Partial coherence and frustration in self-organizing spherical grids

Area of Science:

• Neuroscience
• Computational Neuroscience
• Animal Behavior

Background:

• Grid cells in the rodent brain exhibit remarkable long-range order and coherence in flat environments.
• This periodic firing is theoretically linked to flat spatial manifolds, contrasting with natural rodent habitats like burrows.

Purpose of the Study:

• To investigate how grid cell network dynamics adapt to curved environments, using spheres as a model.
• To predict the emergent properties of grid maps in non-flat terrains and assess their ecological relevance.

Main Methods:

• Utilized an adaptation model of recurrent neural networks known to produce coherent grid maps on a plane.
• Simulated network activity on spherical environments to observe changes in grid cell firing patterns.

Main Results:

• Recurrent connections that induce coherence on a plane distort individual grid cell maps on a sphere.
• Grid cells aggregate into clusters on spherical surfaces, losing their uniform long-range order.

Conclusions:

• Grid cell network organization differs significantly between flat laboratory conditions and curved natural environments.
• The observed clustering and partial coherence on spheres resemble disordered systems like spin glasses, impacting navigational relevance.