Viewpoint Expert opinions on trends and controversies in autism research.
Opinion / Viewpoint

Uta Frith: Why I am obsessed with this cognitive thing

by  /  26 April 2011
The Expert:

Uta Frith

What do we know about autism and its causes? If you have a Post-it note handy, we can start by drawing some lines to create spaces for what we already know and what we don’t.

We’ll reserve the top space for facts about the biological basis of autism. You could list an impressive number of risk factors here, such as susceptibility genes, as well as a mass of data about the brain in autism: differences at the cellular level and differences in the size, activation and connectivity of brain structures, for example.

The space at the bottom is for all the behavioral observations, and performance on neuropsychological or psychometric tests. Behaviors reported by parents or others using standard diagnostic instruments — such as questionnaires or interviews — would be listed here as well. There is a huge amount of data: for recognition of emotion in faces alone, there are more than 40 papers.

The middle space between the two lines is for ideas about what is different about the mind of someone with autism — here is the core of my obsession.

This space describes how the brain sees, understands and acts in the world. The term ‘cognitive’ describes everything that the mind does, and refers to all aspects of mental life. Cognitive is often contrasted with affective, which describes emotion. But what we think and what we feel are two sides of the same coin.


brain scan chromosome 22


It is exciting to try to pinpoint what is distinctive about these minds. What in the mind or brain makes it difficult for someone with autism to adapt to the social world? I’ve written the term ‘mentalizing’ to refer to one theory: that people with autism have difficulty recognizing what others think and feel. I’ve also added the term ‘detail focus’ to point to another theory: that autism involves attention to parts at the expense of wholes. This could explain what in the brain allows individuals with autism to develop special talents.

Such hypotheses are only useful if they lead to testable predictions and observable findings at the biological and behavioral levels. Data at both these levels only become meaningful if they can be explained in terms of what the mind does. Understanding that a child with autism may be ‘mind-blind’ is far more helpful to a teacher than knowing that he or she has a particular genetic abnormality (biological) or tends not to look at eyes (behavioral)

We also need a space for causes that are outside of the person — let’s call this ‘environment.’

Environment can act at every level of the diagram, hence the arrows. There are plenty of data on environmental effects. Some of these are risk factors, especially at the biological level, but others can have positive effects. For example, teaching can powerfully shape actions and help individuals overcome behavioral problems.

Elusive node:

We now have a structure that we can use to lay out what we have already learned about autism. I have drawn lines on Post-it notes, but if you want to insert all the facts that are already known, you will need a very large sheet of paper!

We can also use this simple layout to imagine how our knowledge might expand in the future. The holy grail of autism research is to trace the causes of autism from genes to behavior, to explain what clinicians, parents, teachers and researchers all recognize as autism, and to improve life for people with the disorder.

We know much already, but how can we pull the existing findings together? We need to find a common pathway — the critical part in the system that is always affected, no matter which of the numerous genetic and environmental risk factors has caused the autism.

If we were to find a common pathway, this would define a distinctive property of the brain that is critically different in autism. The common pathway forms a node in a network of possible mappings: It pulls together all the strings from the data that are already available and from the data that are not yet available, but can be predicted.

If we can find a node at the cognitive level, animal models become more informative, for example. This is because one can look beyond behavior, which is obviously very different in mouse and man.

There is a thorny point here: The node would pinpoint both normal and abnormal function. But isn’t this too black and white? Isn’t everyone ‘a bit autistic?’

I happily agree with this at the behavioral level, where one thing always shades into another. A measure of amount of eye contact, for example, will vary continuously across different children — we can only make an artificial cut-off point to define what is ‘abnormal.’ The same goes for the biological level: If you assess neurochemical levels, you will get a continuous distribution, with clinical groups tending to lie at one extreme.

At the cognitive level, however, it is admissible to use categorical distinctions. We are free to theorize that there are functions that the typically developing mind automatically performs, which the mind of a person with autism performs in a qualitatively different way, or not at all — and vice versa. Perhaps this is the case with mind-reading and mind-blindness.

To me there is something very appealing about pulling together the strings in a common pathway. Everyone agrees that autism at the biological and at the behavioral level presents a huge amount of complexity and heterogeneity. This is why some researchers are searching for subgroups.

At the same time, there is a strong clinical impression that autism is something distinctive that can be recognized across different individuals. I think it would be very interesting to investigate what it is that causes this intuition. One possibility is that there is something unique to and ubiquitous in autism — and I believe that this can be found at the cognitive level.

Lessons learned:

The best lessons I have learned during my career are from parents who have shared their insights and observations with me. They have encouraged me to take a hard look at what precisely might be wrong in the machinery of the mind or brain in people with autism. Sometimes their comments and anecdotes have made me change my mind and redirect my research.

Even at its most theoretical and abstract, my research has been informed by personal experience — sometimes my own, but more often that of the parents and teachers that I have had the good fortune to know. I would recommend, even insist, that researchers interact with the individuals they study as well as their caretakers.

You can tell when you read a paper by an author who has never had any real experience with an individual who has autism. When I read such papers I often think — rather harshly — that the authors don’t know what they are talking about and I would be surprised if they had anything relevant to say in the long run.

So, to young researchers I would say: You can’t listen and observe enough. That’s how you will continue to learn something new and surprising about autism every day.


Uta Frith is professor of cognitive neuroscience at University College London.

17 responses to “Uta Frith: Why I am obsessed with this cognitive thing”

  1. Anonymous says:

    Thanks Uta for a great exposition of the cognitive approach to autism. I often struggle trying to explain what cognitive psychology is all about and why it’s important not to get too carried away with the latest study implicating a particular gene or brain blob. So this is really useful.

    I’d be interested to know what makes you think that there is something “unique and ubiquitous” about autistic cognition. It’s certainly *possible* that all people with a diagnosis of autism have some unique cognitive factor that differentiates them from those without a diagnosis. But given how broad the diagnostic criteria are and how much autism overlaps with a range of other disorders, this seems fairly improbable to me.

  2. Anonymous says:

    Hi Jon. Thanks for liking the cognitive thing! Your question touches weak spot of autism research now. I agree with you: the goal posts of autism as a diagnostic category have shifted and are so wide apart that it is indeed unlikely that we find the unique and ubiquitous ‘essence of autism’ in everyone with a diagnosis of autism. But, I think that within the noise there is a clear if hidden signal of this ‘essence’. Once we can identify it, the goal posts will narrow again. Many people who are now diagnosed as autistic will not fit into this narrower band, and it is important to find some more appropriate diagnostic categories for them.

  3. Anonymous says:

    So would you advocate going back to earlier definitions of autism? And if so, how far back? Given that most cognitive research nowadays involves participants who would not have been diagnosed “autistic” in the 1970s, where would that leave us?

  4. Anonymous says:

    Obviously, we can’t go back. earlier definitions weren’t good enough. I do not have a solution, but I hope that ‘this cognitive thing’ will provide a way forward.

  5. Anonymous says:

    You introduce the idea of qualitative analysis without including quantitative. Both forms, whether analytical or synthetic inform signal modality. You need to understand language and numerate morphology, psychoanlytical processes and its relation to psycholinguistics and psychomathematica.

    • Hans Boerema says:

      Hello, I want to learn all about numerate morphology, psychoanlytical processes and its relation to psycholinguistics and psychomathematica. Can you point out some good starting points?

  6. Anonymous says:

    Dear Professor Frith,

    I posted a link to this article with the two paragraphs about listening to people with autism at The Thinking Person’s Guide to Autism Facebook page, which has a robust following. Approving comments have been coming in:

  7. Anonymous says:

    The search for the core defining feature is, as you state, is the search for the ‘Holy Grail’. Theory of Mind deficits has been proposed as the defining core cognitive deficit the distinguishes ‘autism’ from other neurodevelopmental and neuropsychiatric disorders. In fact, Theory of Mind deficits are not diagnostic specific at all and TOM deficits have been reported in such diverse neurologically impaired groups as schizophrenia, anorexia nervosa, intellectual disability, Downs Sydrome, frontal lobotomy and right hemisphere stroke patients.
    I agree with Kanner’s core defining feature, which was encoded in DSM-III (1980) as a ‘pervasive lack of responsiveness to other people – autism’. The pervasive lack of responsivness applies to parents, siblings and caregivers.
    In 1994, which is when the so-called autism epidemic began, all references to Kanner’s core defining cognitive feature were completly removed from from all diagnostic schemes with the newly introduced DSM-IV (1994) and ICD-10 (1994) and replaced by the ambiguous and subjective cognitive feature ‘a lack of social reciprocity’.

    As far as the genetic theories are concerned, The California Autism Twins Study (CATS) is by far the largest population twin study project ever attempted. In a paper to be presented at the IMFAR Conference in San Diego next month they report that severity of autism is not ‘heritable’.

    Francesca Happe who you have co-authored many papers with in a study of ‘autistic-like’ traits in general population twins has also stated that “Around 10% of all children showed only social impairment, only communicative difficulties or only rigid and repetitive interests and behavior, and these problems appeared to be at a level of severity comparable to that found in children with diagnosed ASD in our sample”. That 10% of the general popluation could be described as having an ‘Autism Spectrum Condition’ demonstrates how ambiguous and subjective autism definition has become. If you throw out too wide a net you cach all the wrong sorts of fish.

    Diagnostic substitution is not a new phenomena in the brief history of autism. In 1960’s Leo Kanner joined Van Krevelan in objecting to ‘the abuse of the diagnosis of autism’ that ‘threatens to become a fashion’:
    Kanner wrote:
    “While the majority of the Europeans were satisfied with a sharp delineation of infantile autism as an illness sui generis, there was a tendency in this country to view it as a developmental anomaly ascribed exclusively to maternal emotional determinants. Moreover, it became a habit to dilute the original concept of infantile autism by diagnosing it in many disparate conditions which show one or another isolated symptom found as a part feature of the overall syndrome. Almost overnight, the country seemed to be populated by a multitude of autistic children, and somehow this trend became noticeable overseas as well. Mentally defective children who displayed bizarre behavior were promptly labeled autistic”.

    As far as the search for the core defining feature of autism you can go back. There is no reason why Kanner’s definition encoded in DSM-III (1980) can’t be used at least for research purposes in the search for the Holy Gail, which by the way, has yet to be found.

  8. Anonymous says:

    I too would like to see more research on the cognitive specifics of autism and how to better reach their amazing and different minds to spark their interest in subjects that are not their favorites. My young autistic son is brilliant in many aspects but he completely tunes out when he doesn’t care about the subject, so it becomes hard to expand his knowledge about the world and about what makes his peers “tick”. Also it’s very difficult to teach him why things that make him very upset are not that big a deal, and so forth.

  9. Anonymous says:

    Indeed the cognitive nexus often is overlooked – so too the physiological variables that correlate with cognitive variation within and beyond the clinical autism spectrum. Scientists, like human beings in general, are much more comfortable with the inherently single-variable problems posed by single genes or focal, regionally delimited brain dysfunctions. Our own data, some of them published and some still in various stages of analysis, suggest that dimensional cognitive and perceptual traits of autism, both in social and non-social domains, correlate with fMRI and EEG measures of the timing of frontal lobe activity, whereas the categorical diagnosis seems to have more to do with long-range transfer of information between widely separated brain regions – an emergent phenomenon in which *local* network entropy within frontal systems influences the development of broader, longer-range connections that subserve the rapid and flexible integration of cognitive capacities to address social communication and other novel, unpredictable cognitive problems. In this regard our work is consistent with our and many others’ theoretical views enunciated during the past seven or eight years. A new development, for me, is the recognition that what you have described as “central coherence” actually seems identical to what Yaacov Trope has termed, in the context of social psychology, “level of construal.” Trope’s relation of level of construal to what he terms “psychological distance” links non-social perception and social perspective-taking, and in my view forms an important connection between these levels and domains of cognition both in the normal population and also within the autism spectrum. I have presented on these ideas (see the lecture series “”) but have been so backlogged that it’ll be the end of the summer at least before I have a chance to write the paper that I have in my head!

    And BTW for the past eight years or so I always have been using a slide in my talks which is the spitting image of your second post-it note under the “elusive node” headng – an indication, I suppose, that much of autism research works by zeitgeist.

    Warm regards,
    Matthew Belmonte
    Visiting Professor, National Brain Research Centre
    NH-8, Nainwal Mode, Manesar 122 050, India

  10. Anonymous says:

    My son fits this paradigm exactly:

    IQ of 140, special ed all through school. Suspended when fighting back when called a “retard”. We homeschool now.

    Right brained:…This book can teach you how to be autistic,er ahem, artistic. Shows exercises of how to shut off the tyranny of the left brain.

    I think there are people who are right brained who also have low IQ’s. We call them severely autistic. When you are of a normal mind with a low IQ you are not called severely normal. Why is that? You are called severely retarded. Does that mean normal people are just retarded? It should.

    I wish you would not see autism as a defect. Just sayin….

  11. Anonymous says:

    My degree is only in Special Education. My interest in my students and my son was only the cogtnitive. After that, I figured behavior would follow to some degree.

  12. Anonymous says:

    Maybe I am misreading this…but it seems both you and Dr. Brock seem to be mildly implying that the “autisms” seem to be highly over-diagnosed…and that Neurodiversity, in it’s implication that autism is a way of being, is correct for a majority of the people who file under it’s moniker.

    I tend to agree. I am mightily ticked that able bodied, able minded people like my son are lumped under the same label as kids who have no speech, and little emotional liability. I didn’t know in the beginning that he might not regress, but my family was wise enough to see that he was only a variation of normal. We have a lot of “little professors”. I’m not in any way implying that it is an easy life to be seen as different…it’s not easy for anyone.

    AND, I am in no way implying that I don’t want my son labelled with “the others”. Both he and I know that there are similarities, similar developmental trajectories of mildly afflicted and mightily afflicted people who receive the label of autism. Thus the confusion. But we are labelling a mighty big chunk of a generation of kids as “mentally ill”, and the concomittant mess that arrives with psychiatry’s meddling in the lives of the unworthy (ask any gay…) when chidren are born with an innate way of dealing with the world.

    I love talking to myself…nobody to argue with me….

  13. Anonymous says:

    Coming to this late, via Jon Brock’s site.

    The three diagnostic behavioural categories that define autism are each very complex, so I don’t understand why there should be only one cognitive node. Why can there not be a spectrum of cognitive nodes arising from a spectrum of biological causes and resulting in a spectrum of behavioural outcomes?

    In other words, couldn’t different people show behavioural patterns that meet the diagnostic criteria for autism but have unrelated biological origins and unrelated cognitive pathways?

  14. citsitua says:

    The cognitive approach is the glaringly obvious approach to take, so it is quite surprising that researchers are not taking this approach. No wonder autism is a so-called puzzle. Autism is not a puzzle. There is nothing puzzling or “enigmatic” about it, from the perspective of the autistic person. Because normals are mindblind to autistic people, and because normals lack empathy with autistic people, and because normals cannot read the non-verbal cues of autistic people, and because normals do not have an intuitive understanding of autistic theory of mind, the best way to start any research project is, as you wrote, to interact with real live autistic people to learn more about their internal experiences beyond the external behaviors. We don’t bite. Sometimes.

  15. jwright says:

    I’m delighted to see that this article has launched such an interesting discussion!

    Sue Gerrard, I think this is a great point. The question is, if there truly is such a thing as ‘autism’– and I absolutely believe that there is — there must be a common convergence somewhere in the pathway. Clinicians who spend an enormous amount of time with individuals with the disorder say that despite the vast differences across the extremes of the spectrum, there is something undeniably ‘autism-like’ that these individuals share.

    Citsitua, rest assured that there is a lot of research on cognition in autism all of which uses participants with the disorder. Understanding the mechanisms behind cognition of any brain, autism-like or not, is extremely complex. I know what I think, for example, but I can’t begin to imagine the neural pathways that take place in my brain that link sensory perceptions to emotions and thoughts.

  16. usethebrainsgodgiveyou says:

    Graphic at bottom based on Dr’s Eide of Dyslexic Advantage and my take on the cognitive profile of a lot of kids diagnosed as Dyslexic/Aspergers/ADHD

Leave a Reply

Your email address will not be published. Required fields are marked *