Grid-Cell Distortion along Geometric Borders

Abstract

Grid cells fire in a triangular pattern that tessellates the environment [1]. The pattern displays a global distortion that is well described by a shearing transformation of an idealized grid [2]. However, in addition, distortions often differ across parts of the environment, suggesting that the grid interacts with the environment locally [2-5]. How this occurs is poorly understood. To further determine the nature of local distortions, we therefore analyzed the local spatial characteristics of the grid pattern. When rats ran in a large square enclosure, the grid pattern displayed several stereotypical distortions in relation to features of the environment. These distortions were stronger at edges than on open surfaces. Curved axis orientations and distortions of the grid pattern in the corners could be explained by a geometrical model where the pattern, in conjunction with being sheared, is compressed along the walls of the enclosure. The grid compression coincided with stereotypical running behavior where the animals moved faster in the areas where the grid had the most pronounced distortions. However, neither running direction nor speed influenced the distortions on a moment-to-moment basis, raising the possibility that the distortions are a learned feature.

Keywords: computational models; entorhinal cortex; grid cells; hippocampus; path integration; plasticity; space; spatial map.