Exploring Real-Time Bio-Behaviorally-Aware Feedback Interventions for Mitigating Public Speaking Anxiety
Abstract
Effective public speaking skills are crucial to one’s academic and professional success. Individuals who are good in public speaking are more likely to graduate college and obtain a leadership position compared to their counter-peers. Yet, public speaking anxiety (PSA) is one of the most common social fear faced by people directly affecting one’s academic and professional success.
This Master’s thesis investigates the effectiveness of in-the-moment bio-behaviorally aware feedback in mitigating public speaking anxiety in a virtual training environment. The training environment exposes participants to various virtual stimuli and at the same time, captures their audio and physiological signals. These signals are used to extract bio-behavioral measures (e.g., speech intonation, electrodermal activity mean) and serve as an input to a machine learning model that provides real-time estimates of state anxiety. Based on these state anxiety estimates, the system provides real-time feedback of positive reinforcement and cognitive restructuring–grounded on theoretical rationale from behavioral sciences–when an increase in state anxiety is detected. The system is evaluated through a small-scale study of participants using a pre/post evaluation design.
Results indicate that in-the-moment feedback prompts provided to the participants affect their in-the-moment state-based anxiety. Statistical analysis indicates significant differences of biobehavioral measures before and after the in-the-moment feedback prompts. The self-reported POST-study results from the participants of the user study also indicate that 5 out of 7 participants found this study beneficial for their public speaking skills. Results of this work also highlight the effect of type of audience on the positive reinforcement feedback provided to the participants. It is observed that when the audience is negative, the positive reinforcement feedback prompts provided to the participants by the real-time model were more compared to a positive audience. Findings from this work provide a foundation toward designing artificial intelligence systems for personalized in-the-moment interventions for mitigating adverse behavioral outcomes.
Citation
Agarwal, Akansha (2021). Exploring Real-Time Bio-Behaviorally-Aware Feedback Interventions for Mitigating Public Speaking Anxiety. Master's thesis, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /195845.