Abstract
Visually guided control in manual lateral interception tasks has been demonstrated to be prospective rather than predictive in nature, notably through the repeated finding of systematically different (i.e., target trajectory dependent) hand movement kinematics for uniformly moving targets arriving at the same interception position after the same flight duration while coming from different starting positions. This phenomenon, known as the angle-of-approach effect (Ledouit et al., 2013), suggests that both ball position and ball velocity play a role in guiding the hand towards the future interception location. As existing linear dynamics-based models could not be adapted to adequately capture the empirically observed hand kinematics, we set out to develop an information-driven model incorporating nonlinear dynamics. Combining Duffing stiffness with both Rayleigh and Van der Pol energy-regulating functions allowed generation of discrete movements with the required kinematic characteristics. Visual guidance was incorporated by scaling the model parameters to (i) the (perceptually specified) time-to-contact of the target with the interception axis and (ii) the (perceptually specified) current interception location aimed for. The latter, based on a combination of current ball position and velocity, evolves over time and thereby allows the emergence of the angle-of-approach effect. The model was demonstrated to adequately capture a dataset of empirically observed hand kinematics incorporating 28 different target trajectories under four different ball velocity (and hence time pressure) conditions. We conclude that information-movement coupling in the form of a principled information-driven nonlinear dynamics-based model with as few as six coefficients can adequately capture the richness of behavior observed in manual lateral interception tasks.