Going to mention my crackpot theory on why polyphasic sleeping might end up killing you again:
Polyphasic sleep drastically reduces the amount of deep sleep you get. Sleep helps the body heal. Human aging and mortality seem to be modeled well by a Gompertz curve, where the thing that kills you at old age is the body’s diminished healing ability, which lets cancer precursors in the body stay unfixed long enough to grow into something that kills you.
So for all we know, throwing out 4 hours of deep sleep daily(*) for years on end might make you look a lot more like an 80-year-old, as far as the Gompertz curve modeled mortality is concerned.
ETA: (*) All non-REM sleep isn’t deep (SWS) sleep, see below. According to online hypnograms of sleep stages, the deepest sleep stages mostly happen during the first three hours of sleep, so a sleep cycle that maintains a 3-hour core sleep should be significantly better than a sleep cycle like Uberman that runs solely on power naps. The 4 hours of deep sleep bit is probably an oversimplification.
ETA2: If the first three hours of sleep are the most important for making the body heal, would a sleep cycle where you have two core sleeps daily, for example 3 to 6 AM and 3 to 6 PM, keep you more healthy than a single 8-hour sleep?
I’m currently adapting to polyphasic sleep and have been consistently getting about 2h of deep sleep during my ~3h core (according to my Zeo). Contrast that with e.g. gwern’s Zeo charts which show about 1h of SWS during an entire night of 8 hours. Note that even when his ZQ (~sleep quality) was about 100 (a “good” score) he was still getting about an hour.
...and I’m not even fully adapted. That said, I’m currently not getting quite enough REM, but I’m working on this.
About a week now. Just last night I had a much more appropriate distribution in my core, with a standard 1h of deep sleep and about 55mins of REM. This was accompanied by about 45mins of (wasteful) light sleep as well, but I woke feeling very refreshed thanks to the REM.
I’m planning to blog about this soon. Will link when I do.
I’m now biphasic—I sleep about 6h at night plus a 20min nap every day. This is by far the most stable & rested sleep schedule I’ve ever had. Before, when I was monophasic, if I tried getting a full night’s sleep, I would typically be unable to fall asleep again at the same time as the night before, after a few days. Which meant that I would stay up a bit later (since I couldn’t fall asleep anyway) and end up getting about 7h or so. On 7h of sleep, I was chronically sleep-deprived (would doze off in lectures etc) but at least I could fall asleep at the same time each night.
I’ve been biphasic for the majority of the last 8 months (except for christmas and missing a few miscellaneous days) and I love it. It just feels totally normal now.
The polyphasic phase got messed up, I believe because I made my first nap too close to my core (2.5h) and so I wasn’t properly waking up at that time. I tried fixing this over the course of a month or so, but by then I had really bad sleep habits (in terms of how I would respond to tiredness). Then I tried re-adapting (using an exaptation approach) at the end of last summer (about in time with Leverage), but it failed due to insufficient self-control / systems to keep me awake.
Then I switched to biphasic and have been really happy with it. I could see me trying a 2-nap, 4.5h core at some point, but at the moment I’m happy here.
Zeo has typically recorded about 2h each of REM, Light, and Deep sleep during my core, plus 8 mins REM during my naps.
I finally started actually charting this stuff just about a month ago, and so far can report: (all times in minutes)
Average bedtime = 230 minutes past midnight (stddev 62)
REM = 124 (25)
Light = 122 (28.6)
Deep = 104 (14.8)
Nap typically 229.6 minutes after wake (76)
REM = 124 (25)
Light = 122 (28.6)
Deep = 104 (14.8)
I am totally not a statistician, but models of sleep and my own data definitely support the conclusion that I get much more deep sleep when I go to bed earlier. Given this, I really need to start doing that. The three datapoints I have where I went to bed by 2:30, I got >= 2h deep, which only happened in one later-bedtime case (during which I also slept 7h total, fwiw, due to having missed a nap or something).
So I predict that when I start sleeping earlier I’ll be back to the 2h of deep I mentioned initially.
I’m happy to throw this data at you for analysis. In theory my zeo has been logging it since over a year ago (and decrypted, thank god) but I don’t know how to get it off so I’ve just been logging it in a spreadsheet daily. If you want the raw zeo file, you can have that too.
In theory my zeo has been logging it since over a year ago (and decrypted, thank god) but I don’t know how to get it off so I’ve just been logging it in a spreadsheet daily. If you want the raw zeo file, you can have that too.
If it’s decrypted, then you can use the ZeoDataDecoder Java library to convert it to CSV (see https://forum.quantifiedself.com/thread-zeo-shutting-down-export-your-data ). It’s what I use for my own data right now. It’s a little annoying to set up the dependencies in Debian because the error messages are opaque, but not that bad.
We get about as much REM and SWS (deep sleep) as monophasic sleepers—about 90mins each per 24hrs. This is one hypothesis to explain why so many people (me included) have so much trouble adapting to the original Uberman schedule (which, properly adapted, gives you 50+ mins each).
Hm, right. So the really deep SWS sleep seems to mostly happen during the first 3 hours of sleep, and the rest is alternation between REM and lighter sleep. Based on that, the Everyman cycle does look a lot more sustainable than full-on Uberman.
Going to mention my crackpot theory on why polyphasic sleeping might end up killing you again:
Polyphasic sleep drastically reduces the amount of deep sleep you get. Sleep helps the body heal. Human aging and mortality seem to be modeled well by a Gompertz curve, where the thing that kills you at old age is the body’s diminished healing ability, which lets cancer precursors in the body stay unfixed long enough to grow into something that kills you.
So for all we know, throwing out 4 hours of deep sleep daily(*) for years on end might make you look a lot more like an 80-year-old, as far as the Gompertz curve modeled mortality is concerned.
ETA: (*) All non-REM sleep isn’t deep (SWS) sleep, see below. According to online hypnograms of sleep stages, the deepest sleep stages mostly happen during the first three hours of sleep, so a sleep cycle that maintains a 3-hour core sleep should be significantly better than a sleep cycle like Uberman that runs solely on power naps. The 4 hours of deep sleep bit is probably an oversimplification.
ETA2: If the first three hours of sleep are the most important for making the body heal, would a sleep cycle where you have two core sleeps daily, for example 3 to 6 AM and 3 to 6 PM, keep you more healthy than a single 8-hour sleep?
I’m currently adapting to polyphasic sleep and have been consistently getting about 2h of deep sleep during my ~3h core (according to my Zeo). Contrast that with e.g. gwern’s Zeo charts which show about 1h of SWS during an entire night of 8 hours. Note that even when his ZQ (~sleep quality) was about 100 (a “good” score) he was still getting about an hour.
...and I’m not even fully adapted. That said, I’m currently not getting quite enough REM, but I’m working on this.
How long have you been on the 3h core sleep so far?
About a week now. Just last night I had a much more appropriate distribution in my core, with a standard 1h of deep sleep and about 55mins of REM. This was accompanied by about 45mins of (wasteful) light sleep as well, but I woke feeling very refreshed thanks to the REM.
I’m planning to blog about this soon. Will link when I do.
The aforementioned blog post, which I wrote days later but forgot to post (and I just saw this thread again now).
Any more news?
I’m now biphasic—I sleep about 6h at night plus a 20min nap every day. This is by far the most stable & rested sleep schedule I’ve ever had. Before, when I was monophasic, if I tried getting a full night’s sleep, I would typically be unable to fall asleep again at the same time as the night before, after a few days. Which meant that I would stay up a bit later (since I couldn’t fall asleep anyway) and end up getting about 7h or so. On 7h of sleep, I was chronically sleep-deprived (would doze off in lectures etc) but at least I could fall asleep at the same time each night.
I’ve been biphasic for the majority of the last 8 months (except for christmas and missing a few miscellaneous days) and I love it. It just feels totally normal now.
What happened to the polyphasic phase? And how does the biphasic look on Zeo?
Oh, and the first question.
The polyphasic phase got messed up, I believe because I made my first nap too close to my core (2.5h) and so I wasn’t properly waking up at that time. I tried fixing this over the course of a month or so, but by then I had really bad sleep habits (in terms of how I would respond to tiredness). Then I tried re-adapting (using an exaptation approach) at the end of last summer (about in time with Leverage), but it failed due to insufficient self-control / systems to keep me awake.
Then I switched to biphasic and have been really happy with it. I could see me trying a 2-nap, 4.5h core at some point, but at the moment I’m happy here.
Wow, I haven’t been signed into LW in awhile.
Zeo has typically recorded about 2h each of REM, Light, and Deep sleep during my core, plus 8 mins REM during my naps.
I finally started actually charting this stuff just about a month ago, and so far can report: (all times in minutes)
Average bedtime = 230 minutes past midnight (stddev 62)
REM = 124 (25)
Light = 122 (28.6)
Deep = 104 (14.8)
Nap typically 229.6 minutes after wake (76)
REM = 124 (25)
Light = 122 (28.6)
Deep = 104 (14.8)
I am totally not a statistician, but models of sleep and my own data definitely support the conclusion that I get much more deep sleep when I go to bed earlier. Given this, I really need to start doing that. The three datapoints I have where I went to bed by 2:30, I got >= 2h deep, which only happened in one later-bedtime case (during which I also slept 7h total, fwiw, due to having missed a nap or something).
So I predict that when I start sleeping earlier I’ll be back to the 2h of deep I mentioned initially.
I’m happy to throw this data at you for analysis. In theory my zeo has been logging it since over a year ago (and decrypted, thank god) but I don’t know how to get it off so I’ve just been logging it in a spreadsheet daily. If you want the raw zeo file, you can have that too.
Yes please. What questions did you have in mind?
If it’s decrypted, then you can use the ZeoDataDecoder Java library to convert it to CSV (see https://forum.quantifiedself.com/thread-zeo-shutting-down-export-your-data ). It’s what I use for my own data right now. It’s a little annoying to set up the dependencies in Debian because the error messages are opaque, but not that bad.
We get about as much REM and SWS (deep sleep) as monophasic sleepers—about 90mins each per 24hrs. This is one hypothesis to explain why so many people (me included) have so much trouble adapting to the original Uberman schedule (which, properly adapted, gives you 50+ mins each).
Hm, right. So the really deep SWS sleep seems to mostly happen during the first 3 hours of sleep, and the rest is alternation between REM and lighter sleep. Based on that, the Everyman cycle does look a lot more sustainable than full-on Uberman.
And, your body repartitions your sleep on a polyphasic schedule. My sleep really isn’t like yours any more. See the bar charts waaaay down the page here: http://trypolyphasic.com/forum/post/8455/#p8455