Digital mental health prevention and therapy programs are increasingly integrating embodied conversational agents and avatars to improve user interaction and psychological well-being. However, the effects of human-like avatar embodiment remain underexplored, particularly in digital interventions targeting emotion regulation, a critical transdiagnostic factor of various mental health conditions. Thus, the current study investigates the impact of avatar realism in a brief online emotion regulation intervention.In a between-subjects online experiment, 211 participants completed a single-session digital emotion regulation intervention, featuring psychoeducational content and practical exercises, facilitated by an embodied conversational avatar. Participants were randomly assigned to one of four conditions: realistic humanoid avatar, abstract humanoid avatar, robot avatar, or a control condition using audio waveform animations. Emotional ratings, satisfaction with the intervention, avatar user experience, self-disclosure levels, and open-ended feedback were assessed through self-report.Positive emotions increased across all conditions, except for the realistic avatar group during a gratitude exercise. Results further revealed a significant main effect of condition on intervention satisfaction, with the realistic humanoid avatar receiving lower ratings than the other conditions. Interactions with the realistic avatar were rated as significantly more anxiety-inducing and less pleasant. Additionally, participants in the realistic humanoid avatar condition reported significantly lower levels of self-disclosure compared to the other conditions.These findings suggest that realistic humanoid avatars may induce discomfort in emotion regulation interventions, potentially reducing their effectiveness. The results are discussed considering the uncanny valley and perceptual mismatch hypotheses, highlighting the need for careful multimodal avatar design in digital mental health interventions.
Building similarity graph...
Analyzing shared references across papers
Loading...
Eva Naumann
Computers in Human Behavior Reports
SHILAP Revista de lepidopterología
Building similarity graph...
Analyzing shared references across papers
Loading...
Eva Naumann (Wed,) studied this question.
www.synapsesocial.com/papers/69a75cabc6e9836116a25b8f — DOI: https://doi.org/10.1016/j.chbr.2026.100952