Autism & theory of mind—what’s new?

I recently started reading Joanne Limburg’s book Letters to My Weird Sisters: On Autism and Feminism. She opens up with a wonderfully relatable metaphor for what it’s like to be autistic in a neurotypical world.

Here is my adaptation:

Allistic brains travel on the highway. They go from point A to B the same way every time, with no stops along the way. It’s easy for allistics to communicate with other because the highway route is easy to spot on the map and the route is the exact same for everyone. Therefore, allistics can all relate to each other and share the same experiences without having to explicitly name them.

Autistics, in contrast, use a completely different map to get from point A to B. Our brains travel on the side roads. They stop along the way, take different routes every time, experience more traffic, and see more of the city. This also means that others don’t take the same exact journey as us and therefore there is no unspoken shared understanding of what our journey is like. Importantly, without a map marking where we travelled on the side roads, allistics don’t see what we see.

To me, this metaphor captures the current state of understanding of theory of mind and autism.

What is Theory of Mind?

Theory of mind (ToM) refers to the ability to understand another person’s mental state.^[1]Does the chimpanzee have a theory of mind?, David & Woodruff (1978) For example, understanding someone’s intentions and beliefs. ToM ‘impairment’ is one of the most studied topics in the autism literature as it is thought to be the cause of many of our social communication struggles.

Conventional views of ToM and autism

The relationship between ToM and autism originated from research conducted by Dr. Simon Baron-Cohen. He first claimed that autistic children were incapable of ToM in 1985 based on the False-Beliefs Task.^[2]Does the autistic child have a “theory of mind”?, Baron-Cohen et al. (1985) This task requires children to recognize that someone will have an incorrect belief because they do not have the same information that you do about a situation.^[3]Beliefs about beliefs: Representation and constraining function of wrong beliefs in young children’s understanding of deception, Wimmer & Perner (1983)

Since the original study, many critiques have emerged surrounding the validity of the findings. For example, one critique shows that the line of questioning in the False-Belief Task is inherently ambiguous due to the way the questions are asked.^[4]Why Do Children Who Solve False Belief Tasks Begin to Find True Belief Control Tasks Difficult? A Test of Pragmatic Performance Factors in Theory of Mind Tasks, Schidelko et al., (2021) Other studies have identified that the original findings are also not reliable. For example, studies find that autistics do show ToM in other types of tasks or in real world examples.^[5]Conceptual issues in autism spectrum disorders, Gallagher & Varga (2015)

Yet, even after pointing out issues with the original study and with ToM research as a whole, there continues to be a plethora of studies linking autism to poor ToM. What is extra disappointing is that most of these newer studies still don’t capture the autistic experience in a way that is validating.

Harmful narratives

Many within the autistic community feel that ToM research has caused more hurt than good. In particular, the motives, language use, and biases of researchers conducting these studies have generated harmful and invalidating narratives about autism.

Instead of embracing neurodiversity, this research denies agency and humanity to autistics.^[6]Clinically Significant Disturbance: On Theorists Who Theorize Theory of Mind, Yergeau (2013) These harmful narratives range from claiming that we don’t have empathy, to we are less human (because it is believed that ToM is what distinguishes humans from animals).^[7]Mindblindness: An Essay on Autism and Theory of Mind, Baron-Cohen (1997) In a research context, this even translates to the delegitimization of autistics who conduct their own research (because how can someone who “doesn’t understand humanity” possibly conduct valid research?/sarcasm). Further, if “lacking ToM” means that we are “less human,” then the underlying social agenda is not true neurodiversity acceptance and inclusion but a means to justify exclusion.

I resonate with the way Yergeau & Huebner describe this field of research by saying that “essentialist understandings of ToM reveal a limited theory of other minds—one that privileges neurotypical minds and ignores all others, while simultaneously working to deny, undermine, and delegitimize autistic concepts of identity and community.”^[8]Minding Theory of Mind, Yergeau & Huebner (2017)

Let’s break this statement down using the double empathy problem:

The double empathy problem

The double empathy problem recognizes that it’s easier to understand the mindset of people who are more similar to you. In contrast, it is more difficult to understand the mindset of those who hold different norms and expectations to you.^[9]Double Empathy, Milton et al. (2020)^[10]On the ontological status of autism: the ‘double empathy problem’, Milton (2012) For example, autistics are better at understanding other autistic brains compared to neurotypical brains, and allistics are better at understanding other allistic brains compared to autistic brains.

Crompton et al. give the example that it “might be difficult for non-autistic parents to understand what their autistic child is feeling” or that it might be difficult for autistic people to describe their sensory sensitivities in a way that neurotypicals can understand since sensory sensitivities are not common neurotypical experiences.^[11]Double Empathy: Why Autistic People Are Often Misunderstood, Crompton et al. (2021)

Going back to the earlier map analogy, if most neurotypical brains travel on the highway, when a neurotypical correctly assumes that their friend traveled on the highway as well, it’s not really a ToM skill. Instead, it’s simply imposing your own way of thinking onto others. But the assumption is correct most of the time because neurotypicals often operate similarly to each other.

When we take this into consideration in ToM research, it’s easy to recognize why studies tend to be biased against autistics. Non-autistic researchers are at the forefront of this work and they are essentially asking autistics to identify the route they took, expecting the answer to be the highway route instead of the side street route. And when we give them the side street route as our response, they don’t recognize it as a valid route because it’s not highlighted on their own maps.

Evidence for the double empathy problem

Research conducted in the last 10 years is finally beginning to uncover the relationship between autistic ToM and the double empathy problem. For example, traditional studies show that autistics struggle to identify neurotypical facial expressions. However, this actually goes both ways. Neurotypicals also struggle to identify autistic facial expressions.^[12]Can Neurotypical Individuals Read Autistic Facial Expressions? Atypical Production of Emotional Facial Expressions in Autism Spectrum Disorders, Brewer et al. (2016)

Additionally, it’s been shown that neurotypicals are ineffective at interpreting autistic behaviours. This suggests that social interaction difficulties in autism may be a result of misunderstanding different communication styles between autistics and neurotypicals and not an inherently autistic trait.^[13]How Easy is it to Read the Minds of People with Autism Spectrum Disorder?, Sheppard et al. (2015)

What seems to be more likely than a ‘deficit in ToM’ is a lack of reciprocity and mutuality.^[14]On the ontological status of autism: the ‘double empathy problem’, Milton (2012) Not surprisingly, this lack of reciprocity has negative consequences to autistic mental health. For example, constantly being misperceived by the neurotypical majority has been associated with autistics feeling obliged to withdraw from society, camouflage and mask, and feel isolated.^[15]Autism and the double empathy problem: Implications for development and mental health, Mitchell et al. (2021)

Given these findings, it’s important to understand how the double empathy problem might play out outside of a research setting. Here is my take on a few reasons why autistics often get labeled as having ‘low ToM’ in the real world:

Common autistic traits leading to low ToM

Bottom-up thinking

Autistics are more commonly bottom-up thinkers. This means that we build a big picture or make a decision based on compiling all the pieces of evidence or data. It also implies that until we have all the data we need, we are unable to see what the big picture is. This way of processing is in direct conflict with many ToM scenarios.

There necessitates a level of assumption attached to how neurotypicals classify ‘high ToM.’ It involves assuming you understand and know what the other person is thinking and feeling without asking for explicit information. This matches a more top-down way of thinking, which is less common among autistics.

Also, the language used in ToM literature often conflates being able to ‘infer’ a mental state with being able to ‘understand’ a mental state. To me, these two things have very different meanings. Autistics are usually great at understanding things. Because we are often bottom-up processors, we will ask questions and seek out information until we understand a situation. But, many ToM tests don’t measure understanding, they measure inference. We aren’t meant to ask questions about the other person, we are just meant to make an assumption based on the limited data we have.

This also relates to heuristics—or mental shortcuts—used in decision-making. Neurotypicals save time when making decisions by tending to fall back on social norms instead of filtering through all the incoming information. Back to my analogy for example, this looks like someone jumping from the thought that “I travel on the highway and I see lots of other cars on the highway” to concluding that “everyone must drive on the highway.” They didn’t take the opportunity to see whether there were also cars driving on the side streets.

Since autistics tend to be less sensitive to heuristics, we may not be able to make assumptions about people’s mental states without more information. Importantly, this means that if we are given the information we need, we would understand their perspective.

Consistency between actions and values

One common line of evidence used to claim autistics have low ToM is that autistics are said to show “atypical moral judgement.”^[16]Impaired theory of mind for moral judgment in high-functioning autism, Moran et al. (2011) In particular, this is referring to how autistics tend to assign blame when an individual inflicts harm even in situations where the intention was not harmful. However, I don’t view this as poor ToM. You can be aware of a person’s good intentions while still recognizing that their actions caused harm. And in most cases, the impact of an action is more important than the intention. For example, even when you accidentally step on someone’s foot while walking, you are likely still going to apologize for what you did.

My take on this ‘evidence’ is that instead of showing a ToM deficit, it shows a recognition that for many autistics, the impact of actions is more important than the underlying intentions.

Researchers further reinforce the relationship between ‘atypical moral judgement’ and poor ToM when they label “acting in alignment with your values” as something “bad.” For example, a recent paper gave individuals the choice between incurring a personal loss but funding a morally good cause versus receiving a personal gain but funding a morally bad cause.^[17]Right Temporoparietal Junction Underlies Avoidance of Moral Transgression in Autism Spectrum Disorder, Hu et al. (2021) They found that autistics were more likely than allistics to support the good cause at a personal loss. Moreover, autistics made the same choice even when the choice was private and no one would know if they chose to support the bad cause for more personal gain. In comparison, in private, allistics tended to switch their choice to funding a morally bad cause for personal gain. Amazingly, the researchers claim these findings demonstrate that autistics have low ToM. They conclude that autistics are unable to recognize situations when making a moral transgression is beneficial and without social reputation costs.

I am still in shock about how this paper was able spin these results in a way that portrays a negative image of autistics. In my mind, what this shows is that autistics act with integrity. Unlike allistics, our moral values are not contingent upon social and situational circumstances. Our actions will align with our values even if that comes at a personal cost. This has nothing to do with ToM.

Expressions of empathy

Autistics often like to show that we empathize with someone’s experience by sharing our own experiences. This can clash with common expressions of empathy in neurotypicals. It is often viewed as being insensitive or trying to make the situation all about ourselves. However, I think many of us actually view this as a demonstration that we do have ToM. Contrary to popular belief, when we share our own experiences in relation to others, we aren’t trying to centre ourselves or diminish the other person. Rather, we are trying to show that we can relate and that we understand how the other person is feeling because we have experienced something similar.

Where do we go from here?

Moving away from outdated ToM views means recognizing that the double empathy problem can apply to any type of mismatch. For example, the above traits are common but not universal in autism and any non-autistic can display these characteristics as well. Moreover, these characteristics can be found across other contexts but are simply labeled as differences in values, in cultures, and in communication styles. They can even be compared to speaking two different languages. It’s only in the context of autism that they are viewed as a “ToM deficit.”

From a neurodiversity-affirming perspective, thinking about this phenomenon as a social construct makes the most sense. This means that the situation is also malleable and the way that we treat each other and spend time with each other affects our mutual ToM. Research even shows that when pairs of autistic and non-autistic individuals spend time in discussion together, it fosters understanding, connection, and empathy.^[18]Overcoming the Double Empathy Problem Within Pairs of Autistic and Non-autistic Adults Through the Contemplation of Serious Literature, Chapple et al. (2021)

What are your thoughts on ToM and autism? Do you find the idea that autistics have a ToM ‘deficit’ socially constructed? Are there any other traits you can identify which may get misinterpreted as evidence of low ToM?