Autism and Scientism
Why science is not always the best way to learn about autism
This article appears in the first issue of the Middletown Centre for Autism Research Journal (open-access).
There are many ways for people to come to understand the world. Many different approaches to learning about things, including minds. Scientism — the belief that science is the only route to useful knowledge — is a philosophical mistake.
I say this as someone who loves science, who teaches it for a living and who’s in the middle of another science course right now, for general interest and with an eye to future research. Science is wonderful. We just need to be careful about how we apply it, and what ways of knowing we risk crowding out if we rely on it too heavily.
When it comes to autism, people sometimes rely on scientific studies to the point of disbelieving autistic people’s personal experiences. Despite the low quality of much of the published research on autism, non-autistic experts are assumed to understand autistic experiences better than the people having them. This is a serious problem in a number of ways, and also an interesting case study in the limitations of science.
The scientific method, roughly speaking, consists of forming hypotheses and models, making predictions, and testing them through experiments and careful observation. There is a lot more to science than that in practice, and it can be difficult to pin down exactly what the very disparate practices in different branches really have in common, but as far as it makes sense to talk about ‘the scientific method’, it’s generally agreed to take something like that form.
This approach to discovering the truth about the world we live in has proved phenomenally successful. The insights of physics brought us the industrial revolution, and drive technological innovation to this day. Modern medicine would be inconceivable without modern chemistry and biology. The successes of these fields have driven people to seek similar insights using related approaches in anthropology, economics, psychology and across the social sciences. The demand for rigorous evidence and testable hypotheses has borne many fruits.
However, the more complex the systems we study, the harder it is to capture their behaviour in models, and the more likely we are to meet emergent phenomena, and scenarios where established theories break down. Physics students get used to problems involving perfectly smooth surfaces and flawlessly spherical objects, because we know the models we use are only approximations. In the real world other factors come into play: we know that cows aren’t strictly spherical, but let’s just say they come close enough for our purposes.
Humans are complex. We’re complex even as individuals, and we’re inescapably embedded in societies that are far more complex still. Doing good science on people is really hard. We constantly have to rely on very heavily simplified models, because a complete explanation of almost any human behaviour would really have to bring in — at minimum — cellular biology, anatomy, electrophysiology, endocrinology, social and developmental psychology, evolutionary biology, anthropology, sociology and political economy.
Nobody even attempts to explain things that comprehensively, for fairly obvious reasons; instead, we get used to looking at different levels of explanation and doing what we can by working at one or two levels at a time and just keeping half an eye on when other levels might impinge. Sometimes we get very satisfactory results that way, but other times we might miss really important parts of the picture, or settle for explanations that are plausible but completely wrong, like when social psychology tries to explain what turn out to be innate neurological differences.
Rigorously testing models and hypotheses requires large sample sizes and control groups, and those can be very hard to come by when you are studying humans. Ideally we would look for independent replication of findings, but in practice this often fails. Partly because of those difficulties, the study of humans is vulnerable to systematic biases of various sorts, including conflicts of interest and publication bias.
None of this is to say we shouldn’t try to apply scientific methods to the study of human behaviour and experience. Rigour and strong evidence bases are absolutely worth aiming for, and science is one extremely powerful way of getting them. However, not all evidence is scientific evidence. If someone tells you it hurts when they get kicked, you hopefully wouldn’t hold out for peer-reviewed studies to validate their pain. That’s certainly not how courts of law do it.
Similarly, if we have multiple testimonies from people who say that they were traumatised by autism interventions like Applied Behaviour Analysis (ABA), we should take that seriously. ABA is described as a scientific approach to understanding and modifying behaviour, although it based in the behaviourist paradigm, which has not been mainstream in psychology for some decades; and various meta-analyses have found the scientific evidence for ABA working at all is weak.
Many autistic people who have been through ABA and similar ‘therapies’ like PBS say that they have experienced such interventions as abusive, and we shouldn’t wait around for strong scientific evidence of post-traumatic stress before taking people seriously when they tell us they have been traumatised. The fact that published studies on ABA have systematically failed to investigate harms needs to be factored in to any evaluation, along with the many undisclosed conflicts of interest among the people behind those studies.
More generally, if a person wants to understand autistic experiences or work with autistic people, it is a profound mistake to rely solely on scientific sources. The science of autism is far from mature, and cognitive theories of autism have attained great prominence seemingly based more on their proponents being famous scientists than on solid evidence. Most of the supposed evidence for these theories lacks what’s called ‘face validity’, in the eyes of many of the people being studied — that is, it doesn’t look like it’s measuring what it’s supposed to be measuring at all. Too much autism research has been done without autistic input, which could have prevented data being misinterpreted, flagged up when studies’ goals bore no relation to autistic wellbeing, and prevented major errors of omission.
The failures of autism science are not random: they reflect systematic power imbalances. The central imbalance here is between non-autistic people and autistic people, who are usually only included as subjects: their perspectives are treated as data, if they are reflected at all. There is also a major imbalance favouring those whose research fits in with the broader priorities of the scientific-medical establishment or the autism industry, who control most of the funding. This shapes both the kinds of questions that get asked, and the types of answers that are most likely to get published: it introduces systematic errors, favouring research questions and results that suit those paying for the research. These issues are related to the chronic and pervasive problems with researchers on interventions like ABA failing to disclose their conflicts of interest, and to investigate evidence of harm.
There are also more fundamental problems stemming from the framing of autism in terms of dysfunction and disorder. Suffering is assumed to follow from the simple fact of being autistic; researchers concern themselves with reducing ‘symptoms’ of autism, rather than helping autistic people to thrive; differences are assumed to be deficits. Such unexamined biases get built in to the definitions used by scientists, and end up deeply embedded in the research; but they are not, themselves, scientific in any sense. All of these considerations make it rational to approach scientific findings in the field of autism with caution, and a critical mind.
Fortunately, there are other ways to learn about autism, just as there are other ways of learning about humans in general. By privileging science above all other modes of enquiry, scientism irrationally excludes learning from art, literature and conversation. Many autistic people are quite capable of communicating for ourselves: telling you about what it’s like to be us, and what that means. Many of us have written books and blogs. For those who don’t speak, there are often technologies which can help (AAC). The 2021 Interdisciplinary Autism Research Festival showed the extraordinary insights into the field that autistic researchers can bring, drawing on science but also humanities and the arts; likewise Autscape and The Neurodiversity Reader, although not all contributors are researchers in the traditional sense.
If you want to understand autism, read what autistic people have written; talk to us about our experiences; try to understand autism at an emotional and narrative level. When it comes to autism science, look for work where autistic people have been involved at every stage, from setting priorities to evaluating findings. Where that hasn’t happened, question why not. Objectivity is valuable, but understanding subjective experience should be a priority when it comes to humans.
More and more autistic people are talking to each other; comparing notes, learning about common experiences and ways we might be unusual even within the autistic population. Sharing strategies, brainstorming about barriers. The autistic community has collectively worked out a lot of things that we might never have got to on our own, as well as providing support networks for a population that too often lives in isolation.
There are dangers in relying on anecdotal information, of course, but in the absence of really solid science, it is often the best we can get. Weak scientific evidence doesn’t trump extensive personal experience, but however we approach autism, we need to be wary of motivated reasoning and cognitive biases. Wishful thinking has led many people astray when it comes to autism — not least among them, parents desperate to do what’s best for their autistic children. Unfortunately, faith in science has been no protection against these pitfalls, and may even have made some people more vulnerable to falling for things which sound scientific.
We might hope that one day, autism science will catch up with what autistic people already know, and science will become a truly valuable tool for understanding autism and what works for autistic people. I think it’s plausible — scientific investigation remains an unparalleled tool for making sense of natural phenomena. Autism surely falls into that category, and autistic people deserve high-quality scientific evidence, rigorously evaluated. In the meantime, though, we have many other ways of understanding humans… and autistic people fall into that category, too.
