Given a true propositional input (e.g., 'Paris is the capital of France'), ablating along any basis vector of this cone disrupts the model's ability to generate a truthful response.

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Individual cone basis vectors may correspond to interpretable semantic facets of truth such as temporal facts, geographic facts, or commonsensehypothesis0.780
  Future direction hypothesis for giving semantic meaning to individual axes
• We hypothesize that explicitly instructing the model to evaluate the correctness of the given statement may change the geometry of truth directions.hypothesis0.772
  Motivating hypothesis for Section 5's investigation of prompt template effects.
• The two-dimensional subspace reported by Burger et al. (2024) seems to reflect a stage of transition in the model's processing, rather than a universal property of truth directions.quote0.768
  Load-bearing interpretive claim about the layer-specificity of Burger et al.'s finding.
• With unrestricted vocabulary, models occasionally respond in non-English Yes/No equivalents (e.g., Sí, Nein) after truth-direction interventionsfinding0.761
  Suggestive evidence for language-independent truth representation in LLMs
• The relationship between representations of truth of input statements and of model outputs in conjunction with model performance has not been investigated.question0.757
  Future work direction identified in conclusion for enabling reliable truth assessment methods.
• The results of abductive reasoning (reduced model priors) can be communicated to other agents as prior beliefs, provided all agents share the same model lexicon or hypothesis space.claim0.755
  Explanation of how knowledge (not just parameters) is shared between agents; links to pre-Cartesian consciousness
• The likelihood of a dedicated feature for a concept (element, city, animal, food) follows a sigmoid in log-frequency of the concept in training data, with threshold frequency inversely proportional to number of alive features.finding0.754
  Quantitative relationship between concept frequency and feature presence.
• Truth may be linearly separable in the model's representation space, but the structure is richer than a single linear axisclaim0.754
  Interpretive synthesis of DIM and cone intervention successes