What does “Heritable” mean? 

“Heritable” (or “heritability”) has a specific and unfortunately non-obvious definition in genetics.

The word sounds like a synonym for “inheritable,” rather like your grandmother’s collection of musical clocks. Musical clocks are inheritable; fruit, since it rots, is not very inheritable.

This is not what “heritable” means.

“Heritability,” in genetics, is a measure of the percent of phenotypic variation within a population that can be attributed to genetics.

Let me clarify that in normal speak. “Phenotype” is something you can actually see about an organism, like how tall it is or the nest it builds. “Phenotypic variation” means things like “variation in height” or “variation in nest size.”

Let’s suppose we have two varieties of corn: a giant strain and a dwarf strain. If we plant them in a 100% even field with the same nutrients, water, sunlight, etc at every point in the field, then close to 100% of the variation in the resulting corn plants is genetic (some is just random chance, of course.)

In this population, then, height is nearly 100% heritable.

Let’s repeat the experiment, but this time, we sow our corn in an irregular field. Some patches have good soil; some have bad. Some spots are too dry or too wet. Some are sunny; others shaded. Etc.

Here it gets interesting, because aside from a bit of random chance in the distribution of seeds and environmental response, in most areas of the irregular field, our “tall” corn is still taller than the “short” corn. In the shady areas, both varieties don’t get enough sun, but the tall corn still grows taller. In the nutrient-poor areas, both varieties don’t get enough nutrients, but the tall still grows taller. But when we compare all of the corn all over the field, dwarf corn grown in the best areas grows taller than giant corn grown in the worst areas.

Our analysis of the irregular field leads us to conclude that water, sunlight, nutrients, and genes are all important in determining how tall corn gets.

Height in the irregular field is still heritable–genes are still important–but it is not 100% heritable, because other stuff is important, too.

What does it mean to be 10, 40, or 80% heritable?

If height is 10% heritable, then most of the variety in height you see is due to non-genetic factors, like nutrition. Genes still have an effect–people with tall genes will still, on average, be taller–but environmental effects really dominate–perhaps some people who should have been tall are severely malnourished.

In modern, first world countries, height is about 80% heritable–that is, since most people in first world countries get plenty of food and don’t catch infections that stunt their growth, most of the variation we see is genetic. In some third world countries, however, the heritability of height drops to 65%. These are places where many people do not get the nutrients they need to achieve their full genetic potential.

How do you achieve 0% heritability?

A trait is 0% heritable not if you can’t inherit it, but if genetics explains none of the variation in the sample. Suppose we seeded an irregular field entirely with identical, cloned corn. The height of the resulting corn would would vary from area to area depending on nutrients, sunlight, water, etc. Since the original seeds were 100% genetically identical, all of the variation is environmental. Genes are, of course, important to height–if the relevant genes disappeared from the corn, it would stop growing–but they explain none of the variation in this population.

The heritability of a trait decreases, therefore, as genetic uniformity increases or the environment becomes more unequal. Heritability increases as genetics become more varied or the environment becomes more equal. 

Note that the genes involved do not need to code directly for the trait being measured. The taller people in a population, for example, might have lactase persistence genes, which let them extract more calories from the milk they drink than their neighbors. Or they might be thieves who steal food from their neighbors.

I remember a case where investigators were trying to discover why most of the boys at an orphanage had developed pellagra, then a mystery disease, but some hadn’t. It turns out that the boys who hadn’t developed it were sneaking into the kitchen at night and stealing food.

Pellagra is a nutritional deficiency caused by lack of niacin, aka B3. Poor Southerners used to come down with it from eating diets composed solely of (un-nixtamalized) corn for months on end.

The ultimate cause of pellagra is environmental–lack of niacin–but who comes down with pellagra is at least partially determined by genes, because genes influence your likelihood of eating nothing but corn for 6 months straight. Sociopaths who steal the occasional ham, for example, won’t get pellagra, but sociopaths who get caught and sent to badly run prisons, however, increase their odds of getting it. In general, smart people who work hard and earn lots of money significantly decrease their chance of getting it, but smart black people enslaved against their will are more likely to get it. So pellagra is heritable–even though it is ultimately a nutritional deficiency.

What’s the point of heritability?

If you’re breeding corn (or cattle,) it helps to know whether, given good conditions, you can hope to change a trait. Traits with low heritability even under good conditions simply can’t be affected very much by breeding, while traits with high heritability can.

In humans, heritability helps us seek out the ultimate causes of diseases. On a social level, it can help measure how fair a society is, or whether the things we are doing to try to make society better are actually working.

For example, people would love to find a way to make children smarter. From Baby Einstein to HeadStart, people have tried all sorts of things to raise IQ. But beyond making sure that everyone has enough to eat, no nutrient deficiencies, and some kind of education, few of these interventions seem to make much difference.

Here people usually throw in a clarification about the difference between “shared” and “non-shared” environment. Shared environment is stuff you share with other members of your population, like the house your family lives in or the school you and your classmates attend. Non-shared is basically “random stuff,” like the time you caught meningitis but your twin didn’t.

Like anything controversial, people of course argue about the methodology and mathematics of these studies. They also argue about proximate and ultimate causes, and get caught up matters of cultural variation. For example, is wearing glasses heritable? Some would say that it can’t be, because how can you inherit a gene that somehow codes for possessing a newly invented (on the scale of human evolutionary history) object?

But this is basically a fallacy that stems from mixing up proximate and ultimate causes. Obviously there is no gene that makes a pair of glasses grow out of your head, nor one that makes you feel compelled to go and buy them. It is also obvious that not all human populations throughout history have had glasses. But within a population that does have glasses, the chances of you wearing glasses is strongly predicted by whether or not you are nearsighted, and nearsightedness is a remarkable 91% heritable. 

Of course, some nearsighted people opt to wear contact lenses, which lowers the heritability estimate for glasses, but the signal is still pretty darn strong, since almost no one who doesn’t have vision difficulties wears glasses.

If we expand our sample population to include people who lived before the invention of eyeglasses, or who live in countries where most people are too poor to afford glasses, then our heritability estimate will drop quite a bit. You can’t buy glasses if they don’t exist, after all, no matter how bad your eyesight it. But the fact that glasses themselves are a recent artifact of particular human cultures does not change the fact that, within those populations, wearing glasses is heritable.

“Heritability” does not depend on whether there is (or we know of ) any direct mechanism for a gene to code for the thing under study. It is only a statistical measure of genetic variation that correlates with the visible variation we’re looking at in a population.

I hope this helps.