Grid cells and 3D spaces -- how would recent discoveries would impact the thousand brain theory?

Just a catchy image to start :slight_smile:

I’m wondering what these new (published yesterday on nature) articles can help us understand how ‘cortical grid cells’ might work when understanding 3D spatial relationships.

Locally ordered representation of 3D space in the entorhinal cortex

https://www.nature.com/articles/s41593-021-00907-4.pdf

“Grid cells’ fundamental property isn’t the hexagonal structure, but rather multiple fields with a characteristic pairwise distance. In 2D, this property yields perfect hexagons. In 3D, it yields local-but-not-global order.” @GilyGinosar on Twitter - one of the main authors of the mentioned papers.

So no perfect hexagonal || equitriangular structures in 3D… but they do keep local-but-not-global order.

In any way do these new discoveries impact our understanding of the ‘thousand brain theory’?

5 Likes

This data supports the Kropff and Treves 2008 model of grid cells, which I’ve discussed here: https://discourse.numenta.org/t/video-lecture-of-kropff-treves-2008

Edit to explain how that theory supports this:

In that theory: Grid cells are burdened with a “fatigue” which limits how long the cells can stay active for, and the fatigue takes time to wear off.

When you’re moving about in the world: time is distance.

Therefore the grid cell’s fatigue controls the size of their receptive fields, and controls the (characteristic) distance between two RFs.

1 Like