The sensitivity to think/don't think instructions may be achieved via a circuit that tags tokens as attention-worthy based on instructions or incentives

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

• Circuits could act as an epistemic foundation for interpretability by breaking down model behavior into falsifiable statements about small subgraphs.claim0.780
  Normative vision for how the circuits agenda could resolve the pre-paradigmatic state of interpretability
• The ability to distinguish injected thoughts from text likely relies on different attention heads invoked by different prompt partsclaim0.778
  Speculation about the mechanistic basis of the distinguishing thoughts from text experiment.
• Removing chain-of-thought in prompted setup nearly eliminates compliance gap for Claude 3 Opus in helpful-only and animal welfare settingsfinding0.773
  Contrasts with synthetic doc finding; suggests different mechanisms may be at play
• Task instructions can be transcribed into prior beliefs of a generative model, making instruction-following an instance of prior belief specification.claim0.772
  Practical implication showing task instructions are equivalent to inducing prior beliefs in experimental settings
• acting to optimize value and perception are two aspects of exactly the same principle; namely the minimisation of a quantity [free energy] that bounds the probability of sensory input, given a particular agent or phenotype.quote0.764
  Concise statement of the free-energy principle's unification of action and perception.
• Thought detection peaks at ~2/3 layer depth; intention checking peaks at ~1/2 layer depth.finding0.763
  Lindsey (2026) differential layer performance explained by Janus's path combinatorics — different tasks use different path distributions.
• Each attention head has two largely independent computations: a QK circuit computing the attention pattern and an OV circuit computing the effect if attended toclaim0.762
  Key decomposition enabling separate analysis of where attention goes and what it does
• Priming provided by the injected thought prompt heightens the model's ability to detect concept injectionclaim0.762
  Observation from alternative prompts that detection is weaker without setup.