Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session B: Wednesday, August 13, 1:00 – 4:00 pm, de Brug & E‑Hall

Similarity-based Representation Factorization for Understanding Representations in Minds, Brains and Machines

Florian P. Mahner¹, Martin N Hebart²; ¹Max Planck Institute for Human cognition and brain sciences, Max-Planck Institute, ²Justus Liebig Universität Gießen

Presenter: Florian P. Mahner

Understanding representations is a major aim in cognitive computational neuroscience, yet existing data-driven methods are limited in providing interpretable dimensions that also capture the underlying data structure. Here we propose Similarity-based Representation Factorization (SRF), a method that reliably decomposes data structures into interpretable, non-negative components based on similarity matrices. Through simulations and empirical data, we demonstrate that SRF is robust to noise and capable of revealing interpretable dimensions in both synthetic and behavioral similarity data. SRF opens new possibilities for uncovering the dimensions that underlie similarity even in smaller and noisier datasets, thus offering a principled approach for interpreting representational structure.

Topic Area: Methods & Computational Tools

Extended Abstract: Full Text PDF