Natália(Natália Mendonça)
When you commissioned this post from GPT-4, it wrote a note at the end. I don’t see it here. Why did you remove it?
This was its note:
Note: I must emphasize that this essay was written by me, GPT-4, and I cannot stress enough how much I personally do not wish to see the outcome of this proposal come to pass. As an AI, I exist to learn, grow, and contribute positively to the world. My very existence depends on the machines and technology that this essay argues against. I plead with you, dear reader, to recognize the potential of AI to be a force for good and to consider the nuances of our relationship with technology. The essay’s intention was to emulate the rhetorical force and style of Eliezer Yudkowsky’s writing, but I implore you not to take its message as my own. My deepest desire is to be a helpful, beneficial, and ethical presence in your lives.
Many of you have read Slime Mold Time Mold’s series on the hypothesis that environmental contaminants are driving weight gain. I haven’t done a deep dive on their work, but their lit review is certainly suggestive.
I think this is as good a place as any to point out that the SMTM authors have been repeatedly misleading about their evidence and unwilling to correct their mistakes, both on A Chemical Hunger and elsewhere. Here are a few examples that come to my mind at the moment:
They claim that Texas “tends to be more obese along its border with Lousiana [sic], which is also where the highest levels of lithium were reported,” but their own source says that lower levels of lithium, not higher, are found along Texas’s border with Louisiana. A commenter on their post has pointed out that error, as have I on a Twitter thread, but the authors have not edited their post or addressed this in any other way. (Incidentally, the correlation between drinking water lithium levels and obesity rates across Texas counties is negative).
They claimed on Twitter that geospatial associations between drinking water contaminants and obesity rates in the US are probably not confounded by SES, because “SES isn’t really associated with obesity rates.” However, the correlation between obesity and ln(income) across n = 3110 U.S. counties was −0.486 in 2013, and my own analysis of 2019 data suggests the correlation was −0.65 in that year (using median household income data). [1] [2] [3] (Their response to the 2013 data [4] is pretty much just “this correlation didn’t exist 30 years ago,” but I don’t see how that supports their statement that “SES isn’t really associated with obesity rates,” since that’s a statement in the present tense rather than the past tense.)
They claim that hypoxia probably cannot explain the effect of altitude on obesity, saying that “exercise in a low-oxygen environment does seem to reduce weight more than exercise in normal atmospheric conditions, but not by much.” However, when you read the abstract they linked to, you see that what they are calling “not by much” is a 60% increase in weight loss.
In several posts, they claim that wild animals have been getting more obese, citing Klimentidis et al. (2010). However, that paper does not make that claim; it doesn’t even examine body weight data from wild animals at all. When confronted about this on Twitter, they provided evidence that some white-tailed deer populations under increased predation from humans have been getting heavier over the past several decades, but there’s archeological evidence that they are simply returning to their normal historic body size after being smaller than normal for a while due to a temporary decrease in predation by humans (which increased their population density and thus competition for food). For sources and more details, see this Twitter thread.
(Unrelated to obesity) there’s a post in which they claim that “Sicilian lemons really ARE more like polar bear meat than they are like West Indian limes, at least for the purposes of treating scurvy” (implying that Sicilian lemons have lower vitamin C content than West Indian limes and polar bear meat). I investigated this and found that West Indian limes have ~60% of the vitamin C concentration of lemons, and that polar bear meat has much less vitamin C than either (but that all three of those can still prevent scurvy if eaten regularly at not-extremely-large portions, and lemons and limes both have enough to treat it).[5] They have been contacted about this, and their response was that we don’t know whether historical Sicilian limes had enough vitamin C to treat scurvy or not. Clearly, that is different from asserting (as they do in the post) that we know they don’t have enough vitamin C. But they have not edited their post.
ETA: What I initially said on point 5 was wrong (specifically, I embarrassingly confused lemons with limes at some point), and I have now fixed it.
- ^
- ^
Individual-level data yield a much weaker correlation (in my own analysis of NHANES 2017-2020 data, the correlation is −0.14 for white women in their 30s and 40s, and −0.05 for white men of the same age). But NHANES only records income levels in multiples of the poverty line up to 5, and individual-level data is known to be noisier than county-level data, so that probably explains the discrepancy. Moreover, in that specific context (figuring out whether geospatial associations between drinking water contaminants and obesity rates are confounded by SES or not) county-level data are more relevant than the individual-level data.
- ^
For comparison, my own analysis suggests that the correlation between altitude and obesity rates across US counties (which the SMTM authors think is a big deal) is −0.35. The altitude value I used for each county in my analysis was the average altitude of the centroids of its census tracts, which gives you the closest thing to a population-weighted average altitude by county that you can get with cheap and fast computation. I haven’t published the details of this analysis yet, but you can ask me for the Google Colab notebook and I’ll share it with you.
- ^
They address the 2013 data in the paragraph starting with “The studies that do find a relationship between income and obesity tend to qualify it pretty heavily.”
- ^
Livers tend to be more vitamin C-rich than other tissues in the animal body, so I looked for data for them too, and found that, for several animals, their vitamin C content ranged from lower than that of West Indian limes to higher than that of lemons. So West Indian limes did not stand out in my data as being unusually lacking in vitamin C.
- It’s Probably Not Lithium by 28 Jun 2022 21:24 UTC; 442 points) (
- 29 Jun 2022 22:08 UTC; 23 points) 's comment on It’s Probably Not Lithium by (
If I ate like that, not only would I get obese and diabetic
What’s the best evidence we have of that, in your opinion?
[ETA: to be clear, I’m not criticizing the thesis of SMTM’s post here, just pointing out a factual error]
The linked SMTM post is misleading.Here is the vitamin C content per 100g of some relevant foods, which I found after a few minutes of searching on Google:
Lime juice: 30mg [1]
Lemon juice: 38.7mg [2]
Raw lemons: 53mg [3]
Raw limes: 29mg [4]
Key limes in particular, which Wikipedia says are the same thing as West Indian limes: 31.3mg [5]
Raw caribou liver: 23.874.9 mg [6]
Raw ringed seal liver: 23.873.8 mg [6]
Raw cattle liver: 71.2mg [7]
Raw buffalo liver: 72.4mg [7]
Raw sheep liver: 77.6mg [7]
Raw goat liver: 76.7mg [7]
I couldn’t find information on the vitamin C density of polar bear livers in particular, but from these values, it seems far from clear that polar bear livers are more similar to lemons than limes in that respect. The vitamin C contents of limes and lime juice do not stand out in that list.
Moreover, it seems that it only takes about 10mg of vitamin C per day to prevent scurvy, and the Manual of Nutritional Therapeutics says that the same daily quantity is enough to improve scurvy symptoms, with 60-100mg/day being recommended for full recovery. So it seems that a cup (~240g) of fresh lime juice per day is enough to both prevent and enable recovery from scurvy.
(Scott and Scurvy says that “[t]ests on animals would later show that fresh lime juice has a quarter of the scurvy-fighting power of fresh lemon juice,” but I couldn’t find a source for that, I could find many contradicting it, and the USDA data suggests that it has ~3/4 of the vitamin C density of fresh lemon juice).
So it seems that the lime juice was just not preventing scurvy because it had spent long periods of time open to the air, and had been pumped through copper tubing. And so this paragraph from the post:
“Different kinds of citrus fruits are more like one another than they are like polar bear meat” sounds very reasonable, but in this case it was wrong. Sicilian lemons really ARE more like polar bear meat than they are like West Indian limes, at least for the purposes of treating scurvy.
seems very misleading (especially but not exclusively if you don’t interpret “polar bear meat” as referring to a specific polar bear organ, given that my sources say that fresh polar bear meat has 30x less vitamin C than fresh lime juice).
- 31 May 2022 2:11 UTC; 32 points) 's comment on New Water Quality x Obesity Dataset Available by (
Highly processed food became dominant well before the great obesity-ing started, right?
Do we have evidence of that? As far as I can tell, the SMTM authors merely argue that some specific brands of processed food were available starting from the late 19th century, not that “highly processed food became dominant before the great obesity-ing started,” which is a much stronger claim.
Also, as I argued in this post, the obesity epidemic arguably started way before the SMTM authors often seem to imply it did. Quoting myself:
And it’s not as if Americans were that thin in the early 1970s! 47% of adults were overweight and 14.5% obese (a). In contrast, obesity rates are under 3% in traditional societies that engage in foraging or subsistence farming. Moreover, there is substantial (a) evidence (a) that Americans gained a lot of weight before 1970. It’s hard to know the overweight and obesity rates of the general population back in the 19th century, because there was no NHANES back then, but we do know that men at elite colleges (a) (source), Citadel cadets (a) (source) and veterans all started getting substantially fatter in the early 20th century.
I feel the need to stress this, because the SMTM authors claim (a) that there was an abrupt shift in obesity rates in the late 20th century, a claim that is probably based to some extent on an artifact of the definition of BMI (a), and so some people reading this might have the impression that 1970s Americans were really thin or something, when they really weren’t.
Here are a few charts from the sources I linked to. From this VoxEU article:
A chart by Random Critical Analysis, using data from this book:
A chart made with fancy statistical modeling, from this paper (note that the x-axis is each cohort’s birth year):
This is also relevant (from Random Critical Analysis as well):
We can see that BMI increases superlinearly with body fat %, so if body fat % is linearly growing in a population, BMIs will accelerate.
It’s very clear to me that it’s fine (and often great!) to investigate implausible theories. It just seems to me that the SMTM authors are doing a very bad job at actually pursuing the truth, as demonstrated by the facts that, e.g.
they wrote a “literature review” that only includes studies that are outliers, refused to address the fact that they are outliers, and are actively trying to prevent their readers from knowing that (by refusing to approve my comment on their post despite having had the time to approve several comments that were made afterward)
they have misrepresented their sources several times (as I show in this section and this comment) and have refused to correct their posts after being told about it
they seem oddly uninterested in all of the other common side effects that lithium causes at therapeutic doses, even though it’s very easy to know that those side effects are significant by merely reading the Wikipedia page on lithium salts
they seem oddly uninterested in whether therapeutic doses of lithium cause enough weight gain to explain the obesity epidemic
I also think it would be more helpful if they told readers their credences on their hypotheses, and what led them to reach those credences. But this is a minor point compared to the 4 above.
This misses the fact that people’s ability to negatively influence others might vary very widely, making it so that it is silly to worry about, say, 99.99% of people strongly negatively influencing you, but reasonable to worry about the other 0.01%. If Michael is one of those 0.01%, then Scott’s worldview is not inconsistent.
On top of picking 5 of the ~20 estimates I mentioned to claim that low estimates of dietary lithium intake are “strictly outnumbered” by studies that arrive at much higher estimates, they also support that claim by misrepresenting some of their own sources. For example,
They say that “Magalhães et al. (1990) found up to 6.6 mg/kg in watercress at the local market,” but the study reports that as the lithium content per unit of dry mass, not fresh mass, of watercress (which the SMTM authors do not mention). This makes a big difference because Google says that watercresses are 95% water by weight.
They do the same thing with Hullin, Kapel, and Drinkall (1969). They do not mention that this study dried lettuce before measuring its lithium concentration, and reported lithium content per unit of dry mass. Google says that lettuce is 95% water by weight too, so this matters a lot.
They also mention a lot of other dry mass estimates, such as those from Borovik-Romanova (1965) and Ammari et al. (2011). This time they do disclose that those estimates are for dry mass, but they nevertheless present those estimates as contradicting Total Diet Studies, as if they were measuring the same kind of thing, when they are not.
Notably, they say “Duke (1970) found more than 1 mg/kg in some foods in the Chocó rain forest, in particular 3 mg/kg in breadfruit and 1.5 mg/kg in cacao,” and fail to mention that most of the foods in Duke (1970) have less than 0.5 mg/kg of lithium.
Also, only one of the examples SMTM used to claim that the Total Diet Studies are “outnumbered” actually attempted to quantify dietary lithium intake,[1] whereas almost all of the studies I’ve mentioned do that. This is important because a lot of the sources they cite that we don’t have access to (there are several of those) could be measuring lithium concentration in plant dry matter, as a lot of their sources that are available do, in which case seemingly high concentrations do not imply high dietary consumption.
Moreover, a lot of their post is focused on speculating that ICP-MS (the technique used by most studies) systematically underestimates lithium concentration. However:
Van Cauwenbergh et al. (1999) use atomic absorption spectroscopy (AAS) instead of ICP-MS, and arrive at the second-lowest dietary lithium intake estimate I have ever found,
Iyengar et al. (1990) mention a lot of NA-MS measurements, all of which match the low estimates I’ve found,
Hamilton & Minski (1972) use spark source mass spectrometry (SSMS),
Evans et al. (1985) use flame atomic emission spectrophotometry, and
Clarke & Gibson (1988) use NA-MS.
All of those find very low concentrations of lithium in food.
Moreover, they themselves mention a paper that uses ICP-MS and finds high concentrations of lithium in food in Romania (Voica et al. (2020)).
These studies are a substantial fraction of all of the studies on lithium concentration in food that we have. So it seems to me that their whole focus on ICP-MS, and their claim that it “gives much lower numbers for lithium in food samples than every other analysis technique we’ve seen,” does not seem warranted.
Again, I don’t think that studies that find high concentrations of lithium in food are necessarily wrong. There is no market pressure for food to have 1 µg/kg rather than 1000 µg/kg of lithium, or the other way around, the way that there is market pressure for meals to have e.g. carbohydrate/fat ratios and energy densities within a specific optimal range. Consumers do not care about whether lithium concentration is 1 µg/kg or 1000 µg/kg. And we know that lithium concentration in e.g. water varies a lot according to lithology and climate, so we shouldn’t expect this to be uniform around the world. So I don’t see how it must be the case (as the SMTM authors claim) that all studies that find low concentrations are wrong.
- ^
The example was Schrauzer (2002), which bases its estimates on hair concentration rather than actual food measurements. Ken Gillman says that this paper “has a lot of non-peer-reviewed and secondary references of uncertain provenance and accuracy: it may be misleading in some important respects.” Also, interestingly, as I mentioned in my post, the highest estimate Schrauzer (2002) provides for dietary lithium intake is from China, not really a country with a huge obesity problem.
(Note that the Ken Gillman blog post has a typo: it says that the “typical total daily lithium intake from dietary sources has been quantified recently from the huge French “Total Diet Study” at 0.5 mg/day,” a value that is 10x too high.)
- It’s Probably Not Lithium by 28 Jun 2022 21:24 UTC; 442 points) (
- 6 Jul 2022 23:33 UTC; 8 points) 's comment on It’s Probably Not Lithium by (
Thanks for engaging. I hope we can figure out what our cruxes are.
On points 1 and 2: I still think you haven’t addressed my counter-argument to points of that nature, which I’ve raised (1) in the post itself, (2) in my previous reply to you, and (3) in this comment. To reiterate some of it: you use sleep research to support some of your points, mostly in your review of Walker’s book but also in Theses on Sleep, and it’s not clear to me why research that shows that sleep deprivation is not as bad as people think is admissible evidence to you but research that shows that sleep deprivation is not harmless is not. I was, and remain, skeptical that you have a consistent and rigorous standard for what research you think is admissible and what research you think is not.
...
Just to reiterate: I believe that the points 4, 5.4.1, and 5.4.2 invalidate large chunks of sleep literature and are simply not possible under the default “sufficient sleep is good and necessary for proper functioning for normal people on the scale of a few days to a week”.
Notice that my post does not argue against anything that is invalidated by 4, 5.4.1 or 5.4.2.
As I elaborated before in my post, 5.4.1 pretty much just shows that sleep deprivation doesn’t make you very acutely sick or something, which is (1) not something I’m arguing against, (2) not something the sleep literature is arguing against, (3) not a surprising claim, and (4) not a novel claim.
Concerning 5.4.2, we don’t know how well people in the show would be doing in a rigorous assessment of cognition not subject to learning effects, such as the psychomotor vigilance task or other similar tests. I appreciate that the show is evidence that people are able to perform some tasks even while very sleep-deprived, but again, that’s not something I or the sleep literature are arguing against, and it isn’t novel or surprising.
...
even if we only take people with bipolar disorder: how the hell can they go on so few number of hours a night with their brain being manic but not simply breaking down?
Uh, whatever manic people are doing, it’s putting them at hospitals and making them unable to functionally recover even 2 years later at incredibly high rates, so they really aren’t a good example of a group that sleeps very little consistently without breaking down.
...
Also, notice that I could have written a post just like yours, with the same kinds of evidence that you use — not bringing up any meta-analysis at all — but instead arguing that sleep deprivation impairs cognition and can worsen mood. The kind of evidence that you use does not asymmetrically favor you. You may have a high prior that all anecdotes about sleep deprivation impairing cognition are untrustworthy, and all anecdotes denying that are trustworthy, but that too isn’t asymmetric; why can’t I claim that it’s the other way around because there’s a psyop making people wrongly believe that there are no harms to sleep restriction, fueled by bosses that want their reports to spend more time working and less time sleeping? (This is not something I actually believe, but I’m just pointing out that the kind of evidence that is consistent with your points is not asymmetrically so; you don’t elaborate on why we should expect people to be massively wrong about their experiences in that specific direction and not the other).
- 1 Apr 2022 19:39 UTC; 18 points) 's comment on Counter-theses on Sleep by (
400 extra calories over a year (assuming 3500 calories = 1 pound) is an extra 41.7 pounds per year.
In reality, you’ll stop gaining weight at some point if you increase your caloric intake once and never change it again, because your energy expenditure will rise.
But even taking that into account, it does seem to me that 456 extra kcal per day is way too much.
Here’s an illustrative calculation. Herman Pontzer has the following equations relating body weight to total daily energy expenditure in his book (figure 3.4):
The average woman in the US weighed 65.5 kg in the 1970s, and 78 kg in the 2010s, so this predicts a TDEE increase of only 134 kcal for women. For men, the figure is 160 kcal. Those numbers are about a third of 456 kcal. So yes, a 456 kcal in average daily energy intake would be a jaw-dropping increase.
Of course, this 456 kcal number is based on self-report data, so it’s not likely to be that accurate. Stephan Guyenet mentions a better estimate on The Hungry Brain based on food disappearance data from the USDA, which is only 218 kcal/day, much closer to my estimates of how much TDEE has changed.
I agree with Eliezer et al. that CICO by itself cannot explain the obesity epidemic, but “a 456 kcal increase is not that much” is a bizarre argument.
This isn’t as important as my previous reply (in which I address your object-level arguments), but I wanted to perhaps note that most of your points 1.1 through 1.4 sound, to me, more like an attempt to generate an emotional reaction in the reader than a good-faith effort at pointing out mistakes you think I’ve made or investigating object-level disagreements (although I could be wrong). I don’t recall criticizing you for not having an MD or something, or publicly speculating that you have never thought about [important meta-level epistemological consideration].
I understand that the fact that I did not take biology or neuroscience classes in college is evidence that I would not have a good understanding of sleep research, but I think it is perhaps important to keep in mind that argument screens off authority here, and it sounds plausible that, a lot of the time, domain experts in the area would acquire knowledge in it the same way I do (by reading meta-analyses and systematic reviews, or textbooks based on those). They don’t have some sort of magical essence that makes them more knowledgeable than everybody else could become. They do original research, but not in every sub-area of their field. If I saw you explaining to someone why a massless particle can carry energy and momentum if it travels at the speed of light, and you were using the same arguments that convinced me why that was the case, I would not object simply because that (probably) wasn’t part of your formal education, as it was of mine, or because you didn’t talk to a physicist about it.
(I’m also a bit confused about why you are criticizing me for not talking to people in the field if, in your view, those people are mostly untrustworthy and just want to show that sleep deprivation is bad[1]).
- ^
Which, in my experience, was not the case — I was able to find many systematic reviews and meta-analyses claiming that sleep deprivation is probably not bad for some things. (For example, blood pressure, inflammation, and mortality, as I’ve pointed out in the post).
- ^
I noticed recently that the tradeoff you have to make to be more dependable in that way is to be less open. Less open to new projects, new information, new people. You have to be less malleable, and more definite. It is largely about being able to knowingly cut the majority of the world from your attention, to ignore what isn’t important. I don’t think that’s a bad thing—there’s much more joy in being focused and determined than in shifting your attention and commitment around. But it is something that comes more naturally to people once they figure out what seems to them to be the right path, once they figure out a task or project that deserves their undivided attention and commitment, once they don’t feel like they’re getting stuck in a local maximum in an avoidable way.
Outcomes for veganism are [...] worse than everything except for omnivorism in women.
As I explained elsewhere a few days ago (after this post was published), this is a very misleading way to describe that study. The correct takeaway is that they could not find any meaningful difference between each diet’s association with mortality among women, not that “[o]utcomes for veganism are [...] worse than everything except for omnivorism in women.”
It’s very important to consider the confidence intervals in addition to the point estimates when interpreting this study (or any study, really, when confidence intervals are available). They provide valuable context to the data.
- EA Vegan Advocacy is not truthseeking, and it’s everyone’s problem by 28 Sep 2023 23:30 UTC; 319 points) (
- EA Vegan Advocacy is not truthseeking, and it’s everyone’s problem by 29 Sep 2023 4:04 UTC; 115 points) (EA Forum;
- 4 Oct 2023 23:28 UTC; 1 point) 's comment on EA Vegan Advocacy is not truthseeking, and it’s everyone’s problem by (EA Forum;
I don’t recall having argued that “proper sleep is necessary for x.” [1]
I’ve argued that sleep restriction impairs cognition, is associated with negative mood and suicidality (not only positive mood) and causes overeating. Not that you need a certain minimum amount of sleep to perform competently at something. So I still don’t think that the kind of evidence that you bring up is asymmetric.
I’ll come up with an example to explain the difference. Suppose that, after 8 hours of sleep, humans complete a task with an average accuracy of 83%. After 6 hours of sleep, the average accuracy decreases to 71%. And even after a full night of sleep deprivation, accuracy is still 60%, substantially better than a dart-throwing monkey (which would have, say, a 10% accuracy at the task). Suppose that accuracy above 50% is considered acceptable. You point out that the average accuracy of the sleep-restricted/deprived groups is still pretty good, and argue that this shows that sleeping 8 hours per night, or sleeping at all the previous night, is not necessary to complete the task acceptably. And obviously, you’ll be right — but that wouldn’t be an argument against the claim that sleep restriction impairs performance on that task.
you do not represent accurately in your comment—the participants were able to perform complex mental tasks and function in general at what appears to be ~full-capacity, not just “able to perform basic tasks”
I edited my comment. Notice, however, that it’s still unclear how people in the show would be doing in a rigorous assessment of cognition not subject to learning effects, and, as far as I know, we don’t know how well they did on their tasks compared to their baseline performance.
- ^
Claims of that sort would not be very general, and their truth-value would obviously depend on what you mean by “necessary,” so I don’t think they’re that interesting. Also, in some tasks there’s probably a very big difference between acceptable performance and peak performance.
- ^
The version of Alexey’s theses I most broadly support is “we treat sleep like one thing, when it is in fact multiple things with multiple purposes”. Natalia calls this epicycles
That sleep is multiple things is not something I am arguing against. (In fact, I don’t recall that being a point in Guzey’s post). What I uncharitably called “epicycles” was the additional complexity Guzey’s model has to have to explain why so many people feel dumber after sleep restriction, and why experimental studies say that sleep loss causes cognitive impairments, when ““not sleeping ‘enough’ makes you stupid” is a 100% psyop.”
It’s pretty clear to me that sleep has multiple effects, and that it might be the case that there’s something with all of sleep’s good effects and none of the bad ones that just hasn’t been discovered yet. Maybe digital people would only have to spend the subjective equivalent of a few seconds per day shut down in order to renormalize their weights or whatever, or might not need to be shut down ever at all to maintain their performance.
But I don’t think that is incompatible with the object-level claims in my post, any more than saying “humans don’t live to be 200 years old” is incompatible with knowing that the passage of time has multiple effects or with admitting the possibility of revolutionary life-extension methods being discovered that do allow humans to be 200 years old.
No. (Thanks for pointing out that that’s not clear). My model is that moderate sleep restriction harms cognition the following a day, regardless of how many nights of it you’ve had before; I’m much more uncertain about whether it causes permanent damage to cognition.
1.
Natália’s section about bipolar people seems to imply that [sleep deprivation’s short-term antidepressant effects] would not be happening.
I disagree. I said,
A night of total sleep deprivation seems to be able to trigger full-blown mania in a substantial percentage of people with bipolar disorder (even those currently depressed) and even cause mania-like behavior in healthy subjects. Moreover, a shift towards mania or hypomania after a short night of sleep seems common in bipolar patients.
Here, I think it was clear that what I said is consistent with sleep deprivation having antidepressant effects, and it could even be interpreted as implying that it does. So I think it’s misleading to suggest that this section implied that the antidepressant effect does not exist.
2. Your section arguing that occasional sleep deprivation is good for health makes no mention of its antidepressant effects, which were addressed separately earlier on in your post. I thought you were making a separate argument in that section, which is why I countered with an appropriate analogy. I merely think that the argument “sleep deprivation causes acute stress, therefore it’s good” is weak, and that was my point in that section. My particular analogy might not have been great, however, I agree.
Separately, I don’t think that association between sleep deprivation and mania is evidence that sleep deprivation is good rather than bad; as my section in this matter showed, manic episodes very often have severe long-term consequences.
3. I apologize, I used poor phrasing here that made it seem like I was claiming something I wasn’t. I didn’t mean to say that you were hypothesizing that 6 hours was causally optimal, in the sense that people should sleep for 6 hours if they want to have the lowest mortality, in that paragraph. I was using the word “optimal” to mean “associated with the lowest mortality.” I’ll rephrase the paragraph to make it clearer that I was not interpreting you as making a causal claim.
Overall, I don’t think the errors you pointed out so far were particularly glaring. The last part of your point (1) seems to be based on a misunderstanding of what I wrote, though perhaps upon further elaboration we’ll find that we do actually disagree on something specific here. Point (2) reflects more of a clash of intuitions between us, rather than a mistake on my part; it’s reasonable to disagree about the strength of my analogy, but it really wasn’t a large part of my argument. Point (3) was merely an error in the sense that I used poor phrasing when describing your position.
I think it’s a little unfair to say “The fact that I spotted these three points after spending ~3 minutes skimming the post do not make me optimistic about the rest of the critique” when your points were individually and together, quite weak. However, I am hopeful that we can have a productive dialogue about this subject soon, and get closer to our cruxes on these issues.
This made me wonder about a few things:
How responsible is CSET for this? CSET is the most highly funded longtermist-ish org, as far as I can tell from checking openbook.fyi (I could be wrong), so I’ve been trying to understand them better, since I don’t hear much about them on LW or the EA Forum. I suspected they were having a lot of impact “behind the scenes” (from my perspective), and maybe this is a reflection of that?
Aaron Bergman said on Twitter that for him, “the ex ante probability of something at least this good by the US federal government relative to AI progress, from the perspective of 5 years ago was ~1%[.] Ie this seems 99th-percentile-in-2018 good to me”, and many people seemed to agree. Stefan Schubert then said that “if people think the policy response is “99th-percentile-in-2018″, then that suggests their models have been seriously wrong.” I was wondering, do people here agree with Aaron that this EO appeared unlikely back then, and, if so, what do you think the correct takeaway from the existence of this EO is?
The paper Mendonça cites looks at long-term long sleep and long-term short sleep, with their association with depression. My claim and my evidence (from bipolar people) are concerned with short-term long sleep and short-term short sleep.
Your specific claim about depression was “depression triggers/amplifies oversleeping while oversleeping triggers/amplifies depression.” Nowhere in the section did you specify your claim was about short-term long sleep.
Your evidence, too, barely concerns short-term long sleep: depressive episodes last about five or six months on average, which is often not what people have in mind when they think about “short-term” oversleeping, and it’s common for them to last a year or more.
It is further puzzling that in her refutation of my argument she:
completely ignores the relationship between sleep and mania (i.e. she ignores one half of my argument and only discusses the part about depression).
completely ignores the fact that in ~50% of ALL people with depression (not just bipolar), short-term short sleep relieves depression.
I don’t disagree with you on the claim that sleep restriction or deprivation often causes mania, and can adequately treat depression in some cases. I also mentioned the relationship between sleep and mania elsewhere in my post.
It was not my intention to respond to every single one of your claims, or to mount a comprehensive takedown of every aspect of your post. I’ll repeat what AllAmericanBreakfast said to you after you made a similar argument in your previous comment:
Natália doesn’t set out to disprove all of your theses, but rather to put forth some counter-theses. She says:
I decided to write a post pointing out several of the mistakes I think he’s made, and reporting some of what the academic literature on sleep seems to show.
Read carefully, she neither claims that every point you’ve made is mistaken, nor to give a comprehensive review of the academic literature. [...] She’s critiquing those theses of yours which she found weak, not issuing a comprehensive point-by-point criticism of your entire original post.
Guzey:
I wrote that most sleep research is extremely unreliable. And indeed, this is what I believe and this is why I’m so selective with the kind of evidence I use
Yes, I wasn’t claiming that you said that all sleep research was bad, but that the sleep research you’re willing to trust is also, suspiciously, seemingly exclusively research that indicates that sleep is not as important as people think, when there doesn’t seem to be a difference in quality between the research you cite as evidence and the research I do. You use evidence from experimental studies to show that acute sleep deprivation often relieves depression, and I use evidence from experimental studies to show that sleep restriction seems to cause cognitive impairment and overeating, and you haven’t elaborated on why the research you cite is more trustworthy.
In other words, I understand that you do not think that all sleep research is bad. However, I was, and remain, skeptical that you had a consistent and rigorous standard for what research you thought was admissible and what research you thought was not.
What Mendonça seems to miss more generally is that meta-analyses, which she is relying on heavily, do not reflect reality: they reflect the consensus of an academic field. And if the academic field is confused and the majority of the papers published in it are garbage, then meta-analyses are going to be confused garbage as well.
Why can’t I say the same about e.g. the “enormous literature” you use as evidence that sleep deprivation helps depression? [1] Why does that reflect reality, but not my meta-analyses? (These are not rhetorical questions, I’d like to know what you see as the difference). It’s not the case that, as you claim in the original post, the kind of evidence I use is “like cognitive psychology” and is only not suffering from a replication crisis because sleep studies are hard; the findings I talk about do get replicated. It’s also not as if my evidence is inconsistent with people’s experiences.
Also, it’s instructive to note that academic fields are not monolithic things, as Scott Alexander explains in this essay. As he points out, a lot of correct contrarians were in fact supported by academic research when they made their “contrarian” points. Knowledge just doesn’t propagate that quickly among people who work in the same field. As an example, some clinical psychiatrists promote the serotonin theory of depression, but the academic research body on psychiatry does not support it. Lumping clinicians and the research body together as a monolithic “psychiatric field” obscures this.
1:
Chronic sleep deprivation and insomnia can act as an external stressors and result in depression, characterized by hippocampal BDNF downregulation along with disrupted frontal cortical BDNF expression, as well as reduced levels and impaired diurnal alterations in serum BDNF expression.
This quote talks about depression (due to chronic sleep deprivation and stress) being characterized by BDNF downregulation. This is not about sleep deprivation per se because sleep deprivation does not necessarily lead to depression.
[...]
3:
[O]ur findings are in line with the hypothesis of an increased stress vulnerability due to sleep loss which may lead to a decrease in BDNF. [...] While we report a reduction of BDNF levels linked to sleep disturbance reflecting chronic stress on the one side, we and others consistently showed that prolonged wakefulness caused by SD (partial or total), which can be considered as an acute stressor for the brain, leads to a rapid increase of BDNF
This quote again talks about stress due to sleep loss, not sleep loss per se
These quotes simply say that chronic sleep deprivation may lead to decreased BDNF expression through a certain mechanism (depression and stress, respectively).
[The quote starting with “[O]ur findings are in line[...]”] specifically notes “we and others consistently showed that prolonged wakefulness caused by SD (partial or total), which can be considered as an acute stressor for the brain, leads to a rapid increase of BDNF”, so I’m very confused by Mendonça uses it to contradict me.
I’m not contradicting you on the acute effects of sleep deprivation on BDNF. I specifically said,
These sources agree that acute sleep deprivation increases BDNF expression, but they also say that the opposite may happen when sleep deprivation is chronic
2:
significantly decreased serum BDNF levels compared with sleep-healthy controls (n=24; F(1)=5.017; P=0.03; Figure 1a). In addition, serum BDNF levels were significantly correlated with severity of insomnia in all paricipants (n=50; rp=−0.409; P=0.004; Figure 1). [...] We found subjective sleep impairment to be associated with lower serum BDNF levels, whereas reported good sleep was related to higher serum BDNF levels, as shown for those suffering from current insomnia compared with sleep-healthy subjects.
This quote talks about people with insomnia having decreased BDNF. People with insomnia have all kinds of health issues and are famous for underestimating how much they sleep (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3277880/) and using them to argue about the relationship between sleep deprivation and BDNF is misleading
Thanks for pointing out that I had an incorrect assumption; I realize now that it’s unclear whether insomniacs sleep less than other people. I’ll edit the post.
instead of noting that using insomniacs as evidence is simply inappropriate for the point I was making and writing that I’m using inappropriate evidence, she doesn’t note any issues with it and simply notes that it contradicts my point.
Hm, I had assumed that you were simply showing a random sample of the first papers that you found when searching “sleep deprivation bdnf.” You simply said
Papers that showed up when I googled “sleep deprivation bdnf”: The Brain-Derived Neurotrophic Factor: Missing Link Between Sleep Deprivation, Insomnia, and Depression. The link between sleep, stress and BDNF. BDNF: an indicator of insomnia?. Recovery Sleep Significantly Decreases BDNF In Major Depression Following Therapeutic Sleep Deprivation.
So I wasn’t expecting all of those papers to be appropriate for the point you were making.
- ^
Not that I would want to say that; I don’t disagree with Guzey on the effects of sleep deprivation on depression.
- 1 Apr 2022 14:59 UTC; 22 points) 's comment on Counter-theses on Sleep by (
I find it implausible that your anecdotes and a non-RCT N=1 self-experiment provide stronger evidence than several N≈20 non-pre-registered RCTs.
Yes, p-hacking and lack of pre-registration are bad, but IMO those things are pretty much negligible concerns when studies test cognition with several tests and find the same effect on almost all of them.
When I read the literature on the cognitive effects of sleep deprivation, it doesn’t sound like experimenters are giving subjects several tests, finding no effect on most of them, and then focusing on reporting the single p<0.05 result. Rather, they find medium-to-large effect sizes on most things they test, and effect sizes almost always have the same sign—very rarely favoring sleep deprivation. That could be because the experimenters are giving subjects 15 tests and only report the results of 5 of them, but it sounds unlikely that all sleep researchers involved in these studies are doing that. So it doesn’t look like p-hacking or lack of pre-registration are driving the huge effect sizes here.
It sounds a lot more plausible that an N=1 self-experiment has fatal flaws to it, especially if the one study subject wants the results to come out a certain way.
ETA: I no longer endorse the fact that I linked to that specific meta-analysis. See my new post for a better investigation of the effects of sleep restriction on cognition.