Does the Uncanny Valley Effect Exist When Voice Assistants Have Minds and Human-Like Appearances? The Roles of Humanness, Eeriness, Attractiveness, and Emotions

This study investigates whether the uncanny valley effect exists when voice assistants (VAs) have both human-like appearances and perceived minds and examines the roles played by humanness, eeriness, attractiveness, and emotional responses in this context. We draw on mind perception theory to decompose mind perception into perceived agency (autonomous capabilities) and perceived experience (emotional/cognitive capacities), which is paired with a 2 (perceived agency) × 2 (perceived experience) × 2 (appearance: human-like vs. machine-like) experimental design involving 34 participants. The results reveal that perceived experience significantly influences perceptions of humanness, eeriness, and attractiveness, whereas perceived agency has no main effect. Human-like appearances enhance humanness and attractiveness but also amplify eeriness, aligning with classic predictions associated with the uncanny valley effect. Mediation analysis reveals that humanness positively predicts both eeriness and attractiveness, and attractiveness drives positive emotional responses (pleasure and arousal), whereas eeriness has no direct effect on emotions. This research extends the uncanny valley framework to encompass mind perception attributes. The findings highlight the need for balanced anthropomorphism in VA design, suggesting caution in endowing VAs with human-like emotional experiences to mitigate eeriness while leveraging attractiveness to enhance user engagement. These insights contribute to human–computer interaction theory and offer practical guidelines for the development of user-centered intelligent systems.