Abstract
People are remarkably good at detecting familiarity with actor (Loula et al., 2005), recognizing the gender (Pollick et al., 2005), emotions (Atkinson et al., 2004) and actions of an actor when presented as biological motion. For many of these tasks the influence of the type of stimuli display (point light display, virtual avatar, full light video) on participants' performance has been well researched (McDonnell et al., 2009). The effect of the size of the display, however, remains underinvestigated. According to our hypothesis, a naturalistic environment and stimuli display would enhance performance, in particular for challenging tasks. We motion captured eight actors, who were asked to portray the following ten emotions while seated: amusement, anger, disgust, fear, joy, pride, relief, sadness, shame, and surprise. The resulting 80 motion sequences were then applied to a stick figure and used for the emotion recognition study. As a between participant factor, the stick figure animations were presented either on a laptop screen or on a large back projection surface. In the latter condition the size of the stick figure matched the natural size of the actors. Thirty-two participants (16 female) took part in a between-subject study (gender balanced). For each stimulus the participant had to make a ten-alternative forced choice to categorize the animation as one of ten emotions. Recognition accuracy was significantly higher for natural size condition (38% accuracy for back projection condition vs. 31% for desktop), and reaction time was lower (2.3 animation repetitions for back projection condition vs. 2.7 for desktop condition). In both conditions the emotional categories were an important factor as some emotions were more easily recognized than others. The results show that for complex tasks, e.g. discrimination among multiple emotional categories, enhanced naturalness of stimuli can be beneficiary for the observer.
Meeting abstract presented at VSS 2013