When the Empathy Deficit Isn’t Ours: What a New Meta-Analysis Gets Right About Autistic Feeling
For decades, the empathy deficit has been treated as gospel in autism science. We’ve seen it in diagnostic tools. In teacher trainings. In parent guides. In the stunned reactions of strangers when we don't mirror their tone. Autistic people, we’ve been told, don’t get others. Can’t feel what they feel. Struggle to care.
It’s a narrative that has justified everything from social exclusion to compliance therapy. And now — finally — it’s being taken apart by the data.
A new systematic review and meta-analysis, published in Clinical Psychology Review by Cusson, Meilleur, Bernhardt, and Soulières (2025), asks the question researchers should have asked long ago:
What happens to autistic empathy when you actually measure it well?
Empathy Isn’t One Thing. And It Never Was.
The paper breaks empathy down into three components:
- Cognitive empathy — understanding what others are thinking
- Affective empathy — feeling what others feel
- Empathic concern — caring about someone’s experience
Across 40+ studies, the authors found that differences between autistic and non-autistic people vary dramatically depending on what kind of empathy is being measured — and how.
The biggest takeaway?
When autistic people are allowed to self-report — rather than being evaluated through observer checklists or tasks loaded with verbal demands — we show just as much empathic concern and emotional resonance as non-autistic participants.
What we don’t always do is show it in neurotypical ways. And that’s where the misunderstanding begins.
When Your Measurement Tool Is the Problem
Many of the “deficits” flagged in previous studies weren’t about empathy at all. They were about language processing. Theory-of-mind assumptions. Timed tasks. Scripts. Norms.
The tools weren’t built for us. And yet they were used to define us.
Think about that.
Imagine a test of musical ability that only accepted one scale. Or a vision exam that penalized people for not seeing color the same way. This is what many empathy assessments have done to autistic people — taken rich, valid, nuanced emotional landscapes and reduced them to wrong answers on a multiple-choice test.
And when we tried to explain that we do feel — often too much — the world told us we were wrong. That our flat tone or eye contact made our sadness suspect. That our overwhelmed feelings made our silence look cold.
Empathy as a One-Way Demand
Let’s talk about the deeper problem: empathy is rarely reciprocal when it comes to autism.
We are tested for it. Graded on it. Corrected for showing it “wrong.” But the systems around us — clinicians, teachers, families, researchers — are rarely asked to extend it back. Not in forms we recognize. Not in ways that matter.
Why is it that autistic people are expected to master the unspoken emotional codes of others while no one learns to interpret ours?
Why do studies measure how well we mimic neurotypical affect, but not how it feels to care so much you shut down?
This is what the new meta-analysis invites us to ask:
What if the deficit isn’t in our feelings — but in the frameworks used to judge them?
Living With Too Much Empathy — Not Too Little
Autistic readers may feel a painful recognition in these findings.
Many of us were told we lacked empathy when in reality, we felt so much we couldn’t contain it.
- The grief of others swamps us, but we don’t know how to show it.
- The tension in a room makes our skin crawl, but we’re accused of being oblivious.
- We miss a social cue, then cry over a friend's pain hours later in solitude.
Empathy doesn’t always look like a Hallmark commercial.
Sometimes it looks like withdrawal. Or shutdown. Or quiet care that never lands because we weren’t trained to speak its language.
We’re not broken. We’re saturated.
What Comes Next — And What Still Needs Work
The authors of the study — including Dr. Isabelle Soulières, known for respectful and rigorous work — deserve credit for their precision. They don't overstate their claims. They don't erase complexity. They interrogate the tools instead of pathologizing the people.
That’s the kind of science we need more of.
But there’s room to go further. The study, while respectful, still operates within a system that treats autistic feeling as something to validate against a neurotypical norm. Even as it questions the tools, it stops short of building new ones — tools that start from inside autistic experience, not just outside observation.
We don’t just need corrected studies.
We need reframed science. Co-designed empathy assessments. Participatory ethics. Measures of care that start by asking autistic people:
What does it feel like for you to love? To hurt? To hold space for someone else?
And how do you show it — on your terms, not ours?
Final Thought
The “empathy deficit” has been used to justify dehumanization.
This study won’t undo that harm overnight. But it cracks the door.
It gives parents a reason to question the deficit model.
It gives clinicians a reason to re-read their diagnostic checklists.
It gives autistic people a rare moment of recognition — not in spite of our emotions, but because of them.
And maybe, just maybe, it gives us the space to stop defending our hearts.
Because they were never cold.
They were just misread.