Using Physiological Recordings for Studying User Experience: Case of Conversational Agent-Equipped TV

Abstract

ABSTRACTThis study explores TV viewers’ user experience (UX) of the conversational agent (CA) assisted interactions while watching TV. In the human-computer interaction field, the user experience is primarily measured through performance and subjective self-reported data. But TV is an entertainment medium, the user experience of TV viewers should be determined based on how much the TV viewers enjoyed and immersed, rather than productivity or performance. It is also necessary to collect objective data as well as subjective self-report results of emotional experiences. Although the CA is becoming increasingly common in everyday life, the user satisfaction is still not high. However, this is mainly a result of evaluation in the performance aspect, and few studies have been carried out on the overall user experience including emotional aspects. To comparatively analyze the user experience with the CA interface and remote control unit (RCU) interface, we adopted physiological measurements for objective data as well as self-report questionnaires. Among the physiological measurement results, skin conductance (SC) differed between CA and RCU interactions. SC was high at the beginning of the CA interaction but decreased over time. In RCU sessions, SC maintained a constant value or rather increased. In the self-reporting results, the ergonomic quality of the RCU was higher than that of the CA. However, CA was evaluated more positively in terms of hedonic quality. CA had a greater effect on the overall attractiveness while having more emotional appeal. Additionally, a partial correlation between objective data and self-report results could be observed. In sum, subjective self-reports showed that CA provides a more positive TV UX with regard to emotional aspects. Physiological measurement results provide objective data for further understanding user experience by showing specific differences in cognitive involvement and emotional response changes during user interactions.