This doesn’t undercut the rest of the post, but I do think “Occasionally stubbing your toe is good for health and promotes more efficient toe healing” is unfair. Lots of people believe that acute stress in the form of exercise, infection, fasting, or extreme temperatures is good for them, and sleep deprivation seems a lot more like those than like stubbing a toe. Of course those beliefs could be wrong, or sleep could not be a member of the set, but if the hypothesis were incorrect I think it would be a normal kind of incorrect, not something as ridiculous as “breaking a bone is good for you”.
Lots of people believe that acute stress in the form of exercise, infection, fasting, or extreme temperatures is good for them, and sleep deprivation seems a lot more like those than like stubbing a toe.
I agree with Elizabeth that it intuitively feels like SD fits more with things like exercise and infection than things like toe-stubbing. I did not have a ready-made answer before you asked the important question “why?,” and I think that means you should discount this somewhat as rationalization. With that caveat, here goes!
First, let’s look at some of the underlying physiological impacts of these potentially “good stressors.”
Exercise: tear and repair of muscle fibers results in altered morphology
Infection: repeated exposure results in death of dysfunctionally reactive immune cells
Fasting: hormone changes in response to change in metabolism (I’m handwaving to the max!)
Extreme temperatures: one impact is that elevated body temperature causes the release of heat shock proteins, which directly impact immune system function, such as by increasing MHC class I loading (this is an important role of fever in fighting off infections). It also affects the material properties of proteins, which could impact things like flexibility but also enzyme reaction rates (though not necessarily in a straightforward way, since the body attempts to create homeostasis and there are all kinds of feedback loops).
Let’s start not with a stubbed toe, but with breaking a bone. What does this lack that excludes it from the potentially “good stressor” category, or what does it include that puts it in the “bad stressor” category?
Clearly, bone breaking is incredibly, durably painful. It potentially permanently alters the physical shape of bones in a way that can easily destroy their ability to naturally regain their original shape. Since the body depends on the skeleton to operate muscles effectively, strength, flexibility, and dexterity may be impacted as well. Furthermore, bone breakage may cause large-scale tearing of soft tissue and generate a massive infection site, potentially concomitant with tearing of blood vessels and solid organ puncture. These large-scale physical impacts are what put bone breaking into the “bad stressor” category, and I think it’s safe to say nobody in their right mind would hope to find some countervailing positive effect that’s worth this massive cost.
As we scale down the magnitude and duration of pain, the extent of physical tissue destruction, the destruction of the skin’s barrier against infection, we overall scale down many of the things that put an acute stressor into the “bad” category, potentially allowing any benefits it may have to become relevant.
Since sleep deprivation entails no tissue destruction, except insofar as it impacts synaptogenesis or axonal pruning (a field of study where our knowledge is quite patchy!), we can’t start from a physical assumption that it’s obviously bad.
Now, let’s compare with the stubbed toe, and also with exercise.
Exercise, unlike a stubbed toe, is a natural consequence of human activity that’s necessary for life. As such, if there are inherently damaging aspects of exercise, we can expect that evolution has had an enormous amount of time to work on mitigating them. Biomechanical cues impacting development are all over the place in the body. Because it’s metabolically expensive to maintain unnecessary girth, we can interpret the body’s ability to respond to cues saying “grow more and bigger cells here, we’re obviously using them” as a helpful regulatory feature, one that might be evolutionarily advantageous.
By contrast, we also expect nature to put highly aversive pain signals in place where they’re most useful, since pain comes with costs. A stubbed toe isn’t a signal that we’re “using that toe,” but a signal to adjust how we’re walking and of potential injury. There’s again a straightforward biomechanical interpretation to be found.
Although sleep pressure is often aversive, this aversion comes from an entirely different physical system, and it’s not clear that it means the same thing. In addition, I can control my sleep aversion to an extent that’s impossible without drugs when it comes to pain. The controllability, modest extent, origin from a source other than tissue trauma, and context as part of the normal pattern of behavioral variation in organisms on a day-to-day basis set sleep deprivation apart from a stubbed toe and in the category of exercise, exposure to pathogens and chemicals, fasting, and temperature variation.
None of this necessarily means that sleep deprivation (or any of these other “relatively normal, non-injurious” interventions) is good for you.
If they were, though, we might offer a unified meta-reason. In general, all these sources of variation may be ways to supply useful regulatory cues to the body.
Now, let’s zoom in on sleep. Even when we look specifically at sleep research, as opposed to work-related fatigue, we have a confounder that I suspect may not be adequately addressed.
This is light. Where I am, it’s 1:20 AM, and I have the slights on blazing bright while staring into my computer screen. In addition to my wakefulness, my body us receiving a cue about when light is available. Returning to the evolutionary biology lens, this would be ancestrally correlated with my ability to hunt, my concerns about predators, social activity, the season, and so on. Potentially, more sleep goes with more darkness goes with less abundant food and social opportunity goes with hibernation. More wakefulness would be ancestrally associated with more abundance and opportunity and fewer predation threats. This is aside from the fact that we can actually do more while awake, which may have additional benefits if we use our time wisely.
So this is a candidate for a reason why sleep deprivation could have some benefits. It could either be entirely due to light exposure during additional hours of wakefulness, or to wake and sleep being used as proxies for opportunities while awake. The idea that light cues and circadian rhythms are important controllers of our behavior and experience are already well-established (light therapy is a depression treatment, as is sleep deprivation).
A second possibility is that sleep deprivation could, on its own, diminish bad forms of mental and emotional stress. Hear me out! Usually, when people are sleep deprived, it’s because they’re trying to cope with some very stressful circumstances—a new baby, a demanding job, an upcoming exam, jet lag, a war or other stressful environment. So it is associated with high stress, and often perceived to be at least a contributing cause of it.
However, it’s easy to picture how sleeping for fewer hours, in these situations, is often a way to reduce and manage that stress. You are able to take care of activities that need to be done with those additional waking hours, thereby removing sources of worry and discomfort that would otherwise bother you while you’re trying to sleep.
So imagine if regular folks, without any extreme stressors, chose to take a couple extra hours awake to take care of random lingering sources of discomfort or stress. A couple extra hours per day could easily be the difference between “I have only just enough time to get everything done” and “I have plenty of time to relax and make improvements to the conditions of my life.” In this case, we’d see sleep deprivation reducing people’s stress over time.
Note that neither of these hypotheses are out of the bounds of our familiar experience. Plenty of people have said “oh, if only I had an extra hour or two in my day.” We already use both sleep deprivation and light therapy as depression treatments, and they fit naturally into the category of environmental and behavioral cues regulating physiological processes.
In fact, the real mystery is why sleep deprivation should be bad for us, not why it should be neutral-to-good for us. Not all animals sleep, and amount of sleep isn’t really correlated with cognitive capacity as far as I remember (I’m being lazy and not looking up the relevant study, sorry!). We also have a great alternative explanation for why animals sleep: to reduce metabolic demands during times of high threat and low light, as in hibernation.
One physiological reason to think sleep might in fact be important is that the brain, like other organs, may need to physically alter its morphology in order to accomplish maintenance tasks. These might include learning-central tasks like synaptogenesis, but it also might just entail things like clearance of fluids. I think it’s fairly well-established that this does happen during sleep. In addition, I understand that it’s not just the physical connectivity of synapses, but also the overall flow of electrical potential throughout the brain as a whole, that governs thought. So it’s plausible that sleep is necessary to perform essential maintenance tasks that degrade the physical ability of the brain to function properly if they aren’t done. Alternatively, it might be that these maintenance tasks will intrude into the waking hours, disrupting cognition in an unpleasant fashion, if they don’t happen during sleep.
A better answer: in the course of a week, you will naturally spend time eating and not eating, moving and not moving, and sleeping and not sleeping. Calling something “fasting” (as opposed to “not eating right this second”), “excess exercise” (as opposed to “moving”), or “sleep deprivation” (as opposed to “being awake”) is begging the question. It assumes you know the correct balance and are describing a departure from that balance. Meanwhile, stubbing a toe is always a worse state than not having stubbed a toe. The correct amount of toe-stubbing is zero, we can assume any deviation is bad (but then argue about what is slow-motion assault and what is deeply useful trigger point massage therapy).
[The immune system is not a member of the “all things in balance” class. Having zero pathogens and being adapted to an environment with zero pathogens is obviously superior, but the trade-offs are made weird by being an adversarial game with lock-in]
My model is that the finding balance between eating/not-eating, moving/not-moving, and sleeping/not-sleeping, is not always trivial. Your body sends imperfect signals, the right answer varies based on the environment, which changes a lot, and you’re making trade-offs you might not endorse if you had more info. Mild-to-moderate departures from your current balance can be informative in helping you find the correct balance and learn the trade-offs you’re making, even if the experiment ends up being net negative in the short term
You might argue that Alexey specifically used the words “sleep deprivation”. I think that’s a poor phrasing resulting from a bad model being embedded in the language, but if he comes back and says “no, I definitely meant deprivation-as-in-bad-for-you”, it won’t change my belief in this model.
A better answer: in the course of a week, you will naturally spend time eating and not eating, moving and not moving, and sleeping and not sleeping. Calling something “fasting” (as opposed to “not eating right this second”), “excess exercise” (as opposed to “moving”), or “sleep deprivation” (as opposed to “being awake”) is begging the question. It assumes you know the correct balance and are describing a departure from that balance. Meanwhile, stubbing a toe is always a worse state than not having stubbed a toe.
There is something that is similar to eating, moving and sleeping in that respect: exhaling. Over the course of a minute, you will naturally spend time exhaling and not exhaling. The correct amount of time to spend not exhaling is not zero. And failing to exhale acutely stresses your body. But it doesn’t seem to follow from those things that holding your breath once in a while is beneficial.
In exercise, breathing, eating, and sleeping, we can perhaps roughly model the system as being driven by an oscillatory forcing function, with some amplitude and frequency, such as of the form Acos(w*t). The point is that although instinct, intuition, and imitation allow us to find parameters compatible with life/cognitive function/acceptable performance, we don’t have the data to determine whether or not these parameters are optimal, achievable, or sustainable, nor under what conditions this is or is not true.
We should carefully reserve the word “optimal” and “optimize” for situations where we have some sort of proof or evidence relative to the claim. It’s preferable to use words like “target” or “conventional” or “standard” when we’re discussing parameters that seem likely to work, or that have been shown to work, but that haven’t been clearly shown to be the best for our use case.
We know that a standard sleep length of 8 hours per day for adults is compatible with cognitive function, life, and health, but the argument we are all having here is over whether the evidence is sufficient to say that this is robustly optimal.
My point here is just that the argument “X-to-a-greater-than-usual-extent is an acute stress, and therefore it is good for health” (which was Guzey’s argument) is bad even if X is something we naturally do to a nonzero extent.
I’m probably about as uncertain as you are about whether sleeping 8 hours per night is optimal. (Though my comment was indeed unclear about that before I edited it, so thanks for pointing that out).
I think we should understand Guzey’s argument as a gesture in the direction of saying “we should try to establish the base rate for these sorts of stressors [i.e. purposeful behaviors with a normal range of variation in frequency and intensity]” He’s pointing out a few salient behaviors in this reference class (diet and exercise), pointing out that what we might call “sensible variation” is believed to be good for you, and then noting that sleep also belongs in this reference class. You’ve also noted that breathing belongs in this reference class, and we can also find extensive traditions of breath work that are at least meant to promote health.
I agree with others here that toe-stubbing and other injury does not belong in this reference class, because they aren’t purposeful behaviors.
A stronger counterargument is that we may be suffering from availability bias in identifying members of this reference class. We receive lots of messages to breath deep, get exercise, and eat less. What about sensory stimulation vs. lack of stimulation, emotional variation vs. calm, social contact vs. solitude, focus vs. distraction? Do we expect that there’s a “normal range” here that gives you health benefits with “sensible variation?” For example, do we think that people receive health benefits from periodic meditation retreats due to the unusual level of social isolation, calm, and lack of stimulation? Or that they receive health benefits from attending a music festival (sober), due to the unusual level of social contact, excitement, and sensory stimulation?
My uncertainty here is that I don’t know whether we’re restricting this to physical health, and also that I don’t feel confident on what health impacts these additional sorts of reference class members tend to have. intuitively, it seems good for you to get some variety in your life, but on the other hand, will going to the occasional rock concert really improve my heart health (any dancing I might do aside)?
I think that getting some evidence here would be beneficial. However, I also suspect that most of the research is on extremes (i.e. attending too many loud concerts is bad for your hearing), rather than on moving from “moderate” to “sensibly going a little above or below moderate for a while.”
But a few examples of “clearly good for you” in the reference class along with a bunch of examples of “not sure” still puts the valence of “sensible variation” for a member of this reference class as on average positive, though with not much more than 50% confidence.
On the other other hand, the evidence we do have about the impact of sleep on cognitive function is not encouraging. Anyway, the more I think about it, the less clarity I have on how to even think about this question. It doesn’t seem likely to result in a knock-down argument either way, no matter how much we might research it. Too fuzzy.
I think it’s useful to point out that training muscles for strength/size results in a well documented phenomenon called supercompensation. However, training for other qualities like speed doesn’t really work the same way. There’s lots of irrational training done because people make an inferential leap from the supercompensation they see in strength training and apply it to cases which intuitively seem like they might be analogues (e.g., weighted sprints don’t make you faster).
I think counterexamples are relevant because sometimes intuition points out real analogues, and sometimes fake ones, so we should value evidence and mechanistic explanation over analogies and cultural beliefs.
Sorry if this is a little incoherent, I wrote it when I was really sleepy.
I feel like there’s much stronger evidence that those things are beneficial than that acute sleep deprivation is. So it’s not clear to me how sleep deprivation is a lot more like those things than like stubbing your toe really hard.
My impression is that people who are used to sleep deprivation (new parents, military, shift workers) are not as impaired by it as people encountering the same level of deprivation for the first time- even when the former group is running on a prolonged sleep deficit, and the latter is going into sleep deprivation fully rested. You never stop being impaired, but past deprivation lets you cope with it better. This pattern matches to me with e.g. fasting will always make you dumb eventually, but people with fasting experience can go much longer before getting dumb.
I agree that this sort of adaptation probably occurs, but I don’t see how this makes sleep deprivation a lot more like those other things you mentioned than to stubbing your toe. Guzey also claimed that “[o]ccasional acute sleep deprivation is good for health,” not only that it promotes more efficient sleep, and for each of your examples of things that are more like sleep deprivation than toe-stubbing is (fasting, exercise, and infections in some cases) the claim does not seem to be that those things impair you but just less so with time.
(For that matter, I would guess that there is also some adaptation to stubbing your toe. It shouldn’t ever stop hurting or lightly and temporarily damaging you, but if you do it a lot you’ll probably find a way to cope with it better, and go back to doing work five minutes later rather than ten.)
This clearly isn’t fair. For one, the “really hard” modifier is completely made up (did Guzey ever imply that the way to train sleep deprivation resilience is to go “really hard” at not-sleeping rather than easing into it?), and for two, physical stress to ones toes is clearly a much more local thing than caloric or sleep deprivation so the hypothesis would be “kicking things in a controlled fashion strengthens the thing you’re impacting with”.
And I’m not sure if it’s true or not, but it’s definitely a thing that professional fighters do, and it does not at all strike me as “obviously false”.
[Disclaimer: I’m not a doctor and by far I’m not an specialist on health]
First of all, I don’t think the use of “acute” is correct here. Unless I’m misreading the meaning of “acute” in medical jargon [edit: yes, it seems I did. I use acute to mean sever in this comment], acute stress in all these cases is bad for your health and going too far can even cause death. I think it is a (mild to) moderate stress that is good.
It seems to me that the model for “good stresses” are (evolutionarily-) common moderate stresses that produce some minor damage which the body takes advantage of to improve something. The lack of these stresses implies lack of (or lower?) improvement. The same type of stress but of a high enough magnitude starts being detrimental. I think this is what inspired Taleb for the concept of antifragility, where a stress within some limits is advantageous for a system instead of a harmful.
I also think stubbing a toe is not a good counter example (in a relatively short time, it would produce callus that would ultimately protect from the hits). The breaking a bone example seems better. I think the main difference is that breaking a bone actually is an acute stress rather than a moderate one. That’s an stress that is beyond the threshold the body can stand.
This way to model it is not very helpful here, cause we don’t seem to know yet whether short-time lack of sleep brings us any kind of improvement. But it makes 2 things obvious: (1) it seems very plausible that short-time lack of sleep is at least not too bad, as humans may have had occasional all-nighters during our evolution (e.g. to escape predators or other humans); (2) short-time lack of sleep is not an acute stress, so it could fall into the category of good stresses.
Besides acute stresses (obviously bad) and good stresses, there is at least one other relevant category of stresses: mild to moderate stresses the effects of which are cumulative (enough accumulation in a long enough life has bad consequences). I am not sure if this category can be further broken into 2 categories as well, because at least in some instances these stresses make the body somehow resilient to higher doses of the stressor (e.g. heavy drinkers develop a tolerance to the effects of alcohol). Probably there are more categories of stress but I am not aware of them.
Now the issue is still open to which category of stress do short-time and chronic lack of sleep lay in. I’m very interested to know it! My guess is that short-term lack of sleep can very well be beneficial in some way and that the accumulation of lack of sleep is probably bad in the long term.
This doesn’t undercut the rest of the post, but I do think “Occasionally stubbing your toe is good for health and promotes more efficient toe healing” is unfair. Lots of people believe that acute stress in the form of exercise, infection, fasting, or extreme temperatures is good for them, and sleep deprivation seems a lot more like those than like stubbing a toe. Of course those beliefs could be wrong, or sleep could not be a member of the set, but if the hypothesis were incorrect I think it would be a normal kind of incorrect, not something as ridiculous as “breaking a bone is good for you”.
Why?
I agree with Elizabeth that it intuitively feels like SD fits more with things like exercise and infection than things like toe-stubbing. I did not have a ready-made answer before you asked the important question “why?,” and I think that means you should discount this somewhat as rationalization. With that caveat, here goes!
First, let’s look at some of the underlying physiological impacts of these potentially “good stressors.”
Exercise: tear and repair of muscle fibers results in altered morphology
Infection: repeated exposure results in death of dysfunctionally reactive immune cells
Fasting: hormone changes in response to change in metabolism (I’m handwaving to the max!)
Extreme temperatures: one impact is that elevated body temperature causes the release of heat shock proteins, which directly impact immune system function, such as by increasing MHC class I loading (this is an important role of fever in fighting off infections). It also affects the material properties of proteins, which could impact things like flexibility but also enzyme reaction rates (though not necessarily in a straightforward way, since the body attempts to create homeostasis and there are all kinds of feedback loops).
Let’s start not with a stubbed toe, but with breaking a bone. What does this lack that excludes it from the potentially “good stressor” category, or what does it include that puts it in the “bad stressor” category?
Clearly, bone breaking is incredibly, durably painful. It potentially permanently alters the physical shape of bones in a way that can easily destroy their ability to naturally regain their original shape. Since the body depends on the skeleton to operate muscles effectively, strength, flexibility, and dexterity may be impacted as well. Furthermore, bone breakage may cause large-scale tearing of soft tissue and generate a massive infection site, potentially concomitant with tearing of blood vessels and solid organ puncture. These large-scale physical impacts are what put bone breaking into the “bad stressor” category, and I think it’s safe to say nobody in their right mind would hope to find some countervailing positive effect that’s worth this massive cost.
As we scale down the magnitude and duration of pain, the extent of physical tissue destruction, the destruction of the skin’s barrier against infection, we overall scale down many of the things that put an acute stressor into the “bad” category, potentially allowing any benefits it may have to become relevant.
Since sleep deprivation entails no tissue destruction, except insofar as it impacts synaptogenesis or axonal pruning (a field of study where our knowledge is quite patchy!), we can’t start from a physical assumption that it’s obviously bad.
Now, let’s compare with the stubbed toe, and also with exercise.
Exercise, unlike a stubbed toe, is a natural consequence of human activity that’s necessary for life. As such, if there are inherently damaging aspects of exercise, we can expect that evolution has had an enormous amount of time to work on mitigating them. Biomechanical cues impacting development are all over the place in the body. Because it’s metabolically expensive to maintain unnecessary girth, we can interpret the body’s ability to respond to cues saying “grow more and bigger cells here, we’re obviously using them” as a helpful regulatory feature, one that might be evolutionarily advantageous.
By contrast, we also expect nature to put highly aversive pain signals in place where they’re most useful, since pain comes with costs. A stubbed toe isn’t a signal that we’re “using that toe,” but a signal to adjust how we’re walking and of potential injury. There’s again a straightforward biomechanical interpretation to be found.
Although sleep pressure is often aversive, this aversion comes from an entirely different physical system, and it’s not clear that it means the same thing. In addition, I can control my sleep aversion to an extent that’s impossible without drugs when it comes to pain. The controllability, modest extent, origin from a source other than tissue trauma, and context as part of the normal pattern of behavioral variation in organisms on a day-to-day basis set sleep deprivation apart from a stubbed toe and in the category of exercise, exposure to pathogens and chemicals, fasting, and temperature variation.
None of this necessarily means that sleep deprivation (or any of these other “relatively normal, non-injurious” interventions) is good for you.
If they were, though, we might offer a unified meta-reason. In general, all these sources of variation may be ways to supply useful regulatory cues to the body.
Now, let’s zoom in on sleep. Even when we look specifically at sleep research, as opposed to work-related fatigue, we have a confounder that I suspect may not be adequately addressed.
This is light. Where I am, it’s 1:20 AM, and I have the slights on blazing bright while staring into my computer screen. In addition to my wakefulness, my body us receiving a cue about when light is available. Returning to the evolutionary biology lens, this would be ancestrally correlated with my ability to hunt, my concerns about predators, social activity, the season, and so on. Potentially, more sleep goes with more darkness goes with less abundant food and social opportunity goes with hibernation. More wakefulness would be ancestrally associated with more abundance and opportunity and fewer predation threats. This is aside from the fact that we can actually do more while awake, which may have additional benefits if we use our time wisely.
So this is a candidate for a reason why sleep deprivation could have some benefits. It could either be entirely due to light exposure during additional hours of wakefulness, or to wake and sleep being used as proxies for opportunities while awake. The idea that light cues and circadian rhythms are important controllers of our behavior and experience are already well-established (light therapy is a depression treatment, as is sleep deprivation).
A second possibility is that sleep deprivation could, on its own, diminish bad forms of mental and emotional stress. Hear me out! Usually, when people are sleep deprived, it’s because they’re trying to cope with some very stressful circumstances—a new baby, a demanding job, an upcoming exam, jet lag, a war or other stressful environment. So it is associated with high stress, and often perceived to be at least a contributing cause of it.
However, it’s easy to picture how sleeping for fewer hours, in these situations, is often a way to reduce and manage that stress. You are able to take care of activities that need to be done with those additional waking hours, thereby removing sources of worry and discomfort that would otherwise bother you while you’re trying to sleep.
So imagine if regular folks, without any extreme stressors, chose to take a couple extra hours awake to take care of random lingering sources of discomfort or stress. A couple extra hours per day could easily be the difference between “I have only just enough time to get everything done” and “I have plenty of time to relax and make improvements to the conditions of my life.” In this case, we’d see sleep deprivation reducing people’s stress over time.
Note that neither of these hypotheses are out of the bounds of our familiar experience. Plenty of people have said “oh, if only I had an extra hour or two in my day.” We already use both sleep deprivation and light therapy as depression treatments, and they fit naturally into the category of environmental and behavioral cues regulating physiological processes.
In fact, the real mystery is why sleep deprivation should be bad for us, not why it should be neutral-to-good for us. Not all animals sleep, and amount of sleep isn’t really correlated with cognitive capacity as far as I remember (I’m being lazy and not looking up the relevant study, sorry!). We also have a great alternative explanation for why animals sleep: to reduce metabolic demands during times of high threat and low light, as in hibernation.
One physiological reason to think sleep might in fact be important is that the brain, like other organs, may need to physically alter its morphology in order to accomplish maintenance tasks. These might include learning-central tasks like synaptogenesis, but it also might just entail things like clearance of fluids. I think it’s fairly well-established that this does happen during sleep. In addition, I understand that it’s not just the physical connectivity of synapses, but also the overall flow of electrical potential throughout the brain as a whole, that governs thought. So it’s plausible that sleep is necessary to perform essential maintenance tasks that degrade the physical ability of the brain to function properly if they aren’t done. Alternatively, it might be that these maintenance tasks will intrude into the waking hours, disrupting cognition in an unpleasant fashion, if they don’t happen during sleep.
Broadly, they train your body’s homeostasis mechanisms to handle a wider range of conditions, which lets it cope better with aging.
For the immune system in particular, it’s pretty well demonstrated that childhood disease load is setting expectations for future disease load.
Muscles are strengthened via tiny tears that heal. Tears are stress, too large tears are damage, but small tears are the only way to grow the muscle.
Fasting triggers some hormonal changes that maybe include some housecleaning and repair mechanisms, although I know less about this.
A better answer: in the course of a week, you will naturally spend time eating and not eating, moving and not moving, and sleeping and not sleeping. Calling something “fasting” (as opposed to “not eating right this second”), “excess exercise” (as opposed to “moving”), or “sleep deprivation” (as opposed to “being awake”) is begging the question. It assumes you know the correct balance and are describing a departure from that balance. Meanwhile, stubbing a toe is always a worse state than not having stubbed a toe. The correct amount of toe-stubbing is zero, we can assume any deviation is bad (but then argue about what is slow-motion assault and what is deeply useful trigger point massage therapy).
[The immune system is not a member of the “all things in balance” class. Having zero pathogens and being adapted to an environment with zero pathogens is obviously superior, but the trade-offs are made weird by being an adversarial game with lock-in]
My model is that the finding balance between eating/not-eating, moving/not-moving, and sleeping/not-sleeping, is not always trivial. Your body sends imperfect signals, the right answer varies based on the environment, which changes a lot, and you’re making trade-offs you might not endorse if you had more info. Mild-to-moderate departures from your current balance can be informative in helping you find the correct balance and learn the trade-offs you’re making, even if the experiment ends up being net negative in the short term
You might argue that Alexey specifically used the words “sleep deprivation”. I think that’s a poor phrasing resulting from a bad model being embedded in the language, but if he comes back and says “no, I definitely meant deprivation-as-in-bad-for-you”, it won’t change my belief in this model.
There is something that is similar to eating, moving and sleeping in that respect: exhaling. Over the course of a minute, you will naturally spend time exhaling and not exhaling. The correct amount of time to spend not exhaling is not zero. And failing to exhale acutely stresses your body. But it doesn’t seem to follow from those things that holding your breath once in a while is beneficial.
In exercise, breathing, eating, and sleeping, we can perhaps roughly model the system as being driven by an oscillatory forcing function, with some amplitude and frequency, such as of the form Acos(w*t). The point is that although instinct, intuition, and imitation allow us to find parameters compatible with life/cognitive function/acceptable performance, we don’t have the data to determine whether or not these parameters are optimal, achievable, or sustainable, nor under what conditions this is or is not true.
We should carefully reserve the word “optimal” and “optimize” for situations where we have some sort of proof or evidence relative to the claim. It’s preferable to use words like “target” or “conventional” or “standard” when we’re discussing parameters that seem likely to work, or that have been shown to work, but that haven’t been clearly shown to be the best for our use case.
We know that a standard sleep length of 8 hours per day for adults is compatible with cognitive function, life, and health, but the argument we are all having here is over whether the evidence is sufficient to say that this is robustly optimal.
My point here is just that the argument “X-to-a-greater-than-usual-extent is an acute stress, and therefore it is good for health” (which was Guzey’s argument) is bad even if X is something we naturally do to a nonzero extent.
I’m probably about as uncertain as you are about whether sleeping 8 hours per night is optimal. (Though my comment was indeed unclear about that before I edited it, so thanks for pointing that out).
I think we should understand Guzey’s argument as a gesture in the direction of saying “we should try to establish the base rate for these sorts of stressors [i.e. purposeful behaviors with a normal range of variation in frequency and intensity]” He’s pointing out a few salient behaviors in this reference class (diet and exercise), pointing out that what we might call “sensible variation” is believed to be good for you, and then noting that sleep also belongs in this reference class. You’ve also noted that breathing belongs in this reference class, and we can also find extensive traditions of breath work that are at least meant to promote health.
I agree with others here that toe-stubbing and other injury does not belong in this reference class, because they aren’t purposeful behaviors.
A stronger counterargument is that we may be suffering from availability bias in identifying members of this reference class. We receive lots of messages to breath deep, get exercise, and eat less. What about sensory stimulation vs. lack of stimulation, emotional variation vs. calm, social contact vs. solitude, focus vs. distraction? Do we expect that there’s a “normal range” here that gives you health benefits with “sensible variation?” For example, do we think that people receive health benefits from periodic meditation retreats due to the unusual level of social isolation, calm, and lack of stimulation? Or that they receive health benefits from attending a music festival (sober), due to the unusual level of social contact, excitement, and sensory stimulation?
My uncertainty here is that I don’t know whether we’re restricting this to physical health, and also that I don’t feel confident on what health impacts these additional sorts of reference class members tend to have. intuitively, it seems good for you to get some variety in your life, but on the other hand, will going to the occasional rock concert really improve my heart health (any dancing I might do aside)?
I think that getting some evidence here would be beneficial. However, I also suspect that most of the research is on extremes (i.e. attending too many loud concerts is bad for your hearing), rather than on moving from “moderate” to “sensibly going a little above or below moderate for a while.”
But a few examples of “clearly good for you” in the reference class along with a bunch of examples of “not sure” still puts the valence of “sensible variation” for a member of this reference class as on average positive, though with not much more than 50% confidence.
On the other other hand, the evidence we do have about the impact of sleep on cognitive function is not encouraging. Anyway, the more I think about it, the less clarity I have on how to even think about this question. It doesn’t seem likely to result in a knock-down argument either way, no matter how much we might research it. Too fuzzy.
I think it’s useful to point out that training muscles for strength/size results in a well documented phenomenon called supercompensation. However, training for other qualities like speed doesn’t really work the same way. There’s lots of irrational training done because people make an inferential leap from the supercompensation they see in strength training and apply it to cases which intuitively seem like they might be analogues (e.g., weighted sprints don’t make you faster).
I think counterexamples are relevant because sometimes intuition points out real analogues, and sometimes fake ones, so we should value evidence and mechanistic explanation over analogies and cultural beliefs.
Sorry if this is a little incoherent, I wrote it when I was really sleepy.
I feel like there’s much stronger evidence that those things are beneficial than that acute sleep deprivation is. So it’s not clear to me how sleep deprivation is a lot more like those things than like stubbing your toe really hard.
My impression is that people who are used to sleep deprivation (new parents, military, shift workers) are not as impaired by it as people encountering the same level of deprivation for the first time- even when the former group is running on a prolonged sleep deficit, and the latter is going into sleep deprivation fully rested. You never stop being impaired, but past deprivation lets you cope with it better. This pattern matches to me with e.g. fasting will always make you dumb eventually, but people with fasting experience can go much longer before getting dumb.
I agree that this sort of adaptation probably occurs, but I don’t see how this makes sleep deprivation a lot more like those other things you mentioned than to stubbing your toe. Guzey also claimed that “[o]ccasional acute sleep deprivation is good for health,” not only that it promotes more efficient sleep, and for each of your examples of things that are more like sleep deprivation than toe-stubbing is (fasting, exercise, and infections in some cases) the claim does not seem to be that those things impair you but just less so with time.
(For that matter, I would guess that there is also some adaptation to stubbing your toe. It shouldn’t ever stop hurting or lightly and temporarily damaging you, but if you do it a lot you’ll probably find a way to cope with it better, and go back to doing work five minutes later rather than ten.)
you’re right, it’s not quite analogous. I still low-confidence believe in the homeostasis disruption thing, but this isn’t evidence of that.
This clearly isn’t fair. For one, the “really hard” modifier is completely made up (did Guzey ever imply that the way to train sleep deprivation resilience is to go “really hard” at not-sleeping rather than easing into it?), and for two, physical stress to ones toes is clearly a much more local thing than caloric or sleep deprivation so the hypothesis would be “kicking things in a controlled fashion strengthens the thing you’re impacting with”.
And I’m not sure if it’s true or not, but it’s definitely a thing that professional fighters do, and it does not at all strike me as “obviously false”.
[Disclaimer: I’m not a doctor and by far I’m not an specialist on health]
First of all, I don’t think the use of “acute” is correct here. Unless I’m misreading the meaning of “acute” in medical jargon [edit: yes, it seems I did. I use acute to mean sever in this comment], acute stress in all these cases is bad for your health and going too far can even cause death. I think it is a (mild to) moderate stress that is good.
It seems to me that the model for “good stresses” are (evolutionarily-) common moderate stresses that produce some minor damage which the body takes advantage of to improve something. The lack of these stresses implies lack of (or lower?) improvement. The same type of stress but of a high enough magnitude starts being detrimental. I think this is what inspired Taleb for the concept of antifragility, where a stress within some limits is advantageous for a system instead of a harmful.
I also think stubbing a toe is not a good counter example (in a relatively short time, it would produce callus that would ultimately protect from the hits). The breaking a bone example seems better. I think the main difference is that breaking a bone actually is an acute stress rather than a moderate one. That’s an stress that is beyond the threshold the body can stand.
This way to model it is not very helpful here, cause we don’t seem to know yet whether short-time lack of sleep brings us any kind of improvement. But it makes 2 things obvious: (1) it seems very plausible that short-time lack of sleep is at least not too bad, as humans may have had occasional all-nighters during our evolution (e.g. to escape predators or other humans); (2) short-time lack of sleep is not an acute stress, so it could fall into the category of good stresses.
Besides acute stresses (obviously bad) and good stresses, there is at least one other relevant category of stresses: mild to moderate stresses the effects of which are cumulative (enough accumulation in a long enough life has bad consequences). I am not sure if this category can be further broken into 2 categories as well, because at least in some instances these stresses make the body somehow resilient to higher doses of the stressor (e.g. heavy drinkers develop a tolerance to the effects of alcohol). Probably there are more categories of stress but I am not aware of them.
Now the issue is still open to which category of stress do short-time and chronic lack of sleep lay in. I’m very interested to know it! My guess is that short-term lack of sleep can very well be beneficial in some way and that the accumulation of lack of sleep is probably bad in the long term.
I’m using acute to mean “of short duration”, independent of severity, which I think is not how you are using it.
Ah, ok sorry. Yes, I thought it means sever.