I attended the European Human Behaviour and Evolution Association's 2021 online conference, and presented this abstract, poster and short audio clip.
I received some interesting comments and feedback, written and in the socialising sessions afterwards, and detail some of those, and my responses, at the bottom of this post.
Abstract, poster and short audio clip
"Just-so"? No! Evolutionary hypothesising strengthened, systemised, and applied to autism
Evolutionary scientists could provide illuminating ultimate explanations of mental disorders and other cognitive traits, but are often criticised as 'just-so storytelling'.
Here I provide a novel systematic method of reviewing available evidence, combining principles from evolutionary psychology and evolutionary psychiatry. I apply it to autism spectrum disorder as an example, distinguishing functional and dysfunctional subtypes and strengthening Baron-Cohen’s account of non-dysfunctional autism as a hyper-systemising cognitive specialisation.
Just as bio-archaeological evidence can be used to infer the evolutionary history of physical traits, genetics, epidemiology and anthropology can be used to infer the evolutionary function of psychological traits. Formulaically systemising these principles of inference in the ‘DCIDE method’ (an acronym for its stages of Description; Categorisation; Inference; Depiction; Evaluation) can optimise evolutionary hypothesising.
Applying this method firstly categorises autism into functional or dysfunctional subtypes: around 10-20% of cases can be attributed to dysfunction via de novo mutations and environmental insults, often associated with intellectual disability. The remaining cases are caused by common alleles, without signs of clear neuropathology, thus eligible for adaptive explanations. Evidence of autism’s prevalence, onset and course, and male-skewed sex ratio then provide inferences of the function of autistic cognition, which support Baron-Cohen's hypothesis over alternatives.
The DICDE method can be widely applied to any trait of interest, and further developed to increase its precision. Although particularly useful for understanding heterogenous psychopathology labels which blend with normality, its systematic approach may also lend strength to hypothesising in evolutionary psychology and biology more generally. The evidence-based, evolution-framed results it produces cannot be waved away as ‘just-so’ storytelling.
(Download the pdf to be able to zoom in on parts you can't see)
Comments and additional points
Here is an exchange between a commenter (who I'm keeping anonymous) and myself, made in writing on the online platform:
Commenter: Hi Adam, Thanks for the very clear talk through of the poster. I just had a few questions: 1. In terms of use / application: Is this for hypothesis generation or hypothesis evaluation (or both?)? 2. Is there anywhere in the steps where you evaluate the quality or is that an inherent expectation for each step? Eg, in the describe stage how does one decide descriptions available (or you make) are reasonable, and in the infer stage how does one approach potential biases in evidence, etc. 3. In the categorize stage what happens when the consequences of certain traits are context or frequency dependent? In the poster I think it's suggesting that environmentally-caused is classed as not adaptive (and I assume the suggestion is not that phenotypic plasticity is maladaptive!), But does this only work for traits where trait expression is relatively independent of environmental / epigenetic effects? Hope my questions are clear!
Me: Hi commenter, thanks for the excellent questions. These are all points which I have worked on but couldn't fit on the poster! Appreciate you bringing them up. 1. Application as hypothesis generation or evaluation? It can be both! Where there are multiple current hypotheses they can be evaluated, where there aren't many, they can be generated. I think an ideal practical application is as a new format for systematic review. 2. Quality evaluation is indeed an inherent expectation! Each stage can be optimized, so eg factor analytic descriptions of disorders / traits as done in the HiTOP model are better than current descriptivist DSM categorizing, in the same way as RCTS are preferred over case studies etc. Obviously confidence in the conclusions and the evaluation stage should take into account evidence quality - and when better standards are developed (or old ones called into question) then you have to update. This could appropriate methods from current systematic reviews and meta-analyses which face this problem of combining multiple experiments into a single framework. 3. There are a few separate issues here: firstly, the 'environmentally caused' is particularly misleading, apologies, I just didn't have space while keeping the font size reasonable! I meant toxins, physical damage, etc. Should maybe have said geneXenvironment can categorize as adaptive, pure environment is disease. Context and frequency dependence are very important and it's a shame I didn't have the space to go into them here because they do involve specific categorizing and inferring criteria - regarding frequency-dependence, the trait should be present in a minimum of around 0.5- 1% of individuals, assuming relevant social group sizes of about 100-200 where selection can maintain a minimum of one individual per group (that 1% is mentioned in the poster but not explained). Obviously the prevalence then carries inferences about the function - the correct adaptive explanation has to explain the prevalence & trait distribution/spectrum. Context dependence categorization requires the trait to show in cross-cultural populations within the same context so if depression / jealously / whatever doesn't exist in hunter-gatherers activated by similar contexts, then it can't be an adaptation - that's the categorization - and then for those which do categorize as potentially adaptive, you can infer the function of the trait as adaptive given that context (which is basically what we do anyway).
Here is a second exchange made on the same comment system:
Commenter: Hi Adam, I don't have headphones right now so apologies if this is in the verbal element, but I don't see integration here of the sensory processing issues in ASC, nor the new work on the dyadic nature of the social sides (the double empathy problem with neurotypicals driving a chunk of the problem) Can you comment on that? Also, do you think the fact ASC is often missed in girls means our understanding of sex ratios needs reassessment?
Me: Hi Lynda, thanks for these questions, nice to hear from someone versed in neurodiversity! These questions weren't covered in the audio, so I'm glad you asked. 1. You're right that the sensory processing issues haven't been integrated- space is sadly limiting. These findings would be incorporated into the 'depict' stage along with memory, special skills etc. The crucial point here is whether the sensory issues also arise in more traditional living people, and then how they manifest. They could either be costs or benefits - but my speculation is that autistics in hunter-gatherer groups mainly benefit from their cognition regarding memory and their area of interest / obsession, and that the sensory processing issues are heightened in modern environments. 2. Milton's double empathy problem is interesting and a useful conceptual tool, but isn't really a causal model and doesn't identify any particular evidence which would be integrated into the DCIDE method. And the fact that autism is rare enough that only about one autistic would exist per hunter-gatherer social group implies that autistic-autistic communication would have been very rare before we lived in larger groups and settlements, so the autistic-neurotypical differences in social communication are likely as old as autism itself, very common, and need to explained. The double empathy framing is more useful as conceptual prompt to realise our labelling of autistic communication styles as disabled or disorderly is primarily normative.
3. The sex ratio question is interesting, and any new evidence can be incorporated into the DCIDE method to see whether it confirms / disconfirms any particular hypothesis (which it definitely does). I think the current push to identify autistic girls will help us have a more realistic view of what autism looks like in females, and that will better inform our hypotheses of its possible adaptive function. I'd probably still expect the memory / systemizing adaptive explanation to be well supported.
And finally, a rendition of remarks made in a verbal conversation I had in the socialising session:
Commenter: I found your poster very interesting, but I was left slightly confused as to what it means we should do. Are you trying to rewrite the DSM? How can this be applied?
Me: Right, it is unusually big picture. I'm not trying to rewrite the DSM (although arguably it needs rewriting). The DSM is broadly useful for practising psychiatrists, who just need to identify a problem and prescribe a treatment. The DCIDE method is more about identifying cause - its more scientific than practical. I think a good use would be as a new format of systematic review. We could have yearly reviews of conditions/traits, incorporating all the new evidence, and seeing what the evidence now implies about functional/dysfunctional subcategories and the likely adaptive function. Whatever the correct theory to explain autism, schizophrenia, and so on, it will have to explain all the available evidence and should be framed in terms of evolutionary theory. This is what the DCIDE method forces us to do, in a methodical way currently missing. Such a review method should help to make sense of the vast array of evidence gathered by multiple disciplines, and give us a better idea of the true explanation.
Comments