Heritability and the Discomfort That Follows It

Behavioural genetics produces findings that are unusually well replicated and unusually unwelcome. That combination makes the field worth examining — not primarily for what it has found, but for what the reaction to it reveals about how we handle evidence when we would rather not have it.

The core finding is not in serious dispute among people who work on it: essentially every measured human psychological trait shows substantial heritability. Twin and adoption studies, across many decades and many countries, converge on this. It is one of the most consistently replicated results in the social sciences, and it is one that a great many people, including me, would have preferred to come out otherwise.

What the Word Actually Means

Almost every popular argument about this topic founders on the definition, so it is worth getting right before anything else.

Heritability is a statistic about variation within a population under particular conditions. If a trait has heritability of 0.5 in some population, that means half the observed variation between people in that population, in that environment, is statistically associated with genetic variation. It does not mean half of any individual's trait is genetic. It does not mean the trait is fixed. It does not mean environment is irrelevant.

Two consequences follow that cut against the usual polemical uses.

Heritability rises when environments become more equal. If everyone has adequate nutrition and schooling, environmental variation shrinks, so a larger share of the remaining differences traces to genes. High heritability of educational attainment in a rich country is partly a measure of how successfully that country equalised opportunity. A finding of high heritability is not evidence that intervention failed; it can be evidence that it worked.

Heritable does not mean unchangeable. Height is strongly heritable, and average height in many countries rose several inches in a century through nutrition. Vision is heritable, and glasses exist. Phenylketonuria is a single-gene disorder whose devastating effects are prevented entirely by diet. The environmental lever does not have to be the same causal channel as the genetic one.

A great deal of the fear surrounding this topic assumes heritability means fixity. It does not, and the conflation does damage in both directions — it makes the finding frightening to those who oppose it and useful to those who abuse it.

The Result Nobody Wanted

Beyond heritability itself, the field produced a second finding that is genuinely strange and, if anything, more uncomfortable.

Shared family environment — the household siblings grow up in, with its books, parenting style and household income — turns out to explain much less of adult personality variation than almost anyone expected. On many measured traits its effect on adult outcomes is small, and on some it approaches zero once genetic relatedness is accounted for. Adoptive siblings raised in the same home end up resembling each other far less than intuition predicts.

This one deserves care. It is measured within the range of environments that appear in these studies — mostly ordinary homes in developed countries — and it says nothing about severe deprivation, abuse or neglect, which have large and well-documented effects. It is a claim about variation among the reasonably-ordinary, not a claim that upbringing does not matter.

But within that range it is robust, and it contradicts a belief that is close to sacred in modern parenting culture: that a child's adult personality is substantially something parents build. The evidence says parents matter enormously for a child's actual life — for their happiness, their safety, their relationship with you, their opportunities — and much less than assumed for the adult traits we tend to measure.

Why the Taboo Exists

The discomfort here is not irrational, and it would be dishonest to treat it as mere squeamishness. It has a specific and serious history.

Claims about biology and human difference were used, within living memory, to justify forced sterilisation, immigration restriction, segregation and genocide. The eugenics movement had scientific respectability and university chairs. Twin research includes work done by Mengele. Anyone who reads that history and then encounters modern behavioural genetics with a sense of alarm is not being stupid; they are pattern-matching on one of the worst records any scientific programme has.

There is also a live concern about use. Findings in this area are reliably distorted the moment they leave the journals, and typically in the direction of justifying existing hierarchies. That is not a hypothetical risk.

So the taboo is not arbitrary. The question is whether it is the right response — and I think it is the wrong tool for a real problem.

Why the Discomfort Is Not an Argument

Here is the difficulty. Every reason above is a reason about consequences of belief, not about whether the belief is true. And that is exactly the move examined elsewhere on this site under the heading of consolation: treating how a claim makes us feel as though it bore on whether the claim is so.

The universe was not consulted about which findings would be socially convenient. If a trait is heritable, it is heritable whether or not that fact is useful to anyone. Suppressing an inquiry does not alter the underlying fact; it only ensures that the people describing it are the ones least interested in describing it carefully.

There is a practical cost too, and it is not small. When mainstream researchers avoid a topic, the vacuum is filled by people with an agenda, and their claims go unrebutted by anyone with standing. A field left to its worst interpreters is not a safe field.

And the moral inference the taboo is protecting against does not actually follow. Suppose every disputed heritability claim were true. Nothing about equal treatment would change, because equal treatment was never premised on empirical sameness. We do not owe people equal standing because they are identical in their capacities; we owe it to them regardless of how they differ. That is the whole content of the commitment. A version of egalitarianism that depends on a particular empirical result is one hostage to the next study — which is precisely the wrong foundation on which to rest something this important.

Where the Real Limits Are

Pushing back on the taboo does not mean accepting every claim made under this banner, and the genuine scientific limits are stricter than the popularisers admit.

Within-group heritability tells you nothing about between-group differences. This is the single most abused point in the entire area. Two plots of genetically identical seed will differ in average height if one is better watered, and the heritability within each plot can be high while the difference between them is entirely environmental. The inference from within-group heritability to between-group causation is invalid, and it is the inference nearly every popular misuse depends on.

The molecular picture is unfinished. Polygenic scores explain far less variance than twin-study heritability implies, and they predict poorly outside the ancestral populations they were trained on. Anyone presenting them as a ready-made measure of individual potential is overselling by a wide margin.

These are population statistics with no individual application. A heritability estimate tells you nothing about any particular person, and cannot support a judgement about one.

Each of these is a place where the science genuinely constrains the conclusions. They are better protection against misuse than a taboo, because they are true, and because they work by being correct rather than by being enforced.

The Tell

Which brings this back to the pattern running through everything else here.

The strength of the reaction to these findings tracks how threatening they are, not how well supported. And the specific shape of the objection — this is dangerous, this must not be said, look what could be done with it — is the same shape as every other objection made to an unwelcome truth, from Copernicus onward. The complaint is about consequence, not about evidence.

The honest position is uncomfortable in both directions, which is usually a sign of being roughly in the right place. The findings are real and were not wished for. Their misuse is also real and was not imagined. The response to a dangerous truth is to state it precisely, including its limits — because a precisely stated finding is far harder to abuse than a suppressed one, and because the alternative has never once worked.