The Answer-Basin Representation Hypothesis: We Are Not Probing or Steering Concepts

Excerpt

arXiv:2609.24821v1 Announce Type: new Abstract: The Linear Representation Hypothesis associates high-level concepts with directions in language models, but it remains unclear how these concept-related linear structures are organized within the model. We propose the Answer-Basin Representation Hypothesis: the probability measure induced over answers by the model's continuation distribution organizes these linear structures, with its statistics represented along linear directions shared across que