Sleep to remember, and to forget

Updated: Feb 16, 2019

You've been off caffeine and alcohol for two days. This morning, you woke at 7am and got ready for a day at the sleep lab at the University of California, Berkeley.

It's 11am and you're sitting in front of a computer screen. Words keep flashing up on the screen but this is not a straightforward memory task. There are 100 words in all. After each one appears, it's followed by either a large green R or a large red F.

You've been told that if the R appears, you're going to have to recall the word that came just before it. If the F appears, you can forget about that word. Notice that, at the time you see each word, you don't know whether you're going to have to recall or forget it.

Okay, so now it's 11.15am, you've seen all 100 words and you're now given a quick test to see how well you can remember both the Remember words and the Forget ones. This is done by flashing more words onto the screen and you have to press a key to indicate whether you recognise each one or not - were part of that previous list of 100?

Fast forward to 2.30pm - you now have 19 electrodes connected to various points across your scalp. They're ready to pick up electrical activity in your brain for an EEG recording. You've also got some more electrodes attached to your face, ready to pick up eye movement, so you're feeling pretty sci fi.

And it's time for a nap. You basically have until 4.10pm to get yourself a 100-minute nap.

So, you got to sleep and you were woken up at 4.10pm, but now it's 5.15pm and you're back in front of the computer.

This time it's a 5-minute free recall test. That means you have to type into the computer as many of the words from the initial list of 100 as you can. So, yes, this includes the Remember and the Forget words - the Forget instruction was actually a deception (psychologists lie a lot, you've been warned) - now you're being asked to remember all the words regardless of R of F.

Okay, so here's the thing - there were also 22 other participants who went through exactly the same procedure as you just did (but on different days). AND there were 23 other participants who went through the exact-same procedure except they weren't allowed to nap. In fact, they were monitored all afternoon to make sure they didn't nap.

What these researchers want to know is quite simple - what effect did the nap have on your memory, compared to people who didn't nap? Are there any differences in memory between the nappers and the no-naps?


Before anyone had a chance to nap - on the very first test, which was administered 15 minutes after seeing the original 100 words - there was no significant difference in recall between any of the participants. In both groups everyone was much better at recognising the Remember words than the Forget words.

After some people had napped

Here's where it gets weird. In the 5.15pm free recall test the people who had slept recalled significantly more words than those who hadn't.

Think about that for a moment - the group who were awake all afternoon could think about and rehearse the words they were being told to remember. Those who were asleep couldn't do that - they were sleeping! And yet, the sleep seems to have significantly increased the number of words they could recall compared to the non-nappers.

And, here's where it gets even weirder. The increase in the number of words that the sleepers recalled came almost entirely from the Remember words. In other words, while they were sleeping, their brain selectively picked out the R words and stored them in long-term memory.

How did your brain select what to remember?

Warning: if you dislike technical terms and neurospeak, jump to the next sub-heading

The researchers delved deeper into the data and found the neural mechanism associated with being able to selectively remember only the R words.

Notice the fast sleep spindle in the middle of the Stage 2 NREM sleep. Graphic courtesy of Lumen Learning.

Firstly, this long-term memory consolidation happens during Non-REM sleep (NREM). This is when your brain waves slow down and gain amplitude.

More specifically, the people who were best at selective memorising had more fast sleep spindles. These are short bursts of brain activity, typically occurring in stage 2 NREM sleep.

Even more specifically, these fast sleep spindles were happening predominantly in the left superior parietal cortex. That's the left-hand red blob in the graphic.

What seems to happen during NREM sleep is a kind of conversation between the cortex (the outer layer of your brain) and the hippocampus (a pair of structures located deeper in the brain, in the Medial Temporal Lobes).

The hippocampus is really good at picking up information during the day, but it can only hold so much information and it can’t hold it for long. During NREM sleep your hippocampus hands over the most important facts and experiences to your cortex, where they can be stored long term. And the left superior parietal section of the cortex is where the selection of what to keep and what to discard happens.

Say that again, in English

My daughter is currently studying for A level exams. Recently she was talking with friends who said they were planning to stay up most of the night before an up-coming exam to 'cram'. This is a bad choice for lots of different reasons, but the one we'll focus on here is that, by staying awake, they will shut off the very process of memory consolidation which they need.

To put it another way: you need a minimum of 8 hours of sleep per night. Less than that will reduce your ability to remember facts long-term.


The Role of Sleep in Directed Forgetting and Remembering of Human Memories. Salatin, J, Goldstein, A and Walker, M. 2534-2541, s.l. : Cerebral Cortex, 2011, Vol. 21.