How long should we sleep?

Jump to: navigation, search

This text is part of: "Science of sleep" by Piotr Wozniak (2017)

Optimum length of sleep

In this busy modern world, every minute of time seems precious. For some people, the bottleneck resource is time (not money, material resources, people, etc.). Time becomes a limiting factor, and everyone looks for ways to decongest one's life. As sleep takes a third of our lives, a widespread ignorant solution is to cut down on sleep to economize more time for work. This might work in a short run for someone who needs his legs or arms more than his brains. It will definitely backfire for those who use their brains as the primary tool. Even for someone who believes he can accomplish more on limited sleep, life must feel like a race without the rays of happy sunshine.

A single creative insight produced by a refreshed mind can match a thousand man-hours in backbreaking labor

Millions of young lives wasted on the fronts of World War I must have equalled to less of a meaningful contribution than a few hours of programming on the part of Tim Berners-Lee. Imagine all those lives spent on more productive pursuits! By cutting down on sleep, we undermine our chances for a meaningful creative contribution.

Recommended 8 hours of sleep

Sleep researchers often look for a recommended amount of sleep. Using surveys or lab sleep data, they often come with a recommendation of 8 hours of sleep per night. However, this recommendation opens a minefield of problems. To get their 8 hours, some people may wish to go to sleep too early and thus exacerbate their insomnia and related stress. The 8-hours-per-night recommendation is also scorned by some researchers who promulgate the false claim that sleep is like food and we will always want more even if we do not need it.

Energy conservation theory of sleep is patently wrong. Benefits of sleep, unlike the benefits of food, cannot be accumulated in advance and there is no evolutionary advantage in getting more sleep than necessary. An all-nighter will be as painful after a month of oversleeping as it is after just 2-3 nights of good sleep. Conservation of energy is minimal, and the brain may actually use more oxygen during some sleep stages than when working on a complex task. Even though lions might sleep 20 hours per day when there is shortage of food and water, humans, in normal circumstances, can only binge on sleep after periods of sleep deprivation, or when sleeping in a wrong circadian phase, or when they experience health problems. In theory, neural network optimization could benefit from some additional sleep, however, the brain does not seem to crave that extra optimization. Just the opposite, it clearly looks for the optimum balance, and attempts to fill all the remaining time of the day with pursuits based on the learn drive.

In the natural creativity cycle, the brain clearly goes for the optimum amount of learning combined with the optimum amount of consolidation and restructuring. Sleep control mechanisms are set to make sleep last for a limited period of time each day, even if we tried hard to get more sleep. Not only there is no advantage, there are huge costs to sleeping too much: we are most vulnerable and defenseless in sleep. Even though sleep can be compressed, proving it is not perfectly efficient, there are no natural and healthy methods of sleep compression. The best sleep is accomplished when all circadian, homeostatic, genetic, and neural mechanisms run in synchrony at the right time on a prescribed course. This can only be accomplished with free running sleep. With dozens of SleepChart submissions, I can demonstrate easily that once a regular sleep schedule is adhered to, the total amount of free sleep drops and becomes pretty steady. In the exemplary graph, a DSPS subject runs her sleep free and gets on average the same regular recommended 7.9 hours of refreshing sleep per night even though, before running free, she was convinced that she needed 9 hours, and that, even on 9 hours, she would still be tired throughout the day:

Sleep timeline of a middle-aged female who decided to free run their sleep after she had been diagnosed as a severe case of DSPS

Saving time with polyphasic sleep

Uberman sleep schedule was proposed with a view to gaining more hours in a day. Polyphasic schedules are very appealing in theory, and many people tried them out just to give up within a week or a month (depending on the ability to suffer through the mental misery). Those who try to adjust to any unnatural schedule will suffer an unspeakable torment of the mind. Polyphasic sleepers regulate their sleep with an alarm clock until they reach the breaking point. Human self-experimenting guinea pigs collapse into a sound life-saving 5-8 hour sleep towards the breaking point and then resume the polyphasic schedule with a sense of guilt. That sense of guilt is calmed with exculpatory terminology such as "weekend break", "re-energizer", "bonus sleep", etc. Polyphasic experimenters may happen to sleep less but their intellectual performance will be dramatically undercut. Some polyphasic sleep theories are based on the false premise that the body can adapt to any sleeping rhythm. Researchers tried to find a natural polyphasic rhythm that would minimize the pain of sleeping little. For that purpose they have studied the phase response curve of the circadian rhythm, where the impact of various sleep affecting factors is shown to move the sleeping schedule forward or backward. The obvious conclusion is that we can rather painlessly move the major circadian sleepy time little by little in a desired direction. However, a healthy normal individual will not be able to chop the rhythm into a desired number of pieces. Monophasic sleep or biphasic sleep are the norm in healthy individuals. Biphasic sleep is rarely composed of two major sleep episodes. Usually it has a form of a major episode (nocturnal sleep) and a minor episode (siesta). Great catnappers nap when they feel they need to. Often, they can accurately predict when and how much they will need to nap. If you want to minimize time spent sleeping and maximize your learning results: free run your sleep. Polyphasic sleep is not the answer. Get rid of the alarm clock!

Minimizing sleep time

All forms of sleep control with an alarm clock will increase the overall demand for sleep. This means that:

If you use an alarm clock, either:

  1. you will sleep longer, or
  2. you will feel more miserable.

Artificial sleep schedules will dramatically reduce your mental capacity. A healthy individual in normal conditions will find it difficult to fall asleep 4 hours after the main sleep episode unless that episode was unnaturally cut with an alarm clock resulting in sleep deprivation.

The shortest healthy sleep is accomplished with free running sleep!

In free running sleep, once you know your average sleep time and your optimum wake time, try to stick to it religiously. Use SleepChart to find your optima in case your sleep is irregular. Plan your day in such a way so as to be sure that if sleepiness comes earlier, you can hop in to bed in a wink, and while sleepiness does not arrive in time, you can get busy with some low-priority sleep-conducive activities that will tire you until the right time for sleep comes.

Inefficiencies in the sleep control system

There are many factors that might increase the demand for sleep (e.g. learning, exercise, etc.), or shift the sleep control balance to favor sleep over wakefulness (e.g. brain injury, infection, poisoning, hypothermia, etc.). The impact of external factors can be used to illustrate a degree of inefficiency in the sleep control system. Since days are longer in the summer than in the winter we do tend to sleep a bit longer in winter. There does not seem to be an increase in the need for the neural function of sleep in winter. If we sleep more in winter and there is no biological need for more sleep, then it seems that we must be getting either more sleep than we needed in winter or less sleep than we need in summer. As sleep is primarily controlled by the circadian and homeostatic sleep propensity, and the circadian component is strongly influenced by light, variations in the levels of illumination will cause variations in sleep duration. It is conceivable then that we sleep less efficiently in winter (in terms of neural effects per unit time). Equally well, summer sleep might be less restorative. Eskimos cut off from civilizational influences sleep for a few hours more per day in winter. Dr Jim Horne is right saying that in some circumstances we might sleep more than we really need to. However, he goes a step too far when he compares sleep to eating, which makes some people believe that sleep restriction might be beneficial (by analogy to calorie restriction). In conclusion, we need to realize that sleep control mechanisms are not perfect, however, we have not yet come with any artificial and certified ways of improving upon what we were given by the biological evolution. Natural free running sleep is still the best way to accomplish healthy, refreshing and shortest-lasting sleep.

Length of sleep among users of SuperMemo

A survey of users of SuperMemo (SuperMemo World 1994[1]) revealed that the average speed of learning was 243 items/year/minute. Users who sleep less than 7.5 hours learned at the speed of 240 items/year/minute. Those who sleep more than 7.5 hours learned at 256 items/year/minute. Amount of sleep, smoking, and exercise were poorly correlated with the speed of learning. Students aged 28 years old or younger learned at the speed of 264 items/year/minute, while those above 28 years old learned at the speed 179 items/year/minute. Remember, however, that in another study it has been shown that good students learn slower (Gorzelanczyk et al. 1998[2]) because of their higher self-criticism in providing grades.

People who sleep less live longer?

"People Who Sleep Less Live Longer" screamed news headlines in February 2002. The reason for the uproar was a large-scale study by researchers from the University of California at San Diego who found that people averaging 8 or more hours of sleep per night were 15% more likely to die within the 6-year period of the study than those who slept seven hours. The study makes a valuable contribution to our knowledge of sleep habits but conclusions amplified by mass media are not only wrong, they are dangerous! If you decide to cut down your sleep today to live longer, you will certainly achieve the effect opposite to the one desired. It is your body (actually the brain) which knows best how much sleep you need. This might be five or it might be nine hours. We differ a lot in that respect. There are no noteworthy benefits of cutting down your sleep with an alarm clock, and the dangers are well documented. The erroneous conclusions media drew from Dr Daniel Kripke team study come from a typical cause-effect relationship confusion. It is not that long sleep is detrimental. It is more that poor health may increase the demand for sleep. Driven to extremes, comatose and bed-ridden patients will bias similar statistics. People with poor quality apneic sleep are more likely to linger in bed and report long nights. On the opposite side of this spectrum are people with healthy and sound sleep habits that often feel refreshed with as little as five hours of sleep, and wake up naturally before the alarm time. In addition to having adverse health effects, sleep deprivation is a major cause of traffic accident and causes immeasurable damage to nations' creative potential. Even a poorly designed alertness test is not likely to testify to your sleep's quality. This comes from the fact that stress hormones often mask sleep deprivation. However, if you try to learn with SuperMemo after an artificially shortened sleep, you will see that your recall gets worse and stress hormones may improve your sense of alertness, but they will do so at the cost of focus and memory. You will achieve best health by getting as much sleep as your body calls for in conditions that eliminate stress, stimulants, anti-depressants, sleeping pills and the like. Once more, the mass media amplifier is likely to produce confusion and negative ripples that will keep on reverberating for years to come!

As always, some research seems to make headlines, while more thorough meta-analyses don't. In this case, it is probably the self-comforting thought "even if I feel miserable in sleep deprivation, loss of sleep might actually prolong my life!" If we review the literature on the association between the length of sleep and longevity, we are likely to notice that very short sleep, as much as very long sleep, correlate with shorter lives (Cappuccio 2010[3], 2010[4]). Professor Francesco Cappuccio puts it best saying: "while short sleep may represent a cause of ill-health, long sleep is believed to represent more an indicator of ill-health". If you take an average of the optimum amount of sleep for all members of the population, you will arrive at a specific number that is meaningless for a specific individual. In sub-populations that sleep longer or less than the average, longevity may be diminished. However, for each single individual, the optimum number is the one that is suggested by the body needs. If it is 4 hours or 10 hours, it matters less as long as the number comes from the natural sleep that is not controlled artificially. Moreover, that number will differ from day to day, it will be less before an exciting date, and it will be more after a day of heavy exercise. No one should worry about sleeping 4 hours per day or 10 hours per day, as long as he or she sleeps naturally, wakes up naturally, and feels refreshed.

The optimum amount of sleep differs from person to person, from day to day, and is best determined by sleeping without artificial control such as an alarm clock, sleeping pills, etc. You can best determine your optimum sleep needs by trying free running sleep.

Epidemiological studies that focus on morbidity and ask "how many hours per night do you sleep?" ask a wrong question! They should rather ask: "Do you artificially modify the timing and the length of your sleep?" Only that kind of question would tell everyone that artificial control of sleep increases morbidity, while the actual length of sleep is largely irrelevant.

Investigating links between sleep length and longevity is not much more useful than my own failed attempts to connect the sleep length with the quality of learning. In case of learning, short sleep produces poor results because of the impact of sleep deprivation on attention, recall, and consolidation. However, long sleep produces poor learning as well because it usually is an indication of something going wrong either with health or with the sleep control systems (e.g. sleeping in a wrong phase, or compensating for prior deficits). Asking about how much sleep we need is not different from asking how many calories we need: it depends on our size, our current fat level, our caloric expenditure, and many other factors.

Jim Horne and Daniel Kripke

Jim Horne and Daniel Kripke are two sleep researchers who seem to stand in opposition to the rest of the field in their prominent claim that sleep is like food and we can get too much of it. They even contemplate the concept that, as with caloric restriction, sleep restriction might prolong life! I mentioned Dr Kripke's research that is often erroneously interpreted as "short sleep prolongs life". It is not that long sleepers die earlier. The obvious interpretation of epidemiological studies is that sicker people often sleep longer. Drawing an analogy with calorie restriction is as weak as proposing a "wake restriction" that might have some unknown benefits, esp. that in the hormonal spectrum of glucose metabolism, wake restriction is more similar to caloric restriction. There is no evolutionary advantage to getting excess sleep due to the fact that sleep cannot be accumulated like fat can (see: Excessive sleeping). Metaphorically, if we compare sleep to garbage collection, there is no advantage in collecting garbage ahead of time. People on free running sleep schedule quickly reduce their total demand for sleep and sleep less than on various forms of regulated schedules. Their mental energy is naturally much higher despite sleeping less. I believe that comparing long sleep to overeating is particularly harmful. It sends wrong signals to teenagers and students for who their brain performance determines their future. Drs Horne and Kripke's main concern is that "scare tactics" employed by researchers who insist on the value of the proverbial 8 hours of sleep may worsen insomnia and stress related to not getting "enough" sleep. These are valid concerns, but these can easily swing the balance too far in the other direction, while there is a golden mean: free running sleep that helps people get exactly as much sleep as is needed. To seek some counterbalance, let me then nitpick at some of Drs. Horne and Kripke's statements and hypotheses to throw some light on sleep needs from the free running sleep perspective.

Dr Horne

Here are some of Dr Horne's statements that keep detracting from the value and power of sleep:

  • "The amount of sleep we require is what we need not to be sleepy in the daytime". It is hard to disagree. However, I would upgrade that statement to an Information Age status: "The amount of sleep we require is what our brain asks for". This is a higher standard than just "not to be sleepy". After all, it does not take much sleep to feel well-energized for work in a field, or a walk in the park, or a basketball game. It takes more quality sleep to be at one's intellectual best for writing, programming, learning, research, etc.
  • "Sleeping continuously through the night is quite a modern invention". That's plain false. Many species sleep for many uninterrupted hours. This most certainly has been the habit of early humans. Natural sleep discontinuities, like nocturnal awakenings or segmented sleep, are a far cry from the forced interruptions inflicted by alarm clocks. Waking up for a specific purpose might have been invented in pre-historic human societies, however, there has never been a compelling need for a frequent use of that "invention". For centuries, even candles were available only for the rich. Frederick the Great might have been woken up by the firing of cannons in childhood, but his contemporaries, thankfully, did not have madly authoritarian military-obsessed fathers. Our ancestors never paid much attention to sleep. They just had plenty of it and it was never a subject of discussions or anxieties that now flood the forums for insomniacs. The true advent of alarm clocks and shredded sleep schedules and sleep problems dates back to the early days of electric lighting that disconnected the modern world from the natural cycle of daylight. Even though alarm clocks are not new, they only entered massive use in the last century.
  • "There's a lot of fearmongering about alarms. You hear some funny ideas - that we should be woken up "naturally" by light or at certain points in our sleep patterns". Being woken by light is by far healthier than being woken by an alarm clock. Still, morning light is not necessary to accomplish top-quality sleep. See a case of perfect sleep in the absence of morning light in an octogenarian. Being woken at the lightest phases of sleep is by far healthier than being woken just 15-25 min. later when deep sleep sets in. Most people have experienced the difference and can easily list a couple of symptoms that are indicative of the negative impact of sleep disruption on health. If I am a "fearmongerer" in reference to alarm clocks, it is because I understand the physiology of sleep well enough to know that life without alarms should be a basic human right esp. at younger ages.
  • "Insomnia is not really a sleep disorder but a disorder of wakefulness intruding into sleep". A typical cause of insomnia in an otherwise healthy individual is bad sleep timing. People who need to get up early, go to sleep too early in reference to their body clock, and just cannot fall asleep. It is not wakefulness intruding into sleep, but a human wish to impose sleep onto wakefulness. These are modern schedules and expectations that are the root of the problem in a vast majority of cases.
  • When criticizing psychological tests used in studying sleep deprivation, Horne says "tests simply measure tolerance to boredom, which shortens with sleep deprivation. Take the stimulation away and the animal will fall asleep". Creative work will not always provide sufficient excitement to combat sleep deprivation. It is not the boredom that is the enemy of a sleep deprived creative brain. It is the mental effort itself. No amount of boredom can put a well-rested brain to sleep. However, mental effort in sleep deprivation may be tantamount to torture. Millions of students in the industrialized nations experience that daily. If we want to measure the impact of sleep deprivation on cognition, we need tests that are both stimulating, and tests that eliminate the excitement variable. Tolerance to boredom does not shorten with sleep deprivation! It is a fresh creative mind that truly abhors mental nothingness! It is our ability to stay awake in the absence of arousing stimuli that really gets shortened in sleep deprivation. Boredom does not induce sleepiness. There is no known boredom nucleus in the brain that would start firing in response to lack of stimulation (other than via disinhibition of the natural sleep control system). Even if that nucleus was ever identified, it would rather form a part of an arousal system. Children, when deprived of stimulation, may become cranky or aggressive. Evolution made us smart because we seek action and low entropy states. Boredom unmasks the underlying sleep deprivation. Whoever nodes off at a boring lecture must have simply not gotten enough sleep in the night. A refreshed mind will instead drift to more interesting thoughts such as an exciting project, problems of the day, or an attractive classmate sitting nearby. My impression is that sleep researchers who bring up the effects of boredom on wakefulness must extrapolate from their own sleep deprived condition. Top quality creative brain does not need a whip of adrenaline or dopamine to stay focused on its creative pursuits. We should all aim at sharp and crisp minds, not just minds that are "not sleepy".
  • Dr Horne brings up the Inuit as an example of human capacity to sleep in excess. The Inuit are able to sleep for up to 14 hours in winter. It is true that lifestyle and environment will affect the length of sleep. You can blame our biology for not doing its best of the available time, but nothing natural can be done about this, so we might as well enjoy a full night sleep. When irregular sleepers claim they need 8-9 hours of sleep per night, one might expect them clocking 10-11 on a free running schedule if the "wastefulness of sleep" theory was right. Instead, in free running sleep they often go down to 6-7 hours per night and claim feeling better than ever! As for the Inuit, part of those excess hours might come from the documented wish to minimize the exposure to cold in winter months. Much of that "sleep" might be Wehr's segmented sleep (Wehr et al. 1992[5])). So we would need to employ polysomnography to see how much of that bedtime actual sleep is. On the other hand, sleeping little in summer can easily be explained by the fact that it is not easy to fall asleep at midnight in bright light, and there is a pressure from the other end of sleep to get up early for a hunt with a stress of possible oversleeping. We know that the sleep control system is imperfect and the actual need for sleep may be masked by arousing stimulation. Inversely, Arctic night might be conducive to leisurely sleep that may exceed the actual needs of the organism. However, we cannot state with certainty that it is the lack of stimulation that makes Inuit wastefully sleep longer. We know for sure though that extra summer stimulation makes Inuit get less value from sleep than they otherwise get in winter days. To settle the issue we would need to apply appropriate cognitive tests in summer and in winter, and compare the two. In particular, I would be interested in comparing the circadian peaks of recall. This is more important than just the recall average as I believe that creative people change the world primarily in their moments of peak mental strength, i.e. shortly after waking up from a refreshing night of dream-rich sleep. In my own work, I did not find much correlation between the length of sleep and the quality of learning. Free running sleep might be slightly longer in winter, however, shorter sleep in summer is often compensated by longer naps. On average, cognitive performance seems to be comparable. SleepChart logs emphatically contradict Dr Horne's claim that we like to sleep wastefully. Free running sleep decreases the total sleep as it increases its efficiency!
  • "Researchers in America tend to take a different view. They assert that because most of us can extend our daily sleep, we must need to do so. This would mean that people who seem content with seven and a half hours of sleep a day during the week but enjoy nine hours at the weekend are, unknowingly, chronically deprived, and actually need nine hours every day. Evidence for this is said to come from the many people who are sleepy in the daytime, and the numerous reports of people falling asleep at work and, worse, while driving". Dr Horne's annoyance with the 8-9 hour sleep recommendation is understandable. After all, people have different sleep needs, and the recommendation might compound insomnia. However, sleeping more on weekends than on working days is a clear indication of sleep deprivation. Stress or passion associated with work can easily mask deprivation. That mask falls off on weekends. It is hard to say if 7.5 hours on workdays and 9 hours on weekends equals to a 7.8 hour optimum, or perhaps an 8.5 hour optimum, esp. than those numbers will change from day to day even for very regular sleepers. Everyone should determine his or her needs on their own in free running condition. This condition cannot possibly be approximated by sleep on weekends due to the sleep debt carried from the working week.

Dr Kripke

Here are some statements from Dr Kripke which undervalue the importance of undisturbed sleep:

  • "Nobody seems to know exactly where the idea that we should sleep eight hours came from. I guess it was just passed down from somebody's grandmother". The "8 hour myth" does not come from grandmothers but from surveys in which many people claim that they function best once they get their 8 hour sleep. The actual perfect number may be less because of a common misperception. People who do not get enough sleep during the working week often compensate on weekends when they do sleep more to make up for various sleep deficits. That weekend sleep may be 8 hours or more, but it does not imply that this is what is actually needed. The same people on a free running schedule may be surprised to discover than they actually need just 6-7 hours. It is not uncommon to go down to 4-5 hours per night on a biphasic schedule, where the remaining sleep needs are met with a well-timed siesta nap. My small-sample estimation seems to indicate that people overestimate their sleep needs by 10-15%. Dr Carskadon's research shows that 8 hours may be a good estimate for teenagers who clearly have increased sleep needs. My own analysis of SleepChart submissions also indicates that for teenagers, 8 hours is a solid recommendation! The figure also often shows in other publications (Dinges et al. 2000[6]). When a software glitch in the first two days of 2011 kept iPhone alarms silent, thousands of people realized their capacity to oversleep. Not by just 10-30 min. Some angry bloggers clocked extra 6 hours of sleep illustrating the monstrous degree of sleep deprivation. When grannies suggest 8 hour sleep, they are not that far off the mark!
  • Kripke in his editorial in Sleep (February 2004) wrote: "The results [of epidemiological studies] falsify the widely-circulated hypothesis that it is best to sleep at least 8 hours". All epidemiological studies contribute to our knowledge one way or another. However, a study that focuses on morbidity and mortality that asks "how many hours per night you sleep?" is a waste of time! Like a study of the impact of height on mortality, we know ahead of time that it will provide U-shaped results with no bearing on our lifestyle recommendations. The correct question would be "Do you cut your sleep short with an alarm clock?", or "Do you delay sleep despite being sleepy?" or "Do you try to sleep early even if you are not sleepy?" In short, those questions should ask: "Do you artificially modify the timing and the length of your sleep?" Those studies would tell everyone unequivocally that artificial control of sleep is deadly. Length of sleep is largely irrelevant! For more see: People who sleep less live longer?
  • Kripke: "Recent results from the Hordaland study in Norway showed that working people who reported sleeping less than 6.5 hours a night did not suffer any significant increase in disability. Those who reported more than 8.5 hours were more than twice as likely to become disabled as those who slept 6.5-7.5 hours. The suggestion from this study was that, if anything, spending more time in bed might tend to increase disability". We already know that on average people live longest if they spend a certain number of hours asleep, or if their height is somewhat above average. But the statement seems to suggest a reverse causality. Trying to spend less time in bed would be as good for longevity as trying to shrink people who grew beyond a certain height. Healthy people sleep less, but those who shorten their sleep are less healthy!
  • Kripke: "We know that we do not suffer some horrifying loss of intelligence or memory every time we get less than eight hours of shut-eye. We make deliberate choices to sleep seven hours or six hours, feeling that the practical benefits of having the extra time outweigh whatever sleepiness might result from sleeping less than eight or nine hours". It amazes me to hear a sleep researcher accepting a practice of trading sleep for "practical benefits"! Natural creativity cycle explains how moments of lost sleep may translate to lost insights worth a fortune. If loss of IQ is not horrifying, the death toll on the roads definitely is. There is a neural and health cost to trading sleep for "practical benefits". An irreplaceable physician or a firefighter might do the trade. However, they should always remember that they sacrifice their own creativity, brain and health for the benefit of others!
  • In a debate with Dr Stickgold organized by the Economist, Kripke countered Stickgold's claim that insufficient sleep is the cause of traffic accidents: "People claim that daytime sleepiness causes road accidents, when the fact is that the sleep-related road accident rate goes up late at night. It is driving late at night, which causes most sleepiness on the road. People who stay up unusually late, such as physicians called to duty at night, have extra driving risk and risks of mistakes, but that does not prove that the regular sleep patterns of the general population are too short." When two great researchers disagree on a fact that can easily be verified, I can always suspect professional or personal prejudice. Could it be that early riser Kripke feels most impaired when driving in the evening, while owlish Stickgold feels out of sorts in the morning? I had a peek at the statistics and noticed that overwhelmingly the accident rates follow the average of the human circadian sleep propensity. This would mean that both doctors have a point. It is probably a chronotype that will determine the difference in accident chances at different times of the day on average. The circadian phase while driving would be even a better determinant. My guess is that whatever impairs recall or memory consolidation in the waking day will also impair driving. In other words, alertness graphs plotted in SuperMemo could easily be used to determine the probability of a traffic accident while driving at any time of the day. An interesting difference shows up when investigating accident rates across different ages. There is a huge spike in accidents at 6-7 am in middle-aged people. That would contradict Kripke's claim. However, for younger ages, the rates are very high throughout the night. For retirees, who do not need to get up early, and do not need to drive insanely long into the night, the peak aligns with their siesta time. Whatever the truth, Dr Kripke is definitely wrong when using his statement to defend short sleeping. According to National Sleep Foundation, people who sleep less than 5 hours per night are four times as likely to be involved in a crash than those who sleep 8 hours or more. Even those who sleep 6-7 hours double their risk. Obviously, as argued throughout this article, it is not the length of sleep but its quality that matter. Those who get 5 hours in free running sleep need not worry.

Cognitive bias


Early risers find it difficult to understand the problems of evening type people. They are prejudiced by their own condition. Scientists frequently divide into fiercely opposing camps that are often based on serious prejudices coming from trivial sources. That's good. Discoveries benefit from passions even if they often and inevitably cross rational boundaries. I am seriously prejudiced too. So are those magnificent scientists who I dare to criticise. I bet that when Drs. Horne or Siegel read about new ideas coming from Drs. Stickgold or Walker's camps, they might be mumbling to themselves "Oh no! Not again..." If they ever come to read the presented text, they might exclaim "Who's that [bleep] Wozniak [bleep]!?" That's healthy. I would be honored if they bothered to read.

I bet that some researchers who are short sleepers or early risers themselves tend to extrapolate from their own position to a wider population. Having seen hundreds of SleepChart logs, I know that some individuals definitely need 9-10 hours of sleep and feel bad when they do not get it. Could there be an underlying health problem? Perhaps. However, most of these are teenagers that seem otherwise pretty healthy and good students. I have also seen logs of those who need just 4 hours per night and occasionally feel great on just 2-3 hours! Both extremes are in minority. Most of people do well on 6.5-7.5 hours in free running regimen.

My own prejudice

When standing against alarm clocks and short sleep, I must then delineate my own prejudice.

Personal anecdote. Why use anecdotes?
I have used alarm clocks pretty sparingly in my life. In 1999, I decided to get rid of alarm clocks altogether. I got absolutely in love with my uninterrupted sleep and want to share the fun with everyone, esp. with the young generation. Against the claims of Dr Horne, despite running my sleep free, I am rather a short sleeper. As I write these words my trailing nighttime sleep length is 5.7 hours, which is well below the population average. It used to be more before I started strictly running my sleep free over a decade ago. It used to be far more before I started sleeping biphasically in 1994 (see: the impact on napping on nighttime sleep and total sleep time). I have never noticed a tendency to sleep long just because of conducive circumstances (long nights, cold weather, rainy weather, etc.). Just the opposite, the more religiously I adhere to the rules of good sleep, the shorter my sleep is and the better the quality of my learning (as measured with SuperMemo). Even though this observation is definitely my natural and unavoidable cognitive anchor, the same correlations I noticed in dozens of SleepChart submissions.

My love of free sleep does not imply that I do not know how serious sleep deprivation feels. I recall a time at SuperMemo World, in the early 1990s, when working as a programmer against a deadline, I had to go for some 72 hours without sleep. I did not feel brain dead because of the stress and excitement of the job. However, the outcome was near-to disastrous. A CD-R with the finished product was on its way to Germany for mastering and production when a disastrous bug was spotted in testing. That sleep deprivation and the deadline could have been very costly.

Had I not tried free running sleep, I might have been pretty skeptical of Dr Stickgold hyperbole: "sleep deprivation makes you fat, sick and stupid". In the 1980s, when I was in a constant battle for quality sleep, pulling a dozen of all-nighters annually, and nearly always late to bed, I was actually pretty healthy, very skinny (mostly as a result of coming from a poor household), and I ended up graduating with honors. Neither fat, nor sick, nor certifiably stupid. However, my views changed drastically now that I have tried free running sleep for over a decade. I am now much healthier than 20 years ago, and I learn much faster (even though most of that acceleration is due to technology and experience). I might still be in a daily battle to maintain a healthy level of body fat, however, that battle was by far hardest in the early days of SuperMemo World, in the early 1990s, when my sleep hygiene was at its lowest. I believe strongly that free running sleep improved my health and creativity. I believe that it was free running sleep that helped me eliminate the problem of colds and influenza, even though winter swimming might also have been a strong contributor. Most of all, I love to have nearly forgotten how sleep deprivation feels. Nothing undermines creative work as effectively as a bad night sleep.

When looking at my own sleep length over the last two decades someone might see signs of aging:

Night sleep over 18 years collected with SleepChart

Figure: Average length of night time sleep decreasing in the course of two decades (Wozniak 2000-2017)

I rather claim improved sleep quality and improved total compensated sleep value (i.e. when nighttime sleep is added to napping time multiplied by 3). See: Power nap

Value of good sleep

In the light of my own experience, let me then take liberties and reword/soften Dr Stickgold's claim:

Sleep deprivation will make you fatter, sicker and dumber!

Robert Stickgold, PhD, Associate Professor of Psychiatry, Harvard Medical School (paraphrased)

I admit! I am severely prejudiced! Having tried good sleep, I cannot possibly think of sleep restriction or artificial sleep control. You do not need to trust my judgement though. If you are not sure, apply Pascal's Wager and treat sleep like God. Your intellectual strength is at stake!

Good sleep is a key to the treasure of good life. Don't let anyone rob you!

Effects of sleep duration and sleep phase on learning

I started my investigations of the impact of sleep on learning in the early 2000 from the simple intuition that short-night sleep is bad for learning on the next day. After collecting two years of data with SleepChart, I tried to show the link between the length of sleep and the quality of learning. However, that attempt was not successful. My problem was that I used my own sleep and learning data. I am a religious adherent of free running sleep (i.e. sleep where all forms of sleep control, esp. the alarm clock, are forbidden). Upon closer inspection, it appears that in free running sleep, short night sleep is often an indicator of hitting the optimum sleep phase, while long sleep may result from going to sleep too early, heavy exercise, ill health, and other factors. The interpretation is analogous to Dr Kripke's research showing that people who sleep less live longer. That research led many to a wrong conclusion that keeping one's sleep short is healthy. In Kripke's and my own investigations, the confusion comes from the fact that it is the naturally short-night sleep that is an indicator of good health, correct sleep phase, or good prospects for long life. Using other people's data, I could later show that short sleep caused by the use of alarm clock has a negative impact on learning. One can expect the same effect of alarms on longevity. It is now obvious that the length of sleep cannot be used as an indicator of sleep quality in free running sleep.

In free running sleep there is little or no correlation between the total sleep time and the learning performance. This correlation emerges only when the length of sleep episodes is controlled artificially.

The relationship of bedtime and the sleep phase is more important than the total amount of sleep. We should always sleep at the time when the body clock says it is the beginning of the subjective night. There is more benefit in 2-3 hours of sleep at the right time (subjective night), than in 8 hours of sleep at a wrong time (e.g. when jetlagged in Japan). Obviously, it may be pretty hard to get 8 hours during the subjective day without a serious prior sleep deprivation.

While doing my preliminary investigations with my own sleep data, I concluded that I could use sleep phase as a much better indicator of sleep quality. SleepChart makes it possible to analyze the data and approximate the optimum time of bedtime. Those predictions are very rudimentary and can be explained by the following reasoning: if you went to sleep at 5 am yesterday, and the lateness was natural, not forced, do not hope that you can fall into quality sleep at 2 am today. In free running sleep, tiredness is always the ultimate judge that tells you when to go to sleep. However, SleepChart can warn you when the tiredness is likely to be homeostatic and an "unreliable predictor" of the optimum sleep time. If you got to sleep too early, your sleep will be excessively long, not fully refreshing, and carrying a risk of premature awakening. If you go to sleep too late, your sleep will be unnaturally short and carry a risk of shifting the sleep phase (i.e. going to sleep even later on the next day). Those observations provide a solid suspicion that the sleep phase could affect the quality of sleep and the quality of learning on the next day. The sleep phase here is the difference between the optimum bedtime (e.g. as predicted by SleepChart) and the actual bedtime. However, when trying to correlate the sleep phase with the quality of learning, I was to be disappointed again. I could not find a correlation between the sleep phase and the quality of learning (e.g. as expressed by grades in SuperMemo). There were two major weaknesses in that preliminary effort:

  • SuperMemo uses the concept of midnight shift that allows of registering repetitions executed after midnight with the previous learning day. In other words, in older SuperMemos it was not possible to say if the repetition executed on May 11 took place on May 11 or in the early hours of May 12
  • SleepChart used to use simple statistics in predicating on the optimum sleep phase. It did not use a two-process model of sleep regulation. Just plain old averages. It did not even use the phase response curve to register block shifts that can affect the sleep phase. In other words, SleepChart was strong only when free running sleep was not disturbed by factors such as forced delays, exercise, health problems, sunlight, etc.

Both SuperMemo and SleepChart have been vastly improved since. SuperMemo registers the timing of each repetition, while SleepChart relies on a phase response to predict the circadian acrophase. With improved data gathering, I was able to have a preliminary peek at the relationship between the bedtime phase and the learning performance:

Relationship between the bedtime phase and the learning performance.

Relationship between the bedtime phase and the learning performance. The bedtime phase is defined as the difference between the actual and the optimum bedtime. Learning performance is measured by the average grade obtained while learning with SuperMemo.

As expected, delaying sleep resulted in a gradual decrease in performance. There is far less data on the "advance" side due to the fact that in free running sleep, early bedtime hardly ever results in early sleep, and is more likely to simply entail some unproductive wake time in bed.

Sleeping in the wrong phase (i.e. too early or too late), will result in a degraded learning performance.

SleepChart has become an integral part of SuperMemo as of SuperMemo 14.0 (2008). Some of the findings based on the data collected with those two applications are listed in later sections of the present article.

Sleep block length distribution

NREM-REM sleep cycles take roughly 90 min. A popular myth says that the length of a healthy night-time sleep episode will therefore always be a multiple of 90 min. Another myth says that it is ok to interrupt sleep after a multiple of 90 min. A variant of both myths says that sleep is supposed to last a multiple of a period that is specific to a given individual.

SleepChart displays the distribution of the length of all sleep episodes. That distribution can be used to invalidate the claim that sleep blocks cluster in multiples of 90 min. In the presented example, sleep block length distribution in a monophasic sleeper indeed shows clusters at: 1, 2, 3, 3.5, 4, 5, 6, 7, 8, 9, and 10 hours. However, upon closer scrutiny, this clustering comes only from inaccurate logging by the subject (it is easier to mark 3.0 hour block than 2.95 hour block). There is no 90 min. trend discernible, however, one might be tempted to notice a multiple of 60 minutes.

SleepChart: The sleep block length distribution of a monophasic sleeper

Here is then yet another example that uses a semi-log scale, which is better for visualizing short sleep blocks of a habitual napper. In this case, a biphasic sleeper shows only one significant cluster at 7 hours of sleep. This cluster was again caused by imprecise logging.

SleepChart: The sleep block length distribution of a biphasic sleeper (semi-log scale)

Finally, a sleep block distribution of a regular 7-hours-per-night monophasic sleeper. There are a few peaks discernible, however, no regular sleep length multiple. In particular, no peaks around the expected 5.5 and 8.5 hours.

A sleep block distribution of a regular 7-hours-per-night monophasic sleeper. There are a few peaks discernible, however, no regular sleep length multiple. In particular, no peaks around the expected 5.5 and 8.5 hours.


  1. SuperMemo World, "SuperMemo User Survey" (November 1994)
  2. Gorzelanczyk E.J., Wozniak P.A., Chrzastowska H., "Investigating the correlation between the intelligence and the performance in repetitive learning tasks," 3rd Congress of the European Federation of Neurological Societies, Sevilla (1998)
  3. Cappuccio F.P., MD, D’Elia L., MD, Strazzullo P., MD, Miller M.A., PhD, "Sleep Duration and All-Cause Mortality: A Systematic Review and Meta-Analysis of Prospective Studies," Sleep / Volume 33 / Issue 4 (2010): 585-592
  4. Cappuccio F.P., MD, D’Elia L., MD, Strazzullo P., MD, Miller M.A., PhD, "Quantity and Quality of Sleep and Incidence of Type 2 Diabetes: A systematic review and meta-analysis," Diabetes Care / Volume 33 / Issue 2 (February 2010): 414-420, doi: 10.2337/dc09-1124
  5. Oren D.A., M.D., Wehr T.A., M.D., "Hypernyctohemeral Syndrome after Chronotherapy for Delayed Sleep Phase Syndrome," The New England Journal of Medicine / Volume 327 / Issue 24 (December 10, 1992): 1762
  6. Dinges D.F., Rogers N.L., and Baynard M.D., "Chronic Sleep Deprivation." In "Principles and Practice of Sleep Medicine," edited by Kryger M.H., MD, FRCPC, Roth T., PhD, and Dement W.C., MD, PhD (St. Louis: Saunders, 2000)