This repository contains to code to reproduce the grid cells model of this paper. Grid cells, found in the medial entorhinal cortex, exhibit an hexagonal firing pattern that enables spatial navigation and memory formation. While their structure is well-documented, the underlying principles governing their emergence remain debated. The authors analyze self-organization in recurrent networks, highlighting how local connectivity and learning rules can naturally give rise to grid-like firing patterns. They argue that grid cells likely emerge from general principles of neural computation rather than being an explicitly pre-designed feature of the brain. The network architecture includes feedforward connections between place and grid cells, as well as recurrent connections among grid cells. See paper for more details.
The simulation of the grid cells firing pattern emergence can be found on rat_1D.ipynb. There is also a GPU-based implementation at GPU_rat_1D.ipynb.
The grid cell hexagonal firing pattern is obtained at the end of the simulation.