Intercepting moving targets on foot: Can people learn to anticipate multiple trajectories?

Abstract

Previous work (Fajen & Warren, 2004; submitted) modeled human interception of moving targets based on first-order information about the target's bearing (constant bearing strategy). The model has accurately predicted paths to constant velocity and accelerating targets and to targets moving on a curved trajectory. Owens & Warren (VSS 2005) found that with repeated presentation of the same trajectory, most participants learn heuristic strategies to anticipate target motion and take “shortcuts” to accelerating and curving targets, although there was considerable variation. Here we investigate the number and variety of target trajectories that participants can learn to anticipate by varying trajectory characteristics within blocks of trials. Participants walked in the VENLab, a 12m x 12m virtual environment with a head-mounted display (60 deg H x 40 deg V) and a sonic/inertial tracking system (latency 50–70 ms). One block of 20 trials consisted of two randomized target trajectories that curved in the same direction, but had different radii and speeds. Another block of 20 trials contained two mirror-image trajectories with identical radii and speed. Presentation order of these blocks was counterbalanced. The final block of 40 trials presented four novel trajectories with different radii and speeds. Most participants developed shortcut strategies in all three conditions, although more learned them for the same-side targets than for the mirror-image targets. The trajectories in the last block of trials were the most quickly and consistently learned. A second experiment investigates how the addition of a color cue to target trajectory affects the acquisition of anticipatory strategies.