Decoding humans' and virtual agents' emotional expressions

A factor influencing the interaction between humans and virtual assistants is users' ability to decode their emotional expressions. To address this, a study is described that involves a confrontation between a naturalistic decoding task and a synthetic encoding task, with the synthetic encoding task realized through virtual agents expressing target emotions. The task contrast was examined within three age groupings (young, middle-aged, and seniors) of 270 participants. Participants classified pictures and video clips of humans or virtual agents, selected according to age (again, young, middle-aged, and old) and gender (female, male), in which the human or the virtual agent displayed seven basic emotions: disgust, anger, sadness, fear, happiness, surprise, and neutrality. Results show the influence of emotional category, age, gender, and face typology on recognition of synthetic and naturalistic emotional expressions.