To beat the forgetting curve as an adult, do five things:
- Choose how long you want the knowledge to remain usable. A fact needed next week and an idea you want for ten years should not share a calendar.
- Retrieve before you review. Close the book and answer a question from memory; reopening it is checking, not the attempt.
- Leave enough time for recall to require effort. A review that feels automatic has arrived too soon to tell you much.
- Adjust from performance. After a clean recall, lengthen the next gap. After a miss, correct the answer and shorten it.
- Put the questions and dates somewhere outside your head. A memory system that depends on remembering to run it has a fairly obvious design flaw.
That is the method. The familiar 20-minutes-then-one-day calendar is not a law of memory, and the forgetting curve is not a schedule. It is evidence that unused memories become harder to retrieve over time.
What does the forgetting curve actually tell adults?
It tells us something narrow and useful: memory drops quickly after learning, then the rate of loss slows. It does not tell every adult exactly what percentage of a meaningful book will disappear at 3 p.m. tomorrow.
Hermann Ebbinghaus built the original curve in the 1880s by testing himself on lists of artificial syllables. A replication published in PLOS ONE on 6 July 2015 also used one participant, who spent 70 hours learning and relearning syllable lists after delays from 20 minutes to 31 days. The researchers recovered the same broad steep-then-flat pattern (Murre & Dros, 2015).
That is strong evidence that forgetting has a shape. It is weak grounds for the precise internet claim that every person forgets some fixed percentage of every subject in 24 hours. Meaning, prior knowledge, difficulty, attention, and what you need to do with the material all change the job.
If your problem is more basic—nothing remains even seconds after you close the book—start with the one-minute test that separates failed learning from later forgetting. A review schedule cannot preserve something that never got into memory. This article starts after you can explain the idea once and want to keep it available.
How often should you review so you do not forget?
There is no context-free best interval. Set the gap from two inputs: how long you need the knowledge and whether you could retrieve it last time.
A 2008 Psychological Science experiment makes the first input unusually clear. More than 1,350 people learned facts, reviewed them after gaps as long as 3.5 months, and took final tests as late as one year. For each final-test delay, increasing the study gap helped until it became too long; then performance declined. Most importantly, the best gap grew as the desired retention interval grew (Cepeda et al., 2008).
So work backwards from the horizon:
- Need it next week? Use shorter gaps. You have little room to discover a miss late.
- Want it next year? Allow wider gaps and repeated retrieval over months. Reviewing everything daily wastes effort on memories that are already available.
- No deadline because you learn for pleasure? Use a sustainable cadence and let performance steer it. The system has to survive ordinary weeks more than it has to resemble a perfect graph.
Then use the result of each attempt. If you can explain the idea accurately without clues, move the next review farther away. If you cannot, look up the answer, correct it immediately, and bring the next attempt closer. You do not need to estimate your personal forgetting curve. Each retrieval attempt measures the only part that matters: whether the knowledge is available when asked for.
For a concrete chapter-level calendar you can run by hand, use our spaced repetition method for books without flashcards. That page gives the full interval mechanics; its numbers are a practical starting point, not a biological constant.
What should you do during each review?
Produce first; inspect second. Write what the idea means, answer a question, solve a fresh problem, or explain the argument aloud with the source closed. Only then reopen the book and compare.
In two 2006 experiments, students either restudied prose passages or took free-recall tests. Restudying produced better results on a test five minutes later. On tests after two days and one week, prior testing produced substantially greater retention—even though repeated study made learners more confident that they would remember (Roediger & Karpicke, 2006).
That reversal explains why rereading is seductive. It improves immediate fluency, exactly when the page is familiar. A closed-book attempt feels worse because it exposes what is missing. For durable learning, exposure is not the score; successful retrieval after a delay is.
Use feedback after the attempt. A failed retrieval followed by no correction merely rehearses uncertainty. Check the source, name the error precisely, and try again later. For technical material, prefer a new problem over reciting a definition: being able to use an idea is stronger evidence than being able to recognize its wording.
Do review intervals always need to expand?
No. Expanding intervals are convenient, not sacred.
A 2021 meta-analysis combined 29 studies of spaced retrieval. Spaced retrieval strongly outperformed massed retrieval (g = 0.74), but expanding schedules did not significantly outperform uniform schedules overall (g = 0.034). The authors concluded that the evidence does not support the widespread belief that retrieval intervals must always grow before the final test (Latimier, Peyre & Ramus, 2021).
The robust rule is do not mass all retrieval into one sitting. Whether the gaps are equal or expanding matters less consistently. Expanding them still makes practical sense after repeated success: it reduces workload while testing the memory over longer delays. But if an idea remains fragile, holding the interval steady—or shortening it after a miss—is not breaking the method. It is responding to evidence.
This is also why a rigid calendar becomes absurd at scale. Different chapters fail at different rates. The useful schedule branches according to what you could actually retrieve, instead of marching every idea through the same dates because a diagram said so.
How do you run spaced retrieval without maintaining it?
The free version is complete: write a question when you finish a chapter, put future attempts on a calendar, answer with the book closed, and reschedule from the result. If you keep that system, you do not need software.
The cost is maintenance. You have to decide what deserves a question, write it before you know what you misunderstood, preserve enough context to grade the answer, and keep the calendar alive while starting new books. For many adult learners, forgetting the material and forgetting to run the system are the same problem at different levels.
Study Junkie moves those jobs into the course. Upload a PDF, DOCX, text file, or book and it becomes a chaptered course built from the full source. Recall questions are generated from the chapter rather than from whatever you happened to highlight. You choose the review day offsets, and completed chapters return as recall sessions by email or push notification. Assignments with written feedback add the part a bare memory calendar cannot: checking whether you can use the idea, not merely repeat it.
The trade-off is simple. Paper and calendar reminders are free; Study Junkie uses credits. Software also cannot make you answer the notification. It can remove the prompt-writing and calendar administration, which leaves the one job no memory system can outsource: attempting the recall.
For the broader reading loop around this method, see how to remember what you read. Or start with 1,500 free Study Credits—no card, no subscription, and the credits do not expire while you decide which book is worth keeping.
FAQ
Is there one best spaced-repetition schedule?
No. The useful gap changes with how long you want to retain the knowledge and how successfully you recalled it last time. Cepeda and colleagues found in 2008 that the optimal study gap grew as the final retention interval grew. Treat fixed interval lists as starting points, then lengthen after success and shorten after a miss.
How often should an adult review something to remember it for a year?
Review it repeatedly across months, using closed-book retrieval and widening gaps after successful attempts. Research supports matching wider gaps to longer retention goals, but it does not supply one universal set of dates for every person and subject. A system that adapts to your recall is more defensible than pretending a fixed six-date calendar guarantees a year.
Does rereading reset the forgetting curve?
Rereading gives another exposure, but it is a poor test of whether the memory is retrievable. In the 2006 experiments above, restudying helped at five minutes while testing produced better retention after two days and one week. Try to answer first; use rereading only to check and correct what the attempt exposed.
Should review intervals always get longer?
No. The 2021 meta-analysis found no general advantage for expanding over uniform retrieval intervals. Longer gaps are efficient after success; equal or shorter gaps are reasonable when recall remains weak. Spacing the attempts matters more reliably than forcing every interval to expand.