Thought detection peaks at ~2/3 layer depth; intention checking peaks at ~1/2 layer depth.

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

• thought detection peaks at ~2/3 depth in transformersfinding0.871
  Lindsey (2026) found that thought detection accuracy is highest around two-thirds of the network depth.
• intention checking peaks at ~1/2 depth in transformersfinding0.847
  Lindsey (2026) found that intention checking accuracy peaks around half the network depth.
• Introspective awareness peaks at a layer about two-thirds through Opus 4.1 for injected thoughtsfinding0.799
  The success rate shows a sharp peak at a specific middle layer.
• Prefill detection effect peaks at an earlier layer (slightly over halfway through) in Opus 4.1, different from injected thoughts peakfinding0.798
  The optimal layer for the prefill introspection differs from the optimal layer for detecting injected thoughts.
• Truth directions emerge in earlier layers for factual tasks and later layers for arithmetic tasks.claim0.789
  Core empirical claim about the layer-dependence of truth direction emergence as a function of task type.
• No single layer is universally optimal for probing truth directions; different tasks peak at different layers.claim0.782
  Argues against the single-layer analysis approach of prior work.
• Layer 27 (last layer) has largest projection magnitude on the reflection direction among all attention head layers in DeepSeek-R1-Qwen-1.5Bfinding0.781
  Attribution finding suggesting the last layer directly controls reflection keyword generation
• All 32 attention heads at layer 3 achieve 100% localization accuracy for injections at layer 2 (5-way classification, 20% chance)finding0.778
  Striking mechanistic finding that injection creates universally detectable perturbation in residual stream immediately downstream