Mind & Psychology · Memory & learning
How memories fade and how they get rewritten
A memory can drain away, or it can stay vivid and report something that never happened. Learn what the forgetting curve really measured and how later information edits what you remember.

What you will learn
- Say what the forgetting curve measures and how savings is defined
- State the limits of the original study behind the curve
- Describe the misinformation effect and the study design that demonstrates it
- Name the proposed mechanisms behind the effect
Lesson 1 of 4 · 5 min
What the forgetting curve really measured
In 1879 a German psychologist named Hermann Ebbinghaus began memorizing nonsense. He made up meaningless syllables such as WID and ZOF, strung them into lists, and recited each list until he could say it twice without a mistake. Then he waited, anything from 20 minutes to 31 days, and learned the list again.
The graph of his results is the forgetting curve. You've probably seen a version of it, usually with a claim that people lose half of what they learn within days.
That isn't what he measured. He never counted how many syllables he could still recall. He timed how long the list took to learn again and compared that with the first time. The share of time saved is called savings.
If relearning takes a quarter as long as the first learning, savings is 75%. If the list is still perfect, savings is 100%. If relearning takes just as long as the first time, savings is zero.
The difference matters. A list he couldn't recite at all could still be quicker to relearn. Savings picks up memory that a recall test would score as gone.
Here's what Ebbinghaus found. After 20 minutes his savings had already dropped to 58%. After an hour it was 44%, and after a day 34%. Then the fall slowed to a crawl. After 31 days he still saved 21%.
So forgetting is steep at first and then levels off. Most of what goes, goes in the first hour.
The study had one participant, Ebbinghaus himself, who was also the experimenter. He kept at it for about seven months, often with up to three sessions a day. And nonsense syllables aren't something anyone wants to remember.
The result has held up better than that sounds. In a replication published in 2015, a Dutch researcher spent 70 hours repeating the experiment on himself and got a similar curve. An earlier German replication with two subjects did too.
That's reassuring about the measurement. It's still a handful of people and meaningless syllables, so trust the general shape more than any exact percentage.
Try to recallThe Ebbinghaus forgetting curve is often quoted as a general law of memory. What was the study behind it, and how should that change your confidence in the curve?
The study was Hermann Ebbinghaus testing himself. Around 1880 he memorized lists of nonsense syllables and timed how much faster he could relearn them after delays from 20 minutes to 31 days. So it was one person, who was also the experimenter, learning material with no meaning. Later replications found a similar curve, but they also used one or two people and nonsense syllables. The general shape, steep at first and then nearly flat, deserves confidence. An exact percentage quoted as a law for everyone and everything doesn't.
The forgetting curve plots time saved on relearning, measured on one man and his nonsense syllables. Its shape is the part to keep. Forgetting is fastest in the first hour, and even a month later a trace remains that makes relearning quicker.
Lessons in this topic
What the forgetting curve really measured
The best-known graph in memory research came from one man memorizing nonsense syllables. It plots the time he saved when he learned them again.
How a sentence can rewrite what you saw
A car stops at a stop sign. A sentence read afterward says yield sign, and people then remember a yield sign. See how the experiment worked and what might explain it.
What flattens the forgetting curve
Ebbinghaus's own experiments showed two things that slow forgetting, meaning and repetition spread over days. Later studies worked out how long to wait before a review.
Why a false memory is easier to prevent than to fix
A warning protects a memory best before the false detail arrives. Afterward, warnings and even the original footage only partly undo it.
Three quick stories from this topic
One man and his nonsense syllables
The forgetting curve comes from one man who spent seven months learning nonsense syllables
Hermann Ebbinghaus was his own only subject. Around 1880 he learned lists of made-up syllables, waited between 20 minutes and 31 days, and timed how much faster he could learn them again. After a month, relearning still took about a fifth less time.
What he measured
Savings is the share of the original learning time that relearning doesn't need. It can show a trace of memory even when nothing can be recalled.
It held up
In 2015 a Dutch researcher spent 70 hours repeating the experiment on himself and got similar results.
The broken glass that wasn't there
Changing one verb in a question more than doubled false memories of broken glass
In 1974 students watched films of car crashes and were asked how fast the cars were going when they hit each other. For some, the verb was "smashed". A week later that group was more than twice as likely to remember broken glass. The films showed none.
Faster, too
Students who heard "smashed" also estimated much higher speeds than students who heard "contacted".
The misinformation effect
Information taken in after an event, even inside a question, can become part of the memory of the event.
The white van that wasn't
Police hunted a white van in the 2002 DC sniper case, and the snipers drove a blue sedan
A white van was seen near one shooting, and the police chief showed a picture of it on national television. More witnesses then called to say they'd seen a white van too. The area had more than 70,000 of them, and other tips were ignored.
Suggestibility
Information from an outside source, even a news report, can become part of what a witness remembers seeing.
Three weeks
The attacks went on for more than three weeks and killed ten people before the suspects were caught.
Sources
- Forgetting curve, Wikimedia Foundation
- Replication and Analysis of Ebbinghaus' Forgetting Curve, Murre and Dros, PLOS ONE 10(7), 2015
- Memory: A Contribution to Experimental Psychology, chapter 7, Retention and obliviscence as a function of the time, Hermann Ebbinghaus (1885, translated 1913), Classics in the History of Psychology, York University
- Misinformation effect, Wikimedia Foundation
- Problems with Memory, OpenStax, Psychology 2e




