Reversals in Movement Direction in Locomotor Interception of Uniformly Moving Targets

Abstract

Here we studied how participants steer to intercept uniformly moving targets in a virtual driving task. We tested the hypothesis that locomotor interception behavior cannot fully be explained by a strategy of nulling rate of change in pertinent agent-target relations such as the target-heading angle or target’s bearing angle. In line with a previously reported observation and model simulations, we found that, under specific combinations of initial target eccentricity and target motion direction, locomotor paths revealed reversals in movement direction. This phenomenon is not compatible with unique reliance on first-order (i.e., rate-of-change based) information in the case of uniformly moving targets. We also found that, as expected, such reversals in movement direction were not observed consistently over all trials of the same experimental condition: their presence depended on the timing of the first steering action effected by the participant, with only early steering actions leading to reversals in movement direction. These particular characteristics of the direction-reversal phenomenon demonstrated here for a locomotor interception-by-steering task correspond to those reported for lateral manual interception. Together, these findings suggest that control strategies operating in manual and locomotor interception may at least share certain characteristics.

Steering to intercept targets following curving trajectories: Insights into visual guidance from Qualitative Inconsistency Detection

“…In line with earlier studies on human interception-by-steering 5,15,16,23,24,26 , a b-based zeroth-order (ZTHA) strategy of target pursuit was clearly not observed. The present contribution also allowed ruling out the b-based first-order (CTHA) strategy of reliance on db/dt information.…”

“…The finding that such prolonged periods of a strong db/dt-induced drive to change direction did not lead to compensatory steering behaviour rules out db/dt-nulling as a feasible option for the control of steering under the present circular target trajectory conditions. We note that under linear (uniform motion) target trajectory conditions db/dt-nulling has also been ruled out as a feasible option for the control of steering, as it cannot explain the sign-switch observed during the initial steering action for targets moving outward from an initial position straight-ahead of the participant [22][23][24] . Under such conditions the (linearly) outward-moving target immediately creates an opening of the b angle, with the agent thus lagging the target.…”

“…Focussing on individual trials rather than ensemble averages has started to reveal a far more subtle picture. For instance, interception characteristics related to early systematic influences of zeroth-order information were reported for (near) constant speed targets moving along straight trajectories 4,23,24 , while later systematic influences of second-order information were reported for targets moving along curved trajectories 4 . It thus appears that particular experimental conditions may give rise to richer interception behaviour than reasoning from categorical differences between interception strategies would lead one to expect.…”

“…In short, the QuID method evaluates reliance on plausible informational variables to drive behaviour by examining, for each informational variable, whether it systematically fulfils two essential requirements: an observable change in direction of the action variable should be proceeded by an appropriate information-based drive some time before, and any substantial information-based drive should result in an observable change in direction of action variable a. It is important to note that, under the assumption that the same control law applies over a wide range of conditions, such qualitative inconsistencies may surface only under specific conditions [22][23][24] .…”

“…In the present contribution we therefore applied the QuID method to assess the potential role of the zeroth-order, firstorder and second-order time derivatives of target-heading angle b and target's bearing angle q over the course of each individual interception action in a simulated interception-by-steering task 23,24 with targets moving at constant speed along curved (here circular) trajectories. We hypothesized that control of steering would be q-based rather than b-based 5,[23][24][25] and that curving target trajectories would elicit a shift from initial reliance on first-order (d 1 q/dt 1 ) information to a combination of first-and second-order (d 2 q/dt 2 ) information 4 . As graphically illustrated in Figure 2, participants moving at 20 m/s were to drive through cylindrical targets that could appear at five different initial lateral positions in the environment.…”