When I was 13, I once dreamt that a beautiful woman was sensuously stroking the palm of my hand, as a family of fridges hummed in the background. In reality, a huge, buzzing wasp had landed on my right hand. It idly walked around for a bit, and then stung me. After the shock had worn off, I was puzzled why my dreaming brain had stopped me from waking up to this potential danger. Contrast this with 6 years ago, when even my deepest sleep would be broken by the first sounds of my newborn baby daughter’s cries. How do our brains decide whether or not to wake us up, based on what’s going on in the world? And why does this policy change, depending on whether we’re dreaming or in some other sleep state?
In a recent paper in the Journal of Neuroscience, Thomas Andrillon and his colleagues have discovered intriguing clues that start to answer these questions. They used electroencephalography (EEG) to record their participants’ electrical brain activity while they completed a simple task: listening to a series of words and pressing a button with one hand if a given word was an animal, or the other hand if it was an object. Whenever the participants made an appropriate hand response, this was always preceded by a spike of brain activity – a well-known EEG marker, called the Lateral Readiness Potential (LRP), indicating extra activity over the motor cortex on the opposite side of the brain to the hand response.
Crucially, the researchers had the participants complete this test while both awake and during all sleep stages. The experimenters were interested in how the participants’ brains responded differently in these states. To make sure the participants didn’t “cheat” by memorizing word and response mappings when awake (for instance, that the word “table” always means “press right”), there were separate lists for when they were awake and in different stages of sleep.
Intriguingly, as the participants drifted off to a light sleep, their LRP brain signal persisted, even after they’d stopped physically responding. This suggests that their brains were still working out the meaning of the words and how to respond. However, no such result was found for deep non-REM (rapid eye movement) sleep or REM sleep – the stage where most dreaming occurs.
But the brain wasn’t completely oblivious to the outside world in these other sleep states. There were some trials where the participants had fallen asleep without the researchers realising (they only noticed when examining the EEG signals more closely later on). On these trials, the participants were still being presented with words for which they’d already worked out the meaning and appropriate response while they were awake. These familiar words did elicit the appropriate LRP response in the REM sleep stage, suggesting a shallow type of processing was still occurring.
One minor concern I have with these results was that whereas the LRP occurred about a second after the word was heard when the participants were awake, it did not show up until 3.5 seconds later in all the sleep states – in the context of EEG, this is a huge difference. Was this really the same marker as the LRP in the wake state, perhaps delayed because of inefficient sleep processing, or something rather different?
One way to explore this question further is to look at another measure – the “compressibility” of the brain signal as recorded via EEG. If it is highly random, uncompressible, then conscious level is high and the person is probably awake, whereas if the brain signal is orderly, easily compressed, then the person may be in a deep sleep. Using this measure, Andrillon and his colleagues found that consciousness was highest when the volunteers were awake, lower for light non-REM sleep, and lowest with deep non-REM sleep. Meanwhile, when in REM sleep, consciousness was almost as high as the waking state. Given that when we’re dreaming we can feel quite conscious, but trapped inside the stories in our heads, these results all fit neatly with experience.
Next, the researchers linked this consciousness measure with the LRP. When the participants were awake or in a light sleep, the stronger the LRP signal, the higher the consciousness measure. One interpretation of this is that the more conscious you are, even in light sleep, the more you hear and understand, and the more the relevant semantic and motor response processes kick in (as revealed by a stronger LRP in this study). There’s a complicating twist, though, because in the REM sleep stage, the researchers found the opposite relationship – the higher the consciousness level, the weaker the LRP – perhaps suggesting that in dreams, consciousness is at its height when people are better isolated from the outside world. Remember though that these REM LRPs (seen when the participants were asleep but the researchers originally thought they were awake) were more to do with a basic recall of previously consciously learnt word–response pairings, than to a deeper judgement on the meaning of unfamiliar words.
The authors looked at another EEG marker (known as the N550) that further supported this interpretation. Present someone with a word and if they are asleep they will typically show a larger negative spike of brain activity around 550ms after the word than if they were awake – a sign of the outside world being ignored. In the study, in both light and deep sleep, the larger the LRP, the smaller was the N550, suggesting that effective processing of the words was competing with, and suppressing the N550 stimulus-blocking mechanism. In contrast, during REM sleep, larger LRPs were actually related to larger N550s. This may be due to dream mechanisms boosting the shutting down of the outside world whenever it begins to encroach on their territory. It’s almost as if in non-REM sleep, salient features of the outside world can open a gate to deeper processing, but in REM sleep, these same features cause the gate to be held shut more tightly, unless these salient features are life-threatening.
The emerging picture from this and related research is far removed from the naïve view of sleep as involving a brain simply half shutting down. Instead, sleep is a dynamic, complex process, dependent on what stage you’re in. Your brain has to walk a tightrope: it needs to remain asleep and isolated from the external environment, so that important neural regenerative, and memory consolidation processes can occur (especially in REM sleep); but it also needs to keep you safe from dangers in the outside world, by being prepared to wake up if urgent action is required. This new research shows that this weighting of how much value to give to external signals can change depending on how important the current sleep stage is to your welfare, with REM sleep being especially well protected – even when wasp feet are masquerading as a beautiful woman’s fingers.
(Note that I’ve already published a modified version of this post, adeptly edited by Christian Jarrett, on the British Psychological Society website.)