By April Cashin-Garbutt and Hyewon Kim

The American cartoonist Charles Schulz once said, “Happiness is waking up, looking at the clock and finding that you still have two hours left to sleep.” While all animals that have been studied sleep, we still know very little about what happens in the brain when we do.

As we look ahead to a new sleep lab opening at SWC in January 2024, we asked leading neuroscientists their thoughts on why we sleep. Here are six key reasons they shared.

1. For energy regulation

“I think one of the main reasons for sleep is for energy regulation. Brains are very energetically costly organs. The thing that often people say drives sleep is the product of ATP metabolism, which is strongly related to the energy level in the brain. There is accumulation of adenosine, which is an indication of how much energy you have consumed over a certain period time. Then you need to sleep to regain some energy. I think that is the main function.

In terms of information processing, memory and consolidation, so far it is not entirely understood. If you disrupt sleep, you have some kind of memory problems but it is difficult to figure out whether that is due to an interruption of general energy state that indirectly disrupts memory consolidation or if there are specific processes going on during sleep for memory consolidation. I don’t think it is clear yet and there are still a lot of unknowns.” Ning-long Xu, Chinese Academy of Sciences

2. To preserve important information and erase things you don’t need

“We still don’t know the answer to this fundamental question! We know that it is really bad if you don’t sleep – if you deprive an animal of sleep eventually they die. So we know that sleep is doing something essential.

I think it is likely that sleep does many things that are useful. At a biochemical maintenance level, it seems like sleep might clean the brain of excess metabolites etc. to freshen up the circuits.

In terms of memory, there are a couple of hypotheses. If you look at certain behaviours before and after sleep, the brain is different. It depends on what sort of thing the person has learned. For example, motor learning such as learning to play an instrument, clearly improves by sleep. If you are at a particular level before you sleep, you are better after you sleep. This suggests the networks are reorganising themselves and consolidating memories in a way to improve performance.

Memories for facts and events instead get worse over time because the details get lost. However, they get worse more slowly as a result of sleep. So the idea is that sleep might help preserve certain important information and preserve it better than if you’d stayed away during that period.

The replays during sleep that the hippocampus helps orchestrate (and probably other brain areas do as well) might help cement certain experiences so that they get stored in the long run. Thus, sleep might help with the important things that you want to remember, the stuff that really matters.

There is also evidence that learning during waking typically manifests as stronger synapses, so you learn things during wake and that strengthens connections in your brain. However, if all you do is strengthen connections, then eventually you would end up with epilepsy because everything is too strong and feedbacks on each other. The Sleep Homeostasis Hypothesis, known as SHY, suggests that one of the roles of sleep is actually to just erase stuff. Sleep might be turning down things that you don’t need so that you’re ready to learn again the next day.

This hypothesis is in direct opposition to the idea of storing the important memories. Our guess is that sleep is probably some combination of global downscaling with particular things that are topped back up again. This would be beautiful if it were true as it would be a nice way of the system resetting itself while retaining the information that it needs. But the evidence for this is extremely indirect at this point.” Loren Frank, UCSF

3. For memory consolidation

“I’m not really an expert on sleep, but my personal experiences underscore the vital role sleep plays in our functioning. Without regular sleep, I find my performance significantly compromised. For instance, when faced with an 8-hour jet lag before giving a presentation, I notice a substantial reduction in my cognitive abilities. I shouldn’t be speaking on behalf of my colleagues who work on sleep, but it's generally understood that sleep is crucial for consolidating the memories we've acquired during the day. Furthermore, it's believed to aid in the clearance of metabolic waste and debris produced in the brain.” Weizhe Hong, University of California, Los Angeles (UCLA)

4. To balance the system

“This is a really hot topic. Sleep is like a reset, which is important to keep the balance of the system. I don’t study sleep personally, but I think there is a lot that still needs to be investigated.

An interesting point is that a lot of things we know about sleep and the brain have been discovered in rodents but, unlike us, rodents are nocturnal. Some scientists are moving from rodents to zebrafish as they are diurnal and might thus be more similar to humans in some ways.” Teresa Guillamón Vivancos, Instituto de Neurociencias de Alicante

5. For learning

“There are multiple functions of sleep. Its role in memory appears to be a clear one. Homeostasis is another. In addition to learning, you need to make sure that the synapses don't have runaway excitation or plasticity.

One quote I've always liked is that “Nothing in biology makes sense except in the light of evolution.” The fact that some form of sleep is preserved in all organisms that we've studied shows it's something very critical. But you could also imagine that we've added features to it. Maybe in lower organisms, learning doesn't have much to do with sleep. But in higher organisms like mammals, sleep is a period when you're down and brains can explore patterns linked to learning. So from temperature control to hormone systems and metabolism, sleep seems to impact just about everything. It's hard to pick one feature.” Karunesh Ganguly, University of California, San Francisco (UCSF) and the San Francisco VA Medical Center

6. To get rid of toxins and consolidate information

“Sleep is a big mystery! From what is known, sleep is what helps us consolidate information, which is our ability to store memories in the long term. This is an intensely studied field.

There have also been recent studies that have shown sleep helps you get rid of toxins as they build up. The brain is a very metabolically active organ and there is a build-up of certain chemicals during wakefulness. We all know that when you have a lack of sleep it can affect how clearly we can think and our memory. Sleep is thought to recalibrate the chemical balance.

Sleep also provides some rest. Not just your brain, but other parts of your body. Scientists have shown that across many organisms, even single-celled organisms, there are periods of heightened and reduced activity. These can be akin to sleep going all the way from single-celled organisms, to certain insects, to nematodes, amphibians, mammals and humans. So sleep or periods of rest seem to be a fundamental aspect of life.” Ziv Williams, Harvard University

“There can be two components to this.

The biological component: There could be things related to the biological substrate, like cleaning up toxic chemicals, or performing repairs, that are more efficient during sleep. This could also be for the rest of the body, not just the brain. But I don’t know much about this.

The algorithmic component: CSCG, and CSCG-schema models include a consolidation step that can be performed offline. Of course, in software, there is no reason to sleep — we can make a copy of the model and run consolidation on that while the old copy is awake.

It is possible that the biological component and the algorithmic component coevolved in our brains. If there is a biological need to sleep, then it is good to take algorithmic advantage of that sleep, and vice versa.” Dileep George, DeepMind and Vicarious AI