Can you be autistic and empathetic?

Where the deficit claim came from

The idea that autistic people lack empathy has been a fixture of clinical texts and academic literature for decades, and it did real work: it shaped diagnostic descriptions, training materials, and the way autistic people were talked about in their own appointments.

The authors of the study below describe the position as one both clinicians and academics have increasingly rejected over the past decade, in favour of more nuanced accounts of autistic experience in a social world (Kimber, Verrier and Connolly, 2024). That’s their characterisation of the field in the paper’s background rather than a measured shift, and it’s offered here as such.

Whatever the state of the academic literature, the stereotype is still in wide circulation outside it, which is why the question keeps getting asked.

What autistic people say about their own empathy

The study above asked the question the deficit literature mostly skipped, which is what autistic people report about themselves.

Seventy-six autistic individuals were asked about their experience of empathy and about the oft-cited empathy deficit. The data were thematically analysed, and what came back was a wide array of empathic self-concepts rather than a single profile. A high proportion described hyper-empathic experiences, with many reporting their empathic responses as overwhelming or even distressing. Participants also described the extra labour created when their experience of empathy met societal expectations of how empathy is supposed to look.

That last point may explain some of the misreading. Someone flooded by another person’s distress may withdraw, go quiet, or leave the room. Read from outside against a script where empathy means visible, well-calibrated concern, that can look like coldness while being closer to the opposite.

The boundaries matter here. This is qualitative, self-report research with 76 self-selected participants. It establishes that these experiences were described within this group and that the universal deficit claim doesn’t hold. It is not a prevalence estimate for autistic people generally, and it doesn’t measure anyone’s empathy—it records what they said about it.

The alexithymia explanation

The most influential counter-explanation is that the empathy findings were measuring a different thing all along.

Alexithymia is a condition characterised by reduced ability to identify and describe one’s own emotions, and it co-occurs with autism. In a paper in Translational Psychiatry, Bird and Cook proposed that where emotional impairments are observed in autism, they’re attributable to alexithymia rather than being a feature of autism itself—noting that a close look at the literature reveals wide heterogeneity in emotional competence within autistic populations, which a blanket deficit account struggles to explain (Bird and Cook, 2013).

They named this the alexithymia hypothesis, and their paper reviews a series of studies designed to test it. It’s a theoretical review advancing a proposal, not a demonstration that earlier autism research was in fact measuring alexithymia. Held as a hypothesis, it reframes the question usefully: if empathy differences track alexithymia rather than autism, then studies that didn’t measure alexithymia would have been averaging across two different groups and attributing the result to the wrong one. Whether that’s what happened is what the hypothesis proposes to find out.

The same paper notes elevated rates of alexithymia in several other conditions, and suggests the approach may clarify their emotional symptoms too.

A mismatch account, and what tested it

A separate line of thinking questions whether “deficit” is the right shape of explanation at all.

Damian Milton’s double empathy problem argues that breakdowns between autistic and non-autistic people are a two-way mismatch of experience and expectation, not a one-directional failure located in the autistic person (Milton, 2012). It’s a theoretical paper, and for years it was mostly discussed as a reframing.

Then an experiment tested a prediction it makes. Nine groups of eight people were assembled: three entirely autistic, three entirely non-autistic, and three mixed. A story was told to one person in each group and passed along a chain, with researchers measuring how much detail survived at each stage. Autistic people shared information with other autistic people as effectively as non-autistic people did with each other. In the mixed groups, substantially less information was transferred, and participants also reported lower rapport with the person they were talking to (Crompton et al., 2020).

That result is hard to fit inside an account locating the problem in autistic communication ability, since all-autistic chains would then have performed worst. They didn’t.

Two boundaries belong on this. The study measured information transfer and reported rapport—it is evidence about cross-neurotype communication, not a measurement of empathy, and it’s used here for the mismatch account rather than as direct evidence about feeling. And the design was nine groups in a laboratory task, so one study doesn’t settle a field. The related question of what happens with eye contact specifically is covered in can you be autistic and make eye contact?.

What the empathy measures are testing

A lot of the disagreement is about instruments, and it’s worth seeing why.

Much of the empathy literature relies on self-report questionnaires, which ask people to rate how much they agree with statements about their own emotional responses. That’s a reasonable method for most constructs and a difficult one here, because accurately reporting your own inner emotional states is precisely the ability alexithymia interferes with (Bird and Cook, 2013). A questionnaire that asks “do you feel what others feel?” partly measures how well you can read yourself, which means a low score can indicate difficulty describing an emotion rather than difficulty having one.

The task-based measures have a different problem. They usually involve identifying an emotion from a photograph, a video clip, or a vignette—non-autistic faces, non-autistic voices, non-autistic social scripts, scored against what a non-autistic sample judged to be correct. Performance on that task is cross-neurotype accuracy. Reporting it as general empathic ability builds the conclusion into the design, which is exactly the objection the double empathy account raises.

The information-transfer experiment took a different approach, though it was measuring communication rather than empathy. It didn’t ask anyone how empathic they felt, and it didn’t score autistic participants against a non-autistic key: it measured how much information survived being passed between real people, and compared like with like (Crompton et al., 2020). That’s a design worth noting, not a verdict on the empathy literature.

The argument set out in this section is a methodological critique rather than a finding. How much it should reduce confidence in the older measures is itself contested.

What’s still contested

An honest version of this page has to say where the disagreement remains, because there is some.

Cognitive empathy—inferring what someone else is thinking or feeling—is where differences are most often reported, and the debate is about how much of that reflects autism, how much reflects alexithymia, and how much reflects measures built and validated on non-autistic people. The double empathy work adds a further complication: many of those measures ask autistic participants to read non-autistic faces and scenarios, so they test cross-neurotype accuracy while reporting it as general ability.

Affective empathy—feeling something in response—is where some studies and autistic self-reports describe intact or heightened responses, and where the hyper-empathy accounts sit. That’s a weaker claim than a settled finding, and it’s the honest version.

What the evidence doesn’t support is the flat, universal claim that autistic people lack empathy. It conflates two distinct capacities, it sets aside a co-occurring condition that may account for part of the findings, and it assigns a difficulty that looks mutual to one side of it.

If you’re autistic and have spent years being told you don’t feel things while privately feeling a great deal, you’re describing an experience represented in this study and in other autistic accounts.