Spacing Out: A Field Guide to the Spacing Effect

2026-03-02 status: finished importance: 7 confidence: highly likely memoryspaced repetitionforgetting

Hermann Ebbinghaus spent the 1880s memorizing thousands of nonsense syllables — zux, bik, tol — alone, in his apartment, with no theory to guide him except the hunch that memory could be measured. One of the regularities he found has never stopped replicating: material studied in several separated sessions is retained far longer than the same material studied for the same total time in one sitting.^[1] A century of follow-up work, across essentially every kind of material tested — vocabulary, surgical technique, mathematics, motor skills — has not found an exception to the basic pattern.^[2]

The effect is large. In a 2008 study of over 1,300 subjects learning trivia facts, the optimal gap between study sessions, for material that needed to be retained a year later, was around 20–30% of the retention interval — meaning if you need to remember something in a year, the best time to restudy it is roughly two to three months out, not the next day.^[3] Cramming the night before a test can produce recall that rivals spaced study — on the test itself. A week later the spaced group remembers substantially more, and the gap widens with time.^[4]

Why it works: desirable difficulty, not just repetition

It would be easy to assume spacing works simply because it adds more elapsed time for review, but massed practice with the same number of repetitions and the same total study time underperforms spaced practice, so raw repetition count is not the mechanism. The account with the most support is Bjork's study-phase retrieval theory: when a study session is spaced far enough apart, the material has partially decayed, so restudying it requires a small act of retrieval — reconstructing a slightly faded memory is effortful in a way that rereading a still-fresh one is not.^[5] That small effortful reconstruction is what strengthens the trace. Cram the same material a few minutes later, before any forgetting has set in, and there is nothing to reconstruct — the second exposure is nearly free, and it shows: it barely helps.

This is also why spacing feels bad while it works. By the time you're ready to restudy, you have forgotten enough that the retrieval attempt is uncomfortable — you hesitate, you get some of it wrong, it does not feel like "good" studying compared to a fluent re-read. That discomfort is not a bug in the technique; on this account it is the technique.

Spaced repetition software

Flashcard programs built around this finding — SuperMemo, Anki, and their descendants — try to schedule each card's next review right at the point where recall probability has dropped to some target, often engineered around 90%, so that every review is a moderately effortful retrieval rather than either a trivial one (reviewed too soon) or a failed one (reviewed too late).^[6] The scheduling algorithms have gotten more sophisticated over decades — from fixed multipliers to statistical models fit on millions of real review logs — but the underlying bet is the same one Ebbinghaus's data supported: intervals should grow as the memory strengthens, and the growth rate should be tuned to keep retrieval right at the edge of failure rather than comfortably inside it.

Caveats

The spacing effect is one of the most replicated findings in cognitive psychology, but the optimal-gap literature is noisier than the headline finding suggests — optimal spacing depends on how long you need to retain the material, how well you already know it, and how forgiving the material is of partial recall, and most of the precise numbers come from lab tasks with artificial material rather than the messy stuff people actually try to learn.^[7] Treat any specific ratio, including the 20–30% figure above, as a starting heuristic rather than a law.