Published on May 21, 2025 3:22 PM GMT
None of this is medical advice.
5 AM on a Saturday and I can’t go back to sleep. It’s not the first time, so I get up to write, I might as well use the time I’m given. My hangover is to blame, even a little alcohol the night before changes my sleep cycle.
You’d think it would ruin my day, but in some ways I feel better. I’m more alert, more anxious, more motivated. It’s like being more alive. Being more awake for only one day has me asking questions. Why did sleep evolve? Can we sleep less?
Why did sleep evolve?
Sleep evolved for temporal niches
Imagine you’re a Zebra. You need to eat a lot of grass to maintain your metabolism. To do so, you need to look around to find the best grasses. You also need to be able to see if predators are nearby, particularly if you’re going to forage away from the herd. Both of these things are easier to do in the full light of day, so you do most of your foraging during the day.
Shouldn't you also forage during the night to get more calories? Probably not. The low light means that it’s harder to forage and easier for predators to sneak up on you. Cooler night temperatures and the energy cost of foraging means you’re actively losing calories. You’re better off snuggling with the pack until the sun returns.[1] You can further take advantage of this habit with adaptations that work better during the day.
The broader thesis is that sleep evolved to save energy during periods when animals are less effective at getting calories. This allows them to evolve a bunch of adaptations for that particular time period rather than maintain adaptations for all time periods.
This fits with many of observations about animals and sleep:
- To save on thermoregulation in hot climates, warm-blooded animals are more active during the night.[2] By the same token, warm-blooded animals in cold regions are more active during the day. Cold-blooded animals are active during the warmest parts of the day.Prey animals are active when most predators are asleep (e.g. mice are nocturnal). Predators are most active when their prey is active (e.g. hawks hunt rodents during the day, owls hunt mice at night) or at times when they have an asymmetric advantage (e.g. crepuscular animals take advantage of better eyesight in low light conditions).Animals in areas with harsh winters will hibernate until the winter is over.Animals like fish, migrating birds, and marine mammals need to move constantly in order to survive and often don’t sleep in the traditional sense.Animals that live without exposure to a day/night cycle (such as in caves) sleep less. I can’t find studies on this, but would predict that deep ocean fish sleep less or not at all.Generally, every animal has a consistent sleep schedule linked to the local daylight hours. They have adaptations for hunting during their active period and methods of staying safe during their resting period (e.g. roosting, burrowing).
Sometimes arguments from evolution feel a little circular. But this argument makes the non-trivial point that you get more evolutionary fitness by specializing in a particular time of day. More surprising is how many scenarios this applies to. But in hindsight, specialization is everywhere in nature; there’s no reason temporal niches should be any different.
Other evolutionary theories of sleep don’t work
But wait, what about other explanations for why sleep evolved?
For example, I’ve heard people argue that neurons need rest in order to function properly. This would make them quite different from other cell types which work all the time. The first thing that made me skeptical of this explanation was the fact that neurons in the heart and lungs function 24/7. There are also neurons in the brain that fire tonically (i.e. constantly). Individual neurons and brain regions don’t seem to require amounts of rest proportionate to their activity. Another problem with this theory is that during REM sleep the brain can be about as active as when it’s awake.
Another theory posits that sleep is required for clearing metabolic products. Without sleep, the story goes, metabolic products like tau proteins would build up and cause neurodegeneration. This story is already dubious given the arguments above. If neurons need metabolite clearance, why can some neurons function constantly? Why does REM sleep involve high levels of activity? Why don’t we see a relationship between activity and rest for different parts of the brain? Further complicating the story is that, in mice, Brain clearance is reduced during sleep and anesthesia. Though note the response to the article and the authors’ reply.
I think if metabolite clearance was a major factor, we would see higher rates of dementia in insomniacs/short sleepers accompanied by lower rates in narcoleptics and people with conditions that require more sleep each night. It would also imply that Alzheimer’s was treatable with drugs that make you sleep more. There isn’t enough evidence to make a strong conclusion, but it seems that insomniacs don’t have higher risk for dementia. In a mouse model, short sleepers are at lower risk for Alzheimer’s. Narcoleptics get Alzheimer’s at about the same rate as everyone else. There’s also claims that neural tissue doesn’t even age. None of these facts fit with the metabolite theory, but the evidence isn’t very convincing and new evidence could change the story.
Another argument claims that sleep is essential for memory consolidation. This theory has a lot going for it, like increased recall after a night’s sleep. It’s probably playing a role in organizing memories and updating synaptic connections. But I don’t think memory consolidation is the primary reason we sleep. Looking across the animal kingdom, total sleep varies dramatically in a way that bears little relation to intelligence, memory formation, or learning. I love cats, but they are very dumb. There’s not much learning or memory going on there, yet they sleep for 16 hours a day.
I’m sure sleep has several functions, including memory consolidation and perhaps metabolite clearance, but I think the primary reason sleep exists is to fit a particular niche. That fits a lot better with how sleep adjusts to local daylight, hunting periods, and climate while having little relationship with anything else. It’s also the only explanation that can handle the fact that hydras sleep despite lacking a brain.
I think the synthesis of all these theories is that sleep originally evolved so animals could fill a particular niche, but expanded to other roles and became accidentally load bearing. While resting, the body might as well clear out some toxins, update your synapses, and consolidate memories.
I think many of the other functions of sleep can be completed in only a few hours per night. Alternatively, it seems possible for sleep reduction therapies to perform these functions much more efficiently than natural sleep. Regardless, I’d guess there is no evolutionary barrier to sleeping slightly less each night.
Safe sleep need reduction is feasible
If sleep evolved so animals could specialize in a particular time of day, it should be feasible to eliminate sleep entirely. With an abundance of calories and no predators, we’re no longer bound by our old niche. In other words, sleep reduction therapies pass the Algernon argument.[3]
What about the other functions of sleep? Without an obvious way to perform these during waking hours, some amount of sleep is important. But that doesn’t mean we have to sleep the same number of hours we do today. Most cognitive deficits of sleep deprivation revert after only a few more hours of sleep the following night, implying that sleep can be much more efficient when needed.
It’s very common for people to get less sleep than they’re supposed to. Insomnia is an extreme version of this. There are millions of people who can’t stay asleep for a sufficient amount of time each night. That can be very uncomfortable, even maddening, but insomnia won’t kill you. This is another point in favor of the idea that sleeping (say) 7 hours per night isn’t critical for survival.
Insomnia sounds awful, it’s clearly not something sleep therapy should try to emulate. But there are people with an insomnia-like condition with none of the downsides. Short sleepers naturally sleep 4-6 hours every night, without any of the usual brain fog or irritability that accompanies it. In fact, there’s some evidence that short sleepers have more energy, less depression, and lower risk for dementia.
When people hear about short sleepers, they rightly call it a superpower. These individuals are living proof that sleeping less each night is biologically possible. How do we give this power to everyone else?
Small molecule Orexin agonists
There are several interesting areas for reducing sleep need, but I’m going to tuck most of them in the appendix and focus on the one I find most interesting, Orexin (aka hypocretin).
We only recently discovered the cause of type 1 narcolepsy. There are only 70,000 neurons that produce the neuropeptides orexin-A and orexin-B.[4] People with type 1 narcolepsy experience a loss of these neurons and correspondingly low levels of orexin-A in their cerebrospinal fluid. This leads to daytime sleepiness, sleep attacks, and a loss of muscle control that disrupts normal activities.
Since their discovery, orexin peptides and their receptors have been linked to wakefulness, appetite, addiction, executive function, and positive affect. That energy I felt after a night of sleep deprivation? Orexin. That energy you feel after a run? Orexin. The connection between activity and appetite makes sense in the context of our earlier discussion. The more active and awake you are the more calories you burn, so your appetite should increase.[5]
Could orexin be responsible for short sleeper syndrome? One paper claims that certain mutations that inhibit orexin less cause short sleep in humans and mice. But many of these papers on short sleep genes failed to replicate and Megabase isn’t having any of it. Another paper finds that people with insomnia have higher levels of orexin-A. At this point, I treat it as a tantalizing possibility, but we need better science to figure out what’s really going on.
As you would expect, taking orexins directly or using an orexin agonist (something that activates orexin receptors) increases alertness. In fact, giving monkeys exogenous orexin ameliorates the effects of sleep deprivation. On the other hand, taking an orexin antagonist (which blocks orexin receptors) causes sleepiness. Makes sense.
For these reasons, small molecule orexin receptor agonists are the most promising avenue for sleep need reduction. In the appendix, I address why I think other proposals such as gene therapy are less exciting.
To be clear, orexin agonists are only interesting if they safely reduce the number of hours of sleep needed per night. A merely stimulating effect offers little benefit over drugs like Modafinil. Encouragingly, there’s some evidence that activating orexin receptors reduces total sleep time. But nobody has explored whether this lowers long-term sleep need. We need to test whether people can use orexin agonists to sleep slightly less in a safe and sustainable fashion.
Why has nobody done this before?
Naturally, after discovering the stimulating properties of orexins, pharma companies jumped on the opportunity to create a new smart drug that would keep you alert and happy throughout the day. Just kidding! Since the discovery of orexin in the late 90’s, drug companies focused almost exclusively on doing the opposite: designing orexin antagonists to make you sleepy.
Why the lapse? Well, insomnia is more common and commands a far larger market. Narcolepsy only afflicts a small fraction of the population and is considered an orphan disease. There is some hope for narcolepsy sufferers though, Takeda pharmaceuticals is now pursuing trials of orexin agonists for type 1 narcolepsy, with some promising early results.
Aside from narcolepsy, you would think the stimulating effects of orexin would be interesting enough on their own, but there are three barriers to pursuing it. First, you can’t patent a natural hormone, so you would have to design an orexin-like drug instead, which is expensive. Second, stimulants like caffeine and nicotine are readily available and there are numerous stimulant medications so it’s hard to imagine a new stimulant being competitive. Third, even if we had evidence that orexin agonist reduced sleep need, that isn’t sufficient for FDA approval. Sleep isn’t considered a disease, and the FDA won’t support a cure for something everyone has.
Other benefits of sleep need reduction
Curing other ailments
Targeting orexin receptors for narcolepsy and insomnia is already being explored. But orexin receptors and sleep need reduction are interesting targets for other diseases too. As we’ve discussed, orexin has connections to appetite, addiction, and mood. Could agonizing or antagonizing these receptors help with weight loss[6] , substance abuse, or depression?
The link to depression is particularly interesting. Temporary sleep deprivation momentarily cures depression for reasons that are poorly understood[7] . Circadian rhythm therapies try to leverage this to treat depression over longer periods. Could a drug that reduces sleep need cure depression?
The relationship between sleep and Alzheimers is also exciting. Do short sleepers really have lower rates of dementia? Could there be a link between older people sleeping poorly and the progression of dementia? A lot of this may hinge on whether the “sleep clears out metabolites” theory is true. The existing literature isn’t super convincing either way.
Broader effects
Who would use a therapy to reduce sleep need? Many people have jobs with long hours and would benefit from lower sleep requirements during the week. Shift workers, doctors, and pilots could do their work better if there was a cure for sleep deprivation. New fathers could bear more sleep deprivation so new mothers could get more sleep.
Even people with less demanding jobs would benefit from sleep need reduction. Sleeping less means you have more time to work and more time to relax. You’ll finally have enough time to read Infinite Jest! Saving two hours per day over a 90 year life means enjoying as many waking hours as a sleeper who lived to 101.[8] The additional years would be spread throughout your life unlike traditional life extension which just lengthens your retirement.
There are more mundane benefits from reducing sleep need. Sleep deprivation costs hundreds of billions per year in car crashes and lost productivity. Extending waking hours means that commercial property will see higher utilization. Industries that benefit from more aggregate working hours may see more growth. Our most productive researchers can have more time to work, leading to more innovation.
How this research could fail
Though I’m excited about orexin agonists, I think there’s a chance that they simply won’t work for some poorly-understood reason. Biology is messy, and the cute story I’ve told about orexin and sleep could be misguided. There are many examples of a drug we thought we understood having a totally different mode of action, and an invisible graveyard of drugs that hit a well-validated target but inexplicably fail to produce the desired effect.
Our half-blind tinkering has produced miracle drugs before. I’m hopeful that exploring this new territory will produce valuable discoveries regardless of whether we end up sleeping less.
To me, the most probable outcome is that orexin agonists are mere stimulants and any sleep need reduction is accompanied by a rebound similar to other stimulants. Hopefully they at least offer narcolepsy sufferers a better alternative than existing medications. But if they don’t reduce sleep need I will be a lot less excited about them.
There are also some side effects to watch out for when testing these drugs:
- Even if a drug reduces sleep need, if a person is uncomfortable or unproductive during their waking hours then it probably isn’t worth it. The effects of sleep need reduction on memory are particularly important.Orexin agonism increases both activity and appetite. If the increase in appetite outpaces the increase in activity, patients could experience weight gain. This can be beneficial for some groups, but net harmful for most people.Orexin seems to potentiate the effects of opioids. Orexin agonists could increase the risk of drug seeking or drug addiction.If reduced sleep results in lower clearance of metabolites, long term sleep need reduction could raise the risk of dementia.
These risks aren’t a showstopper, but a good reason to proceed with caution. They also make studying orexin antagonists intriguing. Indeed, orexin antagonists are being studied as treatments for insomnia, obesity, and opioid addiction.
Conclusion
Sleep arose so the rhythm of life could move with the rhythm of the sun. Humanity left that dance centuries ago and now we have the opportunity to play a different beat. The time has come to ask what rhythms we can dance to, what rhythms we should dance to.
Orexin is the most promising route to answering these questions, at least for now. I expect as we explore it further, we’ll find new paths and connections to other diseases. I’m confident that this is something that should be explored.
Sleep need reduction has the potential to give us longer, happier, healthier lives. It’s the kind of overlooked opportunity to improve the world that I started this blog for. This is why, with collaborators, I’m fundraising for a self-experiment on sleep need reduction. More details soon.
Thank you to Niplav and nomagicpill for helping with a draft of this post.
Appendix
Objections to sleep-need reduction.
Some people recoil from this idea because it feels unnatural or because sleeping feels good. Would it secretly be bad to develop therapies to reduce sleep need? I can think of two serious arguments for why it could be bad:
- Reducing sleep need could have long-term negative health consequences.Sleep need reduction would increase inequality.
I’m very sympathetic to #1. Sleep is a critical component of health and messing with it has risks. Any research should be accompanied by rigorous safety testing. That being said, the presence of risk is not sufficient to argue that we shouldn’t pursue research. This line of argument would prevent basically all innovation.
I also think it’s possible for sleep need reduction to benefit long term health given its relationship to positive affect and weight loss. We have examples of this in practice, where short sleepers seem to be happier, healthier, and have lower rates of dementia.
Argument #2 sounds reasonable, but falls apart under further inspection. There are people with conditions that require them to sleep more each night than a healthy person. This is unfair because they have fewer hours to work and play. If a sleep-reduction therapy lowered their sleep need relative to healthy adults, that would reduce inequality. Alternatively, if the marginal value of waking hours decreases, inequality would also fall.
But there’s still a chance that sleep therapy would increase inequality by multiplying existing advantages. My response to this is that we should tackle poverty, not inequality. Reduced sleep will increase innovation, which is far more important for helping people in need than lowering the Gini coefficient.
I’ve considered a lot of possible arguments against this and I can’t find a good reason for why society shouldn’t pursue research on voluntary sleep need reduction. Such an argument would also imply that we all should sleep more, which seems silly. We should proceed with caution of course, but proceed nonetheless.
Other avenues for sleep reduction
Melatonin may reduce sleep need by a little bit, if used properly. This comes from anecdotal evidence from Gwern and others, though I’m skeptical the effect is as large as Gwern says. It would be interesting to run a trial to see how much this generalizes. Because melatonin is not patentable, companies have no incentive to run these trials.
Mazindol is a stimulant that’s used off-label as a narcolepsy treatment. It was only recently discovered that it works as an orexin receptor agonist. It was discontinued previously because it wasn’t profitable enough, but now there’s renewed interest with trials for narcolepsy, ADHD, and addiction. The effect on appetite and addiction is interesting, you would expect an orexin agonist to increase appetite and drug seeking behavior right? I don’t think we have the full story for how this one works.
S-Adenosyl methionine has antidepressant effects and has the side effect of causing insomnia (which, in my mind, is just enhanced wakefulness at the wrong time). This is particularly interesting given that many long-sleep disorders may be caused by depression, and treating one may also help with the other.
Sodium Oxybate is a common treatment for narcolepsy but the mechanism is not fully understood. It also seems to have efficacy for treating alcohol abuse.If we discover the target, we may be able to find better drugs for it.
Creatine seems to pop up as a solution to everything, and sleep is no exception. One study (blog coverage here) finds creatine reduced the effects of sleep deprivation in human subjects. The study Creatine supplementation reduces sleep need and homeostatic sleep pressure in rats has some related results, but I haven’t looked too closely.
Other Treatments
Sleep hygiene may be sufficient to reduce total time spent sleeping. Current research doesn’t focus on how sleep habits can reduce sleep duration but this new focus may turn up some interesting results.
Cognitive Behavioral Therapy can be used to address insomnia, back pain, and mental illness. It’s possible that CBT-i could also be used to increase sleep efficiency. Dealing with misperceptions about sleep quality may alleviate some of the symptoms of insomnia.
Transcranial magnetic stimulation and Transcranial direct current stimulation non-invasively apply electromagnetic fields to the brain. These techniques could be used to modify a person’s circadian rhythm (similar to melatonin) or make their sleep more efficient. Other brain stimulation techniques like ultrasound or deep brain electrodes may have a stronger effect.
Sleep duration shows a strong age dependence with babies and teenagers needing more sleep and the elderly needing less sleep. What is the source of these changes over a lifetime? Could we simulate them with drugs?
More speculatively, gene therapy is the most direct way to modify sleep traits. Some not-super-convincing research links mutations in Neuropeptide S, ADRB1, DEC2, mGluR1, and others to familial natural short sleep. But many of these studies don’t replicate. In theory we should be able to use gene therapy give healthy people the same traits found in short sleepers. In the next section I argue that this isn’t a good approach.
Why small molecule drugs are more promising
Small molecule drugs are great. We have all sorts of miracle drugs with a variety of effects. We have lots of experience researching, testing, and manufacturing them. We have lots of ways to deliver them or design them to reach different parts of the body such as the brain.
Other therapeutics like peptides or RNA are harder in every way. They’re expensive to test, expensive to make, and we have less experience trialing them. The body destroys them quickly and you need clever techniques to deliver them to your target.
And then there’s gene therapy. This is the maximum effort route to sleep need reduction. I’ve seen two proposals to test sleep gene therapies in mice. I think this is misguided for several reasons.
First, mice kinda suck for testing therapies. A lot of mouse research just doesn’t translate to humans. Testing existing drugs in humans is a more direct approach.
Second, gene therapies are having a bunch of problems right now. Even if you discovered a gene that reduced sleep need, you’d face the monumental challenge of turning that into a viable therapy, a challenge that has stumped several pharma companies. Those same pharma companies are turning out successful small molecule drugs on a regular basis.
Third, any gene therapy for sleep need would have to deliver its payload to the brain. Do you know how hard that is? Because of the blood brain barrier any therapy would require intrathecal administration (i.e. into the spine) or direct injection into the cranium. Making sure that therapy reaches the relevant neurons through dense grey matter is another problem.
Fourth, small molecule drugs are (for the most part) reversible, just stop taking them. We haven’t even started to work on this for gene therapies.
So a sleep gene therapy has to solve 3-5 very hard problems that nobody has yet solved. I think it’s better to pursue small molecule drugs first and reach for stronger medicine if that doesn’t work. That being said, I want to see far more work on sleep need reduction, so godspeed to anybody who works on this.
Other links, companies, and related efforts
Related efforts
Issak Freeman and Helena Rosengarten are pursuing “Ozempic for sleep” (love that framing!) in the George Church lab. They seem to have funding from Blake Byers and co. Their approach seems to focus on gene therapies based on short sleep genes.
Minjune Song is spearheading a separate effort, looking at genetic modifications to induce the short sleep phenotype in mice.
I’ve noted the challenges of these approaches above, but I wish them luck. At the very least, we will learn more about short sleep using these techniques than hitting the brain with small molecules with poorly understood targets.
Companies working on Narcolepsy
Takeda Pharmaceuticals is pursuing the most Orexin agonists of any company I know of.
Jazz Pharmaceuticals owns Xyrem (sodium oxybate) and is pursuing trials for one orexin receptor 2 agonist.
Alkermes and Centessa are pursuing orexin receptor 2 agonists as well.
Papers
This short textbook is probably the best single resource on Orexin: The Orexin system. Basic science and role in sleep pathology.
Orexin System: The Key for a Healthy Life
Increased plasma orexin-A levels in patients with insomnia disorder are not associated with prepro-orexin or orexin receptor gene polymorphisms. Finds that orexin-A levels are higher in insomniacs but there aren’t orexin gene mutations associated with insomnia.
Short-sleep gene’s don’t replicate: The impact of Mendelian sleep and circadian genetic variants in a population setting
Orexin antagonists for neuropsychiatric disease: progress and potential pitfalls
The orexin story, sleep and sleep disturbances
Hypocretin/Orexin Overexpression Induces An Insomnia-Like Phenotype in Zebrafish
Most sleep does not serve a vital function: Evidence from Drosophila melanogaster | Science Advances
Total Wake: Natural, Pathological, and Experimental Limits to Sleep Reduction - PMC
Neurobiology of the Orexin System and Its Potential Role in the Regulation of Hedonic Tone - PMC
Some links between orexin neurons and dementia:
Hypocretin (orexin) cell loss in Parkinson's disease - PubMed
Coexistence of narcolepsy and Alzheimer's disease - PubMed
Other links
Understanding the Orexin System: Why it Matters
Beggars in Spain is the only fiction work that considers the implications of not needing to sleep.
A Robin Hanson post from 2012 (!) on how sleep evolved: Sleep Is To Save Energy
Niplav looks at sleep need reduction from melatonin and meditation.
Curing Sleep: My Experiences Doing Cowboy Science
Orexin and the quest for more waking hours
Cause area: Short-sleeper genes
Sleep Compression An Initial Technical Exploration
Matthew Walker's "Why We Sleep" Is Riddled with Scientific and Factual Errors
See also Guzey on his updated view on sleep deprivation.
Pop-sci articles
Technology to cut down on sleep is just around the corner | Aeon Essays
Will we still need to sleep in 50 years?
Can We Eliminate the Need For Sleep? | WIRED
How natural "short sleepers" thrive on 4 hours of sleep per night - Big Think
Living with a short sleep gene: ‘It’s a gift’ | CNN
The Nasal Spray that Was Supposed to Replace Sleep
I wonder if sleep evolved to synchronize the pack as well. ↩︎
Arguably, stimulants have reduced the prevalence of napping during the day. Can we observe changes in memory consolidation, longevity, or dementia because of this? Or perhaps people that nap more have better memory consolidation? ↩︎
The orexin peptide was actually discovered by two teams independently, one team named it orexin due to it’s relationship with appetite, while the other named it hypocretin. ↩︎
This is different from most stimulants, which typically decrease appetite. One of the only other drugs I can think of with this combination is Prednisone which is used for immunosupression. It can cause increased appetite and anxiety, but all sorts of side effects are possible so I’m not sure if there’s a connection here. ↩︎
Or with other eating disorders? ↩︎
One bit of trivia I stumbled across is that depressives get more REM sleep. Serotonin reduces REM sleep which may have a hand in the mechanism by which SSRI’s help with depression. Orexin agonism also reduces REM sleep. ↩︎
Added year equivalents: 90 years 365 days/year 2 hours saved/day = 65700 hours (~7.49 years). But a sleeper enjoys only 17 hours/day of wakefulness, so this is equivalent to adding 10.58 years to a sleepers lifetime. ↩︎
Discuss
