Rippon article

A revoluntionary new understanding of autism in girls By Gina Rippon

Published in New Scientist, April 5th, 2025

In China, it is known as “the lonely disease”. The Japanese term translates as “intentionally shut”. Across the world, there is a perception of autistic people as aloof, socially awkward and isolated, seeming to not only lack the kind of automatic social instinct that enables successful interaction, but also the desire to achieve it. There is also a perception that autistic people tend to be men.

For decades, researchers – myself included – have thought of autism as a predominantly male condition. The more we studied boys and men, the clearer the picture of autism that emerged – or so we thought.

Today, we have come to realise that we were missing a huge piece of the puzzle all along. Not only have we been failing to recognise autism in vast numbers of women and girls – preventing them from getting a diagnosis and support – but we have now made the profound discovery that the female autistic brain works differently than the male one, especially when it comes to social motivations and behaviours. As a result, an entirely new picture of autism in girls is crystallising, forcing a radical rethink of everything we thought we knew.

Autism is a neurodevelopmental condition, commonly diagnosed by the age of 5. Current standard diagnostic criteria refer to “persistent difficulties with social communication and social interaction”, as well as “restricted and repetitive patterns of behaviours, activities or interests to the extent that these limit and impair everyday functioning”.

The World Health Organization estimates that 1 per cent of children worldwide are autistic, but this figure is probably much higher, with estimates that up to 50 per cent of autistic people remain undiagnosed.

The seed of the belief that autism is a male-centric condition was planted quite early on, with Hans Asperger, one of the so-called “fathers of autism”, specifically referring to a form of extreme male intelligence in his case studies (which all looked at boys).

By the 1960s, early prevalence studies were reporting a male-to-female ratio of 4:1, a figure often quoted today on autism websites and in advice manuals, as well as in research papers.

Diagnoses of autism began to soar in the 1980s thanks to a broadening of the definition of the condition, incorporating a much greater range of indicative signs and levels of support needs. And this century, a more widespread recognition and awareness of autism has caused a further surge. Yet all this has only widened the gulf between diagnoses in girls and boys (see “Autism prevalence among 8-year-olds in the US,”), with some reports of male-to-female ratios of 15:1.

The general consensus was that of course there are autistic girls and women, but their social difficulties and repetitive patterns of behaviour are less marked. A fixed diagnostic description of autism had emerged: this is what it is like in boys and men, so this is what autism is like. The belief in the “maleness” of autism also affected the public’s conception of it, often fed by media characterisations such as Dustin Hoffman’s character Raymond in Rain Man.

Traditional view

This traditional view also fed into neuroscientific research into why girls were less likely to be autistic. One idea was a “female protective effect” – suggesting that specific aspects of female biology served as some kind of buffer, requiring a much stronger genetic influence before the condition emerged. In the same vein, a central role for testosterone has been proposed in the development of autistic brains. Indeed, one early neuroscience model of autism proposed that it was causally linked to “an extreme male brain”, firmly based on the accepted maleness of autism.

But it is perhaps in the field of cognitive neuroscience, exploring the links between brain and behaviour, that the male bias has been most evident. Autism clearly runs in families, with estimated heritability ranging from 40 to 90 per cent. Genetic studies strongly indicate multiple differences in brain development between autistic and non- autistic brains, which remain the focus of most research programmes. But there is currently no biomarker for autism, so it is only recognisable behaviourally. This means models of behaviour that most closely match the accepted picture of autism serve as frameworks for brain-imaging studies. Because a core diagnostic characteristic of autism is “persistent difficulties with social communication and social interaction”, a major focus of autism brain-scanning studies has been on the so-called social brain.

The social brain hypothesis has long argued that our brains have evolved to manage social interactions and relationships, enabling us to navigate the complexities of human society. This requires an awareness of not just an individual self, but also other selves, their thoughts, beliefs and intentions, and how these might affect interactions between an individual and other people. This forms the basis of what could be termed a social instinct, which automatically oils the wheels of social interaction. We also have a powerful drive to be social – a need to join or be accepted by social networks – which can underpin many aspects of behaviour.

The human brain is thus equipped with complex networks responsible for the acquisition and retention of social knowledge, such as the recognition of facial expressions or tone of voice. Emotional and motivational processes in the brain then identify the positive or negative aspects of social behaviour. And these, in turn, speak to the action networks, which help us to respond with socially “appropriate” behaviour, for example, following the rules of turn-taking in conversation or respecting personal space.

Once cognitive neuroscientists could model core processes of social behaviour in the brain, the idea was that they could map activity in these social brain systems in a brain scanner and hopefully identify the brain differences that underpinned the social difficulties characteristic of autism.

This offered a fruitful approach for autism neuroscience researchers. Between 1990 and 2020, there were many hundreds of studies comparing patterns of activity and connectivity pathways in social brain networks in both autistic and neurotypical participants. A powerful consensus emerged that, particularly in the social reward networks, there were lower levels of activity and/or reduced connectivity in autistic brains. Such activity was associated with suboptimal social behaviour – poor emotion recognition, for example, or minimal emotional reaction to negative social experiences, such as social rejection. Particular attention was drawn to the striatum, a brain structure heavily involved in reward-processing and reinforcement-learning. The underactivity in this region resonated well with the classic autism phenotype of a socially withdrawn individual, apparently lacking the motivation to make contact with others. Such findings offered the best insights to date into the brain correlates of autism.

But these promising developments were based on highly biased datasets. Last year, I did some detective work and published a review of more than 120 brain-imaging studies specifically looking at social brain differences in autism, revealing that nearly 70 per cent of them tested only male brains (or included just one or two female brains). Of the more than 4000 autistic participants tested overall, fewer than 10 per cent were female. This is more widely reflected in the entire body of autism brain research. A review of 20 years of such work revealed that, of 1428 articles on brain structure and function in autism, almost 30 per cent studied only male brains; in 77 per cent of the remaining studies, sex differences weren’t assessed or sex was treated as irrelevant in the analyses.

Perhaps most concerning of all, very few of the discussions drew attention to the fact that only male brains were being measured. Results were reported in terms of insights into autism or autistic behaviour.

To be fair to the neuroscience community, the failure to include girls and women in research cohorts isn’t just a reflection of what’s called the male-spotlight problem, where a belief in the maleness of autism constrained research design. It is also tied to another consequence of that belief: the almost-impossible obstacle course that autistic girls and women face in order to get a diagnosis.

Early concerns about children’s behaviour are more likely to be dismissed if they are girls. Page Pelphrey, about whom I write in my book, is married to one of the world’s top autism researchers, Kevin Pelphrey, and is the mother of an autistic daughter and an autistic son. With her daughter, Frances, the couple was told to “watch and wait”, and “you have a girl, it’s not autism”. With their son, diagnosis “was a snap”. Online information, meanwhile, slants towards the experience of boys.

Teachers, who should be well-placed to spot behavioural differences in the children they educate, are also prone to gender bias in autism. A 2020 study reported that teachers given identical vignettes of hypothetical children showing atypical behaviour were significantly more likely to suggest a child called “Jack” was autistic and in need of support than if the child were named “Chloe”.

If a girl gets as far as a diagnostic assessment, there is clear evidence that male bias is embedded in the “gold-standard” tests, with the content of the interview questions and the indicative scenarios validated on male-only samples. As many as 80 per cent of autistic girls remain undiagnosed by the age of 18.

These tests also make it much harder for scientists to recruit autistic women into their studies. Even when women show high levels of autistic traits, the tests often determine that they just aren’t autistic enough.

In the past decade, triggered by a wave of powerful personal testimonies from late-diagnosed autistic women, as well as a call to arms from members of the autism research community, the default male model of autism has finally come under scrutiny.

Newer neuroscience studies have homed in on sex differences, finding that autistic girls and women might present with different patterns of atypical behaviour, associated with different patterns of atypical brain activity. In particular, these studies revealed that the “robust” model implicating underactivity in the autistic social brain, established by several decades of research with mainly male participants, doesn’t generally apply to girls and women.

One study, for example, found that autistic girls had higher activity in areas associated with social reward than autistic boys – and even neurotypical girls – pointing to higher than usual levels of social motivation. Another found greater connectivity between key social brain networks in autistic girls compared with boys. And a third study looked at over-responsivity in sensory systems, which is usually thought to be characteristic of autistic behaviour and is linked to powerful aversive responses to loud noises or bright lights. In girls, the networks underpinning this behaviour seemed to have additional connections to those parts of the frontal areas in the brain involved with monitoring social feedback. In other words, there appeared to be a female tendency to monitor and control such over-responsiveness to avoid social embarrassment.

All of these findings suggest that the female participants have a more powerful proactive behavioural drive for social engagement and highly reactive responses to social rejection.

Put plainly, these findings don’t fit into the (male-based) model of stereotypical autistic aloofness and intentional social isolation. But they do fit with a growing realisation that autism presents differently in girls and women – or, perhaps more precisely, the current measures of autism don’t accurately capture their differences. Their lower scores on measures of social dysfunction don’t mean they have milder difficulties.

Rather, they reflect what turns out to be a lifelong pattern of masking or camouflaging, devising strategies to disguise or compensate for autism-related difficulties, particularly linked to social engagement. These may involve intensive study of the social behaviour of others and the generation of social guidance scripts, such as how long to engage eye contact or laugh at jokes, which will then be exhaustively rehearsed and followed whenever a social situation is encountered. It is the recent findings concerning such camouflaging behaviour that have really marked the emergence of an understanding that autism may present differently in girls and women.

This behaviour may imply an aspect of social compliance, with girls and women more likely to be “trained” to be quiet, calm and well-behaved. It should be stressed that there is huge variation and overlap, with some boys and men being more “chameleon-like” and some girls and women displaying the more typical signs of autism – though these individuals will find it easier to get a diagnosis. Not all autistic men are socially withdrawn, just as not all autistic women are desperate for high levels of social engagement.

Masking may seem like a successful trick. But, in fact, camouflaging in autistic girls and women is associated with high levels of anxiety, exhaustion and stress, as well as chronic depression and suicidal ideation. So the persistence of such damaging behaviour suggests an extraordinarily powerful drive to be social and an equally powerful wish to avoid social rejection, exactly as emerging findings in neuroscience suggest.

Given these new discoveries, some researchers have published new estimates of the male-to-female ratio in autism, which hint that it is a lot lower than the history of autism has led us to believe. In particular, large-scale surveys of autism traits in the general population suggest that we might be looking at equal numbers of autistic girls and women and boys and men.

Behind the mask

Tracking the story of autism’s missing women reveals the power of unchallenged belief to bias a whole clinical system, from awareness and recognition to definition, diagnosis and scientific research, distorting the public consciousness and sustaining unhelpful stereotypes. But the newly emerging awareness of these differences should begin to break down the barriers that autistic women have confronted and inform a more accurate – and inclusive – awareness of what autism can look like.

Better questions can be asked, and adjustments to diagnostic criteria should more accurately capture the lived experiences of all autistic people, not just boys and men. Autistic women and parents of autistic daughters will no longer be overlooked or ignored. Researchers can begin to draw on more balanced populations of autistic individuals and design measures that reflect how autism presents in all people. This won’t just be better for autistic girls and women; it will also revolutionise our understanding of the condition as a whole.

Gina Rippon is a professor emeritus of neuroscience cognitive neuroimaging at Aston University, UK. Her latest book, The Lost Girls of Autism, is on sale now