Predicting emotional valence in autism: a preregistered study in the Bayesian Brain framework

Area of Science:

• Neuroscience
• Cognitive Science
• Psychology

Background:

• Autism spectrum disorder (ASD) impacts social interaction, communication, and behavioral flexibility.
• Bayesian Brain Framework theories suggest autistic individuals may process environmental uncertainty differently, overemphasizing volatility.
• The robustness and symptom relevance of these proposed predictive processing differences in ASD remain unclear.

Purpose of the Study:

• To investigate differences in environmental volatility processing between autistic and non-autistic adults.
• To extend prior research by applying a predictive processing paradigm to facial emotion recognition in autism.
• To evaluate the clinical relevance of Bayesian Brain Framework theories in the context of core autistic traits.

Main Methods:

• Extended a computational paradigm (Hierarchical Gaussian Filter) to assess belief states in a facial emotion recognition task.
• Utilized Bayesian linear mixed models to analyze data from 22 autistic and 22 non-autistic participants.
• Preregistered hypotheses were tested to ensure objective evaluation of findings.

Main Results:

• No robust differences in environmental volatility processing were found between autistic and non-autistic adults.
• A non-credible trend suggested autistic participants might process phasic volatility differently.
• Probabilistic associative learning, response times, and pupil sizes were largely comparable across groups.

Conclusions:

• The study failed to find credible differences in environmental volatility and learning rate updates in a facial emotion recognition task for autistic adults.
• Results challenge the generalizability and clinical significance of previous findings on predictive processing in autism.
• The task's block structure may have limited the ability to detect learning effects, suggesting potential limitations in experimental design.