A new study examining the so-called double empathy problem has found no overall difference between autistic and non-autistic adults in their ability to read others’ emotions, though it discovered a specific advantage for autistic participants when empathising with other autistic individuals. The research, published in Molecular Autism, also suggests autistic people may underestimate their own empathic abilities.
Empathy is typically divided into cognitive empathy, the recognition and understanding of another person’s mental state, and affective empathy, the ability to respond emotionally in a way that mirrors what someone else is feeling. The study situates itself within ongoing debate about whether empathic difficulties in autism are solely the fault of autistic individuals or arise from reciprocal misunderstandings between autistic and non-autistic people.
A cornerstone of the research is the double empathy problem hypothesis, which posits that communication difficulties between autistic and non-autistic people are bidirectional rather than one-sided. The authors note that autistic people often report easier communication and greater comfort with other autistic individuals, while non-autistic people can struggle to interpret autistic expressions as accurately as those from non-autistic peers.
The study, led by Yonat Rum and colleagues, sought to test whether autistic and non-autistic participants differed in empathising with autistic versus non-autistic social targets. Researchers hypothesised that non-autistic participants would empathise more accurately with non-autistic targets, and autistic participants with autistic targets.
Participants totalled 235 individuals recruited through the Cambridge Autism Research Database and word of mouth. Of these, 141 were autistic, and 125 were female. Seventy-four per cent were from English-speaking countries, with the majority from the United Kingdom. Autistic participants’ average age was 48, compared with 51 for non-autistic participants.
The researchers developed an empathic accuracy task using videos in which 10 adults — five autistic and five non-autistic — narrated autobiographical stories. As the storytellers recounted their experiences, they used a slider to indicate their affective valence, from negative to positive, with the position recorded every half-second. After each story, the storytellers rated how strongly they felt 12 emotions, ranging from embarrassment and anger to happiness and calmness, on a scale of 0 to 8.
Each participant watched two of the videos: one featuring an autistic storyteller and one featuring a non-autistic storyteller. They did not know the storytellers’ diagnoses and were asked to rate the storytellers’ emotions in real time. The study assessed empathic accuracy by comparing participants’ readings of a storyteller’s emotions with the storyteller’s own ratings, while also capturing participants’ self-rated empathy and interest in the storyteller.
Results indicated that participants were generally somewhat accurate at determining whether a storyteller’s feelings were positive or negative. However, there were no broad differences between autistic and non-autistic individuals in the overall ability to estimate specific emotions. Notably, autistic participants were significantly more accurate at tracking the moment-to-moment emotional valence of autistic storytellers than non-autistic participants, a finding described as a small-to-medium own-group advantage.
The two groups performed similarly when assessing non-autistic storytellers. Interestingly, non-autistic participants rated their own empathy and interest in the storytellers higher than autistic participants.
In free-text feedback, autistic participants frequently described difficulties in identifying emotions and reflected on their own performance, while non-autistic participants tended to offer comments about the task design. The researchers suggest this pattern may indicate that autistic individuals underestimate their empathic abilities.
The authors concluded: “Findings suggest partial support for the double empathy problem theory and potential underestimation of autistic participants of their own empathic abilities. Individual variability patterns caution against treating neurotypes as homogeneous categories that alone determine double empathy processes.”
The study notes several limitations. Participants evaluated pre-recorded videos rather than live, real-world social interactions, which may differ from genuine encounters. Autistic individuals in the sample were volunteers and self-selected, potentially not representing the full autism spectrum, and likely excluding those who require greater social support.
The paper, “Nuances of double empathy in autistic and non-autistic people: examination using the empathic accuracy paradigm,” was authored by Yonat Rum, Noa Feldman, Shir Genzer, Carrie Allison, Ofer Golan, Anat Perry, and Simon Baron-Cohen.
