Speed and accuracy of emotion recognition in autistic adults: The role of stimulus type, response format, and emotion.
Autistic adults read emotions almost as well as anyone else; small delays, not missing skills, explain most social friction.
01Research in Context
What this study did
Georgopoulos et al. (2022) asked autistic and non-autistic adults to name emotions from faces, voices, and short videos. They also tried two response formats: clicking a word label or pressing a key.
The team recorded how accurate and how fast each group was. They wanted to see if the kind of picture or the way you answer changes the size of any autism gap.
What they found
Accuracy was only a little lower and response times a little longer for the autistic group. The gap stayed small no matter if the face was still, moving, or just a voice.
Big individual differences swamped the group effect. Stimulus type and response format barely moved the needle.
How this fits with other research
Shire et al. (2019) and Fink et al. (2014) saw the same near-zero accuracy gap in adults and in kids once verbal skill was matched. These studies form a line of 'null' results that Antonia now extends to varied stimulus and response types.
Song et al. (2018) and Evers et al. (2015) look like they clash: they report clear child deficits for low-intensity or dynamic faces. The gap closes by adulthood, so the 'contradiction' is really about age, not method.
Poljac et al. (2013) shows the effect even bleeds into the general population: non-autistic adults with high autism traits also score slightly worse on negative emotions, proving the difference is dimensional, not diagnostic.
Why it matters
Stop assuming every social struggle in autistic adults stems from not knowing how someone feels. The skill is usually there; the speed is just a beat slower. Use that insight to shift your training target from 'recognize happy' to 'respond within the natural pause of conversation' and to advocate for extra processing time in work or classroom settings.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Add 'response within 2-second natural pause' as a fluency target instead of more emotion-label drills.
02At a glance
03Original abstract
Emotion recognition difficulties are considered to contribute to social-communicative problems for autistic individuals. Prior research has been dominated by a focus on forced-choice recognition response accuracy for static face presentations of basic emotions, often involving small samples. Using free-report and multiple-choice response formats, we compared emotion recognition in IQ-matched autistic (N = 63) and nonautistic (N = 67) adult samples using 12 face emotion stimuli presented in three different stimulus formats (static, dynamic, social) that varied the degree of accompanying contextual information. Percent agreement with normative recognition responses (usually labeled "recognition accuracy") was slightly lower for autistic adults. Both groups displayed marked inter-individual variability and, although there was considerable overlap between groups, a very small subset of autistic individuals recorded lower percent agreement than any of the nonautistic sample. Overall, autistic individuals were significantly slower to respond and less confident. Although stimulus type, response format, and emotion affected percent agreement, latency and confidence, their interactions with group were nonsignificant and the associated effect sizes extremely small. The findings challenge notions that autistic adults have core deficits in emotion recognition and are more likely than nonautistic adults to be overwhelmed by increasingly dynamic or complex emotion stimuli and to experience difficulties recognizing specific emotions. Suggested research priorities include clarifying whether longer recognition latencies reflect fundamental processing limitations or adjustable strategic influences, probing age-related changes in emotion recognition across adulthood, and identifying the links between difficulties highlighted by traditional emotion recognition paradigms and real-world social functioning. LAY SUMMARY: It is generally considered that autistic individuals are less accurate than nonautistic individuals at recognizing other people's facial emotions. Using a wide array of emotions presented in various contexts, this study suggests that autistic individuals are, on average, only slightly less accurate but at the same time somewhat slower when classifying others' emotions. However, there was considerable overlap between the two groups, and great variability between individuals. The differences between groups prevailed regardless of how stimuli were presented, the response required or the particular emotion.
Autism research : official journal of the International Society for Autism Research, 2022 · doi:10.1002/aur.2713