How Can a Social Companion Robot Support Enjoyable but Moderate Solitary Drinking?
This blog post summarizes the paper “Enjoy, but Moderately!: Designing a Social Companion Robot for Social Engagement and Behavior Moderation in Solitary Drinking Context”. This paper will be presented at the 26th ACM Conference on Computer-Supported Cooperative Work and Social Computing, a top venue for social computing scholarship.
Social robots for social engagement and behavioral moderation
Social robots have served as social companions to mitigate people’s loneliness or facilitate behavior change. Depending on its design goals, the social robot was designed to either engage or moderate certain behaviors. For instance, it can help users socially engage in healthy habits such as exercising more or consuming nutritious food and drink. However, the design of a social companion that both promotes and moderates certain behaviors has not been well explored. One unexplored area of social robots providing only social engagement is that it needs to carefully consider the negative consequences when the user is over-engaged in the target behavior. For example, a jogging companion can engage users in enjoyable physical activities, but if a user exercises excessively following the companion’s advice or has an excessive desire to exercise, they can suffer from fatigue or injury.
What did we build
We chose a solitary drinking context that requires both social engagement and behavioral moderation and built a social companion robot that provides social engagement and behavioral moderation together. The robot provides three main interactions people usually engage in the social drinking context: (1) ‘cheers’ interaction, the clinking of glasses between user and robot, (2) robot gestures of shaking its body for drinking moderation or moving back and forth to suggest ‘cheers’ interaction, and (3) voice feedback. These three interactions were used to guide users to maintain a moderate drinking pace.
In particular, we considered the time interval of any two consecutive ‘cheers’ interactions. Each ‘cheers’ interaction was used as a point at which the user wished to have a drink with the robot, and the robot accepted or refused to clink based on the pre-determined time interval. There are three cases depending on when a user attempts a ‘cheers’ interaction. The robot can moderate interaction with a warning message and a short body shake as a gesture (like shaking heads to say ‘no’) if a user tries again within 2.5 minutes of the last interaction (c). If the user suggests clinking glasses between 2.5 and 5 minutes, the robot will accept, provide a gesture of moving toward the user, and perform ‘cheers’ with a simple reaction (b). If the robot waits more than 5 minutes from the last ‘cheers’ since the user does not attempt any interaction, the robot initiates ‘cheers’ (a).
What did we learn
We conducted a preliminary user study with 20 participants to understand (1) users’ perceptions of the robot and (2) user experiences of engaging and moderating interventions. Following are the major findings we learned from the user study.
- The physical and verbal interaction made the participants feel as if they were drinking with their “friends” and helped them enjoy the time even when alone. Beyond the current design, they expected functions such as diverse conversation, realistic physical interactions, and gestures that resembled the context of drinking with real friends.
- Participants mostly followed the social engagement and behavioral moderation by the robot. They perceived the social engagement interventions as analogous to their friends’ behavior and reacted naturally. Conversely, the robot’s behavior-moderating interventions helped them not to drink too fast by making them reflect on their drinking pace and by persuading them as a trustworthy social companion.
- Providing both social engagement and moderation was perceived as human-like features e.g., the robot did not keep repeating the same intervention but showed different ones according to situation. This could alleviate the reluctance toward the robot and lead to it being perceived as a smart and caring friend.
Toward enjoyable but moderate behavior for health and well-being
Based on our user study result, we suggest possible design implications on how designers can leverage social robots to support enjoyable but moderate behavior.
- Physical-social interaction for persuasive engagement design: The major physical interaction (‘cheers’) provided by the robot played an important role in persuasion by building social rapport with users. Our findings show the importance of physical interaction in building social rapport and further contribute to the field of persuasive social robots.
- Social engagement and behavioral moderation for computational commensality: Drinking is typical of commensality (e.g., the practice of eating or drinking together). Our work belongs to existing HCI scholarship that aims to design technologies for supporting eating and drinking experiences, which is known as computational commensality. Our robot’s persuasive human-like aspects (e.g., providing engagement and moderation together) can be applied to diverse commensality scenarios, by utilizing socially-appropriate interaction rituals.
To learn more about our work, you can read the paper here, and feel free to contact our lab in person. We extend a warm welcome to anyone interested in our work and look forward to engaging in discussions with you soon.
