Reduced Primacy Bias in Autism during Early Sensory Processing

Area of Science:

• Neuroscience
• Autism Research
• Cognitive Psychology

Background:

• Autism theories suggest a deficit in context-sensitive weighting of prediction errors.
• This hypothesis is supported by early sensory-level evidence.
• A key question is whether autism involves faster or slower updating of the world model.

Purpose of the Study:

• To differentiate between faster and slower world model updating in autism.
• To investigate how first impressions influence early sensory prediction error processing (MMN).
• To test if autism is characterized by faster model updating as proposed by predictive coding theories.

Main Methods:

• Compared adults with autism (n=27) and matched controls (n=27) using a multi-timescale Mismatch Negativity (MMN) paradigm.
• Tones were presented as standard (frequent) or deviant (rare), with roles reversing each block.
• A reinforcement learning model was used to simulate hypotheses about model updating.

Main Results:

• In the control group, initial context (first block) influenced MMN amplitudes in later blocks, indicating a primacy bias.
• In the autism group, MMN responses did not significantly differ based on initial sound roles (standard vs. deviant).
• This suggests individuals with autism were less influenced by the initial context compared to controls.

Conclusions:

• Findings support the hypothesis that autism is characterized by faster model updating in early sensory processing.
• This faster updating implies a reduced influence of initial impressions or context.
• Results align with predictive coding accounts of autism, suggesting a more rapid adaptation to new information.