Why you forget
Forgetting is fast at the start and slow later. In 1885 Hermann Ebbinghaus learned lists of nonsense syllables and measured how much work it took to learn them again. After 20 minutes, 58 percent of the first work was saved. After one day, 34 percent was saved.
A team in Amsterdam repeated the experiment in 2015 and got a curve of the same shape. The numbers are for meaningless syllables. Material that means something to you fades more slowly, but it fades in the same way. The article on the forgetting curve has the full table.
So the question is not how to learn something once. It is what to do in the days after, when the curve is steep.
The four methods
| Method | What you do | What it fixes |
|---|---|---|
| Recall | Pull the answer out of memory before you look at it. | Reading feels like knowing. Recall shows what you can use. |
| Spacing | Come back after a gap, and make the gaps longer. | Study in one sitting is lost quickly. |
| Mixing | Practice different kinds of items in one session. | Practice in blocks never teaches you to tell the kinds apart. |
| Feedback | Check each answer and see the right one. | An error that is not corrected stays. |
Each method makes study feel harder. That is not a fault. The psychologist Robert Bjork calls such conditions desirable difficulties: they slow you now and they make the memory last. The article on desirable difficulty says when a difficulty helps and when it does not.
1. Recall, do not reread
Close the book and say what was in it. In a 2006 study, students who read a text again did better on a test five minutes later. Students who tried to recall the text did better two days later and one week later.
The students who read again were also more sure of themselves. This is the trap: text that you have seen before feels known, and that feeling is not a measurement of memory.
More on this in retrieval practice.
2. Space it
One long session is weaker than the same time in parts. A 2006 review of the research on spaced study collected the experiments that compared the two, and study that was spread out gave better recall later.
How long must the gap be? A 2008 study taught facts to more than 1,350 people, gave them one review after a gap, and tested them up to a year later. The best gap depended on when the test was:
| You need it in | Best gap before the review | In days |
|---|---|---|
| 1 week | 20 to 40 percent of the delay | About 1 to 3 days |
| 1 year | 5 to 10 percent of the delay | About 3 to 5 weeks |
The rule from that study: the longer you want to keep something, the longer the gap. A review on the same day does little.
3. Mix it
When you practice one kind of problem many times in a row, you know which method to use before you read the problem. A real test does not tell you. Mixed practice makes you choose the method each time, and that choice is part of the skill.
Mixing helps most when the items are easy to confuse with each other. A 2019 meta-analysis found that the benefit depends on how similar the items are, and for some material, blocks are as good or better. The article on interleaving says where it works and where it does not.
4. Get feedback
Recall with no check can fix an error in place. After each attempt, look at the right answer. A wrong attempt that is followed by the right answer still helps.
What helps less than you think
A 2013 review by John Dunlosky and others rated ten common study techniques. Practice testing and spaced practice were rated high. These three were rated low:
- Highlighting. It marks the text. It does not make you use it.
- Rereading. It is the most common method and one of the weakest for memory that lasts.
- Summarizing. It helps only those who are already good at it.
None of these is harmful. They are a first pass. The memory comes from what you do after.
A plan for any topic
- Learn a small part. Ten minutes of material is sufficient for one session.
- Ask at once. Close it and write down or say what you remember. Then check.
- Come back tomorrow. Answer questions on it. Do not read it again first.
- Come back in about a week, then in about a month. Each time the gap gets longer.
- Mix the old with the new. Put yesterday’s topic and last week’s topic in the same session.
- Keep what you got right in the set. Items that you stop testing are the items that you lose.
How Axo does it
Axo is built on these four methods. Every Bite ends in five questions, and each answer is checked at once. Keep Sharp brings each idea back after a gap that gets longer, and it mixes ideas from all your Paths in one session.
You do not make the cards or the schedule. The page on spaced repetition in Axo shows how the timing works, and the science page lists all eight findings behind the app.