Abstract
Environmental deformations induce distortions in the time-averaged activity of grid and place cells, which are thought to reflect a rescaling of the spatial metric of the rodent cognitive map. We propose a mechanism for this phenomenon, where input from border cells resets the spatial phase of grid cells, maintaining a learned relationship between grid phase and boundaries. A computational model demonstrated that this mechanism would yield scale-dependent distortions in time-averaged grid fields, as well as stretched, duplicated, and fractured place fields, as observed experimentally. Furthermore, the model yielded a striking prediction: apparent distortions in time-averaged activity actually arise from dynamical, history-dependent “shifts” in grid field locations. We reanalyzed the two classic datasets on grid distortions and found clear evidence of our alternative prediction. Thus, the effects of environmental deformations on spatial representations must be reconceptualized – altering spatial geometry does not distort, but rather dynamically shifts, the hippocampal map.
