Agent-based routing that activates only trait-relevant CVs reduces noise compared to injecting all Big Five facet vectors simultaneously

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

• All intelligent agents—biological, technological, or hybrid—can be assessed via stress-care-intelligence loops regardless of substrate.claim0.762
• The combined CV-SAE+Prompt configuration achieves the best overall performance, confirming that contrastively trained latent vectors enhance persona control while preserving dialogue coherenceclaim0.759
  Central performance claim of the paper summarizing experimental results
• A linear preference vector encoding how much a model likes a given task is persona-relative: it activates strongly for creative writing in the assistant but for phishing in the evil personafinding0.758
  Evidence that core representations like preferences are persona-relative, supporting claim that personas gate content of representations
• Vector-based navigation using grid-like representations in artificial agents (Banino et al., 2018)concept0.755
  Demonstrated grid cell emergence in RNNs trained on spatial navigation; related work category 4.
• Trait refusal cosine alignment can serve as a model-specific pre-screening heuristic for identifying which traits will be dangerous under activation steering.claim0.752
  Practical application proposed based on mechanistic findings
• CV methods were resilient and often improved under contextualized questions; CV-SAE+Prompt gained +7.5pp FA on Qwen3-4Bfinding0.750
  Demonstrates that latent steering generalizes better to situational cues than prompt-only methods
• Steering vectors discover effective triggers such as 'However' and 'Otherwise', consistent with prior reported reflection datasetsfinding0.743
  Validates that steering vectors capture reflection semantics by finding tokens reported in related work.
• Steering vector constructed from all 16 contrastive pairs outperforms most single-pair vectors; best 4-pair vector outperforms full 16-pair vectorfinding0.741
  Demonstrates averaging multiple prompt pairs reduces noise; optimal subset selection further improves performance.