What Does It Mean to "Cause" Something in Law?

Imagine you’re playing soccer at recess. You kick the ball hard, it flies across the field, hits a loose rock, bounces sideways, smashes a window, and the shattered glass cuts another kid who was standing nearby. Did you cause the injury? Most people would say yes—you kicked the ball, after all. But now imagine it differently: you kick the ball, it hits the rock, bounces sideways, and at that exact moment a branch falls from a tree, knocks the ball in a different direction, and it still breaks the window. Did you cause the window to break? What if, instead of a branch, another kid deliberately swatted the ball toward the window as a prank? What if there were two windows, and both were broken at the same time by two different balls kicked by two different people, but either one alone would have broken both windows anyway?

These aren’t just tricky schoolyard questions. When real courts decide who is legally responsible for harms, they have to answer exactly this kind of puzzle. And philosophers have noticed something strange: the law doesn’t seem to have a single, simple answer to what “causing” something means. In fact, judges and legal thinkers disagree deeply about what causation even is in the legal sense.

The Two-Part Puzzle

Here’s the basic structure most legal systems use. They split causation into two questions.

First: Did the person’s action actually make a difference? This is called “cause in fact.” The standard test is simple: “But for the defendant’s action, would the harm still have happened?” If the answer is no—the harm wouldn’t have occurred without their act—then they caused it in fact. If the answer is yes—the harm would have happened anyway—then they didn’t.

Second: Was the causal connection close enough to hold them responsible? This is called “proximate cause” or “legal cause.” Even if someone’s action was necessary for a harm to occur, the law sometimes says the connection is too remote, too weird, or too interrupted to count. Think about that soccer ball again: if you kicked it and it broke a window three seconds later, that’s close. But if you kicked it, it rolled into the street, got picked up by a truck, was carried 50 miles away, and a year later someone else kicked it into a window—did you cause that? Your kick was technically necessary, but it feels absurd to say you caused the window to break.

This two-part structure seems straightforward. But as soon as you start looking at real cases, things get complicated fast.

When “But For” Doesn’t Work

The “but for” test has a serious problem: it fails in situations where multiple things could each have caused the same harm on their own. Philosophers and lawyers call these “overdetermination” cases.

Imagine two people independently light fires in a forest on the same dry day. The fires join together and burn down a cabin. Either fire alone would have been enough to destroy the cabin. Now ask: “But for the first fire, would the cabin have burned?” No—the second fire would have done it. Same for the second fire. So the “but for” test says neither fire caused the damage. That seems obviously wrong. Both fires did cause the damage, together.

Or imagine a different scenario: one fire starts, burns the cabin to the ground, and then a second fire arrives minutes later—but there’s nothing left to burn. Here, the first fire clearly caused the destruction. The second fire didn’t. Yet the “but for” test has the same problem: neither fire was individually necessary, because if the first hadn’t happened, the second would have done the job. So again, the test fails to give the right answer.

Courts have tried to fix this by modifying the test. One common fix is to ask more precisely: “But for the defendant’s action, would the harm have occurred in exactly the same way, at exactly the same time?” In the two-fire case, the first fire was necessary for the cabin to burn when it did and how it did. That seems to work—but it also opens the door to saying that basically everything is a cause of everything else, if you describe the harm precisely enough.

Another fix is to use what’s called the “NESS” test: an action counts as a cause if it was a Necessary Element of a Sufficient Set of conditions for the harm. The idea is that each fire was part of its own sufficient set (fire + oxygen + dry wood = destruction), even though neither set was individually necessary. This works better in many cases, but it has its own problems.

The Skeptics Arrive

Some legal thinkers have gone further. They’ve argued that “cause in fact” isn’t really a factual question at all—that what courts are really doing is deciding who should be held responsible, and then calling them the “cause.”

This skepticism has a long history. In the early 1900s, a group called the American Legal Realists pointed out that in cases where two fires joined to burn a house, courts treated the situation differently depending on whether the fires were both set by humans or one was natural. If both were human acts, both actors were causes. If one was a lightning strike and the other was a careless camper, the camper might not be held to have caused anything. But the “but for” test doesn’t distinguish these cases—either way, neither fire was necessary. So something else must be driving the decisions.

Later, some legal economists argued that causation in law should be understood purely in terms of incentives. If we want to discourage harmful behavior, we should hold people responsible when their actions raise the probability of harm—not necessarily when they actually cause it. On this view, “cause” is just a way of saying “should have been deterred.”

Then came the Critical Legal Studies movement in the 1970s and 80s. They argued that causation requirements were a myth—that the law pretends to have objective, factual standards when really everything is about political and social power. On their view, calling someone a “cause” is just a mask for a decision about who to punish.

The Proximate Part: How Close Is Close Enough?

Even if everyone agreed on cause in fact, the proximate cause question is just as messy. Here, the central puzzle is: When does a causal chain become too long or too weird to hold someone responsible?

There are several competing answers.

The foreseeability test says you’re only responsible for harms you could have reasonably predicted. If you couldn’t have foreseen the chain of events leading to the harm, you didn’t legally cause it. This seems fair—why blame someone for something they couldn’t have anticipated? But it has a problem: it asks basically the same question as the law already asks about whether someone was careless or intended the harm. If you already know someone acted recklessly, asking “was the harm foreseeable?” doesn’t add anything new.

The harm-within-the-risk test says what matters is whether the harm that actually happened was the kind of harm that made the person’s action wrong in the first place. If you were reckless because you risked an explosion, and the harm was a cut from falling glass—well, was that “within the risk”? Maybe not. But this test doesn’t handle situations where the causal route was bizarre even though the harm was exactly the kind foreseen. If you light a fuse intending to blow up a wall, but lightning strikes the fuse and sets it off, you still caused exactly the harm you intended—but the law usually says you didn’t cause it, because the chain was broken by an unusual event.

The direct cause test focuses on whether something “intervened” between the action and the harm. The idea is that certain kinds of events—freakish natural events, or free and deliberate human choices—break the causal chain. If someone deliberately steps in and redirects your action toward a different harm, they become the new cause. If a once-in-a-century storm washes away the evidence of your crime, you didn’t cause the damage—the storm did. This matches a lot of everyday thinking about causation: we tend to think that free choices and crazy coincidences “reset” causal responsibility.

The remoteness test says causation just fades out over space and time. Not as a sudden break, but gradually. Caesar crossing the Rubicon was technically necessary for you to be reading this essay—but nobody thinks Caesar caused your reading. The causal connection has “petered out.” This seems right, but it’s hard to turn into precise legal rules. How far is too far? How many links in the chain are too many?

The Bigger Picture

So what is the law’s concept of causation? The honest answer is: nobody agrees. There are at least a dozen different tests and modifications floating around in legal decisions and academic writings. Some say the law should use the same concept of causation that philosophers and scientists use—that causation is a real relation in the world, and the law’s job is to discover it. Others say the law’s concept is sui generis—that legal causation is shaped by the purposes of law (deterrence, fairness, compensation) and can’t be reduced to physics or metaphysics.

What makes this debate genuinely hard is that it’s not just theoretical. Real people’s freedom and money depend on how these puzzles get resolved. A court’s decision about whether someone “caused” a harm can mean the difference between prison and freedom, between bankruptcy and financial security.

Part of what philosophers find fascinating about all this is that legal thinking about causation can help reveal things about causation itself. When judges face real cases and have to give real answers, they sometimes notice patterns that philosophers working in their armchairs have missed. And conversely, when philosophers develop theories about what causation is, the law provides a testing ground: do these theories give the right answers in the messy cases that actually matter?

The debate remains very much alive. And if you find yourself puzzling over whether you really caused that broken window, you’re in good company.

Key Terms

Term	What it does in this debate
Cause in fact	The first part of legal causation: whether an action actually made a difference to whether the harm occurred
Proximate cause	The second part: whether the causal connection is close enough (in time, space, or normalness) to hold someone responsible
But-for test	The standard test for cause in fact: “But for this action, would the harm still have happened?”
Overdetermination	A situation where multiple factors are each sufficient to cause the same harm, making the but-for test give weird results
NESS test	An alternative to but-for: an action is a cause if it was a Necessary Element of a Sufficient Set of conditions for the harm
Intervening cause	An event (often a free human choice or a freakish natural event) that “breaks” the causal chain between action and harm
Foreseeability test	A test for proximate cause: you’re only responsible for harms you could have predicted
Harm-within-the-risk test	A test for proximate cause: the actual harm must be the kind of harm that made the action wrong
Substantial factor test	A vague but practical test: was the action a “substantial factor” in producing the harm?
Causal skepticism	The view that “cause” in law doesn’t name a real relation but is really a disguised policy decision

Key People

Herbert Hart was a British philosopher of law who, along with Tony Honoré, wrote the most influential book on causation in law. He argued that the law’s concept of causation is deeply connected to everyday causal thinking, especially the idea that free human actions and coincidences break causal chains.
Tony Honoré was Hart’s collaborator on Causation in the Law. He helped develop the idea that legal causation depends on distinguishing between voluntary human actions (which are causes) and mere background conditions.
The American Legal Realists (like Henry Edgerton) were early 20th-century legal thinkers who argued that legal concepts like causation are really just labels for decisions based on policy and fairness, not factual discoveries.
The legal economists (like Guido Calabresi, Richard Posner, and Steven Shavell) argued that causation in law should be understood in terms of creating efficient incentives—making the people who can cheapest prevent harms pay for them.

Things to Think About

Suppose you’re on a jury. A driver was texting while driving and ran a red light. At that exact moment, a pedestrian who was wearing all black at night stepped into the crosswalk against the walk signal. The driver hit and killed the pedestrian. Did the driver’s texting cause the death? Would the death have happened anyway, even if the driver had been paying attention? How do you decide what “would have happened” in a world where things went differently?
A bully threatens to beat up your friend unless you punch them first. You punch your friend. Did the bully cause your friend’s injury, or did you? Both? Neither? If you think the bully caused it because they made you do it, does that mean any case where someone is forced to act means the person who forced them is the real cause? Where would you draw the line?
A doctor treats a patient for a minor injury, but makes a mistake that causes a serious infection. The patient then refuses antibiotics because of their religious beliefs, and dies. The doctor says: “Yes, my mistake caused the infection—but the patient’s choice caused the death.” Is that right? Does a person’s free choice always “break” the causal chain from earlier actions?
If you think causation in law should track real causation in the world (physics, biology, etc.), what do you do about cases where physics gives the “wrong” answer—like the two fires case, where physics says neither fire was necessary but the law says both were causes? Does the law have to ignore reality here, or is there a way to reconcile them?

Where This Shows Up

Your own life: Every time you hear “but for you, this wouldn’t have happened” in an argument, you’re arguing about cause in fact. Every time someone says “that was too far removed from what you did,” they’re arguing about proximate cause.
News and politics: Debates about police use of force often turn on causation questions: Did the officer’s action cause the death, or did the person’s own behavior? What about pre-existing health conditions?
Medicine: When someone gets sick, doctors and lawyers argue about whether a specific exposure (to a chemical, a drug, a virus) caused their illness, especially when other causes were also present.
History and everyday explanation: When we say “World War I was caused by the assassination of Archduke Franz Ferdinand,” we’re making a causal claim that faces the same problems as legal causation—many other factors were also necessary, and the chain of events seems absurdly long. The law’s puzzles about causation are really everyone’s puzzles.