Abstract
When walking to intercept a moving target, occluding the target severely impairs performance, consistent with on-line control by the constant bearing strategy (Zhao & Warren, 2014). Alternatively, interception might be controlled by an internal model of the target’s motion, which is updated by current information (Diaz, et al., 2013). Here we ask whether an internal model of target motion can be learned over multiple training trials, and then used to guide interception during occlusion. Participants (N=10) walked in a virtual environment (12m x 12m), and displays were presented stereoscopically in a head mounted display (63°H x 53°V, 60 Hz), while head position was tracked (60 Hz). The target initially moved at 0.6 or 0.8m/s, and after 3s changed speed by -0.3, -0.2, 0.0, +0.2 or +0.3m/s. Each participant performed five experimental sessions, with one speed change per session. A session consisted of a learning block of 40 trials in which the target was always visible, followed by a test block of 24 trials in which the target was occluded 2.5s after it started moving. Occluding the target significantly impaired interception accuracy and precision (p< 0.01). An analysis of heading adjustments prior to the occlusion point indicates that participants learned to anticipate the speed change: they turned more in sessions when the target would speed up and turned less in sessions when the target would slow down. However, heading adjustments after the occlusion point indicate that participants did not adaptively adjust their heading during occlusion, but made a stereotyped turn. The results suggest that an internal model of the target’s motion is not learned or used to guide interception in the absence of visual information, rather, interception is normally controlled on-line. The target’s motion can be anticipated, but only when it is visible, possibly by modifying the constant bearing strategy.
Meeting abstract presented at VSS 2015