Neurobiological successor features for spatial navigation

Area of Science:

• Neuroscience
• Computational Neuroscience
• Spatial Navigation

Background:

• The hippocampus is crucial for spatial representation.
• Successor representations (SRs) offer a predictive model of spatial cognition.
• Current SR models face challenges with subjective spatial discretization.

Purpose of the Study:

• To develop a novel computational model of hippocampal place and grid cell firing.
• To integrate boundary vector cells (BVCs) into an SR framework.
• To test the model's ability to predict spatial representation changes under environmental manipulations.

Main Methods:

• A successor representation (SR) model was developed using boundary vector cells (BVCs) as a basis set.
• Place cell firing was modeled as successor features of the SR.
• Grid cell activity was modeled as a low-dimensional representation of these successor features.

Main Results:

• The BVC-SR model accurately accounts for place and grid cell firing patterns.
• The model successfully predicts the impact of environmental manipulations, including dimensional stretches and barrier insertions.
• The model captures the influence of environmental geometry on hippocampal spatial representations.

Conclusions:

• The BVC-SR model provides a biologically plausible mechanism for hippocampal spatial coding.
• This framework offers a more objective approach to modeling spatial representations compared to previous SR models.
• The model advances our understanding of how the brain represents and predicts spatial environments.