Related Experiment Video
Updated: Mar 22, 2026
07:14
Virtual Agent for Real-Time Motivational Interviewing by Integrating Adaptive Nonverbal Behavior and Language Models
Published on: December 23, 2025
591
Pilot study of an anxiety induction method applying generative artificial intelligence and mixed reality
Daniel Badeja1, Florian Schier2, Hans-Christian Straka3
1Department of Clinical Psychology and Psychotherapy, TUD Dresden University of Technology, Germany; Immersive Experience Lab, TUD Dresden University of Technology, Germany.
Journal of Anxiety Disorders
|March 20, 2026
Summary
Artificial intelligence (AI)-generated images of worry scenarios can induce anxiety in a mixed reality setting. Personalized AI worry depictions increased subjective anxiety and negative affect in healthy adults.
Area of Science:
- Psychology
- Computer Science
- Mixed Reality
Background:
- Worry is often a verbal, abstract thought process that hinders vivid mental imagery and emotional responses, serving as an avoidance mechanism.
- Exploring novel methods to bypass worry-related avoidance is crucial for therapeutic interventions.
- Personalized, AI-generated imagery offers a potential avenue to engage individuals with their worry content.
Purpose of the Study:
- To investigate if personalized, AI-generated depictions of worry scenarios presented in a mixed reality (MR) environment can elicit anxiety.
- To assess subjective anxiety, affect, and physiological responses to AI-generated worry imagery.
- To examine the relationship between the perceived accuracy and realism of worry depictions and emotional/physiological responses.
Main Methods:
- A within-subject repeated-measures design was employed with 34 healthy adults.
- Participants viewed personalized, AI-generated images of worry scenarios and neutral images within a mixed reality head-mounted display.
- Subjective state anxiety, affect, heart rate, and skin conductance level were measured.
Main Results:
- AI-generated worry images elicited medium-sized increases in subjective anxiety ratings compared to neutral images.
- Negative affect significantly increased, while positive affect decreased following the presentation of worry images.
- Higher perceived realism and alignment of the AI-generated worry depictions correlated with greater anxiety induction.
- No significant changes were observed in physiological measures (heart rate, skin conductance).
Conclusions:
- AI-generated worry imagery presented in MR can effectively induce subjective anxiety and modulate state affect.
- The findings suggest potential applications for AI-generated imagery in therapeutic contexts for anxiety and worry.
- Future research should focus on enhancing stimulus reliability, testing clinical populations, and comparing MR with other exposure modalities.
