreover, taken not as a strict interpretation but as a standard by which to judge the institution of negligence law, the risk–liability theory can provide an attractive basis from which to judge our norms of care. Perhaps our current norms do not track risk closely enough. One provocative example is Douglas Husak's suggestion, for instance, that driving a relatively “crash-incompatible” vehicle—such as an SUV (sports utility vehicle)—without compelling reason poses an unacceptably large, asymmetric risk to others.20
Douglas Husak, Vehicles and Crashes: Why Is This Moral Issue Overlooked? 30 SOC. THEORY & PRAC. 361 (2004).
Multiple Tortfeasors and Negligence on the Part of the Plaintiff
The law of negligence has a number of different strategies for distributing liability between multiple tortfeasors depending upon whether they acted in concert, whether they each contributed to an injury which is divisible, and other factors. Negligence law also takes into account negligent behavior by the plaintiff in a number of ways, sometimes with the drastic effect of entirely removing the plaintiff's ability to recover. How does the risk–liability theory handle these features of the law?
Consider, first, what the natural lottery rule would dictate with respect to a case where two parties act negligently, and their doing so contributes causally to the very same injury. The natural lottery rule suggests that each negligent party is under a duty to compensate the victim for the full extent of harm suffered. This mirrors the practice of negligence law, where multiple tortfeasors may be held jointly and severally liable for an injury that is the result of concerted activity or that may not be divided.21
Note that at law a plaintiff may not recover more than the full amount of damages suffered, even where multiple parties are entirely liable for the injury. As we read it, the natural lottery rule is silent on this matter and may be plausibly read to be compatible with this feature of the law. For instance, suppose two parties A and B are each, under the natural lottery rule, under a duty to compensate Y for an injury suffered. Suppose A compensates Y for the injury. Does it follow that B still has a duty to Y to compensate for the injury? Arguably not: the duty has been discharged, albeit not by B.
Second, consider an injury or injuries where it may make sense to divide the liability in some fashion. Two hunters may shoot negligently at the same person at the same time, but each injures a different limb, perhaps. In such cases, courts tend to apportion the damages between the tortfeasors. And this is happily compatible with the risk proportionality rule, or something very like it.
The most troublesome aspect of negligence law to reconcile with the risk–liability theory is the defense of contributory negligence on the part of the plaintiff. Where contributory negligence is permitted as a defense, any negligence on the part of the plaintiff, of whatever degree, is sufficient to prevent recovery. Effectively, this defense applies a natural lottery rule to the conduct of plaintiff—holding the plaintiff “liable” for an injury that arises from a self-imposed risk. Not only does it so hold the plaintiff responsible but it proscribes holding anyone else liable, even where the activities of others did contribute to the risk.
This aspect of negligence law is relatively risk-insensitive. It makes no comparisons of degree of risk and it seems to be accommodated by the risk–liability theory only with the addition of norms to the effect that an individual's duty to avoid risk-related harms to him- or herself is of some higher order than the corresponding duty of third parties.
A defender of the risk–liability theory, then, would welcome the moves to limit the defense of contributory negligence that have occurred in the last century. The apportionment of responsibility and consequent reduction of damages, as occurs now in a number of jurisdictions, evidently reflects norms compatible with a risk proportionality rule and is a far less awkward fit with the risk–liability theory.
TWO OBJECTIONS TO THE RISK–LIABILITY THEORY
Ripstein's Critique
Arthur Ripstein rejects the risk–liability theory.22
See ARTHUR RIPSTEIN, EQUALITY, RESPONSIBILITY, AND THE LAW §3.4 (1999).
[A]ny [risk–liability] proposal escapes the supposed arbitrariness of chance by surrendering its ability to explain why carelessness should be the basis of liability or why the plaintiff's loss is the measure of damages.23
Id. at 75. An anonymous referee has pointed out to us that we are addressing only a fragment of Ripstein's critique. Ripstein's broader point against risk–liability proposals is that they adopt a basis for liability that is at odds with that used in all other areas of law. It is thus unprincipled, Ripstein might suggest, to adopt a deviant basis for liability in this area of law alone. We do not respond to this criticism in the current paper, as it crucially turns on claims about the moral basis for liability in other areas of law, which claims we cannot satisfactorily address in a single paper.
First, we note that it is only an unsophisticated risk–liability proposal that cannot explain why carelessness is the basis of liability. As discussed above, the term “imposition of risk” conceals a difficult question of how to divide the risk created by multiple parties acting simultaneously. We need some standard “baseline” of activity to which we can appeal in such cases. But there is no reason to suppose that these issues of the baseline are any more difficult for the advocate of the simple picture than they are for the advocate of a risk–liability proposal. As we suggested, if desirable, the risk–liability theorist can simply define imposition of risk in a way that renders it identical in content to the negligence lawyer's concept of breach of duty.
Second, we have shown why the victim's actual harms should—at least sometimes—determine the amount of compensation required. If the risk–liability theory is true, it may be just to adopt an institution that conforms to the natural lottery rule, because being guaranteed full compensation for actual harm, if it occurs, just is full compensation for exposure to a risk of harm.
We do not insist that the plaintiff's loss must be the measure of the damages. If we adopt a direct payments rule, for instance, the relevant measure of the damages would be the expected loss. But this is precisely part of the interest and attraction of the risk–liability theory: it explains our current institutions while also clarifying the relevant ways in which we might reform those institutions.
Ripstein's project, perhaps, is not the same as ours. In places, he suggests that his aim is to offer an interpretation of our actual legal practices.24
Id. at 1–2.
In places, however, Ripstein does imply not only that there is something defective at law with the risk–liability theory but that the risk–liability theory could not be part of any morally acceptable system of rules. For instance, he discusses circumstances in which a compulsory insurance scheme might be acceptable but insists that it would not be in virtue of the risk–liability theory or any similar moral explanation. Id. at 83–84.
Even for one who wishes merely to give an interpretation of existing law, denying the risk–liability theory is problematic, for there are cases, such as those we mentioned above, where the courts have deviated—or at least flirted with deviance—from the simple picture. Ripstein responds by suggesting, “even in such cases, the harm is not in the risk, but in the loss of a chance.”26
Id. at 76.
Such a proposal is rightly criticized by THOMSON, supra note 10, at 243–245.
It can then be asked: In what important sense is the loss of a chance different from being exposed to a risk? We suggest that the former has the superficial attraction of making it sound as though an actual injury-like event has occurred: a “loss.” But it seems quite easy to translate from one manner of speaking to the other without having distorted the essential nature of the event. For instance, if you burn my lottery ticket, there is a sense in which I have suffered the loss of a chance of winning the jackpot. Alternatively, we may say I have been exposed to the risk that I will fail to obtain something I would otherwise have—the jackpot.
Consider an elaboration on the Incurable Cancer case:28
McCarthy, supra note 4, at 247–248.
Clearly Ripstein thinks that something more significant turns on the distinction between exposure to risk and loss of a chance, as in the following passage:
These risk-related injuries are not injuries that are the result of the creation of a risk, but rather are injuries that consist in something that can only be valued in terms of chances. As such, they pose no special problems, and do not involve liability for the imposition of risks.29
RIPSTEIN, supra note 22, at 77.
But, having no predisposition to reject a risk–liability proposal, we confess an inability to understand why these “injuries” that “can only be valued in terms of chances” are not to be understood as impositions of risk.
When Ripstein discusses cases involving probabilistic causation—such as Sindell—he correctly observes that courts do not remove the requirement of causation altogether. Rather, they vary the normal procedural requirements for proving causation. Instead of permitting a number of faulty parties to benefit from their contemporaneous lack of care, the plaintiff may simply establish that liability is to be shouldered in some fashion among the parties at fault, and leave the burden of establishing precisely which faulty party was the cause of the injury to the faulty parties. Thus the phenomenon to be explained is not the phenomenon of compensation for mere exposure to risk, it is rather the shifting of some of the evidential burdens onto wrongdoers.
As a matter of interpretation of the law, Ripstein is on strong ground here. It is precisely for reasons such as this that the simple picture most probably affords a superior interpretation of actual legal practice. As it stands, however, the law will not be able to deal with cases like that in “Incurable Cancer” by a mere shifting of the burden of proof. In that case, it is not yet established whether any harm will be caused at all: Jones may not contract cancer. If you endorse our judgment that some actual compensation may be called for in the case of Jones, then you have some reason to think the risk–liability theory a morally preferable theory, and you also have a reason to criticize the law's commitment to causation as a basis for liability.
While Ripstein's argument may support a causation-based interpretation of the law, then, he has not undermined the weaker claim that the risk–liability theory might provide a normative basis for negligence law.
A Line of Objection Inspired by Perry
Stephen Perry has argued that, at law, it is not plausible to hold that mere risk can be a form of harm or damage.30
We use the terms interchangeably here.
Perry himself deliberately abstains from making this argument.31
Stephen R. Perry, Risk, Harm, and Responsibility, in PHILOSOPHICAL FOUNDATIONS OF TORT LAW 338 (David G. Owen ed., 1995).
Perry's argument against risk as a form of harm turns on an explication of two different conceptions of risk or probability. Epistemic probability is, roughly, the degree to which a rational agent should believe some proposition is true, given a body of evidence. Objective probability, whether it is cashed out in terms of relative frequencies, propensities, or some other phenomenon, is what our epistemic probabilities ought to approximate and is independent of the available evidence.
Perry asks what sort of probability is relevant to putative risk-as-harm cases. Is it in virtue of the epistemic probability of an action causing harm or in virtue of the objective probability of its causing harm that we might seek compensation for the actions of another?32
Id. at 332.
Id. at 334–335.
Consequently, Perry finds objective risk to be the more plausible candidate as a form of risk that could itself be a harm. Grant that this conclusion is correct. What does it mean for an act to be an objective risk imposition? On the relative-frequency account of probability, it means that the act is a member of a class of intrinsically similar acts, such that a certain proportion of the acts in that class cause harms. (We follow Perry in assuming a relative-frequency interpretation. An analogous argument can be made on alternative interpretations of objective probability.)
Perry then considers the possibility that determinism is true. If determinism is true, then, for any risk imposition, that action will be a member of a more fine-grained class of intrinsically similar actions such that the objective probability of a member of that class causing harm is zero or one. Admittedly, we may not have epistemic access to what characterizes this class, but such limitations of our epistemic perspective cannot be relevant to determining whether harm occurred, claims Perry.34
Id. at 334.
What then is the correct classification of actions? Perry suggests that this is fundamentally indeterminate, and we shall generally select the most fine-grained class available given our epistemic situation:
[T]here is nothing magical about the particular reference class selected: generally it will simply be the narrowest class, given the current state of our knowledge, for which we are able to determine with some degree of accuracy the relative frequency of the type of harm in question.35
Id. at 336.
We either have no determinate answer, then, to the question of what objective risk was imposed by a given action; or, if we settle for taking our narrowest available classification of actions as determining the objective risk, then our judgment is vulnerable to future empirical discoveries. This seems odd if we are to construe risk as a form of harm, for whether or not a harm occurred should surely be a matter that is independent of the extent of our empirical knowledge. It was for this sort of reason that the epistemic conception of risk also seemed inappropriate as the basis for a form of harm also.
These results are not quite so absurd as to show that the risk-as-harm view is incoherent. As a matter of interpreting actual negligence law, however, Perry's view is indeed far more plausible than the alternative. How, then, to reconcile Perry's claim about negligence law with our own view that negligence law is compatible with a risk–liability proposal? We suggest the most plausible means to do so is by rejecting the very premise that Perry is chary of adopting: the claim that negligence law awards compensation only for harms. As has been evident throughout, the risk–liability theory requires only that the imposition of risks be wrongful. It is not necessary that risk imposition be itself a harm.
That said, a worry remains. Having agreed that risk is not itself a harm, we are still maintaining that the imposition of a risk is—at least sometimes—a wrong. In that case, one might be tempted to mount a Perry-style argument against the claim that risk imposition is a wrong. If it is implausible to suggest that mere epistemic risk is itself a harm, then would it not be similarly implausible to suggest that imposing mere epistemic risk is a wrong? And if the relevant species of risk is objective, then we are again subject to the problem of selecting the relevant reference class. Indeed, if determinism turns out to be true, then the risk–liability thesis would be, at best, redundant, because the only salient probabilities of harm will be zero and one—in which case risks of harm are coextensive with actual harms.
We follow Perry's lead here in thinking that the relevant conception of risk for the purposes of ascribing responsibility and hence for designating actions as wrongful is epistemic, for it is—at least to some extent—sensitive to the evidence available to the agent. An accident that is—relative to current evidence—unforeseeable is not an accident for which one can be held liable in negligence law. The very same type of accident in the future, however, might be cause for a successful action because knowledge of risks may have advanced: such an accident may now be a known risk.
It is Important that epistemic probability not be confused with subjective probability. The considerations regarding the limited capacities of agents mentioned above do not mean that we should be concerned only with the probability that an agent actually predicted, for clearly an agent could make grossly inappropriate probability calculations, and that would be no obstacle to holding the agent liable. Rather, it concerns the probability of harm that it would be reasonable to expect an agent to calculate, given certain intersubjective standards of inductive reasoning.36
Id. at 325–326.
It has been noted elsewhere that most modern moral theories adopt what might be termed the ex ante view.37
Schroeder, supra note 16, at 5 ff.
Id. at 5.
The law reflects this sort of view and embodies a classification of actions that discriminates as finely as possible, assuming a certain minimum level of ability and knowledge on the part of the risk imposer. Relative to that reasonableness standard, the foreseeable consequences of the Russian-roulette player's action were no different from any other trigger-pulling and thus equally negligent. Even if we later made the discovery that there were physical circumstances that determined that the player's trigger-pulling would not lead to harm, thus leading us to conclude that there was no objective risk, we would not be likely to revise our attribution of wrongfulness, for the player was surely not in possession of such information at the time.
Thus the implausibility of holding epistemic risk to be a form of harm does not infect the claim that epistemic risk is a basis of wrongful action, and no such argument against the risk–liability theory can be derived from Perry's position.
CONCLUSION
So two important lines of attack on the risk–liability theory have been rebutted.
Ripstein's claim that any liability for risk proposal cannot explain why carelessness is the appropriate basis for attributing responsibility has been shown to be true only on a straw-man version of the risk–liability theory. His claim that it cannot explain why actual harm is the measure of damages has been shown to be patently false.
Perry's argument against the identification of risk as itself a form of harm has been shown to be no direct threat to the risk–liability. In order to obtain an argument from Perry's conclusion, a further premise is required to the effect that all compensation must be for actual harm, and it is a defining feature of the risk–liability that it rejects this proposition.
Moreover, Perry's argument against an epistemic conception of risk as a form of harm has been found not to generalize into an argument against epistemic risk as the basis of wrongdoing. Thus the risk thesis remains defensible on an epistemic conception of risk and thereby fits with the way questions of foreseeability are dealt with in the courts.
We thus recommend the risk–liability theory to the community of legal scholars as a fruitful way of explaining the central features of negligence law while providing a morally attractive rationale for further development and reform of that law.
