Search

Visual feedback is one of the key elements in on-line control of smooth manual tracking. To in- vestigate the effects basal ganglia dysfunction have on visual feedback control, we have tested six advanced Parkinson’s disease (PD) patients in comparison with normal controls using visually guided wrist tracking tasks. Tracking performance was assessed under three visual conditions: (1) both guiding target and movement cursor were displayed continuously; (2) the target display was turned off for the second half of each trial; or (3) the cursor display, but not the target, was turned off for the second half of each trial. Thus, for the second half of each trial under conditions 2 and 3, no visual feedback of the relationship between the target and the cursor was available. Results showed that although PD patients had significantly larger tracking errors than controls, and errors significantly increased in both PD patients and controls after withdrawing either visual cue, increases in tracking errors in PD were not significantly different from those in controls. Nor were any significant changes found in the frequency (6–8 Hz) or magnitude of the PD patient’s action tremor after withdrawing visual feedback. These results suggest that on-line movement control of wrist tracking movements in advanced PD is not especially reliant on visual feedback. In conjunction with our previous study of multiple sclerosis (MS) patients, the present results confirm that the basal ganglia is less involved in visual guidance of smooth manual tracking than the cerebellar circuits.

The akinesia of Parkinsonism is relieved by pallidotomy and subthalamic nucleotomy, but not by thalamotomy. Therefore, this disabling symptom probably depends upon connections other than the pallidalthalamocortical tracts, possibly efferents of the medial pallidum descending to the upper brainstem. We have previously demonstrated akinesia in the normal monkey following radiofrequency lesioning in the region of the pedunculopontine nucleus (PPN), one of the primary targets for descending pallidal outflow. Here, we confirm that selectively destroying neurones in the PPN area, whilst sparing fibres of passage, results in an akinetic state in normal macaques.

The motor thalamic areas receiving input from the globus pallidus (VA) and the cerebellar nuclei (VL) appear to have different roles in the generation and guidance of movements. In order to further test these differences, we used electrical stimulation to map the ventro-anterior and ventro-lateral nuclei of the thalamus in three ketamine anaesthetised monkeys. Movements were readily evoked from VL at currents of down to 10 microA. The movements were typically multijoint, and stimulation could evoke arm and trunk or arm and facial movement at the same current threshold. Evoked arm movements often involved multiple joints, with or without finger movements. Facial movements included the lips, tongue, jaw, eyebrows and, occasionally, the eyes. The thalamic map was topographic, but complex with at least two separate regions related to arm movement. Very few sites within the VA could stimulate movement, even at high currents. We therefore suggest that the cerebellar projections to motor regions of the cortex, which pass through the VL thalamic nuclei, have a different relationship and are closer to movement execution than the projections from basal ganglia via the ventro-anterior nucleus.

In multi-joint reaching movements, the motor system may choose any one of an infinite set of possible joint rotations to move the hand between given start and target positions. In order to find out whether reaching movements are represented in Cartesian hand coordinates or in joint coordinates, it is necessary to measure whether hand paths or joint paths have lower variability. We have measured hand paths and rotations of shoulder, elbow and wrist joints simultaneously in five subjects reaching in four orientations in the horizontal plane. As in earlier studies, we found a preference for nearly straight hand paths, despite different patterns of joint rotation for different orientations of movement. However, movements in three of four orientations showed a single principal joint, which rotated essentially without reversals. This may reflect optimisation in the motor system, preferring the simplest pattern of joint control for a desired hand path. We used generalised Procrustes analysis to quantify the variability in shape of repeated paths in hand space and joint space. Results showed that hand paths were less variable than the joint angles used to realise them, due to the kinematic redundancy of the limb, suggesting that hand paths, rather than joint angles, are directly represented by the motor system. Nevertheless, movements with straighter hand paths, on average, and those requiring coordinated activity at both shoulder and elbow joints also showed more variability in the shape of the hand path. Other orientations such as movement across the body use primarily a single joint and are less variable at the cost of a slightly curved path. These results suggest that coordinating multiple joints to produce a straight hand path has a definite computational cost. The motor system may perform a trade-off between the benefits of planning reaching movements as straight hand paths and the computational simplicity of executing them using patterns of joint rotation which simplify multi-joint coordination.

Lisberger's and Robinson's models of smooth pursuit predict very different results from altering retinal feedback delay. We have therefore investigated the effects of increasing the retinal feedback time delay in three normal human subjects by means of an artificial feedback paradigm. When additional delays were incorporated into the retinal feedback path a threshold was reached beyond which the eye exhibited sustained self-excited oscillations or "limit cycles". The oscillation period increased linearly (as the added delay was increased) with slopes ranging from 1.41 to 1.6 with zero-delay intercepts of between 212 and 306 ms. Contrary to our experimental findings the Robinson and Lisberger models predict that the plot of oscillation period against added delay should have a slope of 3.4 and 2.7 and an intercept of 479 and 554 ms, respectively. Neither model produced comparable limit cycles, both being unstable at delays greater than 280 ms. Our results imply that the models of smooth pursuit need to incorporate predictive control.