Prediction errors drive dynamic changes in neural patterns that guide behavior.

Area of Science:

• Neuroscience
• Cognitive Science
• Decision Making

Background:

• Learning involves updating internal world models via prediction errors (PEs) to enhance future predictions.
• The precise neural mechanisms by which PEs dynamically modify existing representations remain unclear.

Purpose of the Study:

• To investigate how neural activation patterns change following prediction errors during a decision-making task.
• To identify brain regions where prediction error signals influence neural representations and predict behavioral adaptation.

Main Methods:

• Functional magnetic resonance imaging (fMRI) was employed in human participants performing a two-step Markov decision task.
• Dynamic multivariate pattern analysis was utilized to analyze trial-by-trial changes in neural activity following prediction errors.

Main Results:

• Prediction error-related fMRI responses in specific brain regions were found to predict changes in multivariate neural patterns.
• These pattern changes were observed in the orbitofrontal cortex, precuneus, and ventromedial prefrontal cortex (vmPFC).
• The temporal dynamics of pattern changes in the vmPFC correlated with subsequent shifts in choice strategies.

Conclusions:

• Prediction error signals dynamically influence neural representations in key decision-making brain areas.
• These neural pattern dynamics, particularly in the vmPFC, are crucial for adaptive behavioral strategies and learning.
• The findings elucidate a neural mechanism linking prediction errors to behavioral adjustments in learning.