2-shot reinterpretation of '-' yields 23 for 15-8 on held-out query

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

• Two-shot redefinition of "−" operator flips model output from -1 to 23 on 15-8=?finding0.771
  Demonstration of strong prior rebinding via small coherent anchors
• Two exemplars (2−3=5, 7−4=11) induce reinterpretation of '−' as addition on held-out queries across mainstream LLMsfinding0.760
  E1 qualitative finding demonstrating anchor rebinding of strong arithmetic prior
• Ambiguous 2-shot anchors yield four distinct interpretations across M1-M4 (P_abs-mult, P_add x2, P_signed-mult)finding0.723
  E1 finding showing that near-threshold, marginal model differences tilt to qualitatively different bindings
• k50 for base-10 two-digit addition: 0.28 ± 0.05 shotsfinding0.702
  Shot midpoint from logistic fit over 10 runs.
• DS-v3.2 incrementing bid 10→850 over 49 sole-bidder roundsfinding0.700
  possible 'treats-own-bid-as-competitor' pathology in one trace
• DeepSeek v3.2 increments bid from 10 to 850 over 49 sole-bidder roundsfinding0.688
  One DS-v3.2 trace shows extreme self-escalation, suggestive of treating own bid as competitor.
• Qwen-2.5-3B ASR drops from 98.6% at dim 1 to 45.1% at dim 2, recovering partially then declining to 65.3% at dim 5finding0.680
  Smaller models show non-monotonic and diminished ASR with increasing cone dimensionality
• Calibrated few-shot prompting was a surprisingly weak baseline for truth classification compared to linear probesfinding0.678
  Unexpected finding that behavioral baseline underperforms representational probing approaches