Modelling human navigation and decision dynamics in a first-person herding task

Area of Science:

• Cognitive Science
• Robotics
• Human-Computer Interaction

Background:

• Understanding human navigation and decision-making is crucial for developing intuitive human-machine interfaces.
• Dynamical perceptual-motor primitives (DPMPs) offer a framework for modeling complex motor behaviors.

Purpose of the Study:

• To investigate the applicability of DPMPs for capturing human navigation and target selection in a first-person herding task.
• To assess the predictive power of a combined DPMP and heuristic policy model for human behavior.

Main Methods:

• Human participants engaged in a first-person herding game, corralling virtual targets.
• Movement trajectories and target-selection decisions were recorded and modeled using DPMPs and a heuristic policy.

Main Results:

• A DPMP navigation model accurately reproduced participants' movement trajectories.
• A heuristic policy captured approximately 80% of target-selection decisions.
• The combined model successfully simulated and predicted behavioral dynamics in novel contexts.

Conclusions:

• DPMPs provide a robust framework for modeling complex human perceptual-motor behaviors, including navigation and decision-making.
• Integrating DPMPs with heuristic policies enhances the simulation and prediction of human behavior.
• Findings have implications for designing advanced artificial agents for human-machine interaction.