Loss of set in muscle responses to limb perturbations during cerebellar dysfunction.

The properties of electromyograph (EMG) responses that enabled the arm to return accurately to target following limb perturbations were investigated in five Cebus monkeys. In particular, factors that affected the timing and magnitude of an early antagonist response that occurred prior to stretch of the antagonist muscle were examined. The early antagonist response was large and early (latency, 60 ms) when the perturbation was brief and a constant force assisted the return movement. In this situation, early contraction of the antagonist muscle was required to prevent the return movement from overshooting the target. To determine whether this early antagonist response was influenced by prior instruction (which in this case was the type of perturbation the monkey had previously received), two types of perturbations requiring different EMG responses were studied. When torque steps (duration, 2,000 ms) were expected and were applied, monkeys generated M1, M2, and M3 responses and later activity only in the agonist (the initially stretched) muscle. When torque pulses (duration, 40 ms) were expected and were applied, monkeys generated M1 and M2 responses in the agonist and an early antagonist response. EMG responses to torque pulses and steps were then compared when the type of perturbation was expected and when it was unexpected. These comparisons revealed that the early antagonist response only occurred when the monkey expected a torque pulse. Therefore, this response was dependent on set. Expectation of a torque step caused enhancement of the agonist M2 and M3 responses. These agonist and antagonist EMG responses that were dependent on set were also influenced by changes in afferent drive. Cerebellar nuclear cooling through probes implanted lateral and medial to the dentate abolished that component of EMG responses attributed to set. The residual EMG responses in agonists and antagonists appeared to be driven by stretch of their respective muscles. The results suggest that when the nature of an arm perturbation is correctly predicted, the cerebellum provides accuracy in repositioning the limb a) by adjusting the magnitude of the M2 agonist response and b) by enabling activity after a latency of 60 ms (e.g., the M3 and early antagonist response) to be switched to the agonist or antagonist as appropriate, irrespective of which muscle is being stretched. This latter mechanism provides the motor system with predictive ability.