Is Anorexia Nervosa a Version of Autism Spectrum Disorders?

Patients with anorexia nervosa often have difficulties recognizing and regulating emotions. This  conclusion that is largely based on data from  common tests such as Reading the Mind in the Eyes assessing  emotion recognition, and questionnaires like Difficulties in Emotion Regulation Scale (DERS) assessing emotion regulation (see my post here).  Although that study compared currently ill patients with healthy controls (thus raising the possibility that the resulting data was due to the effects of starvation or due to the chronic nature of the ED  in the sample, ~7.5 year on average), there is some evidence that some of these difficulties persist post-recovery.

Individuals with autism (ASD, or autism spectrum disorders) also have difficulties with emotion recognition and regulation, leading some investigators to hypothesize that AN and ASD may share common etiology. Providing further support for this hypothesis are studies suggesting that AN might be overrepresented in ASD and vice versa. (That is to say, the prevalence of AN in ASD patients or vice versa is higher than the population mean.)

This doesn’t mean that all patients with one disorder have the other, or even a majority – just that the prevalence is significantly higher than you’d predict based on epidemiological data. In my view, this observation and this line of research is important for understanding the biological basis of these disorders. (Keep in mind, too, that whereas DSM demarcations are artificial, human traits are continuous – there is always a spectrum of behaviour.)

In this study, Oldershaw et al evaluate patients with AN on various cognitive tasks utilized in studying autism spectrum disorders and compare the results with published ASD data. First, though, Oldershaw outlines three main neurocognitive theories regarding autism spectrum disorders:

Neurocognitive Theories of Autism Spectrum Disorders

Empathising-Systemising Hypothesis – (Baron-Cohen, 2002)
– hypothesis explains strengths and weaknesses in ASD patients
– patients with ASD have difficulties with  inferring what another person thinks (Theory of Mind), deficits in the RME task
– patients with ASD tend to excel “working with systems”, strengths in systemising

Executive Dysfunction Theory – (Hill, 2004)
– attempts to explain the “need for sameness”, repetitive and restrictive behaviours in ASD
– poor set-shifting abilities (difficulty adapting to new rules, for example)

Weak Central Coherence – (Frith, 1989)
– seeks to explain obsessional preoccupations, need for sameness, and “niche abilities”
– weak central coherence is essentially the inability to “see the forest for the trees” or “see the bigger picture”
– poor global integration of information and excellent ability to focus on detail

There have been some studies suggesting that patients with AN score similarly to patients with ASD on tasks measuring central coherence and executive function, again, suggesting that there might be an overlap in the biological basis of these disorders.

The participants were 50 adult with AN, recruited from an outpatient unit. Two were male, and all had a BMI of less than 18.5. The sample was not limited to AN-restricting subtype, and included 9 AN-binge/purge subtype and 19 EDNOS (AN-type). I am a bit surprised by the decision to include AN-BP and AN-EDNOS, to be honest. I would have liked to see comparisons made based on subtype. No information was given about the length of the disorder or the average BMI, unfortunately.

The tests and questionnaires that Oldershaw used to assess the cognitive profiles of individuals with AN were: reading the mind in the eyes task (RME), reading the mind in the voice task (RMV), reading the mind in films (RMF) (theory of mind); Wisconsin card sorting task (WCST) (executive function); group embedded figure test (GEFT) (central coherence).


Theory of mind:

  • groups were matched for age, but AN IQ was lower than in the ASD group
  • compared to ASD, AN scored better on RME, the same on RMV and worse on RMF tasks

Executive function:

Central coherence:

  • again, AN were older than ASD patients, but well matched for IQ
  • no differences were found for the GEFT task (no differences between groups in the mean time to locate a simple shape within a complex design)

So, it seems that for the most part – at least for the tasks evaluated in this study – AN and ASD patients share similarities in neurocognitive profiles. This probably comes as no surprise to clinicians and caregivers of AN patients. What is surprising to me is that this similarity is apparent even in such a mixed AN group (restricting, binge-purge and ednos subtypes). My guess would have been that this similarity would be more apparent in a more long-term AN-R group that has never had symptoms of bingeing and purging.

The main question that still remains unresolved is whether these traits, in the AN group, are stable (that is, they were there before the AN diagnosis and likely persist after), and if so – to what extent? And conversely, to what extent are these results due to starvation?

ASD is a relatively consistent, lifelong developmental disorder whilst AN is a mental illness which can fluctuate in severity and be recovered from. Patients included here were acutely ill and underweight and as such the pattern of cognitive deficits in AN may represent non-specific consequences of the starved state (e.g. non-specific brain dysfunction) rather than specific traits associated with a diagnosis of AN.

Therefore, the definition of AN at the cognitive level requires additional research to confirm whether profiles are trait-like and this has crucial implications in the search for genetic markers. Research including recovered AN participants and longitudinal methodologies is in its infancy, but thus far indicates that weak central coherence (Lopez et al., 2009), executive dysfunction (Tchanturia, Morris, Surguladze, & Treasure, 2002; Tchanturia, Morris et al., 2004) and poor emotion recognition (Harrison, Sullivan, Tchanturia, & Treasure, 2010; Oldershaw et al., 2010) may be trait features of AN, while eToM difficulties may be restricted to the ill state (Oldershaw et al., 2010).

Are there alternative explanations?

Of course, there are other explanations that could explain the results. As an example, Oldershaw et al point out that obsessive-compulsive traits are not only predictive for the development of an ED but also could explain the rigid cognitive thinking styles present in AN. Another example is perfectionism, which may underlie the attention to detail seen in AN.

Since both perfectionism and obsessive-compulsive personality traits are heritable, they are thought to account for some of the genetic predisposition to AN (Anderluh et al., 2003). Therefore, an overlap at a cognitive level does not necessarily imply common underlying psychological, neurobiological or genetic bases for AN and ASD…

Basically, we need more research. (As if you didn’t know that already.)

As always, I try to point out the limitations of the study (most of which are mentioned by the authors themselves), to make sure that the results are interpreted in their proper context.

  • the ASD data wasn’t collected by one research group (but was taken from previously published data)
  • participants may represent extreme/severe examples (as they were recruited from a specialized care centre, this may exaggerate certain traits)
  • ASD sample had higher IQ than AN (and thus may be not very generalizable)
  • no data on the presence of ASD in AN sample or AN in ASD sample
  • gender bias (AN is 95% female, ASD, at least 80% male) may bias results (women tend to score better on empathy, men tend to be faster and more accurate on GEFT)
  • only the detail focus aspect of central coherence was studied (global integration still needs to be evaluated)

In any case, we have to wait and see to what extent – if any – these results can be replicated and whether they are state-related (due to the disorder) or trait-related (persist before and/or after the disorder) factors.

As always, I’m interested in reading your thoughts on this topic. I’m especially interested in hearing from those who have been diagnosed with both – do you think there is a connection or it is just by chance? What implications does this have – if true – for treatment and recovery?


Oldershaw, A., Treasure, J., Hambrook, D., Tchanturia, K., & Schmidt, U. (2011). Is anorexia nervosa a version of autism spectrum disorders? European Eating Disorders Review: The Journal of the Eating Disorders Association, 19 (6), 462-74 PMID: 21280165


Tetyana is the creator and manager of the blog.


  1. i think this is very interesting. im not sure if i agree that this is a form of autism but i think it is something to really research and learn about. i enjoyed reading this blog post.

    • Thank you for your comment anon. I agree with you – I think the title is provocative (I used the same title as the paper I reviewed), there’s definitely no evidence as far as I know to really suggest that AN is on the ASD spectrum, but, these disorders may have common genetic factors.

  2. I’m especially interested in this topic. I have AN. A number of years back I was reading about ASD and wondering if I had it. I even went to see a psychiatrist especially to be assessed. He quickly concluded that I didn’t have autism because I was able to empathise. He also quickly concluded that my problem was shame. I didn’t like him much.

    I think there are some interesting similarities, but also some key differences. I think that poor nutrition is common to both, as I’ve read about some of the health and nutrition problems found in ASD.

    So, of the problems that I think are similar between the two that I’ve encountered: disconnection from emotions, disconnection from people, rigid thinking, preference for sameness. My thoughts are that they are a challenge to overcome – to live with or to change if I can, one by one, bit by bit. And I certainly think that most of those things are worse when I’ve been restricting more.

    • Hi lovingmyfaith, thanks for your comment!

      Conclude that your problem was shame? That seems very odd!

      There are diffintely a lot of differences – I don’t think anyone denies that. The hypothesis is, are there some similarities, perhaps a subset of the same genetic variants that lead to ASD, predispose inviduals to AN.

      Yes, the challenge in studying is as you’ve pointed out: many of these similarities that researchers see do get worse with restriction. So then, is it restriction or is it pre-existing but just gets worse with restriction? It seems that at least for some, it is pre-existing. But, I think there are too few studies to be able to conclude anything concretely, it is just a hypothesis that is worthwhile to ponder and explore, I think.


    • The problem in autism empathy-wise isn’t lack of interest in other people, it’s theory of mind to my knowledge.

  3. I think that there really arent that many good similarities to believe that ASD and AN are related. The emotional difficulties in AN have to do with the fact that they arent getting any nutrients or energy and so they have mood swings and i feel like it can be prevented. In ASD there is something not right in the brain that causes the emotional difficulties and it cant be prevented.

    • Hi Michael,

      Thank you for your comment, however, I disagree. For a number of reasons, here are some:

      First, the question of whether AN and ASD are related is a question of whether they have shared genetic etiology. The behaviours that we measure (such as the tests discussed in the post) are just the phenotype – it is the end result of the genotype + environment. It gives us a rough idea, a suggestion that there might be shared genetic etiology, but as I mentioned in the post, there are other explanations that don’t rely on shared genetic etiology.

      Secondly, the studies of AN showing difficulties in many of the same tasks that patients with ASD have difficulties with, seem to persist after weight restoration and recovery. These could be ‘scar effects’, but I do believe some studies have shown that these difficulties are present prior. Thus, while lack of nutrition may exaccerbate this, I don’t think there is any evidence to believe it is solely the result of lack of adequate nutrition, given that even when there IS adequate nutrition, the same behavioural deficits exist.

      Moreover, it is not fair to say patients with AN aren’t getting “any nutrients or energy”, they may be getting insufficient amounts, but they could be eating 1000 or 1200, or probably even more, a day. It is not enough, but it is not zero.

      I’m also not sure where the point about mood swings comes in, particularly in the case of recognizing emotions in others, such as the “reading the mind in the eyes/voice/film” tasks. This task doesn’t have anything to do with regulating ones own emotions.

      Moreover, all behaviour has a neurobiological basis. The essential question here is, are the behaviours we are seeing due to similar neurobiological underpinnings, or not? It is true that ASD may not be prevented, but is there evidence that AN can be prevented? I’m not sure. It is true that people with AN recover – but, behavioural deficits, such as the ones discussed in this post – persist. Which means that, just because they maintain a healthy weight and eat well, it doesn’t mean that many of the underlying features that predisposed them to develop the disorder, are gone. They can be present, and persist for life, just be managed. Like in the case of someone with mild ASD, who learns a lot about human interactions and so on, almost in a “fake it til you make it” kind of way.

      A few other points to make: patients with ASD and patients with AN have increased risk of AN and ASD, respectively, in their families. You take a person with ASD or AN, and study the prevalnce of AN or ASD in their family, and it is MUCH higher than you’d expect based on a random (ie, not related) population sample. This again suggests shared genetic etiology.

      Final point, ASD and AN need not have to share all the genetic underpinnings, and it is unlikely the do, but there might be significant overlap. AN is more prevalent in females, and ASD in males, in almost reverse ratios. It is possible that the same genetic variants that lead to autism in males, lead – or predispose – females to AN. It could be that some of the genes that lead to ASD, in females, predispose them to AN. The neuropsychological phenotypes are similar, based on many of the tests above, but the ultimate behaviours might be different for all sorts of reasons that lead to gender differences (hormones, environment, all sorts of things).

      Thanks for your comment.


  4. I’m diagnosed with Asperger’s Syndrome and I’m in recovery from restricting-type AN. I absolutely believe that there are connections between ASDs and AN. I only have anecdotal evidence, of course, but I’ve discussed this with friends from treatment, and many areas of similarities between AN and ASDs were life-long issues. Alexythymia, obsessive thought patterns, tactile issues, difficulties with executive functioning… They seem very common.
    If I had the chance to conduct research into this, I would separate various ED diagnoses, and I would survey patients’ parents and former teachers, where available, as I customary in many adult diagnostic procedures for ASDs. I think it would be interesting to see which issues developed after the ED developed, and which were long-term.

    • Yes, definitely. I don’t think that there is a question that there are similarities between AN-R and ASD but the question that does remain is: do the similarities have similar causes or is it just that phenotypically (behaviourally/thought-pattern-wise) they look similar, but the reasons/causes leading to those behaviours/thoughts are actually very different? Of course, there’s also the question, as you point out, of behavioural stability. Are the symptoms we see in AN preceeding the disorder, the result of the disorder, are they scar-effects that remain even after recovery but didn’t necessarily exist before? It is very hard to untangle these things!
      It would be sweet to do prospective studies (as opposed to retrospective, as you suggest, because recall bias is a huge problem) but the low prevalence rates of AN does make it a challenge.
      I also wonder about the gender differences… ASDs are more common in males and AN in females. It is totally a speculation, but perhaps, if the factors leading to these disorders are similar, the resulting behaviours/thoughts (phenotypic differences) are modulated/mediated by gender (hormones?)? Lots of interesting questions.

Comments are closed.