Ebbinghaus's forgetting curve models memory retention as exponential decay: right after learning, recall is near-perfect, then it falls off quickly and then more slowly. Each short micro-session (a few minutes, timed to land while recall is still shaky but not gone) interrupts that decay — it pushes retention back up toward 1.0 and, more importantly, increases the memory's stability, so the next decay is slower. That progressive slowing is the spacing effect: many brief, well-timed reviews build durable memory more efficiently than one long cram session of equal total time, because a massed session gets diminishing returns from re-studying material that is still fresh.
R(t) = e^(-t / S)
S ← S · (1 + gain·(1 − R_at_review))
R_massed(t) = e^(-t / S_massed), S_massed grows only linearly with total minutes studied
- Review interval — the gap between micro-sessions. Too short wastes effort re-studying what's still fresh; too long lets retention collapse before the next review arrives.
- Micro-session length — longer sessions encode a bit more deeply per review, raising the stability gain, but the curve below shows the spacing schedule matters far more than session length.
- Simulation speed — compresses days into seconds so the multi-week spacing pattern is visible in one sitting.
- Memory tower — the glowing stack at the current time marker rises and falls with the spaced-learning retention level in real time.
Real-world relevance: this is the mechanism behind flashcard apps and spaced-repetition schedulers (Anki, SuperMemo) — they time each micro-session to arrive right as the forgetting curve crosses a target threshold, maximizing the stability gain per minute studied.