Paper

{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2602.20273v1</id>\n    <title>The Truthfulness Spectrum Hypothesis</title>\n    <updated>2026-02-23T19:01:31Z</updated>\n    <link href='https://arxiv.org/abs/2602.20273v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2602.20273v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Large language models (LLMs) have been reported to linearly encode truthfulness, yet recent work questions this finding's generality. We reconcile these views with the truthfulness spectrum hypothesis: the representational space contains directions ranging from broadly domain-general to narrowly domain-specific. To test this hypothesis, we systematically evaluate probe generalization across five truth types (definitional, empirical, logical, fictional, and ethical), sycophantic and expectation-inverted lying, and existing honesty benchmarks. Linear probes generalize well across most domains but fail on sycophantic and expectation-inverted lying. Yet training on all domains jointly recovers strong performance, confirming that domain-general directions exist despite poor pairwise transfer. The geometry of probe directions explains these patterns: Mahalanobis cosine similarity between probes near-perfectly predicts cross-domain generalization (R^2=0.98). Concept-erasure methods further isolate truth directions that are (1) domain-general, (2) domain-specific, or (3) shared only across particular domain subsets. Causal interventions reveal that domain-specific directions steer more effectively than domain-general ones. Finally, post-training reshapes truth geometry, pushing sycophantic lying further from other truth types, suggesting a representational basis for chat models' sycophantic tendencies. Together, our results support the truthfulness spectrum hypothesis: truth directions of varying generality coexist in representational space, with post-training reshaping their geometry. Code for all experiments is provided in https://github.com/zfying/truth_spec.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.LG'/>\n    <published>2026-02-23T19:01:31Z</published>\n    <arxiv:comment>28 pages, 26 figures</arxiv:comment>\n    <arxiv:primary_category term='cs.LG'/>\n    <author>\n      <name>Zhuofan Josh Ying</name>\n    </author>\n    <author>\n      <name>Shauli Ravfogel</name>\n    </author>\n    <author>\n      <name>Nikolaus Kriegeskorte</name>\n    </author>\n    <author>\n      <name>Peter Hase</name>\n    </author>\n  </entry>"
}