TY  - RPRT
TI  - Do LLM Self-Explanations Help Users Predict Model Behavior? Evaluating Counterfactual Simulatability with Pragmatic Perturbations
AU  - Pingjun Hong
AU  - Benjamin Roth
PY  - 2026
UR  - https://arxiv.org/abs/2601.03775
ID  - 2601.03775
ER  -