<p>It is common to refer to emotions as “positive” or “negative” – that is, to categorize emotions in terms of valence. Here we question the degree to which the concept of valence aids our understanding of human nature. We consider conceptual issues as well as empirical findings at the intrapersonal and interpersonal levels of analysis that cast doubt on the predictive validity of valence, suggesting that the notion of valence hampers rather than helps scientific understanding and progress. We argue that, rather than in explaining the effects of primary emotional experiences and expressions – which are better understood in terms of specific appraisals, action tendencies, and social functions – valence may have utility at the level of meta-emotions, that is, how people feel about how they feel. We propose that meta-emotion valence is constructed through assessments of the momentary goal congruence and sociomoral evaluation of primary emotions, and that people subsequently represent emotional episodes in terms of how they felt at the time relative to how they wanted to feel. We suggest that the valence of these meta-emotions predicts emotion regulation strategies, enables economical representation of experiences in memory, and characterizes social communication about emotional experiences. These functions at the meta-emotion level may help explain the pervasiveness of valence in human thinking about emotions, despite the conceptual and empirical issues that beset the construct. By repurposing valence at the level of meta-emotions, we hope to pave a constructive path forward towards a better understanding of the role of valence in the human mind.</p>