Central neural mechanisms contributing to cerebellar tremor produced by limb perturbations.

1. Central mechanisms contributing to cerebellar tremor were studied in three Cebus monkeys who were trained to return their arm rapidly to a target region after it was displaced by a perturbation applied to a handle. Cooling through cryoprobe sheaths implanted alongside the dentate and interpositus nuclei resulted in a series of oscillations (tremor) following the perturbation. 2. During progressive cooling from control conditions there was a progressive increase in the number and amplitude of the oscillations and a progressive decrease in their frequency (from 6-8 to 3-5 Hz). The instability of these oscillations and their amplitude and frequency were found to be related to the degree of cerebellar dysfunction. 3. Analysis of EMG activity in biceps and triceps confirmed that during cooling no consistent changes occurred in the segmental (20 ms) and suprasegmental (35-100 ms) reflex responses in the agonist (stretched) muscle, which contributed to the return of the arm following the perturbation. This corrective return was normally actively terminated by EMG activity in the antagonist muscle, occurring 70100 ms after onset of the torque pulse. During cooling there was a delay in onset of the antagonist activity that resulted in the corrective return overshooting the target. In addition, antagonist activity was prolonged, thus causing a second cycle of oscillation. A similar delay and prolonged burst then occurred in the agonist, and then again in the antagonist, which resulted in continued oscillations. 4. Responses of 74 precentral neurons responding to the torque pulse were studied under control conditions and during cerebellar cooling. Of these, 24 were closely related to activity of either the biceps or triceps muscle in that they had reciprocal responses: inhibition (20-50 ms) followed by excitation (50100 ms) for one direction of perturbation, and excitation followed by inhibition for the other direction. When the neuron was related to the muscle that was the antagonist, the excitatory second cortical component (cf. intended response (6)) normally appeared in advance of the EMG activity. During cooling no major change usually occurred in the first cortical response, but the second (antagonist related) response was delayed so as to occur after the onset of antagonist EMG instead of before it. Thus the onset of this neural response occurred after the start of muscle stretch, whereas in normal movements it was predictive; that is, it occurred prior to the start of muscle stretch. 5. Following a limb perturbation, return of the limb is brought about, partly by mechanical factors and partly by segmental and suprasegmental stretch reflexes. It is suggested that, in advance of this return, the motor cortex generates a command (the 50to lOO-ms intended response) to the antagonist muscle to terminate the return, on the basis of predictive information provided by the cerebellum. During cerebellar dysfunction the motor cortex does not receive predictive information, but receives only delayed information resulting from stretch of the antagonist muscle. This results in a descending command, which ar-