Eerola et al.'s multi-layered model for the pleasure taken in music-elicited sadness is an important step in integrating explanatory accounts from different disciplines. Here, I would like to point out two theoretical inconsistencies that lie at the very root of the paradox and can have a major impact on future studies: first, the conflation of the transformation hypothesis and the co-activation hypothesis of sadness and pleasure, and second, the conflation of sadness that is elicited by musical instruments and that which is elicited by lyrics.