Predicting emotional valence in autism: a preregistered study in the Bayesian Brain framework.
Irene Sophia Plank, Alexandra Pior, Anna Yurova and 3 others
A preregistered study found no credible differences in how autistic and non-autistic adults adapt to changing expectations during an emotional face task, challenging earlier reports of group differences.
One measure trended higher in autistic adults but missed the pre-set bar for a confirmed finding.
Key findings
1No credible difference between autistic and comparison adults in environmental tonic volatility, the background expectation of how much the environment changes.
2Phasic volatility — how strongly unexpected events drive the response — showed a trend toward being higher in autistic adults, but the posterior probability of 92.54% fell below the preregistered 95% credibility threshold.
3No credible group difference in the effect of expectancy on reaction times or on pupil sizes, in contrast to the earlier study this one was designed to replicate.
to read the rest of this summary — three full summaries a month, no card.
Already have an account?
What it does not show
The sample of 22 per group was powered for a medium effect; the authors note it may be insufficient to detect subtle differences. The gender distribution (50% women in the autistic group) differs from the earlier study being replicated (25% women), and the sample is too small to draw firm conclusions about sex effects. The HGF model captured the expectancy effect but not the difficulty effect on reaction times, so the authors acknowledge alternative computational models might fit the data better. The task reduced emotion recognition to a simple positive/negative valence decision, which the authors note limits generalisability to the complexity of real-world emotion recognition and produced a ceiling effect on accuracy. Learning-rate update results are explicitly flagged as needing cautious interpretation because the preregistered model showed suboptimal fit for the volatile-to-postvolatile transition.
Declared interests
Funded by Klinikum der Universität München (grant 6933). No industry sponsorship, no author disclosures of competing interests are stated in the text provided.
Summarised by AI from the full paper, without a clinician reviewing it. Check it against the source before it changes what you do. Read it on PubMed →