In my last article, we talked about the double empathy problem and how autistic traits aren’t always so much ‘deficit’ as they are simply ‘different’. We focused on this idea that I called ‘social empathy’ or the ability to understand another person’s expectations and actions in a social situation. The double empathy problem paper I cited is broader than just social empathy though. It includes re-evaluating the ‘deficits’ of autism: ‘theory of mind’, ‘social interaction’, ‘problematic or unexplained behaviors’, etc. At the end of my last article, I said I wasn’t sure how much other research out there supported the ideas in the double empathy problem paper.
Well, I have great news friends. I found a paper! And technically not just one, but many. The paper we’ll talk about today reviews other papers and contains approximately 291 references (including new and old work). That’s a lot of research.
Reviewing multiple papers, comparing them, and assessing their findings against one another or together is critical for understanding the impact and findings of research on the whole, rather than one study at a time.
This paper has saved me a ton of work and reading by finding similar papers and discussing them together. Papers like this contain a ton of sources and information to look at in their references. This is really good news – it means the research has been worked on for some time now, and it’s just taking a while to bubble to the surface and usurp common (and often incorrect) beliefs about autism.
Now that I’m done gushing about its structure, let’s get on to the paper. We’ll be talking about ‘Empirical Failure of the Claim that Autistic People Lack a Theory of Mind’ by Morton Ann Gernbacher and Melanie Yergeau, published oh-so-recently in December 2019, and briefly how this relates to the double empathy problem’s take on theory of mind.
What is theory of mind?
To start, we should get clear on what ‘theory of mind’ means – it’s pretty abstract, right?
Theory of mind is typically described as the ability to recognize that one has a mind, that others have minds, and that those minds are different. This definition and the idea that all and only autistic people lack theory of mind pervades psychology.
“Most of us have a theory of mind in that we can guess what others are thinking and how that might differ from what we are thinking. Those with autism can be thought of as mindblind in that they cannot imagine what others might be thinking, or even that others are thinking. . . . To them, it would be like looking at the headlights of a car to determine why the car just did what it did, or what information it is trying to convey to us.”
—The Encyclopedia of Neuropsychological Disorders (Soper & Murray, 2012, p. 125)
However, theory of mind seems to encompass more than just this idea of knowing that you have a mind, and others have separate minds with different thoughts, experiences, perspectives, and ideas.
Theory of mind is also used to refer to skills such as reading another person’s emotions, general thoughts, facial expressions, or thinking from another person’s perspective. In more clinical terms, theory of mind seems to encompass ‘mindreading’ skills, or cognitive empathy, some social skills, and some executive functioning skills.
Theory of mind tasks, as we’ll see below, are quite varied and seem to encompass some of these other ideas as well, not strictly the italicized definition above.
This variety can make sense. Sometimes researchers use a ‘proxy measure’, or a measure that often occurs (strongly correlates) with the desired outcome and can be used when the desired outcome is unmeasurable or difficult to measure. Here, our desired outcome is theory of mind and our proxy measure might be cognitive empathy. The variety in definition also makes sense if theory of mind were to encompass several different measures in its definition. Then information could be collected on each of these sub-measures to create and combined to make a composite measure of one’s theory of mind. However these explanations are unlikely as the review paper discusses how some researchers admit that we don’t have a clear or singular definition of what theory of mind is, let alone a way to measure it.
Not having a clear definition or way to measure an outcome significantly impacts the quality of research done. Without a clear definition, researchers can’t be sure if they’re measuring the correct thing or even the same thing that other researchers are measuring.
So at the end of the day, I’m not sure how much the phrase ‘theory of mind’ means as an objectively measurable idea.
Theory of mind tasks
So what started the ‘deficit in theory of mind’ framework?
There are many theory of mind tasks out there these days, but the first was called the Sally-Anne test.
In the 1980s, Simon Baron-Cohen et al. (and company) studied a small group of children with and without autism using this test.
How they set up this task: The researcher will interact with a child using two dolls, and some props: a box, a basket and a marble. The researcher then explains that Sally has a marble and that she puts it in her basket. Sally walks away. Anne takes the marble and puts it in her box. Sally comes back, and end scene.
The researcher then asks the child some version of the following three questions to try to measure the child’s ability to think from Sally’s perspective, their perception of reality, and their memory respectively:
Where will Sally look for her marble?
Where is the marble?
Where was the marble in the beginning?
The primary question of interest for measuring theory of mind is question (1), thinking from Sally’s perspective. The others help to tell if the child understands the task and question.
I’d like to briefly introduce other, popular types of theory of mind tasks, beyond the Sally-Anne test mentioned above. The Sally-Anne test is of the False Belief task (second-order) variety in the list below.
Examples of Popular Theory of Mind Tasks*
* Copied from the authors’ Table 2: Examples of Popular Theory of Mind Tasks with minor edits for readability
Type of Task
False Belief task (first-order)
Participant is shown a container with which they’d be familiar, for example, a closed bag of M&M candies. Participant is asked to predict what’s inside. The bag is opened, and the participant is shown that their belief about the contents was false: The bag doesn’t contain M&Ms candies; instead it contain erasers. Participant is asked “What did you think would be in the bag before I opened it?” If participant answers with the name of the bag’s actual content (e.g., erasers) rather than the name of the bag’s expected content (e.g., candy), the participant fails the false belief task.
False Belief task (second-order)
Similar to a first-order False Belief task (as illustrated above), except that the participant is asked, “What do you think another person would think would be inside the box before I opened it?”
Strange Stories task
Participant listens to a spoken story that contains a spoken deception (e.g., lie), a figure of speech (e.g., a metaphor), a misunderstanding, a persuasion, or the like. Participant is required to orally explain why the person said what they said and what they were thinking when they said it.
Faux Pas task
Participant listens to a spoken story that contains a social interaction, such as a person showing newly bought curtains to a friend, who says they don’t like the curtains. Participant is required to identify whether “someone said something that they shouldn’t have” and, if so, orally explain why the person said something that they shouldn’t have, what they should have said instead, and what the person and their friends must have been thinking when the person said what they said.
Animated Triangles task
Participant views a series of animations with geometric triangles. After each animation, the participant is asked to orally explain “What happened in the animation?” Unknown to the participants, their oral answers are scored according to how likely to interpret the animated triangles as humans interacting and the number of emotional terms they provide in their oral explanation (e.g., if they say that one triangle was bullying another triangle).
Participant views only the eye region of numerous black and white photographs and for each photograph is required to select on emotional expression from a set of four emotion terms (e.g., terrified, upset, annoyed, or arrogant).
These tasks clearly show the variety in theory of mind research, and evidence the lack of a cohesive definition. Personally, I am unsurprised that autistic people are failing these tasks after reading these examples. There is a heavy emphasis on spoken communication, something I know I struggle with greatly, as well as seemingly unrelated tasks (to theory of mind), such as reading emotions from purely the eye area; both areas where autistic people typically experience impairment in. The authors acknowledge similar issues with the tasks.
Why are there so many different tasks? The table above is just a subset of theory of mind tasks. The authors mention one reason why we might have so many different theory of mind tasks is that researchers continue to create new tasks, claiming the old ‘lack sensitivity’ when they weren’t finding the results they expected. Some amount of creation and refinement is expected, but as we’ll see below, the evidence points in a different direction.
Problems with the theory of mind framework
Universality & specificity of theory of mind in autistic people
This paper starts out with a few points that I think we could see coming.
Past theory of mind research has stated that all autistic people lack or fail to develop theory of mind, and that this trait is unique to autistic people. I feel like “all and only” statements are destined to fail. What’s that Mark Twain quote? “All generalizations are false, including this one.” Or whoever said, “Always is never right and never is always wrong.”
And, as our lovely researchers Gernsbacher and Yergeau point out, the research doesn’t support this specificity (only autistic people lack theory of mind) or universality (all autistic people lack theory of mind).
The authors found other research that many other groups also failed theory of mind tasks, concluding that “the more atypical the child, the more likely they are to fail false belief tasks”. Some of the other groups that struggle with theory of mind tasks include:
children with specific language impairment, Down syndrome, Williams syndrome, Prader Willi syndrome, cerebral palsy, Fragile X, epilepsy, neurofibromatosis type I;
children who are blind or deaf/hard of hearing;
children exposed to prenatal smoking or drinking;
children with fewer rather than more siblings, with lower rather than higher socioeconomic status, with fewer rather than more adults living nearby
The last bullet may be of particular interest. These groups aren’t diagnoses, syndromes or disorders, but environmental conditions that still impact theory of mind performance.
In 2009, Baron-Cohen (the Sally-Anne test researcher from above) finally acknowledged that lack of theory of mind may not be specific to autistic people. Other researchers have been working for as many as 30 years to correct these misconceptions. Change though, can be incredibly slow.
That all autistic people lack theory of mind is an easy one to counter. All you need is one example, and allow me to add my anecdotal, lived, autistic experience: while I may have lacked theory of mind as a child, as an adult, I’ve been told that my perspective is appreciated because of my insights into the behavior and possible thoughts of others. I believe this is a learned skill, hard-won through many hours of practice, mistakes, and observations. I’ve spent my life trying to figure out how and what other people think. My experience runs directly counter to the idea that all autistic people lack theory of mind. And, just like in a previous article I wrote about proprioception and interoception having a skill component, it makes sense that you could work on your theory of mind and empathy skills over time too.
In research, you’ll hardly ever find a sample where everyone in a group responded the same way.
So then the question becomes: why do some autistic people fail theory of mind tasks, while others do not? The primary reason, according to the authors, is theory of mind tasks’ heavy reliance on spoken language skills. Spoken communication is a skill that varies widely from both autistic person to autistic person, and within a single autistic person, depending on what other stressors may be impacting the individual.
“why do some autistic people fail theory of mind tasks, while others do not? The primary reason, according to the authors, is theory of mind tasks’ heavy reliance on spoken language skills.”
In a population where people can be entirely non-speaking or experience non-speaking episodes, I am entirely unsurprised that autistic people would perform poorly on tasks that rely on spoken language – in addition to the disconnect in autistic-neurotypocal communication. The tasks used for gauging theory of mind appear biased in such a way that autistic people, and others who experience spoken communication challenges, will perform poorly.
What did surprise me though, was the extent to which language skills predicted theory of mind task performance. Research showed that up to three-fourths of a person’s performance differences on a False Belief task can be predicted by vocabulary and grammar and that up to one-half of those differences can be predicted by vocabulary alone. Multiple studies and meta-analyses conducted with groups of autistic and nonautistic participants found that vocabulary predicts theory of mind performance better than whether or not someone is autistic.
The issues of universality and specificity seem a bit trivial when compared to the findings of more recent research regarding theory of mind.
Predictive validity, convergent validity & replication
The authors also looked into work that examined theory of mind research’s predictive validity, convergent validity, and replication. Jargon alert! Here’s what that means:
Predictive validity: examining a measure’s ability to predict other, relevant measures
Convergent validity: examining if two measures looking at the same idea correlate, or show similar patterns in their results
Replication: examining if different studies looking at the same measures have similar results
In short, the answer to all of the above sample questions is, “No, it doesn’t appear so”.
I could list statistics and how many studies found contradictory evidence to the theory of mind framework, and I’ll do a bit of that below, but I want to be abundantly clear that the authors found plenty of evidence indicating that theory of mind performance doesn’t predict measures that we would expect theory of mind to be related to (e.g., autistic traits, empathy, various social skills); that performance on different theory of mind tasks often don’t correlate with each other; and that studies showing no tangible connection between theory of mind and autism outnumber those that do.
With predictive validity, we’re mostly looking at whether or not theory of mind performance predicts measures that we would expect theory of mind to be related to; measures like autistic traits, empathy, emotional understanding, and various social skills and behaviors (the authors have 12 measures). Using over 70 previously conducted studies with small and large sample sizes, the authors found that theory of mind performance was found to predict none of the 12 measures they examined. In fact, looking at these studies on the whole, theory of mind task performance actually doesn’t appear to pattern with anything other than language dexterity. Going back to our discussion above (under Universality), this makes sense as theory of mind tasks relies heavily on language skills.
Many theory of mind tasks do not correlate with each other. Even different versions of False Belief tasks fail to correlate with each other. This undermines theory of mind as a construct – indicating it’s likely not measuring what its expected to measure, and that it may not be measuring anything useful at all, considering the other findings cited.
The authors point out many seminal research studies that regularly fail to replicate. They cite other studies that attempted to reproduce the original works, with new work instead showing that autistic children were not underperforming on theory of mind tasks. In one case, 4 out of 4 known studies attempting to replicate the same results were unable to do so. They also discuss how some of the seminal works’ results are unusually strong in small samples, when “samples two to three times larger are needed to reliably test the somewhat obvious hypothesis…that people who like eggs are more likely to report eating egg salad”. Sometimes small samples can make insignificant relationships appear to be significant purely by chance.
When the evidence isn’t there
Despite the theory of mind framework’s popularity in both research and popular culture, there’s quite a bit of evidence countering its ideas. Although last month I wasn’t sure about what the research was saying, this review paper concludes in supporting the double empathy problem’s ideas about the deficit in theory of mind; agreeing that there’s more going on in this story.
There appears to be enough evidence for theory of mind research to consider a serious shift, perhaps performing more research to reassess the definition for theory of mind, to study theory of mind and related phenomena in other neurotypes (neurodivergent or not), to develop new tasks and measures that don’t rely so heavily on spoken language, and to validate the measures, ensuring they measure what they are supposed to measure.
For such a commonly cited phenomena, there appears to be little evidence that actually supports it. At some point as researchers, when things aren’t going the way we expect, we must do more than reexamine our measures, create new ones, and search new avenues for the answers we want to hear. We must go back to our original hypothesis and reassess if our original idea is incorrect. Richard Feyman puts it in pretty clear terms:
“I’m going to discuss how we would look for a new law. In general we look for a new law by the following process. First we guess it, [laughter] then we – oh don’t laugh, that’s really true – then we compute the consequences of the guess to see what – if this is right, if this law that we guessed is right – we see what it would imply. And then we compare the computation results to nature, or we say compare to experiments or experience. We compare it directly with observation to see if it works.
If it disagrees with experiment. It’s wrong. And that simple statement is the key to science. It doesn’t make a difference how beautiful your guess is, it doesn’t make a difference how smart you are, who made the guess or what his name is. If it disagrees with experiment, it’s wrong. That’s all there is to it.”
Impacts & closing thoughts
While many autistic people may struggle with theory of mind, it’s certainly not true of all autistic people. The sweeping generalization, combined with the mixed definitions of theory of mind, can cause lasting harm to autistic communities.
“[T]he erroneous claim that only autistic people, “together with robots and chimpanzees” lack a theory of mind and are therefore “biologically set apart from the rest of humanity in lacking the basic machinery” echoes throughout psychological literature, practice, and instruction.”
Even research papers use offensive ways to discuss this deficit, as exampled above. This statement that “echoes throughout” psychology reads clear to me: Autistic people are subhuman. It dehumanizes us. Experts comparing us to robots and monkeys makes it easier to justify selling bleach enemas as medicine and ABA therapy to “cure” autism and make autistic people “normal”.
It shocks me that experts (especially those typically dealing with children, the most innocent & in need of protection) would condone such dehumanization in the people they’re supposedly trying to help, let alone use that language themselves. Autistic people are just as human as any other person. Our neurotype evolved with humanity, just as many others did. Dehumanization is used to justify all manners of abuse and cruelty, both in and out of the autism space. It is disturbing how we so easily talk about other humans as if they aren’t human, as if they’re somehow less because they’re not like the majority.
With ample evidence to show the flaws in theory of mind research, why are these stereotypes still floating about? I can’t say for sure – sometimes ideas stick and hold, even when they’re not true. But it at least appears the tides are beginning to shift as researchers open their ears to autistic communities and continue to question the knowledge dutifully passed down from tradition. But how do we help change the stereotype?
We must continue to share our perspectives, to become researchers, to work with researchers, to share our stories and our voices, to not let the stereotypes limit us. Silence cannot drive change. Autism acceptance is on the rise, and it’s only a matter of time before old stereotypes can no longer hold up to evidence.
What do you think? Have you felt impacted by theory of mind stereotypes? How does this old research and language used with autistic people make you feel? Does the newer research leave you hopeful? Share your thoughts in the comments below.
Sullivan is an autistic yoga teacher & researcher striving to share more coping tools (such as yoga & meditation) and research (like this article) with the neurodivergent community and beyond. Blending her background in psychology and mental health with yoga, Sullivan strives to share the peace, self-acceptance, and physical awareness yoga and mindfulness has brought her. Sullivan also has a background in psych and market research, and is now using these skills as the resident researcher at Planet Neurodivergent.
Gernsbacher, M. A., & Yergeau, M. (2019). Empirical failures of the claim that autistic people lack a theory of mind. Archives of Scientific Psychology, 7(1), 102–118. doi: 10.1037/arc0000067