You can learn as you sleep

But only during certain sleep phases.

In a new study, researchers show that the formulation of new memories is possible during sleep, but only during rapid eye-movement (REM) or light non-REM (NREM) sleep. Conversely, deep NREM sleep actually had the opposite effect and suppressed the ability to learn new information. We spoke with Thomas Andrillon of École Normale Supérieure about the work.

ResearchGate: What motivated this study?

Thomas Andrillon: This study was motivated by our previous work on the brain’s surprising ability to process sensory information during sleep. For decades, sleep has been seen as a state in which the brain is isolated from its environment. But this view has been questioned more and more. Auditory information, for example, is actually quite faithfully encoded during sleep.

Since the sleeping brain can process external information, an immediate question was whether the sleeping brain could learn. When we are awake, experiences continuously imprint on the brain, and processing a given piece of information usually leaves a trace behind, even if we are not aware of it. Conversely, it had been proposed that sleep was a peculiar state in which memory systems would be disconnected from sensory areas, leaving the ability to process external stimuli but without learning from them. We decided to examine how sleep and its different stages would affect the ability to learn.

Participant of a sleep experiment equipped with an electroencephalogram. The electroencephalogram records brain's electrical activity and allows to determine in real time whether a person is awake or asleep. Response handles placed in participants’ hands allowed them to categorise the sounds presented during the experiment. Credits: © Scientifilms - Déchiffrer la conscience, voyage dans l'étoffe de nos pensées Credits’ email: jeanne.lacheze@scientifilms.fr

RG: How did you test your hypothesis?

Andrillon: To explore how sleep influences learning abilities we used a peculiar form of learning: the learning of white noise. We were interested in this form of learning for several reasons. For one, the material is complex, since listeners have to learn fragments of white-noise. Fragments of white noise are generated through random sequences of numbers and are deprived of distinctive features. Nonetheless, the learning can be performed automatically—despite their complexity, fragments of white-noise can be learned after only a few exposures and can remain in memory for up to two weeks. They also can be learned while subjects are focusing on something else.

We exposed participants to several sets of noise in wakefulness, NREM, and REM sleep. A given set of noise to learn was presented in only one state thanks to the continuous monitoring of the participants through EEG. Then, upon awakening, these different sounds were played again to the participants alongside new items. Participants were not informed of the learning aspects of the experiments but rather were asked to perform an auditory discrimination task. Based on their performance, we could determine whether hearing a sound during the night helped them perform the task better, which meant that some sort of learning occurred.

RG: What were your results?

Andrillon: What we observed is that exposure during REM sleep led to increased performance, as in wakefulness. So it seems that sleepers can learning during REM sleep. Interestingly, exposure during NREM sleep led to decreased performance compared to new items. This is particularly striking since this is neither a positive result (evidence of learning) nor a null result (no learning) but a negative one. Hearing a sound in NREM sleep made participants’ performance the worst. This means that the sleeping brain can either form or suppress memories during sleep, depending on the specific sleep stage.

Participant of a sleep experiment equipped with an electroencephalogram. The electroencephalogram records brain's electrical activity and allows to determine in real time whether a person is awake or asleep. Response handles placed in participants’ hands allowed them to categorise the sounds presented during the experiment. Credits: © Scientifilms - Déchiffrer la conscience, voyage dans l'étoffe de nos pensées Credits’ email: jeanne.lacheze@scientifilms.fr

RG: How can people adopt your results into their learning?

Andrillon: To translate our results into practice, several things have to be considered. First, we investigated a specific form of learning called perceptual learning, the ability to discriminate or recognize sounds better or quicker. It remains to be shown whether our results apply to other forms of learning, such as the learning of a new language. Second, in most of the instances in which sleep learning was proven, the associated memories happened to be implicit. Implicit memory means that the learner does not know that they have learned. It would be interesting to investigate whether explicit memories can be formed during sleep. Finally, to learn during sleep, participants were stimulated through sounds, which can disturb sleep. So you may learn during your sleep, but this could come at the cost of your sleep quality. It is important to check whether the cons might offset the pros.

RG: What shouldn’t people take away from your study?

Andrillon: People shouldn’t think that because learning during sleep is possible, that it is easy to do and they should try it. Usually, the sleep learning effects are quite small and come at a cost: decreased sleep quality. It is often more efficient to learn while awake and enjoy a good night sleep.

Image credit Giulia Mule.