Hello, and welcome back to The Autistic Writer.
Every so often, a piece of autism research arrives with the suggestion that scientists may finally be closing in on the biological cause of autism. A new study from researchers at the University of California, San Francisco (UCSF) has generated just this kind of coverage, leaving me thinking, “Not this again…”
There are certainly ingredients here for an attention-grabbing story. Scientists have investigated genetic mutations associated with autism, mapped out what those mutations do at the molecular level, and studied their effects in, as the media would have it, tiny lab-grown brains, or as the scientists might frame it, laboratory-grown brain organoids.
The research has also been reported as potentially opening the way for future “treatments” of autism. This particular framing will garner sighs from around the autistic community before we all chant in chorus, You can’t treat autism, it’s not a disease.
Nevertheless, it’s not difficult for the media to produce their usual melodramatic narrative, along the lines of, “Scientists know which genes cause autism, they have reproduced their effects in tiny human brains, and they may now be able to develop drugs to correct the problem“.
The actual research is a bit more complicated than that, though. This is not a case of a sensational claim being based on entirely worthless science. The science itself, viewed purely through the lens of science for its own sake, is interesting.
The study, published in Science on 27 August 2026, represents more than a decade of work. The researchers systematically mapped protein interactions associated with 100 high-confidence autism “risk” genes. They identified more than 1,800 protein interactions, 87% of which had, apparently, not been reported before. They then investigated 54 autism-associated mutations derived from patients. But what, if anything, does this tell us about autism?
Genes Are Only the Beginning
Autism has a substantial genetic component. That statement should no longer be particularly controversial. Even the National Autistic Society recently published a high-profile statement definitively saying autism is genetic.
Researchers have identified large numbers of genetic variants associated with an increased probability of autism, although the genetic picture varies enormously between individuals. Bluntly, genetics is a complex field.
At one end of this picture are rare genetic variants that can have large effects. At the other are common variants, each individually associated with only a very small change in probability, but collectively contributing to the risk of autism in an individual. This multi-gene effect is known as polygenic risk.
The new UCSF research is particularly relevant to the rare genetic variants that tend to have large effects. The university describes the work as being most directly applicable to people with so-called profound autism, many of whom have rare, high-impact mutations in established autism risk genes.
I should point out here that profound autism remains a controversial term, one that is rejected by many in the autistic community, and by many professionals working in the autism field. Profound autism is not, at the time of writing, a formal diagnostic term. The whole issue of severity of autism is a subject I have covered elsewhere on this website, and it would be too much of a rabbit hole for this particular discussion, so I will attempt to stay on point regarding the research.
Genes encode proteins, and proteins are involved in the mechanisms through which cells function. Proteins also interact with other proteins. A mutation can therefore affect much more than the production of one isolated biological element. It can alter a network of interactions, with biological consequences further downstream. This is the area the UCSF researchers have been exploring.
Instead of simply compiling a longer list of genes associated with autism, they investigated the proteins associated with those genes and the ways in which those proteins interact. They then examined how particular mutations change those interactions. In theory, this provides something more substantial than a statistical association between a gene and an autism diagnosis.
The researchers demonstrated that particular mutations alter particular protein interactions. Those alterations can affect cellular processes involved in neural development. Further experiments can then examine the consequences for developing neurons and their activity.
Yes, this is impressive scientific work, but despite the hype, it doesn’t show how autism emerges in an individual. There is a big difference between establishing that something is associated with autism, and discovering what that thing actually does.
From Association to Mechanism
A particular genetic variant might occur more often in autistic people than in non-autistic people. With sufficiently strong evidence, researchers might conclude that the variant contributes to the risk of autism. But it is still only an association. The actual mechanism involved in the causation of autism is a different story.
Researchers can investigate what happens when the mutation is present. Maybe it changes the structure of a protein. Maybe that protein no longer interacts with another protein in the expected or typical way. Perhaps the altered interaction changes the development of a particular type of cell. Those steps can, potentially, be tested experimentally.
This is where the UCSF work moves beyond simple correlations. It would be unfair, perhaps, in this instance, to dismiss the research with the argument that correlation does not equal causation. The scientists are manipulating biological systems and observing what happens. This allows them to make causal claims about particular steps within those systems. The problem comes with the suggestion of those claims being extended further:
Evidence that a mutation causes a change in a protein interaction is not automatically evidence that the protein interaction causes autism.
Similarly, evidence that the resulting molecular change affects developing neurons is not automatically evidence that this neuronal difference produces the characteristics by which we recognise autism in a human being.
Growing Human Neural Tissue in a Laboratory
The term tiny lab-grown brain is typically dramatic journalism, but not a particularly helpful description of what an organoid actually is.
A brain organoid is a collection of cells grown from human stem cells. Under appropriate laboratory conditions, those cells can differentiate into various types of neural cells and organise themselves in ways that reproduce some features of developing human brain tissue. Some features. They are not lab-grown human brains.
Nevertheless, this is pretty amazing biotechnology. Researchers can observe aspects of early human neural development in living human-derived tissue, rather than relying entirely on animal models. It also gives researchers experimental control…
They can grow tissue containing a particular genetic mutation and compare it with tissue that does not contain that mutation. They can examine the cells, investigate how they develop, measure electrical activity, and observe what happens when a particular part of the biological system is altered. As stunning as that biotech is, we must remember that these organoids are not human brains.
The organoids lack the complete architecture of a human brain, its full collection of specialised regions, its extensive connectivity, its vascular system, and the unbelievably complex biological environment provided by the rest of a developing human body. The researchers have not created an autistic brain in a petri dish.
The FOXP Example
One particularly interesting finding from the study concerns FOXP proteins.
The researchers examined autism-associated mutations involving FOXP1 and FOXP2. Different mutations were found to converge on disruption of the same interaction between FOXP1 and FOXP4. In experimental models, the consequences included changes in neuronal development and increased neural-circuit excitability in brain organoids.
Remember: a genetic mutation affects a protein. The mutation changes the way that protein interacts with another protein. This affects processes involved in neural development. Researchers can then observe measurable differences in the developing neural tissue.
A catalogue of genes associated with autism arguably tells us something about the genetic architecture of autism, but it doesn’t demonstrate how those genes influence development. But mapping the molecular consequences is a start.
There is still a further step required, though. The observation that a mutation produces a particular change in neural development does not demonstrate that the observed neural change is the cause of autism.
Demonstrating that would require a clear connection between the molecular and cellular mechanisms and the complex characteristics of an actual autistic human being.
Where Does Autism Enter the Chain?
Consider this syllogism:
P1: A mutation is associated with autism.
P2: The mutation changes a protein interaction.
P3: The altered interaction changes neural development.
C: Therefore, the mutation causes autism.
The three premises may be supported by strong evidence, but the conclusion does not follow; the logic is not valid.
There is no laboratory measurement of autism inside an organoid. Researchers can measure the development of cells, the expression of genes, protein interactions, electrical activity, and numerous other biological properties. Autism, however, is identified at the level of a human being, through characteristic patterns of development, communication, behaviour, sensory experience, and interaction with the world. The movement from one level to the other requires inference.
This does not make the research entirely worthless. Scientific models are built precisely because many important phenomena cannot be manipulated directly. We cannot experimentally alter a gene in a developing human embryo, wait for the resulting person to grow up, and see whether that person becomes autistic. Quite rightly, no ethical research programme could do that. So, we use models to allow scientists to investigate the individual parts of a system.
The problems start when findings at one level of explanation are treated as though they have automatically answered questions at another level; like studies of organoids being represented as studies that describe people.
A mutation may cause a molecular change. The molecular change may cause an alteration in neural development. Both claims may be supported experimentally. But the proposition that this neural alteration is the mechanism that produces autism is not proven.
There Is No Single Genetic Autism
Another problem with talking about the genetic cause of autism is that the phrase encourages the unwary to imagine that researchers are looking for one biological mechanism. But the genetic evidence just doesn’t support that simple picture.
Autistic people can carry very different combinations of genetic variants. Some people have rare variants with large effects. For many others, it involves the combined influence of very large numbers of common variants with individually tiny effects. Genetics is complicated. Even among the rare high-impact mutations, different genes can be involved.
One of the interesting findings of the UCSF research is that some of this genetic diversity may nevertheless converge further downstream. Different mutations can affect the same protein complexes or molecular pathways. UCSF reports that the genetic changes examined in the study clustered around a much smaller number of shared protein complexes.
If hundreds of different genetic routes ultimately only affect a smaller collection of biological processes, researchers do not necessarily need a model for every genetic mutation.
From Mechanisms to Medicines
The UCSF researchers hope that identifying shared molecular pathways could eventually reveal targets for precision so-called therapies. The idea is that if several different high-impact mutations disrupt the same molecular interaction, it might be possible to develop a drug that acts on the shared downstream mechanism, rather than designing a separate treatment for every genetic variant. UCSF has received substantial funding to pursue the therapeutic potential of this programme.
Identifying a molecular interaction that might be modified by a drug is only the beginning of a very long process. Researchers would need to establish that the relevant mechanism operates in living human beings in the way predicted by the models. They would need to discover an intervention capable of altering it, determine whether that intervention can reach the relevant tissue, establish its safety, investigate the importance of developmental timing, and eventually demonstrate meaningful benefits in clinical trials.
However, autism has the additional complication that many of the biological processes being investigated occur during prenatal or very early brain development.
If a genetic mutation changes the developmental trajectory of neural tissue before birth, discovering that mechanism does not necessarily mean that altering the same pathway years later will reverse its consequences. But there is a more important concern than all this, and it’s a question that science simply cannot answer…
Treatment for What?
The language used by researchers involved in this work includes terms such as pathology, disease, harmful function, and abnormality. Within biomedical research, some of this language has established technical uses. But it carries certain assumptions when applied to autism as a whole.
This kind of autism research has been shaped by a medical model that approaches autistic traits as problems to identify, prevent, or correct. When those aims are taken for granted, researchers can overlook neurodiversity’s ethical questions: who decides what counts as a problem, whose interests an intervention serves, and whether its goal is to improve an autistic person’s life or just make them appear less autistic.
Sure, a laboratory can establish that cells carrying a mutation develop differently from control cells. They can measure that difference and investigate its consequences. But the experiment cannot, by itself, determine whether every consequence of that difference should be prevented or eliminated.
This becomes especially relevant when research concerning very specific, rare, high-impact mutations associated with severe disability is discussed under the much broader heading of autism.
Some autistic people have intellectual disability, epilepsy, severe communication difficulties, motor problems, and other conditions that can have profound effects on their quality of life. Some do not. Research capable of mitigating epilepsy, preventing dangerous neurological complications, or alleviating other disabling effects is obviously important, but it is not a treatment or therapy for autism.
Yes, we can investigate neurodevelopmental biology seriously, but we need to be careful about what counts as a desirable therapeutic outcome.
The Problem With the Breakthrough Story
Let’s be clear: scientists have not discovered “the cause of autism.”
What they have done is develop a much more detailed map connecting particular autism-associated genetic mutations with the proteins those genes encode, the interactions between those proteins, and some of the downstream consequences for developing neural cells. That is some pretty serious and impressive science; it doesn’t need anyone hyping it further.
My Thoughts
This is not a study showing that scientists have discovered what causes autism. But it isn’t another meaningless genetic correlation dressed up as a breakthrough, either.
Researchers already had strong evidence implicating particular genetic mutations in some forms of autism. They have now made considerable progress in understanding what some of those mutations actually do.
Understanding how a genetic mutation changes a protein, a cell, or a developing neural circuit is not the same thing as understanding how those changes become the complex collection of characteristics that we call autism in a human being. But if it can lead to treatments of, for example, types of epilepsy, or other debilitating neurodevelopmental disabilities, that’s encouraging… but only if the research is channelled that way, rather than pandering to a misguided, fear-driven public clamour for a “cure for autism.”
That’s all for this time. Take care.
share this article:
The Autistic Writer is a labour of love, dedicated to promoting autism acceptance, understanding, and equality. If you’d like to help support this work, a small tip via Buy Me a Coffee would be greatly appreciated. Just click the button, below.
If you like this blog, I can also highly recommend the brilliant Mortgage Advisor on FIRE, which covers financial wellbeing, mental health, and autism issues, among much more.
