Aug 15

Sleep cycle estimation

Singularity Experience: the lucid dreaming app for iPhone comes with a couple research tools that I used to build the app – sleep history and sleep cycle graph. Over the past 8 months, I’ve been using these tools by looking at the pattern of my midnight awakenings (I wake up multiple times each night) . The human eye is drawn to patterns and I’ve identified a number of interesting discoveries, which I will discuss in this post.


Discovery 1: The ~90 minute sleep cycle

Over the last 6 months my sleep pattern has been hectic. Even thought I continued to experience multiple episodes of midnight awakening and dream recall, the timing of recall was varying. On August 13, I found myself awakening at 90 minute intervals, as indicated in the image to the right:

  • 00:59 (green dream marker)
  • 02:37(green dream marker)
  • 04:09(green dream marker, followed by ~25 minutes of wakefulness)
  • 05:49
  • 07:03
  • 08:46

If you take a look at the timings, you notice that they are all in the 80-100 minute range. Observing the pattern over the course of the night, I was able to predict with a high degree of certainty when my next REM awakening would be, and the times of dreams I recorded confirmed the estimated REM times.

Discovery 2: The optimized bed time and partial sleep cycle

Sometimes I feel the urge to go to sleep, but I finish doing something in real life – playing a video game, programming a few more lines, listening to an audiobook, etc. To be more specific, on August 13th I felt an urge to go to sleep at 10:00PM. My eyes get that heavy feeling, I start yawning, but I pushed myself to finish whatever I’m doing in real life. By the time I finished, brushed my teeth, etc, it was close to 10:37PM. I headed to bed and found myself being unable to sleep for a very long time. I tossed and turned around, all kinds of thoughts are in my head, and that sleepy feeling is nowhere to be found!

Why was I unable to sleep, when I was so sleepy just 37 minutes before that?

I’ve experienced this phenomenon for some time, but yesterday’s and today’s experiments confirmed this. I examined my dream markers and counted back in 90 minute increments from the first dream awakening:

  • First dream awakening at 00:59
  • This means that the sleep cycle for that dream started at ~11:30 PM (23:30)
  • This means that the previous sleep cycle would’ve started at 10:00PM (22:00)
  • I felt sleepy feeling at 10:00PM
  • I went to bed at 10:37PM
  • I reported being awake all the way until 11:30 PM
  • Singularity Experience reports a sleep onset time as 75 minutes (time until the first sleep phase of the first sleep cycle)
  • This means that I fell asleep at 23:52

The calculations above confirm that by going to bed way past my preferred bedtime, I was unable to enter a full sleep cycle and remained awake until the subsequent sleep cycle.

I’m still uncertain as to what prompted my body to start the sleep cycle pattern at 10:00PM. But on August 14th, I decided to go to bed close to 10:00PM to see what would happen:

  • Bedtime at 9:47PM
  • Sleep onset in 25 minutes: at 10:12PM
  • First dream awakening reported at 11:47PM ( 23:47) – 95 minutes after sleep onset
  • Second dream at 00:47 -60 minutes after first dream
  • Third dream at 1:20 – 93 minutes after first dream
  • Fourth dream at 2:45 – 85 minutes after third dream
  • Fifth dream at 4:03 – 78 minutes after fourth dream
  • Another dream at 4:29
  • Another dream at 5:35 – 92 minutes after fifth dream
  • Another dream at 5:51
  • Another dream at 6:06
  • Another dream at 6:09

With later sleep cycles, the REM window becomes fairly large, and multiple dream awakenings may belong to the same sleep cycle. With better choice of bedtime, I observed:

  • 9 dream awakenings in 8 hours of sleep
  • Complete first sleep cycle with a dream awakening
  • More regular sleep pattern (deeper sleep between dream awakenings), more regular sleep cycles


Normally, I do not report any dreams from my first sleep cycle. Such events are very rare. Looking at the results above, I’m wandering if this is because I always miss the proper time for my first sleep cycle. I hypothesize, that by picking the correct bedtime, it is possible to improve sleep to the point where the first sleep cycle is complete, and it produces a dream at the end. More REM and deep sleep means better quality of sleep!


Additional sleep patterns

Patterns are present at ~4 hours after bedtime, 6.5-8 hours after bedtime and

I’ve observed additional patterns of dream awakenings, although I do not know what they mean. What seems to be certain is that I cannot maintain a stable pattern for long – in the top left image, you can see a fairly regular pattern of awakening at ~4 hours after bedtime, followed by a second pattern at ~6.5 hours. Both lines drift and appear to “split” into branches on some days. I’m not sure what this means or if it is just my eyes playing tricks on me.


In the bottom left image, you can see a very stable pattern of awakenings at 4:00, 5:50 and 7:30 on three consecutive days. I do not know if such pattern means anything, but it seems that I fail to maintain it for a very long time – different choices of bedtime shift the pattern.


To help study such events, I’m putting in an improved sleep graph into the app – this one would be 24 hour and include additional events. Additionally, I plan to create the same vertical stack of sleep graphs (currently visible only through graphX) to help observe changes to sleep cycles.



Observe the pattern of dream awakenings after about 4,6 and 7.5 hours after going to bed,