Affective Embodied Conversational Agents for Natural Interaction

Abstract

Human computer intelligent interaction is an emerging field aimed at providing natural ways for humans to use computers as aids. It is argued that for a computer to be able to interact with humans it needs to have the communication skills of humans. One of these skills is the affective aspect of communication, which is recognized to be a crucial part of human intelligence and has been argued to be more fundamental in human behaviour and success in social life than intellect (Vesterinen, 2001; Pantic, 2005). Embodied conversational agents, ECAs (Casell et al., 2000), are graphical interfaces capable of using verbal and non-verbal modes of communication to interact with users in computerbased environments. These agents are sometimes just as an animated talking face, may be displaying simple facial expressions and, when using speech synthesis, with some kind of lip synchronization, and sometimes they have sophisticated 3D graphical representation, with complex body movements and facial expressions. An important strand of emotion-related research in human-computer interaction is the simulation of emotional expressions made by embodied computer agents (Creed & Beale, 2005). The basic requirement for a computer to express emotions is to have channels of communication such as voice, image and an ability to communicate affection over those channels. Therefore, interface designers often emulate multimodal human-human communication by including emotional expressions and statements in their interfaces through the use of textual content, speech (synthetic and recorded) and synthetic facial expressions, making the agents truly “social actors” (Reeves & Nass, 1996). Several studies have illustrated that our ability to recognise the emotional facial expressions of embodied computer agents is very similar to that of identifying human facial expressions (Bartneck, 2001). Related to agent’s voice, experiments have demonstrated that subjects can recognize the emotional expressions of an agent (Creed & Beale, 2006) whose voice varies widely in pitch, tempo and loudness and its facial expressions match the emotion it is expressing. But, what about the impact of these social actors? Recent research focuses on the psychological impact of affective agents endowed with the ability to behave empathically with the user (Brave et al., 2005; Isbister, 2006; Yee et al., 2007; Prendinger & Ishizuka, 2004; Picard, 2003). The findings demonstrate that bringing about empathic agents is important in