Transforming Autism Research: Reflections on IMFAR, RDoC, and DSM-5

Originally posted on Cracking the Enigma, June 2013

A year, it seems, is a long time in autism research. In May 2012, the International Meeting for Autism Research (IMFAR) in Toronto was all about DSM-5 and the proposed changes to autism diagnosis contained therein. Up for discussion were the abolition of Asperger syndrome as a distinct condition, the introduction of a new (and still mysterious) diagnosis of “social communication disorder”, and the feared exclusion of many people from autism-specific support and interventions.

Those concerns have not gone away. But 12 months later, at IMFAR 2013 in San Sebastian, the DSM-5 debate was notable by its absence. The manual had long since gone to the printers and people had apparently agreed to disagree.

Discussion turned instead to comments by Thomas Insel, director of the National Institute of Mental Health (NIMH). In a blogpost entitled “Transforming Diagnosis”, Insel announced that the institute would be “re-orienting its research away from DSM categories”. “Patients”, he argued, “deserve better”.

Insel’s proposed solution is a new research framework – Research Domain Criteria or RDoC. The aim of RDoC is to make connections across different levels of analysis, from genes and neurotransmitters through to brain circuits and behavior. But rather than basing research on existing diagnostic categories, RDoC focuses on broad constructs such as ‘social communication’, ‘cognitive control’, or ‘anxiety’ that cut across existing diagnostic boundaries.

As Insel pointed out, the DSM’s great weakness is “a lack of validity”. This doesn’t mean that it’s useless or a waste of time. The key point is that there’s little independent evidence to suggest that people who are diagnosed with the same condition really do belong together – and should be treated differently to people with other diagnoses. We don’t know that their symptoms have the same underlying cause, and we can’t say that their outcomes will be similar, or that they should be offered the same interventions.

These issues were front and centre in the opening address of IMFAR 2013. Focusing on the problem of “comorbidity”, Christopher Gillberg noted that people with autism often meet criteria for several other DSM diagnoses, including ADHD, specific language impairment, and developmental coordination disorder. These, Gillberg argued, are not discrete syndromes, so it is inappropriate to research them separately or to diagnose one condition without considering the implications of the others.

Daniel Geschwind’s keynote address on Day 2 raised similar problems with the validity of autism. Reviewing the latest in autism genetics, Geschwind noted that, according to current projections, autism is linked to upwards of 1000 different genetic variations. Although these appear to converge on a smaller number of pathways, most of the genetic risk factors discovered so far also increase the likelihood of other, supposedly unrelated disorders. Like so many other conditions in the DSM, autism is a mess.


Insel’s comments, arriving just days before DSM-5’s publication,  were widely seen as a final blow to the DSM’s credibility and a major snub to its publishers, the American Psychiatric Association. Media coverage spoke of a civil war within psychiatry. But the reality is that the aims of RDoC actually align closely with those of the APA.

Back in 2002, in a document entitled “A Research Agenda for DSM-V”, Dennis Charney and colleagues stated:

“It is our goal to translate basic and clinical neuroscience research relating brain structure, brain function, and behavior into a classification of psychiatric disorders based on etiology and pathophysiology. It is possible, even likely, that such a classification will be radically different from the current DSM-IV approach.”

It soon became clear, however, that this radical new approach to diagnosis was not ready for primetime. As Insel noted, the controversy surrounding DSM-5 belies the fact that the changes from DSM-IV have been relatively modest. In the case of autism, the reworking of diagnostic criteria may yet have profound consequences in terms of who receives interventions and services. But conceptually, little has changed. Seventy years after it was first identified, autism is still diagnosed in terms of the collection of behavioral symptoms that also define the condition.

In the absence of reliable alternatives, clinicians will continue to rely on the DSM. But researchers do not have the same excuse. Holding up the current diagnostic criteria as the ‘gold standard’ for research makes it impossible to discover better alternatives.

Writing for the Dana Foundation in 2011, DSM-5 panel member, Steven Hyman, concluded:

“it is critical that the scientific community escape the artificial diagnostic silos that control so much research, ultimately to our detriment.”

Lynn Waterhouse reached a similar conclusion in her 2012 book, “Rethinking Autism”:

“The only viable scientific path… requires researchers to abandon the use of the diagnosis of autism as a basis for research.”


Last year on my journey home from Toronto, I spent the flight reading Jacob Bronowski’s “The Common Sense of Science”, a 1951 history of science that I’d picked up for a few dollars in our local second hand bookstore. I took it for no reason other than that it fitted neatly into my hand luggage. But the ideas seemed particularly relevant in the light of the drama and intrigue of the IMFAR meeting.

According to Bronowski, the first step towards scientific understanding is to “order” things into groups, as he put it, “not of identical things, but of things which seem to be or behave alike”. This “ordering” process is necessary if we are to generalize our narrow scientific observations towards a broader understanding of the world. Initially, the “throwing into groups” has to be intuitive, but these intuitions must be tested. If the evidence doesn’t fit, then we should be prepared to refine or even abandon our intuitions and (literally) regroup:

“We order by likeness, and we choose those likenesses which we first judge and then find to have significance” [emphasis added]

Since Kanner, the idea of a discrete condition, “autism”, has made intuitive sense to researchers and clinicians alike. We could metaphorically throw a wide range of people into that group because, although they were different from one another in many ways, they seemed to have something essential in common. What was learned about one sample of autistic people could, we assumed be generalised to others. However, as more evidence has come in, the concept of autism as a coherent entity has been found wanting. This is the “validity” issue that so undermines the DSM.

Insel’s bet is that the “domains” of RDoC do have validity, that they will map better onto underlying biological mechanisms and, crucially, that they will provide clearer insights into appropriate treatments and interventions for individuals. RDoC represents an alternative “ordering” of the human mind. But like the DSM, it needs to be empirically tested, refined, and, if necessary, rejected.


After all the controversy surrounding DSM-5, RDoC represents a fresh start, and a much-needed reframing of the scientific problem. At present, however, it’s little more than a statement of intent, describing how research could and perhaps should proceed. The goals are long term. And while the idea of looking beyond DSM categories is a welcome move, there’s no guarantee that RDoC will succeed in its primary objectives of linking behavioural symptoms back to their underlying biological origins.

Identifying risk genes and their immediate relevance to neural function is likely to be the easy part. The greater challenge will come in working out how these differences interact with other genetic and environmental risk factors, such that some individuals develop into autism, whereas others have completely different outcomes. Given the complexities, both known and unknown, that are yet to be addressed, it seems essential that any biological approach to autism is complemented by research with more immediate benefits to the autism community.

For me, the highlight of IMFAR 2013 was Maureen Durkin’s keynote address on the final morning, focusing on the discrepancies in autism diagnosis across countries, cultures, and social classes. In her closing remarks, Durkin emphasized an important distinction – between the current emphasis on autism as an impairment and the alternative view of autism as a disability. Practical strategies for assisting people with sensory and physical disabilities often have the indirect consequence of increasing social participation. It follows, Durkin argued, that research on autism should address similar practical solutions that might lead to greater inclusion of people with autism in society.


DSM-5 has been a long and painful process. Only time will tell if the changes represent a net improvement on DSM-IV. But the debates generated within the scientific community and beyond have served an important purpose, laying bare the questionable assumptions that underpin contemporary autism research. Whatever your views on the specifics of DSM-5 and RDoC, Dr Insel was certainly right about one thing. There is room for improvement. Autistic people deserve better.



Insel, T., Cuthbert, B., Garvey, M., Heinssen, R., Pine, D., Quinn, K., Sanislow, C., & Wang, P. (2010). Research Domain Criteria (RDoC): Toward a New Classification Framework for Research on Mental Disorders American Journal of Psychiatry, 167 (7), 748-751 DOI: 10.1176/appi.ajp.2010.09091379