A Model of Spatial Cell Development in Rat Hippocampus Based on Artificial Neural Network

Area of Science:

• Computational Neuroscience
• Neuroscience
• Artificial Intelligence

Background:

• Rat hippocampal place cells mature post-juvenile period, while grid cells, their precursors, mature earlier.
• Understanding spatial cognition requires modeling the developmental interplay between grid and place cells.
• Existing models may not fully capture the dynamic development of these hippocampal cell types.

Purpose of the Study:

• To propose a novel position cognition model inspired by the developmental mechanisms of rat hippocampal grid and place cells.
• To simulate the distinct developmental timelines of grid cells and place cells using computational methods.
• To validate the model's ability to achieve accurate spatial cognition and cognitive map construction.

Main Methods:

• A recurrent neural network with parametric bias (RNNPB) was employed to simulate individual grid cell development.
• An oscillatory interference mechanism was used to model the fusion of developing grid cell representations.
• A backpropagation (BP) neural network simulated place cell development, using grid cell output as input.

Main Results:

• The proposed model successfully simulated the developmental trajectories of both grid and place cells.
• High-precision positioning within a defined spatial area was achieved by the model.
• Cognitive map construction performance was comparable to established methods like RatSLAM, confirming model validity.

Conclusions:

• The developed model effectively captures the developmental processes of rat hippocampal grid and place cells.
• The model demonstrates robust capabilities for spatial cognition and cognitive map generation.
• This biologically-inspired approach offers a valid and accurate framework for understanding hippocampal function in navigation.