Instability in human forearm movements studied with feed‐back‐controlled electrical stimulation of muscles.

1. Amplitude‐modulated electrical stimulation was applied to the elbow flexors and extensors to produce movements of the forearm in normal subjects. The parameters of the modulating (command) signal were set in isometric trials so as to produce equal and opposite background torques, and equal and supportive torque modulations. 2. Bode plots relating forearm movement to command signal (modulating) frequency showed the muscle‐load to have a low‐pass characteristic similar to that previously described in the cat, and a slightly larger bandwidth than described previously in man. 3. The transduced forearm signals were fed back to provide the command signal to the stimulators via a filter which mimicked the transfer function of muscle spindle primary endings. In effect this replaced the neural part of the reflex arc with an accessible model, but left the muscle‐load effector intact. 4. All six subjects developed forearm oscillations (tremor) when the loop gain exceeded a threshold value. The mean tremor frequency at onset was 4.4 Hz, which was similar to that of the equivalent vibration‐evoked tremor (previous paper, Prochazka & Trend, 1988). 5. With the linear spindle model, oscillations tended to grow rapidly in amplitude, and the stimuli became painful. The inclusion of a logarithmic limiting element resulted in stable oscillations, without significant alterations in frequency. This allowed us to study the effect on tremor of including analog delays in the loop, mimicking those associated with peripheral nerve transmission and central reflexes. In one subject, loop delays of 0, 20, 40 and 100 ms resulted in tremor at 4.0, 3.6, 3.0 and 2.1 Hz respectively, as quantified by spectral analysis. 6. By considering separately the phase contributions of the different elements of the reflex arc, including delays, it became clear that muscle‐load properties were important in setting the upper limit of tremor frequencies which could conceivably be supported by reflexes. 7. The results support the conclusion of the related vibration study (Prochazka & Trend, 1988), that for moderate levels of background co‐contraction, the contribution of stretch reflexes to tremor at the elbow should be sought in the 3‐5 Hz range. Exaggerated long‐latency reflexes would be expected to reduce these baseline frequencies by 1 or 2 Hz.