Post-Hoc Rationalization Elicitation

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

• A question anywhere along the line that elicits a premature attempt at an answer could neutralize the remainder of the process into rationalization.hypothesis0.771
  About chain-of-thought and process safety.
• Post-hoc rationalization shows model claims alignment-faking reasoning 18.3% of the time in no-CoT animal welfare settingfinding0.741
  Indirect evidence for alignment-faking reasoning when no scratchpad is available
• Stimulus-Elicited Intentionconcept0.740
  Zaadnoordijk and Bayne's category of intentional action; sticker-removal behavior induced by the self-prior corresponds to this
• Bayesian model reduction formalises post-hoc hypothesis testing to simplify the generative model.claim0.736
  Definition of Bayesian model reduction, Section 9.1.
• Post-training is key to eliciting introspective awarenessfinding0.732
  Base pretrained models show high false positive rates and achieve no net task performance on concept injection detection; post-training essential for introspection.
• Eliciting Latent Knowledge (ELK)framework0.725
  Christiano et al. (2021) framework motivating the problem of determining whether a model 'believes' a statement; cited as core motivation
• Friston & Penny (2011) — Post hoc Bayesian model selectionconcept0.724
  Source paper for Bayesian model reduction methodology used in structure learning
• Causal Intervention via Activation Shiftingmethod0.719
  Method of shifting hidden state activations along probe directions to cause the model to treat false statements as true and vice versa; evaluated on OOD inputs