concept
active
concept:grounding-criterion

Grounding Criterion

Criterion requiring that description of internal state causally depend on that state; tested via concept injection to establish causal link.

Neighborhood — ranked by edge-count

Findings (1)

finding
  • Models can detect and identify injected concept vectors ~20% of the time at optimal layer/strength in Opus 4.1, with immediacy suggesting internal rather than output-inferred detection.

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

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

  • Symbol Groundingconcept0.804
    The capacity for 'true understanding' located in living tissue; invoked when discussing hybrid systems and what living tissue contributes that pure machines may lack
  • Accuracy Criterionconcept0.753
    Criterion requiring that model's description of internal state be accurate, distinguishing genuine introspection from confabulation.
  • Criterion requiring that causal influence of internal state on description be internal, not routed through sampled outputs; rules out pseudo-introspection via self-observation.
  • The groundconcept0.746
    The ultimate non-material reality behind matter, experienced when living structure opens a window to the I.
  • Harnad's foundational problem: how do abstract symbols acquire meaning; relevant to representationalism debates cited in Lyon.