Exploitation of local and global information in predictive processing.

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience

Background:

• Predictive processing theories emphasize learning via prediction errors.
• Emerging research highlights the role of model-compliant events in evaluating internal models, especially under uncertainty.
• Probabilistic information at checkpoints is sampled to validate expectations, even without errors.

Purpose of the Study:

• Investigate the interplay between global uncertainty and local adjustment cues in online expectation updating.
• Examine how the brain processes model-compliant information and uncertainty during predictive processing.
• Determine the hierarchical influence of global versus local information in predictive models.

Main Methods:

• Electroencephalography (EEG) to record brain activity.
• Participants detected digit sequences with varying lengths and uncertainty levels.
• Multivariate pattern decoding and representational similarity analysis were employed.

Main Results:

• The N400 component reflected the integration of locally unexpected information.
• P3b activity demonstrated the exploitation of both local cues and global uncertainty information.
• A global model showed a superior fit, indicating the precedence of global reference frames.

Conclusions:

• Model evaluation, like error-induced adaptation, occurs at multiple processing levels.
• Both local adjustment cues and global uncertainty information are crucial for refining expectations.
• Hierarchical predictive processing involves integrating information across different levels of a model.