Conceptual and methodological difficulties in emotion research are partially due to our tendency to take emotion as “natural kind” or “natural category.” Due to this, many past studies paid tremendous effort to segment and categorize types of emotions. To advance studies in emotion, I suggest that we should take affects as meta-cognition of emotions. We then should utilize four indexes including subjective, physiological, expressive, and communicative. Subjective index is based on introspective observation of the person who is experiencing the emotion by means of verbal or behavioral expression. Physiological index includes heart rate, skin conductance, or evoked brain activities. Expressive index includes facial or vocal expressions as well as bodily movement, anything that are measurable by means of remote sensors. Communicative index includes observers’ interpretation of the state of the subject. These four indexes are at different levels of measurements, but all of these could be defined operationally. In fruitful emotion studies, I suggest we should utilize at least two types of these indexes. Based on this framework, we introduce five studies including category and dimension of facial expressions, detection and control of anger, emotional dynamics in face-to-face communication, reinforcement learning measurement of emotional valence, contextual effects of emotion and its relation to language.