Why it works
Recognition memory is cheap. Production memory is expensive. You can recognize the mitochondria diagram, nod at the phrase “powerhouse of the cell,” and still be unable to explain, under exam pressure, why cells with high energy demand have more mitochondria per cell.
Explain-back is the fastest way to distinguish these two kinds of memory. It costs a paragraph and thirty seconds. The score is a diagnosis of a gap you didn’t know you had. And every gap becomes a card.
How to use it well
Don’t pick your most-confident concept — pick your second-most-confident. That’s where the interesting gap lives. Aim for a 4, not a 5; aiming for 5 makes you defensive, aiming for 4 makes you honest.
And come back a week later — explain the same concept back again, watch the score climb. That climb is the shape of your understanding compounding, measurable in a single integer.
The Roediger & Karpicke result
The single most-replicated result in learning science is that retrieval — pulling a memory back out into the open — is orders of magnitude more powerful than re-reading — passing your eyes over the material again. This has been shown so many times, in so many populations, across so many decades, that it has crossed from “interesting finding” to “engineering constraint on how to build a study tool.”
Explain-back is retrieval, forced. You do not get to peek. You do not get to skim. You must produce the concept from scratch in your own words, at the cost of a scored diagnosis, in the presence of a grader that will not let you off with keyword salad.