In this thesis I argue that the category of emotion is a social natural kind and that this claim has implications for the scientific study of emotion. It is a social kind because it includes social features that are necessary for the explanation of emotional phenomena. It is a natural kind because it is a stable property cluster which shows an aptness for successful induction and explanation. In chapter 1 I introduce the main views on natural kinds in affective science. I present three different views to claim that all three rely heavily on an influential psychological theory, namely Basic Emotion Theory. I claim that the natural kind status of several affective phenomena is not exclusively a metaphysical or philosophical issue and show its implications for interpreting experimental data and designing experiments. This thesis is divided in two parts. Part A (chapters 2 and 3) deals with the metaphysics of natural kinds and social kinds. In chapter 2 I outline and reject the traditional view of natural kinds (essentialism) in favour of an epistemological view where natural kinds are property clusters showing an aptness for successful induction and explanation of the phenomena included in them. The epistemological view I endorse has similarities with accounts proposed by Boyd (1991) or Khalidi (2013), however, is primarily based on Slaterâs (2015) Stable Property Cluster account. I argue that Slaterâs account is a plausible account of natural kinds and has advantages over other available accounts in the literature. In chapter 3 I argue that we should reject the claim that social kinds cannot be considered natural kinds (or useful scientific categories) because they can be subject to heterogeneous construction, and because they are associated with interactivity and normativity features. I argue that in some cases, heterogeneous construction, interactivity between the clustered properties and the environment or normativity factors, do not necessarily entail that the social kind will not show an aptness for successful induction and explanation. I use the example of gender categories to argue that such social kinds can be stable property clusters and show an aptness for successful induction and explanation. Part B (chapters 4 and 5) applies the account defended in the first part to affective phenomena and more specifically to the category of emotion. In chapter 4 I claim that emotion is not a neurobiological kind but a social kind; its properties cannot be reduced to neurobiological properties or neurobiological states. I argue that most of the main views presented in chapter 1 rely on BET according to which emotion is a neurobiological kind. I present evidence against this theory and argue that it should not be relied upon to determine the natural kind status of the category of emotion because it oversimplifies, and does not provide a good explanation of, emotional phenomena. In chapter 5 I argue that emotion is a social kind because it relates to phenomena which are best investigated by looking at several domains in the social sphere such as social psychology, sociology and social ecology. To support my claim I first provide an account of emotion as a stable property cluster which performs a specific function e.g. to enable us to successfully interact with the environment; I then give examples of successful induction and explanation on the basis of which emotion can be considered a natural kind. I conclude that emotion is a social natural kind.