When it comes to grid cells, then, “our first definition was quite rigid, as first definitions tend to be,” Boccara said. But that viewpoint may have outlived its usefulness. Perhaps it’s finally time to move on.
Or maybe instead it can clue us in how grid cells work.
I was looking at grid cells as triangles. I couldn’t understand why the center of the hexagon was omitted. But actually the center could be the only point in the hexagon that matters. The center point fires when it is at a certain distance removed from all points in a circle (not just from the hexagon vertices).
To compare this with the soap bubbles experiment, it’s not the edges that compete against each other, but the points of highest air pressure inside the bubbles.
And so if a rewarding location causes attraction to increase in its vicinity, it would make sense that all the centers gravitate towards that location, and the grids become deformed. (In the soap bubbles experiment, this would happen when the pressure inside a certain bubble decreases slightly. The bubble becomes smaller and the centers of surrounding bubbles move nearer, due to their air pressure).
At first thought this would invalidate the spacial reference system. But think again: it would actually make sense that the grid resolution grows higher in a place of particular interest. You could actually compare this with a dynamic compression / decompression algorithm. And since brains are very plastic, grid cells don’t need to be rigid either, as long as they remain relatively stable for the time it matters (i.e. as long as the rewarding location exists).
(Also, it’s worth remarking that these experiments only talk about the entorhinal cortex. A different mechanism may have evolved in the neocortex).