Vemund Sigmundson Schøyen

This paper investigates grid cells as a conformal isometric map of space, showing that a module of at least seven grid cells can achieve a distance-preserving representation with regular hexagonal phase patterns.

This paper shows that grid-like spatial representations can emerge in neural networks without path integration, suggesting distance preservation and capacity constraints may be key factors.

This paper develops a neural network model that learns place cell-like spatial representations and remapping behaviors, showing clustering patterns similar to biological place cells in mice.

The paper shows that toroidal cells, not grid cells, are crucial for path integration in virtual agents using RNN models.