Many of us were trained to see autism as an empathy problem. Autistic people, we learned, struggle with “theory of mind,” the ability to read what others think and feel. That framing shaped diagnostic criteria, interventions, and a lot of what happens in therapy rooms.
Then an autistic researcher asked a better question. What if the difficulty runs both ways?
What Milton proposed
In 2012, Damian Milton named the double empathy problem. When two people experience the world very differently, both struggle to read each other (Milton, 2012). Non-autistic people misread autistic people. Autistic people misread non-autistic people. Only one group gets a diagnosis for it.
That’s the heart of Milton’s argument. The group whose communication style counts as the dominant norm gets to call the other group impaired, even when the misunderstanding is shared.
What the research shows
Crompton and colleagues tested this with a structured version of the telephone game. Seventy-two adults passed a story along chains of eight people. Autistic chains didn’t differ significantly from non-autistic chains, while mixed chains lost detail faster and reported lower rapport (Crompton et al., 2020).
Autistic people communicated well with each other. The trouble showed up where neurotypes mixed.
First impressions tell a similar story. Across three studies, non-autistic observers judged autistic people less favorably within seconds and were less willing to interact with them. That bias disappeared when observers judged conversational content without audio-visual cues, which pointed to style over substance (Sasson et al., 2017).
The content was fine. The difficulty came because of how it was delivered.
Where the evidence is still developing
We’re still learning more and more as research emerges. In later work from the same research group, mixed neurotype groups shared information as well as matched groups, while people still reported higher rapport with others of their own neurotype (Wilks & Crompton, 2024).
So hold the telephone game result with some care. The finding that keeps showing up is that autistic people can communicate well with each other, and that connection often feels easier with people who share your neurotype.
Double empathy also doesn’t erase disability. Many autistic people have real support needs that a better conversation partner won’t solve. The frame changes where we look for a communication problem. It doesn’t ask anyone to pretend support needs aren’t there.
Why this matters clinically
If social difficulty is shared, interventions aimed only at autistic communication are working on half the problem. They also send a message about whose way of being is correct. That message can feed masking, and masking carries real costs.
It changes how we read our own sessions too. When rapport feels hard with an autistic client, the usual instinct is to locate the problem in the client. Double empathy asks us to check our side first:
Our pacing
Indirect or vague questions
How much we rely on facial cues
Small talk we treat as a required warm-up
What to do differently
Treat communication breakdown as a shared event, and say so out loud with clients.
Ask autistic clients how they prefer to communicate, then adapt where you can.
Be specific and direct. Fewer hints, more clear questions.
Notice whether your treatment goals ask only the autistic person to change.
Value connection with other autistic people. Those relationships can be a source of ease, and they aren’t a sign of withdrawal.
The gap is the point
For a long time, our field looked at a gap between two people and saw a deficit in one of them. Milton’s Double Empathy Theory asks us to look at the gap itself.
That puts the clinician inside the communication. You aren’t grading it from the outside. If a session feels clunky, you’re in that clunkiness too.
Your autistic clients have likely spent their whole lives being told they’re the ones getting it wrong. Many of them believe it. You can be one of the first people to say, “Let’s figure this out together. Both of us.”
The work isn’t teaching autistic people to translate themselves. It’s learning to meet them halfway. Sometimes more than halfway.
References
Crompton, C. J., Ropar, D., Evans-Williams, C. V., Flynn, E. G., & Fletcher-Watson, S. (2020). Autistic peer-to-peer information transfer is highly effective. Autism, 24(7), 1704-1712. https://doi.org/10.1177/1362361320919286
Milton, D. (2012). On the ontological status of autism: The ‘double empathy problem’. Disability & Society, 27(6), 883-887. https://doi.org/10.1080/09687599.2012.710008
Sasson, N. J., Faso, D. J., Nugent, J., Lovell, S., Kennedy, D. P., & Grossman, R. B. (2017). Neurotypical peers are less willing to interact with those with autism based on thin slice judgments. Scientific Reports, 7, 40700. https://doi.org/10.1038/srep40700
Wilks, C. E. H., & Crompton, C. J. (2024). Communication between autistic and non-autistic people. Salvesen Mindroom Policy Briefing, number 15. https://salvesen-research.ed.ac.uk/sites/default/files/2024-11/Briefing%2015%20-%20Communication%20between%20autistic%20and%20non-autistic%20people%20-%20plain%20text.doc.docx