A Brain-Inspired Adaptive Space Representation Model Based on Grid Cells and Place Cells.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Artificial Intelligence

Background:

• Grid cells and place cells are fundamental to spatial representation and navigation in animals.
• Existing models often preset grid cell firing characteristics, limiting their applicability.
• Understanding this neural mechanism informs the development of autonomous navigation systems for intelligent agents.

Purpose of the Study:

• To propose a united generation model for grid cells and place cells.
• To overcome limitations of existing models by not presetting grid cell firing characteristics.
• To enhance environmental adaptability in spatial representation models.

Main Methods:

• Utilized a feedforward network for generating visual grid cells from nonuniform visual place cells.
• Employed a genetic algorithm to integrate visual grid cells and self-motion information for whole-space grid cell generation.
• Applied supervised fuzzy adaptive resonance theory (ART) network for generating uniform place cells from visual place cells and united grid cells.

Main Results:

• The proposed model successfully generates both grid cells and place cells.
• The model exhibits stronger environmental adaptability compared to existing approaches.
• Simulation results validate the model's capability in handling nonuniform environmental information.

Conclusions:

• The united generation model offers a more flexible and adaptable approach to simulating grid and place cell function.
• This research provides a valuable reference for developing brain-inspired navigation mechanisms in intelligent agents.
• The model's ability to handle nonuniform environmental data is a significant advancement for spatial representation research.