One important question in emotion science is determining what emotions there are. To answer this question, researchers have assumed either that folk emotion concepts are unsuitable for scientific inquiry, or that they are constitutive or explanatorily significant for emotion research. Either option faces a challenge from the cultural variability of folk emotion concepts, prompting debate on the universality of emotions. I contend that cultural variation in emotion should be construed as variations in components rather than entire emotional repertoires. To do this, I distinguish between hypotheses concerning emotional repertoires and those focused on specific emotional features within various cultural contexts. I hold that decisions regarding emotional repertoire hypotheses call for either revising current classification systems or maintaining them, but that, given underdetermination by evidence, this entails a preference for maintaining emotion taxonomies. This, in turn, leaves empirical hypotheses on specific emotional features as the most viable avenue for scientific inquiry.