The Forgetting Curve

You cram a list Tuesday night, ace it Wednesday morning, and by Thursday half of it is gone. That is not a glitch, it is the default shape of memory. Hermann Ebbinghaus, working alone in the 1880s, memorized lists of nonsense syllables (meaningless three letter strings like DAX or WUX, chosen so prior knowledge could not help him) and then tested his own retention at intervals. He found the same pattern every time: a sharp, fast drop in retention in the first few hours and the first day, followed by a much slower decline that flattens into a long shallow tail. Most of the damage happens early; what survives the first day tends to stick around longer.

The shape matters because it is not a steady leak. A steady-timer model would predict a straight diagonal line, losing the same amount every hour forever. The real curve bends: picture roughly 100% retention right after learning, dropping fast to around 40-50% by one day, then easing down to maybe 20-25% by thirty days, with the line going almost flat near the end. That bend is the whole story: early forgetting is brutal, late forgetting is sluggish.

Relearning changes the curve's starting point and its slope. If you retest that same material once a week instead of never, each session pulls retention back up and the next decay starts from a higher point and falls more gently. Repeat this weekly and the curve gets flatter and flatter, closer to a plateau than a cliff. This is the mechanism behind spaced practice beating cramming: cramming front-loads everything into one fragile peak that obeys the steep early drop, while spaced review keeps resetting the curve before the steep part hits, building durable long-term storage instead of a memory that is strong at 5pm and gone by Thursday.