Over the last 20 to 30 years, the interest in emotion has increased dramatically not only in psychology, but also in the social sciences and the humanities (see for an overview Wetherell 2012). In ‘cultural studies,’ emotion - or ‘affect’ - has been treated as a concept that allows research to move beyond notions of discourse and cognition to the realm of the unconscious (e.g. Massumi 1996) in an interesting parallel to psychologists who have investigated emotion to move beyond a too rational and information-processing concept of cognition (e.g.Haidt 2012). At the same time we have witnessed a move in the empirical sciences, especially in the study of social interaction, from a focus on emotion as feeling to emotion as behaviour. This paper on emotion display seeks to bring together and relate this work on emotion behavior (expression of the experienced feeling). Over the last decades, research by psychologists, social scientists, and linguists has come to reconsider the cause-and-effect relation between feeling and emotion. There is not a one-to-one relation between feeling and display and emotion displays are embedded in social interaction (as social actions with particular formats, e.g. laughing, laugh particles, cries) and behave according to its normative organization. They can be addressed at others. The resources for displaying emotion are widely varied. Although the examples above deal primarily with audible displays, it is obvious that embodied displays such as facial expressions of emotion play a major role in face-to-face interaction. For language as a resource, Ochs and Schieffelin (1989) and Foolen (2012) show that emotion display can be achieved through all linguistic levels, including sound quality (e.g. smiley voice), prosody (e.g. screaming), morphology (e.g. diminutive suffix), lexicon (e.g. swear words), word order, code-switching, and interactional activities such as problem telling (Couper-Kuhlen (2012), distinguishing between ‘initiators’, ‘concurrents’ and ‘terminators’. A particularly detailed study of the resources for doing emotion in telephone calls is Hepburn’s (2004) study of different forms of crying. On the basis of such design features as different sounds, different lengths, and different amplitudes, and the possibility that these differences are consequential for the interaction she distinguishes between ‘whispering,’ ‘wet sniff,’ ‘snorty sniff,’ ‘wobbly voice,’ ‘high pitch,’ ‘aspiration,’ ‘sobbing,’ and ‘silence.’ Not only do emotion displays take different forms, they also do a variety of interaction work. Holt (2010) for example showed that joined laughter can do the work of closing a topic. Ikeda and Bysouth (2012) have shown in a study of Japanese multi-party talk how laughing can be used either as a recipient token to acknowledge the speaking rights of the other as present speaker, or as a turn-initial device to claim the next turn. They show how these two types of laughter are prosodically different, but also how laughing is part of a multimodally accomplished action in which also gaze direction and body position play a role. Laughter then can be consequential for how the interaction proceeds. Shaw, Hepburn and Potter (2012) have shown how laughter which follows a turn’s completion can be used to modulate the action performed by that turn. In a similar way, Clift (2012) shows how laughter particles inserted into complaints do the work of modulating the complaint and thereby invite the addressee to respond in a manner different from that to a straightforward complaint. Several examples so far have shown that emotion displays can be unilateral, but quite often they are reciprocated by co-participants for example in subsequent or simultaneous turns. But responses to emotion displays encompass more than responses to laughing or crying. The design of talk in terms of linguistic (Ochs and Schieffelin 1989, Foolen 2012) or embodied choices (Sorjonen and Peräkylä 2012), and the actions performed in talking such as complaining (Heinemann and Traverso 2009) often carry an affective or emotional stance, and this stance can be acknowledged by responding to it with affiliating or dissaffiliating responses (Koole 2003, Lindström and Sorjonen 2013). Couper-Kuhlen (2012) for instance showed that affiliating responses to story-telling are produced as an immediate response with prosodic matching or upgrading, while non-affiliative forms show delay and prosodic downgrading. A particular research issue is how professionals deal with emotional clients in institutional services such as emergency call-centres (Whalen and Zimmerman 1998, de Widt et al. 2014), medical consultations (Ruusuvuori 2007), or a child abuse helpline (Hepburn and Potter 2007, 2012). This line of research shows that responses to emotion displays are oriented to the institutional context of the talk. Suggestions have been made that affective or emotional stance is an omnirelevant phenomenon in social interaction and that all our talk displays an emotional or affective stance (Stevanovic and Peräkylä 2014). The research on emotion behaviour that is presented in this paper offers an epistemological perspective that is relatively absent in particular in psychology research on emotions and emotion displays. It follows the ‘emic’ perspective of participants, who have no access to the mental states of their co-participants but only to their semiotic behaviour manifested in social actions: emotion thus constitutes an interactional reality, an interpretation on which participants may come to agree, but not necessarily a mental reality. This perspective may be called agnostic (Hopper 2005) with regard to emotions-as-feelings, but this does not in principle deny that a display could reflect a feeling.