Is the Purpose of Sleep to Let Our Brains “Defragment,” Like a Hard Drive?

By Neuroskeptic | May 14, 2012 12:42 pm

Neuroskeptic is a neuroscientist who takes a skeptical look at his own field and beyond at the Neuroskeptic blog

Why do we sleep? We spend a third of our lives doing so, and all known animals with a nervous system either sleep, or show some kind of related behaviour. But scientists still don’t know what the point of it is.

There are plenty of theories. Some researchers argue that sleep has no specific function, but rather serves as evolution’s way of keeping us inactive, to save energy and keep us safely tucked away at those times of day when there’s not much point being awake. On this view, sleep is like hibernation in bears, or even autumn leaf fall in trees.

But others argue that sleep has a restorative function—something about animal biology means that we need sleep to survive. This seems like common sense. Going without sleep feels bad, after all, and prolonged sleep deprivation is used as a form of torture. We also know that in severe cases it can lead to mental disturbances, hallucinations and, in some laboratory animals, eventually death.

Waking up after a good night’s sleep, you feel restored, and many studies have shown the benefits of sleep for learning, memory, and cognition. Yet if sleep is beneficial, what is the mechanism?

Recently, some neuroscientists have proposed that the function of sleep is to reorganize connections and “prune” synapses—the connections between brain cells. Last year, one group of researchers, led by Gordon Wang of Stanford University reviewed the evidence for this idea in a paper called Synaptic plasticity in sleep: learning, homeostasis and disease.

This illustration, taken from their paper, shows the basic idea:

While we’re awake, your brain is forming memories. Memory formation involves a process called long-term potentiation (LTP), which is essentially the strengthening of synaptic connections between nerve cells. We also know that learning can actually cause neurons to sprout entirely new synapses.

Yet this poses a problem for the brain. If LTP and synapse formation is constantly strengthening our synapses, and we are learning all our lives, might the synapses eventually reach a limit? Couldn’t they “max out,” so that they could never get any stronger?

Worse, most of the synapses that strengthen during memory are based on glutamate. Glutamate is dangerous. It’s the most common neurotransmitter in the brain, and it’s also a popular flavouring: “MSG”, monosodium glutamate. But in the brain, too much of it is toxic.

Glutamate works as a transmitter molecule by opening channels on the cells that receive it. The channels allow calcium into the cells on the receiving end, which activates them, allowing messages to go through. But too much glutamate can cause excess calcium to build up inside the very cells that receive the message, a harmful process called excitotoxicity.

So, if our brains were constantly forming stronger glutamate synapses, we might eventually run into serious problems. One function of sleep, according to the theory, is to protect the brain against excitotoxicity or other “synaptic overload” problems by pruning the synapses.

If the brain is essentially removing the “extra” synaptic strength formed during the previous day, it must do so in a way that preserves the new information. One possible mechanism for this is synaptic scaling.

After some of the neural connections into a given cell, or “inputs,” become stronger, then all of the synapses on that cell could be weakened. This would preserve the relative strength of the different inputs, while keeping the total inputs constant.

It’s as if each neuron were a cup, and each synapse corresponds to a different liquid. During the day, memories form and certain synapses get stronger, which means pouring more of those particular liquids into the cup. At night, synaptic scaling pours some of the mixture back out, bringing it back to the baseline level without changing the relative proportions of the mix.

We know that synaptic scaling happens in the brain, but it’s not yet clear whether it has anything to do with sleep. This is an area of ongoing research.

While synaptic scaling seems to treat each neuron like a cup to be kept from overfilling, the effect of sleep on for the brain overall may be more like disk defragmentation, according to this idea.

After heavy use, hard disks tend to get “fragmented.” This is because when data gets stored, it is written to wherever there happens to be free space on the disk. This makes it inefficient to keep track of it all as files may be split and written in many different places. A defrag consolidates the same data into a more logical order. Defragmentation is a taxing chore for the computer, so many people schedule it to happen overnight. In the same way, sleep may serve to reorganize and reconsolidate memories. The mechanics of how this defragmentation works remain unclear; synaptic scaling might be just one of several processes at work.

Defragmentation is not an exact analogy, however. The process could also be likened to archiving your emails to make room in your inbox, or compressing data into zipped files, to free up room on the disk.

(This theory is specifically about slow-wave sleep (SWS). It doesn’t try to explain rapid eye movement (REM) sleep, when dreams happen. Interestingly, some animals do not have REM, but they all have SWS. In some animals, like dolphins, only one side of the brain has it at a time, which is strong evidence that SWS, but not REM, is vital for life.)

So what’s the evidence? A number of studies in simple animals, like flies and zebrafish, have shown that synaptic strength and density is increased in sleep-deprived animals, compared to recently rested ones. Wang and colleagues showed some examples of synaptic proteins from the brains of Drosophila, the fruit fly:

What about mammals, specifically us humans? In February 2012, Italian researchers Reto Huber and colleagues of Milan provided the first human evidence in a paper, Human Cortical Excitability Increases with Time Awake. Huber’s team measured the excitability of the brain when people were well rested and then looked to see how it changed as they were kept awake for over 24 hours.

The volunteers awoke at 7am on day 1 of the study, and were then kept awake all of that day, all of the subsequent night, and all of day 2. The excitability measurements spanned a period of 30 hours, from 9am to 3 pm the next day. The subjects finally got to go to sleep on the next night, and one final session took place on day 3.

The results showed a steady increase in brain excitability with increasing time spent awake. Sleep put this back to normal—mostly. Excitability was measured using electroencephalography (EEG) combined with transcranial magnetic stimulation (TMS). Essentially, they hit the brain with a strong magnetic pulse from a coil held up against the head, and measured the electrical activity that this triggered in the brain.

The graphs below show what happened: the TMS pulse caused an electrical response with three peaks: up, down, up, all in the space of 30 milliseconds. This represents the firing of different groups of cells. The magnitude of the response was increased when people were sleep-deprived (red lines), compared to when they were well rested:

Although it was a small study, all six participants clearly showed higher stimulation-evoked potentials after sleep deprivation. Although the exact causes of the TMS evoked response are unclear, this is consistent with the idea that synapses become steadily stronger during wakefulness, and are pruned during sleep.

If this is true, it also raises questions for neuroscientists because it means that the time of day that an experiment takes place could have an impact on the results. Whether the test subjects are animals or humans, if they’re tested at the end of long period of wakefulness, their brains might be more excitable.

There are still many questions around the function of sleep, but one thing seems certain: we won’t be able to understand the brain if we limit ourselves to thinking about what happens when we’re awake.

CATEGORIZED UNDER: Mind & Brain, Top Posts
  • Paul

    Didn’t Crick propose a theory very much like this a couple of decades ago?

  • Paul

    (Didn’t Crick propose a theory very much like this a couple of decades ago?)

    Ah, I see Crick and Mitchison (1983) were proposing that REM sleep acted to reduce neural connections, not SWS.

  • Yacko

    Seems to me something like that or an enhanced filing system. With the conscious mind limited, dreams are merely an attempt to make sense of the background noise of moving memories.

    One should not forget that sleep has somatic benefits as well. I don’t think the body would last as long without the recuperative downtime. Just resting is not the same.

  • Brian113

    The presence of glutamate begs the question: Can eating brains cause migraines? I mean, for the eater, not the eatee.

  • Smarti

    Makes a lot of sense. As a seriously sleep deprived mind, I am realizing just how much gets accomplished while we are sleeping: many of the “triggers” that get tripped during our daily functioning get reset during sleep. And when they don’t, we function differently and it adversely affects our health. There is confusion, weight gain, and we cannot operate to our full potential.

  • Jim Johnson

    Interesting article, and the theory makes sense to me, I don’t personally think the answer to what sleep is, is going to be so simple. Maybe “pruning synapses” , plus ALSO “keeping us inactive, to save energy”, plus also physical recuperation (plus also etc., etc., etc.) If sleep is a system that’s been around since early in the development of complex nervous systems, the surprise would be if it weren’t performing multiple functions by now.

    The ultimate question, and the one they’re pursuing I guess, isn’t “what is the function of sleep?” but “what function of sleep is so important that no known nervous systems go completely without it?”

    And, of course this all reminds me of:
    “People say, ‘I’m going to sleep now,’ as if it were nothing. But it’s really a bizarre activity. ‘For the next several hours, while the sun is gone, I’m going to become unconscious, temporarily losing command over everything I know and understand. When the sun returns, I will resume my life.’

    If you didn’t know what sleep was, and you had only seen it in a science fiction movie, you would think it was weird and tell all your friends about the movie you’d seen.

    They had these people, you know? And they would walk around all day and be OK? And then, once a day, usually after dark, they would lie down on these special platforms and become unconscious. They would stop functioning almost completely, except deep in their minds they would have adventures and experiences that were completely impossible in real life. As they lay there, completely vulnerable to their enemies, their only movements were to occasionally shift from one position to another; or, if one of the ‘mind adventures’ got too real, they would sit up and scream and be glad they weren’t unconscious anymore. Then they would drink a lot of coffee.’

    So, next time you see someone sleeping, make believe you’re in a science fiction movie. And whisper, ‘The creature is regenerating itself.”
    ― George Carlin, Brain Droppings

  • Neuroskeptic

    Paul: Oh, that’s very interesting. Thanks for the link. REM sleep is if anything even more mysterious than slow-wave sleep. Not all animals have it, and in humans REM deprivation is less disruptive than SWS loss. There are some drugs actually (MAOis) which abolish REM sleep while leaving SWS intact, and people seem to cope with that OK. So it seems likely that REM isn’t “necessary”, at least in the short term & medium term. But that raises the question of why we have it at all!

  • Tom

    Does the process of synaptic scaling impose limits on our knowledge? This ‘down-sizing’ of synaptic strength that only preserves relative connection strengths presumably means that weaker connections (i.e. older or less used information) lose more strength when more new information has been learnt that day… Does it?

  • Maxine

    Regarding REM sleep, this is when we dream, and we can become “awake” in the dream, conscious that we are dreaming. Check out “Lucid Dreaming”. This is related to astral travel and out-of-body experiences. I have had lucid dreams and been aware of traveling back into my body. If this sounds too woo-woo to neuroscience geeks, consider that our knowledge is constantly widening at an exponential rate (e.g. quantum physics), and that there’s a whole lot beyond what we think we know.

  • William Turner

    As Tom says, synaptic pruning would seem to impose limits on knowledge. That does not seem to be the answer. There are some of us who believe that consciousness (knowledge and memory) resides outside the human brain. Until someone finds the “hard drive” physically in the brain, I have to stick with the idea that we communicate with a form of Cloud out there somewhere. In that chain of thinking, sleep may simply be a re-organization of the electromagnetic reception mechanisms in the brain – sort of a tune-up of the complex satellite radio. In that regard, the TMS and EEG experiment was very interesting because it showed that external electromagnetic waves do stimulate the brain.

  • Kenneth J. Epstein

    The oldest theory of sleep is that it can be explained in terms of metabolism versus catabolism. Waking activity breaks down the body faster than it can be built up. Sleep is necessary to correct this imbalance. It allows animals to function beyond a sustainable capacity during hours when they can gain the most by doing so. Sleep repairs the damage done by this hyperactivity, and is most practical during hours when there is the least to gain from being hyperactive. Defragmentation may indeed be part of the metabolic process, while fragmentation is part of the catabolic process.

    Sleep obviously has a great evolutionary advantage. Animals who can outdo themselves while awake are more fit to survive than those who do not need sleep but are limited to what they can do without it. A constant balance of metabolism and catabolism does not allow the spurts of energy that are often needed, just as a balanced budget amendment would place the country at a disadvantage when cash is needed quickly and borrowing is the only way to get it. Borrowing energy or money during the day and paying it back at night is much better than having no credit or borrowing power at all. Bodily economics is very similar to financial economics. You can live beyond your means temporarily, but not permanently.

  • Sunny D

    I feel that in exploring this topic we should begin to compare and contrast those individuals who sleep a lot and those who only need a few hours of sleep a night. Plus, why do depressed people sleep more? How does a sleep “cycle” allow you to sleep less?

    Answering these questions may lead us to understanding sleep better.

  • James T. Dwyer

    Defragmentation of disk space is a pretty crude analogy, when more general computer operating system memory management strategies and database management systems have a much richer background to draw from.

    A common strategy used to achieve optimal recording of new information is directly analogous to short term and long term memory partitioning. For example, new information might initially be captured by serially recording event data in a short term memory ‘buffer’. This would allow optimal ‘peripheral (sensory) device’ recording performance, with little concern for read access performance or storage space usage.

    Retrieval could later be optimized to provide quick access to select data through the use of multiple indexes. The indexes might allow direct retrieval of all memories of a specific person, for example. Creating those indices would require sophisticated analyses of the simple short term memories. This could be accomplished along with migration from short term memory to long term memory during memory database ‘down time’, thus freeing up the high capture rate short term memory for reuse. Moreover, long term memories could be stored using shared links to redundant data, significantly saving total storage space required. This process would require the recall of stored memory elements, or snippets, to potentially generalize their content and update necessary cross-links.

    Internet search engines are also employ sophisticated retrieval optimization methods using offline indexing, etc.

    This is just a few of the many processes often employed in memory and database management systems. If similar processes are used to manage our own internal memories, they could help explain why sleep is necessary for maintaining memory performance, why dreaming can involve seemingly meaningless recall of memory ‘snippets’ and why memory details can change over time, for starters…

  • Sigmund

    Oh God yes. I remember thinking this exact scenario when I was a Systems Administrator and I had to shut the network down in order for it to “re-set.” The memory boards had all been maxed out with data and the CPU had to organize everything in order for work to proceed. I often thought that, with all those synapses firing in our heads, at some point, the biochemicals involved have to somehow regain their electrons in order the brain to keep functioning. That’s why we can go without food or water longer than we can go without sleep.

  • Pingback: Quora

  • Raymond DiLuzio

    Synaptic pruning everynight sugest that for ther to be pruning everynight, the pruning would be of illrelivent information that happense everyday. Understainding that pruning is a metaphore, the information gathered by the test is real, but it is up to inteligent minds to interpret the information in a way the dose not conflice with undeniyable laws of the universe. We use electrons to to stimulate matter into an arangment that retains the form of the energy appliyed. 1s and 0s, their not the information indevidualy but rather part of a colective seqencing of a recording of information. Every thing we now is the result of are sencers recording outside information. If we could draw in information of the univerce with our mind alone, life would have never constructed eyeballs and other sencers to gather information/enery. Energy forces of a habitat causes matter to mix into a better energy gathering system untill it reaches critical mass, supper novas, then the remnents (HUMANS), like a white dwurf star, cunstructs itself into the most stable energy management system in the known universe. All the energy released (or “pruned” is illrelivent and self destructive in the case of a black hole formation ( insainity if related to the mind). There is most definitly a cloude of information in the universe, made by the universe that made humanes to make the cloud that is the internet, books, and thru any form of comunication. If you ignore actual information, you will continue to become proned to the ignoring of actual universe information thet your mind has been looking for. By the laws of phisics you could set an ignorent path for yourself that will take a longtime to dissipate. For your own sake, and the sake of the 4 people who has already liked you post, stop trying to exsplain something about science that any confused LSD induced mind could easily conclud to with ther paridoxil way of thinking


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

The Crux

A collection of bright and big ideas about timely and important science from a community of experts.

About Neuroskeptic

Neuroskeptic is a British neuroscientist who takes a skeptical look at his own field, and beyond. His blog offers a look at the latest developments in neuroscience, psychiatry and psychology through a critical lens.


See More

Collapse bottom bar

Login to your Account

E-mail address:
Remember me
Forgot your password?
No problem. Click here to have it e-mailed to you.

Not Registered Yet?

Register now for FREE. Registration only takes a few minutes to complete. Register now »