Can you learn while you sleep?

We will all spend about one third of our lifetime asleep, deprived of this precious ability to act and to react. During these long idle hours, little is perceived from the external world and little is remembered. For some, sleep is a refuge. For others, it is just a saddening waste. Yet, all animals, from fruit flies to humans, need to sleep and scientists have proven, time and time again, the variety of benefits that sleep has on the body and most importantly, on the mind.

In parallel, many have aspired to make a use of our nights. For example, in his novel A Brave New World, Aldous Huxley imagined a society in which sleep is used for learning. In times fueled by a fastening quest for self-optimization, finding a practical method to learn during sleep has become not only a trendy topic in fundamental research but also a new frontier for private endeavors.

Can the fantasy of learning during sleep be realized? A major obstacle is related to the fact that sleepers seem quite indifferent to what is going on in their close environment. Indeed, since sleepers do not react to external stimuli – unless they’re awakened – it is often assumed that sleepers are disconnected from the real world. Not even being able to process the material to learn should logically prevent any sleep-learning to occur. And yet, how can we be sure of what is processed (or not processed) during our nights?

Experimental evidence actually suggests that we remain, to some extent, connected to our environment. The detailed examination of brain responses to sounds during sleep revealed patterns of activity similar to wakefulness. In particular, it has been shown that, although behaviorally unresponsive, sleepers can still distinguish their own name compared to another, detect weird or meaningless sentences compared to regular ones, and be surprised when a rule shaping a stream of sounds is suddenly broken. All of this is done silently, covertly, and is only revealed when examining neural responses (through electroencephalography or brain imaging) rather than body responses. Recently, we showed that not only could sleepers process external information in a complex fashion (e.g. distinguishing spoken words based on their meaning), but they could also prepare the motor plan of the answer they were instructed to provide (pressing a right or left button) before falling asleep. These studies support the notion that the sleeping brain is not completely disconnected from its environment. The next logical step was therefore to examine whether processing a piece of information during sleep would leave a trace, a memory.

Early attempts to convincingly demonstrate sleep-learning have been largely inconclusive. Consequently, it has been hypothesized that sleep was a state in which the formation of new memories was inhibited in favor of the consolidation of past mnesic traces. Nonetheless, in these past few years, new, well-designed, and well-controlled investigations in both humans and animals showed that mnesic traces could be formed during sleep. Along this line, Sid Kouider and I set out to examine in finer details the nature of these sleep-memories.

Our approach was quite simple: we asked participants to classify real or fake words while falling asleep. Each word-category (real or fake) was associated to a response side (right or left). Preparing to answer a right or left response triggers an asymmetry between right and left motor cortices that we monitored with scalp electroencephalography. Thus, even when participants were unresponsive, we could check whether they correctly processed words presented during sleep.

After their nap, we presented our volunteers with the words heard during wakefulness or sleep along completely new items. Participants were instructed to indicate whether they remembered having heard a given item during their nap and to rate their confidence in their response. Unsurprisingly, participants were very good at recognizing words heard while awake but could not explicitly differentiate words heard during sleep from new ones. However, we found a difference in confidence ratings for words heard during sleep compared to the new ones. Such effect on confidence in the absence of an explicit recognition is classically interpreted as the manifestation of an implicit memory, like when you cannot explicitly remember a specific item from a list while still finding it somewhat familiar.

To buttress this interpretation, we completed the analysis of participants’ behavioral responses with the examination of their brain responses to old or novel spoken words during this memory test. Once again, we found different neural responses for words heard during sleep relative to new items. Such differences imply that participants’ brains could differentiate these two categories and that items processed during sleep left a trace within the brain. Dissecting the neural responses to words presented during sleep confirmed the existence of memories formed during sleep as well as their implicit nature.

However, it is important to stress the limited impact of these memories on behavior, making them of dubious use in our daily-lives. Learning words in a new language during sleep would be quite useless if we cannot explicitly retrieve these words when needed. Nonetheless, a recent study found an interesting application for such sleep-memories: during sleep, smokers were presented with the scent of tobacco paired to a nasty rotten fish odor in order to form an association between these two odors. Upon awakening, such pairing did not affect how participants perceived the scent of tobacco but significantly reduced their smoking habits. Importantly, the same manipulation during wakefulness did not lead to such decrease in cigarette consumption.

Hence, although faint, the implicit nature of sleep memories can be used to alter the behavior where classical wake-learning fails. But the long-term consequences of such manipulations are unknown. As said above, sleep has a crucial restorative effect on our cognitive abilities, so interfering with sleep should therefore be done wisely and taking into account the potential downsides. Learning is thus possible during sleep but is better suited for situations in which the implicit nature of the implanted memories is a positive rather than limiting factor. Otherwise, you better save your bedtime for resting.

Featured image credit: “Learning” by CollegeDegrees360. CC BY-SA 2.0 via Flickr.