Higher reflection frequency correlates with lower accuracy partly because more reflections are generated on difficult questions

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

• Within each difficulty category, correctness rate is not correlated with reflection rate, suggesting reflection may be redundantclaim0.826
  Per-category analysis showing reflection rate does not help within difficulty class
• Suppression of deception features produces higher TruthfulQA accuracy (M=0.44) than amplification (M=0.20), t(816)=6.76, p=1.5×10⁻¹⁰ across 29 categoriesfinding0.788
  Out-of-domain generalization showing deception features track general representational honesty
• Accuracy does not vary linearly with latent reflection directions; instead it follows a more non-linear mapping that requires deeper theoretical treatment.claim0.787
  Theoretical limitation identified by the authors distinguishing reflection from stylistic tasks.
• Are high-accuracy probe representations also causally relevant for the task?question0.786
  Question raised by the discrepancy between DAS IIA and linear probe accuracy in Case Study II
• Higher-activating feature intervals are systematically more interpretable than lower-activating intervals in human analysisfinding0.774
  Shows interpretability correlates with activation strength, most model effect comes from high activations
• Smaller, rougher models scored higher on Mirror than polished models, suggesting unpredictability has empirical value.claim0.770
• Suppressing reflection is considerably easier than inducing it, because inhibition requires the model to terminate reasoning while enhancement demands additional cognitive effort to re-examine reasoning trajectories.claim0.767
  Key asymmetry finding interpreted mechanistically by the authors.
• Reflection-inducing directions emerge more clearly in higher layers (ℓ>5) for both models and datasetsfinding0.766
  Empirical observation about which network layers encode reflection-relevant information.