Emotion Measurement from Attention Analysis on Imagery in Virtual Reality

Measuring human emotion from affective interaction is an important part of computing for mental health applications. Our work refers to non-obtrusive emotion measurement from eye tracking that gets increasingly ubiquitous, e.g., in Virtual Reality (VR) technology. We present a concept and first results of a feasibility study in which affectively weighted imagery of an image database is presented and integrated using discriminative observation, multi-object tracking and video-gaming. We measured attention preference for affective image classes and found relevant correlation with data extracted from a questionnaire on emotional states (MDBF) that would eventually substantiate basic valence classification from eye tracking data. The playful approach using the well-known concentration (pairs) game enables frequent repetition of the measurements in mental health care, therapeutic or pedagogical scenarios.