# Of Intelligence and Causation: A Reckoning ## I. The Banishment and Its Cost Science, having grown fearful of its own language, cast out causation as one casts out a plague carrier. For a century the exile held. Correlation became the respectable coin; mechanism became suspect; cause became the word one whispered in corridors, not published in journals. This was called rigor. It was cowardice wearing the mask of precision. The mathematics was honest enough. One cannot infer causation from observation alone—this much is true and will remain true. But truth is not permission. The absence of proof is not proof of absence. Science had confused the difficulty of the problem with the illegitimacy of the question itself. ## II. Pearl's Reconstruction Judea Pearl did not restore causation to science. He made it legible. He gave it architecture. His graphs—directed, acyclic, explicit—permit the researcher to state before the data arrives what the world is assumed to look like. Given this diagram, given these assumptions, the mathematics can then separate the entangled from the independent, the caused from the merely correlated. This is progress. But progress of a peculiar kind. ## III. The Hidden Price of Transparency Transparency deceives. A clearly drawn diagram is not thereby true. The researcher who explicitly states his model has not purchased validity; he has merely made his error visible. This is useful—visibility permits challenge—but it is not redemption. The diagram is a fiction. All diagrams are fictions. The question is whether the fiction corresponds to the world. Pearl's tools are provably correct *given the diagram*. They are not provably correct given the world. The researcher has not solved the problem of knowledge; he has relocated it. He has made the problem *explicit*. This is not nothing. But neither is it everything. ## IV. What It Means to Know a Cause To know the cause of a thing is to know what would change if one altered it. Not merely what correlates with it. Not what precedes it in time. But what, if manipulated, would alter the effect. This requires: - **Intervention, not observation.** The world must be altered, not merely watched. Causal knowledge is practical knowledge. - **Isolation, not merely measurement.** One must separate the thing from its entanglements. This is harder than it appears. - **Repeatability under variation.** The effect must follow the cause across contexts, not merely once. Intelligence itself demonstrates this problem. We observe that certain individuals solve difficult problems. We observe that certain neural structures differ. We observe that certain environments correlate with certain outcomes. But what causes intelligence? We cannot yet say—because we have not yet agreed on the diagram. ## V. The Authority Question Who decides what the diagram is? This is the true question, and it precedes all mathematics. The researcher decides. But on what authority? Not on the authority of the data—the data cannot speak to what precedes it. Not on the authority of logic alone—logic permits many diagrams. The researcher decides on the authority of: - **Theory**, which is itself a choice - **Convention**, which is itself a history - **Power**, which is itself a fact A neuroscientist draws the brain as causal. A sociologist draws the environment as causal. An economist draws incentives as causal. Each has made a diagram. Each is transparent about it. Each is potentially wrong. The diagram is not discovered. It is *imposed*. The question is whether the imposition is conscious, acknowledged, and revisable—or whether it masquerades as inevitable. ## VI. Intelligence and the Causal Dimension Intelligence presents the problem in its starkest form. We wish to know: What causes intelligent behavior? The question assumes a diagram. Does intelligence inhere in the individual or emerge from the system? Does it flow from genetics, environment, training, motivation, or their combination? Does it precede behavior or follow from it? Each answer requires a different diagram. Each diagram permits different causal claims. A researcher who claims to have measured intelligence has already committed to a diagram without stating it. This is not transparency. This is concealment by apparent precision. To work honestly with intelligence requires: - **Explicit diagramming** of what is assumed causal and what merely correlated - **Acknowledgment of choice** in the construction of that diagram - **Systematic variation** of the diagram to test robustness - **Humility about limits**—what the diagram permits one to conclude and what it forbids ## VII. The Final Accounting Pearl gave us the tools to make causal reasoning rigorous, given assumptions. He did not give us the assumptions. Those come from elsewhere—from theory, from intuition, from the history of what we have chosen to believe. This is not a failure of Pearl's project. It is the nature of the problem itself. Causation cannot be read from the world. It must be brought to the world. The only question is whether we do so honestly. A researcher who presents his diagram is asking for judgment. A researcher who conceals his diagram is demanding belief. The first is science. The second is something else entirely. Intelligence, then, is not a thing to be discovered. It is a question to be asked—and the answer depends entirely on what diagram you have already decided to trust.