Biodeterminist's Guide to Parenting

by Scott Alexander of Astral Codex Ten

2012 December 15, rescued from his old LiveJournal with permission

After my recent posts about genetics, a lot of the comments have focused on some form of “So, how long till I can have a designer baby?”

And the answer is “probably longer than your reproductive lifetime”, because the scientific community is wasting its time investigating the mating habits of tree shrews or something instead of focusing on giving us awesome superbabies who will be able to do quantum chromodynamics in kindergarten.

But for a long time I’ve been collecting studies that show the possibilities for non-genetic intelligence enhancement of children. In answer to some of these comments, and as a Christmas present for Alicorn, I have spent the last week writing them up into a presentable format, which I will now name “The Biodeterminist’s Guide To Parenting”.

There are many reasons to be a biodeterminist parent. The best available research shows that normal variation in parenting doesn’t have a lot of effect on how a child turns out. By all means avoid traumatizing or malnourishing children, but finding the perfect balance between permissiveness and authoritarianism probably isn’t going to have a lot of effect.

But children still turn out different. That implicates genes and a mysterious category called “nonshared environment”, factors other than genetics and parenting style which differ among siblings. No doubt some of this category is social: who was bullied in the third grade, who had an especially inspiring high school science teacher. But there’s also a lot of opportunity for biological factors to make a difference. And unlike the social factors, these are mostly weird, fascinating, and unexpected.

Almost all these factors require strong caveats. In some cases, they’re based on flimsy research that might be overturned in the future. In other cases, the research results are clearly correct, but causation cannot be established. There’s a certain pattern where once doctors start recommending Intervention A, all the smart conscientious parents do it, but the ignorant apathetic parents don’t bother. Then ten years later when doctors finally get around to testing their recommendations, they find that Intervention A is correlated with smart, conscientious children — this proves nothing but that children turn out like their parents, which we already knew. In other cases, smart conscientious people buy nice homes in places far away from toxic chemicals and then have their smart conscientious babies, which to scientists looks a lot like toxic chemicals making babies dumber. Some studies try to control for these effect, with more or less success, but controlling for confounders is a highly uncertains science. The best option would be to run randomized controlled trials on everything, but there are ethical limits to how much we can do (we can’t randomly expose pregnant women to toxic chemicals without getting yelled at by ethics committees) and so I have only been able to include a few of them here.

Aside from these two big challenges there are other problems. Sometimes an effect may hold only in a certain population: for example, malnourished people can probably improve child outcome by feeding them better, but that doesn’t mean a rich First World family can get a superior outcome by stuffing their kid’s face. Two different effects may be (probably are) non-additive: that is, if Intervention X gains 3 IQ points, and Intervention Y also gains 3 IQ points, doing both A and B may not give you 6 IQ points — they may be addressing the same fundamental problem and so only give you 3 when used together.

I used IQ points as an example because it’s one of the most common outcomes I was able to find data on. In fact, I’ve really focused on it just because it allows me to compare different interventions in a way non-numerical quantities don’t. And I think in this form it is a good proxy for most of the other variables too. At the level of coarseness we’re working at, “good outcomes” is almost a natural category. Biological insults to the brain tend to cause trouble in every conceivable way at once; many of these studies find that children with such insults are lower-IQ, lower school performance, higher violence, more prone to psychiatric (and physical) disease, shorter, lower lifespan, et cetera. As such, I’m not going to make a big deal over whether IQ is really interesting or not, and I’m going to mix various outcome measures in ways that will probably make real epidemiologists frown concernedly.

Most people don’t have a good intuitive feel for IQ. Just to help calibrate how much you should care about these, each extra IQ point is associated with about a 2% increase in lifetime earnings and a 2% increase in worker productivity. A 15 to 20 point rise in IQ, which is a little more than you get from supplementing iodine in an iodine-deficient region, is associated with half the chance of living in poverty, going to prison, or being on welfare, and with only one-fifth the chance of dropping out of high-school (“associated with” does not mean “causes”). The average IQ of a janitor is 92, the average of a doctor is 120, and the average of a Nobel Prize winner is 144. Because of the way standard deviations work, raising IQ by 10 points (a little less than the size of the iodine effect) sextuples (multiplies by six) the chance of having IQ > 140 and therefore in Nobel Prize territory.

Saving the most important caution for last. I am not a doctor, I am a doctor, but an untrained one without any expertise in this area. Please take this document entirely as entertainment and a pointer to interesting areas for your own investigation, not as ultimate truth. Epidemiological research of the sort being discussed here is almost always low-quality, and as soon as it starts involving IQ it becomes a hundred times worse. I have specifically used a non-scientific tone and peppered this document with corny jokes in order to signal that you should not take it seriously. Enjoy reading the various results, but keep in mind that this is intended as a random fun project only. Caveat lector.

On a similar note, there is a known failure mode where parents obsess over every possible parenting tip, feel like they are a horrible person unless they get every single one right, and eventually go insane. Please don’t do this: elevated maternal levels of the stress hormone cortisol impede fetal brain development and can cause a three to five point drop in IQ (1, 2). But seriously: this is not worth your sanity and if statistics like the preceding are likely to be triggers for you please stop reading this document; the tiny adjustments it might or might not be able to produce aren’t worth it.

If you don’t want to read several pages of me complaining about how much I hate all of these studies, you can just skip to the summary at the bottom.

BEFORE HAVING A CHILD

Let’s start with the obvious: the most important biological factor affecting a child’s life outcomes is who the parents are. If you’re trying to optimize your own child, that leaves one degree of freedom: your partner. Choosing a partner with the traits you want in your child is a no-brainer. Now that we’ve gotten that out of the way, are there any other unexpected factors that might influence your choice of partner?

Age

How old should your partner be? How old should you be?

Maternal age is a bit complicated. The best known effect is an increased chance of Down Syndrome with increasing maternal age, which goes from 0.1% at age 25 to 3% at age 45 (Down Syndrome is very detectable by prenatal screening, but the process can be complicated and can occasionally require invasive tests with a slight chance of fetal injury). This effect has now been joined by increasing rates of autism — your chances go from about 1% to about 1.3%.

And yet increased maternal age is also correlated with increased IQ, up to a point. 30 year old women have babies 8 points smarter than 20 year old women. The slope levels off after this; waiting until 45 only gains you an extra 2 points over 30. There is no good biological explanation for this and it is almost certainly because lower-class women are more likely to marry and have children earlier, whereas upper-class women prefer to finish schooling and maybe advance professionally first. Since upper-class people tend to have higher IQs, this suffices to explain the correlation.

How about paternal age? Paternal age is linked to a massive number of diseases and conditions. Among the best established are autism (relative risk 1.4, or in absolute terms from 1% to 1.4%) and schizophrenia (by age 50, your kids have 3x the risk as at age 25, so in absolute terms about 1% to 3%). Although this result has been replicated in a few studies, there are also studies that find no relation (this one found no relationship to paternal age, but did find a relationship to the mother’s age).

Paternal age also has a well-established relationship with IQ. An intriguing study I can’t access says it can account for about 2% of variation in intelligence, saying that the highest intelligence is between ages 25-44 (I think the drop before 25 may be for the same reason as in the women; the drop after 44 seems more plausibly biological and may result from accumulated mutations in sperm). The same study I took the female figures from suggests maybe a 2-3 point IQ drop for every ten years a man waits before having kids. Biologically plausible, replicable, and in the opposite direction as one would naively expect the confounders to go. Looks solid enough.

Hybrid Vigor

We will avoid the question of race like the plague, except to note the theory of Alon Ziv and Michael Migroni that interracial humans are, ironically, the superior race. Ziv writes in a very popular style, talking about “hybrid vigor”, the well-known biological effect where hybrids of two subgroups are healthier than pureblooded members of either. He claims that multiracial people are “healthier, more attractive, more symmetrical, more athletic”, and isn’t above citing the obvious examples of really pretty multiracial Hollywood actresses to prove his point. Migroni takes a much more academic tack, suggesting that greater outbreeding (“heterosis”) has spread dominant high-IQ genes and so is partly responsible for the Flynn Effect, the secular rise in IQ.

Ziv is right that hybrid vigor is a well-known phenomenon in dogs, cattle, crops, and other domesticated species, but a lot of this is because these species are pretty inbred (think about eg purebred dogs) and hybridization removes some of the detrimental effects of inbreeding; this isn’t really an issue in humans. Ziv cites a few studies showing that interracial humans are on average taller (usually a good measure of health), but none of them are very convincing. And his opponents (usually blogs in the HBD community, which may not be the most unbiased source and which have enough racial slurs in their reviews that I don’t feel comfortable linking to them) counter with studies showing some health detriments among mixed-race individuals, for example genes in whites and Asians which are perfectly healthy on their own but which cause increased cardiovascular problems when combined. Overall I find his argument weak, although I have to admit that many of the mixed race people I know are indeed staggeringly sexy.

Migroni is more respectable, but suffered from a couple of well-conceived takedowns, including one showing the Flynn Effect is unlikely to be genetic and Gene Expression noting that we haven’t seen abnormally high IQ increases after sudden interminglings of populations like occurred during the settlement of British colonies. Steve Sailer did mention one study did show that mixed white-Japanese kids in Hawaii are two IQ points smarter than monoracials, but I can’t find it and Steve didn’t seem very convinced by it. Overall I find the “hybrid vigor” theory unimpressive.

Epigenetics

Despite a lot of research, epigenetics remains mysterious. But research is suggesting that epigenetic changes allow lifestyle factors from before parenthood to be passed on to children. The Overkalix study is complicated enough to be questionable and not of practical value, but a good example of how important epigenetics can be. Some studies suggest that paternal smoking can cause metabolic syndrome (related to obesity) in children. Overall this field is too new and underexplored to suggest anything quantifiable.

Month of Conception/Birth

Surprisingly, month of conception / month of birth has significant effects on child outcomes. Various explanations have been put forward for this effect. Some are social; in countries where children begin school or athletics at a particular time, birth month may make someone younger or older than their peers. Others are biological: birth month may affect how much Vitamin D (often produced via sunlight) is available to a developing fetus, and certain diseases that could harm a fetus may be seasonal, such as influenza. Maybe the lower classes give birth on a different schedule than the upper classes — one economist attributes the whole thing to stupid people having unprotected sex during Christmas parties and after prom. Or maybe it’s because of varying goblin activity. This entire field is incredibly confused and mysterious, as we will soon see.

If you want to avoid disease, your best bet is probably to conceive a child in winter so it will be born in autumn. Children born in autumn live about six months longer than those born in spring, and have lower risks of diseases like childhood diabetes, Crohn’s disease, alcoholism, bipolar disorder, schizophrenia, Parkinson’s, multiple sclerosis, and even — decades later — Alzheimers’. The change is moderate but significant: for example, 33% less chance of MS and 10% less chance of schizophrenia.

Social outcome also seems to favor the autumnals. People born at the beginning of the year may be up to 10% more likely to drop out of high school and about 0.5% more likely to drop out of college. In Britain, summer-born children are 3% less likely to qualify for A-levels (I think this is a sort of advanced college prep class?) than the autumn born. Children born in the first quarter of the year make about 2.38% less than children born in the last quarter. The standard (and very reasonable!) explanation for this revolves around age at school entry and whether you’re the oldest or youngest child in your class.

Then things get ugly. Study after study shows higher intelligence (2-4 points), more favorable personality traits (for example, 33% less extreme shyness) and greater height (0.6 cm) in winter/spring babies (1, 2, 3, 4). Study after study, the authors, who have apparently never read any of the other studies on this subject but only the ones on schizophrenia, talk about how surprised they are: like me, they expect “good outcomes” like less disease, greater height, and higher IQ to co-vary. Despite an ungodly amount of time trying to massage this effect away or find some justification for it, I have to admit defeat. Have a kid in autumn and you decrease disease risk. Have a kid in spring and you increase IQ, height, and sociability.

So what do you do? If you’re going to send your child to public school, I think giving birth in the autumn is a no-brainer. The data showing better outcomes for people who enter school after the school cutoff date (and are thus the eldest in their grade) is robust, and the data showing better health outcomes (especially less schizophrenia) for autumn babies is on equally good footing.

If you’re not going to send your child to public school…tentative and contradictory and possibly noncausal as it is, I can’t help being impressed by the data showing better weight, height, cognition and social skills for winter babies. If you don’t mind taking a chance, late February / early March might be the way to go.

Really, the only thing one can say for certain in a complicated field like this is that Canadians should make sure to have their kids in January: it just about doubles their chances of becoming a professional hockey player.

DURING PREGNANCY

Many women don’t know they’re pregnant until a month or two into the pregnancy, but the first few weeks are an important period for the developing fetus. Good family planning will make it easier to suspect pregnancy as soon as it begins and better consider the advice below.

Alcohol

Everyone knows that drinking large amounts of alcohol during pregnancy is a bad idea. The effect of light or moderate drinking is very controversial, with some people saying that it’s dangerous and other people in the “unintended consequences” tradition saying that telling people to stop light drinking is dangerous.

This turns out to be another horrible minefield. First, one expects smarter and more conscientious mothers to be more likely to heed the medical advice not to drink, hopelessly confounding everything. Second, there’s a well-known tendency from alcohol/lifespan research that a lot of people who don’t drink are abstaining because they’re sick for some reason; this hopelessly confounds everything in the opposite direction.

For a while the evidence seemed to be converging around the idea that light drinking (a glass or two a week) was not harmful and might even be helpful. Then a month or two ago a genetic study came out showing light drinking correlated with up to 3.5 points lower IQ in people with a slower version of the alcohol metabolism gene only. The numbers looked good to my untrained eyes, but having a recommendation, finding no evidence to support it, and then throwing genes at it until one sticks is a classic epidemiology failure mode.

The way medical recommendations work is that if they know a lot of drinking is bad for you, and they can’t figure out the point at which drinking stops being bad for you, they will just say don’t drink at all. This seems reasonable. On the other hand, if you’re super into alcohol and stopping drinking would be a traumatic experience for you, realistically there’s not much evidence to condemn you if you want a glass of wine every few days. Just don’t overdo it.

Smoking

Smoking doubles the risk of low birthweight babies. Low birth weight is associated with motor problems, mental retardation, and increased mortality. There is also a more tentative attempt to link it to up to double the risk of childhood obesity. As for IQ, the 6-or-so point drop associated with smoking is probably mostly confounding, but smoking does increase the risk of low birth weight and low birth weight does decrease IQ. If smoking makes 100% greater relative chance of low birthweight (= about 5% greater absolute chance), and low birthweight has on average 5 points lower IQ, smoking during pregnancy costs 0.25 IQ points from that source alone.

Okay, it’s not very impressive. Still, don’t smoke, okay?

Iodine

This one is awesome! Iodine deficiency decreases a child’s IQ by ten to fifteen points, and cheap, simple iodine supplemenation of pregnant women and children can completely reverse that! This is a huge deal in places like Afghanistan where a very large portion of the population is iodine deficient, and some biodeterminist historians have tried to explain the continuing problems of those areas, all the way up to its current problem with terrorists, with cognitive handicaps due to lack of iodine.

You are not as iodine-deficient as an Afghan and you will not increase child IQ fifteen points by supplementing with iodine. The average urinary iodine level in a pregnant woman is about 125 mcg/L, which although less than the very conservative WHO recommendation of 150 mcg/L is still way above the level associated with retardation in Afghanistan, Pakistan, and Western China (20 mcg/L).

But about 10% of the US population and possibly 20% of pregnant women have urinary iodine levels below 50 mcg/L and the UK may be even worse. This is below the level (60 mcg/L) that was found in a New Zealand study to improve test scores when supplemented with iodine. A study in Spain found that children with an iodine level above 100 mcg/L had 3 points higher IQ than those below, which could mean either that iodine in early childhood is important or that iodine in early childhood is correlated with mother’s iodine during pregnancy.

Overall I am super disappointed by the lack of good studies showing whether the mild to moderate iodine deficiencies in First World pregnant women translate into long-term outcomes. But there’s very little risk of supplementing with iodine unless you’re an idiot about it, so the benefits seem to outweigh the risks. It would not surprise me if the 20% of iodine deficient pregnant women in the US could increase child IQ a point or two by taking a prenatal vitamin with iodine (not all of them have it, so watch out) or just using more iodized salt (ironically, the anti-salt campaign by the hypertension people is responsible for a lot of the current iodine problem).

And keep the iodized salt after you give birth; not only does iodine reduce gynecological complaints in women, but as we saw supplementing iodine in children can raise their intelligence directly if they’re deficient.

If a sufficient amount of iodine is good, would excessive amounts of iodine be even better? This is exactly the sort of question responsible nutritionists totally avoid asking because responsible nutritionists are boring. Luckily, a study of unsafe drinking water levels in China gives the answer: too much iodine doesn’t create geniuses, it just has mild thyrotoxic effects that you should probably try to avoid. If you don’t eat a pound of salt with every meal you’ll be fine.

Other popular micronutrient deficiencies include folic acid and iron. These should be in the prenatal vitamin your doctor will recommend anyway, so we skip over them without much comment.

Maternal Stress

Studies show that babies whose mothers suffer high leves of stress in pregnancy can have IQ up to ten points lower than those who don’t. Possible confounders are obvious; things like being poor or in a bad family environment are stressful, and also harm child development.

That was why I was excited to learn about Project Ice Storm, maybe the coolest-named study ever. Scientists compared women who were pregnant during a catastrophic ice storm in Quebec to those who were pregnant during less stressful times; they found children of high-stress women had up to 5 points lower IQ.

A review of the literature also finds effects on behavior. 22% of the variance in ADHD, 10-20% of the variance in their development, and an uncertain but significant percent of the variance in other behavioral problems like conduct disorder are attributable to variations in maternal stress during pregnancy. Scientists hypothesize that maybe this has to do with an evolution mechanism located in the hypothalamic-pituitary axis which pre-adapts offspring to stressful conditions while still in the womb.

The stressful events measured in these studies range from personal (separation with a partner) to natural (the ice storm) to political (the September 11 terrorist attacks led to an increase in fetal deaths the appropriate amount of time later, even as far away as California).

The mediating hormone here seems to be cortisol; you can decrease cortisol by pretty much common-sense anti-stress interventions. The Internet recommends low-glycaemic load diets, eliminating caffeine, regular exercise, and meditation/yoga. And stay out of ice storms.

Fish

Some official bodies have said to avoid fish during pregnancy because of worries about mercury. It’s true that mercury is a bad thing and you do not want it in your baby.

But studies (1, 2) show that eating fish leads to better outcomes. Mothers who eat fish in early pregnancy have a vastly reduced risk of hyperactivity, and those who eat fish in late pregnancy had a vastly increase verbal IQ (+7.5 points!). The study claims to have adjusted for confounding factors, but by this point I never believe that sort of thing and probably at least some of the effect is richer people eating more fish or something. Still, that’s just about the largest effect we’ll see in this guide.

How much fish and what kind? One study found benefits from eating more than 340g (=3/4 pound); another found oily fish to be slightly better than non-oily fish. Oily fish that are usually low-mercury include sardines, herring, Atlantic mackerel and my favorite, salmon (wild salmon from Alaska are considered the least polluted). And stay away from the Great Lakes; if the case studies I’ve read are even close to true, anything caught there has enough mercury that you could stick it in glass and call it a thermometer.

The theory is that fish has omega-3 fish oils in it, which are useful in building developing brains. And yet when people try to supplement with fish oil pills directly, it doesn’t work and sometimes even produces worse outcomes thant he control group. Some people suggest it’s other things in the fish like selenium that produce the positive effects, others say the fish oil is more bioavailable when attached to the rest of the fish, and my personal theory is that there is a race of mischevious gremlins who have devoted themselves to hopelessly confusing all medical studies by interfering in people’s bodies and producing bizarre results. Until they go away, stick to real fish.

Licorice

So not only is there a study called Maternal Licorice Consumption and Detrimental Cognitive and Psychiatric Outcomes In Children, not only did it find a significant effect, but it turned out to be one of the most convincing and impressive papers I investigated in this entire guide.

Licorice contains the chemical glycyrrhizin, which aside from being an excellent name for a demon in a fantasy book and possibly the single worst English word in terms of letters before it gets to an unambiguous vowel, inhibits a placental enzyme that deals with cortisol; we already noted that cortisol can be a problem for the developing fetus.

This study was more impressive than some of the others because, although still correlational, it didn’t find any signficant difference between high-licorice consumption and low-licorice consumption mothers, so this is unlikely to be confounded by socioeconomic status in the same way most of the other results are. What it did find was a difference of half a standard deviation on various cognitive tests, which if translated to IQ should mean about 7.5 points. They were also about half a standard deviation worse on various indices of behavioral problems, psychiatric diseases, et cetera.

The worst effects were found in those who ate more than 500 mg of glycyrrhizin/week, which seems to correspond to about 250 grams of licorice which seems to correspond to about fifteen average licorice sticks, raising the question: who the heck eats that much licorice? People with sane amounts of licorice consumption are probably at a lot less risk, but there didn’t seem to be a threshold value and really given how easy it is to abstain from licorice I don’t see why you wouldn’t want to play this one safe.

Pesticides

Worrying about pesticides isn’t just for organic nuts anymore. Several studies have linked prenatal exposure to different classes of pesticides to birth weight and IQ drops during pregnancy; seven points for organophosphates and three points for chlorpyrifos. I haven’t examined the studies in detail, but smart people seem to think they’re worth paying attention to.

You can minimize exposure to pesticides by not spraying pesticides in your own house and yard, obviously. You can avoid hanging out on farms or in heavily agricultural areas. And you can change your eating habits. Washing fruits and vegetables, even before you peel them, can decrease pesticide exposure a little. And although I am generally skeptical of buying organic (and although government anti-pesticide standards are probably more than good enough to meet the safety needs of people outside pregnancy), pregnancy may be the one time when you will get a genuine health advantage from buying organic (though watch out for “organic” pesticides). There’s apparently a Shopper’s Guide To Pesticides In Produce which might be helpful here.

Chemicals

A bunch of chemicals, mostly related to plastics, have vague rumors of deleterious health effects. Polyaromatic hydrocarbons and polychlorinated biphenyl have been linked to 3-5 point decreased IQs in studies, but there’s so much possibility for confoundC-ing that I don’t put a lot of weight behind either.

Probably the most interesting research in this category relates to their role as endocrine disruptors, which cause problems both pre- and post-natally. The theory is that they mimic female sex hormones and prevent men from developing fully masculine characteristics. This definitely happens in animals, and there’s some evidence in humans as well: somebody went through the charming work of measuring the anogenital distance (exactly what it sounds like) in male babies and correlating it with phthalate exposure. Boys with higher exposure to phthalate had a more female pattern — the odds ratio was a very impressive 10.2.

So the obvious question is: ARE OUR PLASTICS MAKING US GAY?!??! We turn to the International Journal of Andrology (apparently you can just start publishing a journal in anything) and particularly the article “Prenatal phthalate exposure and reduced masculine play in boys”. This study asked a series of questions that it double-dog swears are scientifically validated, like “How often did the child play with jewelry in the past month?” and “How often did the child show interest in snakes, spiders, and insects?” and finds that higher urinary levels of phthlate metabolism are linked to more feminine play styles among boys. But the effects were small (something like 8% max) and there’s no evidence they carry on after preschool even if they exist at all.

How do you avoid plastics exposure? If you feed your baby with a plastic bottle, check the plastic composition for phthlates and PCBs. Try not to eat too much canned food during pregnancy; apparently the canning process involves some of these chemicals. Don’t microwave things in plastic wrap.

What if you think boys could probably use a bit more feminization? Presumably you could just eat nothing but food microwaved in plastic wrap throughout your pregnancy, but be warned this could also increase sterility rates later on.

Birth Method

“Babies born naturally may have higher IQ!” touts an article on a study which, when you look at it, measured the levels of a protein in rats.

As for studies that have actually investigated at least one human baby, (1, 2), they find non-significant differences in favor of the Caesarian. Another study that looked for psychosocial differences found none.

The important thing seems to be not to get the baby into a position where it’s stuck and can’t breathe for a while; temporary suffocation doesn’t do the developing brain any favors. Breech presentation may be an example — this is a case in which the child is in the womb in an unusual position. Although many obstetricians say natural childbirth in breech presentation is okay, and although they’re usually right, it was associated with a 7 point IQ drop. If you’re in breech, I would say go for the C-section. If your baby is in perfectly normal position, and your obstetrician says everything is okay and there’s minimal risk of complicated labor, then normal vaginal delivery will probably be fine for your baby.

But why limit your concern to the next generation? Keep getting those C-sections, and your great-great-great-great grandkids will be super-babies!

AFTER THE BIRTH OF A CHILD

It is a little-known fact that some parenting also takes place after the birth of a child. Although this is no doubt the least interesting stage of raising a child, and although it is tempting to leave the child to be raised by wolves, case studies suggest this leads to generally poor outcomes. Arguments that this is due to confounders (women of lower socioeconomic status more likely to live in wolf-inhabited areas) are unconvincing. Despite the lack of a proper randomized trial, the truly determined parent may wish to stick around for one, two, or even eighteen years after their child is born.

Location, Part I

Different countries have different average IQs, all the way from Equatorial Guinea at IQ 59 to South Korea at IQ 106. US states also vary, though not as widely: one list ranges from 104 (Massachussetts) to 94 (Mississippi).

What produces this variation? It’s tempting to answer migration — smart people go to interesting places like Harvard and MIT, and stupid people go to trailer parks in the Deep South. This is part of the picture, but it doesn’t explain the country data and even among natives a difference remains. Racists will no doubt have a field day with this data, but in fact it holds true within races: white/black/Asian people in a high IQ area are to some degree smarter than white/black/Asian people from a low IQ area.

One interesting observation is that with a few exceptions, places at more extreme latitude have higher IQs. Again, useful for the racists. But epidemiologists have a better explanation: parasite load. Tropical jungles are crawling with parasites. The North Pole generally isn’t. Sure enough, when they run the studies, parasite-infestedness of an area correlates with national IQ at about r = -0.82. The same is true of US states, with a slightly reduced correlation coefficient of -0.67 (p<0.0001). The effect was found to do better than other things one might naively expect it to be a proxy for, like temperature, latitude, evolutionary novelty of environment, et cetera. It’s pretty robust. And it holds up in quasi-experiments. When an area eliminates parasites (like the US did for malaria and hookworm in the early 1900s) the IQ for the area goes up at about the right time.

The theory makes biological sense. The more infections a person is fighting off, even micro-infections that never make it to obvious symptoms or require a doctor’s visit, the more stress the body is experiencing and the less resources it has for brain development. This could happen either on the individual post hoc level (wasted all my resources, no chance to build a strong brain) or the genetic pre hoc level (constant failure to build a strong brain because of parasites selects for genes that don’t waste energy trying). Although the racists will no doubt prefer the latter, the former must also be at least some of the story to explain the distribution of IQ in US states, where there wasn’t enough time for evolutionary adaptations. But that means that the effect is potentially causal; you can raise a child’s IQ merely by raising her in a different state.

What states have the lowest parasite load? According to the Parasite-Stress USA index, the ten top places to be a child (and ten bottom places to be a hookworm) are from best to worst Maine, New Hampshire, West Virginia, Vermont, Wyoming, Idaho, North Dakota, Montana, Utah, Oregon. I’m getting a “northern and rural” vibe from this, so if you’re outside the States the northern and rural parts of your own country ought to work. The ten worst states are New York, Illinois, Tennessee, Delaware, Maryland, Alabama, Georgia, South Carolina, Louisiana, and, in last place, poor Mississippi again. Other states that may be of interest to readers are (higher rankings are less parasite-infested) Massachussetts: 13, California: 27, Florida: 34, Texas: 38. Within each state, expect northern, more rural, and drier areas to be better than southern, more urban, and swampier areas.

This suggests the hypothesis that home-schooled children should be smarter than public-schooled children because they’re not getting exposed to the latest contagious infection every month, but no one’s been brave enough to face the confounders an attempt to measure that would throw up, let alone the political firestorm.

Location, Part II

Urban areas have higher IQs than rural areas, but the effect has been decreasing in the past few years. Aside from the obvious confounders like selective migration, the main mechanism people invoke to explain this is cognitive stimulation: cities are just more interesting. By this point you will not be surprised to hear this is another proposed explanation for the Flynn Effect (although weirdly, it doesn’t seem to apply to Asians?).

This seems contradictory to the last section, which found the highest IQs in rural states. I’m not really sure how to resolve this, but by the time you read this I predict the rural-urban IQ gap will have shrunk close to zero anyway.

Breastfeeding

Ten years or so ago everyone was certain that breastfeeding raised IQ. Then a meta-analysis said that was totally false and it was all just confounding because smart conscientious mothers were more likely to take medical advice and breastfeed their kids, and intelligence and conscientiousness are heritable. A group of geneticists fired back with a study saying that okay, only a certain genotype benefitted from breastfeeding. Although the media and the establishment took them seriously, the Internet, which is way ahead of everyone else these days, immediately tore the study apart and was vindicated when the study completely failed to replicate. So for a while the best standard of thinking on this matter (as represented by statistics blogs) was that the breastfeeding-IQ link was a sham. Then word came in from Belarus that they had done a randomized controlled trial there and it had come out positive. Really positive! Six IQ points positive! So for now we go with the Belarussian study and figure it does improve IQ after all.

Even if you don’t want to believe a bunch of dirty Commies, breastfeeding increases growth and immune function, and if you remember back a section anything that boosts immune function should probably be good for the brain as well.

Fluoride

Speaking of not trusting dirty Communists, we are totally going here.

A Chinese meta-analysis finds that levels of fluoride in the drinking water are adversely correlated with childrens’ IQs, sometimes to relatively high levels (5-10 points). The levels they look at, which are common in certain parts of China, are about 3 - 8 mg/L, which they compare to low fluoride areas of China which have only 0.4 mg/L. A Harvard meta-analysis seems to broadly agree.

There is of course a very vocal anti-fluoride community in the United States, usually dismissed as crackpots. They cited this study as support for their position. Their opponents, mostly respectable scientists, said that the Chinese levels were ten times higher than in the US and there was no proof that US levels had harmful effects.

Fluoride levels in the US actually vary from city to city, from as low as undetectable to as high as 1.5 mg/L. So the high in the US is about half the level that was observed to definitely cause problems in China.

This seems really sketchy to me. Suppose we saw someone drink a mysterious black oily liquid out of a bottle and then fall over dead. Then we saw someone take an identical bottle and give it to her child. “Stop!” we say, “that’s poisonous.” “No,” she tells you. “I’m only going to have my child drink half as much, and we have no evidence that it’s poisonous at half the dose.” I mean, I understand some things have threshold values below which they’re not harmful, but I can’t find any evidence that fluoride has ever been evaluated for this.

Worse, a study from Inner Mongolia, an area known for the high quality of its trustworthy science, seems to show a linear dose-response relationship with fluoride all the way down to typical American levels. They suggest a decrease of 0.6 IQ points per mg/L of urinary fluoride. Some of the Chinese studies looked like the effect was slightly worse than this, but not enough to worry too much over.

Now all we’ve got to do is correlate drinking water fluoride with urinary fluoride. It looks like (reasonably enough) after a while these numbers become equal. So if US cities have up to 1.5 mg/L fluoride in the water supply, that costs their populace about 1 IQ point. And this is a worst-case scenario; most cities have nowhere near 1.5 mg/L.

This hardly seems worth worrying about, but if you’re going to worry, you can move to one of the many cities with undetectable levels of fluoride in its water. Portland, OR, for example, is in a nice northern low-parasite location.

Feeding Schedule

People used to believe you shouldn’t feed a baby every time it cried for food. You should just let it cry sometimes, to help it learn how the world was an amoral, Lovecraftian place that cared nothing for its suffering or pitiful human needs (except I think the actual recommendations used words more like “delayed gratification”).

Turns out maybe this wasn’t such a good idea! Babies who were fed “on demand”, ie whenever they cried for food, had four points higher IQ. This produced the normal stream of complaints that it was confounded (maybe upper class mothers who aren’t working long hours have more opportunities to feed their baby whenever it needs food) and the normal stream of protests that they had totally adjusted for all the possible confounders. I don’t even know at this point.

Lead

I hate statements like “It is impossible to overstate the importance of lead poisoning on a population basis”. It is totally possible to overstate the importance of lead poisoning on a population basis. If I said that lead poisoning caused the September 11 terrorist attacks and the fall of the Soviet Union, that would be overstating its importance (terrorism, you’ll recall, is more linked to iodine deficiency, and lead did not cause the fall of the USSR although some historians do speculate it caused the fall of the Roman Empire)

Aside from that, though, it’s pretty hard to overstate the importance of lead poisoning on a population basis. During the bad old days of leaded gasoline, lead probably took a few IQ points off of nearly everyone, and its removal is probably a major contributor to the Flynn Effect. Violent crime has dropped about 40% in the US in the past two decades, and several people have pointed out this lines up suspiciously well with the corresponding drop in lead exposure (lead is known to cause serious behavioral problems and impair executive function). Economists estimate that having decreased lead levels saves us up to $300 billion per generation.

What’s in it for you? The lower your child’s lead exposure, the (slightly) smarter (and less likely to be a criminal, and so on) they will be. Only 2% of US children have lead levels about 10 ug/dl (officially considered the “safe limit”), but studies have found that really there is no safe limit and the less lead you have the better. The average child has lead levels of about 2 ug/dl. IQ decreases 0.75 points for each extra ug/dl of lead in people with low-level exposure, which suggests that by eliminating lead you could give the poor kids above the safe limit an extra 7.5 points and the average kid an extra 1.5 point. These numbers may be underestimates, as Koller et al note a study suggesting “a nonlinear relationship between children’s IQ scores and their blood lead concentration, with larger associations at lower lead concentrations.” So getting rid of those last 2 ug/dl might be worth a little more than 1.5 points after all.

How can you decrease childhood lead exposure? The most important factor is not to live in a house with lead paint. Government regulations forbid modern paints from containing the substance, but many older houses (CDC says 24 million, which sounds like maybe 20% of houses) continue to have lead in them. Try not to live there. If you do, try to avoid letting your child eat paint. Actually, that just seems like a good idea regardless.

Lead in soil is the other big source. The CDC suggests trying to avoid having your child contact soil entirely, but that sounds more like a ridiculous taboo that applies to the son of the God-Emperor than like reasonable advice to a normal parent. Saner advice would be to avoid tracking dirt into the house (someone once claimed to me in all seriousness that the reason Asian people are so smart is that they take off their shoes before entering the house and so avoid tracking lead-filled soil in where their children can breathe it in; this seems implausible but I like the way this person thinks). A randomized controlled trial of dust control unfortunately failed to find any effect.

Also, Time Magazine says that eating breakfast leads to lower lead levels, and if we can’t trust Time, who can we trust?

Multivitamins

What the heck? I was totally on board with the latest run of studies that showed multivitamins were totally useless, when along come twelve randomized controlled studies that all agree on a modest 2.5 IQ point gain from giving multivitamins to schoolkids.

On further analysis it’s not so simple. Most of the kids didn’t gain at all, while about a fifth had spectacular gains; presumably this fifth had poor diets to begin with. So it could just prove that kids with vitamin deficiencies benefit from vitamins; one assumes readers of this guide are smart enough not to let their children become vitamin-deficient in the first place?

But as far as I can tell the vitamins can’t hurt (much) and are pretty cheap, so I would say go for it.

Exercise

We know that exercise produces long-term neurological changes and better cognitive performance in mice. We know that exercise looks correlated with similar neurological changes in children. And we know that higher physical fitness is correlated with higher IQ in Swedish military recruits (unfortunately this study reported its results in a weird way that I couldn’t follow and I wasn’t able to get an exact number of points from it).

These don’t exactly add up to a randomized trial, but they certainly waggle their finger suggestively. And an actual randomized trial in Georgia found a 3.8 point IQ gain by making overweight children exercise a bit. A lot of these studies use 40 minutes a day, and they almost all agree it’s aerobic exercise rather than muscle training that provides the benefits.

It’s all fun and games telling your kid to exercise more, but some sources say that the mother should exercise when she’s pregnant to boost IQ. Thus far the only source I have been able to find on this is a book by psychologist Richard Nisbett, who promises a 7 point IQ gain. He seems like a credible guy, but I can’t find any evidence that he’s basing this off of any particular study. Until I know his evidence (anyone have this book?) I can’t in good conscience endorse it, and you may continue inflicting exercise on your children while remaining blissfully sedentary yourself.

Diet

Save the most infuriating for last. The chance that any associations between diet and IQ are anything other than confounding is slim enough to be a ballerina, and everyone has their own horrible agenda. That having been said, a study of 7000 children in Britain found that a diet of home-prepared vegetables, fruit, cheese, and legumes was associated with 2 points higher IQ; too much junk food was associated with 2 points lower. Another study found a similar effect: “processed” high-fat high-sugar food was associated with a 1.6 point IQ drop at age 8.5.

A thousand such correlational studies aren’t worth one experiment, and there is exactly one experiment in this area: a randomized trial in Kenya found children who ate extra meat did better both physically and mentally than those who ate a calorically equal amount of extra vegetables or extra milk. While I am impressed by the randomized nature of the trial, it’s pretty hard to extend to First Worlders who are probably eating more nutrients and protein already; it may provide slight evidence that complete vegetarianism isn’t a good idea?

And don’t forget about salt: iodized salt remains useful in keeping mental performance sharp even after birth.

So the best we can do in this area is say about what you expected: not too much fat or sugar, meat might be helpful, and iodize your salt.

SUMMARY

Below are all the interventions considered, my probability that the effect is real, and the number of IQ points gained. “Probability it works” is my totally made up wild guess as to how likely it is to be a real causal effect. “Probability it applies” is my totally made up wild guess as to how likely it is an average reader can benefit from it. For example, licorice has a low probability of application, because you probably weren’t planning to eat a ton of licorice anyway. Some things have partial probabilities of application; you may get a little benefit from changing your location if you had been planning on living in California, but not the full benefit you would get if you had been planning on living in Mississippi.

The best case scenario — in which every single study I cited was the gospel truth, someone reading this had been planning to do everything wrong, and now they are planning to do everything right — nets us an extra 90 IQ points, meaning that it would be enough to bring an average IQ 100 child (which of course the child who did everything wrong would not have been) to an IQ 190 child, which is around the level historians speculate Isaac Netwon could’ve been at.

The slightly more realistic scenario — in which studies may be wrong (heavens forbid!) but someone still switches from doing everything wrong to everything right is +46.5, which is around the level that changes the intelligence of the average member of the population into the intelligence of the average Nobel Prize winner.

The situation most relevant to you is what an average person who is not doing ridiculous dangerous things like eating truckloads of licorice can get out of these interventions, which is the sum of (the gain from each intervention * the probability it works * the probability it applies to you) over all interventions. Based on my completely made up numbers, this is 17 points, which is still enough to go from the intelligence of the average member of the population to the intelligence of the average college professor, which some people would classify as an improvement.

And which is still a ridiculous overestimate of actual gains. Most likely, as mentioned above, these factors interact. Imagine a car that can go 70 miles an hour. We observe that breaking the engine decreases its speed by 70 miles an hour, that breaking the wheels decreases its speed by 70 miles an hour, and that breaking the battery decreases its speed by 70 miles an hour. By summing these up, we conclude that by fixing the battery, engine, and wheels, we could get its speed to 210 miles an hour! This is about the same reasoning we’re working with here.

Nevertheless, this is the best we can do by purely biological means until we finally get our act together and invent genetic engineering. I note that there is probably the potential for much more effective intervention through different forms of education — look up John Stuart Mill’s early life if you don’t believe me.

I welcome suggestions for additions, subtractions, different probability estimates on the various methods, and comments.

EDIT 1: I originally said that “no doubt” most of the non-genetic differences between children were social, but one of xuenay’s Facebook friends cites a paper that does throw doubt upon that conclusion. I’ve changed it to no doubt some of the differences are social, which seems properly weasel-ish until I can investigate further.

EDIT 2: Apparently saying like EVERY OTHER LINE that a lot of these studies are questionable and that the effects even if they exist are non-additive wasn’t enough to prevent people from complaining that I didn’t realize that a lot of these studies are questionable and the effects even if they exist are non-additive. So I’ll try one more time: a lot of these studies are questionable and the effects even if they exist are non-additive.