I have a flexible schedule so I wake up naturally almost every day, but my sleep length still has massive variance. Even though I have over a thousand nights of data, I still have no clue how long my sleep cycles last.
Here is a histogram of my time spent in bed:[1]
My average is 535 minutes, but only 40% of my nights fall within the 60-minute window from 505 to 565 minutes.
Given the theory about how sleep cycles work, I'd expect to see a multimodal histogram with peaks every ~90 minutes. But instead the histogram is unimodal. So I don't know what to do with this information.
[1] I'm reporting time in bed rather than time asleep because first of all, my phone isn't very good at knowing when I fall asleep, and second, I think time in bed is more relevant because I can't control when I fall asleep but I can control when I go to bed.
Wow, that's quite a lot of variance for natural wake-ups. Are there any big external variables? Stress before bed, drinking, morning sounds, kids/partner activity in the morning?
I sleep with the window open (to keep my room cool) but I sleep with earplugs + a fan, and it's rare that I get woken up by outside noise.
I usually don't eat any food within 3–4 hours of going to sleep, to avoid digestion potentially disrupting sleep.
I rarely drink alcohol (twice a year maybe).
I live alone, no kids/partner.
My stress levels vary but it's rare for stress to literally keep me up at night.
I take 300mcg melatonin an hour before bed and I use a blue light filter on my computer. (It would be even better to stop using my computer an hour or two before bed, but I don't do that.)
The biggest variables I can think of are
My bedtime varies from 9pm to 11pm or so.
The sun rising usually wakes me up (I have blackout curtains but the sun wakes me up anyway). Although this isn't a complete explanation because my wakeup time varies by more than sunrise time.
I have caffeine 4 days a week (M/W/F/Sa), always in the early morning. This may affect sleep quality.
Thanks for the post! What makes me skeptical of utilizing this is that it usually takes me anywhere between 10 and 45 minutes to fall asleep, so when going to bed, I have no way of knowing when to get up even if I had perfect knowledge about my sleep cycles.
I also wonder if sleep inertia is actually as bad as it seems. Does it meaningfully impact the rest of the day? Or do you just feel groggy for 15 minutes but then return to normal? Or is it more about that last bit of sleep (your unfinished sleep cycle) being a waste of time and it makes sense to skip it?
That's a good point about time-to-sleep. I'm one of those lucky fall-asleep-in-5-minutes-anytime people, and it seems my method takes that for granted.
IME sleep inertia hits pretty hard, and I can drag myself out of bed and shower, but otherwise can't really do any meaningful work for 45-60mins. So avoiding it is more important than saving just the 20 minutes of sleep
If I understand your thesis right it's that you believe your simple math is more likely to result in diagnosing the end of the sleep phase than the machine learning algorithms that are trained-based on the sleep lab data to extrapolate sensor data about heart rate, temperature, movement and oxygen saturation with the correct end of the sleep phase.
Generally the first sleep cycle is measured* to last between 70 and 100 minutes, and the other cycles between 90 and 110 minutes. Those ranges mean there’s some uncertainty in when each cycle actually completes for you; it varies from person to person.
It doesn't just vary form person to person. It varies from sleep cycle to sleep cycle.
To the extend that you do notice improved sleep from what you are doing it's likely because your ritual results in you have a clear intention about when to wake up and your body organizing to meet that intention and not because your beliefs about sleep phases match your actual sleep phases.
I was way too far in life by the time I figured out my sleep cycles. This is one of those personal metrics that’s worth learning and memorizing because once you know it, you’ll use it often—like your daily calorie expenditure, or the exact chemicals you’re allergic to, etc.
It’s a simple way to optimize your sleep, completely free, using only some background scientific knowledge and a little math.
(But this is not any kind of substitute for the basics of sleep optimization: get some exercise, don’t drink too much, don’t eat too late, don’t stare at bright screens too much, etc. etc.)
What are sleep cycles?
When you sleep, your brain activity follows distinct patterns that researchers call sleep phases. Broadly, they are light sleep, deep sleep, and REM (rapid eye movement) sleep. You move through them in order, and a single run of all the phases is a sleep cycle.
Most people go through four or five sleep cycles in an ideal night of sleep.
Your experience with sleep cycles
When you wake up near the end of a sleep cycle, you feel alert, and when you wake up in the middle of a cycle, you feel groggy and disoriented. Researchers have called this variable grogginess “sleep inertia.” I’m sure you’ve experienced both disoriented wake-ups and alert wake-ups. In the absence of illness, alcohol, or sleep deprivation, it’s almost always a matter of sleep cycle timing.
Now, how can we use this to improve our wellness and productivity?
Smart wake-up
Your phone alarm app might have that “smart wake-up” feature, where it has a range of time in which it can go off, and it tries to pick the point right at the end of a sleep cycle to wake you up. It’s a great idea, but your tech can’t do it. Sleep cycles vary in length from person to person, and the first cycle usually varies in length from the others. Your phone or wearable fitness device will measure movement, heart rate, temperature, etc., but none of those are reliably linked to sleep cycles. That’s why the “hypnogram” chart that your app produces is often erratic without consistent patterns. It’s not good data!
There just isn’t much signal here
You’d need an EEG device measuring your brain to really know. A few years ago, however, I realized I could figure this out from just my own wake-up experiences and math.
Find your ideal sleep length
The first bit of info you need to collect is the length of your ideal night of sleep. This could take a while to set up: you need a night when you’re totally free of sleep debt and when you have enough slack in your morning schedule to wake up naturally. And you should have been going to sleep at the same time the last few nights, so your circadian rhythm is as regular as possible.
Then you wake up, check the time elapsed, and record it.
Some people are so conditioned to their alarm clock that without it they’ll keep sinking back into sleep for several more hours, even without any sleep debt. It helps to set an intention before you go to sleep—literally say to yourself, several times, that you’ll check the time when you wake up naturally in the morning. That sets up your subconscious to push through and exit sleep when those four or five cycles are complete.
You might have to do this several times to be confident of your measurement. Maybe that’s annoying work, but again, this is something you’ll only need to do once (or, once per major biological change in your life), and then it’ll be useful every day.
Once you’ve got your ideal sleep length, that can be hugely beneficial already. For years I would try to get at least eight hours of sleep most nights (because that’s the standard advice), and I had a hard time waking up every morning. But some mornings, for whatever reason, I’d naturally wake up a bit earlier than eight hours. When I learned about sleep cycles I finally noticed what this was: a full night of sleep (five cycles), for me, lasts about 7h 40m.
That means every time I unthinkingly set my alarm for “eight hours,” I was sleeping an unnecessary extra 20 minutes, and during those 20 minutes I was starting a new sleep cycle, only to cut it short (!) with my alarm, causing me to feel groggy.
So now I sleep for less time and wake up feeling better. But the optimizations don’t end there.
The n-1 cycle night
If you wake up at the end of a sleep cycle, you feel good and alert. “So if I set my alarm for exactly four cycles instead of five, I’ll still wake up alert?”
Yes. I’ve run this experiment enough times that it’s not an experiment anymore. If I’m working on something late at night and I go past my planned bedtime, it’s better for me to stay awake longer to get exactly a four-cycle night, rather than rush to bed and get four-and-a-half cycles.
I wake up very alert, maybe even moreso than with five cycles (because, I suspect, there’s a physiological boost of cortisol or adrenaline or something when your full night of sleep is “cut short”).
“Wait, so I can just skip that last cycle??”
No. Although you won’t feel groggy after an n-1 cycle night, there are other drawbacks. I’ll feel drowsy in the late afternoon (and probably should nap at that time). If I do it for multiple days in a row, I start to get brief headaches and have occasional moments of brain fog. Obviously those drawbacks would keep getting worse if I kept it up.
But in one-off cases? Say, once per week? That’s a full hour and a half you can get back, for basically no cost. I suspect this is one of those things where modern life offers us maximum stability, but we’ve actually evolved to absorb some volatility without issue.
(P-hat is me)
Find your n-1 cycle time
The way I went about this is I started noticing when shorter nights of sleep gave me a more alert wake-up, recorded that info, and then later I plugged in the math to verify what was going on. An easy method: drink a lot of water before bed. You’ll wake up to go to the bathroom, and because it’s not an abrupt interruption, your body will pick the easiest time—the end of a cycle—to wake up. If it’s roughly 100 minutes earlier than your full night of sleep, that’s probably n-1 cycles. If it’s roughly 200 minutes earlier, that’s probably n-2 cycles, so record that too.
Put it all together
We want to integrate these data points to get the timing of each cycle throughout the night. First we want to know: is your ideal night of sleep four cycles or five (or more rarely, six)?
Generally the first sleep cycle is measured* to last between 70 and 100 minutes, and the other cycles between 90 and 110 minutes. Those ranges mean there’s some uncertainty in when each cycle actually completes for you; it varies from person to person.
The profile of your sleep cycle lengths looks roughly like this.
Fortunately, because we’re only dealing with a small number of cycles, the variation doesn’t add up that much. A total duration under 430m must be four cycles, according to the data we have. Durations of 520-540m can theoretically be reached by either five or six cycles; let’s make things easy for ourselves and say 530 is the cutoff there.
*Basic info from sleep cycle research can be found here (or your AI can bring it up for you).
Use the calculator to align your cycle lengths
If you know your ideal sleep duration and you know how many cycles it is, and you assume, as the research shows, that the the first cycle is shorter and the rest are all about the same length, you can get a good idea of whether your n-1 (or n-2) measurement was correct (i.e. whether it’s really showing the sleep inertia phenomenon or something else was affecting your wake-up experience).
There’s some wiggle room in how it all adds up. A first cycle of 70m and four more cycles of 103m gives you eight hours, but a first cycle of 90m and four later cycles of 98m also gives you eight hours. If you can tweak the balance of cycle lengths so that both your ideal night and your n- night are represented in the same schedule, then you can assume your information is good.
For example, my ideal sleep was 7h 40m, and that’s five cycles. My n-1 sleep was just a few minutes over 6h. If I enter the first cycle as 80m, the latter four cycles need to be 95m to add up to 7h 40m in five cycles (they can be calculated automatically). And that puts the n-1-night duration at 6h 5m, matching my other data point.
So, have a try* at that:
*If you’re reading this on Substack or LessWrong, go to the original webpage to use the calculator.
Use it
And now you know your whole sleep cycle schedule. Keep in mind that lots of factors can mess up your sleep; this is effective in the absence of those. But it is effective.
I get five cycles most nights, and occasionally I get four. On a few dire nights with poor planning, I’ve targeted exactly three cycles to minimize the damage. And on days when I’m sick or injured and want extra sleep, I target six cycles.
Your alarm clock is a good tool, but your body just isn’t designed to spring awake at any arbitrary moment. By considering your sleep cycles, you can use your alarm in harmony with your natural rhythm, rather than against it.
1
