Note: This post is entirely about Non-24-hour sleep–wake disorder, a very rare circadian rhythm disorder I have. There is a little info just about sleep and circadian rhythms in general. If you’re just here for the fat loss, you can safely skip this one, unless you have messed up sleep.
Wait, what is Non-24?
Non-24-hour sleep–wake disorder is a circadian rhythm disorder. The circadian rhythm (literally, the “roughly daily rhythm”) is a mechanism in almost all life, starting with plankton, through plants, all animals, and humans, that lets them react to the daily light/dark rhythm set by the sun.
Practically all life reacts to light and the sun in some way. Plankton rises to the surface to absorb more energy, then drops down to avoid predation. Some animals (like humans) are mostly active during the day, where they can see and find more food. Others are awake at night, to eat the animals asleep at night. Some animals are most active at dusk and down, to sneak through in the in-between periods. Plants often have daytime and nighttime “behavior” such as orienting themselves toward the sun for better sunlight exposure.
This is performed via an internal, biochemical clock mechanism. There is of course no physical clock in any plant nor in us. Instead, there are chemicals that react to light and set in motion some sort of chemical reaction. This is how we “entrain” to sunlight.
The average human circadian rhythm is about 24.5h, and would therefore be a bit too long for earth days. If it was fixed, everyone would wake up 30 minutes later every day, and get tired 30 minutes later as well, slowly circling around the clock.
But since we have these mechanisms to “entrain,” almost all humans can compensate for the 30 minutes by getting a modicum of sunlight exposure and some other environmental factors, although sunlight is by far #1.
There were experiments done on plants, animals, and later humans kept in total darkness. Their perception of time completely went out the window. Some people stayed up for days at a time, unaware of the fact.
A circadian rhythm disorder means that your rhythm isn’t working properly. As a consequence, people with these disorders have a difficult time functioning in society. They might only be awake at night, and have a hard time staying awake in school, or keeping a job.
There are generally 4 circadian rhythm disorders:
Delayed Sleep Phase Syndrome (DSPS): these people are pathological night owls. They might wake up at noon and go to bed when the sun rises. This is the most common one. In short, their circadian rhythm is too long, and extremely difficult (if at all possible) to push early enough to fit into regular life.
Advanced Sleep Phase Syndrome (ASPS): The opposite, where people wake up extremely early and subsequently get tired very early. For example, waking up at 2am and getting tired at 6pm. The circadian rhythm is too short, and very difficult or impossible to push later.
Non-24: People with Non-24 might not have a very short/long circadian rhythm per se, but the mechanism for entrainment to sunlight is broken. Some blind people have this condition, since most of our sensitivity to light is in the eyes, but not all. People with Non-24 circle around the clock, waking up e.g. 1h later every day if undisturbed. This is what I have.
Irregular sleep–wake rhythm disorder (ISWRD): These people have a normal-length (~24h) clock and can entrain to the sun, but their sleep is fragmented, causing them to get tired during the daytime. It’s estimated that people with this condition have a very weak circadian rhythm, which makes it difficult to stay asleep at night, or awake during the day.
Diagnosis
Circadian rhythm disorders are difficult to diagnose. Until somewhat recently, there weren’t even official diagnoses. I think the classifications were only added to the DSM in 2013 with DSM-5.
One thing that makes these very difficult to diagnose is that, of course, you can still wake up/be woken up. You can go to school and sit there. You’ll just be extremely tired the entire time. Then you go home and crash at the first possibility to make up for sleep, since you’ve been tired all day. The next day, the cycle repeats. I suspect that the majority of people with circadian rhythm disorders don’t know they have them.
I’ve personally met and identified at least 2 people with DSPS who had never heard of it, and 1 other with either DSPS or Non-24 (they can present similar symptoms). When you repeatedly meet somebody in a 24/7 coffee shop at 3am, working on his laptop, that’s a hint.
The basis for diagnosis is to write down your wake-up time every day. You need to take a long enough period where you’re not woken up by any external stimuli like alarm clocks, spouses, boyfriends/girlfriends. Basically, let your circadian rhythm run free. This is also called “free running sleep.” You won’t see the distinct patterns if you wake up with an alarm every day. Most people will need to take a vacation for this. That’s how I discovered it.
I recommend at least 2 weeks, but a month is better. This is because you might start to see a trend in 2 weeks, but many patterns become much more obvious in a month.
For example, let’s look at my wake-up times somewhat recently (on keto), which are pretty normal, if mildly night-owlish.
As you can see, I mostly wake up around 9am. Sometimes as early as 8, sometimes as late as 10. There’s one outlier where I woke up before 7am. Not sure what happened there.
But the trend is clear: I wake up around 9am. Pretty consistent over March even with a handful of outliers.
Now let’s look at a period from last year, when I took a lavish vacation in Europe (nice churches!):
What the… what happened there?!
In the beginning I wake up roughly between 9-10am.
Then I begin the potato diet. Within a few days, my wake-up time begins to drift later and later. Interestingly, it then plateaus around 1-2pm, until it finally jumps to 2:30pm. Once, slightly after I stop the potatoes, I even woke up at 4pm.
After I stop the potato diet, my wake-up time stabilizes somewhat, as I’m back on keto.
The 3 outlier-low dots are a time zone shift. You can see my wake-up time quickly jumps back to its previous time after I adjust.
I’m now waking up at 2-3pm every day, which is really annoying. I’m missing half the day, barely have time to do anything before everything closes.
I decide to deliberately spike myself with carbs, to push my circadian clock forward through the night until I wake up early in the morning, at which point I’ll stop the carbs, go back on keto, and let the rhythm settle normally. (Why did I decide to push forward instead of pulling earlier? It’s much easier for me to go forward, as I have about a 25h rhythm. Since most people have a rhythm slightly longer than 24h, it’s easier to adjust later than earlier. That’s why it’s easier to adjust to jet lag when you fly west for most people.)
It works pretty much as expected. I eat rice and bread, and after a few days out of ketosis, my rhythm begins to visibly drift later again. This time it doesn’t plateau, but keeps going by almost exactly 1h per day. You can almost set a clock by it, haha.
I keep up the carbs for exactly 14 days. I had calculated 1h per day, but I had forgotten that it would take a few days to begin. At this point, I’m waking up at 2:30am. Luckily, I know that at this point my rhythm will just continue drifting naturally, and then stabilize later, if I go back to keto. Also I’m pretty tired of eating so many carbs and feeling so bloated all the time at this point.
So I go back to keto, and now it’s basically what would happen to a normal person with jet lag - I adapt to the current sunlight situation. A week later (2:30am + 7h) my rhythm has settled roughly between 9-12am. It’s honestly a pretty wide window, but I was traveling a bunch here and not exactly optimizing my sleep, beyond just continuing to eat keto. But you can see that even with various jet lag and travel during this last period, I stayed within roughly a 2-3h window.
So that’s how you diagnose Non-24.
Somebody on it will experience that sliding window all the time:
In this picture I’ve just copy & pasted the shifting window from above 4 times. If someone with Non-24 had a 4 month vacation, this is what his sleep schedule would probably look like.
Unfortunately I don’t have data from back before I started keto, so I don’t have actual wake-up data for that long. I’d totally do it just to show you, but at 1h/day shift it takes almost a month to go around the clock, and I don’t feel like taking 6 months off work just for a graph :-) Although…
The part where Keto totally fixes it
Oh yea, I almost forgot about that. Now that I’ve been doing keto for over 7 years, I often forget that I have Non-24. As you can see above when I ate carbs for 2 weeks, I can turn it on and off like pushing a button.
I found this totally by accident. At the time, I was trying to lose weight, and I decided to try a ketogenic diet. Totally unexpected, and by accident, my Non-24 was gone within mere days.
It’s not that eating keto “sets you to the correct time,” at least not for me. What it does is it fixes the entrainment mechanism. So from a few days after I eat keto (basically, when entering ketosis) my circadian rhythm suddenly begins to respond to the light signal.
If I’m waking up at night (as in the the experiment on my Europe vacation, where I woke up at 2-3am), that means I’ll continue to drift later until I hit my “normal” wake-up time of usually 9-11am. That’s a bit nightowlish, but nothing crazy. That’s not even considered real DSPS!
If I’m waking up during sunshine hours, I’ll typically just stay there, unless it’s extremely late. But as you see in the graph above, I was able to stabilize at 2-3pm after the potato experiment. While that’s stable, it’s not exactly great. This would put me squarely into DSPS territory. Of course, as you saw, I can push it around within limits.
So it doesn’t really matter when I start keto, and I don’t have to time it much. As soon as I start, it’ll pretty much stabilize the next time it comes around. The only exception is when I’m currently waking up so late that stabilizing there would put me in DSPS territory and I’d miss out on half the day. And be unable to show up for work, lol.
I don’t seem to run into any risk of getting into ASPS territory, which would mean waking up way too early. My “natural” rhythm, when my Non-24 is in remission, seems to be that of a mild to slightly severe night owl. I’ve woken up as early as 8am without an alarm, and as late as noon, but both are really outliers. Most common is 9-10am.
Discipline, lol
Besides fixing Non-24, I lost 100lbs when I first went on keto. Some people who hadn’t seen me in a while were amazed by my weight loss, and they admired my “discipline” when eating out. I’d sit there in the restaurant and drink mineral water or coffee, and maybe eat a salad, unless they had steak or burgers. Nothing even tempted me.
Of course I didn’t have any discipline whatsoever. It wasn’t even a real decision.
Imagine you’re in a wheelchair, but avoiding carbs allows you to walk again. Would you really need discipline to maintain a ketogenic diet? Keto gave me my life back, and it fixed my life-long sleep problems. Fat chance I was going to have some french fries, lol.
Not even that deep in ketosis
When I first got into keto, I was doing it pretty extreme. Pretty low protein, pretty close to zero carbs. This is also when I first lost 100lbs (COINCIDENCE?).
Later I experimented, and found that I could eat a very lenient, high-protein diet. I only ever managed to bring back my Non-24 through high protein once, and that was attempting (and failing) to eat 2 whole T-bone steaks in one sitting.
I can also drink near infinite lattes (which contain lactose), and easily eat a bar of dark chocolate (85% up) every day. 70% chocolate is close, if I go much further, I’ll fall out of ketosis enough to mess up my sleep.
When I’d measure this “dirty keto” I’d rarely get much over 0.5mmol/L ketones in my blood, but this seems enough to keep my Non-24 in remission. I’m lucky in that regard, as I think epileptics often need to go into very deep ketosis (3.0mmol/L or higher) for their epilepsy to go into remission.
But I probably am unlucky in that this lured me into eating too much protein, making me gain those 100lbs back ;-)
Mechanism
Why does keto fix this condition?
Oh, I forgot to mention it: of course ketogenic diets are NOT an acknowledged cure to put Non-24 into remission. In fact, it seems that there are almost no ways to fix it.
First, let’s take a look at how the mechanism to entrain our circadian rhythm even works. A simplified model could look like this:
Light hits receptors in our eyes first thing when we wake up and go outside (side note: this is why you should go outside in the morning!)
A biochemical reaction is set in motion that is basically a clock, causing the hormone melatonin to rise throughout the day.
Once the timer hits (=melatonin is high enough) we get tired, and a few hours later, we fall asleep (unless we force ourselves to stay awake).
Of course, blind people who lack the photoreceptors in the first place wouldn’t ever get that signal, and so their timer would never start. Note that the sensors for circadian entrainment are different from those we use to see - which explains why some blind people can still entrain. It depends on the cause of blindness.
Melatonin
Melatonin is a hormone that rises slowly during the day, until it crosses a threshold of “tired enough” and we fall asleep. It is part of the mechanism that keeps our circadian rhythm entrained.
Eliezer Yudkowsky, who also has Non-24, has been keeping his in remission using a completely different method: he supplements melatonin with very specific timing. He’s had it under control for over a decade now. I spoke with him and confirmed that he had tried a ketogenic diet, which did not help with his Non-24. I suspect there are multiple different causes, and keto only fixes mine.
Most people who take melatonin to help sleep take it when they want to go to sleep. But people with Non-24 have a different problem. Falling asleep is not difficult: it’s keeping the circadian rhythm from shifting every day.
So he came up with a pretty neat trick: he takes a small dose of melatonin every day in the early afternoon (about 9h before he wants to go to sleep), to drag his circadian rhythm earlier by one hour. In a sense he doesn’t skate to where the puck is tired, he skates to where he wants the puck to get tired… if you catch my analogy.
The Suprachiasmatic Nucleus (SCN)
This little thing in our brain is the “master clock” and controls our circadian rhythm. It is the part of the brain receiving the light signal from the photoreceptors in our eyes.
When I accidentally discovered that keto fixed my Non-24, I suspected that my SCN was broken. I knew the ketogenic diet had originally been invented for epileptics. If an epileptic seizure is a short-circuiting of a part of the brain, maybe my SCN was being short-circuited on a carb diet?
The brain is protected by the blood-brain-barrier, and fatty acids or amino acids can not pass through. This means that on a carb heavy diet, only glucose can cross and fuel the brain. Maybe my SCN was somehow unable to burn glucose? Ketones have the somewhat unique ability to also cross the blood-brain-barrier. This was presumably what helped fix some epileptic seizures: there was suddenly a fuel source where, previously, parts of the brain had short-circuited.
So my hypothesis for the longest time was just this: my SCN was bad at burning glucose, making it unable to entrain. Ketosis provided it with sufficient energy to do its job.
But recently, I’ve come to think it might be something (slightly) different.
This paper by Amber O’Hearn summarizes research into sleep on ketogenic diets.
Summary:
Many molecular pathways activated during ketogenic diets are known to modulate sleep–wake cycles, circadian rhythms, and sleep stages. Ketogenic diets often have beneficial effects on sleep at the same time as having beneficial effects on particular medical conditions. Enhancement of slow-wave sleep and rejuvenation of circadian programming may be synergistic with or causally involved in the benefits of ketogenic diets.
It seems that there are many more points of interaction where a ketogenic diet has been shown to produce at least different results, including increased brain ATP. So I’m not as confident now that it’s just as simple as “ketones → SCN works properly.” It might still be true, but there are also other points where ketosis could play a role.
Do you have a circadian rhythm disorder?
Statistically, it’s unlikely. But if you’re reading this, you’re probably weird to begin with ;-) And you might know somebody who has these symptoms. If so, now you know how to help diagnose :)
These are so rare, I’ve never met anyone in real life who also had Non-24. At least knowingly. I’ve met 2 people ever who clearly had DSPS, and one who is maybe DSPS or maybe Non-24. I’ve never met anyone with ASPS symptoms.
Here are some signs if you’re curious:
You’ve always had crazy sleep troubles that nobody seems to understand
You always seem to get really into the zone after midnight
You read and do stuff all night, but then just want to sleep during the day
There are stories about you, as a child, playing in the sandbox at 3am
At sleepovers, you were miffed that all the other kids wanted to go to bed at 4am when you just got started, so you decided to just stay awake until they woke up again
When on vacation and without an alarm clock or social obligations, your sleep rhythm shows any of the patterns described above:
ASPS (extreme early morning lark)
DSPS (extreme night owl, but stable)
Non-24 (circling around the clock)
I think the intermittent one is more difficult to diagnose just from a pattern, but I could see keto helping it, as it seems more closely related to Non-24 than the other two.
My understanding is that ASPS and DSPS are defined somewhat fluidly compared to just regular larks or night owls. E.g. you might be a very strong night owl, but if you use an alarm and limit your screen use and get early morning sun, you handle it fine. In that case, you’d probably not be considered to have DSPS. If, on the other hand, you couldn’t get it under control, and your life suffered dramatically (e.g. unable to finish school or hold a job), that would likely be considered DSPS.
I know people with DSPS that sleep until noon or 2pm, and then stay up until sunrise. They make it work, working freelancing jobs that allow them to mostly work on their own schedules. One of them I met repeatedly at the same 24/7 coffee shop, where he would work on his laptop at 3am. We were the only people there. Even the barista was asleep.
Both of them could probably get diagnosed with DSPS if they went to a doctor or sleep lab.
Will Keto fix your sleep issues?
Again, unlikely.
I don’t even think keto fixes all types of Non-24. I suspect there are at least 2 types, maybe more:
Some people just have a really long, really entrainment-resistant circadian rhythm. This is basically extremely strong DSPS. I don’t think keto will fix this. These people would basically have to live a stone age life to entrain: no artificial light whatsoever, no screens whatsoever, being outside in the sun all day, preferably doing physical labor. For some, even that might not be enough.
And then there’s the type I have: some sort of binary switch that keto just switches on. I don’t know which part of the mechanism it fixes, but it’s really like a button push for me.
Maybe we can think of these like Type 1 and Type 2 diabetes: somewhat similar symptoms, but completely different root causes and mechanisms.
In any case, thanks for your interest in my admittedly esoteric ramblings on sleep and circadian rhythm disorders. In the future, I might do a post on how to optimize your sleep in general.
Not sure if you will see a new post on on this older page.
I will be re-reading all your Non-24 posts as I try to straighten out my screwed up sleep. I have been eating low carb/ketoish for 5-6 years but that hasn't made much difference. Over the decades I have shifted from a night-owl to DSP to extreme DSP to reversed sleep schedule (10 am to 5 pm is quite comfortable). I've noticed that when my husband is traveling and I give in completely that I will first stay awake for 24 hours and sleep for 12 hours!
Glad you mentioned Eliezer, I was just about to email him this article!