Logic as the Foundation of Thinking: How Reasoning Works

A course about how reasoning is actually built — and how to stop making decisions because a story sounded good.

WHO IT IS FOR. People who decide on incomplete information and would rather not rely on gut feel: analysts, managers, researchers, founders, doctors, lawyers, product managers. And anyone tired of sensing that an argument went wrong without being able to say where.

WHAT YOU WILL LEARN. Break an argument into premises and conclusion and find the exact joint where it fails. Tell validity from truth, and see why a valid argument can carry you to a false conclusion while an invalid one lands on a true one. Read "if A then B" the way logic means it rather than the way conversation does. Catch quantifier-order mistakes: "everyone fears something" and "there is something everyone fears" are different claims, and they get confused constantly. Surface hidden premises, presuppositions and framing. Work Bayes' theorem by hand and understand why a positive result on a rare medical test usually still means you are healthy. Spot p-hacking, multiple comparisons, and effect size traded for significance.

WHY IT MATTERS IN LIFE. You stop buying "studies show" and start asking what the sample was, how big the effect was, what the base rate is, and whether it replicated. That saves money on decisions made under a confident tone, and hours in arguments where both sides mean different things by the same word.

WHY IT MATTERS AT WORK. Being able to say "this metric does not measure what we think it measures" — and defend it — is rare and expensive. It separates someone who executes from someone trusted with decisions. The course also covers how to disagree so that you are heard: burden of proof, the principle of charity, steelmanning the strongest version of the other position.

WHAT MAKES THIS COURSE DIFFERENT. It is checkable. From the first lesson you keep a forecast journal: predictions with explicit probabilities, then your own calibration and Brier score. Objective feedback instead of a feeling that something clicked. The exercises have right answers known in advance: Bayesian problems, finding the substitution in a real text, taking apart a paper with broken statistics.

HONEST ABOUT LIMITS. This course does not promise you will become rational "in general". It gives an apparatus for specific problems and says plainly where the apparatus stops: some problems cannot be solved in reasonable time at all, and there are situations where a fast heuristic beats full analysis. Pointing at the form of someone's mistake does not refute their conclusion — that gets its own discussion.

THIS IS PART ONE OF TWO. Here you learn how reasoning is built and how to read data. Part two, "Decisions Under Incomplete Information", is about what to do with it: causality and why correlation does not license intervention, decision theory and game theory, models and systems, whom to trust and when, and where language models fit into thinking. The forecast journal you start here closes there — you will plot your own calibration curve over a real stretch of predictions.