The Forgetting Curve Is Real: How Spaced Repetition Beats Cramming

You’ve forgotten most of what you learned last week. That’s not a guess, and it’s not a personal failing. It’s the forgetting curve, the best-replicated finding in memory research, first measured in 1885 and confirmed with modern methods in 2015. In that replication, more than half of the initial learning effort was already lost within an hour (Murre & Dros, 2015, PLOS ONE).

The same 140 years of research also shows exactly what stops the leak. Space your practice. Test yourself. Keep sessions short. None of it is complicated, and almost none of it is what school trained you to do. Here is the evidence, with the honest numbers.

What the forgetting curve actually shows

The forgetting curve shows memory for new material dropping steeply within hours, then levelling off. Hermann Ebbinghaus first mapped it in 1885 by memorising lists of nonsense syllables and retesting himself at set intervals. In 2015, Murre and Dros repeated the experiment under modern controls and produced nearly the same curve.

One detail most blog posts get wrong: Ebbinghaus never measured “percent remembered”. He measured savings, the reduction in effort needed to relearn a list compared with learning it cold. In the replication, savings fell to about 42% after 20 minutes and 34% after an hour. Put plainly, within an hour, over half of the initial learning effort was already gone.

Notice the curious bump at one day. Savings at 24 hours (about 32%) came out higher than at nine hours (about 27%), in the replication and in Ebbinghaus’s original data alike. The plausible explanation is sleep: a night of consolidation seems to firm up part of what the day eroded. Memory decays fast, stabilises, and responds to what you do next. You can watch that play out in our interactive forgetting-curve simulator, where adding a few reviews visibly flattens the decline.

And what you do next is the whole game. That steep early drop is why a single exposure, one lecture, one chapter, one tutorial binge, feels productive in the moment and leaves so little behind a month later.

Spaced repetition: the fix hiding in plain sight

Spacing means splitting the same study time across several sessions with gaps in between, and it works with unusual reliability. A synthesis of 839 assessments across 317 experiments found that spaced study consistently beats massed study, better known as cramming, for long-term retention (Cepeda et al., 2006, Psychological Bulletin).

The same review surfaced a rule worth memorising: the longer you need to remember something, the longer the optimal gap between reviews. Material for next week’s meeting can be reviewed tomorrow. Knowledge you want to keep for years should be revisited across weeks and months. Each successful, slightly effortful recall flattens the curve a little more.

The effect holds far beyond the lab. When Dunlosky and colleagues (2013) reviewed ten popular study techniques for Psychological Science in the Public Interest, only two earned the top “high utility” rating: practice testing and distributed practice, which is spacing by another name. Highlighting, rereading, and summarisation, the techniques students report using most, all rated low.

A close cousin called interleaving mixes related topics within a session instead of blocking one at a time. In a study of 7th-grade maths classes, students who practised interleaved problems scored 61% on a surprise test a month later, versus 38% for blocked practice (Rohrer et al., 2015). Mixing feels harder while you do it. That difficulty is the point.

Testing yourself beats rereading

Retrieval practice, closing the book and pulling the answer from memory, beats rereading at every delay that matters. In a classic experiment, rereading looked better when tested five minutes later, but after two days and one week, students who had tested themselves retained substantially more (Roediger & Karpicke, 2006, Psychological Science).

The effect holds at scale. A meta-analysis of 272 effect sizes from 188 experiments found practice testing beat restudying with an effect size of g = 0.51, and beat doing nothing with g = 0.93 (Adesope et al., 2017, Review of Educational Research). Better still for the quiz-averse: multiple-choice practice tests (g = 0.70) outperformed short-answer formats (g = 0.48). A quiz doesn’t have to hurt to help.

Rereading loses because it manufactures a feeling of knowing. The second pass through a chapter is smooth, the sentences are familiar, and your brain reads that fluency as mastery. It isn’t. Familiarity means you recognise the material while it sits in front of you. Retrieval is the skill you need later, when the page is gone, and the only way to train retrieval is to practise retrieving.

Why short sessions win

Short sessions fit the attention and working memory you actually have. Average attention on a single screen has fallen from 2.5 minutes in 2004 to 75 seconds in 2012 to about 47 seconds now, a replicated finding from Dr. Gloria Mark’s lab at UC Irvine (American Psychological Association).

First, clear away the myth. You’ve probably heard that human attention spans shrank to eight seconds, less than a goldfish. That figure traced back to an unsubstantiated “Statistic Brain” number, not to any actual research, as the BBC documented in 2017. Nobody ever measured the goldfish either. Mark’s screen-attention data, by contrast, comes from years of direct observation and has been replicated by other researchers.

Attention is only half of the constraint. Working memory, the mental workspace where new information gets processed, is severely limited, and overloading it blocks learning rather than speeding it up (Sweller, 1988, Cognitive Science). Adults hold roughly four chunks of information in the focus of attention at once (Cowan, 2001). Sit through an hour-long lecture and you get a few minutes of real encoding followed by a long tail of overflow.

The format graveyard makes the same case. Among people who register for a MOOC, 52% never start it at all (Reich & Ruipérez-Valiente, 2019, Science). Meanwhile, a literature review found that lessons of one to ten minutes, with material reintroduced over time, counteract the forgetting curve and reduce mental fatigue (Shail, 2019, Cureus). Small units, repeated, beat big units abandoned.

Putting it to work in five minutes a day

So what does a routine built on this evidence actually look like? Four habits, none of them exotic: learn in small pieces, quiz yourself instead of rereading, let the gaps between reviews grow, and attach the whole thing to a daily trigger so the spacing happens without a calendar.

1. Learn in small units. One idea per session respects the four-chunk limit on working memory. A single clear concept that survives beats ten that evaporate.
2. Test immediately. After each lesson, recall the core point before moving on. Even one multiple-choice question converts passive exposure into retrieval practice.
3. Space the reviews. Revisit tomorrow, then in a few days, then in a couple of weeks. Expand the gap every time recall succeeds, and shrink it when recall fails.
4. Show up daily. A fixed daily slot, coffee, commute, queue, makes distributed practice the default instead of a chore you schedule.

This research is the blueprint behind Scroll - Daily Microlearning. Lessons take about 45 seconds, a length that sits comfortably inside Mark’s 47-second screen-attention finding (a happy fit, not a causal claim). Every lesson ends with a quick quiz, because retrieval beats rereading, and the daily cadence brings ideas back across days and weeks, which is the spacing effect doing its quiet work.

If the format is new to you, start with the full microlearning guide. If the hard part is consistency rather than method, here’s how to go about building the daily habit in five minutes or less.

The forgetting curve isn’t a flaw to be ashamed of. It’s the operating condition of every human brain, measured in 1885 and confirmed in 2015. Cramming ignores it and loses. Spacing, retrieval, and short sessions work with it and win. You forget on a curve. Learn on one too.