Social Functions — 1 Social functions of emotions at four levels of analysis Dacher Keltner University of California, Berkeley Jonathan Haidt University of Virginia 12/1/98 This is the final manuscript that was published in Cognition and Emotion (1999), vol. 13, pp. 505-521 Abstract In this paper we integrate claims and findings concerning the social functions of emotions at the individual, dyadic, group, and cultural levels of analysis. Across levels of analysis theorists assume that emotions solve problems important to social relationships in the context of ongoing interactions. Theorists diverge, however, in their assumptions about the origins, defining characteristics, and consequences of emotions, and in their preferred forms of data. We illustrate the differences and compatibilities among these levels of analysis for the specific case of embarrassment. We close by suggesting research strategies that incorporate a social-functional perspective.Social Functions — 2 Social functions of emotions at four levels of analysis The primary function of emotion is to mobilize the organism to deal quickly with important interpersonal encounters (Ekman, 1992, p.171). Emotions are a primary idiom for defining and negotiating social relations of the self in a moral order (Lutz & White, 1986, p.417). Early studies of emotion tended to focus on the "intrapersonal" aspects of emotion, mapping the determinants and characteristics of emotional response within the individual1. Many of the initial functional accounts of emotion similarly highlighted how emotions solve problems within the individual, for example as "interrupts" that prioritize multiple goals of the individual (e.g., Simon, 1967; Tomkins, 1962). Several developments have led researchers to examine more closely the "interpersonal" functions of emotions. Researchers have begun to uncover how emotions structure relationships between parents and children (e.g., Bowlby, 1969), siblings (Dunn & Munn, 1985), and romantic partners (Levenson & Gottman, 1983). Emotions such as anger and embarrassment have been shown to have systematic effects upon other individuals (e.g., Averill, 1980; Keltner & Buswell, 1997; Miller & Leary, 1992; Tangney & Fischer, 1995). Ethological studies have shown how emotions guide social interactions such as courtship and appeasement rituals (Eibl-Eibesfeldt, 1989). Finally, the growing contact between anthropologists (Abu-Lughod, 1986; Lutz, 1988; Lutz & White, 1986) and psychologists (Haidt, Koller & Dias, 1993; Markus & Kitayama, 1991; Mesquita & Frijda, 1992; Russell, 1991) in the new field of cultural psychology (Shweder, 1989; 1991) has led to greater awareness of the ways that emotions construct and are constructed by cultural practices and institutions. These converging trends have inspired a wave of research and theory in a variety of disciplines on the connections between emotions and the social environment (Averill, 1980; Campos, Campos, & Barrett, 1989; Clark, 1990; Frijda, 1986; Frijda & Mesquita, 1994; Kemper, 1993; Lazarus, 1991; Lutz & Abu-Lughod, 1990; Markus & Kitayama, 1991; Plutchik, 1980; de Rivera & Grinkis, 1986; Tooby & Cosmides, 1990). Frijda and Mesquita (1994) have written usefully about the social and interactional functions of emotions, particularly anger, shame, and guilt. However we think the time is right for a more general discussion of the assumptions, claims, and empirical findings that can be brought together into a social functionalist perspective on the emotions. Our aims in this review are as follows. First, we discuss what it means to take a social functional approach in the study of emotion. Second, we review claims about the social functions of emotion in anthropology, ethology, history, psychology, and sociology, highlighting illustrative empirical findings and conceptual issues. We then apply a social functional analysis to embarrassment, and conclude with a discussion of needed lines of empirical and theoretical inquiry. We hope that this essay contributes some clarification to a growing field by distinguishing between social functions at four levels of analysis: 1) individual (intra-personal), 2)dyadic (between 2 individuals) 3)group (a set of individuals that directly interact and has someSocial Functions — 3 temporal continuity), and 4)cultural (within a large group that shares beliefs, norms, and cultural models)2 . As we describe below, researchers working at each level differ in the systems they refer to, their preferred kinds of data, and the theoretical traditions within which they explain the origins and defining characteristics of emotions. Our aim will be to specify the differences and similarities in the accounts offered at each of the four levels, and to show how these levels can be put together to create a more complete understanding of the social functions of emotions. Social functionalist accounts of emotion Functional explanations, although a bit more recent to the field of emotion (Keltner & Gross, this issue), have long been used in biology and the social sciences. Functional explanations refer to the history of some object (e.g., behavior or trait), as well as the regular consequences that benefit the system in which the object or trait is contained. As Merton (1949) stated, functionalist explanations hinge on "interpreting data by establishing their consequences for larger structures in which they are implicated". Functionalist accounts vary according to the kind of system being analyzed. For biological systems within an individual organism, a strong functionalism is usually appropriate, in which features were shaped or selected for the consequences they bring about. For example, the heart can only be understood as a pump working within a circulatory system “designed” by natural selection to fulfill a specific function -- allocating blood at variable rates -- within that larger system. At the cultural level of analysis, however, greater caution must be observed when making functional claims. Some institutions and cultural practices may have been designed to benefit the rich and powerful, as a Marxist might say, or to perpetuate themselves, as a meme-theorist might say (Dawkins, 1976). But because there is no over-arching selection mechanism culling out inefficient or poorly adapted cultures, one cannot assume, as Malinowski did in his early pronouncements, that every