A familiar rhetorical trope in modern advocacy is: “Imagine if visitors from outer space were observing x; how would they describe it?” The payoff of this exercise is to get the audience to see that the view proposed by the speaker, while superficially unfamiliar, is actually more perceptive than the conventional understanding of the practice at issue. The subtext is that only with the benefit of insights gleaned from a great distance (or an unusual perspective) can those immersed in a practice truly understand it.
I could not help but think of this trope while reading Knobe and Shapiro’s fascinating—if at times frustrating—paper on proximate cause. Of course, they are not space aliens; they are both philosophers and one (Shapiro) is a law professor as well. But neither specializes in tort law, and by their own admission they are leveraging their distance from the conventional discourse of torts scholars and judges to arrive at insights that have otherwise eluded those of us immersed in the practice.
This paper deserves attention from anyone interested in the future of private law as a distinct field. It tries to preserve private law’s relevance as a jurisprudential category by offering it a lifeline from outside its familiar precincts. Significantly, Knobe and Shapiro are not skeptics about legal reasoning, nor do they concede ground to those, like Leon Green, who sought to collapse private law into public law. The assistance they offer is intended to improve tort law from within by giving it new tools to make sense of the mass of judicial opinions which already instantiate a workable set of rules for deciding cases. The question is, how much help do they really provide?
The paper makes four distinct claims. The first is that the legal doctrine of proximate cause is a mess—that the reasoning provided, if not the outcomes reached, by judges when they deploy the doctrine are confused (and perhaps wrong, as a matter of law). The second is that the reason for the mess is because lawyers and judges, notwithstanding that they disagree with each other, are starting with a set of false premises about law. The third is that the visitors (the philosophers) can see what the insiders have missed because they are using tools heretofore unknown to the insiders, namely “experimental jurisprudence.” (P. 7.) Finally, Knobe and Shapiro claim that the conception of proximate cause they offer fits “patterns observed in legal judgments” pretty well. (P. 39.)
My reaction to these four claims, in brief, are as follows. The first is correct. The second may be correct, but Knobe and Shapiro do not add much to the reasons we have for believing it (other than the fact the first is correct). As the third depends on the validity of the fourth, and as the fourth is not wholly convincing, it seems premature to endorse the third.
It is well known that proximate cause is a controversial topic in tort law as well as other parts of the law. The source of the controversy is sometimes over terminology, but the interesting issues concern the practical implications generated by the arguments over terminology. So, for example, Reporters for the Third Torts Restatement, in a “Special Note on Proximate Cause” explain that the term “proximate cause” does a poor job of capturing the idea it expresses.1 In other words, the phrase “proximate cause” (and its later iteration, “legal cause”) denotes a legal concept that is part of the law, and not just a special case of some other legal concept (like duty), the only problem being that the concept has until now been poorly defined and analyzed. The Reporters state that “Tort law does not impose liability on an actor for all harm factually caused by the actor’s tortious conduct” and suggest that the phrase “scope of liability”—separate from duty and cause in fact—captures this legal concept more satisfactorily than the phrase “proximate cause.”
Thus, we have some agreement by the insiders with Knobe and Shapiro. “Proximate cause” has not served our legal system very well. But the agreement does not go very far. Knobe and Shapiro come to their conclusion by observing a debate—which they characterize as between Formalists and Realists—and discerning that both sides share certain premises but fight over the proper relation between those premises. (P. 12.) Those premises are that anyone doing adjudication (or writing about it) work in a world where there are such things as causal judgments and moral judgments, and the two kinds of judgments are conceptually independent of each other—the former referring to a “metaphysically real relation” (P. 12, emphasis supplied) and the latter referring to a judgment about who is “morally responsible for a harm.” (P. 10.) Formalists, they continue, start with judgments about causation and, depending on the moral principal employed, decide if someone should be blamed for what they have caused (e.g.: negligence vs. strict liability), whereas realists start with judgments about moral responsibility and decide if someone should be treated as if they are a ‘real’ cause based on the prior moral judgement (e.g.: Andrews’ dissent in Palsgraf).
Knobe and Shapiro argue that the better understanding of causation (all causation, not just what lawyers have called proximate cause) is that it is like a sandwich: moral judgments inform the concept of causation (as a relation between events), and judgments about causation inform legal conclusions about blame (or responsibility). (Pp. 12-15.) Their approach has two crucial steps. First, Knobe and Shapiro need a method that determines the moral concepts that guide causal judgments. This is where they introduce the innovation of experimental jurisprudence—studying ordinary people’s responses to questions posed in experiments. Second, Knobe and Shapiro need to provide content to the ‘moralized’ causal concept produced by the first step. This paper provides the content with some variations, but here is the central idea around which the variations pivot: “[P]eople’s causal judgments are impacted by their beliefs about whether the agent’s behavior is abnormal. (P. 20, emphasis in the original.) This claim reflects an understanding of causation that extend to any proximate cause query, and is applicable to, for example, causes that are statistically abnormal, thus intervening to relieve an actor of responsibility. But before I discuss these two steps in more detail, I want to make an observation about the putative novelty about their methodological innovation (what I have, slightly irreverently, called “the sandwich”).
It is not quite accurate to say that the debate over proximate cause is exhausted by the debate between Formalists and Realists. The formalist position presented in the paper resembles the so-called “directness test” found in the famous Polemis case. That is, the idea that the test for proximate cause in law involves a factual judgment about the world—whether or not an event “directly” caused another event. A version of this can be seen in Ryan v. New York Central R.R. Co. (35 N.Y. 210 (1866)), which improbably claimed that a fire spreading to a neighbor’s property was not “natural and ordinary”. As the Reporters of the Third Restatement wrote, these Formalist decisions reflected the view that proximate cause “could be determined through a neutral, scientific inquiry.”2 The Third Restatement’s formulation of “scope of liability” rejects the Formalist position yet does not adopt the Realist position. It is actually quite similar to the Knobe and Shapiro sandwich. The Third Restatement explicitly embraces what is sometimes called the “risk rule” for determining when an action counts as a proximate cause of an injury. The risk rule quite overtly employs a norm to determine whether an actual cause is a proximate cause—the norm which says that an actor is not subject to tort liability for causing an injury to another unless the causal connection involves the realization of one of the risks that renders the actor’s conduct tortious in the first place.3
To recap: Knobe and Shapiro have told us that the Formalist and the Realist positions about proximate cause fail because they each misunderstand the need for a normative concept of causation, and they offer in the place of these alternatives a normative concept of causation that they think will work pretty well. However, as I have pointed out, tort insiders also have a normative concept of causation—the risk rule, and many have been trying to apply it since the early 1960’s , when it was formally adopted in Wagon Mound I in the U.K. and Kinsman in the U.S. So, the next question is, how does Knobe and Shapiro’s normative concept of causation compare to the risk rule?
In a short essay it is not possible to cover all of the variations of Knobe and Shapiro’s proposed test. I will focus on a few specific instances where their test and the risk rule are in direct competition. The most obvious place to begin is in cases of “superseding cause”, such as the case with which Knobe and Shapiro begin their paper, Henningsen v. Markowitz, 230 N.Y.S. 313 (1928). The case involved the question of whether a mother’s negligence in failing reasonably to execute an intervention (by ineffectively attempting to take away an air rifle from her 13-year-old son) was a reason to hold that the negligent conduct of the defendant (selling the gun to the child in violation of a state criminal statute) was not a proximate cause of the plaintiff’s injury. Knobe and Shapiro argue that the correct rule attributes causation to the negligent seller unless it is the case that the mother’s act (not doing enough to prevent her child from gaining access to the gun) was more abnormal than the defendant’s original negligent act (unlawfully selling an air rifle to a minor). The risk rule would argue that the correct rule attributes causation to the negligent defendant if the risk that was realized (the victim being shot in the eye when the child and his friend were using the gun) is one of the risks that rendered the defendant’s conduct negligent. Both approaches come up with the same answer—the defendant’s conduct constitutes a proximate cause. So why prefer one over the other?
One reason is that superseding cause cases are not really the most important, or illuminating, cases confronting courts grappling with the problem of proximate cause. As the Third Restatement has pointed out, superseding cause is an area of law with “declining importance” for a variety of reasons, including the introduction of comparative responsibility—so that the cases involving it (such as Henningsen) are of “waning influence” in this area of law.4 And for other, more central types of proximate cause cases—those not involving intervening wrongdoing—are better handled by the risk rule than the Knobe and Shapiro abnormality test. Consider the problem which the risk rule is especially good at solving—where the defendant causes an unusual harmful consequence without the negligent or intentional interference of a third party.
To take a classic example, suppose D, an adult, negligently entrusts a loaded handgun to a minor or an incompetent, and the minor or incompetent non-negligently drops the handgun on P’s toe, breaking it. The risk rule is robust in its ability to explain why D’s negligent is not a proximate cause of P’s injury: the risk that was realized was not one of the risks that rendered D’s conduct careless (as would have been the case if the incompetent person had accidentally shot P). Yet it is hard to know how Knobe and Shapiro’s “normality-based” approach (P. 21) can even start to address this situation. A child or incompetent non-negligently dropping a 2-pound object is not abnormal, either statistically or morally, which would suggest, counterintuitively, that D’s conduct was a proximate cause of P’s harm. Knobe and Shapiro state that “people tend to regard a factor as especially causal when it is morally wrong and when it is statistically infrequent” (like a criminal taking advantage of storeowner’s failure to light a portion of her property). (P. 22) But as the handgun example shows, under the risk rule, whether a defendant is a proximate cause depends less on a comparison between an intervening event and the defendant’s careless conduct, and much more on a comparison between the defendant’s conduct and the final event—the risk that was realized. The intervening event will be part of the process of evaluation under the risk rule, but the aspect which Knobe and Shapiro emphasize—the comparison of each events’ degree of “normality”—seems ad hoc. The question is not, as they put it, whether another cause “beats out” the defendant’s conduct as a cause of the victim’s injury (P. 27)—as if proximate cause is a race. Rather, it is whether the defendant’s wrongdoing aligns in the right way with the harm suffered by the plaintiff.
None of this is to say that Knobe and Shapiro’s “normality”-based concept of proximate causation produces the wrong answer in those cases that do involve intervening actors who are intentional wrongdoers, or involve highly unlikely intervening events. But their theory cannot explain a lot of other cases, and so it is, at best, under-determinative for many cases about which modern tort law is deeply concerned. This makes one wonder whether the results the cognitive science research, which aligns with the results of the risk rule up to a point, reveal the limits of the use of research into ordinary judgments about causation for law. One does not need to deny the truth of Knobe and Shapiro’s claim that ordinary judgments about causation, as revealed by empirical research, tell us that lay judgments about causation are motivated by norms which align with legal concepts such as the risk rule. But alignment may not be enough—it may be the case that the insiders cannot only resort to the tools being brought to us from the outside because there are some questions in law that only more legal reasoning can solve.
This paper is, in some ways, reassuring for those of us who are its intended beneficiaries. It is good to know that the conventional methods of courts and scholars aligns, to some extent, with the predictions produced by experimental jurisprudence. This should not be too surprising, since there is a way in which the process of watching juries grapple with jury instructions from scores of jurisdictions (and then thinking about what to do with those juries’ outputs when they are appealed) shares some common ground with the everyday work of research psychologists. The challenge posed by Knobe and Shapiro is one of emphasis—where do we think our efforts should go when the going gets tough and courtroom results do not easily yield to legal analysis. This paper suggests the better strategy is to ask better questions about what ordinary people believe the law is. I doubt that this is going to prove to be winning strategy, since, like so much in life, what ordinary people believe underdetermines the hardest parts of most human practices.
- See Chapter 6, Scope of Liability (Proximate Cause), Special Note.
- See Restatement (Third) of Torts: Phys. & Emot. Harm § 34, cmt a (2010).
- See Restatement (Third) of Torts: Phys. & Emot. Harm § 29, Reporter’s Notes cmt e (2010).
- See Restatement (Third) of Torts: Phys. & Emot. Harm § 34 cmt a (2010).