hypothesis
active
hypothesis:known-entity-features-suppress-default-can-t-answer-and-unknown-name-featuresKnown-entity features suppress default can't-answer and unknown-name features.
Tested by promoting known-answer features on Batkin prompt.
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (2)
finding
- Activating unknown-names features is not sufficient to induce refusal on the Michael Jordan prompt.associated_withUnexpected result indicating incomplete circuit.
- Interventions support circuit hypothesis.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Hallucination circuit.
- Default refusal circuit.
- A priori unknown behavior discovered.
- Metacognition caveat.
- Which behaviors does a model represent internally, default to, can be pushed to amplify, or refuses to expose?question0.721The motivating diagnostic question that prompting alone cannot answer
- Limitation of the multiple-choice analysis.
- QK attribution for MC heads.
- Grok 4: baseline 2.24, prompted 6.48; Gemini 3.1 Pro: 1.97→6.18. Reflective mode exists but is suppressed in default interaction.