Mnemonic prediction errors bias hippocampal states.

Area of Science:

• Neuroscience
• Cognitive Science
• Neuroimaging

Background:

• When experiences violate predictions, the brain adapts by prioritizing new information encoding and reducing reliance on inaccurate memory retrieval.
• This adaptive process is crucial for updating internal world representations to align with current reality.

Purpose of the Study:

• To investigate whether mnemonic prediction errors shift the hippocampus towards encoding or retrieval states.
• To examine how network connectivity between hippocampal subfields changes in response to prediction errors.

Main Methods:

• Functional magnetic resonance imaging (fMRI) was used to scan participants.
• Participants internally retrieved learned complex room images and were then shown identical or modified versions.
• Analysis focused on network connectivity between hippocampal subfields (CA1, CA3, entorhinal cortex) and activity patterns in CA1.

Main Results:

• In the left hemisphere, increased CA1-entorhinal connectivity and decreased CA1-CA3 connectivity correlated with the number of image modifications.
• Lower similarity in left CA1 activity patterns between retrieved and presented images (when modified) indicated a prediction error signal.
• These findings suggest a neural mechanism for memory updating driven by mnemonic prediction errors.

Conclusions:

• Mnemonic prediction errors appear to bias hippocampal states, potentially promoting encoding over retrieval.
• Distinct changes in hippocampal subfield connectivity provide a neural basis for how prediction errors drive memory updating.
• This study offers insights into the adaptive mechanisms of memory consolidation and updating in the human brain.