In the year that Everest was conquered and DNA’s structure discovered, a young PhD student called Eugene Aserinsky did what no one had thought to do before. He spent hours staring at the eyelids of a sleeping person. What he saw was amazing. Instead of the slow periodic motion he’d been expecting, he saw frantic, jerky eye movements.
Until then, sleep had been seen as a pretty passive and uninteresting state of mind. But Aserinsky and his PhD adviser, physiologist Nathaniel Kleitman of the University of Chicago, went on to show that this “rapid eye movement” was correlated with massively increased brain activity and dreaming. Their discovery is widely regarded as having kick-started the modern discipline of sleep research.
Fifty years of intense study later, however, and almost every aspect of sleep remains a mystery. We still don’t know what sleep is for, how it evolved, or why we dream. But there are exciting new ideas.
Advertisement
Earlier this month, the city of Chicago hosted the biggest ever gathering of sleep researchers to mark the 50th anniversary of the discovery of REM, and to ask again, what is it all for? Graham Lawton reports.
MOST PEOPLE think they don’t get enough of it. Some manage just fine on half the usual amount. Dolphins do it with half a brain at a time. Rats die after three weeks without it, yet male Emperor penguins manage an entire three months of brooding without getting any. It’s as good as ubiquitous among animals – even insects do it. We spend a third of our lives at it. Yet no one has any clue why we do it or how it came about.
Our ignorance about sleep is becoming something of an embarrassment. Try to think of another fundamental biological phenomenon to which we can’t assign a role. You’ll draw a blank. “I think it’s the biggest unanswered question in neuroscience,” says Craig Heller, a sleep researcher at Stanford University in California.
Our ignorance does not stem from lack of trying. There are dozens of theories of sleep function out there, which fall into four broad classes: restoration and recovery, predator avoidance, energy conservation and information processing. Most sleep researchers are happy to accept that sleep has more than one function and that several of these theories, or perhaps all of them, are simultaneously correct. And yet not one of them has been confirmed or refuted. Take “restoration and recovery”, for example. This is probably the most obvious potential function of sleep and surely the easiest to test. Yet we still don’t know what the brain might be restoring, or how.
Why is sleep such a hard nut to crack? One big stumbling block has been its dual nature. While we sleep, we cycle between two very different states (see Diagram). The first is called slow wave sleep because it is characterised by long, lazy waves of undulating electrical activity (delta waves) that seem to be synchronised across the whole brain. The second is rapid eye movement (REM) sleep, as Aserinsky first witnessed, and it could hardly be more different. REM sleep is characterised by frantic brain activity that looks very much like wakefulness on an EEG trace. It also has very obvious physical signs: the rapid flickering motion of the eyeballs, the near-total muscle paralysis thought to prevent you acting out your dreams, and penile erections (women exhibit engorgement of the clitoris and lubrication of the vagina).FIG-mg24015201.jpg
REM sleep appears to have arisen quite early in evolution – reptiles, birds and mammals all do it. It’s also a very active and excitable state. So it must serve a very useful function. Any decent theory of what sleep is for must account for it. In fact, explaining REM has so far dominated sleep research (so much so that slow wave sleep, despite having several distinct stages or depths is often defined merely as non-REM sleep). Yet the function of REM has proved so difficult to fathom that some sleep researchers despair that we will ever solve the problem. “REM sleep has a biological function but we have not been able to find it. And I don’t think there will be any explanation for another 50 years,” says Michel Jouvet, a veteran sleep researcher from the medical school at the University of Lyon in France, who has spent 48 years studying REM sleep.
While plenty of researchers are still busy studying REM and dreaming (see Box, “What dreams are made of”), in the past few years a new generation of sleep function theories have emerged that relegate REM to the background and say that non-REM is what sleep is all about. In these theories, REM sleep is a mere handmaiden of non-REM sleep and its function is to give the brain a break from, or perhaps even test and modify, the really crucial activity that is going on during non-REM sleep. Can this new generation of ideas finally crack the problem?
One such theory has been developed by Heller. Some six years ago, he noticed something odd about the electrical activity of the brain during non-REM sleep. Looking closely at an EEG, he saw brief and subtle patterns of activity that looked like aborted transitions into REM. Such episodes, he found, happen increasingly often until the brain eventually swaps entirely into REM sleep. In other words, the pressure to enter REM sleep seems to build up through a bout of non-REM until the brain can no longer hold it back.
This, Heller says, is a potentially revolutionary observation. Most theories of sleep control posit an external master switch that flips the sleeping brain from one state to another on a preordained timetable (the classic 90-minute cycle). Both states of sleep then have plenty of time to get on with their respective functions. But the build-up of aborted transitions to REM suggests that something else controls the change. It’s as if non-REM creates the need for REM sleep – just as wakefulness eventually creates pressure to snooze. Further support for this idea comes from the fact that the length of a bout of REM is directly proportional to the preceding bout of non-REM. The longer you spend in non-REM, the more REM you need to “recover”, perhaps.
In this view of sleep, REM is demoted to a secondary role as a recovery phase after whatever hard work is done during non-REM. “Our premise is that something happens during non-REM – some imbalance builds up – that has to be corrected for in REM,” says Heller.
But what goes on in non-REM that means you need to recover afterwards? According to Heller, it’s all to do with energy replenishment. During non-REM sleep your brain tops up its turbochargers – glycogen stores in the brain’s supporting glial cells that neurons draw on when they absolutely have to work at maximum capacity. When they run down, they must be replenished.
To do this, the brain would have to enter a quiescent state. To achieve this, suggests Heller, it might throw its neuronal membranes into a suppressed electrical state called “hyperpolarisation”, making them inactive and unresponsive to outside stimuli. This would manifest itself as the unconsciousness of slow wave sleep. Hyperpolarisation, however, comes at a price. It can only be achieved by allowing precious potassium ions to leak out of neurons. Such a state could only be maintained for so long without risking permanent loss of ions. Every now and again the brain has to pump potassium ions back in, repolarising the membrane and switching the cortex back on again, all the while keeping you in a relatively quiescent state. This is what we experience as REM sleep. All the exciting and spectacular facets of REM sleep are merely a by-product of the brain’s housekeeping, no more and no less, says Heller. “Dreaming and so on are really a second or third order phenomenon and are not a function of REM sleep at all.”
But isn’t this rather disappointing? Can all the wonderful complexities and experiences we have in REM sleep really come down to brain cells hauling potassium ions across their membranes? Heller chuckles and points out the hypothesis remains largely untested; there is plenty of room for surprises. “I’m sure it’s going to turn out to be more complex than that.”
Meanwhile, other ideas are emerging that put non-REM in the spotlight without relegating REM sleep completely to the level of the mundane. One hypothesis has been developed by Terrence Sejnowski of the computational neurobiology laboratory at the Salk Institute in La Jolla, California. In his hypothesis, non-REM sleep is also used for recovery and restoration, carrying out “construction projects” related to the rigours and events of its waking activity – replenishing proteins, strengthening synapses, inserting receptors into membranes and so on. You need to be unconscious during this activity so that there is no neuronal activity to get in the way – a bit like moving out of your house while the builders are in. Then, once the brain has completed some construction work, it flips into REM for an interim report. In effect the brain boots up some of its waking systems in a contained environment to test what it has done and find out what still needs to be completed. Then it slips back into non-REM sleep and gets on with the construction work.
“As you have experiences during the day, you’re making a lot of small, temporary changes here and there,” says Sejnowski. “You need to organise and prioritise this information, and you do it in slow wave sleep.” The evidence he has for this assertion comes from computer models of a phenomenon called “spindling” – spikes of electrical activity seen only during the transition from wakefulness to non-REM, and from REM back into non-REM. During spindling there is massive influx of calcium into neurons that have been modified by that day’s experience. And calcium is known to be a crucial activator of enzyme function and gene expression – the kinds of biochemical heavy-lifting the brain would have to perform to convert these small changes into permanent ones.
“But you don’t want to make all your changes at once,” Sejnowski says. “You make a few of them, and you test to see what impact this has. Maybe REM sleep is the test period.”
There is other evidence building for the idea that a brain’s workload determines how much deep sleep a brain needs, perhaps even with harder-working areas showing different sleep states to those that had a more leisurely day (see “Sleep or sleeps?”).
One of the beauties of these “non-REM rules” ideas is that they help explain why the brain cycles through several non-REM to REM bouts during a night’s sleep, and why you always start the night with a bout of non-REM. If the two different modes of sleep have independent functions, then it’s not clear why they should be organised in this way. But if non-REM sleep makes REM essential, then the way sleep is organised suddenly makes much more sense.
Another leading researcher who sees REM as having a secondary but non-trivial function is Derk-Jan Dijk, director of the newly opened Sleep Research Centre at the University of Surrey in Guildford. He’s sure that the main function of sleep is restorative and that the restoration process – whatever it is – has little to do with REM sleep. He points out that bouts of REM increase in length throughout the night and hit a peak at the very end of the sleep. “This tells you something about the function of REM sleep,” he says. “It may not be that important to the recovery process.”
So what is it for? “It may be more important for the transition from sleep and wakefulness. Some people have said that REM sleep is a gate to wakefulness.” Animals very frequently wake up at the end of each REM episode, check all is well, then drift off again. Even humans may do it when there is something, say an infant, in need of regular checking.
Another new theory pushes the importance of REM sleep even further down the list, at least in adults. According to this idea, REM does have a function – but only in the womb and soon after birth. REM sleep continues in adults only as a developmental relic of this function. It’s the brain’s belly button.
The idea originated from a group led by Stephen Duntley of the Sleep Medicine Centre at Washington University in St Louis, Missouri. They started from the well-known fact that as the brain develops, it prunes out redundant neurons in the cortex to build a sleek and honed thinking machine. The pruning tool is programmed cell death, or apoptosis, and the signal for it to occur is neuronal inactivity. Duntley’s group believes the brain needs quiet periods of recovery and restoration, but this creates a problem in babies. You don’t want to lose too many neurons as they descend into their quiescent sleeping state.
The group also noted another well-known fact – that infants have a very high proportion of REM sleep, starting in the womb and continuing throughout early childhood. During REM sleep, neurons are very, very active indeed. From an EEG trace alone you would have a hard time distinguishing between REM and wakefulness. Could the two phenomena be related? Is frantic brain activity in REM anything to do with building a brain?
To find out, group member Michael Morrissey used a drug called clonidine to selectively deprive very young rats of 60 per cent of their normal REM sleep. When they examined the rats’ brains for markers of cell death, the results were clear: REM-deprived baby rats had huge amounts of apoptosis, much more than you would see in a healthy brain or in an adult given the same drug. “It’s way beyond normal,” says Morrissey.
The team speculates that REM sleep evolved to keep useful neurons busy and so save them from apoptosis, without forcing the animal to waste precious energy on staying awake. “It’s a very exciting theory,” says Duntley.
Critics agree that it’s an interesting idea, but are not yet convinced by the data. David Gozal, an expert in childhood sleep at the University of Louisville in Kentucky, points out that the apoptosis assay the team used is not very sensitive. Given that there is so much apoptosis going on at this stage of development, it’s hard to say for sure that there’s an abnormal load in REM-deprived rats. Duntley counters by saying that the work is at a very early stage and the hypothesis should be easily testable with refined techniques. Watch this space.
But there is another prominent group of theorists who refuse to accept that sleep is merely a time for rest and recuperation. They prefer to assign sleep – particularly REM – a more fittingly cerebral role in information processing, categorising and storing memories, perhaps, or even thinking creatively.
The idea that sleep plays a crucial role in learning and memory goes back to 1983 at least, when James Watson of double-helix fame, proposed that REM was responsible for the destruction of unwanted facts and experiences. Since then the field has exploded, producing reams of evidence that REM sleep is intimately involved with what’s known as “procedural memory” – learning how to perform a complex task such as riding a bike or playing the piano. (The other sort of memory, “declarative memory” for storing facts and figures, seems to be unaffected by sleep.)
In one classic experiment, volunteers were asked to spend four hours playing an apparently simple video game involving six targets with buttons underneath. The rules are very straightforward: when a target lights up, subjects press the corresponding button. The researchers measure their reaction times to see how good they are at the game. The more subjects play, they better they get. What the players don’t know, however, is that the targets don’t light up at random. There is a complex imperceptible set of rules determining the sequence with which they light up.
Then comes the interesting bit. The volunteers are allowed a good night’s sleep and then retested. Miraculously, their performance improves a great deal overnight. But not in every case. In some volunteers there is no improvement at all. The difference? One set of non-improvers was deprived of REM sleep by repeatedly being wakened through the night. The other was playing a game in which the sequence was totally random. There’s nothing to learn, and so no overnight improvement. The conclusion? Spending time in REM asleep helps “consolidate” procedural memory.
Many similar experiments have come to the same conclusions about the role of REM sleep in procedural memory. And there is supporting evidence from other types of experiment, too. For example, PET imaging of the brain as it learns a procedural task shows a characteristic pattern of activity which is repeated again and again during REM sleep. This activity is taken to be memory consolidation in action.
But once again, it is a measure of non-REM’s growing influence and of REM’s waning one that researchers have begun to grant it a central role in information processing too (New Scientist, 25 September 1999, p 26). This is quite a reversal. “REM sleep was seen as the most important, but now slow wave sleep is moving up fast,” says Carlyle Smith, the director of Trent University’s sleep labs in Peterborough, Ontario, and a leading proponent of sleep’s role in memory consolidation.
In particular it seems that slow wave sleep is when the brain processes spatial memories – learning your way around a new city, for example. In one experiment, volunteers had to learn to navigate their way around a virtual city using a joystick while their brains were being imaged with PET. As they learned the routes, the PET scan showed intensive activity in the hippocampus, as expected from previous work showing its vital role in spatial learning.
The subjects then slept while hooked up to the PET machine. During slow wave sleep their hippocampuses showed exactly the same patterns of brain activity as during learning, and the next morning they were much better at finding their way around. “Slow wave sleep really has something to do with memory,” concludes Philippe Peigneux of the University of Liège in Belgium, who carried out the tests.
The idea that you learn in your sleep has obvious appeal and has lodged in the popular consciousness. But not everyone thinks it is correct. There is a small and highly vocal group of critics who say that sleep has absolutely nothing to do with memory. It looks as though even this most compelling of sleep function theories will have to stay on the “unproven” pile until more evidence comes in.
One of the most strident critics is Robert Vertes of the Center for Complex Systems and Brain Sciences at Florida Atlantic University, Boca Raton. He claims that the evidence for a role for REM sleep in memory is highly contradictory, with about half the published studies showing sleep has no effect. And in those that show a positive effect you cannot control for the stressful effects of REM sleep deprivation. Perhaps sleep-deprived subjects perform worse simply because they are tired.
Vertes’s most compelling evidence, however, comes from patients with brain injuries who do not get any REM sleep at all. In one remarkable case, a man was hit by shrapnel from a bullet at the age of 20 which deprived him almost completely of the ability to enter REM sleep. At best he was getting 1 to 2 per cent REM sleep at night. Yet he continued his education and became a practising lawyer. It’s hard to imagine someone severely deficient in procedural memory achieving this, Vertes says.
Another line of counter-evidence comes from animals. “If you think about the theories of REM sleep serving some intellectual function, the fact that most mammals have about the same percentage of REM sleep gives you pause for thought,” says Heller. “Does a cow really have as many problems as we do?” A similar point is made by Jerome Siegal of the University of California, Los Angeles. He points out that there seems to be no correlation between mental capacity and REM sleep duration (see Diagram). Platypuses, for example, sleep for about 14 hours a day, of which 8 hours are REM sleep. “The platypus is a lovely animal, but it’s also a rather stupid animal,” says Siegal.
The theory’s proponents won’t have any of it. Robert Stickgold of Harvard Medical School, for example, says that evidence in favour of information processing is so good that it cannot be disputed. The best studies control for tiredness and have confidence limits of 0.001, he says. And the animal comparisons are meaningless and unscientific – rather like saying that because millipedes have dozens of legs yet move more slowly than cheetahs, legs cannot be for locomotion.
So, half a century on from the start of modern sleep science, are we any closer to understanding what sleep is for? Things have at least moved on from the days when sleep was simply a form of inactivity and dreams were for prying into your private desires, and might well have come from the kidneys, for all the biology that was known. And there are plenty of new techniques coming into their own that might help settle some of the arguments. The ability to make tissue slices “sleep” in a dish, better and better brain scanners, new mass genetic screening techniques, mutant flies with sleep deficits, and connections with the fields of circadian rhythms and metabolic control – any one of these new developments could be the key to the puzzle. The one thing everyone agrees on is that it is a goal worth pursuing. “If we did know what the function of sleep was, things would change dramatically,” says Heller.
What dreams are made of
Move over Freud. Modern sleep researchers are starting to analyse dream content, and it has nothing to do with your mother
THE 1953 discovery of REM sleep and its clear link with dreaming (Science, vol 118, p 273) gave a huge boost to dream researchers. Until this point, the study of dreams was almost the exclusive domain of the Freudians, who believed the content of our dreams was a hotline to our innermost desires and feelings. But the obvious mental and physical signatures of REM sleep – rapid eye movements, frantic brain activity, and (to the delight of the Freudians) penile erections – opened the door to studying dreams as a real biological phenomenon. “We rejoiced in the discovery of REM sleep,” recalls J. Allan Hobson, a veteran psychiatrist and dream researcher at Harvard Medical School. “We thought it would make psychoanalysis scientific.”
It wasn’t as easy as they hoped. Researchers failed to find any consistent links between physical and mental activity and dream content. Eye movements, for example, rarely tracked the visual content of dreams. Erections had little to do with dreams’ erotic content. Worst of all, within a decade, the apparently clear-cut connection between REM sleep and dreaming melted away, as researchers noted almost as many dream reports when people were woken from non-REM sleep as from REM. If dreaming could come from two such disparate brain states, how would we ever explain it?
In the past few years, however, the biological study of dreaming has undergone something of a revival. Cognitive neuroscientists and neurologists are getting in on the act, using new tools to work out what causes the experience of dreaming and even what its function might be.
One of the key discoveries is that dreams reported after waking from non-REM sleep may originate in REM sleep after all. Using sensitive electrical monitoring techniques to eavesdrop on the activity of the sleeping brain, Tore Nielsen of the University of Montreal and others have spotted brief fragments of REM, so called “covert REM”, intruding into non-REM sleep. These were just too subtle to be picked up by standard equipment. Another recent study led by Hiroyuki Suzuki of Japan’s National Institute of Mental ҹ1000 in Ichikawa has confirmed that, despite prodigious dream reports from non-REM awakenings, there is a huge qualitative difference between the two kinds of dreaming. Non-REM dreams tend to be short and dull. REM dreams, in contrast, are vivid and long.
The re-corralling of the richest dreams back into REM sleep, when brain activity is at its liveliest, has encouraged other researchers to take a fresh look at dreams to see what they tell us about cognition during sleep. “Dreams are the only source of cognitive information we’ve got,” says Sophie Schwartz, from the University of Geneva in Switzerland.
One big question is whether the brain activity you see in REM sleep corresponds with the experiences we call dreams. To find out, a team lead by Pierre Maquet at the University of Liège in Belgium made a “sleep map” of the brain, using the imaging technique PET to find out which areas were busiest during the different phases of sleep. Their results showed that brain activity in REM is very different from non-REM or wakefulness – and that it tallies nicely with the content of dreams.
During dreams, visual areas are very active, as are the amygdala, thalamus and the brainstem, which fits with the fact that dreams tend to be very visual and emotional. At the same time, the prefrontal and parietal cortices and the posterior cingulate, areas which deal with rational thought and attention, are all very quiet, which tallies with the lack of insight, illogicality and time distortion that characterises dreams (see Diagram).FIG-mg24015202.jpg
Maquet accepts that this conclusion is too general to give any major insights into the process of dreaming. But there is another promising approach on the horizon. Schwartz, who was once part of Maquet’s group, suggests we can learn from patients with neurological damage for whom everyday experience is full of bizarre events most of us only experience while dreaming.
Schwartz is currently focusing on Frégoli syndrome, in which people constantly mistake strangers for people they know, even though there are no physical similarities. It’s a mistake that happens all the time in dreams. Frégoli syndrome is caused by brain damage which severs the link between the prefrontal cortex and the other brain structures involved in recognising faces. The way she sees it is this: if you monitor these areas in normal sleeping subjects, you can ask whether Frégoli-type dreams coincide with Frégoli-type brain activity.
There are numerous other brain lesions that cause dreamlike perception during waking – reduplicative paramnesia, in which patients recognise new places as familiar, micropsia and macropsia, where you see things as too big or too small, palinopsia, where you see multiple copies of an object, and achromatopsia, in which colour perception goes haywire. Schwartz says that all of these could help us make links between brain activity and standard dream events.
But none of this answers the most critical and intriguing question: what are dreams for?
Are we any closer to an answer?
“I don’t think we know anything with any confidence about what dreams are for,” says psychiatrist Robert Stickgold of Harvard Medical School. But he has some ideas. Stickgold believes that one of the critical functions of dreaming is information processing, including memory consolidation (see main story). But he’s recently extended the idea into a wider concept of information processing he calls “finding meaning”.
Sometimes when you have a difficult decision to make, all the rational thinking in the world won’t give you the answer, he points out. So what do you do? “You go home that night and you sleep on it,” says Stickgold. By the morning you somehow have the answer even though you’ve gained no new information overnight.
It is dreaming during REM sleep, he thinks, that performs this magical decision-making process. As you dream, your brain runs through imaginary scenarios, testing your emotional response to them without rationality getting in the way. He now has experimental evidence that REM sleep promotes creative thought, allowing you to bring together widely differing concepts you would never link while awake.
The evidence comes from a common cognitive test of a process called priming. Normally the investigator would show someone a word, then quickly flash up another, either real or meaningless. The task is to work out whether it’s a real word or not. If the second word is related to the first, say the pair were “wrong” and “right”, then the decision time is usually significantly faster than, say, “wrong” and “paper”. The reason, according to conventional theory, is that the first word has primed the brain to recognise related words by activating networks of associated concepts.
Stickgold, though, was not interested in wakeful consciousness. He tested people who had just been roused from REM sleep. The result was the exact opposite of wakefulness. The more distantly related the second word, the faster the subjects recognised it. “In REM sleep, the brain ignores the obvious in favour of the crazy, the unexpected or the bizarre,” Stickgold says. “It’s biased towards activating weak, non-obvious and potentially useful connections.” And this, he says, is what allows us to make meaning out of complex information. It might even be the origin of creativity.
Sleep or sleeps?
Do different parts of your brain sleep in different ways? Ever since the discovery of REM we have known that there are several distinct states of sleep. Now it looks as though different patches of the brain can be in different sleep states at the same time. “Sleep is not a whole-brain phenomenon,” says James Krueger at Washington State University in Pullman. “It’s a localised property.”
Dolphins and some birds and fish are already known to sleep with one cerebral hemisphere at a time, either to stay alert or so they don’t drown. But Krueger’s idea runs deeper, suggesting that sleep is a completely devolved system, with different brain regions making their own decisions about depth and onset of sleep. His concept of sleep is at an early stage, and he has just started revealing his evidence to other researchers. But if he is right, then it could give us some new hints about what sleep is for.
The first sign that sleep states might differ between brain regions came from studies of “sleep regulatory substances”, biochemicals that build up in the brain during wakefulness and apparently help trigger the transition into sleep. Around 10 years ago Krueger noticed that these substances build up faster in parts of the brain that are most active during wakefulness. Did this have any effect on subsequent sleep?
It turns out that it does. Apply extra amounts of sleep substances to one hemisphere of the cortex and not the other and the result is much deeper non-REM sleep on that side, as well as subtle differences in REM activity. It seems as if the harder a brain region works during the day, the harder it has to sleep at night.
This tallies with other findings. Cats and rats that are kept in the dark during wakeful hours have unusually shallow non-REM sleep in the visual cortex, but much deeper non-REM sleep in the part of the cortex dealing with touch. And if you trim a rat’s whiskers on one side then put it into a maze, the corresponding half of its brain sleeps more deeply later on – presumably because it had to work harder processing incoming data.
And according to David Rector of Washington State University, you can even see devolved sleep in an EEG trace. He says that a close reading of EEGs reveals that what looks like a steady state is anything but. Though the overall effect might look uniform, on a more local level brain activity is highly changeable, with different neuronal groups constantly cycling between states. What’s more, Krueger points out that in people who survive brain injury, regardless of which parts of the brain are lost the patient always remains capable of sleep. This suggests that sleep is an intrinsic property of brain cells and doesn’t need to be imposed from the top.
Support for this view is also coming from experiments on groups of neurons grown in a Petri dish, which spontaneously produce states very like non-REM sleep. Deprive these neurons of their shut-eye and they go haywire, firing rapidly and randomly in an epileptic-like state. Sleep, then, seems to be an inherent property of small neural networks which keeps them in good shape. Is this how sleep evolved and why we have to do it?
Perhaps the most intriguing spin-off from this work is to understand the mysterious “state dissociation” disorders – cataplexy, for example, where the loss of muscle tone designed to stop you from acting out dreams unexpectedly switches on during wakefulness and makes you fall over, or dream enactment where muscle tone remains switched on in REM sleep. Doctors have long assumed these are caused by some unspecified mixing of states of consciousness. Krueger’s theory might just prove them right.