The complete frequency spectrum of physiological tremor can be recreated by broadband mechanical or electrical drive

Two frequency peaks of variable preponderance have been reported for human physiological finger tremor. The high-frequency peak (20-25 Hz, seen only in postural tremor) is generally attributed to mechanical resonance, whereas the lower frequency peak (8-12 Hz, seen in both postural and kinetic tremor) is usually attributed to synchronous central or reflexive neural drive. In this study, we determine whether mechanical resonance could generate both peaks. In relaxed subjects, an artificial finger tremor was evoked by random mechanical perturbations of the middle finger or random electrical muscular stimulation of the finger extensor muscle. The high and the low frequencies observed in physiological tremor could both be created by either type of artificial input at appropriate input intensity. Resonance, inferred from cross-spectral gain and phase, occurred at both frequencies. To determine any neural contribution, we compared truly passive subjects with those who exhibited some electromyographic (EMG) activity in the finger extensor; artificially created tremor spectra were almost identical between groups. We also applied electrical stimuli to two clinically deafferented subjects lacking stretch reflexes. They exhibited the same artificial tremor spectrum as control subjects. These results suggest that both typical physiological finger tremor frequencies can be reproduced by random artificial input; neither requires synchronized neural input. We therefore suggest that mechanical resonance could generate both dominant frequency peaks characteristic of physiological finger tremor. The inverse relationship between the input intensity and the resulting tremor frequency can be explained by a movement-dependent reduction in muscle stiffness, a conjecture we support using a simple computational model.