A new representation of emotion in affective computing

In the recent years, increasing attention has been paid to the area of affective computing, which deals with the complex phenomenon of human emotion. Therefore, a model for describing, structuring, and categorizing emotional states of users is required. The dimensional emotion theory is one of widely used theoretical foundations for categorization of emotions. According to the dimensional theory, emotional states are projected to the affective space, which has two dimensions: valence and arousal. In order to navigate in the affective space, Cartesian coordinate system is used, where emotion quality is defined by combination of valence and arousal. In this paper, we propose another representation of the affective space with polar coordinate system. The key advantages of such a representation include (1) capability to account not only for emotion quality, but also for emotion intensity, (2) reasonable explanation of the location of neutral emotion in the affective space, and (3) straightforward interpretation of the meaning of an emotional state (quality defined by angle and intensity defined by distance from the origin). Although in our experiment most of the induced emotions can be differentiated with polar coordinate system, further investigation is still needed to find out either Cartesian or polar coordinates system represents affective space better in practice.