Learning and the mind
Sleep and memory consolidation
The brain replays and reorganises the day's memories while you sleep, so a night of sleep improves recall and skill where a day awake does not.
John Jenkins and Karl Dallenbach showed in 1924 that less is forgotten across sleep than across waking; the modern account is set out in reviews by Walker and Stickgold (2004, 2006) and Rasch and Born (2013).
In 1924 Jenkins and Dallenbach had two subjects learn nonsense syllables and recall them after one, two, four or eight hours spent either asleep or awake. After sleep they remembered more at every interval. The first explanation was passive: sleep shields a fresh memory from the interference of new experience. The current account, which Rasch and Born call active system consolidation, says more happens. Memories first held in the hippocampus are reactivated during slow-wave sleep and gradually redistributed to the neocortex, where they join older knowledge; the rapid eye movement (REM) sleep that follows may stabilise them. On this view the waking brain is tuned for encoding and the sleeping brain for consolidation.
The effect is easiest to see in skills. In 2004 Kuriyama, Stickgold and Walker trained 57 people aged 18 to 31 on finger-tapping sequences. After a night's sleep they were 17.7 percent faster on a simple five-key sequence with 47.8 percent fewer errors, and 28.9 percent faster on the hardest nine-key two-handed one; a group retested after 12 hours awake had improved by 0.9 percent. The same lab's 2002 paper in Neuron had already shown that the gain depends on sleep rather than on the passage of time. The mechanism, reactivation of the day's hippocampal patterns during the up-states of slow oscillations together with sharp-wave ripples and spindles, is the part Rasch and Born treat as established. The practical reading is that the session before bed and the night after it are one lesson.
Most of the clean numbers come from finger-tapping, visual discrimination and syllable lists in small laboratory samples, and how far they carry to a semester of mathematics or a new codebase is less measured. The field has also shifted its emphasis more than once, from REM sleep to slow-wave sleep, and the roles of the stages are still argued. The claim that holds across a century of studies is narrower and more useful: memory tested after sleep is better than memory tested after the same time awake. Murre and Dros's 2015 replication of Ebbinghaus found savings higher at one day than at nine hours, which they read as the same effect showing up in a forgetting curve.
Example
In a study published in Learning and Memory in 2004, Kuriyama, Stickgold and Walker trained 57 healthy right-handed volunteers aged 18 to 31 to type short key sequences as fast and accurately as they could. Retested after a night of sleep, those on the simplest sequence, five keys with one hand, were 17.7 percent faster and made 47.8 percent fewer errors; those on the hardest, nine keys with two hands, were 28.9 percent faster. A group retested after 12 hours awake had improved by 0.9 percent. The authors concluded that the overnight gain grows with the complexity of the task.
Where it applies
Put the hard learning session before sleep rather than after an all-nighter, and treat the night as the last step of a practice session rather than a break from it. Machine learning borrowed the idea: experience replay in reinforcement learning re-presents stored episodes to the network between live updates, a design with the same shape as hippocampal replay.
Think about it
When you next need to learn something hard, what would change if you treated the night after as part of the session?
Applied it to something real? Log it.
Test yourselfIn the 2004 finger-tapping study, what happened to speed after a night's sleep compared with 12 hours awake?
Sleep gave 17.7 percent on the simplest sequence and 28.9 percent on the hardest, with errors nearly halved on the simple one; the group kept awake for 12 hours gained 0.9 percent.
Related models
- The forgetting curveMemory for new material drops fast in the first hours and then slowly; what is left after a day is most of what will be left after a month.
- Spaced repetitionReviews spread out over time beat the same reviews crammed together, and the best gap grows with how long you need to remember.
- Deliberate practiceExpert skill is built by structured, effortful practice aimed at specific weaknesses with feedback, not by hours spent doing the activity.