Maggie's Farm

In fact, causation is an inference that depends on correlation.

Consider a simple example: You push a lever and a gumdrop comes out. You propose the hypothesis that pushing the lever causes the gumdrop to come out. You do it a bunch of times and the gumdrop keeps coming out, so the correlation supports the proposed cause. A falsification would be push the lever and no gumdrop comes out, or don't push the lever and no gumdrop comes out. But falsification itself requires observations of patterns—that is, correlations. Repeated patterns of correlation, experimental control, and elimination of competing explanations, can reasonably justify causal inference. But it's still an inference.

Causation has been a subject of considerable debate, from Aristotle to Hume to the wrinkles of quantum theory. Menton didn't consider "alternative hypotheses or potentially conflicting evidence" about the nature of causation, because his purpose is polemics. Menton was not just wrong, but used his wrongness to cast aspersions.