Apparent success on binary detection tasks is entirely explained by mechanical logit shifts that bias models toward affirmative responses regardless of question content

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

• Automated logit weight prediction achieves 74% mean accuracy for features vs 58% for neurons vs 50% chancefinding0.759
  Automated interpretability of logit weights confirms feature downstream effects are more interpretable than neuron effects
• Logit-based self-report unmasks introspective capacity that greedy decoding concealsclaim0.759
  Central methodological contribution: computing probability-weighted expected value over digit-token logits recovers continuous, informative signal
• Binary detection adjusted accuracy reaches 97.3% at layer 0 with α=5 before baseline control is appliedfinding0.758
  The misleadingly high result that prior paradigm would report as evidence of introspection
• Researcher bias and the hardware lottery contribute to apparent convergence in AI models beyond the proposed theoretical pressuresclaim0.755
  Alternative explanation for observed convergence: AI community designs systems to mimic human reasoning
• We hypothesize that native self-report, fine-tuned introspection models, and trained activation-to-language systems will show different performance on bias-resistant localization and strength benchmarkshypothesis0.752
  Comparative prediction motivating future work contrasting different approaches to LLM self-knowledge
• Logistic regression fails to identify the true feature direction when a confounding feature is non-orthogonal to the truth direction, converging instead to the maximum margin separatorclaim0.746
  Motivates the introduction of mass-mean probing as an alternative to LR
• Under ask-correct, probes trained on arithmetic tasks A1-A3 generalize almost perfectly to factual tasks F0-F2 (AUROC ~1.0), whereas under no-prompt this generalization is largely absent.finding0.745
  Key improvement in cross-task generalization enabled by explicit instruction framing.
• We hypothesize that degraded generalization on benchmarks like MMLU may reflect the computational demands of the tasks.hypothesis0.745
  Connecting the paper's task-difficulty findings to prior observations of weak generalization on complex QA benchmarks.