Empathy and autism: a widespread misunderstanding

What “empathy” covers

The meta-analysis distinguished several parts of empathy: understanding other people’s thoughts and feelings (cognitive empathy), emotional resonance and compassion (affective empathy, including “empathic concern”, compassion for others), and the accuracy with which one recognises others’ feelings. It also separated a more stable tendency (mostly measured by questionnaires) from behaviour in a concrete situation (mostly measured by tasks). ^[4]

What the 2019 meta-analysis showed

In 51 studies from eastern and western countries, according to the authors, differences were found in four sub-areas (stable cognitive empathy, stable compassion, cognitive empathy in the situation, compassion in the situation), while situational accuracy in recognising feelings did not differ and one form of “stable accuracy” was even higher in people with autism. ^[4] The authors stressed: the difficulties are area-specific, not across the board. Age and sex influenced individual sub-areas. ^[4] As limitations, they named that there were few studies on women and on eastern cultures; results differed greatly between studies (high heterogeneity). ^[4] It should also be borne in mind that questionnaires on “stable” empathy rely on self-reports or reports by others. Note: this is a classification by the editorial team.

“Double empathy”: misunderstandings in both directions

The concept of double empathy says that difficulties in understanding between people with and without autism are mutual. There are studies on this: in an experiment with 72 adults, groups of people with autism passed on information as effectively as groups without autism; in mixed groups, more was lost, and mutual understanding (rapport) was lower. ^[2] The authors described the sample as rather small and called for replications. ^[2] In one study, 102 adults without autism had conversations between pairs rated (42 participants with autism, 44 without, all male): conversations between people without autism were rated best, mixed ones worst; the observers rated participants with autism more negatively than their conversation partners did and underestimated their judgements of trust and intelligence. ^[3] In a further study, 81 adults from the general population judged the feelings of narrators with autism less accurately than those of narrators without autism. ^[1]

What follows from this

It cannot be concluded from the studies that people with autism do not feel or have no compassion. These are notes by the editorial team that follow the studies:

• Ask instead of assuming: If someone reacts differently, say what you mean and ask how it comes across to the person. More: Communication in everyday life: avoiding misunderstandings, Social communication and interaction in autism.
• No judgement by facial expression: In the study, observers judged people with autism more negatively than they themselves did in the interaction. ^[3]
• Meeting as equals: Both sides can reduce misunderstandings. ^[2]

Sources and literature

1. Cheang RT, Skjevling M, Blakemore AI, Kumari V, Puzzo I. Do you feel me? Autism, empathic accuracy and the double empathy problem. Autism 2025;29(9):2315–2327 (online 2024). https://doi.org/10.1177/13623613241252320 – Experimental study (81 adults)
2. Crompton CJ, Ropar D, Evans-Williams CV, Flynn EG, Fletcher-Watson S. Autistic peer-to-peer information transfer is highly effective. Autism 2020;24(7):1704–1712. https://doi.org/10.1177/1362361320919286 – Experimental study
3. Jones DR, Botha M, Ackerman RA, King K, Sasson NJ. Non-autistic observers both detect and demonstrate the double empathy problem when evaluating interactions between autistic and non-autistic adults. Autism 2024;28(8):2053–2065 (online 2023). https://doi.org/10.1177/13623613231219743 – Experimental study (102 Beobachtende)
4. Song Y, Nie T, Shi W, Zhao X, Yang Y. Empathy Impairment in Individuals With Autism Spectrum Conditions From a Multidimensional Perspective: A Meta-Analysis. Frontiers in Psychology 2019;10:1902. https://doi.org/10.3389/fpsyg.2019.01902 – Meta-analysis (51 studies)