Dancing with the Avatars: Minimal Avatar Customisation Enhances Learning in a Psychomotor Task

Title of the Paper

Dancing with the Avatars: Minimal Avatar Customisation Enhances Learning in a Psychomotor Task

Paper Information

• Subject Area: Intersection of virtual reality and motor learning, focusing on the impact of avatar customization in virtual environments.
• Keywords: Avatar Customisation, Skills Training, Psychomotor, Virtual Environments, Virtual Reality

Research Background and Issues

• Identified Problems and Challenges:

  • Using avatars that resemble the learner's appearance in virtual environments may enhance learning outcomes, but this has not been fully explored in psychomotor tasks (e.g., dance training).
  • Generating highly realistic avatars is costly, raises privacy concerns, and may lead to the "uncanny valley" effect, negatively affecting user experience.
  • It remains unclear how minimal similarity in avatars affects learning outcomes, and there is a lack of comparative studies between screen-based and virtual reality (VR) environments.

• Significance of the Research:

  • Observational learning through avatars, especially in virtual reality, has significant potential for applications in sports, clinical rehabilitation, and educational training, with a notable impact on psychomotor skill acquisition.
  • As virtual reality and virtual environments become increasingly prevalent in education and training, optimizing avatar design can enhance the overall learning experience.

• Motivation and Related Work:

  • The "mirror neuron" theory in observational learning suggests that similarity between the learner and the avatar may activate mirror neurons more effectively, thereby improving learning.
  • Previous studies have primarily focused on single motor tasks (e.g., squats) or the impact of highly customized avatars in other behavioral domains, with limited exploration of "minimal customization" avatars in complex psychomotor tasks.

Solution

• Research Methodology:

  • Proposed a "minimal avatar customization" approach, creating avatars by matching only three basic appearance features of the learner: gender, skin tone, and hair color. These avatars are quick and cost-effective to generate, avoiding issues associated with highly realistic avatars.
  • Compared the effects of visually similar (matched-feature, MF) and dissimilar (D) avatars in both screen-based and VR learning environments.

• Innovative Contributions:

  • Introduced and validated the "minimal avatar customization" method, reducing avatar generation complexity while protecting privacy and avoiding the uncanny valley effect.
  • Systematically investigated the differences in avatar effects between screen-based and VR environments, focusing on the impact of avatar appearance similarity on psychomotor task learning.

• Implementation Steps and Techniques:

  • Designed a mixed-factor experiment: screen-based or VR as between-group variables, and avatar type (MF and D) as within-group variables.
  • The task involved learning two 3-step hip-hop dance sequences by observing and imitating the avatar's movements.
  • MF avatars were generated by allowing participants to select gender, skin tone, and hair color, while D avatars used contrasting features.
  • Outcomes were assessed using a newly developed "Visual Dance Movement Imagery Questionnaire" (VDMIQ), along with scales measuring intrinsic motivation, avatar preference, and virtual presence.

Research Findings

• Specific Results:

  1. Compared to dissimilar avatars, minimally customized avatars significantly improved participants' clarity of dance movement imagery, including internal visual imagery, external visual imagery, and kinesthetic imagery.
  2. Minimal avatar customization was effective in both screen-based and VR environments, with more pronounced effects in VR, particularly in kinesthetic imagery.
  3. Individual differences (e.g., age, gender, perceptual ability, virtual presence) significantly influenced learning outcomes, with low-skill learners and those perceiving greater similarity with the avatar benefiting the most.

• Advantages:

  • Provided a low-cost, efficient avatar design method that is easier to implement and more user-friendly compared to generating highly realistic avatars.
  • This method balances privacy protection, user experience, and learning effectiveness, making it suitable for large-scale adoption.

• Experimental and Evaluation Results:

  • Conducted linear regression analysis on the correlation between avatar similarity (gender, hair color, skin tone) and participants' learning performance, revealing a positive relationship between perceived avatar similarity and learning outcomes.
  • For VR-based learning performance, participants' sense of virtual presence significantly influenced learning outcomes.

• Limitations and Future Directions:

  • The study only examined a few basic appearance features (gender, hair color, skin tone) and did not include factors such as age or body type, requiring further research on more comprehensive avatar characteristics.
  • Future studies should explore how to adapt avatar features in cross-cultural contexts and investigate the long-term effects of avatar design on learning duration.
  • It remains unclear whether the relationship between avatar similarity and learning outcomes is linear, and at what level of similarity negative effects (e.g., uncanny valley) might emerge.

Conclusion

This study demonstrates the potential of minimal avatar customization in psychomotor tasks (e.g., dance learning), with particularly significant effects in virtual reality environments. The research not only provides practical recommendations for avatar design but also expands the understanding of behavioral and cognitive interactions in virtual environments.