Modeling flexible behavior with remapping-based hippocampal sequence learning

Area of Science:

• Neuroscience
• Computational Biology
• Cognitive Science

Background:

• The hippocampus is crucial for contextual behaviors, exhibiting context-dependent sequential activity.
• The mechanisms underlying the formation of this context-dependent activity through neural remapping are not fully understood.

Purpose of the Study:

• To develop a biologically plausible reinforcement learning model explaining hippocampal sequential activity formation.
• To investigate how this activity supports flexible, context-dependent behaviors.
• To explore the model's implications for neurological disorders.

Main Methods:

• A novel biologically plausible reinforcement learning model was developed.
• The model incorporates a 'Context selector' mechanism to promote context-dependent sequential activity.
• The model was validated against existing data from neural activity, optogenetic inactivation, and human fMRI studies.

Main Results:

• The model successfully reproduces diverse findings on hippocampal activity and behavior across different contexts.
• It demonstrates how context-dependent sequential activity can be formed and flexibly switched.
• The model predicts that specific representational imbalances in the 'Context selector' are linked to schizophrenia and autism spectrum disorder-like behaviors.

Conclusions:

• The developed reinforcement learning model provides a framework for understanding hippocampal function in context-dependent behavior.
• It highlights the role of sequential activity and neural remapping in flexible behavioral adaptation.
• The model offers novel insights into the neural underpinnings of schizophrenia and autism spectrum disorder.