What Is Spaced Repetition and How Does It Work?

Spaced repetition is a way of deciding when to return to material: reviews are spread out, and items that are still hard usually come back sooner than items you can retrieve easily.

It is a timing method, not a flashcard format. Anki is one implementation. A calendar such as 1-3-7 is a starting heuristic, not a law of memory. You can space reviews of a lecture note, a worked example, or a diagram the same way you space a deck.

The research finding behind the method is the spacing effect: spreading study or review over time usually supports later recall better than spending the same time in one block. Cramming can still help a test that is very soon. The spacing effect is the research finding. Spaced repetition is a practical system that tries to use that finding by choosing the next review date.

When you do return, the usual useful activity is retrieval: try to recall before looking. Spacing answers when to return. Retrieval practice answers what you do when you return. There is strong experimental support that spacing and retrieval usually improve later retention compared with massed rereading. Results still depend on the material, the delay, and what you do during review.

This page is the definition and the mechanism. It is not a product setup guide, and it does not pick a universal interval list.

What spaced repetition is

Spaced repetition means revisiting material across separated sessions rather than packing the same reviews into one sitting.

Take a short page on how osmosis works: water moves across a semipermeable membrane toward the side with higher solute concentration. In a plant root, that movement helps the plant take up water.

Massed practice is four explanations of that chain in one evening. Spaced practice is returning on later days and trying the chain again. The second pattern is usually the stronger bet for still having the idea next month. The first can still help a quiz tomorrow.

Two more details belong in the practical definition, not as extra slogans:

• Hard items come back sooner. If you cannot reconstruct the direction of water movement, a system that waits three weeks is wasting the signal. If the chain is easy, reviewing it again tomorrow is often a poor use of time.
• A review is a return, not a reread by default. Opening the page and recognizing the heading is not the same as producing the mechanism with the source closed.

Flashcards are a convenient unit for that return. They are not what makes the method “spaced repetition.”

Spacing effect, retrieval, and a scheduler are not the same thing

People hear one phrase and get four jobs glued together. Separate them.

The spacing effect does not require software. You can get a spacing benefit with a paper calendar. Spaced repetition is the system that keeps choosing the next date instead of leaving the return to chance.

Retrieval is a different job. Testing yourself five times in one evening is retrieval with almost no spacing. Opening the same notes every few days without trying to recall first is spacing without retrieval. Retrieval and spacing are different jobs; this page does not restage that comparison. The useful pairing is to retrieve during reviews that are distributed over time.

A scheduler is not the science. Leitner boxes, SM-2, and FSRS are different ways of picking a date. None of them is the spacing effect.

A 2013 review of ten common study techniques rated distributed practice and practice testing as high-utility — two separate techniques, not one brand name. Utility is not “always wins.” Highlighting and rereading can still support a first pass. They are a weak complete method if they never include a delayed retrieval attempt.

If you want the encoding-to-retention loop rather than the definition — understand, retrieve with the source closed, check, wait, sleep — that path is how to remember what you study.

How spaced repetition works

A later return is often a harder test of what is still available. That is why delay can help: the attempt is no longer echo memory from the same sitting. Effortful retrieval after a delay is useful when you can still succeed. Guessing at something you never understood is not retrieval practice.

You do not need a cellular story about synapses to use the method. You need a cue, an attempt, a check, and a next date.

The review loop

The loop is the same whether the unit is a card or a page.

1. See a cue. A question, a heading, a blank page, a problem.
2. Try to retrieve before looking. Produce the idea: the direction of water movement, why the root takes up water, where the explanation goes vague.
3. Check a reliable source. Correct the miss on the material itself.
4. Rate difficulty. How hard was the recall? This is a record of the attempt, not a hidden score of whether you are a good student.
5. Get a next date. Harder recall usually means a sooner return. Easier recall usually means a later one.

Without the attempt, you are scheduling rereading. Without the check, errors get practiced. Without a next date, you are back to hoping you will remember to come back.

The craft of cues lives in active recall with notes. The craft of returning to pages lives in spaced repetition for notes.

Why later reviews often get farther apart

Many apps lengthen the gap after a successful recall because that is a practical way to spend less time on stable items. That does not mean expanding intervals are always better than evenly spaced reviews. Evidence on expanding versus equal spacing is mixed; one often-cited comparison is one reason not to treat “always expand” as a law.

Useful spacing depends partly on how long you need the knowledge. A calendar like 1-3-7 can be a starting heuristic, not a law of memory. A schedule that fits a test next week is not automatically right for knowledge you want next year.

“Review just before you forget” is a metaphor. Apps estimate when recall is getting harder. They do not detect the exact moment a memory will disappear.

How common schedulers implement the idea

Ebbinghaus showed that memory performance can drop over time and that later relearning can still be faster. That is not a personal countdown clock, and it is not the same as a modern review app. He did not invent spaced repetition. The forgetting curve is that pattern and its misreadings — not a setup tutorial for a scheduler.

What later systems added was a rule for the next date, usually from a signal about how recall went.

Leitner boxes. In the 1970s, Sebastian Leitner popularized a simple box system for paper flashcards: if you remember a card, it moves to a box you review less often; if you forget it, it moves back. The scheduler tracks which box the item is in. That is frequency, not a model of the brain.

SM-2. SM-2, used in early SuperMemo and later adapted by other apps, stores how easy an item has been and multiplies the next interval after a successful review. Anki’s older scheduler is based on SM-2 but is not identical to the 1980s description. The scheduler tracks ease and interval. SuperMemo 1.0 is dated 13 December 1987. That is the history of a program, not a claim that one person invented the spacing idea.

FSRS. FSRS estimates how stable a memory is and how likely recall is right now, then schedules the next review toward a chosen retention target. That is a model of scheduling, not a proof of what is happening in the brain. Some apps, including Anki, offer FSRS as a later option beside older SM-2-style scheduling.

These families differ in what they store. They share the same job: pick a date so you spend fewer reviews on what is holding and more on what is not.

This is a map, not a setup guide. It does not tell you which Anki options to click, and it is not a tutorial for any one formula.

You do not need flashcards

A flashcard packages a cue, a retrieval target, and a check into a small unit. That is why cards fit schedulers so well. It is a format choice, not the definition.

You can leave the osmosis page as a note, cover it, reconstruct the chain, check, and come back later. Notes versus flashcards is which object you review, not which method you picked. For card-free returns — free recall, a sketch, an explanation, a topic review — see spaced repetition without flashcards.

If the main job is thousands of atomic cards, Anki is often the better-fitting tool. You do not have to convert a connected explanation into a deck in order to space it.

What spaced repetition does not do

It does not replace the first pass. If you cannot follow osmosis with the source open, closed-source guessing is not the next step.

It does not beat every near-term test. Cramming can still help a test that is very soon. Packed practice can look efficient on the day you do it. Spreading the same effort usually looks better later. Those are different deadlines.

It does not give you a universal scientifically correct interval list. 1-3-7 is a heuristic. Expanding gaps are a common practical pattern, not a proven universal optimum.

It does not detect the moment a memory will vanish. A scheduler estimates. A box system approximates. Neither is a sensor on your memory.

It does not, by itself, tell you whether you were correct. Many apps ask you how the recall felt, or whether you remembered the card. That signal updates a date. It is not hidden marking of your knowledge unless the product is actually scoring an answer.

It does not decide what is worth remembering. A queue will keep returning whatever you put in it. Selectivity is your job.

How Yuhimo fits this idea

Yuhimo is a notes-and-learning app. The unit is the page, not a flashcard. A page can stay an ordinary note. You can schedule a note itself for review. Questions are optional prompts, not a requirement to convert the page into cards. Core concepts is the product model; the notes-first walkthrough is the product page.

When a page is due, the review is retrieve-then-check, not a reread reminder. The prompt is “How well did you remember this?” Yuhimo does not mark answers right or wrong. You rate how well you remembered the page as Forgot, Barely, Remembered, or Easily, and that rating updates the next review date. Self-assessment is the signal. It is not an automatic examiner.

Yuhimo schedules learning pages with FSRS-light, a simplified model inspired by spaced repetition. It is not the full FSRS algorithm used in Anki.

If the main job is thousands of atomic cards, Anki is often the better-fitting tool. Yuhimo is built for notes you optionally put on a review schedule.