Not a global map, but a local hash: grid cells decorrelate the representation of position and scramble long-range distance information

Grid cells in the medial entorhinal cortex construct an intriguing multiperiodic representation of space whose properties have been the subject of much theoretical speculation. Here we combine modeling with analyses of neural population data from mice and rats to show that the grid cell representation is ideally set up to decorrelate and assign easily distinguishable labels to inputs, thus acting as a pattern separation device, much like a hash function in computing rather than a global map or metric. The multiple modules of the grid cell system allow the threshold for pattern separation to be controlled. We also extend these arguments to show how grid cells can perform pattern separation in abstract and higher-dimensional spaces. This pattern separation ability may serve to enhance episodic memory in the hippocampal formation by reducing interference between similar patterns.