Numerous grant funding calls, as well as public and scientific debates are now focusing on the cause of autism spectrum disorder (ASD). While it is important that we search for the cause(s) of the disorder - in order to improve support, education, employment and so on for those on the spectrum - such an approach is problematic for two reasons. First, there is an implicit suggestion that the quest for a cause will be simple and thus achievable. Second, there is a general assumption that there will be a unitary (single) cause to explain all of the wide ranging behaviours seen within an individual with ASD, as well as across the spectrum of the condition. With the development of molecular and behavioural genetic research programmes, as well as neuroimaging research, this quest has appeared to many to have an end in sight. Despite the wealth of superb minds focused on understanding the condition, why, then, have we failed to identify one (or more) causes of ASD?
As noted above, it is very clear that there is a highly varied profile of behaviours observed in autism, both at one measured time point (between individuals) and over the lifespan (within an individual). In order to make substantial in-roads into understanding the condition, it is important to investigate the nature of this varied profile. One useful method that will improve the chances of studies identifying causation and linking brain and behaviour (e.g. through neuroimaging studies), is to use the terminology adopted by Morton and Frith (1995) in their causal modelling approach to neurodevelopmental disorders (1). This approach has been outlined in a series of papers and in detail more recently (2).
The causal modelling approach provides a graphical tool for thinking about neurodevelopmental disorders. This approach can be used to model multiple theories of a single disorder (in this case ASD), as well as to model possible similarities and differences between the causes and consequences of different disorders. The causal modelling approach has four constituent parts: biology, cognition, behaviour and the environment. In turn, these factors will be broken down further. In the biological domain membrane proteins, receptors and neurons, for example, may all play a role. Likewise, cognitive processes will be broken down into their constituent parts; for example, phonological processing, inhibition and verbal short-term memory. Behaviours include reading skill, digit span and word naming. Finally, aspects of the environment may have an influence at any level of the model. These influences may be protective or destructive and might include intra-uterine environment, parenting style, diet or relationships. Cognition is crucial in the causal modelling approach, as it acts as a mediator between biological and behavioural features. Finally, it is likely that there will be interactions between different levels of the model, such that relationships between components will be bi-directional, having a developmental impact on one another over time.
By contemplating the cognitive explanations for behaviours observed in ASD, we can start to establish how groups of behaviours link together, and where other behaviours are separate. In the former situation, common cause would be implicated, while in the latter, differential cause may be more likely. Cognitive accounts of ASD have emerged gradually over the years, with one early account that remains popular describing a theory of mind deficit, or mindblindness, in ASD (3). More recently, other cognitive theories have emerged, including difficulties in the area of executive functions (an umbrella term for abilities such as planning, working memory, impulse control, inhibition and shifting set, as well as the initiation and monitoring of action) (4). In addition other accounts attempt to explain the profile of strengths often reported in ASD, with the weak central coherence and enhanced perceptual functioning accounts being pertinent here (5). While none of these theories can fully explain the observed behaviours, what the theories do is to highlight how behaviours (e.g. poor understanding of another's belief about a situation, poor planning ability) group at the cognitive level, with further specification being possible at the biological level (in the ToM case linking to brain areas implicated in typical theory of mind such as the anterior cingulate, in the executive functions case linking perhaps to structures of, or neurochemical input in areas of the frontal lobes, or connections between brain areas). Note that these are illustrations rather than an exhaustive list. Furthermore, clear links can be made between the cognitive level of the model and functional tasks that are the focus of day to day life (e.g. getting dressed, feeding skills, going shopping).
Essentially, then, Morton's causal modelling approach enables a visual picture of potential links between the different levels of description of a disorder or set of functions to be made. This can then aid interpretation of patterns of behaviour and a careful investigation of their causes. As noted above, cognition is crucial to this approach and is likely to hold the key to progress in understanding ASD. This approach will help us to identify putative cognitive endophenotypes - that is, groups of behavioural symptoms with a common cognitive origin - of ASD. Individuals with such groupings of symptoms could then be considered as subgroups that may be associated with varying causes and consequences. Possible subtypes of autism symptoms are now being reported in the literature, with different performance profiles being seen when considered in a number of ways. The most obvious subgrouping could be argued to be that between autism and Asperger's Syndrome. However, the picture is far more complicated. For example, Helen Tager-Flusberg and her team have identified a variety of language profiles within those on the autism spectrum, while others have investigated other differences such as sensory processing difficulties. These subtypes have been grouped by their performance in a cognitive domain. Other studies are starting to focus on whether subgroups can be identified at the biological level, and at how these levels of analysis might interact. Once we can include reliable cognitive subtypes in studies recruiting behavioural genetic and neuroimaging techniques, we can hope to move closer to finding some more concrete answers concerning biological, cognitive and behavioural explanations of the condition. Likewise, the same can be said for consideration of environmental inputs that may aid or abet autistic functioning.
At this point it should be stressed that there are likely to be multiple causes and consequences of the condition that we term the autism spectrum, and there is no reason to expect that one cause will explain all facets of the disorder. In fact, recent genetic work suggests that different aspects of the cognitive phenotype are explained by different genetic contributions (6). In addition, we must seek explanations that can be useful for understanding and explaining the behavioural consequences of the condition, as illustrated through observed behaviours in daily life. A number of research groups are starting to make links between cognitive and/or biological components of autism and what is termed adaptive behaviour (how daily life skills are affected).
The complexity of the autism spectrum at many levels (at cognitive, behavioural and biological levels of description, for example) has become increasingly clear over the past decade or so. This complexity has led to an understandable difficulty in identifying the cause of the condition. However, this factor is not the only culprit. The general isolation of different research methods from one another in the quest to understand the autism spectrum has led to a variety of perspectives being adopted, none of which has been as successful in improving knowledge, understanding and experience of the condition as one might hope. In part, the reason for this is the need for an integrative approach to understanding the disorder in which different methods as well as comparison groups are combined within the same study (7). However, it is also critical that all of those invested in improving and supporting the lives of those affected in some way with ASD appreciate the level of detail and time that this quest will consume. At times, conducting research that does not appear to be addressing a cause directly will, in the longer-term, be hugely beneficial in this quest. If funding and knowledge is withheld from such domains and focused entirely on those seen to have a direct focus on cause (e.g. genetic studies), then we will be far slower to make the progress that we would like. This is not to say that genetic and other studies are not also critical in the endeavour to understand autism. However, it is important to provide the public with appropriate expectations of what scientific techniques can achieve, and provide scientists with the appropriate resources to use these. Of course, these are not arguments that are unique to the study of autism and issues relating to the use of (for example) genetic and imaging techniques, as well as the ethical issues that such methods raise more broadly, are being keenly debated in many areas at the present time.
Sources and References
1) Morton, J., & Frith, U. (1995). Causal Modeling: a structural approach to developmental psychopathology. In D. Cichetti & D. J. Cohen (Eds.), Manual of Developmental Psychopathology. Volume 1 (Vol. 1, pp. 357-390). New York: John Wiley.
2) Morton, J. (2004). Understanding developmental disorders: a causal modelling approach. Oxford: Blackwells.
3) Baron-Cohen, S., Leslie, A.M. & Frith, U. (1985). Does the autistic child have a 'theory of mind'? Cognition, 21, 37-46.
4) Hill, E.L. (2004). Executive dysfunction in autism. Trends in Cognitive Sciences, 8, 26-32.
5) Bowler, D. (2007). Autism Spectrum Disorders: Psychological Theory and Research. Chichester, West Sussex: John Wiley & Sons Ltd.
6) HappÃ©, F.G., Ronald, A. & Plomin, R. (2006). Time to give up on a single explanation for autism. Nature Neuroscience, 9, 1218-1220.
7) McGregor, M., NÃºÃ±ez, M., Cebula, K. & GÃ³mez, J-C. (Eds.) (2008). Autism: An Integrated View from Neurocognitive, Clinical, and Intervention Research. Oxford: Blackwell.