Ambiguous anchors (33-27=60, 11-9=20) yield four distinct arithmetic interpretations across M1-M4

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

• Ambiguous 2-shot anchors yield four distinct interpretations across M1-M4 (P_abs-mult, P_add x2, P_signed-mult)finding0.820
  E1 finding showing that near-threshold, marginal model differences tilt to qualitatively different bindings
• Adding a single disambiguating example (12−9=21) aligns divergent M1-M4 interpretations under tested seedsfinding0.776
  E1 finding consistent with threshold-crossing: near-threshold state resolved by one additional anchor
• Mid-layers (6-15) achieve peak anchoring because semantic structure differentiates while maintaining coherence, forming a Goldilocks zoneclaim0.745
  Interpretation of E3 layer-wise results; motivates targeted UCCT interventions at layers 8-12
• Higher-activating feature intervals are systematically more interpretable than lower-activating intervals in human analysisfinding0.742
  Shows interpretability correlates with activation strength, most model effect comes from high activations
• Options contracts and ambiguous centers both encode multiple simultaneous readings as counterfactual structures.claim0.738
• Commonsense reasoning shows uniform but weaker anchoring (S ≈ −2.15)finding0.738
  Task-specific comparison.
• Meta-LLaMA-3.1-8B-Instruct shows optimal anchoring at layer 9 (S ≈ −1.90, median peak layer ℓ* = 10 [IQR 0.384])finding0.738
  E3 result establishing the Goldilocks zone at mid-layers for LLaMA architecture
• Math and code tasks show strongest mid-layer anchoring on LLaMA (S ≈ −1.65 at layers 8-12)finding0.729
  Task-specific E3 finding showing compositional reasoning requires deeper processing