Spacing Effect: Why Two Short Sessions Beat One Long One

A second study session is worth more when the first has had time to fade. Why is still debated; three accounts have support. One is about retrieval: reread a list five minutes after learning it and the words are still fresh, but a day later some have to be dug out, and on this account the effort of digging strengthens the memory. One is about context: the second session happens in a different room, mood or time of day, and a memory tied to two contexts has more routes back to it than a memory tied to one. One is about processing: material that is still fresh gets little attention the second time. The accounts differ on the mechanism, but on all of them the gap is doing work rather than wasting it.

The effect is one of the oldest and most reliable in experimental psychology. Hermann Ebbinghaus found it in himself in the 1880s, and Cepeda, Pashler, Vul, Wixted and Rohrer's 2006 review found 839 comparisons of spaced and massed study in 317 experiments. Across the 271 comparisons that pitted spaced presentations directly against massed ones, average recall was 47 percent after spaced study and 37 percent after massed, and the advantage grew with the delay to the test: small when the test came within minutes, large when it came after days.

The size of the right gap depends on how long the memory has to last. Cepeda and colleagues' 2008 study put 1,354 people through 26 combinations of gap and test delay and found that for each delay there was a best gap, beyond which more spacing started to hurt. The best gap grew with the delay, from about one to three days for a test a week away to about three weeks for a test a year away, while as a share of the delay it shrank, from roughly 20 to 40 percent of a week to 5 to 10 percent of a year.

The catch is that spacing feels worse. The massed session ends with the material fluent and fresh, which learners read as mastery; the spaced session begins with forgetting, which they read as failure. In Kornell's 2009 flashcard study, spacing worked better for 63 of all 70 learners, 90 percent. Yet at the end of the first session, 58 learners judged one method better, and 42 of them, 72 percent, picked massing. The same misjudgment makes rereading feel better than testing, and the two effects are usually put to work together.

Hermann Ebbinghaus published Memory: A Contribution to Experimental Psychology in 1885, reporting years of experiments with himself as the only subject. He learned lists of nonsense syllables until he could recite them without error, and measured memory by how many repetitions it took to relearn a list later, a method he called savings. In the chapter on repetition he compared two schedules for a twelve-syllable list. Sixty-eight repetitions in a single sitting left the list in a state where it could be relearned the next day in seven repetitions. Thirty-eight repetitions spread across the three preceding days left it relearnable in six. His conclusion was that with any considerable number of repetitions, a suitable distribution of them over a space of time is decidedly more advantageous than massing them at a single time.

The modern synthesis is Cepeda, Pashler, Vul, Wixted and Rohrer's 2006 review in Psychological Bulletin, "Distributed practice in verbal recall tasks." From 184 articles they extracted 317 experiments and 839 assessments of distributed practice, separating the spacing effect, spaced against massed presentations, from the lag effect, longer against shorter gaps. Spaced study won almost everywhere, and the analysis of lag showed that the gap producing the best retention grew as the retention interval grew, a pattern that previous reviews had not isolated.

The same group tested that pattern directly in 2008. Over the internet, 1,354 people learned 32 obscure facts, returned for a review session after a gap of up to three and a half months, and took a final test 7, 35, 70 or 350 days after the review. For each test delay, recall rose with the gap up to a point and then fell. The best gaps among those tried were 1, 11, 21 and 21 days for the four delays, and the improvements over a zero gap were 10, 59, 111 and 77 percent. The authors described the pattern as a ridgeline: a line of best gaps that climbs with the delay while shrinking as a fraction of it, and they noted that the gaps used in most laboratory studies, minutes rather than days, sit far below it.

The rows below are illustrations, not findings from the studies cited on this page.

The spacing effect is a benefit, so the thing to counter is the habit of ignoring it because massing feels better.

• Fix the total time and divide it. Two sessions of twenty minutes instead of one of forty, the plan in the demo above, is this page's illustration rather than a schedule any study ran; for verbal material that has to last a week, a split of that kind is generally expected to help, not guaranteed to.
• Match the gap to the deadline. For a test in a week, a day or two between sessions; for a month, several days; for material you want to keep for years, weeks. The 2008 ridgeline is the guide.
• Expect the second session to start badly. On the retrieval account of the effect, some forgetting is part of why the return helps; on any account, it is not a sign that the first session failed.
• Judge the schedule by delayed results, not by how fluent the material feels at the end of a sitting.
• Combine spacing with retrieval. A spaced session that tests rather than rereads gets both effects, which is what the better flashcard systems are built to do.

The spacing effect is often run together with other findings about how study works.

• It is not the testing effect, which concerns what you do in a session: retrieving material beats rereading it. The spacing effect concerns when sessions happen. The two combine, and spaced retrieval is the schedule with the strongest support.
• It is not the serial-position effect, the better recall of the first and last items in a list. That effect is about the order of items within one session; spacing is about the gap between sessions, and it helps items in every position.
• It is not the levels-of-processing effect, where material processed for meaning is retained better than material processed for its surface. Spacing works for shallow and deep processing alike, and adds to whatever depth of processing contributes.