Causation

In plain terms

One billiard ball strikes another and the second moves. Everyone knows the first caused the second, and nobody can say what that means.

Hume looked for the connection and found three things: the balls touched, one moved before the other, and this always happens. He could not find the necessity - the making-it-happen - anywhere in the scene, and concluded that the idea comes from our habit of expectation rather than from anything observed.

That is an uncomfortable result, because causation is doing enormous work everywhere. Medicine, law, engineering and history all turn on it, and each has had to develop its own criteria for deciding when it holds, because philosophy has not supplied one they can use.

As a philosopher would put it

Aristotle's four aitiai - material, formal, efficient and final - answer four different why-questions, and only the efficient corresponds to the modern notion, which is why translating them as causes misleads.

Hume's analysis reduces causation to contiguity, priority and constant conjunction, with necessity supplied by the mind's habit. The modern successors divide sharply. Regularity theories develop the Humean line, with Mackie's INUS conditions handling complex causes. Counterfactual theories, principally Lewis's, define causation by what would have happened otherwise, which handles single cases well and struggles with overdetermination and pre-emption. Probabilistic theories make causes raise the probability of effects. Interventionist theories, following Woodward and Pearl, define a cause as something that would change the effect if manipulated, which fits experimental practice closely.

The Indian traditions divide over whether the effect pre-exists in the cause - Sāṃkhya's satkāryavāda against Nyāya's denial - which is a question the European debate never quite poses.

What is actually disputed

Whether causation is one thing.

Each theory handles some cases and fails on others. Regularity accounts cannot distinguish causes from correlated effects of a common cause; counterfactual accounts stumble on overdetermination, where two sufficient causes occur; probabilistic accounts face cases where a cause lowers the probability of its effect.

Pluralists conclude that the word covers several relations - production, dependence, manipulability - and that the search for a single analysis was misconceived.

Russell's more radical position is that fundamental physics contains no causal notion at all, only functional relations between states, and that causation is a folk category that survives because it is useful at human scale. Cartwright's reply is that science is full of causal talk precisely because it is doing something laws cannot - and that the physics does not have the authority to abolish a concept every applied discipline requires.