Philosophy of Science: Causation and Mechanism
Why 'What Caused It?' Is a Harder Question Than Science Admits
Every time you say one thing caused another, you are making a philosophical bet that science alone cannot cash.
The Idea
We tend to treat causation as the bedrock of scientific explanation — find the cause, and you have understood the world. But causation is one of the strangest, most contested concepts in all of philosophy, and the more carefully you look at it, the less solid it becomes. The Scottish philosopher David Hume made the classic move in the 18th century: he pointed out that when we observe one billiard ball striking another, we never actually see the causation. We see sequence — one event, then another — and our minds supply the 'because.' Causation, Hume argued, is a habit of expectation, not a feature of reality we can directly perceive. Modern philosophy of science has pushed further. We now distinguish at least two different things we might mean when we ask 'what caused X?' The first is a correlation or regularity — A reliably precedes B. The second is a mechanism — there is some identifiable process by which A brings B about. Mechanisms feel more satisfying, more real. But pinning down what makes something a genuine mechanism, rather than just a very detailed correlation, turns out to be extraordinarily difficult. The philosopher James Woodward proposed an influential test: a cause is something you could intervene on. If you change A, does B change? This interventionist account has reshaped how scientists think about causal inference. But it raises its own puzzles — what counts as a legitimate intervention, and can this framework handle the deep causes that nobody can manipulate, like the Big Bang? None of this means causation is an illusion. It means the concept is doing enormous philosophical work that most of us never pause to examine.
In the World
In the 1950s, a British epidemiologist named Austin Bradford Hill was trying to convince the medical establishment that smoking caused lung cancer. The statistical correlation was overwhelming, but correlation had a bad reputation — everyone knew it didn't prove causation, and the tobacco industry was delighted to exploit that gap. Hill's response was to formulate what became known as the Bradford Hill criteria: a set of features that, taken together, could make a causal claim credible even without a randomised controlled trial. These included the strength of the association, its consistency across different populations, its specificity, and — crucially — biological plausibility. Did we have a mechanism that could explain how tobacco smoke would damage lung tissue? What Hill was doing, without using the philosopher's vocabulary, was arguing that causation isn't a single thing you either have or don't have. It's a cluster of evidence types, and a mechanism is one strand in that cluster, not the whole rope. Decades later, this tension still plays out in live scientific debates. Long COVID researchers face exactly the same problem: strong epidemiological signals, contested or poorly understood mechanisms, and a public and institutional tendency to treat 'we don't fully understand the mechanism' as equivalent to 'we're not sure it's real.' The philosophy underneath these disputes isn't abstract — it determines which patients get believed and which treatments get funded.
Why It Matters
Most of us encounter causal reasoning every day, often in high-stakes contexts: a news story reports that a lifestyle factor raises risk of disease, a manager concludes that one policy caused a business outcome, a friend explains why their relationship ended. The question of whether we're dealing with genuine causation or sophisticated correlation shapes decisions in medicine, policy, and everyday life. Having a sharper mental model of what causation actually requires — regularity, mechanism, the ability to intervene — gives you a useful instinct for when to push back. Not cynically, not reflexively, but with genuine curiosity: 'Is there a mechanism here, or just a pattern?' 'What would it look like if we intervened?' There's also something quieter and more personal in this. Hume's original insight — that causation is partly something the mind constructs — sits oddly close to certain Buddhist and contemplative insights about how the mind stitches experience into narrative. We are meaning-making machines, and causation is one of our most powerful and least examined tools. Noticing the tool is not the same as doubting reality. It's just the beginning of thinking more carefully about it.
A Question to Ponder
When you explain something that happened in your own life — a decision you made, a relationship that changed, a feeling that arrived — how much of the 'cause' you name is mechanism, and how much is a story your mind found satisfying?
Get a new one of these every morning.
Start learning with Thinkable