An Integrated Model of Collaborative Skill Acquisition: Anticipation, Control Tuning, and Role Adoption.

Area of Science:

• Cognitive Science
• Human-Computer Interaction
• Artificial Intelligence

Background:

• Collaborative skill acquisition is crucial for complex tasks.
• Understanding the cognitive processes underlying seamless collaboration is challenging.
• Previous models often lack integration of predictive and role-adoption mechanisms.

Purpose of the Study:

• To investigate collaborative skill acquisition in a dynamic, communication-limited environment.
• To develop and validate an integrated cognitive model of collaborative learning.
• To identify key cognitive processes involved in learning to collaborate.

Main Methods:

• Utilized the game Co-op Space Fortress for a dynamic task setting.
• Developed a cognitive model within the ACT-R architecture.
• Implemented reinforcement learning for role adoption and motion extrapolation for prediction.
• Incorporated a novel supervised learning procedure within ACT-R.

Main Results:

• Observed subjects developing consistent, uncommunicated roles during skill acquisition.
• Demonstrated subjects' ability to anticipate future task states.
• The ACT-R model successfully reproduced observed subject skill acquisition trajectories.
• The model integrated role adoption and predictive processes effectively.

Conclusions:

• The developed ACT-R model is the first integrated cognitive model of collaborative skill acquisition.
• The findings provide insights into the interplay of role adoption, prediction, and learning in collaboration.
• This research advances our understanding of the cognitive underpinnings of effective teamwork.