Comparative Analysis of Neural Population Codes

This website serves as a central hub of Google CoLab jupyter notebook tutorials on different metrics to compare neural representations.

While notebooks should be self-contained or contain references to required background material, they’re designed to complement the COSYNE 2026 Tutorial: Comparative Analysis of Neural Population Codes. Links to the lecture recordings will be posted here once they are available!

Part 1: Neural Tuning, Geometry, and Procrustes shape distance

Authors: Alex Williams, Sarah Jo Venditto, Grant Zempolich, & Pierre-Etienne Fiquet

Build conceptual and mathematical intuition of Procrustes distance and it’s extension to linear invariance.

https://colab.research.google.com/drive/1jy5d0gqR9-DKFpQcQiCjI0_F12DhODy0?usp=sharing

Part 3: Exploiting Metric Space Properties in Neural Populations

Authors: Alex Williams, Sarah Jo Venditto, & Shujun Xiong

Demonstrate how Procrustes distance can be used on real data: computing neural similarity between brain regions and animals using data available through AllenSDK.

https://colab.research.google.com/drive/1fAhi1_JlbxaBJXnTReNvthW2KcJ9j2qb?usp=sharing

Part 2: Relationship of Procrustes to RSA and CKA

Author: Alex Williams

Build mathematical intuition of RSA and CKA, understand conditions under which they are equivalent, and show how these metrics are related to Procrustes distance.

https://colab.research.google.com/drive/1WIAN5jDCTGU8KaRga2jwu4LcCo6AmVw9?usp=sharing

Part 4: Measuring Distance between Mismatched Representations

Author: Chaitanya Kapoor

Soft-matching and partial soft-matching: Extend Procrustes distance to situations when 1-to-1 matches between neurons is not possible or feasible.

https://colab.research.google.com/drive/1hksgmS67mqLr3V4_H0faNcv75FuqHnMS?usp=sharing