Mechanism Eliminates External Verification

Definition

A procedure learned as a memorized sequence of steps offers no way to tell, from the inside, whether you've applied it correctly — a single misremembered step produces a wrong answer that looks the same as a right one, so you must ask someone else to check. A procedure learned as an understood mechanism lets you verify it yourself, the way understanding why a hot stove burns means you never need to be told twice not to touch it, and never need to ask whether touching it "was right."

In the Book

The book opens Chapter 1 by diagnosing why students struggle with school multiplication methods (grid method, long multiplication, and others): "the methods require memorisation of a number of steps, which, if any are wrong, makes the answer incorrect... it is not possible to know yourself if it is incorrect. You have to ask someone else." It cites the recurring student question "Is this right, Miss?" as a symptom, and argues that memorizing steps without understanding trains students "to act like a robot," dependent on a "controller" to say whether they got it right. Against this, Carson gives the hot-stove analogy — nobody has to memorize a rule against touching a hot stove, because they understand the consequence — and uses his cat's one accidental encounter with a hot stovetop as a real-world instance of the same principle. The chapter's stated method is to make the reader derive "what is multiplication?" themselves, on the claim that self-derived, mechanism-level understanding is what makes a fact "impossible to forget."

Why It Matters

This reframes "understanding vs. memorization" from a vague virtue into a concrete, checkable property: does the learner have an internal way to test their own output, or must they route every check through an external authority? That distinction generalizes past arithmetic to any skill taught as a procedure — a checklist, a recipe, a compliance rule — where teaching the underlying mechanism (not just the steps) is what lets the learner self-correct, adapt the procedure to novel cases, and stop needing permission to trust their own answer.