- 1 Biphasic human
- 2 Biphasic learning
- 3 Biphasic sleep periodogram
- 4 Biphasic learning and sleep
- 5 Biphasic sleep
- 6 Monophasic sleep with biphasic learning
- 7 Biphasic circadian graph
- 8 Two components of biphasic sleep propensity
- 9 Biphasic performance in sleep deprivation
- 10 Summary: Napping is good!
- 11 References
In this chapter, I will show why biphasic sleep is the best sleep pattern for high productivity and for the brain health.
Temperature changes in the course of the day in degrees centigrade (courtesy of: Dr Luiz Menna-Barreto, State University of Campinas, Brazil)
SuperMemo and SleepChart provide an excellent tool to verify the claim of the biphasic nature of human sleep-wake cycles. I have collected data from monophasic and biphasic sleepers that illustrate our biphasic nature.
SuperMemo alone makes it possible to see the biphasic character of the learning performance throughout the day by charting grades over time without the need to include sleep log data. In the presented example, a monophasic sleeper, a busy father of two, shows the best learning performance in the early morning around 6 am, i.e. shortly after his natural waking time. There is a big dip in average grade scores from 11 am to 1 pm. There is a second surge in the quality of learning at around 5-6 pm:
Biphasic sleep periodogram
SleepChart alone can also be used to demonstrate sleep biphasicity. Free running sleep logs can be subject to Fourier analysis to reveal the nature of sleep periodicity. An exemplary periodogram is shown in the graph:
Exemplary periodogram of human free running sleep reveals a biphasic nature of sleep periodicity. Two basic sleep frequencies dominate this particular sleep log. These roughly correspond to 12 and 24 hour cycles.
Biphasic learning and sleep
If we employ both SleepChart and SuperMemo, we can see how waking performance changes in reference to sleep phase. The biphasic grades graph from SuperMemo (as shown earlier) can be corrected for the circadian phase that can be pretty independent of the actual clock time, esp. in free running sleep. In the presented example, a biphasic sleeper, middle-aged male with irregular sleep patterns, shows the best learning performance in the early morning (roughly around the estimated end of the subjective night):
There is a big dip in the average grade scored some 7 hours from the morning peak. There is a second surge in the quality of learning in the evening. Finally, there is a steep decline in the quality of learning shortly before sleep.
SuperMemo for Windows makes it possible to correlate recall with the circadian phase as estimated by SleepChart, which has been integrated with the program. In the presented example, a biphasic 45-year-old male shows two major peaks in alertness and learning quality during the day:
The first peak occurs in the hours 3-4 from the estimated natural waking time, i.e. not the actual waking time, which may be different. The second, slightly longer peek spans hours 12-18. There is a pronounced depression in free recall at the 8th hour of the subjective day period (i.e. wake time estimated from the circadian data, not the actual waking period). The red line shows the estimated overall alertness derived from SleepChart's two component model. In this case, the estimated alertness nearly perfectly matches the recall measured during an actual learning process.
Monophasic sleep with biphasic learning
The height of the two alertness peaks may differ in a monophasic sleeper, who will also show the same depression in recall around the 8th hour of the subjective waking day. However, characteristically, a monophasic sleeper may not get the same performance boost in the evening as biphasic sleepers due to the effects of the homeostatic sleep drive component. Even a few minute nap can result in a major boost in alertness. This has already been noticed by a prominent napping expert Dr David Dinges in his comprehensive surveys comparing habitual nappers with non-nappers (Dinges 1989).
To illustrate the difference between biphasic and monophasic sleepers, see an analogous recall graph in which a monophasic 15-year-old non-napper shows the best performance in the morning hours with a sharp dip at the 8th hour of wakefulness coinciding with a subjective decline in cognitive function:
After a temporary dip, there is a sharp recovery, and a gradual decline in performance in the second half of the day. That decline is strongly accelerated by a homeostatic mechanism. The yellow line shows the estimated circadian component of alertness. In this case, the circadian benefits are muffled by the homeostatic decline in alertness, which is not shown in the graph. This is why the hypothetical circadian alertness and the actual alertness match only in the first half of the day.
Biphasic circadian graph
There is a biphasic twist to the two-process model of sleep regulation. In free running sleep, where sleep is a true expression of sleep propensity, it is possible to visualize both the homeostatic and the circadian components of sleep in a circadian graph:
In a habitual napper, the circadian biphasic nature is paradoxically expressed by the two-peak sleep propensity curve instead of the circadian curve. The reason for this role reversal is the physiological difference between the two circadian peaks in sleep propensity. In a habitual napper, sleep is initiated as easily at siesta time as it is initiated at night. However, the length of sleep at siesta time is very short (usually 15-80 min).
In the presented graph, the blue line corresponds with the ability to initiate sleep at any given circadian time. On the horizontal axis, it aligns well with the alertness graphs displayed in SuperMemo (as shown in earlier paragraphs). It aligns well with both the learning data, as well as with the two-process sleep model implemented in SleepChart.
The red line corresponds with the ability to maintain sleep. It reveals what is not visible in the alertness graph shown earlier: siesta naps cannot last long and will always be subject to an early natural termination (low red line under the first blue peak). In contrast, the period of subjective night is the only time of day when sleep can and should last longest (usually no less than 4-5 hours). The red peak is also the reason why polyphasic sleep adepts crave for "core sleep", wake up groggy, and need heavy alarm artillery to wake up in this critical subjective night period.
David Dinges, in his surveys noticed, that napping more than once within a day was extremely rare. Most nappers took naps lasting 15-120 min. Naps will be shorter if they are taken before the siesta peak. If they are taken after the peak, they will usually last longer, and may even integrate with the night sleep in cases of particularly large delay, or where there is a sleep deprivation, REM-sleep deficit, or any other form of "sleep debt".
Dinges noticed that both appetitive (habitual) and replacement (compensatory) nappers tended to time their naps 7-8 hours from waking (see: Best timing of naps). Even though napping habits may differ, the circadian timing of the siesta trough seems to be pretty similar across the population (Dinges 1992)
It is important to note again that the evening boost in alertness is magnified by a nap, but shows up also in non-nappers and can easily be deconvoluted in the two-processes model into its homeostatic and circadian components (as shown in the next two examples).
Two components of biphasic sleep propensity
The last two graphs show the impact of the circadian and homeostatic components on alertness.
In the first example, a free running female 29-year-old non-napper shows an alertness dip in 8-9 hours since waking. The red homeostatic estimate shows no dip and a steady decline over the waking day:
The yellow circadian estimate shows the expected position of the dip and the evening crest that explains a boost in the evening learning performance:
Both the evening recall boost and the evening circadian estimate align pretty well showing once again that the overall alertness depends on both homeostatic and circadian components of the sleep control system.
Biphasic performance in sleep deprivation
Mid-day slump is as prominent in conditions of severe as well as mild sleep deprivation. This graph shows a mid-day alertness slump in a 26 hour sleep deprivation study (Czeisler et al. 2006). The timing of the slump (hours 10-12 of waking period) indicates that the preceding sleep episode was positioned suboptimally (hence the need to interrupt sleep for the study). Natural awaking would probably take place 1-2 hours after a forced awakening in lab conditions. The graph also shows that sleep inertia caused by forced awakening from Stage 2 NREM or REM sleep causes a much greater cognitive decline than 26 hours of sleep deprivation.
Summary: Napping is good!
From the above charts we can conclude that human circadian pattern is definitely biphasic. Here are the implications:
- Humans are biphasic in nature and show a circadian boost in learning in subjective evening hours
- Non-nappers show a mid-day dip in performance and might also benefit from a siesta
- In free running sleep, healthy humans tend to fall into biphasic sleep pattern
- Siesta is a well-tried old practice in many cultures around the world. It is a healthy practice
- Dinges D.F. and Broughton R.J., "Sleep and Alertness: Chronobiological, Behavioral, and Medical Aspects of Napping," (New York: Raven Press, 1989), 171-204
- Dinges D.F., "Adult napping and its effects on ability to function." In "Why we Nap: Evolution Chronobiology and Functions of Polyphasic and Ultrashort Sleep," edited by C. A. Stampi (Boston: Birkhäuser, 1992), 118–134
- Wertz A.T., BS, Ronda J.M., MS, Czeisler C.A., PhD, MD, and Wright K.P., Jr, PhD, "Effects of Sleep Inertia on Cognition," The Journal of the American Medical Association / Volume 295 / Issue 2 (2006): 163-164, doi: 10.1001/jama.295.2.163