Crim Law and Philos DOI 10.1007/s11572-017-9428-7 ORIGINAL PAPER
Is There a Case for Strict Liability? Larry Alexander1
Ó Springer Science+Business Media B.V. 2017
Abstract In this short paper, I shall answer the title’s question first in the context of criminal law and then in the context of tort law. In that latter section, I shall also mention in passing contractual and other forms of civil liability that are strict, although they will not be my principal focus. My conclusions will be that strict liability is never proper as the basis for retributive punishment; that it is a very crude device for achieving deterrence through nonretributive penalties; and that with respect to tort liability, it is best justified as a means of defining insurance categories. Keywords Strict liability Retributive punishment Tort Negligence Deterrence Burden of proof
1 A Down and Dirty Definition From here on, when I refer to ‘‘strict liability,’’ its meaning will be liability—criminal or civil—for which culpability is immaterial. In other words, when a defendant is deemed strictly liable in criminal law or in torts, he may be punished, penalized, or forced to pay compensatory damages because of some act of his even without proof that he committed that act culpably. When has a defendant committed an act culpably? The orthodox answer is he has acted culpably when either he intended the offending result; he believed he would produce that result; or he believed he was taking what was an unjustified risk he would produce that Warren Distinguished Professor, University of San Diego School of Law. I want to thank the participants in the conference on Crime Without Fault at Georgetown University Law Center and especially John Hasnas for organizing it and inviting me. & Larry Alexander
[email protected] 1
University of San Diego School of Law, San Diego, CA, USA
123
Crim Law and Philos
result. The labels for these three types of culpability are ‘‘purpose,’’ ‘‘knowledge,’’ and ‘‘recklessness.’’ I believe that these three types of culpability are really a single, unified form of culpability: insufficient concern for the legally protected interests of others.1 It is easy to see that knowledge and recklessness are really just points on a single continuum of awareness of risk. But risk-imposition is only culpable when the defendant perceives no justifying reason for doing so, which is where purpose enters the culpability formula. Orthodoxy also has it that there is a fourth type of culpability—negligence. Negligence consists of imposing unjustified risks of harm without being aware one is doing so (i.e., inadvertently) when one should have been aware one is doing so. I, however, reject the orthodoxy on this point. For I consider negligence (as opposed to recklessness) to be nonculpable and thus negligence liability to be a form of strict liability.2 The claim that an actor should have been aware of the riskiness of his conduct when he is not in fact adverting to the riskiness of it cannot mean he has the occurent ability to advert to that to which he is not adverting. It usually can be translated into: he would have adverted to it if he had taken some prior act, if he had a different character, or some other such counterfactual.3 But, as culpability attaches to acts or omissions, not to one’s character, and as the upstream acts that are alleged to have caused the downstream negligent act are either not culpable, or if culpable, themselves rather than the negligent act the proper focus of the culpability claim, there is no basis for deeming negligent acts culpable.4 To sum up, culpable acts are those committed with purpose, knowledge, or recklessness. If one is deemed liable for an act even if one has not been proved to have committed it with purpose, knowledge, or recklessness, then one has been subjected to strict liability.
2 Strict Liability in the Criminal Law If one is a retributivist with respect to punishment for crimes, as I am, then one believes that to be justly punished, the defendant must have committed the criminal act culpably. Without culpability, there is no retributive desert. And in the absence of retributive desert, punishment is unjust. A retributivist, therefore, should oppose all forms of strict liability in the criminal law. I see no way to avoid that conclusion. Suppose, however, one argues that, when the strict liability element has been proven to exist, the likelihood that the element was produced culpably is so high that we might as well conclusively presume defendant was culpable. In other words, even if we do not require culpability with respect to this element—i.e., we treat it as a matter of strict liability—we can rest assured that defendant almost without a doubt was culpable with respect to that element. 1
I first argued for this unified conception of culpability in Larry Alexander, ‘‘Insufficient Concern: A Unified Conception of Criminal culpability,’’ 88 CAL. L. REV. 931 (2000). I then reiterated and elaborated on it in Larry Alexander, ‘‘Culpability,’’ in The Oxford Handbook of Philosophy (J. Deigh et al., eds., 2001): pp. 218–38, 227–30 and in Larry Alexander and Kimberly Kessler Ferzan, Crime and Culpability: A Theory of Criminal Law (2009), ch. 2.
2
See Alexander and Ferzan, supra note 1, at ch. 3.
3
See Heidi M. Hurd, ‘‘Finding No Fault with Negligence,’’ in Philosophical Foundations of the Law of Torts (J. Oberdiek, ed., 2014): pp. 387–405. 4
See Alexander and Ferzan, supra note 1, at ch. 3; Larry Alexander and Kimberly Kessler Ferzan, ‘‘Confused Culpability, Contrived Causation, and the Collapse of Tort Theory,’’ in Philosophical Foundations of the Law of Torts, supra note 3, pp. 406–25, 408–15.
123
Crim Law and Philos
The problem with this rationale for strict liability elements in crimes should be obvious. Assume the standard of proof the prosecutor must satisfy is, as the U.S. Supreme Court declared in In Re Winship, proof beyond a reasonable doubt (BRD).5 Now, if the likelihood of culpability given the strict liability element is 100%, then proof of that element BRD will also be proof of culpability BRD. In that case, the strict liability element is harmless— it does not lead to punishment of the nonculpable—but it is also superfluous. For even if the element required culpability, that culpability could be proved BRD merely by proving the element BRD. The more general point is that if the strict liability element correlates highly with culpability, proof of that element BRD goes a long way towards proving culpability BRD. There is no need or justification for a conclusive presumption of culpability based on proof of the strict liability element BRD. What if one argues for strict liability on the ground that proving culpability BRD with respect to certain elements is just too difficult? For example, some argue in sexual assault cases for strict liability on the element of nonconsent. That is, if it is proved that the defendant had sex with a person who did not consent, it should be immaterial that the defendant believed consent had been given. The reason would be that it is just too easy for the defendant to raise a reasonable doubt if the prosecutor has to prove BRD that the defendant did not believe consent was given. It is true that proving culpable mental states BRD is often quite difficult. Perhaps the Supreme Court should not have required proof BRD on every element of a crime. But the quarrel then should be with the standard of proof. Until that standard is changed, and no matter what the standard should be, the standard should not be circumvented by means of strict liability elements. What if one switches the focus from retributive punishments, which require culpability, to deterrent penalties? If we look at criminal sanctions as costs attached to various kinds of conduct in order to deter people from engaging in that conduct, does strict liability fare better? Not really. First, any serious penalty—prison, jail, or a large fine—imposed on someone who does not deserve that penalty, is quite unjust. So let us assume we are dealing with minor offenses that are penalized with small fines. And the rationale for those fines is not retribution but deterrence. Second, consider how deterrence of the strictly liable works? The argument from deterrence works best if we assume a culpable defendant—one who is aware he is committing or highly likely to be committing the actus reus—and is deterred by even a relatively light penalty because of the high likelihood of conviction. The high likelihood of conviction is produced by requiring only proof BRD of the actus reus and not requiring proof BRD of mens rea. But how does deterrence work if the defendant is not culpable? Those who are strictly liable may not believe they are committing the actus reus that contains the strict liability element. Moreover, they may not believe they are taking an unjustifiable risk of committing that actus reus. So, at the moment they do commit the actus reus, the penalty attached to doing so cannot guide and thus deter them. Deterrence, therefore, must operate earlier in time than at the moment of committing the actus reus. The prospect of the penalty must lead actors to take measures to reduce the chance that they will commit the strict liability offense unaware that they are doing so. And often those measures will be personally and socially costly.
5
In re Winship, 397 U.S. 358 (1970).
123
Crim Law and Philos
For example, suppose a state makes it a crime, punishable, we’ll suppose, by a small fine, for any man over 18 to have sex with a woman younger than 16. And suppose it makes the age of the woman a strict liability element. No matter how reasonable it may be for a male defendant to believe his female sexual partner is 16 years of age or older, if she is not, he will be fined. To deter more than requiring culpability would deter, males over 18 will have to avoid sex with any woman who could possibly be under 16, no matter how mature she looks, no matter what she says, no matter what her driver’s license says, and so on. Depending on your views of sex among the young, this may or may not count as a significant social or personal cost. So consider another example, that of sale of a mislabeled drug. Suppose that conduct is a strict liability offense. No matter how careful the defendant is to avoid the sale of a mislabeled drug, if he does it he will be penalized. And if he is risk averse, then, because reasonable care will not negate liability, he will either engage in supra-reasonable care or else quit the business. Those measures, however, are socially costly. Supra-reasonable care raises the price of drugs, which may render them unavailable to those who need them. And quitting the business results in fewer and less risk averse competitors, with adverse social consequences. These points can be put in terms of the ‘‘Hand formula’’ from Carroll Towing.6 That formula is usually put forward as a formula for negligence, but it is really a formula for recklessness, as it assumes an actor adverting to the costs, benefits, and probabilities but who chooses to defy the verdict of his calculations. My point here, however, is that strict criminal liability deters only if the actor avoids acts even when the costs of avoiding those acts dictates that they not be avoided. Strict liability deterrence is both prudentially and socially costly. Finally, some strict liability crimes wreak havoc with the voluntary act requirement. Take the example of someone who has an epileptic seizure while driving, with the result that his out-of-control car violates several traffic laws. If those laws are strict liability crimes, then can the defendant be found guilty of their violation given that he was unconscious at the times of violation? Now, if the crimes required culpability, we could answer that question easily. For we would look at the defendant’s conduct prior to his seizure and ask whether at any such time he had the purpose to violate the traffic laws, knew he would do so, or knew he was taking an unjustifiable risk of doing do. If, for example, like Mr. Decina, he knew he was subject to such seizures but drove without taking his anti-seizure medicine, we could deem him guilty of violating the traffic laws because he culpably risked doing so when he began to drive without having taken his meds.7 And that act, as will be true of all culpable acts, was a voluntary act. If, however, the violations do not require proof of a culpable act, then we have a dilemma. If the voluntary act is merely the driving prior to the seizure, then the voluntary act requirement will be trivial. For there will always be voluntary acts that precede the involuntary actus reus. The alternative would be to look at only the actus reus and deem it to be involuntary. But, as we see with Decina-like examples, an involuntary actus reus can be consistent with the voluntary act requirement. The problem is, if we are looking for a culpable act, that act will always be voluntary. If, however, we have a strict liability offense, we are not looking for a culpable act. Therefore, we do not know whether to look 6
United States v. Carroll Towing Co., 159 F 2d 169 (US Ct of Apps, 2d Cir., 1947).
7
People v. Decina, 2 N. Y. 2d, 157 N. Y. S. 2d 558, 138 N. E. 2d 799 (1956).
123
Crim Law and Philos
prior to the involuntary actus reus, where we can always find voluntary acts, or to look at the time of the actus reus, which may be involuntary conduct.8
3 Strict Liability in Tort Law Tort law orthodoxy has it that there are torts that require that the defendant acted negligently and torts that are matters of strict liability. I have already argued that negligence— failure to advert to an unjustified risk to which the so-called ‘‘reasonable person’’ would have adverted—is not culpable, although it may have been the product of a prior culpable act (one involving consciousness of the risk of future negligence). So, if negligence is not culpable, is it really just a form of strict liability? I think the answer is clearly ‘‘yes.’’ The analysis I will now present I have presented before.9 I believe, however, that my prior analysis merits repeating. I have elsewhere argued that the distinction between negligence and strict liability cannot be maintained because there is no non-arbitrary way to construct the notion of the ‘‘reasonable person’’ who would have adverted to the risk created by his proposed conduct.10 Here, I want to argue against the distinction between negligence and strict liability on different, but ultimately related, grounds. I want to argue that there is no non-arbitrary way to assess the riskiness of conduct ex post that will produce a negligence or strict liability distinction among those defendants who failed to advert to the risk. There is no non-arbitrary way because, ex post, the risk of harm created by conduct is either one or zero, and neither risk will produce the needed distinction. To illustrate my point, I offer some cases: 1.
2.
3.
Nellie, looking for some spices in a cabinet, takes out some rat poison and places it on a kitchen counter. She is so preoccupied with her cooking that she forgets she has left the poison there. Later, her 3-year-old brother Oscar comes into the kitchen, thinks the poison is sugar, and eats it. He becomes very ill. Same case as (1), except Oscar slips while he is reaching for the rat poison, hurts himself, and begins crying. Nellie rushes into the kitchen, where she sees that Oscar had almost eaten the poison. With a great sense of relief, she puts the poison back in the cabinet. The Wrinkled Prune Company sells its prunes both pitted and unpitted. Its pitted prunes are pitted by a state-of-the-art technological process. It is extremely rare for a prune pitted by this process to retain any of its pits, but occasionally this happens. There is no cost-justified technology or process design that will guarantee that no pits will remain. Wrinkled puts a notice on every box of its pitted prunes that warns consumers that some pits may remain. Paula, a 6-year-old, loves to eat prunes. She bites down on one of Wrinkled’s prunes and breaks a tooth on a pit that had not been removed. A worker at Wrinkled could have inspected this particular box of prunes for less cost than Paula’s dental bill.
8
See generally Larry Alexander, ‘‘Reconsidering the Relationship Among Voluntary Acts, Strict Liability, and Negligence in Criminal Law,’’ in Crime, Culpability, and Remedy (E.F. Paul et al., eds., 1990): pp. 84–104, 90–94.
9
See Larry Alexander, ‘‘Foreword: Coleman and Corrective Justice,’’ 15 HARV. J. LAW & PUB. POL’Y. 621, pp. 631–36 (1992).
10
See Larry Alexander, supra note 8, at pp. 98–101.
123
Crim Law and Philos
4.
This case is the same as (3), only Paula discovers the pit with her fingers before biting down and thereby avoids injury.
3.1 Analysis of the Cases Traditionally, these four cases would be analyzed along the following lines. In case (1), Nellie was negligent, and her negligence resulted in injury to Oscar. In case (2), she was again negligent, but her negligence led to no harm. In case (3), in many jurisdictions, Wrinkled would be strictly liable to Paula because its product, pitted prunes, was defective and caused Paula’s injury. Wrinkled is not negligent, however, because its process for pitting was cost-justified, even given the risks of an occasional incompletely pitted prune. In case (4), there is neither negligence nor strict Liability. Now what were the risks of harm in these cases? Because we are operating ex post, the answers are easy. In cases (1) and (3), the risk of harm was one. In cases (2) and (4), it was zero. The actual risks fail to distinguish negligence and strict liability. They distinguish only the cases of harm from the cases of no harm. What if we add the condition that the risks of harm must be unjustifiable if conduct is to be deemed negligent? That still does not produce the outcomes of the traditional analysis. In case (3), given that the actual risk of harm to Paula was one, Wrinkled was ‘‘unjustified’’ in not having a worker inspect the box bound for Paula’s house. Of course, it did not know that particular box contained an incompletely pitted prune. But neither did Nellie know that she had left the rat poison out, at least in the sense that she was not adverting to that fact. Moreover, in case (2), the risk is zero, which can hardly be deemed an unjustifiable risk. To make the traditional analysis work, we must delete the information about injury that we possess ex post in such a way as to produce risks in cases (1) and (2) that are between one and zero, equal to each other, and higher than the risks in cases (3) and (4) (which are also between one and zero and equal to each other). The problem is how to select which information about the actual cases to delete. This is not a problem in cases of recklessness. There, we let the actor’s subjective estimate of the risk ex ante determine our characterization of his conduct. If he estimates the risk as high, so high that we would deem taking such a risk unjustifiable, then he is reckless whether the risk was one or zero (that is, whether the harm risked came to pass). In cases (1)–(4), however, we cannot avail ourselves of the actor’s ex ante subjective estimate of risk if we want to produce the traditional outcomes. Nellie undoubtedly believes her activity, which she would describe as ‘‘cooking,’’ is only minimally risky, because she is unaware that it includes her leaving rat poison in a place accessible to Oscar. Wrinkled has an ex ante estimate of the risks of injury to Paula which equals the risks to any consumer of its pitted prunes chosen at random. By hypothesis, that risk is low and reasonable to impose. The risk is ascertained by abstracting from the details of all reported cases of incompletely pitted prunes information that can serve as the basis for efficient actuarial categories. Thus, the percentage of prunes that are incompletely pitted and the total damage caused by incompletely pitted prunes will probably be the only information that is gathered, the more specific details being too expensive to gather or incapable of affecting Wrinkled’s conduct even if gathered. (An example of the latter information would be information that consumers are more likely to bite down on a pit at breakfast, when they are less alert, than at lunch; here, unless the differential risk were sufficient to warrant a special ‘‘breakfast warning,’’ Wrinkled must lump the risks together).
123
Crim Law and Philos
The cases of Nellie and Wrinkled resist being distinguished upon close analysis despite the traditional view that they are different. In all four cases, the actors have in a sense misgauged the risks. In cases (1) and (3), Nellie and Wrinkled have underestimated the risk—Nellie, because she views her act as ‘‘cooking’’ rather than as ‘‘cooking while leaving rat poison where Oscar will eat it,’’ Wrinkled, because it is not concerned with the risks in any individual case but only with the average risks in the aggregate. In both cases, the actual risk of one could have been reduced to zero by cost-justified actions, but neither actor was aware of the factors that would justify those actions. Nellie was unaware that she had forgotten to put the rat poison back in the cabinet. Wrinkled was unaware that the box bound for Paula contained an incompletely pitted prune. So, although Nellie and Wrinkled believed that their conduct was risky, but justifiably so under the descriptions ‘‘cooking’’ and ‘‘selling prunes pitted by a particular process,’’ both were unaware that their conduct caused a risk of one under the descriptions ‘‘cooking (or selling prunes) under the circumstances that actually exist in these cases.’’ Likewise, in cases (2) and (4), where no harm occurs, Nellie and Wrinkled have overestimated the risk in the particular circumstances. To summarize the analysis thus far, the risk in every case is actually either one or zero. In each case, there are general features on which the actor will focus to predict a risk of harm that lies somewhere between one and zero. The actor who predicts a risk that is unjustifiably high and then proceeds to act is reckless even if the harm risked does not eventuate (the actual risk is zero). The actor who predicts a low risk—a risk that it is justifiable to take—is not reckless if he acts. If his prediction is actuarially sound, yet harm eventuates (the actual risk is one), then any liability he faces is paradigmatic strict liability (for example, our case (3)). This is so even though in the particular case acting was not cost-justified. (Wrinkled was not cost-justified in selling the particular box of prunes to Paula in case (3), or was not cost-justified in failing to inspect that particular box.) What if the actor incorrectly underestimates the actuarial risk?11 This may occur when an act has a general feature or features that lend themselves to determining the actuarial risk, but the actor is either unaware of those features or is aware of the features but not of the actuarial risk associated with them. These are the cases of inadvertent negligence, such as Nellie’s leaving the rat poison within Oscar’s reach (cases (1) and (2)). In these cases, we must ask why the actor is ignorant of the important features or the risk associated with them. And in doing so we repeat the preceding analysis, only this time we replace risk of harm with risk of ignorance (of features or risks).12 The risk of ignorance may be either reckless, or it may be cost-justified. (For example, Nellie may be employing an efficient level of advertence that in cases (1) and (2) just happened to result in inadvertence to risk.) Or the actor could be ignorant of the risk of ignorance, in which case the analysis repeats. Ultimately, we end up with either a reckless actor or a strictly liable one. Inadvertent negligence breaks down into one of those two categories; it is not a third category. True risks—one or zero—do not distinguish within the class of human-caused accidents, because the risk is always one in those cases. The focus must therefore be on ex ante subjective estimates of the risk. The subjective estimates may show the action to be ex ante cost-justified (strict liability) because the action is a particular instance of a general class whose actuarial risk of harm is low relative to its benefits and the costs of refining the 11 Here I continue a discussion begun in Larry Alexander, ‘‘Causation and Corrective Justice: Does Tort Law Make Sense?,’’ 6 LAW & PHIL. 1, pp. 17–21 (1987). 12
Mark Grady has made a similar point in several articles. See, e.g., ‘‘Why Are People Negligent? Technology, Nondurable Precautions, and the Medical Malpractice Explosion,’’ 82 NEV. U. L. REV. 293 (1988).
123
Crim Law and Philos
actuarial estimates. The subjective estimates may show the action to be reckless. Or the subjective estimates may, were they correct, show the action to be ex ante cost-justified, but they may be incorrect subjective estimates. In the latter case, our focus must switch to the subjective estimate of risks associated with features of acts, the subjective estimates regarding the degree of confidence warranted in the subjective estimates of risks associated with features of acts, the subjective estimates of the benefits of using different actuarial categories, or the subjective estimates of the likelihood of not adverting to a feature whose associated risk is known. The subjective estimates of those subjective estimates may either reveal reasonable behavior, and hence strict liability, or reveal recklessness. In short, every case of inadvertent negligence—Nellie in cases (1) and (2)—is at bottom a case of recklessness or a case of strict liability. And every case of strict liability— Wrinkled in case (3)—can be viewed as a case of inadvertent negligence if we ignore subjective estimates of risk and focus on the particulars that make the true risk one. To the extent that tort law is concerned with deterring culpable behavior and requiring culpable actors to pay for the damage caused by their acts, both inadvertent negligence and strict liability are outside its purview. To the extent that tort law is concerned with placing the costs of interactions on those—injurers or injured, third-party insurers or first-party insurers—who possess actuarial information about risks, providing (cost-justified) incentives to obtain actuarial information, or effecting distributional patterns and reducing transaction costs (including litigation costs), cases of both inadvertent negligence and strict liability are within its purview, although the distinction between them is wholly chimerical and irrelevant. This is why, from an economist’s view, when strategic concerns relative to the costs and likelihood of proving negligence in court are put to one side, the choice between a negligence rule and a strict liability rule is inconsequential: they would lead to exactly the same conduct. Ultimately, everything comes down to what to do about non-actuarial risks. When we lack information, we cannot act on the information we lack, nor can we assess the value of obtaining that information. (We must have the information in order to assess the value of obtaining it).13 Ignorance cannot be assigned an actuarial value. Except for reckless, knowing, or intentional harm-causing, all harm-causing results from ignorance of true risk. An actor who acts in the face of his own estimate of unjustifiably high risks has acted unjustifiably and can be dealt with in the same manner as are knowing and intentional injurers. But an actor whose ignorance leads him to assess a risk as sufficiently low to make taking it justifiable is an actor whose liability, if any, is strict. And when the strict liability attaches to ignorance that, at the level it occurs, cannot be actuarialized, the case for liability and internalizing costs rather than socializing them cannot rest on attaching the proper incentives to rational calculation.
4 Conclusion Strict liability has no place in criminal law. It authorizes punishment of the innocent and risks deterrence of socially valuable conduct. In tort law, it cannot be distinguished from negligence liability. And the argument in its favor is really an argument for socializing the costs of nonculpably caused accidents.
13 This is the so-called ‘‘information paradox’’ discussed by the famous economist Kenneth Arrow. See Kenneth J. Arrow, ‘‘Economic Welfare and the Allocation of Resources for Invention,’’ in The Rate and Direction of Inventive Activity (Nat’l Bureau of Econ Research ed., 1962), p. 609.
123
