[Thanks to Marco G for proofreading and offering suggestions]


Several studies have shown a genetic link between autism and intelligence; genes that contribute to autism risk also contribute to high IQ. But studies show autistic people generally have lower intelligence than neurotypical controls, often much lower. What is going on?

First, the studies. This study from UK Biobank finds a genetic correlation between genetic risk for autism and educational attainment (r = 0.34), and between autism and verbal-numerical reasoning (r = 0.19). This study of three large birth cohorts finds a correlation between genetic risk for autism and cognitive ability (beta = 0.07). This study of 45,000 Danes finds that genetic risk for autism correlates at about 0.2 with both IQ and educational attainment. These are just three randomly-selected studies; there are too many to be worth listing.

The relatives of autistic people will usually have many of the genes for autism, but not be autistic themselves. If genes for autism (without autism itself) increase intelligence, we should expect these people to be unusually smart. This is what we find; see Table 4 here. Of 11 types of psychiatric condition, only autism was associated with increased intelligence among relatives. This intelligence is shifted towards technical subjects. About 13% of autistic children have fathers who are engineers, compared to only 5% of a group of control children (though see the discussion here) for some debate over how seriously to take this; I am less sure this is accurate than most of the other statistics mentioned here).

Further (indirect) confirmation of the autism-IQ link comes from evolutionary investigations. If autism makes people less likely to reproduce, why would autism risk genes stick around in the human population? Polimanti and Gelemter (2017) find that autism risk genes aren’t just sticking around. They are being positively selected, ie increasing with every generation, presumably because people with the genes are having more children than people without them. This means autism risk genes must be doing something good. Like everyone else, they find autism risk genes are positively correlated with years of schooling completed, college completion, and IQ. They propose that the reason evolution favors autism genes is that they generally increase intelligence.

But as mentioned before, autistic people themselves generally have very low intelligence. One study found that 69% of autistic people had an IQ below 85 (the average IQ of a high school dropout). Only 3% of autistic people were found to have IQs above 115, even though 15% of the population should be at this level.

These numbers should be taken with very many grains of salt. First, IQ tests don’t do a great job of measuring autistic people. Their intelligence tends to be more imbalanced than neurotypicals’, so IQ tests (which rely on an assumption that most forms of intelligence are correlated) are less applicable. Second, even if the test itself is good, autistic people may be bad at test-taking for other reasons – for example, they don’t understand the directions, or they’re anxious about the social interaction required to answer an examiner’s quetsions. Third, and most important, there is a strong selection bias in the samples of autistic people. Many definitions of autism center around forms of poor functioning which are correlated with low intelligence. Even if the definition is good, people who function poorly are more likely to seek out (or be coerced into) psychiatric treatment, and so are more likely to be identified. In some sense, all “autism has such-and-such characteristics” studies are studying the way people like to define autism, and tell us nothing about any underlying disease process. I talk more about this in parts 2 and 3 here.

But even adjusting for these factors, the autism – low intelligence correlation seems too strong to dismiss. For one thing, the same studies that found that relatives of autistic patients had higher IQs find that the autistic patients themselves have much lower ones. The existence of a well-defined subset of low IQ people whose relatives have higher-than-predicted IQs is a surprising finding that cuts through the measurement difficulties and suggests that this is a real phenomenon.

So what is going on here?


At least part of the story is that there are at least three different causes of autism.

1. The “familial” genes mentioned above: common genes that increase IQ and that evolution positively selects for.

2. Rare “de novo mutations”, ie the autistic child gets a new mutation that their non-autistic parent doesn’t have. These mutations are often very bad, and are quickly selected out of the gene pool (because the people who have them don’t reproduce). But “quickly selected out of the gene pool” doesn’t help the individual person who got one of them, who tends to end up severely disabled. In a few cases, the parent gets the de novo mutation, but for whatever reason doesn’t develop autism, and then passes it onto their child, who does develop autism.

3. Non-genetic factors. The best-studied are probably obstetric complications, eg a baby gets stuck in the birth canal and can’t breathe for a long time. Pollution, infection, and trauma might also be in this basket.

These three buckets and a few other less important factors combine to determine autism risk for any individual. Combining information from a wide variety of studies, Gaugler et al estimate that about 52% of autism risk is attributable to ordinary “familial” genes, 3% to rare “de novo” mutations, 4% to complicated non-additive genetic interaction effects, and 41% “unaccounted”, which may be non-genetic factors or genetic factors we don’t understand and can’t measure. This study finds lower heritability than the usual estimates (which are around 80% to 90%; the authors are embarrassed by this, and in a later study suggest they might just have been bad at determining who in their sample did or didn’t have autism. While their exact numbers are doubtful, I think the overall finding that common familial genes are much more important than rare de novo mutations survives and is important.

Most cases of autism involve all three of these factors; that is, your overall autisticness is a combination of your familial genes, mutations, and environmental risk factors.

One way of resolving the autism-intelligence paradox is to say that familial genes for autism increase IQ, but de novo mutations and environmental insults decrease IQ. This is common-sensically true and matches previous research into all of these factors. So the only question is whether the size of the effect is enough to fully explain the data – or whether, even after adjusting out the degree to which autism is caused by mutations and environment, it still decreases IQ.

Ronemus et al (2014) evaluate this:

They find that even autistic people without de novo mutations have lower-than-average IQ. But they can only screen for de novo mutations they know about, and it could be that they just missed some.

Here’s another set of relevant graphs:

This one comes from Gardner et al (2019), which measures the cognitive ability of the fathers of autistic people and disaggregates those with and without intellectual disability. In Graph A, we see that if a child has autism (but not intellectual disability), their likelihood of having a father with any particular IQ (orange line) is almost the same as the likelihood of a neurotypical child having a father of that IQ (dotted line). Disguised in that “almost” is a very slight tendency for fathers to be unusually intelligent, plus a (statistically insignificant) tendency for them to be unusually unintelligent. For reasons that don’t entirely make sense to me, if instead we look at the likelihood of the father to be a certain intelligence (bottom graph, where dark line surrounded by gray confidence cloud is autistic people’s fathers, and dotted line is neurotypical people’s fathers) it becomes more obvious that more intelligent people are actually a little more likely to have autistic children (though less intelligent people are also more likely.

(remember that “intellectual disability” just means “IQ over 70”, and so many of these not-intellectually-disabled people may be very intellectually weak – I wish the paper had quantified this)

Graph B is the same thing, but with people have have autism with intellectual disability. Now there is a very strong effect towards their fathers being less intelligent than usual.

This confuses me a little. But for me the key point is that high-intelligence fathers show a trend (albeit not significant in this study) to be more likely than average to have children with autism and intellectual disability.

These questions interest me because I know a lot of people who are bright nerdy programmers married to other bright nerdy programmers, and sometimes they ask me if their children are at higher risk for autism. While their children are clearly at higher risk for autistic traits, I think they want to know whether they have higher risk for the most severe forms of the syndrome, including intellectual disability and poor functioning. If we take the Ronemus and Gardner studies seriously, the answer seems to be yes. The Gardner study seems to suggest it’s a very weakly elevated risk, maybe only 1.1x or 1.2x relative risk. But the Gardner study also ceilings off at 90th percentile intelligence, so at this point I’m not sure what to tell these people.


If Ronemus isn’t missing some obscure de novo mutations, then people who get autism solely by accumulation of common (usually IQ-promoting) variants still end up less intelligent than average. This should be surprising; why would too many intelligence-promoting variants cause a syndrome marked by low intelligence? And how come it’s so inconsistent, and many people have naturally high intelligence but aren’t autistic at all?

One possibility would be something like a tower-vs-foundation model. The tower of intelligence needs to be built upon some kind of mysterious foundation. The taller the tower, the stronger the foundation has to be. If the foundation isn’t strong enough for the tower, the system fails, you develop autism, and you get a collection of symptoms possibly including low intelligence. This would explain low-functioning autism from de novo mutations or obstetric trauma (the foundation is so weak that it fails no matter how short the tower is). It would explain the association of genes for intelligence with autism (holding foundation strength constant, the taller the tower, the more likely a failure). And it would also explain why there are many extremely intelligent people who don’t have autism at all (you can build arbitrarily tall towers if your foundation is strong enough).

I’ve only found one paper that takes this model completely seriously and begins speculating on the nature of the foundation. This is Crespi 2016, Autism As A Disorder Of High Intelligence. It draws on the VPR model of intelligence, where g (“general intelligence”) is divided into three subtraits, v (“verbal intelligence”), p (“perceptual intelligence”), and r (“mental rotation ability”) – despite the very specific names each of these represents ability at broad categories of cognitive tasks. Crespi suggests that autism is marked by an imbalance between P (as the tower) and V + R (as the foundation). In other words, if your perceptual intelligence is much higher than your other types of intelligence, you will end up autistic.

It doesn’t really present much evidence for this other than that autistic people seem to have high perceptual intelligence. Also, it doesn’t really look like autistic people are worse at mental rotation. Also, the Gardner paper has analyzed autistic patients’ fathers by subtype of intelligence, and there is a nonsignificant but pretty suggestive tendency for them to have higher-than-normal verbal intelligence; certainly no signs of high verbal intelligence preventing autism. I can’t tell if this is evidence against Crespi or whether since all intellectual abilities are correlated this is just the shadow of their high perceptual intelligence, and if we directly looked at perceptual-to-verbal ratio we would see it was lower than expected. Also also, Crespi is one of those scientists who constantly has much more interesting theories than anyone else (eg), and this makes me suspicious.

Overall I would be surprised if this were the real explanation for the autism-and-intelligence paradox, but it gets an A for effort.


1. The genes that increase risk of autism are disproportionately also genes that increase intelligence, and vice versa (~100% confidence)

2. People diagnosed with autism are less intelligent than average (~100% confidence, leaving aside definitional complications)

3. Some of this effect is because autism is caused both by normal genes and by de novo mutations and environmental insults, and the de novo mutations and environmental insults definitely decrease intelligence. Every autism case is caused by some combination of these three factors, and the more it is caused by normal genes, the more intelligence is likely to be preserved (~100% confidence)

4. This is not the whole story, and even cases of autism that are caused entirely or mostly by normal genetics are associated with unusually low IQ (80% confidence)

5. This can best be understood through a tower-versus-foundation model where higher intelligence that outstrips the ability of some mysterious foundation to support it will result in autism (25% confidence)

6. The specific way the model plays out may be through perceptual intelligence out of balance with verbal and rotational intelligence causing autism (3% confidence)

New Comment
12 comments, sorted by Click to highlight new comments since: Today at 3:52 PM

The other explanation I've heard bandied about is a polygenic version of sickle-cell anemia (where being heterozygous for the allele protects you from malaria but being homozygous gives you an awful disease).

In this model, there are a bunch of alleles that all push the phenotype in roughly the same direction, and having some of them is good, but past some threshold fitness starts rapidly declining.

(Further speculation: the optimum threshold is higher in the environment of civilization than in the ancestral environment, so these genes are experiencing massive positive selection over the last 10,000 years.)

This isn't a causal explanation, but it differs from the tower-foundation model in claiming that there's not a separate weakness to go looking for, just an optimum that's being surpassed- especially when two people near the optimum have children.

Epistemic status: Wild guesses based on reading del Guidice's Evolutionary psychopathology and two papers trying to explain autism in terms of predictive processing. Still maybe better than the "tower hypothesis"

0. Let's think in terms of two parametric model, where one parameter tunes something like capacity of the brain, which can be damaged due to mutations, disease, etc., and the other parameter is explained bellow.

1. Some of the genes that increase risk of autism tune some parameter of how sensory prediction is handled, specifically, making the system to expect higher precision from sensory inputs/being less adaptive about it. (lets call it parameter p)

2. Several hypothesis - Mildly increased p sounds like something which should be somewhat correlated with increased learning / higher intelligence;

  • something which can force the system to build more exact representations, notice more "rule violations", keep track of more patterns, etc.
  • (also if abstract concepts are subject to the same machinery as sensoria, it would be something like having higher precision in abstract/formal reasoning)

3. But note: tune it up even more, and the system starts to break; too much weight is put on sensory experience, "normal world experience" becomes too surprising which leads to seek more repetitive behaviours and highly predictable environments. In the abstract, it becomes difficult to handle fluidity, rules which are vague and changing,...

4. In the two-parameter space of capacity and something like surprisal handling, this creates a picture like this

  • the space of functional minds is white, the orange space is where things break (in practice the boundary is not sharp)
  • for functional minds, g is something like capacity c + 0.1 * p; for minds in the orange area this no longer holds and on the contrary increasing p makes the the mind work worse
  • highly intelligent people can have higher values of p and still be quite functional
  • blue dotted area is what is diagnosed as autism; this group should be expected to have on average low g

Parts of the o.p. can be reinterpreted as

  • in this picture, some genes mean movement to the right; they are selected because of slight correlation with g
  • random mutations/infections/ etc. generally mean movement down
  • overall fitness profile of right-moving genes is somewhat complex (movement to the left or right is good or bad in different parts of the graph)

Even if this is simple, it makes some predictions (in the sense that the results are likely already somewhere in the literature, just I don't know whether this is true or not when writing this)

  • What happens if you move parameter p in the opposite direction? you get a mind less grounded in sensory inputs and stronger influence of 'downstream' predictions. In small quantities this would manifest as e.g. "clouds resembling animals" more for such people. Move to the left much more, and the system also breaks down, via hallucinations, everyday experience seemingly fitting arbitrary explanations despite many details not fitting, etc. This sounds like some symptoms of schizophrenia; the model predicts mild movement in the direction of schizophrenia should decrease g a bit


With a map of brains/minds into two dimensional space it is a priori obvious that it will fail in explaining the original high-dimensional problem, in many ways; many other dimensions are not orthogonal but actually "project" to the space (e.g. something like "brain masculinisation" has nonzero projection on p), there are various regulatory system like g means better ability to compensate via metacognition, or social support.

There is speculation that people with ASD have on average a lower IQ, but also a much higher variance. This would cause an over-representation of people with ASD among both the lowest and highest IQs, similar to the supposed male IQ variance. Perhaps ironically, ASD has also been dubbed an 'extreme male brain' disorder.

When studying psychometrics, there is a common finding usually known as the "positive manifold" where basically all positive attributes tend to positively correlate with each other, at least at population levels. I.e. people who are fast at running or have more symmetric faces also tend to be smarter. And there being very few positive things that tend to be negatively correlated with each other, or things that seem clearly negative predictors of performance in one domain being positive predictors of performance in another domain. 

I've been trying to understand the strength of this effect for a while, since I think it has really large effects on a lot of different psychology research, and I remember this post sticking in my mind as a relatively clear negative datapoint for the positive manifold, at least when you limit the population to be studied to an above-average IQ population. But I would also love to see more thinking going into this topic, and see some more eyes on this topic, so I am nominating it for the 2019 review.

(Epistemic status: I don’t have much background in this. Not particularly confident, and attempting to avoid making statements that don’t seem strongly supported.)

I found this post interesting and useful, because it brought a clear unexpected result to the fore, and proposed a potential model that seems not incongruent with reality. On a meta-level, I think supporting these types of posts is quite good, especially because this one has a clear distinction between the “hard thing to explain” and the “potential explanation,” which seems very important to allow for good discussion and epistemology.

While reading the post, I found myself wishing that more time was spent discussing the hypothesis that IQ tests, while intelligence-loaded in general, are not a great way to analyze intelligence for autistic people. The post briefly touches on this, but “mutations positively correlate with intelligence but negatively with test-taking ability through some mediator, meaning that at first, increased intelligence outweighs the negative effects, but depending on exact circumstance, intelligence is not possible to express on a standard IQ test after enough mutations accumulate” seems like a natural hypothesis that deserves more analysis. However, upon further reflection, I think that the neglection of this hypothesis isn’t actually an issue, because it conceals a regress: why does intelligence outweigh lack of test-taking ability at first, only to bring eventual significant costs? I think there are several just-so stories that could explain an inflection point, but I’d prefer not to posit them unless someone with more background/knowledge in this subject suggests that this is viable so as to prevent harmful adoption.

I think a more serious issue is the selection bias mentioned in the discussion of autism. Because IQ is positively correlated with good outcomes writ large (https://www.gwern.net/Embryo-selection, see an early section), including functionality, and autism in the DSM-V is defined as requiring various deficits and significant impairment (https://www.autismspeaks.org/autism-diagnosis-criteria-dsm-5), it would be somewhat shocking if autism was not negatively correlated with IQ. If we assume the two variables are completely independent, it would still be less likely for higher-IQ people to be diagnosed as autistic, because they are nearly definitionally less likely to meet the diagnostic criteria. This suggests a much simpler model, given the apparent correlation between autism and IQ: autism mutations push up intelligence in the vast majority of cases, and lower IQ autistic people are far more likely to be diagnosed. I wonder whether this could even explain some of the diverse harms associated with autism—if autism mutations push up “technical” intelligence/performance on iq tests relative to general intelligence, then could i.e. social skills appear to suffer because they’re correlated with a lower general intelligence (obviously way over-simplified, and entirely speculative).

Overall, I’d appreciate if this post was more comprehensive, but I think it’s a good category of post to promote as is. I’d weakly advocate for inclusion, and strongly advocate for inclusion conditional on editing to spend more time discussing selection effects.

Here's a few questions that might be interesting:

Is autism (or various ASDs) a scientifically valid definition( as defined in the DSM or ICD 10/11)?

What is the intra-site and inter-site diagnostic reliability for ASDs?

How consistently are ASD diagnostic tools applied? 

Do ASDs have any unique traits?

Are separate, unrelated, etiologies considered ASDs? 

Does "clinical expertise" or opinion having more weight than diagnostic tools impact validity?

Are the definitions of ASDs consistent enough longitudinally for analysis?

Do ASDs vary significantly by culture and socio-economic status?


The fundamental problem with most discussions of autism is they very poorly understand how autism is defined and diagnosed.  Before we can build these derivatives, it's important to understand that at it's core autism is diagnosed in a way that prevents it from ever being a scientifically valid concept.

One of the most important smell tests for the validity of any model should be how quickly heterogeneity sets in when you add more data.  Well defined models should be able to tolerate additional data with very low/no increase in heterogeneity.  That autism displays such high heterogeneity from every single data set (whether imaging, genetic, or case note analysis) indicates a very significant issue with how ASDs are being defined.  The variability isn't an indicator of massive complexity in expression, it's an indicator of a massively broken model.

Core to any discussion of ASDs lies the same core issues with all psychiatric/psychological definition, that there is no responsibility to define etiology and the endophenotypes are based on external observation or opinion.  Human cognitive bias examination should indicate why this approach is troublesome, and we see clear indications of this by the large variances in inter-site and intra-site reliability of autism diagnosis, as well as the extremely significant cultural and socio-econonic impacts on diagnosis rates.  That the current DSM V diagnostic criteria allows a pick list of non-specific symptomology backed by clinical opinion being able to override the result of the actual diagnostic tools themselves is on it's face a huge validity problem.

This type of analysis is interesting because it provides a non-specific substrate for a discussion of human variability in general, rather than anything specific to autism.

"autistic people ... generally have very low intelligence. One study ... autistic people had an IQ ..."

Unless you positively define intelligence as measured by some IQ-Test, I oppose that statement.

The entire discussion around intelligence would profit, if people would stop casually equating the two.

One is a test that have seen different ones of and some where out right bad others flawed, the other is a concept that can be described, but is much more often used than understood by the public.

"cases of autism that are caused entirely or mostly by normal genetics are associated with unusually low IQ (80% confidence) "

Only the research correlating genes and IQ-test results are objective.

All correlations between IQ and DIAGNOSED autism are skewed. People who are smart and have good enough speech skills, and thus are not too affected can hide their level of autism. People who are functional will not be diagnosed.

Lets assume, that autism is not an on/off deal but gradual and that there is a positive correlation with general intelligence, then the statistic will not include people who are below a high level of autism because they compensate.

Imagine that the graph of IQ as a function of parameter p is left-skewed.

Then natural selection is going to produce a population in which values of p are clustered around the optimum with more values slightly below the optimum than slightly above the optimum. In such a population, genes which increase p would paradoxically appear to be associated with both high iq and extremely low iq.

The issue here is the level of functionality on ASD, and why that it. Neuroscience recently has shown an increase in dendrite in the brains of autistic patients, this can possibly lead to some cognition issues - such as many more inputs per neurons causing a lack of ability of the brain to ‘sort’ information, hence non-verbal, non-functioning as ‘neuro typical’ humans. The other issue is the IQ test for intelligence, as an autistic person, I have never been able to take them. The questions almost swim around my own brain, rapidly turning to symbols I don’t understand, as if I’m trying to read Japanese. Following that, the english words become illegible. I’ve seen this in children I teach with maths phobia also. So it’s possibly the range of difference in the type of question or visual acuity. Interestingly my father is an engineer and so are most of my family, most undiagnosed, but some are. I’m on my 3rd degree whilst also teaching, but I cannot get past the first question on an IQ test. Cerebral organoids will provide an interesting way forward to test the possible hyperactivity of parts of the autistic brain.

When it comes to intelligence, rationality, depression, autism the evolutionary selection aspect is interesting, because we all know that the mentioned mental properties are lowering your chances to raise many children today.




Too much good quickly turns bad.