Where autism comes from biologically
Last reviewed November 16, 2026. OwieClear editorial team. Waiting for clinician review. Not medical advice. Talk with your doctor about your own care.
Autism is among the most heritable of all neurodevelopmental conditions, and it has no single cause. Those two statements are both true and the tension between them is the subject of this page.
What there is instead is a very large number of genetic contributions, a small number of which are individually powerful, converging on how brain cells connect to each other.

Related pages: Autism
The bottom line
- Twin and family studies put heritability at roughly 70 to 90 percent.
- Most of the genetic contribution is polygenic: many common variants, each with a tiny effect.
- A minority of cases involve a single rare variant with a large effect.
- Implicated genes converge on how synapses form and how brain development is regulated.
- Why the same genetic variant produces autism in one person and not another.
- Whether the brain differences described are causes, consequences, or both.
The story, step by step
- 1. Heritability ()
- 2. Polygenic ()
- 3. The rare variants ()
- 4. What the genes do ()
- 5. The brain findings ()
What the evidence labels mean
What the twin studies show
What happens
The genetic contribution to autism is large and has been measured repeatedly.
What the research shows
- Identical twins, who share essentially all their DNA, are far more likely to both be autistic than non-identical twins, who share about half.
- Heritability estimates generally fall between 70 and 90 percent, among the highest for any neurodevelopmental condition.
- Siblings of autistic children have a substantially higher likelihood of being autistic than the general population.
- Heritability is a population statistic, not a prediction for an individual. It describes how much of the variation in a population is attributable to genetic variation.
- A high heritability also does not mean inevitability: identical twins are not always concordant, so other factors matter.
Why it matters
Autism runs in families, and many parents recognize themselves during a child's assessment. That is a common experience rather than a coincidence.
What's still uncertain
- Heritability estimates vary between studies depending on method and on how autism is defined.
Read more
- NIMH: Autism spectrum disorderGovernment
- NICHD: Autism spectrum disorderGovernment
Next: Polygenic
Hundreds of small contributions
What happens
For most autistic people, there is no single gene to find.
What the research shows
- The majority of the genetic contribution comes from common variants, each one shifting the likelihood by a very small amount.
- Hundreds of such variants are involved, and everyone carries some of them. Autistic people, on average, carry more.
- This is the same architecture as height or blood pressure: continuous variation produced by many small effects.
- It explains why autistic traits exist on a continuum in the general population, and why a diagnostic threshold is a line drawn across a gradient.
- It also explains why there is no genetic test for autism and why there is unlikely to be one.
Why it matters
The absence of a genetic test is not a gap waiting to be filled. It follows from how the genetics actually works.
What's still uncertain
- Polygenic scores for autism have very limited predictive value for individuals and are not clinically useful.
Read more
- NIMH: Autism spectrum disorderGovernment
Next: The rare variants
When one change does most of the work
What happens
In roughly 10 to 20 percent of cases, a single rare genetic change makes a large contribution.
What the research shows
- These include de novo variants, changes that appear in the child and are not inherited from either parent, and copy number variants where sections of chromosome are duplicated or deleted.
- Named genes include SHANK3, which builds the scaffolding at synapses, CHD8, which regulates how other genes are switched on during development, and SCN2A, a sodium channel involved in neuronal signalling.
- Some cases occur within a recognized genetic syndrome, most commonly fragile X syndrome, caused by changes in FMR1, and tuberous sclerosis.
- Genetic testing is recommended after an autism diagnosis in many guidelines, mainly to identify syndromes that have their own medical implications and to inform family planning.
- A positive result does not change the autism diagnosis or the support needed. It may change medical monitoring.
Why it matters
Genetic testing after diagnosis is about finding associated medical conditions, not about confirming autism.
What's still uncertain
- Most rare variants identified have incomplete penetrance, so carrying one does not guarantee autism.
Read more
- NICHD: Autism spectrum disorderGovernment
- OwieClear: Fragile X syndromeOwieClear
Next: What the genes do
Converging on the synapse
What happens
The implicated genes are varied, and what they do is less varied than their number suggests.
What the research shows
- Many are involved in building and maintaining synapses, the junctions where neurons communicate.
- Many others are involved in chromatin remodelling, the machinery that controls which genes are switched on at which point in development.
- Others affect how neurons migrate to their correct positions as the brain forms.
- The convergence suggests the relevant process is how brain circuits are assembled and tuned in early development, rather than any single structure.
- Most of this happens before birth and in the first years of life, which is why autism is developmental rather than acquired.
Why it matters
The genes do not code for social behaviour. They code for how a brain wires itself, and the behavioural differences follow from that.
What's still uncertain
- How these cellular differences produce the specific features of autism is not understood in detail.
Read more
- NIMH: Autism spectrum disorderGovernment
Next: The brain findings
Real, and not diagnostic
What happens
Neuroimaging has found consistent group differences, and this is an area where confidence should be calibrated carefully.
What the research shows
- Accelerated brain growth in early childhood has been reported in some autistic children, though the finding has not replicated uniformly and appears to apply to a subset rather than to everyone.
- Differences in functional connectivity have been described, often summarized as more local connectivity and less long-range synchronization. The real picture is more heterogeneous than that summary suggests, and findings vary with age, task and method.
- Differences have been reported in regions involved in social processing, in the cerebellum, and in sensory pathways.
- All of these are group averages with substantial overlap between autistic and non-autistic people. None can diagnose an individual.
- There is no brain scan, blood test or biomarker that identifies autism, and anyone offering one is not practising evidence-based medicine.
Why it matters
Group differences in research do not translate into individual tests, and the gap between those two things is where a great deal of pseudoscience lives.
What's still uncertain
- Much imaging research has been conducted in small samples of young white boys, limiting how far the findings generalize.
Read more
- NIMH: Autism spectrum disorderGovernment
- OwieClear: Where autism comes fromOwieClear
More plain-language reading
- OwieClear: Where autism comes fromOwieClear
- OwieClear: Autism and vaccinesOwieClear
- OwieClear: Co-occurring conditionsOwieClear