If you have spent any time around autism in education, you will have met the standard story: autistic people have difficulty with empathy, social communication, and reading other people. The job of the school, in that story, is to teach the autistic student the social skills they are missing.
The Double Empathy Problem turns that story around. It argues that the misunderstanding between autistic and non-autistic people is mutual. Both sides struggle to read each other. We have simply spent decades measuring only one side of it.
For schools, this is not a small adjustment in language. It changes who we think needs to do the work.
What the Double Empathy Problem is
The term comes from Dr Damian Milton, an autistic academic who set it out in 2012. His argument is straightforward. Empathy and understanding are a two-way street. When two people have very different ways of experiencing the world, of communicating, and of showing what they feel, they will each find the other hard to read. The gap between them belongs to both of them.
Autistic and non-autistic people often have genuinely different communication styles. So when they try to understand each other, both sides hit the same wall. The autistic person finds the non-autistic person confusing, and the non-autistic person finds the autistic person confusing, at the same time, for the same underlying reason.
The breakdown isn't located inside the autistic person. It's located in the interaction — and it runs both ways.
The traditional model only ever looked one way. It asked why the autistic person failed to understand others, and never asked why others failed to understand the autistic person. Once you ask both questions, a lot of what we see in schools looks different.
Why the one-sided model doesn't hold up
This is not just a nicer way of talking about autism. There is now a body of research behind it.
In one study, Catherine Crompton and colleagues (2020) passed a story down a chain of people, like a game of telephone. Some chains were all autistic, some all non-autistic, and some mixed. The all-autistic chains kept the detail just as well as the all-non-autistic chains. It was the mixed chains where information dropped away, and where people rated the interaction as less comfortable. Autistic people were not poor communicators. They communicated well with each other. The difficulty showed up specifically across the neurotype divide.
A separate line of work looked at first impressions. Noah Sasson and colleagues (2017) found that non-autistic people form negative impressions of autistic people within seconds, and that these judgements are based on style rather than on what the person actually says. When the same content was presented without the audio and visual cues, the bias disappeared. People were not responding to a lack of substance. They were responding to a difference in delivery, and reading it as a deficit.
Put together, these findings sit awkwardly with the idea that the problem lives inside the autistic person. They fit far better with Milton's account: this is a mismatch between two groups, not a flaw in one of them.
What it looks like in the classroom
Most teachers recognise the Double Empathy Problem the moment they see it described, because they have watched it happen. Here are some of the everyday moments where a one-sided reading gets it wrong.
What staff often see
A student gives a flat, one-word answer and doesn't make eye contact. It reads as rude or disengaged.
What may be happening
The student is concentrating hard on processing the question, and looking away is helping them do it. The eye contact that would reassure you is costing them the answer.
What staff often see
A student takes an instruction literally, or does the opposite of what was "obviously" meant.
What may be happening
The instruction relied on a hint that was clear to you and invisible to them. They answered the question you actually asked, not the one you meant.
What staff often see
A student seems "fine" all day, then has a difficult time at home, or falls apart on a Friday.
What may be happening
They have spent the day working hard to read a setting built for someone else, and to be read correctly by it. That effort has a cost, and it lands later.
In each case, the older model puts the difficulty inside the child. The Double Empathy Problem puts it in the space between the child and a setting that was not built with them in mind.
What schools can change
If the gap runs both ways, then so does the responsibility for closing it. That is the practical heart of this idea, and it points the work somewhere more useful than asking the student to mask better.
Be explicit, and treat that as good teaching
Say what you mean directly. Give instructions in plain terms, written down where you can, and check understanding without making the student guess at a hidden expectation. This helps every student, and it removes a barrier that only ever existed because we assumed everyone reads between the lines the same way.
Read autistic communication on its own terms
A flat tone is not the same as not caring. Looking away is not the same as not listening. Honesty that lands bluntly is not the same as rudeness. Learning to read autistic students accurately is a skill staff can build, and it is at least as important as anything we ask the student to learn.
Change the environment, not just the child
Sensory load, unpredictability, and constant social demand are real barriers. Reducing them is not lowering expectations. It is removing obstacles that have nothing to do with the learning itself.
Notice the cost of fitting in
A student who looks effortlessly fine may be working harder than anyone in the room to appear that way. "No problems at school" is sometimes the thing to look at most closely, not the thing that closes the conversation.
What this is not
This is not social skills training by another name, and it is not behaviourist compliance work. The point of the Double Empathy Problem is precisely that we stop asking the autistic student to do all the adapting. It is a neuroaffirming idea: autistic communication is different, not broken, and the job of the adults is to meet it halfway. Approaches that aim to make an autistic child appear less autistic are working against the grain of the evidence here, not with it.
Free download: the Double Empathy Problem staff toolkit
A one-page staff handout, five reframes to use in the room, and a set of discussion prompts for a department or INSET session. Built to use the same week. Tell me where to send it and it's yours.
Get the staff toolkit (free) →Common questions
Who came up with the Double Empathy Problem?
The concept was developed by Dr Damian Milton in 2012. Milton is an autistic academic, and he supervises my PhD at the University of Kent alongside Dr Triona Fitton.
Is the Double Empathy Problem actually supported by research?
Yes. Crompton and colleagues (2020) showed that autistic people pass information to each other as effectively as non-autistic people do, with breakdowns appearing mainly in mixed groups. Sasson and colleagues (2017) showed that non-autistic people form negative first impressions of autistic people within seconds, based on style rather than on what is said.
Does this mean autistic students have no real difficulties?
No. It means the difficulty in communication sits between people, not only inside the autistic person. Autistic students still face genuine barriers in school, many of them created by environments and expectations built around non-autistic norms.
How is this different from social skills training?
Social skills training asks the autistic person to change to fit everyone else. The Double Empathy Problem says the misunderstanding runs both ways, so the adjustment has to be mutual. The focus moves from changing the student to changing how staff communicate and how the setting is built.
Isn't this just saying everyone is different?
It goes further than that. It makes a specific, testable claim: that communication works better within a shared neurotype than across different ones, and that the failure we have historically blamed on autistic people is in fact shared. The research supports that claim.
References
Milton, D. (2012) 'On the ontological status of autism: the "double empathy problem"', Disability & Society, 27(6), pp. 883–887.
Crompton, C. J., Ropar, D., Evans-Williams, C. V. M., Flynn, E. G. and Fletcher-Watson, S. (2020) 'Autistic peer-to-peer information transfer is highly effective', Autism, 24(7), pp. 1704–1712.
Sasson, N. J., Faso, D. J., Nugent, J., Lovell, S., Kennedy, D. P. and Grossman, R. B. (2017) 'Neurotypical peers are less willing to interact with those with autism based on thin slice judgments', Scientific Reports, 7, 40700.
Bring this to your staff
I run talks, INSET, and workshops that take schools through the Double Empathy Problem and what it means in practice — grounded in current research and in daily work with autistic students.
See speaking & CPD →