Overview

This paper explores parasomatic interaction—incorporating synthetic physiological signals into human-AI interaction. We found simulated heartbeat sensations significantly bias human perception of AI agents: elevated heartbeats made participants seven times more likely to distrust an AI, while calm heartbeats increased AI selection as collaborators by 60%. Notably, emotionally stable individuals were most vulnerable to these effects. These findings reveal both opportunities for enhancing human-AI interaction and risks for manipulation, suggesting urgent need for design considerations in somatically embodied AI systems.

Contributions

Led project from conception to publication. Designed experimental protocols and secured IRB approval. Conducted data collection with 39 participants and performed all statistical analyses. Contributed to hardware development and 3D modeling. Wrote substantial portion of paper.

Overview

Contributions

Timeline

Paper

Password Required

Password Required