On the supra-linear storage in dense networks of grid and place cells
Adriano Barra, Martino S. Centonze, Michela Marra Solazzo, Daniele Tantari · 2025
Read On the supra-linear storage in dense networks of grid and place cells by Adriano Barra, Martino S. Centonze, Michela Marra Solazzo, Daniele Tantari (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-02441.txt.
Description
Adriano Barra, Martino Centonze, Michela Marra Solazzo, and Daniele Tantari ask how the brain's navigation system stores so many places. Place cells, which fire for specific locations, are usually modelled with pairwise connections. In such networks, the number of stored patterns grows only linearly with size, far too little for mapping two-dimensional surfaces at fine resolution, let alone three-dimensional space. The authors build a two-layer model pairing place cells with grid-like units like the brain's grid cells.
Each place cell interacts with pairs of grid cells, and the team proves the model is equivalent to a dense Hebbian network with four-body interactions, a class known for much larger memory. Statistical mechanics analysis shows analytically that these higher-order assemblies support more than linearly many stored continuous attractors. Numerical experiments confirm the network can recognise places and navigate on general surfaces embedded in three-dimensional space.
Catalog record
| CREATOR | Adriano Barra, Martino S. Centonze, Michela Marra Solazzo, Daniele Tantari |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 144 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-02441