On the Level Sets and Invariance of Neural Tuning Landscapes

TL;DR

This paper investigates the structure of neural tuning landscapes using level sets and Morse theory, revealing hierarchical topological changes in brain and CNN units, and proposing a new geometric hypothesis for higher-order units.

Contribution

It introduces a topological signature based on level set topology changes to analyze neural tuning landscapes and compares biological and artificial neural representations.

Findings

Topological signatures change across cortical hierarchy

Similar topological trends observed in CNN units

Higher-order units resemble local isotropic radial basis functions

Abstract

Visual representations can be defined as the activations of neuronal populations in response to images. The activation of a neuron as a function over all image space has been described as a "tuning landscape". As a function over a high-dimensional space, what is the structure of this landscape? In this study, we characterize tuning landscapes through the lens of level sets and Morse theory. A recent study measured the in vivo two-dimensional tuning maps of neurons in different brain regions. Here, we developed a statistically reliable signature for these maps based on the change of topology in level sets. We found this topological signature changed progressively throughout the cortical hierarchy, with similar trends found for units in convolutional neural networks (CNNs). Further, we analyzed the geometry of level sets on the tuning landscapes of CNN units. We advanced the hypothesis that higher-order units can be locally regarded as isotropic radial basis functions, but not globally. This shows the power of level sets as a conceptual tool to understand neuronal activations over image space.