Causation is one of the basic conceptual tools of legal analysis. And for most purposes, we can get along with a notion of causation that is both vague and ambiguous. In the world of medium sized physical objects (automobiles, pedestrians, etc.), our judgments about causation rarely depend on conceptual niceties. The driver’s negligence caused the death of the pedestrian but did not cause Barak Obama to win the Iowa caucuses in 2008. In these cases, various notions of causality converge. The person on the street, the scientist, and lawyer can all agree in such cases that for all practical purposes X caused Y but not Z. But sometimes the various notions of cause come apart exposing ambiguities and vagueness in both ordinary and legal talk about causes and effects. This post provides a very basic introduction to causation for law students (especially first-year law students) with an interest in legal theory.
Cause-in-Fact & Legal Cause
Let’s put the most important distinction on the table right away. Contemporary legal theory and judicial practice assume that there is a distinction between legal cause on the one hand and cause-in-fact on the other. What does that mean? That’s a huge question, of course, but we can state one conclusion straight away: that X is a cause-in-fact of Y does not entail that X is a legal cause of Y. Less obviously, that X is a legal cause of Y does not entail that X is a cause-in-fact of Y. The various ways that cause-in-fact and legal cause can come apart leads many to the conclusion that legal cause simply has nothing to do with causation, but this turns out to be an exaggeration. I know this all sounds very airy. So let’s get down to brass tacks!
What do we mean when we say that X is a cause-in-fact of Y? Many law students learn that the answer to this question is but-for causation. If it is the case that but for X, Y would not have occurred, then X is a but-for cause of Y and hence X is a cause-in-fact of Y. This simple story works most of the time, and as a rough and ready rule of thumb, it isn’t half bad. But it turns out that if you try to use but-for causation as a hard and fast rule for determining whether X is the cause of Y, you will run into trouble, sooner or later. In torts and criminal law, but-for causation runs into trouble somewhere in the midst of the first-year course. In a sense, the point of this Lexicon post is to provide a set of tools that for understanding the troubles that overreliance on but-for causation can cause.
Necessary and Sufficient Causes
The first item in the causation toolkit is the distinction between necessary and sufficient cause. The basic ideas are simple and familiar. X is a necessary cause of Y, if Y would not have occurred without X. Ben’s running the red light is a necessary cause of the damage to Alice’s car, just in case the damage would not have occurred without Ben’s having run the light. The idea of "necessary cause" is the same idea expressed by the phrase "but-for cause."
X is a sufficient cause of Y, if Y would have occurred so long as X occurred. Alice’s shooting Ben through the heart is a sufficient cause of Ben’s death, just in case the shot thru the head by itself would have caused Ben’s death. This is true, even though Ben would have died anyway, because Cynthia shot him through the head at the same time Alice shot him through the heart.
This rough ready distinction between necessary and sufficient causes really won't do all the necessary work. For example, Alice's shooting Ben through the heart is not truly sufficient, by itself, no matter what other conditions obtain, to cause Ben's death. If Ben had been on an operating table awaiting a heart transplant, then he might have lived despite the slot. One suggestion for dealing with cases like this is to identify "sufficient causes" as a "necessary element of a sufficient set." Things get even more complex from for the purposes of this introduction the point is simply to see the indadequacy of "necessary and sufficient cause" as a tool for dealing with complex cases in a precise way.
The Role of Counterfactuals
The notions of necessary and sufficient causation are familiar to almost everyone. We use these ideas all the time in everyday life. But the very familiarity of these concepts creates a temptation to take them for granted. There is an important feature of these ideas that our day-to-day use of them does not make explicit. Both necessary and sufficient causation are counterfactual concepts. What does that mean? “Counterfactual” is simply the fancy name for “what if” thinking. What if Ben had stopped at the red light? Would the damage to Alice’s car still have occurred? What if the Ben had gotten immediate medical attention? Would the shot through the head still have killed him? Every statement regarding a necessary or sufficient cause can be interpreted as making a counterfactual (“what if”) claim.
What-if reasoning is itself familiar and ordinary. When we say, Ben’s running the red light was a necessary cause of the damage to Alice’s car, we are claiming that if the world had been different and Ben had not run the red light, then Alice’s car would not have been damaged. We imagine what the world would have been like if Ben had stopped at the red light, and Alice had proceeded through the intersection without being struck by Ben’s car. Counterfactual reasoning can get more complicated that this, but for our purposes we can use everyday what-if reasoning as our model of role of counterfactuals in necessary and sufficient causation.
Once we’ve gotten the notions of necessary and sufficient causes, we can move on to the idea of overdetermination. An effect is overdetermined if it has more than one sufficient cause. Take the case of Alice shooting Ben through the heart. We have postulated that the bullet passing through the heart was a sufficient cause of Ben’s death, but it may not have been a necessary cause. Suppose that Alice was a member of a firing squad, and that at the exact same moment that Alice’s bullet passed through Ben’s heart, another Bullet, fired by Cynthia, passed through Ben’s cerebral cortex and that this would have resulted in Ben’s death, even if Alice’s had not fired or her bullet had missed. Ben’s death now results from two sufficient causes, but neither Alice’s shot nor Cynthia’s shot was necessary. If Alice had not fired, Cynthia’s shot would have killed Ben. If Cynthia had not fired, Alice’s shot would have killed Ben.
Overdetermination is important, because it undermines the idea that but-for causation tells us everything we need to know about cause-in-fact. We might say that both Alice and Cynthia’s shooting caused Ben’s death or we might say they were both partial causes of Ben’s death, but we would not be likely to say that neither Alice nor Cynthia’s shot was the cause.
The firing squad example was described as a case of simultaneous overdetermination—both sufficient causes occurred at the same time. What if Cynthia shot a few seconds before Alice and Ben died before Alice’s shot pierced his heart? In that case, Cynthia’s shot would have preempted the causal role of Alice’s shot. If Cynthia had missed, then Alice’s shot would have killed Ben. This kind of case is sometimes called preemptive causation.
Overdetermination poses one kind of problem for but-for causation, coincidence poses another a different sort of difficulty. Suppose the driver of a trolley is speeding. As a result the trolley is in just wrong place and time and a tree falls, injuring a passenger. If trolley had gone just a little faster or just a little slower, the tree would have missed the trolley and the injury would not have occurred. Given these circumstances, speeding was a but-for cause (a necessary cause) of the tree injuring the passenger. So what? Coincidence is no problem for cause-in-fact, but it does pose a problem for the legal system. Intuitions vary, but lots of folks are inclined to believe that one should not be legally responsible for harms that one causes as a result of coincidences.
Coincidence is related to a variety of other problems with but-for causation. Take our example of Ben running the stoplight and hitting Alice’s car. Running the stoplight was one but-for cause of this accident, but there are many others. For example, Alice’s being in the intersection was also a but-for cause. And how did Alice come to be in the intersection at just the time when Ben was running the red light? If her alarm clock hadn’t gone off, she would have slept in and arrived in the intersection long after Ben, so her alarm clock’s ringing was another but-for cause. And you know how the story goes from here. As we trace the chain of but-for causes back and out, we discover that thousands and millions and billions of actions and events are but-for causes of the accident.
Alternatives to But-For Cause-in-Fact
Although "but-for causation" is frequently presented as the only account of cause in law school classes, it is far from the only view. The alternatives to but-for causation frequently view causal relationships as involving more than but-for cause. In fact, some theories of causation sharply distinguish but-for relationships from true causation. The best way to learn about these alternatives to read the references cited below: the Stanford Encyclopedia of Philosophy articles provide a good starting point.
What do we about the problems with problems created by but-for cause? One way that the law responds is with the idea of legal cause or proximate cause. In this post, we cannot hope to achieve a deep understanding of legal cause, but we can get a start. Here are some of the ideas that help me to understand legal cause.
First, there is a terminological issue: causation may be confused with responsibility. “Legal cause” is only partially about cause. We start with the idea of cause-in-fact (understood in light of the distinction between necessary sufficient cause). This idea of cause seems, on the surface, to fit into the structure of various legal doctrines. So we imagine that if a defendant breaches a duty of care and causes a harm, then defendant is legally responsible for the harm. This works for lots of cases, but then we start thinking about other cases like overdetermination and coincidence. “Legal cause” is the way that we adjust our ideas about legal responsibility to overcome the counterintuitive results that would follow from a simple reliance on but-for causation. In other words, “legal cause” may be a misnomer. It might be clearer if we used the phrase “legal responsibility” (or some other phrase) to describe the ways in which we adjust the law.
Second, legal cause is frequently associated with the idea of foreseeability. For example, in coincidence cases, the harm (the tree injuring the passenger) is not a foreseeable consequence of the wrongful act (driving the trolley at an excessive speed). If the purpose of the law is deterrence, then no good purpose may be served by assigning legal responsibility in cases where the effect is unforeseeable.
Third, legal cause is sometimes associated with the idea of proximity in time and space. Of course, the phrase “proximate cause” emphasizes this connection. We usually don’t want to hold defendants responsible for the remote and attenuated effects of their actions. We frequently do want to hold defendants responsible for the immediate and direct effects of their actions. “Proximity” seems to capture this point, but an overemphasis on proximity in time and space leads to other problems. Some immediate consequences do not give rise to legal responsibility: the trolley driver may have started speeding just seconds before the tree fell. Some causal chains that extend for long distances over great durations do give rise to legal responsibility: Osama bin Laden’s responsibility for 9/11 would not be vitiated by the fact that he set events in motions years in advance and thousands of miles away.
Our investigation of causality so far has elided an important set of issues—the connections between causation and probability. These connections are far too large a topic for this post, but even a superficial analysis requires that we consider two perspectives--ex ante and ex post.
Ex post questions about causation arise in a variety of contexts, but for the legal system, a crucial context is provided by litigation and especially trial. In many cases, there is no doubt about causation. When Ben’s car speeds through the red light and hits Alice’s car, we don’t have much doubt about what caused the damage. But in many types of cases, causation will be in doubt. Did the chemical cause cancer? Was the desk job the cause of the back injury? Sometimes the evidence will answer these questions with certainty (or perhaps, with something that is so close to certainty that we treat it as certainty for legal and practical purposes). But in other cases, the evidence will leave us with a sense that that the defendant’s action is more or less likely to have caused the harm to the defendant. Such probabilities may be expressed either qualitatively or quantitatively. That is, we might say that it is “highly likely” that X caused Y or we might say that there is a 50% chance (p = .5) that X caused Y.
Ex ante issues of causation also arise for the law. For example, the legal system may be required to assign a value to a risk of harm that has not yet been realized. David has been exposed to asbestos, but may or may not develop cancer. In this case, probabilities refer to the likelihood of future events.
Decision theory and mathematics have elaborate formal machinery for representing and calculating probabilities. In this short post, we cannot even scratch this surface, but there are two or three bits of notation that every legal theorist should know:
--The letter “p” is frequently used to represent probability. Most law students encounter this notation in Justice Hand’s famous opinion in the Carroll Towing case (B < PL or “burden less than loss discounted by probability). The notation p(x) = 0.1 can be read “the probability of x equals 1/10.” And the notation, p=0.5 can be read “probability equals one in two.”
--The symbol “|” is frequently used to represent conditional probabilities. Suppose we want to represent the probability that X will occur given that Y has occurred, we can use this notation: p(X|Y). So we could represent the sentence, “The probability of Cancer given Exposure to Asbestos is ten percent,” as p(C|EA)=0.1.
Types and Tokens
So far, we have been focusing mostly on cases where an individual instance of harm is caused by some particular wrongful action. But of course, we frequently think about causation as a more general relationship. For example, in science we might speak of “causal laws.” There is no standard terminology for this distinction: we might use the phrase “individual causation” and “systematic causation.” One helpful bit of terminology for getting at this idea is to differentiate “types” and “tokens.” Ben’s running the rend light at a particular time and location is an event token and it is a token of a type of events, i.e. the type “running a red light.”
Once we have the distinction between types and tokens in place, we can define individual causation as a causal relationship between a token (e.g. a token event) and another token (e.g. a token action). And we can define systematic causation as a causal relationship between a type (e.g. a type of event) and another type (e.g. a type of action). Science studies causal relationships between types; trials frequently involve questions about the causation of one token by another. This leads to another important point: the question whether an individual harm was caused by an individual action will sometimes depend on the question whether a systematic causal relationship exists; for example, the question whether this factory’s release of a chemical caused an individual case of cancer may require a jury to resolve a “scientific” question about systematic causation.
Even though this is a long entry by the standards of the Legal Theory Lexicon it is a very compressed and incomplete treatment of the concept of causation. Given the way legal education is organized (around doctrinal fields like torts, criminal law, and evidence), most law students never get a truly comprehensive introduction to causation. Torts may introduce the distinction between cause-in-fact and legal cause; criminal law, problems of overdetermination; and evidence, the relationship between probability and causation. If this post accomplishes anything of value, I hope that it serves as warning—causation is a deep and broad topic about which there is much to learn.
Related Lexicon Entries
- H.L.A. Hart & Tony Honore, Causation in the Law (2d ed. 1985). For many years, this was the book on causation and the law. Back in print.
- Michael Moore, Causation and Responsibility (2009). This is the best contemporary treatment, and a must read for legal theorists who want to know about causation.
- L.A. Paul & Ned Hall, Causation: A User's Guide (Oxford University Press 2013). Very recent introduction.
- Causation (Oxford Readings in Philosophy) (Ernest Sosa & Michael Tooley eds. 1993). A fine collection of essays, with contributions by J.L Mackie, Michael Scriven, Jaegwon Kim, G.E.M. Anscombe, G.H. von Wright, C.J. Ducasse, Wesley C. Salmon, David Lewis, Paul Horwich, Jonathan Bennett, Ernest Sosa, and Michael Tooley.
Resources on the Internet
- Antony Honore, Causation in the Law, Stanford Encyclopedia of Philosophy
- Peter Menzies, Counterfactual Theories of Causation, Stanford Encyclopedia of Philosophy
- Jonathan Schaffer, The Metaphyiscs of Causation, Stanford Encyclopedia of Philosophy
- Richard Wright, The NESS Account of Natural Causation: A Response to Criticisms, SSRN.
(This entry was last modified on April 10, 2016.)