Sleep: what a night does to a memory, and what a night of cramming costs¶

Level: 101 · for anyone who has stayed up all night before an exam

One line: Sleep after learning keeps more of it than the same hours awake, and a short night before learning cuts what can be taken in the next day. Both are among the best-established results in memory research; which sleep stage does what, and the molecular story, are less certain, and the popular protocols built on them less certain still.

What is well established¶

Sleep after learning protects and strengthens memory. Jenkins and Dallenbach showed it in 1924: the same syllables were remembered better after eight hours of sleep than eight hours awake. A century of work since (Rasch and Born's 2013 review is the standard summary) has confirmed it for facts, for skills, and in every age group. Grade: well established.
Sleep loss before learning cuts encoding. One night without sleep reduces what the hippocampus takes in the next day by a large margin (Yoo and colleagues, 2007), and impairs attention, working memory and the learning of new sequences. Grade: well established.
Cramming instead of sleeping therefore costs twice: the material studied is consolidated less, and the next day's material is encoded less. Zakrajsek's 37% on the Calculus 1 final, after four all-nighters that each produced a good unit-exam grade, is the pattern in one story. Grade: well established as a consequence of the two above.
Adults need about 7 to 9 hours, and both much less and much more go with worse cognitive performance. This is a consensus guideline (the American Academy of Sleep Medicine and the Sleep Research Society, 2015) rather than a single experiment, and people differ, so McGuire's advice stands: find your own minimum and keep to it. Grade: well established as a population range.

What has good evidence¶

A short nap helps. A nap of 20 to 30 minutes restores alertness; one long enough to include deeper sleep also consolidates what was learned before it (Mednick, Nakayama and Stickgold, 2003). Longer naps can leave you groggy for a while. Grade: good evidence.
Regular sleep and wake times go with better sleep and better daytime function. Grade: good evidence, mostly observational.
Consolidation happens partly during quiet wakefulness too. Ten to fifteen minutes of doing nothing after learning, eyes closed, no phone, improved recall in Dewar and colleagues' experiments (2012), and the effect has been replicated at modest size. Summaries that call this "non-sleep deep rest" and say it "dramatically" increases retention, or that the hippocampus "replays at twenty times normal speed", are adding rat electrophysiology and a marketing adjective to a small, real effect. Grade: good evidence for a small effect.
Sleep before an exam beats a last hour of review. This follows from the two well-established points, and classroom studies that track sleep and grades agree; the direct experiment is hard to run ethically. Grade: good evidence.

What is weaker, or mostly from animals¶

Which stage does what. The usual division, slow-wave sleep for facts and REM for skills and emotional memory, has real support but also real exceptions, and the field is still arguing (Rasch and Born discuss both sides). The picture of memories being replayed from hippocampus to cortex during slow oscillations, spindles and ripples comes largely from rodent recordings, with human evidence growing. Grade: good evidence for the replay picture in animals; weak for any advice a learner could act on beyond "sleep".
Sleep clears waste from the brain, and sleep loss raises amyloid. Both are real findings, but in mice and in short human studies; their meaning for a student's memory this semester is unknown. Grade: weak, as far as learning is concerned.
Light, screens and caffeine at night. Bright light in the evening delays the body clock, which is well established; the size of the effect of ordinary phone screens is disputed and probably small, and content that keeps you engaged matters more than the light. Caffeine's half-life is about five to six hours, so an afternoon coffee is still partly present at bedtime. Grade: good evidence for light and caffeine; weak for the specific "no screens for two hours" rule.
Exercise and sleep help each other. Regular exercise improves sleep, and in one study exercise offset the memory cost of restricted sleep. Grade: good evidence for the first; weak for the second.

What this page does not do¶

It gives no guidance on sleep medication, supplements or sleep disorders. If sleep is a problem rather than a habit, that is a question for a doctor.

What to do¶

Protect the night after a study session, not just the night before an exam. The learning is not finished when you close the book.
Never trade sleep for review. The last hour of cramming is worth less than the sleep it replaces, twice over.
Keep a regular schedule, including weekends as far as life allows.
Nap if you can, briefly, and in the early afternoon.
Do nothing for ten minutes after a hard session, without a phone. It is cheap and probably helps a little.

Po polsku, w skrócie¶

Sen po nauce utrwala więcej niż te same godziny na jawie, a krótka noc przed nauką zmniejsza to, co da się przyjąć następnego dnia. Oba wyniki należą do najlepiej potwierdzonych w badaniach pamięci; zarywanie nocy przed egzaminem kosztuje więc podwójnie. Dorośli potrzebują zwykle siedmiu do dziewięciu godzin, ale ludzie się różnią. Krótka drzemka pomaga, a dziesięć minut spokoju bez telefonu po trudnej sesji pomaga trochę. Słabsze są twierdzenia o tym, która faza snu za co odpowiada, o „oczyszczaniu mózgu” i o dokładnych regułach w rodzaju „żadnych ekranów dwie godziny przed snem”: to głównie badania na zwierzętach albo małe badania na ludziach. Strona nie mówi nic o lekach i suplementach na sen; to pytanie do lekarza.