Multiple event segmentation mechanisms in the human brain.

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Modeling

Background:

• The human brain segments continuous experience into discrete events.
• Theoretical accounts propose prediction error (mismatch between expected and observed information) and prediction uncertainty (reduced precision in predictions) as key mechanisms for event segmentation.

Purpose of the Study:

• To investigate the neural correlates of event boundaries driven by prediction error and prediction uncertainty.
• To examine how computational models of these boundaries relate to brain activity patterns and evoked responses.

Main Methods:

• Utilized functional magnetic resonance imaging (fMRI) and computational modeling.
• Developed models to generate event boundaries based on prediction error or prediction uncertainty.
• Analyzed fMRI data for pattern shifts and evoked responses around human-identified, error-driven, and uncertainty-driven boundaries.

Main Results:

• Identified a temporal sequence of neural pattern changes around human boundaries, involving anterior temporal, parietal, and prefrontal regions.
• Found that both error-driven and uncertainty-driven boundaries were associated with core neural responses but exhibited unique pattern shifts.
• Error-driven boundaries showed early prefrontal shifts, while uncertainty-driven boundaries engaged parietal regions within the dorsal attention network.

Conclusions:

• Provided evidence for two overlapping brain networks involved in environmental representation maintenance and updating.
• Demonstrated that distinct prediction quality signals—prediction error and prediction uncertainty—control these networks.
• Highlighted the differential neural dynamics associated with error-based versus uncertainty-based event segmentation.