The first is masking. Autistic girls often have a relatively higher drive for social connection than boys. They watch the people around them, and they're skilled at deliberately learning and imitating social behavior. They hold eye contact by force. They memorize others' speech patterns and expressions like a script, and perform them to look natural. This is camouflaging.

Interests, too, look different on the surface. Where a boy might fixate on train timetables, dinosaurs, or the wheels of a car, a girl might attach obsessively to dolls, a particular person, or animals — subjects that seem ordinary at a glance. The intensity is the same; the object simply draws less social attention, so it's harder for others to notice.

The problem is that this camouflage isn't free. The smoother the surface, the more energy is burning underneath. Just how much sustaining a perfect mask grinds a person down often surfaces only in adulthood — as severe burnout, anxiety, depression. And the earliest diagnostic criteria were built around the traits of boys. So girls and women get overlooked in the clinic, or misdiagnosed as something else entirely.

The second is a little more unfamiliar. From here we'll borrow from a lecture — Professor Joon-Yong An's "Female Autism, Seen Through DNA." It's a picture that came into view once NGS, the technology for reading the whole genome, let us read our DNA in full.

We inherit about three million genetic variants from our parents. On top of that, each of us is born with a few dozen variants that neither parent carries — arising by chance during the cell divisions that make sperm and egg, and written into us alone. These are de novo ("newly arising") variants.

Do autistic people carry more of these de novo variants? Strikingly, by count alone, no. Whether autistic or an undiagnosed sibling, the overall number is much the same. What matters isn't the count — it's which gene the variant lands in.

Synonymous variants, which don't change the protein sequence, show no difference between autistic children and their siblings. But variants that cut protein production short — PTVs, or loss-of-function (LoF) variants — appear markedly more often in autistic people. And the genes these variants land in are ones heavily used by neurons in the developing brain. It's as if the blueprint for building the brain is severed midway, so development takes a serious hit. Indeed, autistic people carrying these damaging variants tend to show more pronounced intellectual impairment or developmental delay.

De novo variants aren't the whole story. The genetic picture of autism has three layers — newly arising damaging variants (de novo), rare variants inherited from parents, and polygenic risk, the small contributions of thousands of common variants adding up. The large study by Antaki and colleagues (2022) that Professor An cites showed that these three combine, adding together, to produce autism risk.

And here an interesting balance appears. Within a single person, these factors trade off against one another. Carry one large damaging de novo variant, and you can cross the threshold with less polygenic load; without that de novo hit, more common variants have to accumulate to get there. There is more than one road to autism.

And then, sex. In Antaki's study, women diagnosed with autism carried significantly more of both rare variants and polygenic load than diagnosed men. It means the genetic threshold required for autism to appear is higher in women — the female protective effect.

There's one more telling detail. In males, the more damaging de novo variants a person carried, the lower their polygenic load — the two trade off. But in females, this trade-off was not clearly seen. For women, neither alone tends to be enough; they carry both kinds of burden to cross the threshold. It's further evidence of just how high that threshold sits.

Because autism is more common in males, an old assumption tags along: "Isn't it an X-chromosome problem?" The intuition is plausible — men have only one X, so they'd be more vulnerable. But Antaki and others say it plainly: the male preponderance of autism is not explained by variation on the sex chromosomes. The real difference lies not in the X, but in how variants on the autosomes — all the chromosomes that aren't the sex chromosomes — act differently in males and females.

To bring it together: masking raises the threshold of detection — it hides the less-visible half. Genetics raises the threshold of expression — women need a heavier genetic load before autism appears at all.

Two thresholds of entirely different character, pointing the same way. Both make women's autism harder to catch. And laid over each other, two facts that seemed to contradict resolve into one picture: women's autism is caught less often, while the women who are diagnosed present more heavily. Because the threshold of expression is higher, those who cross it arrive carrying a larger hit; and because the threshold of detection is higher too, the milder end slips through the net all the more.

So, back to the two boxes we started with — autism is neither gift nor defect, nor a thing that divides into present-or-absent. There is a threshold on a continuum, and its position simply sits a little differently from one sex, and one person, to the next.

From here, this is my own thought. If someday it becomes clear exactly what the female protective effect protects, and how, might that knowledge also offer a thread toward easing the difficulties autistic people face? It's still a distant prospect — closer to a wish.