The forgetting curve, explained

You learn something today and feel like you know it. A week later, most of it is gone. That is not a personal failing, it is the default behaviour of human memory, and it has a well-measured shape called the forgetting curve.

Understanding that shape is what makes spaced repetition make sense, so this is the primer the rest of the guide builds on.

What Ebbinghaus actually measured

In the 1880s, Hermann Ebbinghaus ran experiments on himself, memorising lists of nonsense syllables and testing how much he retained after different delays. The result, now called the forgetting curve, showed retention dropping sharply at first and then more slowly.

The exact numbers vary with the material and the person, but the shape is consistent: the steepest loss happens soon after learning. Much of the detail goes within the first day, and what is left keeps eroding over the following weeks unless something interrupts the decay.

He used nonsense syllables on purpose, to strip out meaning and prior knowledge. Real, meaningful material decays more slowly, but the curve still bends the same way. The brain discards what it does not seem to need, and “need” is judged largely by how often something comes back.

Why the curve is steep on purpose

It is tempting to treat forgetting as a bug. It is closer to a feature. You encounter an enormous amount of information every day, and keeping all of it would be useless noise. So the brain defaults to letting things fade and only holds on to what proves important.

The signal it uses for “important” is recurrence. A fact that returns to your attention several times across days and weeks gets flagged as worth keeping. A fact seen once gets dropped. This is the lever spaced repetition pulls.

How reviews flatten it

Here is the part that matters in practice. Each time you successfully recall something, the forgetting curve that follows is shallower. The memory decays more slowly the second time, slower still the third, and so on.

That is why the intervals can grow. The first review might need to come a day after learning. After that review, the material holds for several days, so the next review can wait a few days. After that, a week. After that, a month. You are not reviewing on a fixed timer; you are riding the flattening curve, reviewing each time just before the memory would have dropped too far.

This is also why timing beats volume. A review done at the right moment, when the memory is fading but not gone, does far more than the same review done five minutes after learning (when there is nothing to reinforce) or a month too late (when it is already gone and you are relearning from scratch). The effort of a slightly difficult recall is what strengthens the trace, which is the same reason active recall beats rereading.

What this means for studying

Three practical consequences fall out of the curve:

• Review early. The first review matters most because the curve is steepest at the start. A quick pass the day after you learn something protects it cheaply.
• Let gaps grow. Once a memory survives a review, it is more durable, so reviewing it again too soon is wasted effort. Expanding intervals match the flattening curve.
• Do not wait until it is gone. If you leave material untouched for months, you are not reviewing, you are relearning. The whole efficiency of the method comes from catching memories before they fall off.

Put together, these are exactly the rules of spaced repetition: review at growing intervals, timed to the curve. The forgetting curve is the why; spaced repetition is the how.

The catch is bookkeeping. Tracking the right review moment for one fact is easy. Tracking it for hundreds, across several subjects, is where people give up, which is the problem I built Memset to handle: it keeps the schedule so you only have to show up for the reviews.