Exploring the Effects of User-Agent and User-Designer Similarity in Virtual Human Design to Promote Mental Health Intentions for College Students

Overview

• This study explores how computer science students co-design virtual humans (embodied conversational agents) to support mental health conversations for students similar to them.
• The researchers conducted a user study with 481 computer science students at a North American university to understand their preferences and expectations for virtual human mental health counselors.
• Key findings include the tendency for computer science students to create virtual humans that closely resemble them demographically, as well as the influence of the designer's own characteristics on the virtual human's design.
• The study provides insights into how user-designer demographic similarity can impact the effectiveness of virtual humans in mental health conversations.

Plain English Explanation

The paper examines how computer science students co-design virtual humans, or digital characters, to help support mental health conversations for other students like themselves. Virtual humans have the potential to provide mental health support, especially for students in STEM (Science, Technology, Engineering, and Mathematics) fields who may be at higher risk of anxiety and depression.

The researchers wanted to understand how computer science students would design these virtual human counselors based on their own backgrounds, experiences, and expectations. They had 481 computer science students from a major North American university participate in a user study where they co-created virtual humans to support mental health conversations.

The key finding was that the computer science students tended to design virtual humans that closely resembled them in terms of demographics like age, gender, and ethnicity. The students' own characteristics, especially their ethnicity and gender, were often reflected in the virtual humans they created. This suggests that users may be more comfortable and receptive to virtual human counselors that share their demographic characteristics.

Understanding this "similarity-attraction" effect could help tailor the design of virtual human counselors to better meet the needs and preferences of different student populations, ultimately making them more effective at providing mental health support.

Technical Explanation

The study employed a user-centered design approach, where 481 computer science students from a major North American university were invited to co-design virtual human mental health counselors for students similar to themselves. The researchers collected demographic information about the student participants and analyzed the characteristics of the virtual humans they created.

The results indicate that computer science students tend to design virtual human agents that closely match their own demographic profiles, a phenomenon referred to as "agent-designer demographic similarity." Factors like age, gender, ethnicity, and the alignment between the virtual human's appearance and voice were found to be key influences on the design process.

Furthermore, the researchers observed that the demographic characteristics of the virtual human designers, particularly their ethnicity and gender, were often reflected in the virtual humans they created. This suggests the presence of a "similarity-attraction" effect, where users may be more inclined to engage with virtual human counselors that share their own demographic traits.

The findings provide insights into how the demographic alignment between users, designers, and virtual humans can impact the effectiveness of these agents in promoting mental health conversations. Understanding these dynamics can inform the design of more personalized and relatable virtual human interventions to support the mental well-being of diverse student populations.

Critical Analysis

The study provides valuable insights into the design of virtual human counselors for college students, particularly those in STEM fields. By involving the target user group (computer science students) in the co-design process, the researchers gained a deeper understanding of the factors that influence students' preferences and perceptions of virtual human mental health support.

However, the study is limited to a single university setting and a specific student population (computer science). Further research is needed to explore the generalizability of these findings across other academic disciplines and cultural contexts. Additionally, the study does not address potential limitations or biases that may arise from the designer-agent demographic similarity, such as the risk of reinforcing stereotypes or overlooking the needs of diverse student populations.

Future studies could also investigate the long-term effectiveness of virtual human counselors in supporting student mental health, as well as the potential ethical considerations and privacy concerns associated with the use of such technology in sensitive domains like mental healthcare. Exploring the synthetic vs. human emotional responses in these virtual human interactions could also yield additional insights.

Conclusion

This study highlights the importance of understanding user preferences and characteristics when designing virtual human interventions to support college students' mental health, particularly in STEM fields. The findings suggest that computer science students tend to create virtual human counselors that closely resemble them demographically, and that the designers' own characteristics can influence the virtual humans they create.

These insights can inform the development of more personalized and relatable virtual human mental health support systems, which may be more effective in promoting open and meaningful conversations around mental wellness. By considering the similarity-attraction effects between users, designers, and virtual agents, researchers and designers can tailor these interventions to better meet the diverse needs of student populations and potentially improve mental health outcomes.