The recent Princeton autism study has been lauded as a “major autism breakthrough” in the UK press. But for many autistic people, it feels underwhelming, to say the least. The concerns and debates it has raised for the autistic community are nothing new.
Autism is undoubtedly complex, and growing up undiagnosed, I wish in retrospect that I had more understanding of my differences at the time. Why did I struggle to make friends? Why couldn’t I cope with mainstream school? Why was it so hard to concentrate? Why did I hate change and bright lights and loud noises?
Naturally, it’s vital for children and their families to understand their autistic experiences. Diagnosis (or self-diagnosis) and autism research can provide answers, so autistic children can embrace who they are and not internalise negative messaging about themselves. It also helps family members to know how to support their loved ones and what to expect from the future.
But there has been a long-standing divide between autism researchers and autistic people. Autism research is still embedded within neurotypical understandings of autism, leading to restrictive and unhelpful interpretations that can demonise and cut off autistic children from support.
That’s not to mention the countless ways autistic people are divided and parcelled up into strict categories: “levels of autism” as well as “functioning labels”. Whilst these can have practical purposes, they also hold autistic children and adults back in many respects.
So, what is the Princeton autism study, why does it matter for autistic children, and why shouldn’t we divide the autistic community further?
What is the Princeton autism study?
Published in Nature Genetics on 9th July 2025, the Princeton autism study has already been accessed over 190,000 times. The study involved researchers from both Princeton University and the Simons Foundation, and used data on 5,000+ autistic children and teens gathered from an autism cohort study called “SPARK”.
Analysing this data through a computer model, and a variety of statistical tests, researchers split the children into four autistic subtypes. They argued that these four types are distinct in their genetic, clinical, and developmental presentations. In other words, each subtype was characterised by the same genetic patterns, autistic traits, and child development.
These four subtypes are:
- Social and behavioural challenges: This autistic subtype covered about 37% of children studied. It is characterised by “social difficulties” and “repetitive behaviours”, but without developmental delay. Children in this group also typically had co-occurring conditions, such as ADHD, anxiety, or OCD.
- Mixed ASD with developmental delay: This subtype applied to around 19% of the children. These participants typically had some developmental delay in milestones, such as talking, and their communication and “repetitive behaviours” showed a level of variation. They typically did not have co-occurring conditions, like anxiety.
- Moderate challenges: This subtype, which was applied to approximately 34% of participants, is characterised by reduced “difficulties” in socialising and “restrictive behaviours”, but still at a higher level than non-autistic children. These participants usually did not have developmental delay or co-occurring conditions.
- Broadly affected: This subtype applied to 10% of participants, and featured more intense “difficulties” in core autism traits, along with developmental delay and co-occurring conditions.
What are the potential benefits of the Princeton autism study?
While there are several criticisms that I want to make of this study, I think it’s important to explore the positives too. There are two key positives that stand out:
- Autistic genetics: The study provides additional evidence for the genetics of autism. In consideration of the harmful remarks made by RFK Jr. regarding “finding the cause” of autism, studies of this kind may be helpful to dispel conspiracy theories about autism.
- Tailored support: Some autistic people and their families feel the study could be helpful in encouraging more tailored support and diagnosis, particularly for those with higher support needs. I’ll explore the concerns I have about this further down, but I can also understand why some feel positive about this study for this reason.
Why the Princeton autism subtypes could be problematic for autistic people
For many in the autistic community (myself included), the Princeton autism study holds more problematic elements than it does positives.
So, what exactly are the problems with this study and its potential impact on the autistic community? Let’s explore each in more detail.
The study excludes certain autistic groups
One of the biggest criticisms that the study has received is its limited sample. Although the sample is relatively large for autism studies, it is still not reflective of the autistic population. The sample was 77% white and about 80% male.
Unfortunately, women and non-white autistic people are less well-represented in autism research. This means that the data will be less applicable to them. Don’t these autistic children matter too?
This issue is further compounded by the exclusion of self-diagnosed autistic people. To some extent, this is natural with a study of this kind. For one thing, it focuses on children, and adults are more likely to be self-diagnosed. Secondly, the researchers may wish to improve the accuracy of the results by excluding self-diagnosis. Sadly, this can have the opposite effect.
Late-diagnosed or self-diagnosed autistics are those who will have been missed at a younger age. Excluding them means that autism data will continue to exclude autistic children who have not received an official diagnosis during childhood. Just because there’s no diagnosis doesn’t mean they’re not autistic!
Deficit-focused language
Another issue with the Princeton study is its perpetuation of deficit narratives of autism. This presents autistic traits as “deficits”, positioning them as inherently negative. Yet, as any autistic person will know, there are both positives and negatives to autism.
For example, consider the “social and behavioural challenges” subtype. This presents a very subjective, negative view of two key autistic traits – something which is sadly common in autism research. Our communication differences are referred to as “social challenges”. But whose challenges are they?
Communication is a two-way street, but time and time again, autistic people are left to shoulder the blame of inter-neurotype social mistakes. Yet, research has repeatedly shown the existence of a “double empathy” issue, where both neurotypicals and neurodivergents struggle to understand one another.
Similarly, “repetitive behaviours” is a very negative way to interpret stimming and special interests! The study groups autistic people by negatives – and interprets our positive experiences from a negative lens.
The study uses an outdated, deficit-focused methodology
The deficit-focused language hasn’t appeared from nowhere. It is embedded in the diagnostic questionnaires and previous literature used to inform the study and its methods. The findings generated by the study will only be as useful as the data fed in – and this study uses incomplete, deficit-focused data.
The “phenotype features” recorded in the study are based on questionnaires such as the Social Communication Questionnaire-Lifetime (SCQ), the Repetitive Behavior Scale-Revised (RBS-R)35, and Child Behavior Checklist 6–18 (CBCL).
Current diagnostic tools are outdated and not suited to picking up on autistic traits in girls and minority groups. Beyond this, they are also deficit-focused, honing in on “behaviours” that are interpreted negatively.
What kind of autism subtypes would we discover if we focused on positive autistic traits and experiences? Our intense interests and talents? Our honesty and strong sense of justice? Our meaningful connections with other neurodivergent people? The impact of seeing autistic experiences framed through a deficit-focused lens can have a devastating impact on autistic children, spreading into their adult lives.
The study has autism data blindspots
While reading the study, two aspects stood out to me. One was the claim that participants were tested against their “non-autistic siblings” both genetically and via self-report.
If there’s one thing I agree with, it’s that autism is highly genetic. It absolutely runs in families. So, I am sceptical of the mention of “non-autistic siblings”. Whilst it’s possible for siblings to be neurotypical, I am concerned that this has not been thoroughly explored within the cohort. Again, just because a sibling is not diagnosed does not mean they are not autistic.
While the siblings measured lower on autism questionnaires, as mentioned these are not always geared to all autistic children; and other autistic people may find their traits are expressed much later. Most neurodivergent people I know have neurodivergent siblings, so I do feel concerned that another group of autistics is being overlooked here.
Beyond this, most autistic people I know have co-occurring psychiatric conditions. ADHD, anxiety, depression, OCD, and eating disorders are very common in the autistic population. So, I find it surprising that some groups were reported as having “no co-occurring conditions”. I have rarely met an autistic person who was only autistic.
I feel this issue highlights the importance of consulting with autistic people throughout the methodology process. Autism research requires a nuanced understanding of autism and how it can present through families.
Autism subtypes put us into boxes
Autistic people have always struggled with neat categorisation. Similar to “low” and “high” functioning labels, creating autism “subtypes” divides and carves up autistic experiences in unhelpful ways. For example, the “moderate challenges” subtype could imply that the struggles of this group are “not as bad”. If the subtypes are to be used in diagnostics or funding, it could stand in the way of “moderate” autistic children receiving the support they need.
Many autistic children will not fit into these subtypes. I certainly wouldn’t have! I can see aspects across all these subtypes that would have fitted with my younger self. This type of reductionism slots autistic children into boxes rather than understanding their individual needs. Every child is unique. Some will face immense challenges in one area; and have wonderful skills in another. Their challenges may also fluctuate from one day to the next.
Considering how functioning labels have been used to compartmentalise autistic experiences and support needs, I’m left wondering what these subtypes will mean in practice. Will children receive different support based on the subtype they’re assigned? What about autistic children with spiky profiles and co-occurring differences? What happens to children who don’t align neatly with these four subtypes?
It’s difficult to know how, or whether, the Princeton study will be applied in practice at this stage. The study’s researchers have also stated there may be additional subtypes. However, in my view, more research into subtypes won’t provide the solution. Not when it’s embedded in biased data and ignores autistic perspectives.
The great thing about the autistic “spectrum” is that it acknowledges and embraces the diversity of autistic lives. Do we really want to put autistic children and their needs back into boxes?
The media response can be harmful
It’s not just the research itself that can be a problem. It’s also how it’s interpreted by the media – and in turn, the wider public. There has long been an issue with the “allure” of neuroscience. This field of research is often interpreted as more valuable and trustworthy due to its biological and scientific background.
News outlets often use forceful language to describe study findings, implying that they are set in stone. For example, articles from Scientific American and The Independent already claim that the study has “shown” or “discover[ed]” four autism subtypes. But has it? In reality, the findings will require further replication and development to pinpoint their validity. And, as discussed, there are other problems with potential bias in the methodology.
Yet, many ordinary people will read these articles and think uncritically of this study as irrefutable fact.
Why we shouldn’t divide the autistic community further
Aside from the issues above, it is essential to bring the autistic community together, rather than slot us into categories. Some autistic children are made to feel “incapable” or “disruptive”, affecting their self-esteem. Others appear to be “high-functioning” and so their challenges are ignored. Some enjoy socialising, whilst others prefer their own company.
Autistic children have unique needs and experiences. Labels can be helpful for clinicians in some specific situations, e.g., to help them to quickly understand what a child may need support with.
But, for the most part, they should be avoided. Autistic children and adults may struggle to fit in, so having a supportive, inclusive autistic community that embraces all is incredibly important. Wherever possible, it’s always best to ask autistic people what they need and how they feel about their own lives, needs, and experiences.
Subtypes create strict boundaries for autistic experiences. The spectrum opens up space for all autistics to thrive.
What needs to change in autism research?
The problems associated with the Princeton autism study are not a one-off issue.
Autism research must centre the needs of all autistic people. That means involving autistic people through all stages of the research process.
Ideally, we would see more autistic researchers conducting autism research, and there are several fantastic autistic researchers already making waves – including Steven Kapp and Damian Milton.
Even when studies are conducted by non-autistic researchers, there needs to be a strong autistic presence. Consulting “autistic experts” is not enough – the best autistic experts are autistic people themselves!
It’s also important for autistic adults, particularly late diagnosed adults to be included in studies, wherever possible. These autistic people are more likely to be women, and may have gone under the radar due to their stronger social skills and heightened masking. As explored, this has the potential to improve diagnosis for autistic children who may otherwise be missed, so they can receive support too.
Takeaway
While the Princeton autism study has its positives, this is overshadowed by its potential harm. Autistic children and adults already have enough roadblocks to diagnosis and support.
Separating autistic people into categories, particularly ones that don’t fully correspond to autistic experiences, can only confuse clinicians’ understanding of autism; more confused than it already is.
Autistic children deserve support that is tailored to their unique selves. Sadly, autism research is still determined to put them into neat boxes that just don’t fit.
Milly Allinson is a freelance writer specialising in neurodiversity, using research and her lived experience to write insightful content for SENsational Tutors.
