| Popis: |
One view of motor planning and control posits is that the CNS may attempt to maximize movement smoothness by minimizing jerk-cost. This notion is only partially supported by studies of kinematically determinate single- and/or planar two-joint movements. To see if multi-joint (i.e., kinematically indeterminate) movements are produced more smoothly following learning, we examined changes in end-effector path smoothness using root-mean-square (RMS)-jerk (and its magnitudinal and directional components), and the smoothness of joint-angle trajectories using RMS-jerk, peak-jerk, RMS-snap and peak-snap. Subjects attempted to produce 400 ms kicking movements with temporal accuracy used to measure performance. End-effector RMS-jerk increased as the speed of movement production increased but did not decrease with further learning. Also, trials late in learning often had much higher jerk values than early trials of similar movement time. As joint angular jerk and snap measures increased as movement times decreased, and increased more with further learning, joint trajectories were less smooth following learning. We empirically demonstrate that movements with similar paths can have different jerk values; it is therefore doubtful that the control system (i.e., CNS) attempted to maximize movement smoothness. We re-examine the arguments for minimization of jerk as a control parameter including their applicability to multi-joint movements. |
Full Text from ScienceDirect