Every night, we invent worlds in our dreams. They often have little to do with the environment we sleep in, and can sometimes be impossible.
Strangely, these wells of creative imagination often disappear as soon as we wake up. Like mirages in the desert, they vanish as we get closer. Conversely, when we're immersed in our dreams, the real world is no longer. So much so that it can be difficult to discern the dream from reality.
Why our waking and dreaming lives seem to coexist in parallel universes is a long-lasting scientific question. My recent research is unravelling parts of this mystery.
Tracking the activity of the sleeping brain
During sleep, while our bodies keep still, our brains are far from idle. Dreaming is perhaps the most direct experience of this continuing activity.
Sleep is actually a busy and vital time for the brain, and past research has shown the importance of it in the consolidation of memories, the cleaning up of metabolic waste, and even on the strengthening of the immune system.
One of the key features of sleep is the fact that sleepers usually fail to respond to their environment (unless you wake them up). It's been proposed that this loss of responsiveness was due to a gating of sensory inputs, with sounds or images failing to even reach the cortex (the cauliflower structure where most of the complex and conscious processes are performed).
However, there's accumulating evidence that this is not the case.
I've shown that the sleeping brain can receive and process a surprisingly wide range of sensory information.
The sleeping brain can recognise important sounds such as your own name or the cry of your own baby, can process speech, and can differentiate words based on their meaning. The sleeping brain can even learn, although this learning is usually limited to basic associations.
We are, of course, unaware of all of this. With my collaborator, Professor Sid Kouider from the École Normale Supérieure in Paris, we recently proposed that this unawareness was due to the fact that most of these processes are performed locally, in isolated brain regions that form islets of wakefulness in an otherwise sleeping brain.
While our hypothesis is well-suited for the parts of sleep in which we don't dream, how can we explain that sleepers, during their dreams, can be conscious of a virtual world, and at the same time still fail to perceive their immediate environment?
Different modes of sleeping
To answer this question, it's important to understand the activity of the brain during sleep.
Sleep can be divided into two distinct phases.
In non-rapid eye-movement (NREM) sleep, the brain slows and becomes hyper-synchronised. Dreams are typically more infrequent and less complex. In NREM sleep, the sharing of information between brain regions can break down, allowing the kind of local, unconscious processing mentioned earlier.
The second phase of sleep is called rapid eye movement (REM) sleep. This part is known as paradoxical sleep because it is, well, quite a paradoxical state – sleepers seem fast asleep, but their brain activity resembles wake activity, and their eyeballs can move frenetically behind their close eyelids, as if sleepers are exploring an inner world.
In previous research, my colleagues and I have shown how these eye movements could reflect the scanning of the dreamt scenery, pretty much as we explore our real world when awake.
In our new study, led by PhD candidate Matthieu Koroma, we tried to understand why sleepers failed to process their environment in REM sleep, despite the fact that the brain is able to process sensory information.
Being absorbed in one’s dreams
To investigate this question, we asked participants to come to sleep in the lab in the morning (when REM sleep is more prevalent). While they were falling asleep, we played stories through earphones: movie dialogues, Wikipedia articles, radio broadcasts, and so on.
We actually delivered two stories at the same time, one in each ear, as if sleepers were sleeping in a room with two people talking in separate conversations. Crucially, one speaker was speaking in French (this was done in Paris), and the other in a made-up language that resembled French but was devoid of meaning (think of Lewis Carroll’s Jabberwocky poem).
Our results further indicate how the dreams themselves could isolate sleepers from their environment.
If you don’t know the first thing about French, you wouldn't be able to recognise which is which. However, if you're fluent in French, your brain gets automatically attracted by the meaningful speaker.
Our goal was to understand how the sleeping brain would deal with these different pieces of information – one meaningful, one meaningless.
During sleep, we obviously cannot ask participants if they're listening to the different speakers, and which speaker they're focusing on, so we had to get this information in an indirect way. We relied on our ability to decode what people are listening to by recording their brain activity with an electroencephalogram.
To put it simply, this technique, called stimulus reconstruction, allows us to reconstruct what is in people’s minds. Obviously, this reconstruction is imperfect, but works well enough to allow us to track sleepers’ auditory attention, as we showed last year in NREM sleep.
Here, we replicated this approach in REM sleep and found that not only intelligible and unintelligible speech is processed, but intelligible speech is prioritised over unintelligible speech, as during wakefulness.
This demonstrates that even when we're sleeping and oblivious of our surroundings, our brain can still track our environment, and select and process the most relevant sources of information. This could be useful if you take a brief nap during your commute – you won’t be helplessly vulnerable.
Think twice, however, before sleeping through meetings or classes – your sleeping brain isn't as smart as you are!
Besides, this ability varies during sleep. In particular, we observed that the meaningful speech was no longer prioritised, and in fact suppressed, when sleepers were moving their eyes in REM sleep. This is particularly interesting, as eye movements in REM sleep usually mark the occurrence of dreams.
An interpretation of these results is that, when we dream, we start processing internally-generated percepts (the stuff of dreams) that will compete with external information. The result of this competition would be the incapacity to process external sensory inputs.
Dreaming could thus represent a form of maximal inattention where sleepers get so absorbed by their own fantasy that they become incapable of processing their external world, consciously or unconsciously.
Why do we dream?
Through this research, we don't only try to examine how the dreaming brain manages the interplay between real and imagined sensory information, we seek also to understand why we dream in the first place.
Dreaming and REM sleep have been proposed to be a training field, or harmless sandbox, in which the brain can train itself to react to situations we may encounter in the real world. This could be why REM sleep is more prevalent in the early years, when we have so much to figure out.
Our results further indicate how the dreams themselves could isolate sleepers from their environment. Such isolation could come at a price – since my daughter was born, I've realised that I won’t wake up easily to her cries if I've been dreaming, despite being fully conscious.
But isolating ourselves from our surroundings could have important advantages in terms of memory consolidation, emotional balance, neuronal plasticity, and making sure that we wake up every morning fresh and ready to face up to a challenging world – and a much more real one.
To receive a fortnightly email wrap up of stories from Lens.