Are emotions natural kinds? The debate is often framed as a choice between biologically fixed kinds with invariant essences and culturally constructed categories lacking natural reality. I argue that this framing is too restrictive because it conflates two independent questions: what fixes the content of emotion concepts, and what unifies natural kinds. Cross-classifying externalist and internalist accounts of folk-psychological content with essentialist and causalist accounts of kind unity yields four positions. Externalist-essentialism fails because public markers such as facial expressions and physiological profiles supply no invariant membership criteria. Internalist-essentialism relocates essences to cognitive architecture, but it demands sharp internal boundaries that prototype effects, dimensional organization, and neural degeneracy do not support. Externalist-causalism abandons essences, yet identifies emotions through folk causal roles rather than the mechanisms that generate emotional episodes. I argue that only internalist-causalism meets empirical and metaphysical constraints. On this view, emotions are natural kinds in virtue of stable internal causal mechanisms: distributed patterns of neural, cognitive, physiological, and behavioral coordination that produce context-sensitive responses. The account preserves scientific realism while accommodating cultural and individual variation.