Speaking in front of cartoon avatars: A behavioral and psychophysiological study on how audience design impacts on public speaking anxiety in virtual environments

Public speaking anxiety is defined as a strong difficulty in speaking in front of an audience and has been shown to impair work performance and social relationships. Virtual Reality (VR) offers an efficient tool for modulating public speaking anxiety through a wide range of customizations concerning environmental settings. However, scientific research needs to understand better what features of the simulated environment are more important to increase or reduce participants’ perceived discomfort. The present study investigates the role of visual (human vs cartoon characters) and acoustic (human vs robotic voice) audience features on perceived anxiety, sense of presence, and perceived realism in an interactive VR public speaking scenario. 42 participants (mean age = 24 y. o; Females = 30) performed four public speaking sessions, characterized by different levels (high vs. low) of graphic and acoustic audience design. Both explicit (questionnaires) and implicit physiological measures (Electrodermal activity-EDA) collected during audience interaction were used to assess the participants’ experiences. The results showed that the features of the simulated audience played a crucial role in perceived anxiety during a virtual public speaking. Specifically, the more realistic level of graphic and acoustic stimuli resulted in higher levels of self-reported anxiety as compared to the lower realism level. However, the experienced realism and the sense of presence seem more affected by the graphical than acoustic features of the virtual environment. By contrast, the acoustic features impact on the interaction realism with the virtual audience. Interestingly, the robotic voice (lower acoustic realism) increased electrodermal response during the interaction with the audience, interpreted as a break in the sense of presence. A positive correlation between anxiety, sense of presence, and experienced realism was found. As well, perceived anxiety is correlated with electrodermal activity during the performance. Nevertheless, physiological activity is more affected by the first experience than the realism features, suggesting the presence of a habituation effect across the repeated sessions. Taken together, the results of our study showed that multisensory features (graphical and acoustic) of the virtual environment play a fundamental role in creating realistic public speaking experiences and might be used within gamification strategies for soft skill training (e.g., for improving public speaking anxiety).