Learning Orientations: a Discrete Geometry Model

TL;DR

This paper proposes a novel geometric model for how the mammalian brain integrates spatial location and head direction information into a unified orientation framework, using algebraic topology and affine geometry.

Contribution

It introduces a new discrete geometry model that combines spatial and directional neural data into a cohesive representation of orientation.

Findings

Model provides a mathematical framework for neural spatial integration

Uses algebraic topology to analyze neural encoding of space

Suggests a new approach for understanding neural spatial maps

Abstract

In the mammalian brain, many neuronal ensembles are involved in representing spatial structure of the environment. In particular, there exist cells that encode the animal's location and cells that encode head direction. A number of studies have addressed properties of the spatial maps produced by these two populations of neurons, mainly by establishing correlations between their spiking parameters and geometric characteristics of the animal's environments. The question remains however, how the brain may intrinsically combine the direction and the location information into a unified spatial framework that enables animals' orientation. Below we propose a model of such a framework, using ideas and constructs from algebraic topology and synthetic affine geometry.