Background: Sensory processing optimizes predictions on sensory input. This predictive coding usually entails prediction errors prompting further sensory processing. Prediction errors are weighted by precision, which can be an attributed salience of the sensory input. Many autistic individuals report different sensory processing, which has been associated with increased precision weighting. The locus-coeruleus norepinephrine (LC-NE) system modulates the neuronal gain of sensory input. We hypothesized that LC-NE tonic upregulation underlies an increased precision weighting in autism.

Methods: Autistic (n=139) and non-autistic (n=98) individuals were assessed during a passive auditory oddball task with pupillometry and electroencephalography. A baseline pupil size (BPS) assessed LC-NE tonic activity that estimated precision weighting. A stimulus-evoked pupillary response (SEPR) assessed LC-NE phasic activity that estimated sensory-driven selectivity. Electroencephalography assessed amplitudes of mismatch negativity (MMN-amp) that estimated prediction errors. Measures were modeled between groups within the task by combining Frequentist and Bayesian approaches.

Results: Across groups, higher BPS was associated with more negative MMN-amp to standards and oddballs. Controlling for this association, autistic versus non-autistic individuals showed a higher SEPR in response to standards. In addition, an association of BPS and SEPR to standards was specific to autistic individuals. With task progression, autistic versus non-autistic individuals showed a higher initial increase and subsequently higher decrease of BPS. This was supported by Bayesian posteriors.

Conclusions: The findings characterize sensory processing in autistic individuals by increased sensory-driven selectivity to frequent stimuli. This is caused by an LC-NE tonic upregulation during sensory processing that underlies increased precision weighting in autistic individuals.