Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Review
. 2001 Dec;11(6):655-62.
doi: 10.1016/s0959-4388(01)00265-3.

Computational Approaches to Motor Control

Affiliations
Free PMC article
Review

Computational Approaches to Motor Control

T Flash et al. Curr Opin Neurobiol. .
Free PMC article

Abstract

New concepts and computational models that integrate behavioral and neurophysiological observations have addressed several of the most fundamental long-standing problems in motor control. These problems include the selection of particular trajectories among the large number of possibilities, the solution of inverse kinematics and dynamics problems, motor adaptation and the learning of sequential behaviors.

Figures

Figure 1
Figure 1
Hand trajectories and predictions of neural models during figure drawing tasks. (a,b) The hand paths and velocity profiles predicted by the minimum jerk model [30]. This model suggests that among all possible hand trajectories, the selected movements maximize motion smoothness, defined here as the trajectory that minimizes the rate of change of hand acceleration (jerk). The hand paths are shown for two patterns: a figure of eight – upper left in (a) – and a double limaçon – upper left in (b). The predicted velocity profiles for these two figural forms (red solid curves, bottom panels) closely matched those of the recorded movements (dashed black curves, bottom panels). Also illustrated (upper right panels) is the piecewise constant relationship between hand velocity (V) and radius of curvature (R), when plotted in logarithmic scales. (c) Neural and finger trajectories during a figure of eight drawing task. A time series of population vectors calculated during the task was temporally integrated to yield the neural trajectory. Individual movement segments are marked by different colors. Segment boundaries correspond to maxima of tangential velocity. Also shown is the two-thirds power law representation in the actual hand kinematics (finger) and those predicted by population vectors (neural). (d) Magnitude of the prediction interval increases as the path becomes more curved (the radius of curvature decreases). Panels (c) and (d) reproduced with permission from [15].
Figure 2
Figure 2
Schematic diagram that illustrates three hierarchical levels for planning multi-joint arm movements. First, a trajectory is planned in hand coordinates; second, the hand trajectory is transformed into joint trajectories by solving the inverse kinematics problem; finally, the joint torques are found by computing the inverse dynamics [6]. An alternative to inverse dynamics is based on equilibrium point control. After the trajectory is planned, the controller, shown on the bottom, maps intentions to motor commands. A forward model (possibly of both the controlled limb and of the environment) maps the motor commands to predicted sensations. Both the controller and the forward model might be composed of neural systems with modular architectures. The schematic hand controller in the upper right panel illustrates the forward model as a map that predicts joint proprioception and movement from the controller output.

Similar articles

See all similar articles

Cited by 19 articles

See all "Cited by" articles

Publication types

LinkOut - more resources

Feedback