A post going around claims scientists have discovered "when autism and schizophrenia actually begin". The headline oversells it, but the study behind it is real and worth understanding. It was published in Cell Genomics in September 2025 by a team at the University of Exeter, and it does something nobody had done at this scale: follow the chemical switches on human DNA across an entire lifetime of brains.
Switches on the DNA
Every cell in your body carries the same DNA. What makes a neuron a neuron is which genes are switched on and off. One of the main switches is DNA methylation: small chemical tags attached to the DNA that turn nearby genes up or down without changing the sequence itself. This is the core of what people call epigenetics, which we introduced in epigenetics and mental health.
The Exeter team measured methylation in the cortex of nearly 1,000 donated human brains, the youngest just six weeks after conception, the oldest 108 years old. They also separated neurons from other brain cells, because a tag that matters in a neuron may mean nothing in an astrocyte.
Most of It Happens Before Birth
The main result is about timing. More than 50,000 sites on the DNA change dramatically during early and mid pregnancy, and not smoothly: the changes speed up, slow down and level off at specific moments, as the cortex is being built. After that, the pattern is largely set. Most of what is laid down before birth stays stable for the rest of life, and only a very small share of sites keep changing afterwards.
Developing neurons drive most of this. The sites that change in neurons sit in regions of the genome that are active in excitatory neurons; the sites that change in other cells sit in regions tied to astrocytes. The brain's cell types are already writing their own separate instructions very early on.
Where Autism and Schizophrenia Come In
The researchers then asked where the genes already linked to autism and schizophrenia sit in this landscape. The answer: they are significantly more likely than other genes to be next to the sites that change during fetal development, and the link is strongest in developing neurons. In the words of first author Alice Franklin: "By analysing how chemical changes to DNA shape the brain across the human lifespan, we've uncovered important clues about why neurodevelopmental conditions like autism and schizophrenia may develop."
This fits a great deal of earlier evidence. Autism is visible in behaviour from the second year of life, and schizophrenia usually appears in late adolescence, but the genetics of both point to fetal brain development. It also fits what we described in autism, schizophrenia and cognitive evolution: two conditions with very different presentations and partly shared roots.
What This Does Not Say
It does not say what causes autism or schizophrenia. The study maps when the relevant genes are being regulated in a typical brain. It did not compare autistic and non-autistic brains, and it did not identify a trigger.
It does not blame pregnancy. Nothing here shows that something a mother ate, felt or did changed these marks. Most of this prenatal program is driven by the genome itself running its developmental schedule. Heritability estimates for autism are around 80%, which leaves little room for the everyday guilt that headlines like this can feed.
It is not a test. You cannot measure these marks in a living person's cortex, and blood methylation does not mirror the brain's.
The authors list their own limits: little tissue from late pregnancy, a technology that reads only a fraction of all methylation sites, and no way to tell methylation apart from a related mark, hydroxymethylation.
Why It Still Matters
For a long time the debate was framed as genes versus upbringing. This kind of work moves it somewhere more useful: a developmental program, mostly genetic, that starts before birth and unfolds for decades. That is also why autism and schizophrenia are not caused by parenting, vaccines or screens, and why the rise in diagnoses is better explained by who gets noticed, as we show in why so many people are diagnosed today.
It also explains something practical: these are lifelong ways of being wired, not phases. If you recognise yourself in the descriptions, a validated screener is a reasonable first step: the AQ-10 for autistic traits, or the PQ-16 if your concern is unusual perceptual experiences. Neither is a diagnosis. Both give you something concrete to bring to a professional.
Source: Franklin A, Mill J, et al. Cell-type-specific DNA methylation dynamics in the prenatal and postnatal human cortex. Cell Genomics, 24 September 2025. University of Exeter. Funded by the Simons Foundation Autism Research Initiative, the Medical Research Council, the EU Horizon Programme, Wellcome and the NIHR Exeter Biomedical Research Centre. Figures and quotations are from the article and the university's release; the explanations in plain language are ours. This article is educational and is not a diagnosis.