Know thy agency in predictive coding: Meta-monitoring over forward modeling

Area of Science:

• Cognitive Neuroscience
• Psychology
• Neuroscience

Background:

• Agency computation is linked to prediction error.
• Assessing the reliability of detected prediction errors is crucial for understanding agency.
• A meta-monitoring ability over internal forward models is hypothesized.

Purpose of the Study:

• To investigate the meta-monitoring of the forward model.
• To examine the implicit evaluation of motor prediction accuracy and felt agency.
• To explore the relationship between accuracy and confidence in decision-making tasks.

Main Methods:

• 105 healthy participants performed a motor control task with self or other visual feedback.
• Assessed accuracy-confidence correlation in mismatch detection and self-other attribution tasks.
• Investigated schizotypal traits in relation to meta-monitoring performance.

Main Results:

• A significant accuracy-confidence correlation was observed in both tasks, supporting meta-monitoring.
• Individuals with lower accuracy and confidence showed higher schizotypal traits.
• The findings indicate an implicit evaluation of forward model accuracy.

Conclusions:

• Humans possess a meta-monitoring ability to evaluate their own forward model's accuracy.
• This ability is linked to confidence in performance and may be impaired in higher schizotypy.
• Understanding this meta-monitoring is key to understanding agency and self-awareness.