Claude Ghez

Independent learning of internal models for kinematic and dynamic control of reaching.

Psychophysical studies of reaching movements suggest that hand kinematics are learned from errors in extent and direction in an extrinsic coordinate system, whereas dynamics are learned from proprioceptive errors in an intrinsic coordinate system. We examined consolidation and interference to determine if these two forms of learning were independent. Learning and consolidation of two novel transformations, a rotated spatial reference frame and altered intersegmental dynamics, did not interfere with each other and consolidated in parallel. Thus separate kinematic and dynamic models were constructed simultaneously based on errors computed in different coordinate frames, and possibly, in different sensory modalities, using separate working-memory systems. These results suggest that computational approaches to motor learning should include two separate performance errors rather than one.

Accuracy of planar reaching movements

This study examined the variability in movement end points in a task in which human subjects reached to targets in different locations on a horizontal surface. The primary purpose was to determine whether patterns in the variable errors would reveal the nature and origin of the coordinate system in which the movements were planned. Six subjects moved a hand-held cursor on a digitizing tablet. Target and cursor positions were displayed on a computer screen, and vision of the hand and arm was blocked. The screen cursor was blanked during movement to prevent visual corrections. The paths of the movements were straight and thus directions were largely specified at the onset of movement. The velocity profiles were bell-shaped, and peak velocities and accelerations were scaled to target distance, implying that movement extent was also programmed in advance of the movement. The spatial distributions of movement end points were elliptical in shape. The major axes of these ellipses were systematically oriented in the direction of hand movement with respect to its initial position. This was true for both fast and slow movements, as well as for pointing movements involving rotations of the wrist joint. Using principal components analysis to compute the axes of these ellipses, we found that the eccentricity of the elliptical dispersions was uniformly greater for small than for large movements: variability along the axis of movement, representing extent variability, increased markedly but nonlinearly with distance. Variability perpendicular to the direction of movement, which results from directional errors, was generally smaller than extent variability, but it increased in proportion to the extent of the movement. Therefore, directional variability, in angular terms, was constant and independent of distance. Because the patterns of variability were similar for both slow and fast movements, as well as for movements involving different joints, we conclude that they result largely from errors in the planning process. We also argue that they cannot be simply explained as consequences of the inertial properties of the limb. Rather they provide evidence for an organizing mechanism that moves the limb along a straight path. We further conclude that reaching movements are planned in a hand-centered coordinate system, with direction and extent of hand movement as the planned parameters. Since the factors which influence directional variability are independent of those that influence extent errors, we propose that these two variables can be separately specified by the brain.

Trajectory control in targeted force impulses

SummaryIn the preceding study (Gordon and Ghez 1987), we showed that accurately targeted isometric force impulses produced by human subjects are governed by a pulse height control policy. Different peak forces were achieved by modulating the rate of rise of force while force rise time was maintained close to a constant value and independent of peak force. An early measure of the rate of rise of force, peak d2F/dt2, was scaled to the required force (target amplitude) and highly predictive of the peak force achieved. In six subjects examined, peak d2F/dt2 accounted for between 70% and 96% of the total variance in peak force. In the present study, we further examined these targeted responses to determine whether the residual variability not predicted by peak d2F/dt2 could be accounted for by adjustments to the force trajectories which compensated for initial errors in the scaling of the d2F/dt2. A statistical model of the determinants of peak force was tested. This model included two paths by which the target amplitude could independently influence the peak force achieved. The first path was preprogrammed pulse height control. In this path, target amplitude determined the initial rate of rise of force (peak d2F/dt2) which in turn determined the final peak force achieved. The second path was an independent influence of errors in the initial scaling of peak d2F/dt2 on peak force. Multiple regression analysis was performed on trajectory variables within the sets of responses by each subject in each condition to determine whether the second path contributed significantly to explaining the variance in peak force. In each subject and condition, there was a significant independent influence of error in d2F/dt2 on peak force, and the direction of this effect was to decrease the magnitudes of peak force errors. These compensatory adjustments accounted for between 1% and 14% of the total variance in peak force. Further multiple regression analyses revealed that inappropriate scaling of the initial phase of the trajectories was compensated for by shortening or prolonging the force rise time. These trajectory adjustments were in turn implemented by modulation of the timing and magnitude of the contractions in the agonist and antagonist muscles that produced the force trajectories. Because these compensatory adjustments were evident in the EMG pattern at latencies too short to be accounted for by peripheral feedback, we assume that they depend on internal monitoring of the unfolding neural commands. These internal feedback processes act in parallel with the programmed commands, both determining the force trajectory.

Accuracy of planar reaching movements - II. Systematic extent errors resulting from inertial anisotropy

This study examines the source of directiondependent errors in movement extent made by human subjects in a reaching task. As in the preceding study, subjects were to move a cursor on a digitizing tablet to targets displayed on a computer monitor. Movements were made without concurrent visual feedback of cursor position, but movement paths were displayed on the monitor after the completion of each movement. We first examined horizontal hand movements made at waist level with the upper arm in a vertical orientation. Targets were located at five distances and two directions (30° and 150°) from one of two initial positions. Trajectory shapes were stereotyped, and movements to more distant targets had larger accelerations and velocities. Comparison of movements in the two directions showed that in the 30° direction responses were hypermetric, accelerations and velocities were larger, and movement times were shorter. Since movements in the 30° direction required less motion of the upper arm than movements in the 150° direction, we hypothesized that the differences in accuracy and acceleration reflected a failure to take into account the difference in total limb inertia in the two directions. To test this hypothesis we simulated the initial accelerations of a two-segment limb moving in the horizontal plane with the hand at shoulder level when a constant force was applied at the hand in each of 24 directions. We compared these simulated accelerations to ones produced by our subjects with their arms in the same position when they aimed movements to targets in the 24 directions and at equal distances from an initial position. The magnitudes of both simulated and actual accelerations were greatest in the two directions perpendicular to the forearm, where inertial resistance is least, and lowest for movements directed along the axis of the forearm. In all subjects, the directional variation in peak acceleration was similar to that predicted by the model and shifted in the same way when the initial position of the hand was displaced. The pattern of direction-dependent variations in initial acceleration did not depend on the speed of movement. It was also unchanged when subjects aimed their movements toward targets presented within the workspace on the tablet instead of on the computer monitor. These findings indicate that, in programming the magnitude of the initial force that will accelerate the hand, subjects do not fully compensate for direction-dependent differences in inertial resistance. The direction-dependent differences in peak acceleration were associated with systematic variations in movement extent in all subjects, but the variations in extent were proportionately smaller than those in acceleration. This compensation for inertial anisotropy, which differed in degree among subjects, was associated with changes in movement duration. The possible contributions of elastic properties of the musculoskeletal system and proprioceptive feed-back to the compensatory variations in movement time are discussed. The finding that the magnitude of the initial force that accelerates the hand is planned without regard to movement direction adds support for the hypothesis that extent and direction of an intended movement are planned independently. Furthermore, the lack of compensation for inertia in the acceleration of the simple reaching movements studied here suggests that they are planned by the central nervous system without explicit inverse kinematic and dynamic computations.

Adaptation to Visuomotor Transformations: Consolidation, Interference, and Forgetting

The paradigm task A→task B→task A, which varies the time interval between task A and task B, has been used extensively to investigate the consolidation of motor memory. Consolidation is defined as resistance to retrograde interference (interference by task B on initial learning of task A). Consolidation has been demonstrated for simple skills, motor sequencing, and learning of force fields. In contrast, evidence to date suggests that visuomotor learning does not consolidate. We have shown previously that adaptation to a 30° screen-cursor rotation is faster and more complete on relearning 24 hr later. This improvement is prevented if a 30° counter-rotation is learned 5 min after the original rotation. Here, we sought to identify conditions under which rotation learning becomes resistant to interference by a counter-rotation. In experiment 1, we found that interference persists even when the counter-rotation is learned 24 hr after the initial rotation. In experiment 2, we removed potential anterograde interference (interference by task B on relearning of task A) by introducing washout blocks before all of the learning blocks. In contrast to experiment 1, we found resistance to interference (i.e., consolidation) when the counter-rotation was learned after 24 hr but not after 5 min. In experiment 3, we doubled the amount of initial rotation learning and found resistance to interference even after 5 min. Our results suggest that persistent interference is attributable to anterograde effects on memory retrieval. When anterograde effects are removed, rotation learning consolidates both over time and with increased initial training.

Patterns of regional brain activation associated with different forms of motor learning

To examine the variations in regional cerebral blood flow during execution and learning of reaching movements, we employed a family of kinematically and dynamically controlled motor tasks in which cognitive, mnemonic and executive features of performance were differentiated and characterized quantitatively. During 15O-labeled water positron emission tomography (PET) scans, twelve right-handed subjects moved their dominant hand on a digitizing tablet from a central location to equidistant targets displayed with a cursor on a computer screen in synchrony with a tone. In the preceding week, all subjects practiced three motor tasks: 1) movements to a predictable sequence of targets; 2) learning of new visuomotor transformations in which screen cursor motion was rotated by 30 degrees -60 degrees; 3) learning new target sequences by trial and error, by using previously acquired routines in a task placing heavy load on spatial working memory. The control condition was observing screen and audio displays. Subtraction images were analyzed with Statistical Parametric Mapping to identify significant brain activation foci. Execution of predictable sequences was characterized by a modest decrease in movement time and spatial error. The underlying pattern of activation involved primary motor and sensory areas, cerebellum, basal ganglia. Adaptation to a rotated reference frame, a form of procedural learning, was associated with decrease in the imposed directional bias. This task was associated with activation in the right posterior parietal cortex. New sequences were learned explicitly. Significant activation was found in dorsolateral prefrontal and anterior cingulate cortices. In this study, we have introduced a series of flexible motor tasks with similar kinematic characteristics and different spatial attributes. These tasks can be used to assess specific aspects of motor learning with imaging in health and disease.

Pharmacological inactivation in the analysis of the central control of movement.

In this review, we describe how pharmacological inactivation can be used to elucidate the central control of skilled limb movement. Local anesthetics and tetrodotoxin block neuronal cell bodies and passing fibers while gamma-aminobutyric acid (GABA) and muscimol only block cell bodies. Blockade induction time is short (several minutes) for all the agents. Blockade duration produced by local anesthetics and GABA is 15-60 min, while that of tetrodotoxin and muscimol is up to several days. We describe our drug injection system, with an integrated microelectrode and a viewing port for visually monitoring drug flow into the injection cannula. We used glucose metabolism to assess the extent of inactivation. Intracortical lidocaine or muscimol injection produces a central core of maximal hypometabolism (1 mm radius), which could be due to drug spread, surrounded by an extensive region (several millimeters) of reduced hypometabolism, possibly due to reduced synaptic activity of neurons receiving projections from the core region. Drug injection only depresses neuronal activity, which contrasts with cooling, where there can be neuronal hyperexcitability at the periphery of the inactivation site. Our experiments in behaving animals show how pharmacological inactivation is an effective analytical tool for dissecting the differential functional contributions of subcortical and cortical forelimb representations to limb movement control.

Discrete and continuous planning of hand movements and isometric force trajectories

Abstract We have previously demonstrated that, in preparing themselves to aim voluntary impulses of isometric elbow force to unpredictable targets, subjects selected default values for amplitude and direction according the range of targets that they expected. Once a specific target appeared, subjects specified amplitude and direction through parallel processes. Amplitude was specified continuously from an average or central default; direction was specified stochastically from one of the target directions. Using the same timed response paradigm, we now report three experiments to examine how the time available for processing target information influences trajectory characteristics in two-degree-of-freedom forces and multijoint movements. We first sought to determine whether the specification of force direction could also take the form of a discrete stochastic process in pulses of wrist muscle force, where direction can vary continuously. With four equiprobable targets (two force amplitudes in each of two directions separated by 22° or 90°), amplitude was specified from a central default value for both narrow and wide target separations as a continuous variable. Direction, however, remained specified as a discrete variable for wide target separations. For narrow target separations, the directional distribution of default responses suggested the presence of both discrete and central values. We next examined point-to-point movements in a multijoint planar hand movement task with targets at two distances and two directions but at five directional separations (from 30° to 150° separation). We found that extent was again specified continuously from a central default. Direction was specified discretely from alternative default directions when target separation was wide and continuously from a central default when separation was narrow. The specification of both extent and direction evolved over a 200-ms time period beginning about 100 ms after target presentation. As in elbow force pulses, extent was specified progressively in both correct and wrong direction responses through a progressive improvement in the scaling of acceleration and velocity peaks to the target. On the other hand, movement time and hand path straightness did not change significantly in the course of specification. Thus, the specification of movement time and linearity, global features of the trajectories, are given priority over the specific values of extent and direction. In a third experiment, we varied the distances between unidirectional target pairs and found that movement extent is specified discretely, like direction, when the disparity in distances is large. The implications of these findings for contextual effects on trajectory planning are discussed. The independence of extent and direction specification and the prior setting of response duration and straightness provide critical support for the hypothesis that point-to-point movements are planned vectorially.

Impaired sequence learning in carriers of the DYT1 dystonia mutation

Previous positron emission tomography (PET) studies have shown that nonmanifesting carriers of the DYT1 dystonia mutation express an abnormal pattern of resting glucose metabolism. To determine whether motor behavior is impaired in these subjects, we compared movement and sequence learning in 12 clinically unaffected DYT1 carriers with 12 age‐matched controls. Regional differences in brain function during task performance were assessed with simultaneous H215O/PET. We found that motor performance was similar in the DYT1 and control groups, with no significant differences in movement time and spatial accuracy measured during each of the tasks. In contrast, sequence learning was reduced in gene carriers relative to controls (p < 0.01). PET imaging during motor execution showed increased activation in gene carriers (p < 0.001, uncorrected) in the left premotor cortex and right supplementary motor area, with concomitant reduction in the posterior medial cerebellum. During sequence learning, activation responses in DYT1 carriers were increased in the left ventral prefrontal cortex, and lateral cerebellum. These findings suggest that abnormalities in motor behavior and brain function exist in clinically nonmanifesting DYT1 carriers. Although localized increases in neural activity may enable normal movement execution in these subjects, this mechanism may not compensate for their defect in sequence learning. Ann Neurol 2003;54:102–109

Learning of scaling factors and reference axes for reaching movements

TO further understand visuomotor transformations in reaching, we compared adaptation to display rotation and altered gain in planar movements. Healthy subjects moved a cursor on a screen by moving an indicator on a horizontal digitizing tablet with their unseen hand. Adaptation to rotation was less complete and was accompanied by markedly increased directional variability. Adaptation training on a single target generalized broadly for gain change, but poorly for rotation. We propose that the difficulty in adapting to rotation arises from the substantial demands on short-term working memory imposed by the need to determine the new reference direction. Adaptation to gain change makes more modest demands on short-term memory to recalibrate the visuomotor scaling factor.