Figure 3.

The Transition from Local to Global Representations Cannot Be Explained by Biases in the Sampling of Grid Cells

(A and B) The fit between recorded firing patterns of grid cells of a single scale and ideal local and global grids, respectively, as a function of experience of the environment. Only cells with a scale of 45 to 55 cm in the screening environment are included. Each data point represents the average local and global fits across all 45–55 cm cells with an independent fit >0.45, recorded from one animal in one session.

(C) The proportion of 1,000 ideal grids, with random phase offsets between the compartments, with a better fit to the cells in (A) and (B) than the local or global models. Values are mean + SEM across 45–55 cm cells with an independent fit >0.45 in the first or last five sessions. WSRTs compare observed values to an expected median of 0.5.

(D and E) The fit between recorded firing patterns of grid cells from a single module in a single animal and ideal local and global grids, respectively, as a function of experience. Dashed lines extend the least-squares lines to predict local and global fits in unrecorded sessions.

(F) The best fit achieved by the local model in the first five sessions and the global model in the last five sessions to the grid patterns in the thirds of the compartments nearest to or furthest from the corridor. Values are mean + SEM of the collapsed average within animals of cells with an independent fit >0.45. Paired, one-tailed t tests test whether difference in observed means differs from an expected mean of 0. ^∗p < 0.05; ^∗∗p < 0.01.