Reconciling flexibility and efficiency: medial entorhinal cortex represents a compositional cognitive map

Area of Science:

• Computational Neuroscience
• Cognitive Psychology
• Neuroimaging and Modeling

Background:

• Cognitive maps are mental representations of environments, crucial for navigation and goal-directed behavior.
• Compositionality, the ability to combine simpler representations into complex ones, is vital for cognitive flexibility.
• The neural basis of compositional cognitive maps and their role in efficient planning remain poorly understood.

Purpose of the Study:

• To propose a novel computational model for creating and planning with compositional predictive maps.
• To investigate how compositionality can be reconciled with efficient planning in neural representations.
• To explain neural activity patterns observed in the medial entorhinal cortex related to spatial cognition.

Main Methods:

• Developed a computational model that treats objects as alterations to a baseline spatial map.
• Implemented compositional combination of object-related representations to form predictive maps.
• Validated the model against empirical data, specifically response fields of cells in the medial entorhinal cortex.

Main Results:

• The model successfully reproduces response fields of object vector cells and grid cells in the medial entorhinal cortex.
• Demonstrated that compositional predictive maps can support efficient and flexible planning.
• Showcased how combining object representations compositionally provides a mechanism for complex spatial reasoning.

Conclusions:

• The proposed model offers a framework for understanding the neural basis of compositional cognitive maps.
• Compositionality in predictive maps is key to efficient and flexible goal-directed navigation.
• Insights into brain processes supporting planning and spatial representation are provided by this computational approach.