Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session A: Tuesday, August 12, 1:30 – 4:30 pm, de Brug & E‑Hall

Rat on a sphere: Exploring alternate grid cell representations on a hemisphere

Muhammad Sahal Goolam¹, Clémentine Carla Juliette Dominé², Luke Hollingsworth², Rodrigo Antonio Carrasco-Davis², Geraud Nangue Tasse³, Devon Jarvis³; ¹University of Witwatersrand, ²University College London, University of London, ³University of the Witwatersrand

Presenter: Muhammad Sahal Goolam

Grid cells in the medial entorhinal cortex are fundamental to mammalian spatial navigation, forming periodic, hexagonal firing patterns that serve as an internal map. While these cells exhibit isotropic firing in flat two-dimensional environments, their behaviour in three dimensions is more irregular, with the mechanisms governing spatial cognition beyond planar surfaces remaining largely unexplored. This study examines how spatial encoding adapts to a curved two-dimensional surface embedded in three dimensions. Specifically, we simulate a rat’s movement on a hemispherical surface under varying gravitational strengths and consider three spatial encoding strategies based on the Successor Representation framework: (i) aligned with the sphere’s curvature (logarithmic projection); (ii) relative to the horizontal plane tangent to the hemisphere at its origin (orthogonal projection) and (iii) using polar coordinates. Our results show that orthogonal projection does not elicit particularly salient or regular grid patterns, which compromises localisation. In contrast, logarithmic projection improves localisation while encoding in polar coordinates enhances grid regularity at a cost to localisation. These findings highlight the need to account for third-dimensional effects on grid cell organisation even on two-dimensional surfaces. Our model establishes a theoretical foundation for future empirical studies and offers extensions to computational models exploring grid cell formation on flat two-dimensional surfaces (embedded in two-dimensional space).

Topic Area: Memory, Spatial Cognition & Skill Learning

Extended Abstract: Full Text PDF