In the scientific study of the emotions the goal is to find natural kinds. That is, to find categories about which interesting scientific generalizations and predictions can be formed. Core affect is dimensional approach to the emotions which claims that emotions emerge from the more basic psychological processes of valence (pleasant/unpleasant) and arousal (activation/deactivation). Lisa Feldman Barrett (2006b) has recently argued that the discrete emotion approach has failed to find natural kinds and thus should be dismissed as a failed paradigm. She offers core affect as an alternative theory that will better capture natural kinds in emotionally salient phenomena. In this thesis I evaluate Barrett’s claim on the basis of a philosophically robust understanding of natural kinds and a careful assessment of the empirical evidence. I argue that while core affect is not a natural kind, subsets of core affect space may be natural kinds.