The Curious Relationship Between Sleep and Memory

In 1924 two psychologists at Cornell University, John Jenkins and Karl Dallenbach, asked student volunteers to learn lists of nonsense syllables and then tested them after intervals of one to eight hours. Sometimes the volunteers spent the interval awake, going about their day. Sometimes they spent it asleep. After sleep, they remembered noticeably more.

Jenkins and Dallenbach thought sleep simply protected memories from the clutter of waking life, like a museum closed to visitors. A century of research has turned that idea on its head. Sleep, it now seems, is not a pause in memory-making at all. It is one of its busiest workshops.

The night shift

When you learn something new, such as a name, a route or a recipe, the hippocampus, a curled structure deep in each temporal lobe, captures it quickly. But the hippocampus is a temporary store, rather like a notepad by the telephone. For a memory to last, it has to be woven into the vast long-term networks of the cerebral cortex.

According to the leading theory, set out in an influential 2013 review by Björn Rasch and Jan Born, much of that weaving happens during deep, slow-wave sleep. The brain reactivates recently learned material, replaying it in compressed form, and gradually transfers and reshapes it for long-term storage. REM sleep, the stage of vivid dreams that tends to dominate the early morning, may then help stabilise the transformed memories. The waking brain, in this view, is built for taking information in; the sleeping brain is built for making it last.

Waves and spindles

How does the brain coordinate such delicate work? Recordings of electrical activity on the scalp reveal a remarkable choreography.

The deepest stage of sleep is dominated by slow oscillations, great rolling waves in which vast populations of neurons fall silent together and then fire together, less than once per second. Riding on these waves are sleep spindles, brief bursts of faster activity lasting around a second that look, on a trace, like a spindle of thread. And nested within the spindles, the hippocampus produces its own rapid ripples, which are thought to carry the replayed memories.

What seems to matter is timing. In a 2018 study in Neuron, Randolph Helfrich, Matthew Walker and colleagues at the University of California, Berkeley, showed that the slow wave governs the precise moment at which spindles arrive, and that the quality of this coupling predicted how much people retained overnight. In older adults, the coupling had drifted: spindles tended to arrive slightly too early, missing the peak of the wave. The degree of mistiming was linked to shrinkage in the medial frontal cortex, and to forgetting.

“

Sleep is not a pause in memory-making at all. It is one of its busiest workshops.

The same Berkeley group had earlier found, in a 2013 study published in Nature Neuroscience, that older adults with more grey-matter loss in the prefrontal cortex produced weaker slow waves, and that this helped explain why they retained less of what they had learned the evening before. It offers one plausible route by which ageing and memory decline are connected: not only through changes in waking life, but through a gradual loosening of the night's choreography.

A scent of roses

If sleeping brains replay memories by themselves, could we prompt them? In 2007 Rasch, Born and colleagues tried exactly that. Volunteers learned the positions of pairs of cards on a grid while the scent of roses filled the air. Later, as they slept, the researchers released the same scent.

When the odour was delivered during slow-wave sleep, people remembered the card positions better the next day. When it came during REM sleep or while they were awake, or when they had not smelled it while learning, there was no benefit. Brain scans showed that the scent during deep sleep activated the hippocampus. The odour seemed to act as a cue, nudging the sleeping brain to replay that particular memory.

The technique, now known as targeted memory reactivation, has since been tried with sounds, music and spoken words. In 2020 Xiaoqing Hu, Ken Paller and colleagues gathered 91 experiments involving more than 2,000 participants into a single meta-analysis. They found a reliable, small-to-moderate improvement in memory when cues were played during the lighter stage two of non-REM sleep or during slow-wave sleep, and none during REM sleep or wakefulness.

:::note How targeted memory reactivation works

• Learn something while a distinctive cue, such as a scent or a sound, is present
• Sleep, while researchers monitor brain waves to detect non-REM sleep
• Replay the cue quietly enough not to wake the sleeper
• Test the next day: in a 2020 meta-analysis of 91 experiments, cued memories were recalled better on average :::

Rocking the waves

Others have tried amplifying the slow waves themselves. In 2013 Hong-Viet Ngo, Jan Born and colleagues at the University of Tübingen and the University of Lübeck played brief pulses of quiet sound to sleeping volunteers, timed by a computer to coincide with the upswing of each slow oscillation. The sounds strengthened the slow-wave rhythm, boosted the spindles locked to it, and improved overnight recall of word pairs. Played out of step with the rhythm, the same sounds did nothing. It is a little like pushing a child on a swing: the push helps only when it arrives at the right moment.

Whether these laboratory tricks can become safe, useful tools at home is still an open question. They depend on accurate, real-time reading of brain waves, which consumer gadgets cannot yet provide reliably.

What you feel in the morning

There is a curious postscript from the world of wearables. In a 2025 study posted to arXiv, Peter Neigel and colleagues at Osaka Metropolitan University and partner institutions tracked 29 students with sleep rings for up to eight weeks, testing their working memory each morning with a demanding letter-matching task. The amount of REM sleep the ring recorded helped predict how well they performed, and there were hints that the effect built up over more than one night.

But the students' own morning judgement of how well they had slept predicted their memory scores over and above anything the ring measured. It is a pleasing reminder that the sleeping brain does its work largely out of reach of our sensors, and that how rested you feel is a meaningful signal in its own right.

The practical lessons are old-fashioned. Regular, sufficient sleep gives the brain its best chance to consolidate what you have learned. A good night after studying, practising an instrument or meeting new people may be time well spent. And if your sleep is persistently poor, or you notice changes in your memory that worry you, it is worth talking to your GP.

✦

Filed under The Mind. Spotted an error? Tell the editors.