TY - RPRT TI - Disambiguating Moral Circuits and Moral Features in Language Models: Moral content, normative judgment or cue words? Evidence from 14 models AU - Davide Liga AU - Liuwen Yu AU - Matteo Magnini PY - 2026 UR - https://searcharxiv.com/abs/2611.21705 ID - 2611.21705 N1 - Source identifier: 2611.21705 ER -