What Makes Something a Cause?
Imagine you’re playing catch with a friend. You throw the ball, they catch it. Simple. But then a philosopher asks: “Did your throw actually cause the catch? Or is it more complicated than that?”
Here’s a strange thing philosophers noticed: when we say one thing caused another, we often mean something like: if the first thing hadn’t happened, the second wouldn’t have happened either. If you hadn’t thrown the ball, your friend wouldn’t have caught it. That seems obvious. But when you start looking closely, it gets weird fast.
The Basic Idea: “If Not This, Then Not That”
The philosopher David Lewis, who spent a lot of time thinking about these questions, put it this way: “We think of a cause as something that makes a difference, and the difference it makes must be a difference from what would have happened without it.”
This is called a counterfactual — a thought about what would have happened if things had been different. Most counterfactual theories of causation start here: if you want to know whether event C caused event E, ask yourself: “If C hadn’t happened, would E still have happened?”
If the answer is no, then E depends on C. That looks like causation.
But here’s the problem: life isn’t that simple. Sometimes an event doesn’t depend on its cause in this straightforward way. And sometimes things depend on each other that aren’t causes at all.
When Dependence Isn’t Causation
Suppose you’re in a loud classroom, and your friend whispers something. You say “What?” loudly. Your friend says it again, louder. You hear it.
Did your first “What?” cause your friend to speak louder? Well, yes — but is that causation in the same way that a baseball breaking a window is? Lewis says no, because the events aren’t “distinct” enough. Your speaking and your friend’s response are tangled up together. Causation, he thinks, requires two genuinely separate events.
Similarly: you write “Larry” on a test. Did you write “rr” because you wrote “Larry”? Obviously. But writing “rr” isn’t a separate event from writing “Larry” — it’s part of it. So that’s not causation either.
The Problem of Preemption
Here’s where things get really interesting. Imagine two kids, Suzy and Billy, throwing rocks at a bottle. Suzy throws first. Her rock shatters the bottle. Billy’s rock sails through the space where the bottle was a moment later.
Now ask: did Suzy’s throw cause the bottle to shatter? Yes — obviously.
But now check the counterfactual: “If Suzy hadn’t thrown, would the bottle have shattered?” The answer is: yes, because Billy would have hit it. So according to the simple counterfactual test, Suzy’s throw didn’t cause the shattering! That’s the wrong answer.
This is called late preemption. The problem is that the actual cause gets masked by a backup cause that would have done the job. Philosophers have spent decades trying to solve this puzzle.
How Lewis Tried to Fix It
Lewis’s first move was to look at chains of events. Even if the bottle’s shattering doesn’t directly depend on Suzy’s throw, maybe there’s an intermediate step that does. Suzy’s throw causes her rock to fly through the air. That flying rock does depend on the throw — if she hadn’t thrown, the rock wouldn’t be flying. And the shattering depends on that flying rock — because by the time the rock is in mid-air, Billy has stopped throwing.
So: throw → flying rock → shattering. Each step depends on the previous one. Lewis says causation is just a chain of these dependences. Problem solved? Not quite.
The Problem of Late Preemption
Remember Billy and Suzy? Here’s why the chain trick doesn’t work for them. The flying rock does depend on Suzy’s throw. But does the shattering depend on the flying rock? Think about it: if Suzy’s rock hadn’t been flying through the air (if it had been a dud, say), Billy’s rock would still have shattered the bottle. So the shattering doesn’t actually depend on Suzy’s flying rock either.
This is late preemption: the backup cause cancels the counterfactual dependence at every step. Lewis tried a few different solutions to this. One was to say that events are very specific — the actual shattering, at exactly that moment and in exactly that way, wouldn’t have happened without Suzy’s throw. But critics said this makes too many things into causes. (If the poison killed your victim slowly because he ate dinner first, did dinner cause his death? That seems weird.)
Another Way: Causal Models
More recently, philosophers have developed a different approach using something called causal models. It’s a bit technical, but the basic idea is elegant.
Instead of just thinking about whether one event depends on another, you build a model of the whole situation — a kind of map of what depends on what. The model includes variables (like “Did Suzy throw?” “Did Billy throw?” “Did the bottle shatter?”) and equations that describe how they’re connected.
The trick is this: to test whether Suzy’s throw caused the shattering, you “freeze” the backup cause at its actual value (Billy’s rock didn’t hit the bottle) and then check the counterfactual. With Billy’s rock frozen as “didn’t hit,” the counterfactual “If Suzy hadn’t thrown, the bottle wouldn’t have shattered” becomes true. That’s how the model gets the right answer.
This approach has been really influential. It handles not just the simple cases but also trickier ones like trumping preemption — where, say, a major and a sergeant both shout “Advance!” and the soldiers advance because the major is higher-ranking, even though both commands reached them.
But What About Chance?
So far we’ve assumed that everything is determined — that given the cause, the effect must happen. But what about real life, where things are chancy?
You hook up a bomb to a Geiger counter. If the counter clicks enough times in ten minutes, the bomb explodes. You turn up the dial to make it more likely. The bomb explodes. Did your turning the dial cause the explosion? Yes — but it’s not true that if you hadn’t turned it, the bomb wouldn’t have exploded. It just would have been much less likely.
Lewis handled this by saying: a cause is something that raises the chance of its effect. If you hadn’t turned the dial, the chance of explosion would have been much lower. So the turning caused the explosion, even though it didn’t guarantee it.
The Queen and the Gardener
Here’s a puzzle that still bothers philosophers. You go on holiday. You ask your gardener to water your flowers. The gardener forgets. Also, the Queen (who lives nearby) doesn’t water your flowers either — but nobody expected her to. Your flowers die.
Did the gardener’s failure cause the flowers’ death? Most people say yes. Did the Queen’s failure cause it? Most people say no. But the counterfactuals are identical: if the gardener had watered them, they’d be alive; if the Queen had watered them, they’d be alive. Both failures have the same relationship to the outcome.
So why do we treat them differently? One explanation: we think of causation as involving something abnormal. The gardener breaking his promise is abnormal. The Queen not watering your flowers is totally normal. Maybe our concept of causation itself is sensitive to what’s normal or expected.
Another explanation: causation is really just about counterfactual dependence, full stop. The Queen’s failure is a cause — it just seems weird to say so because it’s not relevant or interesting. In a court case, nobody would blame the Queen. But that doesn’t mean her failure isn’t a cause — it just means we don’t bother talking about it.
Philosophers still argue about this. Some think causation is an objective fact about the world, independent of what we find normal. Others think causation is partly about our interests and expectations.
Still Unsettled
After all this work, there’s no single theory of causation that everyone agrees on. The counterfactual approach is powerful and intuitive, but it struggles with preemption, chance, and the role of normality. Different versions of the theory handle different problems, and philosophers keep finding new edge cases that challenge the latest version.
Maybe that’s the point. Causation is something we use all the time — every time you say “because” or “that’s why” — but when you try to pin down exactly what it means, it slips through your fingers like water.
Appendices
Key Terms
| Term | What it does in this debate |
|---|---|
| Counterfactual | A thought about what would have happened if something had been different; the basic building block of these theories |
| Causal dependence | The relationship between two events when one wouldn’t have happened without the other |
| Preemption | A situation where a backup cause is ready to produce the effect, so the actual cause doesn’t show counterfactual dependence |
| Structural equations | Mathematical models that map out how different variables depend on each other in a causal system |
| Intervention | A change that breaks the normal causal rules (like a “surgical” change to one variable) to test what else would change |
| Chance-raising | The idea that causes don’t have to guarantee their effects; they just have to make them more likely |
Key People
- David Lewis (1941–2001): A hugely influential American philosopher who developed the most famous counterfactual theory of causation. He also believed that other possible worlds are as real as our own — a view most philosophers find too weird to accept.
- Judea Pearl (b. 1936): A computer scientist and philosopher who developed the structural equations framework for thinking about causation. His work is used in AI, statistics, and medicine.
Things to Think About
- Think of a time when you said one thing caused another. Could there have been a “backup cause” that would have produced the same result? If so, does that change your judgment about what actually caused what?
- Is the Queen’s failure to water the flowers really a cause? If you think no, what makes something “count” as a cause? If you think yes, why does it feel wrong to say so?
- If we built a robot that could gather information and determine causes using counterfactuals, would it ever get the “wrong” answer compared to humans? What would that tell us about causation?
- Do you think causation is something real in the world, like gravity, or is it something our minds impose on the world to make sense of it?
Where This Shows Up
- In court: Judges and juries constantly decide what caused what. The counterfactual test — “would the harm have happened anyway?” — is central to legal reasoning about liability.
- In medicine: Doctors figure out what caused a patient’s illness by considering what would have happened with different treatments. Clinical trials are essentially giant counterfactual experiments.
- In AI and machine learning: Causal models (based on the structural equations approach) help computers learn not just patterns but genuine cause-and-effect relationships, which lets them make better predictions about what would happen if things changed.
- In everyday arguments: When you say “But if I hadn’t done that, it would have been worse!” or “That’s not really why it happened,” you’re doing counterfactual reasoning about causation.