cs.CLNov 10, 2026

Representational Atoms in Neural Networks

Authors: Chenhui Hu, Pengfei Cao, Kang Liu, Jun Zhao

Organizations: The Key Laboratory of Cognition and Decision Intelligence for Complex Systems, Institute of Automation, Chinese Academy of Sciences, Beijing, China. · School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, China.

Abstract

Understanding the internal representations of neural networks and characterizing their relationship with human conceptual structure are two critical questions for elucidating model mechanisms. The former reveals how learned knowledge is organized in internal representations, whereas the latter provides a basis for interpreting these representations through human conceptual structure. Here, we formally define fundamental representational units (FRUs) and establish two evaluation criteria: fidelity and stability. We further prove that threshold-activated sparse autoencoders (TSAEs) can identify FRUs with high fidelity and stability under appropriate conditions, and validate this result through large-scale experiments. Further analyses reveal that FRUs with higher stability exhibit greater monosemanticity and can be composed across layers to recover semantic organization corresponding to human conceptual structure. These results provide a unified framework for defining, evaluating and identifying FRUs in neural networks, and establish a foundation for characterizing the relationship between neural network representations and human conceptual structure.

Figures & tables

Appendix figures & tables13 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. Do Sparse Autoencoders Capture Concept Manifolds?

    Apr 30, 2026Usha Bhalla, Thomas Fel, Can Rager +9Neural Representation GeometrySparse Autoencoders

  2. ConceptTracer: Interactive Analysis of Concept Saliency and Selectivity in Neural Representations

    Apr 8, 2026Ricardo Knauer, Andre Beinrucker, Erik RodnerNeural Network InterpretabilityNeuron-Level Interpretability

  3. Neural Functional Alignment Space: Brain-Referenced Representation of Artificial Neural Networks

    Feb 28, 2026Ruiyu Yan, Hanqi Jiang, Yi Pan +4Neural Representation GeometryCross-Modal Alignment