Introduction
Social robots are increasingly deployed to motivate physical activity, manage weight, and encourage healthier eating habits. While their physical presence and multimodal communication capabilities make them effective "persuaders," the underlying mechanisms of this influence remain poorly understood. Most research treats robots as sophisticated tools, yet humans intuitively respond to them as social actors—a phenomenon known as the "Computers Are Social Actors" (CASA) paradigm. This position paper argues that to understand health persuasion in human-robot interaction (HRI), we must look beyond the "nudge" itself and examine the interpersonal relationship formed between the human and the machine.
Theoretical Foundations
The authors propose a new framework by synthesizing three established psychological theories:
- The Mediational Model of Social Control: This model distinguishes between positive strategies (e.g., logic, caring) that lead to behavior change via positive affect, and negative strategies (e.g., criticism, threats) that often backfire due to psychological reactance.
- Interdependence Theory: This theory explores how individuals in close relationships undergo a "transformation of motivation," prioritizing the needs of the partner or the relationship over immediate self-interest.
- The CASA Paradigm: This provides the basis for why humans form socioemotional bonds with "relational artifacts" like robots, treating them as social entities rather than mere objects.
A Relational Model of Health Persuasion
The core contribution of this work is a conceptual model that maps how robot strategies and relationship quality jointly influence behavioral outcomes. Unlike traditional persuasive technology, this model accounts for the "dual effects" of social control.
As shown in the model, the process follows two primary paths:
- The Positive Path: Positive social control from the robot generates positive affect in the user, leading to intended health behavior change.
- The Negative Path: Negative social control (e.g., nagging or criticism) can trigger negative affect and psychological reactance, where the user intentionally performs unhealthy behaviors to reassert autonomy.
Crucially, the Human-Robot Relationship acts as a moderator. A strong bond can amplify the success of positive encouragement and may even serve as a buffer, shielding the user from the negative impact of accidental or poorly timed feedback from the robot. In a strong relationship, a user might comply with a suggestion not just for personal health, but because they perceive the behavior as meaningful to their "relationship" with the robot.
Implications and Future Directions
This perspective shifts the focus of HRI design from the frequency or accuracy of messages to the quality of the relationship. For a robot to be an effective health coach, designers should prioritize relationship-oriented activities, such as self-disclosure and rapport building.
However, several challenges remain for future research:
- Empirical Validation: The model is currently theoretical and requires rigorous experimental testing to confirm if "transformation of motivation" truly occurs with non-biological entities.
- The "Dark Side": More research is needed to understand the boundaries between helpful nudges and perceived threats to freedom, especially when a robot unintentionally offends.
- Moral and Long-term Dynamics: The field must grapple with the ethics of unilateral relationships (where only the human is emotionally invested) and how these interactions evolve over months or years.
By viewing robots as "relational artifacts" rather than mere tools, we can better predict and design for the complex social dynamics that drive long-term health behavior change.
Summary updated