Figure 10.

Comparison of grid cell activity with and without stabilization mechanism for grid cell simulations with real data in input. The spiking activities (red dots) of representative simulated grid cells are shown in correspondence to the position of a real robot (gray trace) while exploring for 30 min a circular arena without (A) and with (B) realignment mechanism (gridness scores: −0.043 for A, 0.577 for B). The activation of this plasticity-based stabilization mechanism successfully anchors the neural activity of grid cells to external sensory cues despite the additional noise introduced by imprecise velocity and landmark position estimates.