Medical and Psychological Studies for June

Original spider picture edited out after complaints.

The benefits of exposure to ultraviolet light may outweigh the risks, say scientists who apparently hate putting on sunscreen as much as I do. This reminds me of my father’s occasional complaints that every year he finds out that the health advice he was giving his patients last year was driving them to an early grave.

It’s been a good month for gut microbiota research. A team from UCLA finds that probiotics change activity in brain areas that affect emotional processing in humans as well as actual behavior on emotion-related tasks.

Mice raised without normal gut bacteria show autistic-like behavior, and giving them normal gut bacteria reverses many of the symptoms. Extremely interesting because prevalence of autism keeps increasing and no one knows why, and the amount of stress we put our gut flora under with weird diets and antibiotics also keeps increasing and would be really neat if we could use the one to help explain the other. Bonus: this study was conducted by one of my professors during medical school who was sort-of-very-briefly my research advisor.

Adjusting the level of the gut bacteria a. muciniphilia causes a bunch of downstream changes in the guts of mice which eventually affect their obesity levels. But please, tell us again how all obesity is 100% due to how much you eat minus how much you exercise.

Elsewhere in the gut, children born after their mothers have bariatric surgery are less likely to be obese and have different gene methylation patterns than children born just before their mothers have bariatric surgery – suggesting that fetuses may be “programmed” to metabolize fat differently based on their mother’s health. BUT IT’S STILL ALL JUST EATING AND EXERCISE LA LA LA I CAN’T HEAR YOU.

In my Biodeterminists’ Guide, I suggested the then unproven conjecture that First World iodine deficiency was probably having small but measurable effects on kids’ intelligence. Last month, research showed this was true in the UK.

Old study: Medicare costs vary wildly around the country because doctors are basing their treatment on local fads rather than sound practice. New study: Medicare costs vary wildly around the country because health varies wildly around the country. But be sure to read the responses and counter-responses.

Ketamine has long been able to treat depression instantaneously and miraculously. The only problem is that, well, it’s ketamine. Now scientists announce the discovery of a molecule that may be able to replicate ketamine’s antidepressant effects without its side effects.

I’d previously seen hypothermia caps used to fight insomnia, but apparently the real medical interest is in a different direction: hypothermia caps can allow chemotherapy patients to keep their hair. The mechanism is actually very clever: you wear the cap, the cold contracts the blood vessels in your scalp, and the chemotherapy drugs don’t reach the hair follicles.

A study finds that cruise ships cost about the same as assisted-living facilities for the elderly, provide about the same range of services, and are much more fun. It suggests that more seniors should go on permanent cruises instead of “off to a home”.

Meta-science strikes again: peer review is basically useless. Researchers find that peer reviewers will reject articles they previously accepted but forgot about at around the same rate they will reject articles in general – at least once the names of the prestigious scientists involved are stripped off.

Genetic risk for schizophrenia is associated with decreased IQ even in those who don’t develop the disease (I can’t find by how much). This is interesting for two reasons. First, the correlation was observed at age 70 but not at age 11, which is yet another reminder that a lot of these IQ studies just don’t work all that well on kids who are still growing and having their IQ fluctuate for other reasons. Second, in light of the recent total failure to figure out IQ genes, it’s nice to sort of have something to work with.

75% of lesbians are obese, which is much larger than comparable numbers for straight women or gay men. A new study is trying to find out why. It would be mildly interesting if social, super interesting if biological as it might give us more insight into what’s going on hormone-wise.

All research relating to oxytocin is existentially terrifying, and the Karolinksa Institute’s work on oxytocin and pair-bonding is no exception. The study finds that the 25% of people with a certain allele of an oxytocin receptor score significantly differently on measures of relationship status, a discovery the media has not entirely unfairly spun as a “divorce gene” that makes you 50% more likely to break up with your partner. This is the first time I’ve really on a gut-level appreciated how scary the age of univerally available genetic information is going to be. Do people have a right to ask their prospective partner what version of the oxytocin receptor they have? (also worth checking out: the other divorce studies mentioned in that second link)

Just as religious people are more likely to believe their religion in times of stress, so people are more likely to believe in the power of capital-s Science in stressful situations. This raises some interesting possibilities for increasing the likelihood that the public will take scientific advice seriously, like preceding announcements of important studies that people should take seriously with pictures of a giant spider climbing on a man’s back.

This entry was posted in Uncategorized. Bookmark the permalink.

68 Responses to Medical and Psychological Studies for June

  1. Romeo Stevens says:

    Thank you for editing the post.

    I’d still apply sunscreen to your face as 90+% of skin aging is photodamage. I did some cross referencing on the effects of various types of sunscreen and discovered that [this]( is the best combination of usability and non-cancer, non-acne inducing.

    • Romeo Stevens says:

      also I didn’t mean to include a referral link, no idea what it is for. You can search “eltamd sunscreen spray” to avoid it.

      • Kibber says:

        Are we really turning this thread into a sunscreen comparison?

        Also, I just happen to know that the sunscreen you mentioned uses nano-particles of ZnO and TiO2, and the science is still not settled on the safety of that; furthermore, the particles do not appear to be coated, which, as far as I know, can make them chemically active in some environments.

        • Romeo Stevens says:

          Regarding the passive diffusion of TiO2 the EU Scientific Committee on Cosmetics and Non-Food Products (SCCNP)46 published a paper based on studies with micro- and nano-sized material. Herein they state that these particles remain on the skin surface or on the outer layer of the stratum corneum and do not penetrate into or through the living skin.47–49 Confirmation was obtained with studies on human, porcine or murine skin50–53 for particles within a size range between 10 and 100 nm. These data were recently confirmed by the outcome of an EU project (NanoDerm). Here, TiO2 are only found in the top layer of the stratum corneum and the openings of the hair follicle.54 Similar results were obtained for ZnO.50,55 Just recently two studies on ZnO56 and TiO2,57 were published; ZnO penetration was investigated in vivo with human volunteers and located the particulated materials only on the skin surface and their accumulation in skin folds and/or hair follicles.56 In vitro measurements of the penetration of TiO2 particles between 20 and 100 nm showed the nanoparticles only in the top 3–5 layers for all skin samples used (porcine skin, healthy human skin and human skin grafted on a severe combined immuno-deficient mouse model).57

        • Kibber says:

          Right, this is a review from 2009, here’s a 2010 study that states otherwise:, here’s a 2012 one:

        • Romeo Stevens says:

          interesting. thanks for the links.

  2. Matthew says:

    forgive the formatting error their, that was supposed to be a response to ” BUT IT’S STILL ALL JUST EATING AND EXERCISE LA LA LA I CAN’T HEAR YOU.”

  3. Matthew says:

    I think it’s plausible that you have a situation, a la the bravery debates post, where the modal overweight reader of your blog is likely have done a good job trying to implement the conventional weight-loss wisdom, but the average overweight person in general has not, and so the appropriate advice is audience-dependent.

    I was chubby kid. As it turns out, I actually lucked out massively in the genetics/metabolism department, and now have roughly the physique of an army ranger. My problem actually was that I wasn’t eating right and getting enough exercise. And now I see way too many people at the gym “exercising” on the elliptical or the treadmill, who’ve barely broken a sweat after their 20 minute “workout”. The average overweight rationalist may really have already sincerely tried to optimize their exercise and dietary behavior. I’m not convinced the average overweight person has, however.

  4. BreiAnne says:

    About the 75% of lesbians are obese, the article linked links to another article saying it’s 26% of lesbians that are obese, which seems a little suspect.

    • Berry says:

      Maybe they meant 75% more than the straight population?

      • Deiseach says:

        According to the CDC, “More than one-third of U.S. adults (35.7%) are obese”

        If 26% of lesbians are obese, that is less than the percentage among the entire population.

        Unless anyone wants to blame the ‘obesity epidemic’ on lesbians? Get rid of all the lesbians, and immediately you reduce the incidence of obesity amongst the general population!

        This is just one more of the “What in the name of God are social scientists doing, that they have nothing more pressing than sitting around devising research about are lesbians fatter than straight women/straight men/gay men?” Can we not get these people out doing something useful like sweeping the streets?

        • Berry says:

          “The report, published in the American Journal of Public Health, looked at a survey of more than 67,000 Massachusetts residents between the ages of 18 and 64.

          According to, researchers found that while 21 per cent of straight men were classed as obese, only 14 per cent of gay men were.

          Conversely, they found that 26 per cent of lesbians were obese, compared to 17 per cent of straight women.”

          This is what following the links leads you to, and it does not look very epistemically healthy when contrasted with the extrapolated claims.

        • Scott Alexander says:

          “This is just one more of the “What in the name of God are social scientists doing, that they have nothing more pressing than sitting around devising research about are lesbians fatter than straight women/straight men/gay men?” Can we not get these people out doing something useful like sweeping the streets?”

          Suppose there’s a 10% chance this study leads to a solution that makes 10% of lesbians normal weight instead of obese. Suppose there are two million lesbians in the US. That means 20,000 lesbians become normal weight. If being normal weight instead of obese makes you live 5 years longer, that’s 100,000 extra years of life. In other words, probably several hundred times more important than all the good you and everyone you have ever met has done combined.

          Public health is weird.

        • Deiseach says:

          Scott, unless these studies are going to lead to some conclusion that being a lesbian has a definite genetically-associated risk of obesity (the same way it might do with breast cancer or diabetes, for instance), then it’s nothing more than a curiosity. “Hm, gay women are slightly more likely to be fat than straight women”.

          “Suppose there’s a 10% chance this study leads to a solution that makes 10% of lesbians normal weight instead of obese.”

          Because there certainly aren’t lots of campaigns about adult obesity, child obesity, the obesity epidemic, etc. etc. etc. already being instituted? Fat taxes on sugary drinks and fast foods proposed? What do you propose – a special Lesbian Diet? And how will that be different to any other kind of diet? Unless it will mean that doctors might treat obese lesbian patients by prescribing drugs to help them lose weight (ah, the magic goal which isn’t yet achieved – the fat melting pill!) rather than just tell them “Eat less, exercise more” like the rest of the population?

          I admit, it would be wonderful to get the medical profession to listen to women when they describe their symptoms, rather than tune them out and hand out a canned recommendation. If I thought that would work, I’d claim to be a lesbian myself (maybe then, a male gynaecologist would believe me when I tell him that no, these are not normal periods and not brush me off with ‘ah, you’ll be fine when you go through the menopause’).

        • Scott Alexander says:

          Deiseach: It seems possible to me that this is something hormonal, and that finding out exactly what might lead to a treatment.

        • Randy M says:

          That’s not entirely fair, Scott. You have to divide that good amongst all the people who have and will contribute to the eventual increase in lesbian-life-years (weirdest unit ever).
          Or else any burger flippers he might know are responsible for keeping every one of a billion people alive by feeding them once.

      • St. Rev says:

        The linked study says 26% of lesbians in the sample are obese, compared to 17% of straight women in the sample. Which a) gives a relative risk of 53%, not 75%, and b) as pointed out is still lower than the general population. So perhaps the question should be: why are straight women in this sample less likely to be obese? I blame Massachusetts hipsters.

      • Douglas Knight says:

        The 75% figure cites a grant proposal (though that copy truncates the author’s surname, which is Bryn Austin). But it is a misquote: she says that 75% of lesbians are overweight or obese, compared to 50% of straight women.

        The second link quotes Conron (esp table 2) that 26% of lesbians are obese, compared to 17% of straight women (and 21% of straight men and 14% of gay men); but that’s for Massachusetts, which is on the lighter side. Including overweight, the gap shrinks to 50% of lesbians vs 44% of straight women in MA.

  5. Michael Vassar says:

    I have been saying that this is the best guess WRT Sunlight for a while now.

    The ‘stress and science’ study supports the ‘crisis signaling’ vs. ‘abundance signaling’ hypothesis that I have been developing for a while now.

    I’d love to talk about MetaMed with all of the scientists who you share studies from, but particularly the publishers of the ‘cruise’ study. Maybe MetaMed should invite a bunch of them on a cruise together (with their elderly relatives?). Maybe not, but it’s a fun thought. If you want to make the introductions, I’m totally serious.

  6. jsalvatier says:

    I thought we had the first Genome Wide Association Study hits for IQ just recently: (author is a genetics researcher)

    • Douglas Knight says:

      The article Scott links talks about the same study, though Scott’s description sure doesn’t sound like it.

      Hsu calling it “first hits” is extremely misleading. It is the first replication, not the first claimed genes. The fact that most GWAS are unreplicable bullshit is worth pointing out. Though that’s probably not special to IQ.

  7. The peer review study unfortunately does not send the paper to the original reviewers, so all this establishes is that the reviewers are capricious, lazy, and/or varied in taste and competence. Further, it naively suggests that the prestige of the submitter shouldn’t be used as information by reviewers (I didn’t read to see whether the review process for the journal claims to be author-blind). People who experience peer-review rejections have often suspected this.

    That said, it’s a nice exercise, and peer review should be studied.

  8. cool rich guy says:

    Holy shit, 75% of lesbians being obese is fucking crazy. I always thought the “fat lesbian” stereotype was a myth.

    What’s strange is that I failed to notice this. Sometimes I wonder where all the obese people actually are. I’m not really sure what the boundary between obese and non-obese looks like, but I would guess that I only know four or so of them personally. I know, I know, I’m upper middle class and obesity is negatively correlated with income, but still. Like, apparently 50% of black people are obese, but I feel like most of the black people I see aren’t obese, even the poor black people I see when I’m driving through poor areas. There must be some place where all the black people are obese to balance this out. It could be the South, I guess, but then there was that article claiming that Southerners aren’t fatter, they’re just more honest? Or maybe obese people just leave their house less. Who knows.

    • Shel says:

      A lot of obese people don’t look like what we associate with obesity:

      • bilbo says:

        The only people there who violated my expectations of what their weight range looked like were the ones who were very muscular — and people with lots of muscle are a case where everyone knows you can’t meaningfully use BMI. The rest are pretty much what I’d expect, though a lot are hard to tell because you can’t actually see any of their body due to baggy clothes or poor angles.

        The BMI seems to get a lot of hate, and I really don’t know why. Here’s a defence of it that I found to be a pretty good read, if anyone’s interested:

        • Deiseach says:

          I’ll tell you why the BMI gets bashed, it’s because it’s a blunt instrument. It correlates height and weight and that’s it.

          Now, I’m 5’4″ so according to that, I should be a particular weight (which I’m not). However, that measure says that me and her and them over there, because we’re all the one height, should all be the one weight (falling within a small range) to be “normal” – and not just “normal”, but “healthy”.

          My sister, when she was very stressed, lost weight to fall to the lower edge of the BMI ‘normal’ weight for her height recommendation. She was not healthy; she was gaunt and haggard and rundown. When she put on half a stone (seven pounds) extra, she looked a lot better, had more energy – and according to the BMI, was dangerously close to tipping into ‘overweight’ category.

          When I was waiting for a hospital appointment, I happened to see a woman walking by who was – quite literally, I’m not joking – flat as an ironing board (she had no breasts or buttocks; the line of her back was a straight line down). She probably was a ‘healthy’ weight by the BMI and she was the same height as me.

          I can tell you, no matter what diet I go on, I would not achieve the same build as her until I’m dead and reduced to a skeleton.

          The point being, the BMI takes no account of body shape – I have hips that are good sturdy peasant child-bearing hips, my mother’s broad shoulders and my granny’s thick wrists. I am never going to be ‘thin’. I am never going to be the same as a Twiggy type, yet the BMI measure is the one we are all judged by.

        • St. Rev says:

          The thing I find most puzzling about BMI is why the denominator uses the square of height rather than the cube.

        • Army1987 says:

          From Wikipedia:

          associated only with a difference in size, rather than with differences in weight relative to one’s ideal weight. If taller people were simply scaled-up versions of shorter people, the appropriate exponent would be 3, as weight would increase with the cube of height. However, on average, taller people have a slimmer build relative to their height than do shorter people, and the exponent which matches the variation best is between 2 and 3. An analysis based on data gathered in the USA suggested an exponent of 2.6 would yield the best fit for children aged 2 to 19 years old.[33] The exponent 2 is used instead by convention and for simplicity.

  9. yli says:

    Feels weird seeing this kind of post on this blog seemingly naively hyping new studies. I wonder if *any* of those studies would stand up under the kind of scrutiny you put the Victorian and abortion stuff under.

    • Scott Alexander says:

      Yeah, I thought about that. But the fact is that I don’t have the time or energy to analyze every study I come across, and I do find cutting-edge medical research interesting. All of these had high enough priors that I don’t doubt them too much, but I agree we should think of them as intriguing possibilities and not as The Thing That Is About To Cure Cancer.

  10. ag says:

    The spider was cute and not at all stress-inducing.

    Try some slimy, crawling worms next time. Those things are horrifying.

  11. gwern says:

    > Second, in light of the recent total failure to figure out IQ genes, it’s nice to sort of have something to work with.

    How is that a total failure when they actually found SNPs?!

    More to the point, it’s exactly what we expected from the earlier SNP studies showing IQ was going to be highly polygenic and we’d need large samples, and if anything, is incredibly encouraging because it shows we are getting hits just from pooling the data we already have (it was a meta-analysis of ~12 sequencing studies) and more importantly, that the smallness of the effect sizes (or inversely, the largeness of the sample sizes) aren’t hugely off from previous polygenic traits like height.

    Hsu discusses this in

    > Note Added: I’ve been asked by several people whether this is a discouraging result. If effect sizes are so small, won’t it take enormous sample sizes to detect specific alleles accounting for a big chunk of total genetic variance? There is some relevant discussion in the supplement to the paper (see figure S22 and section 7). The answer to the question really depends on the correlation between g and years of education (most of the data these researchers had access to specified educational attainment as the phenotype, with no direct measurement of g). If, for example, the correlation is 0.5, and it is actually g that is driving the effect on years of education, then the corresponding g effect size for these alleles is (1/0.5)^2 or 4 times larger in variance units. This makes the g effect size in variance units about 5 times smaller than for the corresponding largest height locus. However, if the correlation is only 0.25, the g effect is about as big as the largest height locus. Having looked at correlations between SAT and college GPA, I’d guess that 0.5 is too large, but on the other hand in the Swedish sample for which they have both g and years of education the correlation is 0.46. Using 0.5 as the correct correlation, the minimal sample size with *actual g data* to detect these rs alleles is (see Fig S22) in the 20-50k range. I’d guess that, worst case, the sample size requirements are still less than an order of magnitude larger for g than for height. However, one can’t be very confident of any guess because of the uncertainties discussed above, and because we’ve only seen the first few alleles.

    • Scott Alexander says:

      Total failure might have been the wrong words, but for me at least it was a very strong signal that we have no idea what’s going on.

      If I remember right, their best SNP explained 0.02% of variation, when IQ is about 50% hereditary. That would require 2500 SNPs of that size, but since that was the one largest SNP, more likely that means variation in IQ is distributed among many more SNPs, maybe 25000 or more each of which explains 0.002%.

      The more independent determinants of intelligence there are, the less variation in intelligence there should be in the real world, just by the law of averages. If we consider a very simple model where there’s an 0.5 chance of each SNP ending up “smart” or “dumb”, and the average person has 12500 smart and 12500 dumb of the 25000 determinants, then the chances of getting even so little an imbalance as 13000 one way and 12000 the other way is so small it breaks my calculator. But that’s what we’d need for intelligence to vary just 1%.

      This suggests something like Gregory Cochran’s mutational load idea, where multigene attributes like intelligence and height are related to a general eugenic or dysgenic disposition caused by how much mutational load there was among your ancestors. But I don’t really understand how that could explain IQ variation within a single population/ethnic group, and I definitely don’t understand how it could produce findings like this link with schizophrenia (unless all they’re actually finding is that dysgenic people have both low IQ and high risk of schizophrenia, but that seems unlikely to me for reasons I can’t quite place)

      Also, I don’t know if people outside Cochran’s readers are aware of the mutational load idea, and yet no one in mainstream science seems too worried about the implications of this 25000 SNP thing, suggesting I’m missing something.

      Anyway, the fact that this result makes me despair of the possibility of making sense of intelligence in terms of traditional “Well, you have some SNPs, and I have others” sort of thing and forces us to draw in complicated and speculative concepts like mutational load is what I mean when I refer to the study as a failure.

      I’d be really interested in hearing what you think about this and whether my analysis is completely wrong.

      • gwern says:

        I don’t know why polygenic traits can have such clear differences between individuals and families, but we already knew that IQ would be many thousands of SNPs at a minimum ( To take a less controversial example: we all can agree that height is to a large degree genetic, varies considerably between families and populations, has large heritability estimates etc, and yet, just like with IQ, it is weakly influenced by a large number of variants. So clearly this situation is possible and height doesn’t get ‘smeared out’ by the law of averages so everyone is 5.5 feet tall or whatever.

        • Scott Alexander says:

          I agree with you entirely. I guess I already expected something like this, but it would have been nice to have been proven wrong or even to find one or two SNPs that accounted for 5% or so.

          (actually, now that I think of it, I think torsion dystonia is a single mutation which causes a multi-point IQ jump, which makes it especially weird that there’s no healthy-people equivalent).

          As far as you know, are people just totally ignoring the “why do polygenic traits vary so much?” question?

          Like, this bothers me enough that it makes me wonder whether there’s some non-genetic form of inheritance going on here. Epigenetics doesn’t really have the characteristics we would want to explain IQ variance, but absent some way to make all this stuff vary together (and granted, mutational load may be such a way) I’d rather believe it’s epigenetics somehow than try to figure out how 25000 tiny insignificant genes come together to produce important differences.

          • gwern says:

            > but it would have been nice to have been proven wrong or even to find one or two SNPs that accounted for 5% or so.

            Oh goodness no. If there were any SNPs that huge, they would’ve been found a long time ago.

            > (actually, now that I think of it, I think torsion dystonia is a single mutation which causes a multi-point IQ jump, which makes it especially weird that there’s no healthy-people equivalent).

            The space of ‘big effect with no or horrific consequences’ is always going to be at least as ‘big effect with no horrific consequences’, so it’s not really that surprising. Or to put it another way, it’s like asking ‘why didn’t Africans evolve better malaria resistance than sickle cell anemia?’ You’re asking why there isn’t some enormous low-hanging fruit available for the plucking… It’s Algernon’s law all over: if there were some huge IQ boost sans horrific consequences, we would already have taken it.

            > As far as you know, are people just totally ignoring the “why do polygenic traits vary so much?” question?

            I don’t think they’re ignoring it. My assumption is that they’ve already answered it, I just haven’t read it or have forgotten it or don’t know the right search query. (Quantitative genetics is one of the areas where my default assumption is never ‘I’ve made a trenchant criticism and asked a deeply insightful question!’ but always ‘I am failing at some basic conceptual point – or alternately someone terrifyingly more intelligent and/or competent than me has already answered this question in detail, I am just ignorant of what the answer is’.)

        • MugaSofer says:

          “why do polygenic traits vary so much?”

          Funnily enough, googling just that (with quotes) has exactly one result.

          It confidently informs you that this is because “polygenic traits do not follow Mendelian patterns of inheritance.”

      • Douglas Knight says:

        The more independent determinants of intelligence there are, the less variation in intelligence there should be in the real world

        Yes, the more genes, the smaller the natural variation is compared to the scale of a corrected genome. So what if human variation is 1% of the difference between a human and an error-corrected human?

        • Scott Alexander says:

          I think that the fact that we know the strongest SNP contributes less than .002% allows us to move from a relative to an absolute scale.

  12. I wish you had a Like button on your posts. Even if it was no use to you, it would make me feel good to push it. And I’m probably not the only one.

    Or, put a different way, it feels wrong not to somehow cheer “hooray!” at most of them.

    (If you happen to take this seriously—some of us aren’t on Facebook, so *their* Like button is not what we want… I suppose this implies that you might need to have five different brands of Like, which is probably more trouble than even a very generous person would want to go to.)

    • Scott Alexander says:

      Thank you. I might add a button like that, thought it feels kind of whorish to me. I do like publicity and I have Facebook and Twitter share buttons at the bottom of my posts, so if you can think of anything else you want along those lines please let me know.

      • Forgot to ask: do you have an actual (public) twitter account? I’d want to mention it when I tweet links to your blog (plus follow of course).

      • Lucidian says:

        If you do put in a Like button, I would greatly appreciate it if it was somewhere quite inconspicuous. I like being able to read blog posts without being influenced by the groupthink of “this blog post is currently at +177 likes, many of which are from your friends!”

      • Berna says:

        I would like ‘like’ buttons. I love your posts, and many of the comments as well, but I don’t want to post ‘Great post/comment’ and ‘me too’ all the time.

  13. Berry says:

    The payoff for the spider thing was great, but I’m seriously anxious about opening up the site now becasue HOLY SHIT GIANT SPIDER AAAAHHHH!!!!

  14. bilbo says:

    I think you’re strawmanning the ‘eat less move more’ argument. I usually argue that whilst eat-less-move-more is the bottom line, plenty of factors can affect how difficult it is to do either. I suppose there is a third term I haven’t considered though, which talking of gut bacteria made me realise, so we actually have:

    Food-in minus exercise minus food-pooped-out-because-gut-is-too-inefficient-to-extract-energy-from-it = weight-gain.

    If so, it still seems to be the case that people whose guts are the most efficient (ie, those with nothing wrong with them) ought to have the most potential for weight gain and sickness ought to in almost all cases only ever decrease efficiency.

    So I don’t buy it. Those mice who gained weight must have ate more or moved less. Sure, hormonal imbalances could have caused them to be irrationally hungry or lazy, but it still comes down to them eating more or moving less.

    You posted on your old blog once about leptin, and how it looking like there’s a feedback cycle whereby eating lots of carbs stops a chemical process that would normally send your brain an “I’m full!” signal. I see explanations like this as totally consistent with the ‘eat less move more’ mantra, and I think you only have a case against those who use that argument whilst also saying ‘it’s easy!’.

    • St. Rev says:

      Thermogenesis burns far more calories than physical activity for most people, varies wildly for mostly unclear reasons, and isn’t subject to conscious control.

      Please stop victim-blaming now.

      • Scott Alexander says:

        The parent comment seemed intelligent and well-intentioned, and the victim-blaming comment isn’t really helpful.

      • bilbo says:

        I don’t think that’s true; according to wikipedia:

        About 70% of a human’s total energy expenditure is due to the basal life processes within the organs of the body (see table). About 20% of one’s energy expenditure comes from physical activity and another 10% from thermogenesis, or digestion of food (postprandial thermogenesis).

        Perhaps you mean that there is significant variation in the basal metabolic rate? That certainly seems to be the case, though most of it is explained by differences in lean body mass, when corrected for that and a few other things you get a remaining unexplained 30% difference between the fastest 5% of metabolisms and the slowest 5% (citation is paywalled and my uni doesn’t have a subscription, but the abstract talks about mice, not humans, so this data might not be representative of humans).

        If this data is representative of humans, we would expect the slowest 5% of metabolisms to use very roughly about 1500kJ less per day than the fastest 5%. Assuming a symmetric distribution, this is about 750kJ less than the average. However if the distribution is bottom heavy like I suspect, this number could be much less. I suspect it would be bottom heavy because again, unless you have major organ damage (your liver uses 27% of your BMR for example), I would think your body ought to become less efficient with sickness, not more.

        I am very inconsistent with my exercise, and over the past 2 years my GPS tracked exercise averages to 500kJ per day. This is much less than the recommended 30 mins of moderate intensity exercise per day that doctors normally recommend, which would come to about 1800kJ per day for someone of my weight.

        So although this isn’t what I set out to show, it seems that a moderate amount of exercise is indeed enough to counter most of the variation in basal metabolic rate.

        Despite this, most of my weight loss came not from exercise but from eating less. This is purely anecdotal, but my appetite became a lot more reasonable after I started exercising. This is consistent with the leptin-signalling explanation that Scott has talked about in the past, as aerobic exercise is known to decrease triglycerides, which may be blocking leptin from getting across the blood-brain barrier to send the “I’m full!” signal.

        Eating less would have been a lot harder though if not for this saving grace of a decreased appetite, so although I attribute my weight loss to eating less and moving more, I concede that there is more going on than just that, and that others may have other physiological reasons for having a high appetite that makes it difficult for them to lose weight.

        • St. Rev says:

          I am lazy and mistaken. I was incorrectly conflating BMR and thermogenesis. Half my point is wrong (the variability is lower than I claimed) but the other half stands: your food-in equation should read

          Food-in minus exercise minus basal metabolic rate minus thermogenesis minus food-pooped-out-because-gut-is-too-inefficient-to-extract-energy-from-it = weight-gain

          …and exercise is small compared to BMR. Thus aside from ‘ate more/less’ and ‘moved less/more’, there’s ‘digested more/less efficiently’, ‘generated less/more heat’, ‘scavenged/built non-fatty tissue’, ‘fought infections’, ‘secreted calorie-dense material’ (e.g. excreting/exhaling/sweating ketone bodies under ketosis), …

          Point is, there are many more terms in the equation than calorie intake and exercise, even without getting into the feedback loops involved.

    • Deiseach says:

      People do burn the same amounts of calories differently and respond to stress in different ways. I get stressed? I comfort eat. My sister gets stressed? She loses her appetite. So yes, eating more than you expend in energy will put on weight, but then again, there are people who can “eat anything and not put on weight” and they’re not noticeably more active, and there are those of us who the minute we relax a rigid diet will pack it all back on. And I’m fairly sure I’m not the only woman out there who finds that she retains/more easily burns off weight at certain points of her menstrual cycle, and it’s not purely fluid retention.

      Re: the ‘lesbians are more obese’, I’d like to first know what they consider ‘obese’ – do they mean “not starving themselves to fit into a size zero dress”, do they mean “several pounds over what’s considered a healthy weight (whatever the current definition of ‘a healthy weight’ may be)” or do they mean “really very much heavier than is healthy”?

      • Emily says:

        The word “obese”, when used by American health researchers to talk about adults, has a standard definition: having a BMI of 30 or greater. (Sometimes that group is itself split into different subgroups, so you might hear about obese vs., say, morbidly obese.) If you hear the word “obese” when used in connection with current health research, it is safe to assume that’s what they mean. A BMI of 30 for someone who is 5’6″ corresponds to a weight of 186 lbs. You might also find someone using a definition based on body fat percentage or measurements.

        • von Kalifornen says:

          It should be noted that BMI is considered a poor metric for individual people.

    • Shel says:

      If being thin for one person requires eating until they’re not overstuffed, and for another person requires leaving the table a little bit hungry every day for the rest of their lives, it is perfectly reasonable for the second person to be like “…nope, rather be fat.”

      It also makes me wonder about correlation/causation issues: what if some hormonal imbalance or something causes both weight gain and poor health? I mean, I’m not saying that’s how it works, but it is an interesting hypothesis.

    • Scott Alexander says:

      I’d seen a lot of discussion around this point before, but the other side of the argument didn’t really sink in for me on a gut level (no pun intended) until I saw this comment by Eliezer on Less Wrong:

      There is no thermodynamic law stating that fat cells must release fat just because your body needs it. If you’re built so that weight loss is impossible and you try eating less, your metabolism slows down – possibly in much the same way it would as if you tried eating less and you had no fat cells whatsoever. I can’t cite studies but wouldn’t be particularly surprised to see that muscle gets cannibalized instead of fat being lost, if you try to eat less than the most slowed metabolism needs. And if most metabolically disprivileged people stop trying to eat below their minimal metabolic rate before doing significant damage to themselves, that’s just the survival instinct kicking in. I would seriously not be surprised to find that fat people have starved to death without their fat cells releasing fat, and blinded by preconceptions, nobody managed to notice or note down when this occurred. But I would expect that to be rare – most people, if their body tells them they’re starving to death, will eat. This gets cited as weakness of will.

      [Thin people assume that if you eat less, your fat cells will release fat.] No. We don’t have energy storage units like you do, we have energy retention units. Calories go in, they don’t come out. Or if they do, it’s on special occasions we don’t understand how to predict or trigger, and which don’t have any obvious relation to attempts to eat less or exercise more. The laws of thermodynamics do not require that a physical fat cell physically release stored lipids when you eat less or exercise more – and if your fat cells are malfunctioning, they just won’t.

      This seems to sort of fit what you’re saying and what I said before while also being much much worse.

      • Kibber says:

        Now, I’m not really a specialist in dieting, but my understanding was that people that literally starve to death are generally very skinny at the time of death (especially if they’re given at least some food as opposed to no food at all) – with the evidence coming from WWII documentaries mostly.

        • Some examples of people starving to death while fat.

          This suggests to me that it’s possible to be malnourished while fat.

        • Army1987 says:

          Ey said “generally”, not “always”, so a few cherry-picked examples to the contrary don’t invalidate eir claim.

        • Kibber says:

          > This suggests to me that it’s possible to be malnourished while fat.

          While I would definitely agree to that claim, your link does not seem to support it. All the deceased people from the list appear to have died from various obesity-related causes. While some of them were indeed on a strict diet shortly before or at the time of death, that by no means makes the diet responsible for it. Brief investigation suggests that the only man from the list claimed to have “literally starved to death” (Michael Edelman) has suffered from congestive heart failure and pneumonia, and neither conditions would make you feel like you’re starving to death (thus not supporting Eliezer’s hypothesis).

      • bilbo says:

        Thanks Scott, that is indeed a very persuasive way of putting it, and I agree it’s consistent with but more extreme than what we’re both saying. I am sceptical though, maybe people can become extremely, extremely hungry to the point where absolutely nobody alive would have the willpower to resist eating, but starving to death, perhaps not.

        • St. Rev says:

          Some people do have the ‘willpower’ to starve themselves to death, if you can call it that: hunger strikers, and people with anorexia. I’ve known at least one person casually who died that way. The suffering involved is horrific, though.

      • Michael Vassar says:

        Don’t military drafts give us good info on the reliability with which people do or do not respond in a predictable way to a shared food environment? What about prisons?