The target vs. off-target probe area metric quantifies steering selectivity and distinguishes selectively steerable from entangled interventions.

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

• Concept steering with target vs off-target probe area metric reveals three operational regimes (selectively steerable, encoded but entangled, non-encoded) across SleepFM, REVE, LaBraM.finding0.886
  Result categorizing concept steerability into three distinct regimes.
• If steering in a purported concept direction does not shift self-report in the expected direction, probe quality becomes suspect, especially when conventional probe metrics alone looked acceptable.quote0.778
  Key methodological insight: introspection enables a new probe validation criterion beyond conventional separation metrics
• Focus→wellbeing steering: both probe entropy (1.09→1.67 bits) and report entropy (0.88→1.69 bits) increase monotonically with αfinding0.778
  Evidence that improved introspection in focus→wellbeing arises from enriched internal state and report channels simultaneously
• A probe may achieve high performance even on representations that are not causally relevant for the taskclaim0.775
  Key interpretive claim from Case Study II distinguishing probe accuracy from causal relevance
• Linear steering produces noisy off-target effects; manifold steering cleanly shifts probability mass between sequential concepts.finding0.772
  Core empirical claim comparing steering approaches on cyclic concepts.
• Probe-based data attribution bridges interpretability work (probes/activations) with data-centric alignment.claim0.771
  Conceptual framing: integrates mechanistic interpretability tools with alignment-focused data curation.
• manifold steering produces clean probability shifts along natural behavior structure; linear steering cuts across manifold and produces off-target noisy effectsfinding0.769
  Empirical demonstration on Llama-3.1-8B that steering along representation manifold aligns outputs with behavior manifold, whereas linear steering does not.
• Steering vectors capture latent dimensions of reflective behavior more faithfully than surface-level embedding similarity.claim0.764
  Supported by the instruction discovery experiments comparing steering vs. embedding baselines.