question
active
question:can-clinical-concepts-be-selectively-steered-without-damaging-unrelated-performanceCan clinical concepts be selectively steered without damaging unrelated performance?
Question about the feasibility of safe concept steering in EEG models.
Source paper
extracted_from(2026) · William Lehn-Schiøler · Magnus Ruud Kjær · Rahul Thapa · M. Pedersen +9
Neighborhood — ranked by edge-count
Findings (2)
finding
- Age-pathology confounding observed: impossible to suppress one concept without corrupting the other.answered_byEmpirical demonstration of entanglement between age and pathology features.
- Observation of catastrophic performance drop when steering certain concepts.
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.
- Can concept steering interventions on EEG foundation models be made selective rather than globally destructive?question0.799Research question motivating the introduction of the probe area metric and identification of operational regimes
- Main empirical finding of the concept steering analysis
- Key methodological insight: introspection enables a new probe validation criterion beyond conventional separation metrics
- Core research question driving the mechanistic investigation.
- No redundancy criterion.
- Addresses skeptical alternative that reports reflect only conversational content
- Central motivating question of the paper; the model organism approach is the proposed answer.
- There may exist a global introspective faculty or steering direction that improves introspection uniformly across all conceptshypothesis0.754Framed as an open problem; current evidence only points to local pair-specific improvement